A vehicle control method, apparatus, vehicle and medium

Abstract

The application discloses a vehicle control method, device, vehicle and medium, the vehicle comprises a power system, and the method comprises the following steps: acquiring a power state parameter and a current vehicle speed of the vehicle at a current time, determining a power mode, a mode cumulative change degree and a driving intention of the driver of the vehicle according to the power state parameter and the current vehicle speed, acquiring a historical vehicle speed prediction value, determining a vehicle speed residual change degree according to the current vehicle speed and the historical vehicle speed prediction value, acquiring an initial state transition probability corresponding to the power mode and the driving intention, adjusting the initial state transition probability according to the mode cumulative change degree and the vehicle speed residual change degree to obtain a target state transition probability, determining a predicted vehicle speed sequence in a prediction time domain according to the current vehicle speed and the target state transition probability, generating a control instruction, and adjusting the operating parameter of the power system according to the control instruction. The state transition probability of the application can be adaptively updated following the working condition change, and the accuracy of vehicle control and user experience are improved.

Description

Technical Field

[0001] This application belongs to the field of vehicle control technology, specifically relating to a vehicle control method, device, vehicle, and medium. Background Technology

[0002] As the automotive industry moves towards intelligent and electric vehicles, traditional vehicle control systems face new challenges and opportunities, particularly in areas such as energy management, torque control, and NVH (Noise, Vibration, Harshness) control. These areas demand more efficient and precise control strategies to adapt to increasingly complex operating conditions. In such controls, the controller often needs to anticipate short-term future driving demands, such as vehicle speed sequences, to rapidly generate SOC (State of Charge) reference trajectories, optimize energy management solutions, and mitigate noise conditions.

[0003] Currently, vehicle speed prediction generally relies on historical speed data and powertrain information. However, existing prediction methods often fail to adequately consider the impact of different power modes and varying driver intentions on future speed, leading to significant discrepancies between predicted and actual speeds in certain situations. Particularly in complex road conditions and multi-mode driving scenarios, the prediction accuracy of existing methods falls short of practical requirements, consequently affecting vehicle handling accuracy and user experience. Summary of the Invention

[0004] The purpose of this application is to provide a vehicle control method, device, vehicle, and medium, which is achieved as follows: In a first aspect, embodiments of this application provide a vehicle control method, the vehicle including a power system, the method comprising: The vehicle's power state parameters and current speed are obtained at the current moment. Based on the power state parameters and current speed, the vehicle's power mode, cumulative mode change degree, and driver's driving intention are determined. Obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual change degree based on the current vehicle speed and the historical vehicle speed prediction value; Obtain the initial state transition probabilities corresponding to the power mode and the driving intention from a pre-constructed set of state transition probabilities; The initial state transition probability is adjusted based on the cumulative change of the mode and the change of the vehicle speed residual to obtain the target state transition probability; The predicted future vehicle speed at each prediction time point within the prediction time domain is determined based on the current vehicle speed and the target state transition probability. The predicted vehicle speed values ​​at each of the predicted times are combined in chronological order to obtain the predicted vehicle speed sequence in the prediction time domain. Control commands are generated based on the predicted vehicle speed sequence, and the operating parameters of the power system are adjusted according to the control commands.

[0005] Secondly, embodiments of this application provide a vehicle control device, the vehicle including a power system, the device comprising: The parameter acquisition module is used to acquire the vehicle's power state parameters and current speed at the current moment, and determine the vehicle's power mode, cumulative mode change degree, and driver's driving intention based on the power state parameters and current speed. The residual variation determination module is used to obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual variation value based on the current vehicle speed and the historical vehicle speed prediction value. The initial transition probability acquisition module is used to acquire the initial state transition probability corresponding to the power mode and the driving intention from a pre-built set of state transition probabilities; The target transition probability acquisition module is used to adjust the initial state transition probability according to the cumulative change degree of the mode and the change degree of the vehicle speed residual to obtain the target state transition probability; The vehicle speed prediction module is used to determine the future vehicle speed prediction value at each prediction time in the prediction time domain based on the current vehicle speed and the target state transition probability. The vehicle speed sequence prediction module is used to combine the future vehicle speed prediction values ​​at each prediction time according to the time sequence of the prediction time to obtain the predicted vehicle speed sequence in the prediction time domain. The vehicle control module is used to generate control commands based on the predicted vehicle speed sequence and adjust the operating parameters of the power system according to the control commands.

[0006] Thirdly, embodiments of this application provide a vehicle including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the method described above.

[0007] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, and when the program or instructions are executed by a processor, they implement the method described above.

[0008] The embodiments of this application have the following advantages: This application achieves real-time perception and adaptive operation of driving behavior by acquiring the vehicle's dynamic state parameters and current speed at the current moment, and determining the vehicle's power mode, cumulative mode change, and driver's driving intention based on these parameters and speed. It also achieves online closed-loop feedback of prediction errors by acquiring the historical speed prediction value corresponding to the current moment and determining the speed residual change based on the current and historical speed prediction values. Furthermore, it obtains the initial state transition probabilities corresponding to the power mode and driving intention from a pre-constructed set of state transition probabilities, and adjusts these probabilities based on the cumulative mode change and speed residual change to obtain the target state transition probability, enabling the transition probability to adaptively update with changes in operating conditions. Finally, it determines the future speed prediction value for each prediction moment within the prediction time domain based on the current speed and the target state transition probability, utilizing the historical statistical patterns reflected in the target state transition probability while satisfying the physical constraints of the current power system, effectively improving prediction accuracy under complex road conditions and multi-mode operating conditions. The predicted vehicle speed values ​​at each prediction time are combined in chronological order to obtain a predicted vehicle speed sequence within the prediction time domain. Control commands are then generated based on this sequence, and the operating parameters of the powertrain are adjusted accordingly. This allows the controller to anticipate future vehicle speed demands, proactively optimizing energy management strategies within the prediction time domain, thereby improving vehicle handling accuracy and user experience. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0010] Figure 1 This is a flowchart illustrating the steps of a vehicle control method according to an embodiment of this application; Figure 2 This is a schematic diagram of a conditional transition tensor provided in an embodiment of this application; Figure 3 This is a schematic diagram of an reachability domain projection correction provided in an embodiment of this application; Figure 4 This is a diagram of a range-extended vehicle powertrain system provided in one embodiment of this application; Figure 5 This is a logical schematic diagram of a vehicle control method provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a vehicle control device provided in an embodiment of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been presented in the various embodiments of this application to enable readers to better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and updates based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0012] In existing technologies, the accuracy and stability of vehicle speed prediction cannot meet actual needs under different driving modes and operating conditions. Under dynamic changing conditions such as range extender start-stop and battery SOC limit, existing methods are difficult to adapt to the above changes, and the prediction results have large errors and cannot reflect the real vehicle driving status.

[0013] Existing vehicle speed prediction methods do not fully consider the impact of driving intentions such as acceleration, deceleration, and coasting on vehicle speed prediction. Vehicle speed not only depends on historical data, but is also closely coupled with current driving behavior. There is a lack of effective recognition of driving intentions, and the prediction accuracy is significantly reduced under non-steady conditions such as rapid acceleration and sudden braking.

[0014] Vehicle speed prediction is affected by factors such as driving habits and road conditions. The system fails to assess the reliability of the prediction output in real time, lacks self-correction logic for closed-loop compensation based on historical prediction errors, and cannot continuously iterate and optimize in response to environmental changes. The prediction error accumulates and adds up as the prediction time domain extends, affecting the subsequent decision-making level's acceptance of the prediction results.

[0015] Existing vehicle speed prediction methods do not consider the acceleration and deceleration capabilities of the power system, ignore the inherent physical limits of the vehicle power system such as the maximum output power of the motor, torque saturation characteristics, and the current SOC limit of the power battery. The prediction results exceed the actual capabilities of the vehicle, and the predicted acceleration or vehicle speed under low speed or power-limited conditions does not conform to reality.

[0016] Therefore, this application provides a vehicle control method, device, vehicle, and medium that can determine the power mode, mode cumulative change degree, and driving intention based on power state parameters and current vehicle speed, determine the vehicle speed residual change degree based on current vehicle speed and historical vehicle speed prediction values, dynamically adjust the state transition probability using mode cumulative change degree and vehicle speed residual change degree, recursively predict future vehicle speed sequences, and generate control commands. This achieves adaptive following of driving mode and operating condition changes by the prediction model, ensuring that the predicted vehicle speed is always within the vehicle's physical capabilities, thereby improving the accuracy of vehicle control and user experience.

[0017] Reference Figure 1The diagram shows a flowchart of the steps of a vehicle control method provided in an embodiment of this application.

[0018] In this embodiment, the vehicle includes a power system, which refers to the assembly that provides energy and driving force for the vehicle's movement, including a range extender, a motor, a power battery, and a corresponding transmission mechanism. The range extender is a combination of an engine and a generator, which starts generating electricity when the power battery is low, either to provide power to the motor or to charge the power battery.

[0019] This method may specifically include the following steps: Step 101: Obtain the vehicle's current power state parameters and current speed. Based on the power state parameters and current speed, determine the vehicle's power mode, cumulative mode change, and the driver's driving intention.

[0020] In this embodiment of the application, the current dynamic state parameters and the current vehicle speed can be obtained.

[0021] Among them, the dynamic state parameters are a set of physical quantities describing the current operating state of the vehicle's powertrain, including the motor output torque, brake pedal opening, and accelerator pedal opening. Current vehicle speed is the vehicle's actual speed relative to the ground at the current moment.

[0022] In this embodiment, the vehicle's power mode, cumulative mode change, and driver's driving intention can be determined based on the power state parameters and the current vehicle speed.

[0023] Among them, the power mode is the dominant operating type of the vehicle's power system. In specific implementation, the power mode can include drive mode, compound braking mode and braking mode.

[0024] The cumulative change in the pattern is a quantitative value of the degree to which the dynamic state characteristics have continuously deviated from the normal baseline. This value accumulates positively over time and reflects the severity of changes in the dynamic pattern.

[0025] The driver's driving intention is the driver's subjective expectation of the vehicle's longitudinal movement at the current moment. In practice, driving intention can include acceleration intention, cruising intention, coasting intention, and braking intention.

[0026] Step 102: Obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual change degree based on the current vehicle speed and the historical vehicle speed prediction value.

[0027] In this embodiment of the application, the historical vehicle speed prediction value corresponding to the current moment can be obtained. The historical vehicle speed prediction value is the vehicle speed prediction result for the current moment calculated in the previous round of vehicle speed prediction process prior to the current moment.

[0028] The method in this application runs continuously from the moment the vehicle starts, performing a complete vehicle speed prediction process at each sampling moment. The historical vehicle speed prediction value corresponding to the current moment is actually the vehicle speed prediction result calculated in advance when the prediction process was executed at the previous sampling moment.

[0029] For example, the sampling period is 0.1 seconds. At time 0.1 seconds, the vehicle performs a speed prediction, resulting in a predicted speed sequence that includes predicted speeds for future times such as 0.2 seconds, 0.3 seconds, 0.4 seconds, and so on. When time progresses to time 0.2 seconds, the predicted speed value corresponding to time 0.2 seconds from the previous prediction process—that is, the prediction executed at time 0.1 seconds—is the historical speed prediction value for the current time 0.2 seconds.

[0030] In other words, when the vehicle reaches the current moment, the previous prediction process has already provided the vehicle speed prediction result for the current moment, which is the historical vehicle speed prediction value.

[0031] At the first sampling moment after vehicle startup, since the vehicle speed prediction process has not been executed during this trip, there are no historical vehicle speed prediction values. At this time, the initial value of the historical vehicle speed prediction value can be set as the current vehicle speed. Then, the difference between the current vehicle speed and the historical vehicle speed prediction value is zero, the vehicle speed residual change is zero, and no adjustment is triggered. From the second sampling moment onwards, the historical vehicle speed prediction value exists normally, and the vehicle speed residual change is calculated normally.

[0032] In this embodiment, the degree of variation of the vehicle speed residual can be determined based on the current vehicle speed and historical vehicle speed prediction values. The degree of variation of the vehicle speed residual is a quantified value of the abnormality of the current prediction error relative to historical fluctuation levels.

[0033] Specifically, the current vehicle speed is compared with historical predicted vehicle speeds; the degree of difference between the two is the current prediction error. Based on the relative relationship between the fluctuation levels of the current prediction error and the historical prediction error, the degree of change in the vehicle speed residual can be determined. By comparing the current actual vehicle speed with historical predicted vehicle speeds, the accuracy of the previous prediction can be evaluated in real time.

[0034] Step 103: Obtain the initial state transition probabilities corresponding to the power mode and driving intention from the pre-constructed set of state transition probabilities.

[0035] In this embodiment of the application, a pre-constructed set of state transition probabilities can be obtained.

[0036] Among them, the state transition probability set is a pre-built data structure that records the statistical probability of vehicle speed transitioning between adjacent time points under different power modes and different driving intentions.

[0037] In this embodiment of the application, the initial state transition probabilities corresponding to the power mode and driving intention can be obtained from the set of state transition probabilities.

[0038] The initial state transition probability is the state transition probability data obtained from the state transition probability set based on the current power mode and the current driving intention. The initial state transition probability is used to describe the statistical law of the evolution of vehicle speed under the current operating conditions and serves as the basis for subsequent adaptive adjustments.

[0039] Step 104: Adjust the initial state transition probability according to the cumulative change degree of the mode and the change degree of the vehicle speed residual to obtain the target state transition probability.

[0040] In this embodiment, the degree of drastic change in the current operating condition can be determined based on the cumulative change in the power mode and the residual change in vehicle speed. The cumulative change in the power mode reflects the drastic change in the power mode, while the residual change in vehicle speed reflects the degree of abnormality in the current prediction error relative to historical fluctuation levels. Together, they constitute a comprehensive judgment on the stability of the current operating condition.

[0041] In this embodiment, the initial state transition probability can be adaptively adjusted according to the degree of change in the current working condition to obtain the target state transition probability.

[0042] The target state transition probability is the transition probability data obtained after correcting the initial state transition probability based on the current prediction error and the degree of drastic change in the model.

[0043] In practical implementation, when the cumulative change in mode or the change in vehicle speed residual increases, it indicates a significant change in the current operating condition, and the reference value of historical data in the initial state transition probability decreases. In this case, the adjustment range of the initial state transition probability is increased to make it respond more quickly to changes in operating conditions. Conversely, when both changes are at low levels, it indicates that the current operating condition is relatively stable, and the reference value of historical data in the initial state transition probability is high. In this case, the adjustment range is decreased to maintain the stability of the transition probability. This achieves adaptive following of the initial state transition probability to changes in driving mode and operating conditions.

[0044] Step 105: Determine the predicted future vehicle speed at each prediction time within the prediction time domain based on the current vehicle speed and the target state transition probability.

[0045] In the embodiments of this application, the vehicle speed transfer pattern corresponding to each prediction time in the prediction time domain can be obtained.

[0046] The prediction time domain is a continuous time range extending from the current moment into the future, which includes multiple discrete prediction moments.

[0047] In the embodiments of this application, the future vehicle speed prediction value at each prediction time can be determined based on the current vehicle speed and the vehicle speed transfer pattern at each prediction time within the prediction time domain.

[0048] Among them, the future vehicle speed prediction value refers to the estimated vehicle speed at a certain prediction time within the prediction time domain.

[0049] In practical implementation, the current vehicle speed can be used as the starting value for prediction. By utilizing the vehicle speed state evolution law described by the target state transition probability, the vehicle speed value at each prediction time in the prediction time domain can be calculated step by step to obtain the future vehicle speed prediction value at each prediction time.

[0050] Step 106: Combine the predicted vehicle speed values ​​at each prediction time according to the chronological order of the prediction times to obtain the predicted vehicle speed sequence in the prediction time domain.

[0051] In this embodiment of the application, the predicted future vehicle speed values ​​at each prediction time can be combined according to the chronological order of the prediction times.

[0052] In this context, each prediction time refers to a series of discrete time points arranged sequentially from the current time within the prediction time domain, with each time point corresponding to a predicted future vehicle speed value. Arranging these predicted future vehicle speed values ​​sequentially from earliest to latest time forms a complete sequence of vehicle speed changes over time, which is the predicted vehicle speed sequence.

[0053] Step 107: Generate control commands based on the predicted vehicle speed sequence, and adjust the operating parameters of the power system according to the control commands.

[0054] In this embodiment of the application, control commands can be generated based on the predicted vehicle speed sequence.

[0055] Among them, the control command is a specific operation command generated by the controller based on the predicted vehicle speed sequence, used to adjust the working state of the power system.

[0056] In this embodiment, the operating parameters of the power system can be adjusted according to control commands.

[0057] Among them, the operating parameters are the specific values ​​that can be adjusted when the various components of the power system are working. Adjusting the operating parameters means changing the actual output values ​​of the various actuators in the power system so that the actual operating state of the vehicle matches the future driving needs reflected by the predicted vehicle speed sequence.

[0058] In practice, adjusting operating parameters means changing the actual output values ​​of each actuator in the power system in advance based on the future speed change trend reflected by the predicted vehicle speed sequence, so that the actual operating state of the vehicle matches the future driving needs.

[0059] For example, when the predicted vehicle speed sequence indicates an upcoming acceleration phase, the range extender's power generation can be adjusted in advance to store electrical energy, increasing the motor's output torque limit and ensuring timely response to acceleration demands. When the predicted vehicle speed sequence indicates an upcoming deceleration or coasting phase, the range extender's operating state can be adjusted in advance to increase energy recovery intensity, converting braking energy into electrical energy stored in the battery. This achieves proactive optimization and adjustment of the powertrain's operating parameters within the prediction time domain.

[0060] This application achieves real-time perception and adaptive operation of driving behavior by acquiring the vehicle's dynamic state parameters and current speed at the current moment, and determining the vehicle's power mode, cumulative mode change, and driver's driving intention based on these parameters and speed. It also achieves online closed-loop feedback of prediction errors by acquiring the historical speed prediction value corresponding to the current moment and determining the speed residual change based on the current and historical speed prediction values. Furthermore, it obtains the initial state transition probabilities corresponding to the power mode and driving intention from a pre-constructed set of state transition probabilities, and adjusts these probabilities based on the cumulative mode change and speed residual change to obtain the target state transition probability, enabling the transition probability to adaptively update with changes in operating conditions. Finally, it determines the future speed prediction value for each prediction moment within the prediction time domain based on the current speed and the target state transition probability, utilizing the historical statistical patterns reflected in the target state transition probability while satisfying the physical constraints of the current power system, effectively improving prediction accuracy under complex road conditions and multi-mode operating conditions. The predicted vehicle speed values ​​at each prediction time are combined in chronological order to obtain a predicted vehicle speed sequence within the prediction time domain. Control commands are then generated based on this sequence, and the operating parameters of the powertrain are adjusted accordingly. This allows the controller to anticipate future vehicle speed demands, proactively optimizing energy management strategies within the prediction time domain, thereby improving vehicle handling accuracy and user experience.

[0061] Optionally, the dynamic state parameters include motor output torque, brake pedal opening, and accelerator pedal opening.

[0062] In this embodiment, the motor output torque can refer to the actual mechanical torque output by the motor at the current moment. The brake pedal opening can refer to the proportion of the driver's travel when pressing the brake pedal. The accelerator pedal opening can refer to the proportion of the driver's travel when pressing the accelerator pedal.

[0063] The steps for determining the vehicle's power mode, cumulative mode change, and driver's driving intention based on power state parameters and current vehicle speed include: S1011 determines the power mode based on the motor output torque and brake pedal opening. S1012 determines the dynamic state characteristics based on the motor output torque, brake pedal opening and current vehicle speed; S1013, Determine the cumulative degree of change of the model based on the dynamic state characteristics; S1014, determine the driver's driving intention based on the current vehicle speed, brake pedal opening, and accelerator pedal opening.

[0064] In this embodiment, the vehicle's power mode can be determined based on the motor output torque and the brake pedal opening. Specifically, when the brake pedal opening is small and the motor output torque is positive, it indicates the vehicle is in a driving state, and the power mode is determined to be the driving mode. When the brake pedal opening is small and the motor output torque is negative, it indicates the vehicle is in an energy recovery state, and the power mode is determined to be the combined braking mode. When the brake pedal opening is large, it indicates mechanical braking has engaged, and the power mode is determined to be the braking mode. This completes the determination of the vehicle's current dominant operating type.

[0065] In this embodiment, the dynamic state characteristics can be determined based on the motor output torque, brake pedal opening, and current vehicle speed. These dynamic state characteristics are a comprehensive descriptive quantity composed of the motor output torque, brake pedal opening, and current vehicle speed, used to characterize the vehicle's dynamic state at the current moment. This characteristic provides a quantitative input basis for subsequent steps.

[0066] In this embodiment, the cumulative degree of change of the mode can be determined based on the dynamic state characteristics. The cumulative degree of change of the mode is a quantified value of the cumulative extent to which the dynamic state characteristics have continuously deviated from their normal level, and is used to characterize the severity of the change in the dynamic mode.

[0067] In this embodiment, the driver's driving intention can be determined based on the current vehicle speed, brake pedal opening, and accelerator pedal opening. The driver's driving intention is the driver's subjective expectation of the vehicle's longitudinal motion at the current moment, representing whether the driver wants the vehicle to accelerate, maintain speed, or decelerate. In specific implementation, after calculating the current acceleration based on the current vehicle speed, the driver's driving intention at the current moment can be determined by combining the accelerator pedal opening, brake pedal opening, and the current acceleration with their respective preset thresholds.

[0068] In practical implementation, the dynamic state characteristics can be constructed in the following ways: At discrete time k, signals closely related to the changes in the power system state are acquired from the CAN bus: motor output torque, braking signal, and vehicle speed. To unify the dimensions and facilitate the fusion of different signals, a normalized feature vector is constructed:

[0069] in, It represents the characteristics of dynamic state. This indicates the motor's output torque. Indicates the brake pedal opening. Indicates the current vehicle speed. This indicates the motor's output torque at the previous moment. This indicates the brake pedal opening at the previous moment. This represents the global torque upper limit constant used for normalization. This represents the global brake pedal opening upper limit constant used for normalization. This represents the global vehicle speed upper limit constant used for normalization.

[0070] The aforementioned global calibration upper limit constant is used to unify dynamic state parameters with different dimensions into a single scale system, ensuring the comparability of characteristic quantities at different times and under different modes. The value of the global calibration upper limit constant can be set according to the signal range of the vehicle platform. In one embodiment, It can be set to 400 N·m. It can be set to 100%. It can be set to 180km / h. Among them, this... Used for feature normalization, it can be understood as a global reference torque upper limit, which is distinct from the upper limit of the maximum driving torque at the wheel end that varies with the power system mode.

[0071] This application achieves comprehensive recognition of vehicle operating status and driver's operating intention by determining the power mode based on motor output torque and brake pedal opening, determining the cumulative change degree of the mode based on power state characteristics, and determining the driving intention based on current vehicle speed and pedal signal. It solves the technical problem that existing methods cannot distinguish between different driving modes and driving intentions, and provides an accurate working condition perception basis for subsequent vehicle speed prediction.

[0072] Optionally, the step of determining the power mode based on the motor output torque and brake pedal opening includes: S1021, Obtain the preset torque threshold and braking mode opening threshold; S1022, If the brake pedal opening is less than the braking mode opening threshold and the motor output torque is greater than or equal to the torque threshold, then the power mode is determined to be the drive mode. S1023, If the brake pedal opening is less than the braking mode opening threshold and the motor output torque is less than or equal to the negative torque threshold, then the power mode is determined to be the compound braking mode. S1024, if the brake pedal opening is greater than or equal to the braking mode opening threshold, then the power mode is determined to be the braking mode.

[0073] In this embodiment, preset torque thresholds and braking mode opening thresholds can be obtained. The torque threshold is a dividing value used to determine whether the motor output torque is in a driving or braking state, and the braking mode opening threshold is a dividing value used to determine whether the brake pedal is effectively depressed. Both thresholds are pre-calibrated and stored as fixed parameters in the controller.

[0074] In practical implementation, the torque threshold can be set to 20-30 N·m, for example, T0 = 30 N·m. The braking mode opening threshold can be set to 3%-5%, for example, 5%.

[0075] In this embodiment, when the brake pedal opening is less than the braking mode opening threshold, it indicates that the driver has not pressed the brake pedal, and mechanical braking has not engaged. When the motor output torque is greater than or equal to the torque threshold, it indicates that the motor outputs sufficient driving torque in the positive direction. When both conditions are met simultaneously, the vehicle is in a driving state of active acceleration or maintaining vehicle speed, and the power mode is determined as the driving mode.

[0076] In this embodiment, when the brake pedal opening is less than the braking mode opening threshold, it indicates that the driver has not pressed the brake pedal, and mechanical braking has not engaged. When the motor output torque is less than or equal to the negative torque threshold, it indicates that the motor is outputting negative torque, and the motor is working as a generator, converting the vehicle's kinetic energy into electrical energy for recovery. When both conditions are met simultaneously, the vehicle is in an energy recovery state, and the power mode is determined to be a composite braking mode.

[0077] In this embodiment, when the brake pedal opening is greater than or equal to the braking mode opening threshold, it indicates that the driver has pressed the brake pedal and the mechanical braking system has engaged. At this time, regardless of the motor output torque value, the vehicle is in a mechanical braking state, and the power mode is determined as the braking mode. The braking mode has the highest priority and, once triggered, overrides the determination results of other modes.

[0078] In the specific implementation, candidate power mode labels are constructed based on the current torque and braking signal. This is used to describe the dominant operating state of the current power system, and the specific formula is as follows:

[0079] in, These labels indicate the driving mode. Mode 1 indicates the driving mode. Mode 2 indicates the combined braking mode. Mode 3 indicates the braking mode. Indicates the torque threshold. Indicates the braking mode opening threshold. This indicates the motor's output torque. This indicates the brake pedal opening.

[0080] In practical implementation, taking a torque threshold of 30 N·m and a braking mode opening threshold of 5% as an example, when the brake pedal opening is less than 5% and the motor output torque is greater than or equal to 30 N·m, it indicates that the driver has not pressed the brake pedal and the motor outputs sufficient driving torque in the positive direction. In this case, the power mode is determined to be the drive mode. When the brake pedal opening is less than 5% and the motor output torque is less than or equal to -30 N·m, it indicates that the driver has not pressed the brake pedal and the motor outputs negative torque for energy recovery. In this case, the power mode is determined to be the compound braking mode. When the brake pedal opening is greater than or equal to 5%, it indicates that mechanical braking has been engaged. In this case, the power mode is determined to be the braking mode.

[0081] By setting the above threshold, it is possible to effectively avoid frequent misjudgments of power mode caused by small fluctuations in motor output torque or slight vibrations in the brake pedal signal, thus ensuring the stability and reliability of the mode determination results.

[0082] To avoid mode jitter, let the time of the last mode switch be... The minimum number of steps to hold is When the following conditions are met: , , When a pattern change occurs, determine that a pattern variable has occurred and update the pattern variables:

[0083] in, This represents the cumulative feature offset. This indicates the threshold for determining the change point. Indicates candidate dynamic modes, Indicates the power mode. Let k represent the previous dynamic mode and k represent the current time. Indicates the time of the last mode switch. This represents the minimum number of steps to hold.

[0084] A mode change is determined only when all three conditions are met simultaneously: the cumulative feature offset is greater than or equal to the change-point determination threshold, the candidate dynamic mode is different from the previous dynamic mode, and the current time is greater than or equal to the minimum hold-up step. In this case, the candidate dynamic mode is taken as the current dynamic mode, the current time is updated to the previous mode switch time, and the cumulative feature offset is cleared to zero. The aforementioned minimum hold-up step is used to avoid frequent dynamic mode switching and ensure the stability of the mode determination results.

[0085] If the above conditions are not met, the pattern remains unchanged:

[0086] In other words, the candidate dynamic mode remains the same as the dynamic mode at the previous moment.

[0087] In one embodiment, when the sampling period Minimum hold steps when the hold time is 0.1s Option 5 is acceptable, meaning the candidate mode must persist for at least 0.5 seconds before a formal switch is allowed. Only when all three conditions are met simultaneously is a stable mode change confirmed, and the power mode is officially updated. At the same time, the cumulative feature offset cumulative statistics will be calculated. The process is reset to zero to restart the next round of change point accumulation. If any condition is not met, it is considered that the current fluctuation is only short-term or the dynamic pattern has not yet stabilized, and the previous dynamic pattern remains unchanged.

[0088] This application obtains preset torque thresholds and braking mode opening thresholds, and classifies power modes into drive mode, compound braking mode, and braking mode based on the comparison results of brake pedal opening and braking mode opening thresholds and motor output torque and torque thresholds. This solves the technical problem of not being able to accurately distinguish different power modes of the vehicle, and provides a working condition classification basis that matches the actual working state of the power system for subsequent vehicle speed prediction.

[0089] Optionally, the step of determining the cumulative degree of change of the model based on the dynamic state characteristics includes: S1031, determine the feature offset based on the difference between the dynamic state characteristics and the baseline characteristics; wherein, the baseline characteristics are obtained by exponential moving average of historical dynamic state characteristics; S1032, the feature offset is positively accumulated over time to obtain the cumulative feature offset; S1033, Determine the initial mode cumulative change degree based on the ratio of the cumulative feature offset to the preset change point determination threshold; S1034, If the initial cumulative change of the mode is greater than the preset cumulative change threshold of the mode, then the cumulative change threshold of the mode is used as the cumulative change of the mode. S1035, if the initial cumulative change of the mode is less than or equal to the cumulative change threshold of the mode, then the initial cumulative change of the mode is taken as the cumulative change of the mode.

[0090] In this embodiment, the feature offset can be determined based on the difference between the dynamic state characteristics and the baseline characteristics.

[0091] Among them, the dynamic state characteristics are a comprehensive descriptive quantity composed of motor output torque, brake pedal opening and current vehicle speed, used to characterize the dynamic state of the vehicle at the current moment.

[0092] The baseline feature is obtained by exponentially moving average of historical dynamic state features, representing the characteristic level of the vehicle under normal operating conditions. Comparing the dynamic state features with the baseline feature, the difference between the two is the feature offset, which reflects the direction and degree of deviation of the vehicle's dynamic state from the normal level at the current moment.

[0093] In this embodiment, the feature offset can be positively accumulated over time to obtain the accumulated feature offset. Positive accumulation refers to adding up the portion of the feature offset exceeding a preset drift compensation constant at each time point. This accumulation operation is performed continuously over time, and the accumulation result reflects the continuous accumulation effect of the feature offset over a period of time.

[0094] When the feature offset is less than the drift compensation constant, the cumulative value gradually decreases to avoid the long-term impact of small random fluctuations on the cumulative result. The resulting cumulative feature offset is used to characterize the degree to which the dynamic state characteristics deviate from the normal level.

[0095] In this embodiment of the application, the cumulative change degree of the initial mode can be determined based on the ratio of the cumulative feature offset to the preset change point determination threshold.

[0096] The change point determination threshold is a preset boundary value used to determine whether the dynamic mode has changed significantly. The ratio obtained by dividing the cumulative feature offset by the change point determination threshold is the initial mode cumulative change degree, which reflects the relationship between the current cumulative offset degree and the determination threshold.

[0097] In this embodiment of the application, when the initial cumulative change degree of the pattern is greater than the preset cumulative change degree threshold of the pattern, it indicates that the current cumulative change degree of the pattern has reached the upper limit. The cumulative change degree threshold of the pattern is directly used as the cumulative change degree of the pattern to avoid the cumulative change degree of the pattern from exceeding a reasonable range.

[0098] In this embodiment, when the initial cumulative mode change is less than or equal to a preset cumulative mode change threshold, it indicates that the current cumulative mode change is within a reasonable range, and the initial cumulative mode change is directly used as the cumulative mode change. Through the above comparison and selection operations, the cumulative mode change is limited to the range from zero to the cumulative mode change threshold, giving the value a uniform quantitative scale.

[0099] In practical implementation, the cumulative degree of pattern change can be determined in the following ways: To characterize the feature level under normal conditions, and to maintain the feature baseline vector, an exponential sliding recursion is used:

[0100] in, Indicates baseline characteristics, Represents the smoothing coefficient. Indicates dynamic state characteristics, This represents the baseline characteristics at the previous time step. The value can be 0.05. The baseline vector can be initialized to the first sampling point x0, or it can be initialized to the mean of the first 20 sampling points after the vehicle starts, so as to reduce the impact of anomalies in a single initial sampling point on the baseline.

[0101] When the dynamic system state changes, the eigenvector will continuously shift relative to the baseline. (Definition time) Offset:

[0102] in, Indicates time The feature offset is defined as follows; the meanings of the other parameters can be found in the preceding explanation. for Norm, representing the sum of the absolute values ​​of the elements of the vector within parentheses. Used to converge multidimensional offsets into a single scalar sequence.

[0103] Further utilize the concept of cumulative sums to construct change point statistics:

[0104] in, Indicates time The cumulative feature offset, Indicates time The cumulative feature offset of the previous time step. Indicates time The feature offset, This represents the drift compensation constant. A value greater than 0 is used to eliminate the long-term cumulative effect of small random fluctuations on the statistics, and the cumulative feature offset is initialized to 0. max() indicates that the maximum value of the value in parentheses is taken.

[0105] In one embodiment, when the normalized feature vector uses an L1 norm convergence offset... A value of 0.03 can be used to filter out sensor noise and small random fluctuations, so that only continuous offsets will accumulate to form change point statistics.

[0106] To facilitate subsequent fusion and threshold setting, the formula for the cumulative change of the pattern is defined as follows:

[0107] in, Indicates the cumulative degree of change of the pattern. This represents the cumulative feature offset. This indicates the threshold for determining the change point. min() means taking the minimum value within the parentheses. Greater than 0, therefore .

[0108] In one embodiment, A value of 0.15 can be used. When the cumulative feature offset reaches or exceeds 0.15, it is considered that the current dynamic system mode change has reached a definite level.

[0109] This application determines the feature offset based on the difference between dynamic state characteristics and baseline characteristics, and accumulates the feature offset over time to obtain the cumulative feature offset. Then, it determines the cumulative change degree of the mode based on the ratio of the cumulative feature offset to the change point determination threshold. This achieves a quantitative assessment of the severity of dynamic mode changes, solves the technical problem of not being able to accurately identify the intensity of dynamic mode changes, and provides a reliable quantitative basis for the adaptive adjustment of subsequent transition probabilities.

[0110] Optionally, the step of determining the driver's driving intention based on the current vehicle speed, brake pedal opening, and accelerator pedal opening includes: S1041, Obtain the preset acceleration threshold, accelerator pedal opening threshold, and brake intention opening threshold; S1042, determine the current acceleration based on the current vehicle speed; S1043, if the accelerator pedal opening is greater than or equal to the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is greater than or equal to the acceleration threshold, then the driving intention is determined to be an acceleration intention. S1044, If the brake pedal opening is greater than or equal to the braking intention opening threshold, then the driving intention is determined to be the braking intention; S1045, if the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the absolute value of the current acceleration is less than the acceleration threshold, then the driving intention is determined to be cruise intention. S1046 If the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is less than or equal to a negative acceleration threshold, then the driving intention is determined to be a coasting intention.

[0111] In the embodiments of this application, preset acceleration threshold, accelerator pedal opening threshold, and braking intention opening threshold can be obtained.

[0112] Among them, the acceleration threshold is the dividing value used to determine whether the vehicle is in a state of significant acceleration or deceleration. The accelerator pedal opening threshold is the dividing value used to determine whether the accelerator pedal has been effectively depressed, and the brake intention opening threshold is the dividing value used to determine whether the brake pedal has been effectively depressed. All of the above thresholds are fixed parameters that are pre-calibrated and stored in the controller.

[0113] In this embodiment, the current acceleration can be determined based on the current vehicle speed. The current acceleration is the longitudinal acceleration value of the vehicle at the current moment, which is obtained by dividing the difference between the vehicle speed at the current moment and the vehicle speed at the previous moment by the sampling time interval, and is used to characterize the degree of acceleration or deceleration of the vehicle at the current moment.

[0114] In this embodiment, when the accelerator pedal opening is greater than or equal to an accelerator pedal opening threshold, it indicates that the driver has actively pressed the accelerator pedal. When the brake pedal opening is less than a braking intention threshold, it indicates that the driver has not pressed the brake pedal. When the current acceleration is greater than or equal to an acceleration threshold, it indicates that the vehicle is in a state of significant acceleration. When all three conditions are met simultaneously, the driving intention is determined to be an acceleration intention.

[0115] In this embodiment, when the brake pedal opening is greater than or equal to the braking intention opening threshold, it indicates that the driver has pressed the brake pedal and mechanical braking has been initiated. At this time, regardless of the accelerator pedal opening or the current acceleration state of the vehicle, the driving intention is determined as the braking intention, which has the highest priority.

[0116] In this embodiment, when the accelerator pedal opening is less than an accelerator pedal opening threshold, it indicates that the driver has not pressed the accelerator pedal. When the brake pedal opening is less than a braking intention threshold, it indicates that the driver has not pressed the brake pedal. When the absolute value of the current acceleration is less than an acceleration threshold, it indicates that the vehicle speed change is minimal, approaching constant speed. When all three conditions are met simultaneously, the driving intention is determined to be a cruise intention.

[0117] In this embodiment, when the accelerator pedal opening is less than the accelerator pedal opening threshold, it indicates that the driver has not pressed the accelerator pedal. When the brake pedal opening is less than the braking intention opening threshold, it indicates that the driver has not pressed the brake pedal. When the current acceleration is less than or equal to a negative acceleration threshold, it indicates that the vehicle is in a state of significant deceleration and that this deceleration is due to driving resistance rather than mechanical braking. When all three conditions are met simultaneously, the driving intention is determined to be a coasting intention.

[0118] In practical implementation, driving intent can be determined in the following ways: Based on the current vehicle speed and related longitudinal control signals, the driver's driving intention can be inferred to characterize the vehicle's dominant longitudinal behavior at the current moment, which can be used to index the intention dimension of the conditional transfer tensor in subsequent steps.

[0119] First, the current acceleration can be determined:

[0120] in, Indicates the current acceleration. Indicates the current vehicle speed. Indicates the vehicle speed at the previous moment. Indicates the sampling time interval.

[0121] The threshold set for the decision is constructed as follows:

[0122] in, Represents the threshold set, Indicates the acceleration threshold. Indicates the accelerator pedal opening threshold. This indicates the threshold for the braking intention.

[0123] In one embodiment, if the vehicle speed is measured in meters per second and the sampling period is 0.1 seconds, the acceleration threshold can be set to 0.2 meters per second squared, the accelerator pedal opening threshold can be set to 5%, and the braking intention opening threshold can be set to 3%. The acceleration threshold is used to filter out misjudgments caused by small fluctuations in vehicle speed, while the accelerator pedal opening threshold and braking intention opening threshold are used to distinguish whether the driver has made a clear acceleration or braking input.

[0124] The set of hidden states for operating conditions is defined as follows;

[0125] in, The values ​​indicate driving intentions: 1 indicates acceleration intention, 2 indicates cruise intention, 3 indicates coasting intention, and 4 indicates braking intention.

[0126] The driving intention is inferred from the longitudinal acceleration and pedal signal as follows:

[0127] in, The AND operator represents the logical AND statement; the meanings of each parameter can be found in the previous explanation.

[0128] Specifically, driving intention 1 indicates acceleration intention, corresponding to an accelerator pedal opening greater than or equal to an accelerator pedal opening threshold, a brake pedal opening less than a braking intention threshold, and current acceleration greater than or equal to an acceleration threshold. Driving intention 2 indicates cruising intention, corresponding to an accelerator pedal opening less than an accelerator pedal opening threshold, a brake pedal opening less than a braking intention threshold, and current absolute acceleration less than an acceleration threshold. Driving intention 3 indicates coasting intention, corresponding to an accelerator pedal opening less than an accelerator pedal opening threshold, a brake pedal opening less than a braking intention threshold, and current acceleration less than or equal to a negative acceleration threshold. Driving intention 4 indicates braking intention, corresponding to a brake pedal opening greater than or equal to a braking intention threshold; braking intention has the highest priority.

[0129] This application achieves accurate identification of the driver's longitudinal operation intention by acquiring acceleration threshold, accelerator pedal opening threshold, and brake intention opening threshold, determining the current acceleration based on the current vehicle speed, and then determining acceleration intention, braking intention, cruise intention, and coasting intention by comparing the accelerator pedal opening, brake pedal opening, and current acceleration with their respective thresholds. This solves the technical problem that existing methods cannot distinguish between different driving intentions and provides a condition classification basis that matches the driver's actual behavior for vehicle speed prediction.

[0130] Optionally, the step of determining the degree of change of vehicle speed residual based on the current vehicle speed and historical vehicle speed prediction values ​​includes: S1051, the absolute value of the difference between the current vehicle speed and the historical vehicle speed prediction value is used as the current residual; S1052, Perform sliding statistical processing on the current residuals to obtain the standard deviation of the residuals; S1053, determine the initial residual variation based on the ratio of the current residual to the standard deviation of the residual; S1054, If the initial residual variation is greater than the preset residual strength threshold, then the residual strength threshold is used as the vehicle speed residual variation. S1055, if the initial residual variation is less than or equal to the residual strength threshold, then the initial residual variation is taken as the vehicle speed residual variation.

[0131] In this embodiment of the application, the absolute value of the difference between the current vehicle speed and the historical vehicle speed prediction value can be used as the current residual.

[0132] Among them, the historical vehicle speed prediction value is the vehicle speed prediction result for the current moment, calculated in the previous prediction process before the current moment.

[0133] The difference between the current vehicle speed and the historical vehicle speed prediction reflects the magnitude of the deviation of the previous prediction from the actual vehicle speed. Taking the absolute value yields the current residual, which is used to quantify the absolute error of the previous prediction.

[0134] In this embodiment, the standard deviation of the residuals can be obtained by performing sliding statistical processing on the current residuals. Sliding statistical processing involves incorporating the current residuals into the historical residual sequence and calculating the standard deviation of the residuals using an exponential sliding method. The standard deviation of the residuals reflects the fluctuation range of the historical residual sequence and represents the normal fluctuation level of the prediction error.

[0135] In this embodiment, the initial residual variation can be determined based on the ratio of the current residual to the standard deviation of the residual. The larger the ratio of the current residual to the standard deviation of the residual, the more abnormal the current prediction error is relative to the normal fluctuation level; this ratio is the initial residual variation.

[0136] In this embodiment of the application, when the initial residual change is greater than the preset residual strength threshold, it indicates that the current prediction error has reached its upper limit. The residual strength threshold is directly used as the vehicle speed residual change to avoid the vehicle speed residual change from exceeding a reasonable range.

[0137] In this embodiment of the application, when the initial residual change is less than or equal to the preset residual strength threshold, it indicates that the current prediction error is within a reasonable range, and the initial residual change is directly used as the vehicle speed residual change.

[0138] In practical implementation, the degree of change in vehicle speed residual can be determined in the following ways: First, the vehicle is sampled discretely at a sampling period, using discrete time indexing. After obtaining the actual vehicle speed from the vehicle bus speed signal at the current time, a residual closed-loop pre-update is first performed to form a residual statistic reflecting the current prediction error level, providing a priori basis for subsequent steps of online update intensity adjustment and confidence assessment.

[0139] First, we can calculate the current residual:

[0140] in, Indicates the current residual. Indicates the current vehicle speed. This represents historical vehicle speed predictions, which are based on existing errors and do not involve future information. Therefore, they can be used for pre-updates before the current prediction round.

[0141] To balance real-time performance and robustness, an exponential sliding method can be used to estimate the residual mean and residual variance online.

[0142] in, This represents the mean of the residuals at the current time. Represents the residual update coefficient. This represents the mean of the residuals at the previous time step. This represents the current residual.

[0143]

[0144] in, This represents the variance of the residuals at the current time. This represents the variance of the residuals at the previous time step. The meanings of the other parameters can be found in the preceding explanation.

[0145] The residual mean represents the average level estimate of the residuals at the current time, the residual variance represents the estimate of the volatility of the residuals at the current time, and the residual update coefficient ranges from 0 to 1. A larger residual update coefficient indicates a faster response to the latest residuals, while a smaller value indicates a stronger memory of historical residuals. In one embodiment, the residual update coefficient ranges from 0.01 to 0.2.

[0146] The residual magnitudes and standard deviations are as follows:

[0147] in, This represents the current absolute value of the residual. Indicates the current residual. This represents the standard deviation of the residuals.

[0148]

[0149] in, Indicates the degree of change in vehicle speed residual. This represents the proportionality coefficient. This represents a constant to prevent the denominator from being zero. The meanings of the other parameters can be found in the preceding explanation.

[0150] The absolute value of the current residual is the absolute value of the prediction residual for the current step. The proportionality coefficient is greater than zero, and the zero-prevention constant is a constant greater than zero. The rate of change of the vehicle speed residual ranges from 0 to 1, and is used to characterize the strength of the current residual relative to the level of fluctuation.

[0151] In one embodiment, the scaling factor can be 2.0, indicating that when the one-step prediction residual reaches twice the sliding standard deviation of the residual, the variability of the vehicle speed residual is close to saturation. (Zero-constant protection) The setting can be adjusted according to the vehicle speed unit. For example, when the vehicle speed is calculated in meters per second, use 0.02 meters per second; when the vehicle speed is calculated in kilometers per hour, use... A value of 0.1 km / h can be used to avoid numerical instability caused by the residual standard deviation approaching zero. The above value is only an illustrative example; the actual value can be calibrated according to the resolution and sampling period of the vehicle speed sensor.

[0152] This application uses the absolute value of the difference between the current vehicle speed and the historical vehicle speed prediction value as the current residual, and performs sliding statistical processing on the current residual to obtain the residual standard deviation. Then, it determines the degree of change of the vehicle speed residual based on the ratio of the current residual to the residual standard deviation, thereby realizing a quantitative assessment of the degree of anomaly in the prediction error. This solves the technical problem that existing methods cannot effectively quantify the prediction error and provides a reliable error feedback basis for the adaptive adjustment of the subsequent state transition probability.

[0153] Optionally, the set of state transition probabilities is constructed in the following manner: S1061, Obtain the historical vehicle speed sequence, which includes the historical vehicle speed corresponding to different historical moments; S1062, obtain the vehicle's dynamic state parameters at each historical moment; S1063, for any historical moment, determine the vehicle's power mode and the driver's driving intention based on the power state parameters and historical vehicle speed corresponding to the historical moment. S1064, Determine the speed increment sequence based on the historical vehicle speed sequence; the speed increment sequence includes speed increments corresponding to different historical moments; S1065, obtain several preset discrete increment intervals; wherein, each discrete increment interval corresponds to a discrete increment state; S1066, determine the discrete increment intervals to which the velocity increments at different historical moments belong in the velocity increment sequence, and combine the discrete increment states corresponding to the discrete increment intervals in the order of historical moments to obtain the discrete increment state sequence. S1067, associate and save the discrete incremental state corresponding to any historical moment in the discrete incremental state sequence with the vehicle's power mode and the driver's driving intention at the corresponding historical moment. S1068, take any power mode and any driving intention as a set of working condition combinations, find the discrete incremental state that matches the working condition combination from the discrete incremental state sequence, and combine them in the order of historical time to obtain the discrete incremental state subsequence corresponding to the working condition combination. S1069, For any combination of operating conditions, perform a first-order Markov property test on the discrete incremental state subsequence corresponding to the combination of operating conditions. S1070, If the test passes, then determine the subset of state transition probabilities under the working condition combination based on the discrete incremental state subsequence; S1071, combine the subsets of state transition probabilities under each combination of working conditions to obtain the set of state transition probabilities.

[0154] In this embodiment of the application, a historical vehicle speed sequence can be obtained. The historical vehicle speed sequence is a series of vehicle speed values ​​recorded during the historical operation of the vehicle. Each vehicle speed in the sequence corresponds to a historical moment and is used to describe the speed change of the vehicle at different times.

[0155] In this embodiment, the vehicle's power state parameters at each historical moment can be obtained. These power state parameters are a set of physical quantities describing the operating state of the vehicle's powertrain, including motor output torque, brake pedal opening, and accelerator pedal opening, used to subsequently determine the power mode and driving intention at each historical moment.

[0156] In this embodiment of the application, for any historical moment, the power mode of the vehicle and the driver's driving intention can be determined based on the power state parameters and historical vehicle speed corresponding to that historical moment.

[0157] Among these, the power mode is the dominant operating type of the vehicle's powertrain system, including drive mode, combined braking mode, and braking mode. The driver's driving intention is the driver's subjective expectation of the vehicle's longitudinal movement, including acceleration intention, cruising intention, coasting intention, and braking intention.

[0158] In practical implementation, for any given historical moment, the power mode can be determined based on the motor output torque and brake pedal opening at that moment. When the brake pedal opening is small and the motor output torque is positive, the power mode for that historical moment is determined to be the drive mode. When the brake pedal opening is small and the motor output torque is negative, the power mode for that historical moment is determined to be the combined braking mode. When the brake pedal opening is large, the power mode for that historical moment is determined to be the braking mode.

[0159] For any given historical moment, the current acceleration at that historical moment is determined based on the current vehicle speed. Then, by combining the accelerator pedal opening, brake pedal opening, and current acceleration at that historical moment with their respective preset thresholds, the driver's driving intention at that historical moment is determined.

[0160] When the accelerator pedal opening is greater than or equal to the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is greater than or equal to the acceleration threshold, the driving intention at that historical moment is determined to be the acceleration intention.

[0161] When the brake pedal opening is greater than or equal to the braking intention opening threshold, the driving intention at that historical moment is determined as the braking intention.

[0162] When the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the absolute value of the current acceleration is less than the acceleration threshold, the driving intention at that historical moment is determined to be the cruise intention.

[0163] When the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is less than or equal to the negative acceleration threshold, the driving intention at that historical moment is determined to be a coasting intention.

[0164] Since the historical moment and the current moment use the same judgment logic, only the data processed comes from historical records rather than real-time collection, this application can also accurately identify the power mode and driving intention at each moment using historical data, providing a reliable data foundation for subsequent grouping and statistics by working condition combination.

[0165] In this embodiment of the application, a speed increment sequence can be determined based on a historical vehicle speed sequence. The speed increment sequence is a sequence composed of the differences between vehicle speeds at adjacent historical moments. Each speed increment in this sequence corresponds to a historical moment and is used to describe the change in vehicle speed between different moments.

[0166] In the embodiments of this application, several preset discrete increment intervals can be obtained.

[0167] The discrete increment interval divides the range of velocity increment values ​​into multiple continuous sub-intervals. Each sub-interval has a clear upper and lower boundary. All sub-intervals together cover the complete range of velocity increment values. Each discrete increment interval corresponds to a discrete increment state, which is used to map continuous velocity increment values ​​to discrete state numbers.

[0168] In this embodiment, the discrete increment interval to which each speed increment belongs in the speed increment sequence can be determined. Each speed increment is divided into a corresponding discrete increment interval according to its numerical value, and the discrete increment state corresponding to that interval is the state to which that speed increment belongs. All discrete increment states corresponding to all speed increments are arranged in chronological order according to their corresponding historical moments, and combined to obtain a discrete increment state sequence. The discrete increment state sequence is used to describe the trajectory of the vehicle speed increment in the discrete state space.

[0169] In this embodiment of the application, the discrete incremental state corresponding to any historical moment in the discrete incremental state sequence can be associated with and saved with the vehicle's power mode and the driver's driving intention at that historical moment.

[0170] Each discrete incremental state in the discrete incremental state sequence corresponds to a historical moment, which also corresponds to a power mode and a driving intention. Associating these three elements according to the same historical moment allows each discrete incremental state to trace its associated power mode and driving intention, which is then used for subsequent grouping and statistical analysis based on operating conditions.

[0171] In this embodiment, any power mode and any driving intention can be considered as a set of operating conditions. Discrete incremental states matching the current operating condition combination are searched from the associated and saved data; that is, all discrete incremental states that simultaneously satisfy the power mode and the driving intention are searched. The found discrete incremental states are arranged in chronological order according to historical time, and combined to obtain the discrete incremental state subsequence corresponding to the operating condition combination. Each set of operating conditions corresponds to a discrete incremental state subsequence, used to statistically analyze the state transition patterns under different operating conditions.

[0172] In the embodiments of this application, a first-order Markov property test can be performed on the discrete incremental state subsequence corresponding to any combination of operating conditions.

[0173] The first-order Markov property test is used to verify whether a discrete incremental state sequence satisfies the first-order Markov property, that is, whether the state at the current moment depends only on the state at the previous moment. It is used to confirm whether a first-order Markov model can be used to describe the state transition law under this working condition.

[0174] In this embodiment of the application, if the verification passes, the state transition probability subset under the working condition combination is determined based on the discrete incremental state subsequence. The state transition probability subset is the set of transition probabilities from all source discrete incremental states to the target discrete incremental state under this working condition combination, used to describe the transition law of the discrete incremental states under this working condition.

[0175] In this embodiment, the state transition probability subsets under each operating condition combination can be combined to obtain a state transition probability set. The state transition probability set contains the state transition probability subsets under all operating condition combinations, and this set is used in subsequent steps to determine the initial state transition probabilities corresponding to the current power mode and driving intention index.

[0176] In practical implementation, the state transition probability set can be constructed in the following ways: First, a velocity increment sequence can be constructed:

[0177] in, Indicates the speed increment. Indicates the current vehicle speed. This indicates the vehicle speed at the previous moment.

[0178] Then, the discrete bin boundaries can be defined. Let the number of discrete states for the velocity increment be K, and define the lower and upper bounds of the global bins as follows: and and the interval Divide the space into K equal boxes.

[0179]

[0180] Where w represents the width of the discrete increment interval. Indicates the upper bound of the velocity increment. This represents the lower bound of the velocity increment, and K represents the number of discrete increment states.

[0181] Calculate the boundary of the discrete increment interval:

[0182] in, This represents the boundary of the j-th discrete increment interval. Let w represent the lower bound of the velocity increment, w represent the width of the discrete increment interval, and K represent the number of discrete increment states. The boundary of the first discrete increment interval is equal to the lower bound of the velocity increment, and the boundary of the Kth discrete increment interval is equal to the upper bound of the velocity increment.

[0183] Discrete states can be defined According to the following rules Mapped to status number

[0184] in, Represents discrete incremental states. Indicates the speed increment. This represents the boundary of the first discrete increment interval. Indicates the j-th One discrete increment interval boundary, This represents the boundary of the j-th discrete increment interval. Indicates the Kth One discrete increment interval boundary, where K represents the number of discrete increment states.

[0185] When the velocity increment is less than the boundary of the first discrete increment interval, the discrete increment state is 1. When the velocity increment is greater than or equal to the j-th interval... When the discrete increment interval boundary is less than the j-th discrete increment interval boundary, the discrete increment state is j. When the velocity increment is greater than or equal to the k-th discrete increment interval boundary... When there is one discrete increment interval boundary, the discrete increment state is K.

[0186] In one embodiment, the number of discrete incremental states can be 5, and the velocity increment range can be defined as a lower bound of -1 km / h and an upper bound of 1 km / h. The width of the discrete increment interval is then 0.4 km / h, and the boundaries of the discrete increment interval are successively -1 km / h, -0.6 km / h, -0.2 km / h, 0.2 km / h, 0.6 km / h, and 1 km / h. If the continuous velocity increment sequence is 0.3 km / h, 0.4 km / h, -0.1 km / h, and -0.5 km / h, it can be mapped to the discrete incremental state sequence 4, 4, 3, and 2.

[0187] This mapping transforms continuous vehicle speed variation curves into finite state sequences suitable for real-time statistics and recursion by the onboard controller.

[0188] Then the conditional sample set and the first-order transition count can be determined. For a given condition... To perform a statistical transition, first define the conditional indicator:

[0189] Among them, conditional indication quantity This indicates that the value is 1 when the current power mode is m and the driving intention is z, and 0 in other cases.

[0190] Under conditions The following is a statistical count of first-order transitions:

[0191] Among them, the number of state transitions This represents the total number of transitions from discrete incremental state i to discrete incremental state j under the conditions of power mode m and driving intention z. The total sample length T represents the total number of sampling points in the discrete incremental state sequence. Indicator function. This indicates that the value is 1 when the discrete incremental state at time k equals i, and 0 otherwise. Indicator function This indicates that the value is 1 when the discrete incremental state at time k+1 is equal to j, and 0 otherwise.

[0192] Conditional indication quantity This indicates that the value is 1 when the current power mode is m and the driving intention is z, and 0 in other cases.

[0193] This formula counts all adjacent time pairs that satisfy the following conditions: current power mode is m, current driving intention is z, current discrete increment state is i, and next discrete increment state is j, to obtain the number of state transitions under the corresponding working condition combination.

[0194] Define state The number of times the condition appears is as follows:

[0195] in, Let K represent the total number of times discrete incremental state i occurs under the conditions of power mode m and driving intention z. K represents the number of discrete incremental states. This represents the number of state transitions from discrete incremental state i to discrete incremental state j under the conditions of power mode m and driving intention z. This formula sums the number of transitions from discrete incremental state i to all target discrete incremental states j under the same power mode and driving intention, yielding the total number of times discrete incremental state i occurs under this combination of conditions.

[0196] Therefore, the maximum likelihood estimate of the conditional first-order transition probability is obtained as follows:

[0197] in, This represents the first-order transition probability estimate for transitioning from discrete incremental state i to discrete incremental state j under the conditions of power mode m and driving intention z.

[0198] This formula divides the number of state transitions by the total number of times the state occurs, yielding a probability estimate of the transition from the current discrete incremental state to the next discrete incremental state under the current operating condition combination. The meanings of the remaining parameters can be found in the preceding explanation.

[0199] To test the first-order Markov property, under the condition The following is a statistical count of second-order ternary transitions:

[0200] in, This represents the total number of ternary transitions from discrete incremental state p to discrete incremental state i and then to discrete incremental state j, under the conditions of power mode m and driving intention z. The total sample length T represents the total number of sampling points in the discrete incremental state sequence. Conditional indication quantity. This indicates that the value is 1 when the current power mode is m and the driving intention is z, and 0 otherwise. Indicator function This indicates that the value is 1 when the discrete incremental state at time k-1 equals p, and 0 otherwise. Indicator function This indicates that the value is 1 when the discrete incremental state at time k equals i, and 0 otherwise. Indicator function This formula represents a value of 1 when the discrete incremental state at time k+1 equals j, and a value of 0 otherwise. It counts the number of ternary transitions occurring under the corresponding working condition combination, where the state is p to i in the past two steps, the current state is i, and the next state is j.

[0201] And define binary prefix counting:

[0202] in, Let K represent the total number of occurrences of the preceding state pair from p to i, given a power mode of m and a driving intention of z. K represents the number of discrete incremental states. This represents the total number of ternary transitions from discrete incremental state p to discrete incremental state i and then to discrete incremental state j, under the conditions of power mode m and driving intention z. This formula sums the ternary transition counts corresponding to all target states j to obtain the total number of occurrences of the preceding state pair from p to i under this operating condition combination.

[0203] This yields the maximum likelihood estimate of the conditional second-order transition probability:

[0204] in, This represents the second-order transition probability estimate for the transition from discrete incremental state p to discrete incremental state i and then to discrete incremental state j, under the conditions of power mode m and driving intention z. This represents the total number of ternary transitions from discrete incremental state p to discrete incremental state i and then to discrete incremental state j, under the conditions of power mode m and driving intention z. This represents the total number of occurrences of the preceding state pair from p to i under the conditions of power mode m and driving intention z.

[0205] This formula divides the number of ternary transitions by the total number of occurrences of previous state pairs to obtain the probability estimate of transitioning to state j in the next step when the previous two steps were from state p to state i under the current working condition combination.

[0206] Construct the first-order Markov likelihood ratio test statistic:

[0207] in, This represents the first-order Markov likelihood ratio test statistic, where K represents the number of discrete increment states. This represents the total number of ternary transitions from discrete incremental state p to discrete incremental state i and then to discrete incremental state j, under the conditions of power mode m and driving intention z. This represents the second-order transition probability estimate for the transition from discrete incremental state p to discrete incremental state i and then to discrete incremental state j under the conditions of power mode m and driving intention z. Let ln(m) represent the first-order transition probability estimate for transitioning from discrete incremental state i to discrete incremental state j under the conditions of power mode m and driving intention z. ln() is the natural logarithm. When the ternary transition number is zero, the corresponding summation term takes the value of zero.

[0208] This formula calculates the logarithm of the ratio of the second-order transition probability to the first-order transition probability, then sums the results using the number of ternary transitions and multiplies by 2 to obtain the test statistic used to determine whether the first-order Markov property holds. When there is no significant difference between the second-order and first-order models, the statistic takes a smaller value; when the second-order model is significantly better than the first-order model, the statistic takes a larger value.

[0209] Degrees of freedom are calculated using the following formula:

[0210] Wherein, the degree of freedom ν is equal to the number of discrete incremental states multiplied by the square of the number of discrete incremental states minus one.

[0211]

[0212] in, This indicates the conclusion of the first-order Markov property test. This represents the first-order Markov likelihood ratio test statistic. This represents a chi-square distribution with ν degrees of freedom at a confidence level. The quantiles below. The significance level is defined as follows: When the first-order Markov property likelihood ratio test statistic is less than or equal to the chi-square distribution quantile, the first-order Markov property indicator value is 1, indicating that the first-order Markov property is satisfied under the corresponding working condition combination. When the first-order Markov property likelihood ratio test statistic is greater than the chi-square distribution quantile, the first-order Markov property indicator value is 0, indicating that the first-order Markov property is not satisfied under the corresponding working condition combination.

[0213] This application determines the power mode and driving intention at each historical moment by acquiring historical vehicle speed sequences and power state parameters at each historical moment. Based on the historical vehicle speed sequences, a speed increment sequence is determined, and the speed increments are mapped to discrete increment intervals to obtain a discrete increment state sequence. The discrete increment states at each historical moment are then associated and saved with the corresponding power mode and driving intention. Matching discrete increment states are searched according to the working condition combination to obtain the discrete increment state subsequence corresponding to each working condition combination. A first-order Markov property test is performed on the discrete increment state subsequences corresponding to each working condition combination. Based on the discrete increment state subsequences that pass the test, a subset of state transition probabilities under each working condition combination is determined and combined into a state transition probability set. This solves the technical problem of how to construct a state transition probability set covering different working conditions, providing a complete statistical probability foundation for subsequent vehicle speed prediction.

[0214] Optionally, the step of determining the subset of state transition probabilities under the working condition combination based on the discrete incremental state subsequence includes: S1081, In the discrete incremental state subsequence corresponding to the working condition combination, for each adjacent time pair, the discrete incremental state of the previous time is taken as the source discrete incremental state, and the discrete incremental state of the next time is taken as the target discrete incremental state. S1082, for any source discrete incremental state, determine the number of state transitions from the discrete incremental state subsequence to the next time step as the target discrete incremental state. S1083, Based on the number of state transitions, determine the transition probability of the discrete incremental state from the source discrete incremental state to the target discrete incremental state under the working condition combination. S1084 combines the transition probabilities from each source discrete incremental state to each target discrete incremental state under the working condition combination to obtain the state transition probability subset corresponding to the working condition combination.

[0215] In the embodiments of this application, the source state and target state can be divided for each adjacent time point in the discrete incremental state subsequence corresponding to the working condition combination.

[0216] A discrete incremental state subsequence is the portion of a discrete incremental state sequence that belongs to the same working condition combination. Each adjacent time pair contains two discrete incremental states: the previous time and the next time. The discrete incremental state of the previous time is taken as the source discrete incremental state, and the discrete incremental state of the next time is taken as the target discrete incremental state, which is used to describe the directionality of the state transition.

[0217] In the embodiments of this application, for any source discrete incremental state, the number of state transitions can be counted from the discrete incremental state subsequence corresponding to the combination of operating conditions.

[0218] In the specific implementation, for any source discrete incremental state, the number of all adjacent time pairs in the current working condition combination discrete incremental state subsequence where the source discrete incremental state appears and the next time step is a certain target discrete incremental state is counted to obtain the number of state transitions.

[0219] In this embodiment, the transition probability under a given operating condition combination can be determined based on the number of state transitions. The transition probability from the source discrete incremental state to the target discrete incremental state is obtained by dividing the number of state transitions from a source discrete incremental state to a target discrete incremental state under the given operating condition combination by the total number of times the source discrete incremental state appears in the current operating condition combination.

[0220] In this embodiment, the transition probabilities from each source discrete incremental state to each target discrete incremental state under the given operating condition combination can be combined. The transition probabilities from all source discrete incremental states to all target discrete incremental states together constitute the state transition probability subset corresponding to the operating condition combination, which is used to describe the complete transition law of discrete incremental states under the given power mode and driving intention conditions.

[0221] In practical implementation, the subset of state transition probabilities corresponding to the current working condition combination can be obtained in the following way: First, define the elements in the conditional transition counting tensor. , which represents the weighted count of the transition from the source discrete incremental state to the target discrete incremental state under the corresponding working condition combination up to the current time. , , K represents the number of discrete states; M represents the number of modes; and Z represents the number of hidden states under a given condition. The initialization time is set to... .

[0222] To accelerate the model's response to new data when residuals increase or modes switch, a forgetting factor is defined as follows:

[0223] in, , where is the forgetting factor at time k. This represents the upper limit of the forgetting factor. This represents the lower limit of the forgetting factor. Indicates the baseline forgetting value. Indicates the residual strength weight. Indicates the intensity weight at the variable point. Indicates the degree of change in vehicle speed residual. Indicates the cumulative degree of change in the pattern.

[0224] In one embodiment, the lower limit of the forgetting factor can be 0.01, the upper limit can be 0.20, the baseline forgetting value can be 0.03, the residual strength weight can be 0.10, and the change point strength weight can be 0.07. When both the vehicle speed residual variability and the pattern cumulative variability are low, the forgetting factor is close to the baseline forgetting value of 0.03, and the model mainly retains historical statistical information. When the vehicle speed residual variability or the pattern cumulative variability increases, the forgetting factor increases accordingly, allowing the conditional transition count to respond to new data more quickly. The above values ​​are example values ​​and can be set according to actual conditions.

[0225] At any moment slice of current condition Execution index forgotten, add one to update:

[0226] in, It represents the weighted count from the source discrete incremental state i to the target discrete incremental state j under the current working condition combination at the current time. This represents the forgetting factor. This represents the weighted count from the source discrete incremental state i to the target discrete incremental state j under the current operating condition combination at the previous time step. Indicator function This indicates that the value is 1 when the discrete increment state at the previous time step equals i, and 0 otherwise. Indicator function. The value is 1 when the discrete increment state at the current time is equal to j, and 0 otherwise. This formula performs an exponential forgetting update on the weighted count of the current working condition combination, multiplying the old count by a minus forgetting factor, and adding the currently observed state transition indicator value.

[0227] Slices not based on the current condition remain unchanged: ,in .

[0228] To avoid the zero probability problem caused by small samples, a smoothing constant is introduced to calculate the conditional transition probability:

[0229] in, This represents the estimated conditional transition probability of transitioning from the source discrete incremental state i to the target discrete incremental state j under the corresponding working condition combination at the current moment. It represents the weighted count from the source discrete incremental state i to the target discrete incremental state j under the corresponding working condition combination at the current time. This represents the smoothing constant, used to avoid the zero-probability problem caused by small samples. K represents the number of discrete incremental states. In the denominator... It represents the weighted sum of counts from the source discrete incremental state i to all target discrete incremental states under the corresponding working condition combination at the current moment.

[0230] Stacking the transition probabilities of all conditional slices along the conditional dimension yields the conditional transition tensor. as follows:

[0231] in, The conditional transition tensor represents the conditional transition probabilities corresponding to each combination of operating conditions, each source discrete incremental state, and each target discrete incremental state. Combination of working conditions The three-dimensional data structure obtained by stacking the dimension, source state dimension, and target state dimension is used to combine the corresponding transition probability slices according to the current working condition during subsequent multi-step prediction.

[0232] Reference Figure 2This diagram illustrates a conditional transition tensor according to an embodiment of this application. The conditional transition tensor is composed of stacked state transition matrices corresponding to multiple driving condition combinations. Each driving condition combination corresponds to a K-row, K-column transition matrix. The rows of the matrix represent the current discrete incremental state, the columns represent the next discrete incremental state, and the values ​​in the cells represent the transition probabilities from the current discrete incremental state to the next discrete incremental state. After determining the current driving condition combination based on the current power mode and current driving intention, a slice of the transition matrix corresponding to that driving condition combination is selected from the conditional transition tensor by index. The selected slice is the set of transition probabilities from all source discrete incremental states to all target discrete incremental states under the current driving condition combination, used for recursive calculation of the state distribution in subsequent multi-step predictions.

[0233] Slice of conditional transition probabilities for each combination of operating conditions It is a K-row, K-column transition matrix.

[0234] A conditional transition probability slice represents a two-dimensional data block consisting of the transition probabilities from all source discrete incremental states to all target discrete incremental states under a specific set of operating conditions. In a K-row, K-column transition matrix, each row corresponds to a source discrete incremental state, and each column corresponds to a target discrete incremental state. The value in the i-th row and j-th column of the matrix represents the probability of transitioning from source discrete incremental state i to target discrete incremental state j.

[0235] This transition matrix is ​​used to characterize the transition law of discrete incremental states under this combination of operating conditions. The transition law of discrete incremental states refers to the statistical pattern of the change of vehicle speed increment between different discrete intervals. This pattern reflects the typical characteristics of vehicle speed change under specific power modes and specific driving intentions.

[0236] The transition probabilities from each source discrete incremental state to each target discrete incremental state under the same working condition combination are combined to obtain the state transition probability subset corresponding to the current working condition combination.

[0237] This application achieves complete statistical analysis of the discrete incremental state transition patterns under various working condition combinations by determining the source and target discrete incremental states for each adjacent time step in the discrete incremental state subsequence, counting the number of state transitions for any source discrete incremental state under that working condition combination, determining the transition probability based on the number of state transitions, and finally combining the transition probabilities from all source states to the target state under that working condition combination into a subset of state transition probabilities corresponding to that working condition combination. This provides a probabilistic data foundation that matches the working conditions for subsequent vehicle speed prediction.

[0238] Optionally, the step of obtaining the initial state transition probabilities corresponding to the power mode and driving intention from a pre-built set of state transition probabilities includes: S1091, determine the current operating condition combination based on the power mode and driving intention; S1092, Select all transition probabilities from any source discrete incremental state to any target discrete incremental state under the current working condition combination from the set of state transition probabilities, and use them as the initial state transition probabilities.

[0239] In this embodiment, the current operating condition combination can be determined based on the power mode and driving intention. The current power mode and driving intention are combined to obtain the current operating condition combination, which is used to index the corresponding transition probability data from the state transition probability set later.

[0240] In this embodiment, all transition probabilities under the current operating condition combination can be selected from the state transition probability set as initial state transition probabilities. The state transition probability set contains the transition probabilities from all source discrete incremental states to all target discrete incremental states under all operating condition combinations. Data corresponding to the current operating condition combination is indexed from this set; this data contains all transition probabilities from any source discrete incremental state to any target discrete incremental state under the current operating condition combination. This data is used as the initial state transition probability for subsequent adaptive adjustment steps.

[0241] This application determines the current operating condition combination based on the power mode and driving intention, and selects the transition probabilities from all source discrete incremental states to the target discrete incremental state under the current operating condition combination from the state transition probability set as the initial state transition probabilities. This achieves accurate indexing of the transition probability data matching the current operating condition, solves the technical problem of how to quickly obtain the corresponding transition probability data of the current operating condition from a pre-constructed set, and provides a data foundation for the adaptive adjustment of the subsequent transition probabilities.

[0242] Optionally, the step of adjusting the initial state transition probability based on the cumulative change degree of the mode and the change degree of the vehicle speed residual to obtain the target state transition probability includes: S1101, obtain the preset lower limit of forgetting factor, upper limit of forgetting factor, baseline forgetting value, residual strength weight and change point strength weight; S1102, calculate the first product of the residual strength weight and the vehicle speed residual variability; S1103, calculate the second product of the change point intensity weight and the cumulative change of the mode; S1104, add the baseline forgetting value, the first product value, and the second product value to obtain the initial forgetting factor value; S1105, If the initial forgetting factor value is less than the lower limit of the forgetting factor, then the lower limit of the forgetting factor is used as the forgetting factor value; S1106, If the initial forgetting factor value is greater than the upper limit of the forgetting factor, then the upper limit of the forgetting factor is taken as the target forgetting factor value; S1107, If the initial forgetting factor value is greater than the lower limit of the forgetting factor and less than the upper limit of the forgetting factor, then the initial forgetting factor value shall be used as the target forgetting factor value. S1108, based on the target forgetting factor value, the initial state transition probability is updated exponentially weighted to obtain the target state transition probability.

[0243] In this embodiment, preset lower limit, upper limit, baseline forgetting value, residual strength weight, and change point strength weight can be obtained. The lower limit is the minimum possible value of the forgetting factor, the upper limit is the maximum possible value, the baseline forgetting value is the baseline level of the forgetting factor under normal conditions, the residual strength weight controls the influence of the vehicle speed residual change on the forgetting factor, and the change point strength weight controls the influence of the cumulative change in the mode on the forgetting factor. All of the above parameters are pre-calibrated and stored as fixed values ​​in the controller.

[0244] In this embodiment, the product of the residual strength weight and the vehicle speed residual variability can be calculated as the first product value. This product value reflects the contribution of the vehicle speed residual variability to the forgetting factor. The greater the vehicle speed residual variability, the larger the first product value, and the greater the forgetting factor.

[0245] In this embodiment, the product of the change point intensity weight and the cumulative change of the pattern can be calculated as the second product value. This product value reflects the contribution of the cumulative change of the pattern to the forgetting factor. The greater the cumulative change of the pattern, the greater the second product value, and the greater the forgetting factor.

[0246] In this embodiment, the initial forgetting factor value can be obtained by adding the baseline forgetting value, the first product value, and the second product value. In a specific implementation, the initial forgetting factor value can be... .

[0247] The baseline forgetting value provides the basic level of the forgetting factor. The first and second product values ​​are respectively increased on the baseline value by increments proportional to the change in vehicle speed residual and the change in pattern cumulative variation, so that the initial forgetting factor value can increase with the increase of the two changes.

[0248] In this embodiment of the application, when the initial forgetting factor value is less than the lower limit of the forgetting factor, it indicates that the calculated initial value is lower than the minimum allowable value. The lower limit of the forgetting factor is directly used as the target forgetting factor value to ensure that the forgetting factor is not too small.

[0249] In the example provided, when the initial forgetting factor value is greater than the upper limit of the forgetting factor, it indicates that the calculated initial value is higher than the maximum allowed value. The upper limit of the forgetting factor is then directly used as the target forgetting factor value to ensure that the forgetting factor is not too large.

[0250] In this embodiment of the application, when the initial forgetting factor value is greater than the lower limit of the forgetting factor and less than the upper limit of the forgetting factor, it indicates that the calculated initial value is within the allowable range, and the initial forgetting factor value is directly used as the target forgetting factor value.

[0251] In this embodiment, the target state transition probability can be obtained by exponentially weighting the initial state transition probability based on the target forgetting factor value. Exponential weighting update refers to using the target forgetting factor to perform a weighted summation of historical weighted counts and the currently observed state transitions. The result of multiplying the old count by one minus the target forgetting factor, plus the result of multiplying the current observation indicator value by the target forgetting factor, yields a new weighted count. The transition probability is then recalculated based on this new weighted count. A larger target forgetting factor value results in faster decay of historical data weights, leading to a more rapid model response to current data. A smaller target forgetting factor value results in slower decay of historical data weights, leading to a more stable model.

[0252] This application obtains the lower limit, upper limit, baseline forgetting value, residual strength weight, and change point strength weight of the forgetting factor. It calculates the first product of the residual strength weight and the degree of change of the vehicle speed residual, and the second product of the change point strength weight and the cumulative degree of change of the mode. The baseline forgetting value is added to the two products, and then the target forgetting factor value is obtained by limiting it to the range of the lower and upper limits of the forgetting factor. The initial state transition probability is then updated exponentially based on the target forgetting factor value. This achieves adaptive following of the state transition probability to changes in driving mode and operating condition, and solves the technical problem of increased prediction error caused by response lag in existing prediction models when switching between multiple modes. This enables the model to respond quickly when residuals or modes change, and maintain historical statistical regularity when stable.

[0253] Optionally, the step of determining the predicted future vehicle speed at each prediction time point within the prediction time domain based on the current vehicle speed and the target state transition probability includes: S1111, determine the current speed increment based on the difference between the current speed and the previous speed; S1112, Determine the original vehicle speed prediction value at each prediction time based on the current speed increment and the target state transition probability; S1113, obtain the upper limit and lower limit of vehicle speed at each prediction time within the prediction time domain; S1114, For any prediction time, if the original vehicle speed prediction value is greater than or equal to the vehicle speed limit value, then the vehicle speed limit value is determined as the future vehicle speed prediction value at the prediction time. S1115, If the original predicted vehicle speed is less than or equal to the lower limit of vehicle speed, then the lower limit of vehicle speed shall be determined as the predicted future vehicle speed at the time of prediction. S1116 If the original vehicle speed prediction value is greater than the lower limit of vehicle speed and less than the upper limit of vehicle speed, then the original vehicle speed prediction value is determined as the future vehicle speed prediction value at the prediction time.

[0254] In this embodiment, the current speed increment can be determined based on the difference between the current vehicle speed and the previous vehicle speed. The current vehicle speed is the actual speed of the vehicle relative to the ground at the current moment, and the previous vehicle speed is the actual speed of the vehicle at the previous sampling moment. Subtracting the previous vehicle speed from the current vehicle speed yields the current speed increment, which describes the change in the vehicle's speed at the current moment.

[0255] In this embodiment, the original vehicle speed prediction value at each prediction time can be determined based on the current speed increment and the target state transition probability. The target state transition probability is adaptively adjusted transition probability data, which describes the transition law of discrete incremental states under the current operating condition combination.

[0256] Starting with the discrete incremental state corresponding to the current speed increment, the probability distribution of the discrete incremental state at each prediction time is obtained by recursively using the target state transition probability. Then, the expected value of the speed increment at each prediction time is calculated based on the probability distribution of the discrete incremental state. The expected value of the speed increment is accumulated step by step with the current vehicle speed as the starting value to obtain the original vehicle speed prediction value at each prediction time.

[0257] In this embodiment of the application, the upper limit and lower limit of vehicle speed at each prediction time within the prediction time domain can be obtained. The upper limit and lower limit of vehicle speed constitute the speed reachable interval at each prediction time within the prediction time domain. This interval is determined by the physical constraints of the vehicle's power system and is used for subsequent physical feasibility verification of the original vehicle speed prediction value.

[0258] In this embodiment, the relationship between the original predicted vehicle speed and the upper limit of vehicle speed can be determined for any prediction time. When the original predicted vehicle speed is greater than or equal to the upper limit of vehicle speed, it indicates that the original prediction result exceeds the maximum speed that the vehicle's physical capabilities can achieve. The upper limit of vehicle speed is then directly used as the future predicted vehicle speed for that prediction time to ensure that the prediction result does not exceed the actual capabilities of the vehicle.

[0259] In this embodiment of the application, when the original predicted vehicle speed is less than or equal to the lower limit of vehicle speed, it indicates that the original predicted result is lower than the minimum speed that the vehicle's physical capabilities can achieve. The lower limit of vehicle speed is directly used as the future predicted vehicle speed at that prediction time to ensure that the prediction result is not lower than the actual capability range of the vehicle.

[0260] In this embodiment of the application, when the original vehicle speed prediction value is greater than the lower limit of vehicle speed and less than the upper limit of vehicle speed, it indicates that the original prediction result is within the speed range allowed by the vehicle's physical capabilities. The original vehicle speed prediction value is directly used as the future vehicle speed prediction value at that prediction time without correction.

[0261] This application determines the current speed increment based on the difference between the current vehicle speed and the previous vehicle speed, and determines the original vehicle speed prediction value based on the current speed increment and the target state transition probability. Then, it obtains the upper and lower limits of the vehicle speed to verify the physical feasibility of the original vehicle speed prediction value. This solves the technical problem that existing prediction methods do not consider the physical constraints of the vehicle's power system, which leads to prediction results exceeding the actual capabilities of the vehicle. It ensures that the predicted vehicle speed sequence is always within the vehicle's physical reach, thus improving the reliability and engineering usability of vehicle speed prediction.

[0262] Optionally, the step of determining the original vehicle speed prediction value at each prediction time based on the current speed increment and the target state transition probability includes: S1121, Determine the discrete increment interval corresponding to the current speed based on the current speed increment, and obtain the current discrete increment state corresponding to the current speed; S1122, Based on the current discrete incremental state and the target state transition probability, the discrete incremental state probability distribution data at each prediction time in the prediction time domain is recursively obtained; S1123, Calculate the expected value of velocity increment at each prediction time based on the discrete incremental state probability distribution data at each prediction time; S1124: Based on the current vehicle speed and the expected speed increment at each prediction time, the original predicted vehicle speed value at each prediction time is accumulated.

[0263] In this embodiment, the discrete increment interval corresponding to the current speed can be determined based on the current speed increment. The speed increment is the difference between the current vehicle speed and the vehicle speed at the previous moment.

[0264] Discrete increment intervals divide the range of velocity increment values ​​into multiple continuous sub-intervals. Each velocity increment is assigned to a corresponding discrete increment interval based on its numerical value. The discrete increment state corresponding to this interval is the state to which the current velocity increment belongs, and it is used to convert continuous velocity increment values ​​into discrete state numbers.

[0265] In this embodiment, the probability distribution data of the discrete incremental state at each prediction time point within the prediction time domain can be recursively obtained based on the current discrete incremental state and the target state transition probability. The target state transition probability is the adaptively adjusted transition probability data, which describes the transition law of the discrete incremental state under the current operating condition combination.

[0266] Using the current discrete incremental state as the starting state for prediction, the probability distribution data of the discrete incremental state at each prediction time is obtained by recursively applying the target state transition probability at each time step. This distribution data describes the probability that the vehicle speed increment is in each discrete incremental state at each prediction time step.

[0267] In the embodiments of this application, the expected value of the velocity increment at each prediction time can be calculated based on the discrete incremental state probability distribution data at each prediction time.

[0268] Each discrete increment state corresponds to a velocity increment value. The probability distribution data of the discrete increment state at each prediction time is weighted and summed with the corresponding velocity increment value to obtain the expected velocity increment value at that prediction time, which is used to characterize the most likely velocity increment size at that prediction time.

[0269] In this embodiment, the original vehicle speed prediction value for each prediction time can be obtained by accumulating the current vehicle speed and the expected speed increment values ​​for each prediction time. Using the current vehicle speed as the prediction starting value, the expected speed increment values ​​for each prediction time are accumulated sequentially according to time to obtain the vehicle speed value corresponding to each prediction time. These vehicle speed values ​​are the original vehicle speed prediction values, which are used for subsequent physical feasibility verification.

[0270] In practice, the original vehicle speed prediction values ​​at each prediction time can be obtained in the following way.

[0271] Define the discrete state probability vector at time k:

[0272] in, This represents the discrete state probability vector at the current moment. This vector represents the probability that the discrete state is i at the current moment, K represents the number of discrete incremental states, and the superscript T denotes the vector transpose. Only one element in this vector has a value of 1, while the rest have values ​​of 0, representing a definite and unique degenerate distribution of the discrete state at the current moment.

[0273]

[0274] in, This represents the probability that the discrete state is i at the current moment. Let represent the indicator function, which takes a value of 1 when the current discrete state is equal to i, and a value of 0 otherwise. K represents the number of discrete incremental states. This formula indicates that the state distribution at the current time is deterministic, that is, the current discrete state is known and unique.

[0275]

[0276] in, This represents the prediction of the discrete state distribution at the (k+h)th prediction time based on information from the current time. This represents the conditional transition probability matrix corresponding to the current working condition combination, with the superscript T indicating matrix transpose. This indicates that based on the information at the current time, the (k+h)th time... 1. Prediction of discrete state distribution at the prediction time.

[0277] This formula multiplies the state distribution at the previous prediction time with the transition probability matrix under the current working condition combination to obtain the state distribution at the current prediction time. It then iterates time by time to achieve recursive prediction of the discrete state distribution at multiple future times.

[0278] Define each discrete state The corresponding speed increment represents the center of the bin:

[0279] in, This represents the velocity increment value corresponding to discrete incremental state i. Indicates the i-th One discrete increment interval boundary, This represents the boundary of the i-th discrete increment interval. The formula takes the average of the left and right boundaries of the discrete increment interval as the representative value of the velocity increment corresponding to that interval, used to map the discrete state back to a continuous velocity increment value.

[0280] The formula for calculating the expected value of the velocity increment is as follows:

[0281] in, This represents the expected value of the velocity increment at the predicted time k+h. This represents the probability that the discrete incremental state at the predicted time k+h is i, based on the information at the current time. Let represent the representative value of the velocity increment corresponding to discrete incremental state i, and K represent the number of discrete incremental states. This formula multiplies the probability of each discrete incremental state by its corresponding representative value of velocity increment and then sums the results to obtain the statistical expectation value of the velocity increment at the predicted time.

[0282] The formula for accumulating the original predicted vehicle speed values ​​is as follows:

[0283] in, This represents the original predicted vehicle speed at prediction time k+h. This indicates that for the k+h-th... 1. The original predicted vehicle speed at the predicted time. This represents the expected speed increment at the predicted time k+h. The formula adds the predicted speed at the previous time to the expected speed increment at the current time to obtain the predicted speed at the current time.

[0284] The formula for predicting the initial value of the original vehicle speed is as follows:

[0285] in, This represents the original predicted vehicle speed at the start of the prediction process. This represents the actual vehicle speed at the current moment. This formula uses the actual vehicle speed at the current moment as the starting value for prediction, and then uses it to calculate the vehicle speed at each subsequent prediction moment by adding it up. Let represent the closed interval formed by the lower limit and upper limit of vehicle speed at the predicted time k+h.

[0286] The projection correction formula is as follows:

[0287] in, This represents the predicted vehicle speed after projection correction at the predicted time k+h. The meanings of the other parameters can be found in the previous explanation.

[0288] Reference Figure 3 This diagram illustrates an reachability domain projection correction schematic according to an embodiment of this application. The input area includes the current vehicle speed, the current discrete incremental state, the conditional transition probability slice corresponding to the current operating condition combination, and the speed reachability domain channel.

[0289] The multi-step recursive prediction module recursively obtains the discrete incremental state probability distribution at each prediction time in the prediction time domain based on the current discrete incremental state and the conditional transition probability slice, and calculates the expected value of the speed increment. Starting with the current vehicle speed, it accumulates to obtain the original vehicle speed prediction value.

[0290] The reachability domain projection correction module projects the original vehicle speed prediction values ​​onto the speed reachability domain channel time-by-time to obtain the corrected vehicle speed prediction values ​​and calculates the projection default. The confidence calculation module integrates the vehicle speed residual change, mode cumulative change, and projection default to construct the prediction confidence. The output area outputs the projection-corrected predicted vehicle speed sequence and the prediction confidence.

[0291] This application determines the current discrete increment state based on the current speed increment, and recursively obtains the discrete increment state probability distribution data for each prediction time in the prediction time domain based on the current discrete increment state and the target state transition probability. Then, it calculates the expected value of the speed increment based on the probability distribution data for each prediction time, and finally obtains the original vehicle speed prediction value by summing the current vehicle speed and the expected values ​​of each speed increment. This achieves the prediction of the statistical law of vehicle speed change, solves the technical problem that existing methods cannot perform recursive prediction based on the current actual state, and ensures that the prediction result conforms to the statistical law of vehicle speed change.

[0292] Optionally, the steps of obtaining the upper and lower limits of vehicle speed at each prediction time within the prediction time domain include: S1131, determine the upper limit of the driving torque and the upper limit of the braking force corresponding to the power mode according to the power mode; S1132, the current vehicle speed is used as the upper and lower speed limits of the vehicle at the current moment; where the current moment is the start moment of the prediction time domain; S1133, obtain the road resistance value of the road where the vehicle is currently located; S1134, For any predicted time other than the current time, calculate the increment of the upper limit of acceleration at the predicted time based on the upper limit of vehicle speed, upper limit of driving torque and road resistance at the previous time. S1135, calculate the decrease value of the lower limit of acceleration at the prediction time based on the lower limit of vehicle speed, upper limit of braking force and road resistance of the previous time. S1136, The upper limit of vehicle speed at the previous time of prediction is added to the upper limit of acceleration at the time of prediction to obtain the upper limit of vehicle speed at the time of prediction. S1137, the lower limit of vehicle speed at the predicted time is obtained by subtracting the lower limit of acceleration at the predicted time from the previous time.

[0293] In this embodiment, the upper limits of drive torque and braking force corresponding to a power mode can be determined based on the power mode. The power mode is the current dominant operating type of the vehicle's powertrain. The upper limit of drive torque is the maximum drive torque that the vehicle can output in the current power mode, and the upper limit of braking force is the maximum braking force that the vehicle can provide in the current power mode. The upper limits of drive torque and braking force differ in different power modes, and these values ​​are determined using a calibration table.

[0294] In this embodiment, the current vehicle speed can be used as both the upper and lower speed limits at the current moment. The prediction time domain starts from the current moment and includes multiple prediction moments. The current moment is the starting moment of the prediction time domain, and the vehicle speed at the current moment is a definite true value. This true value is used as both the upper and lower speed limits for subsequent recursive calculation of the achievable speed range at each prediction moment within the prediction time domain.

[0295] In this embodiment, the road resistance value of the road where the vehicle is currently located can be obtained. Road resistance is the obstructive force exerted on the vehicle by the road during its movement, consisting of rolling resistance and air resistance, and is used to subsequently calculate the upper limit increase and lower limit decrease of acceleration.

[0296] In the specific implementation, for any prediction time within the prediction time domain, the previous time refers to the time before that prediction time. If the prediction time is the first prediction time within the prediction time domain, then the previous time is the current time; if the prediction time is a subsequent prediction time within the prediction time domain, then the previous time is the previous prediction time, and the upper and lower speed limits of the vehicle at the previous prediction time have already been calculated in the previous round of recursion.

[0297] Taking a sampling period of 0.1 seconds as an example, the current time is k, and the predicted times in the prediction time domain are k+1, k+2, k+3, and so on. For prediction time k+1, its previous time is the current time k, and the upper and lower speed limits of the vehicle at the current time k are both known actual vehicle speeds. For prediction time k+2, its previous time is prediction time k+1, and the upper and lower speed limits of the vehicle at prediction time k+1 have already been obtained when calculating prediction time k+1. And so on, the upper and lower speed limits of each prediction time depend on the calculation results of its previous time, forming a recursive process time-by-time, until all prediction times in the prediction time domain are covered.

[0298] Therefore, when calculating the upper limit increase and lower limit decrease of acceleration at any prediction time, the upper limit and lower limit of vehicle speed at the previous time on which it is relied is the actual vehicle speed at the current time in the first prediction time, and the upper and lower limit values ​​of vehicle speed calculated at the previous prediction time in subsequent prediction times.

[0299] In this embodiment, the acceleration upper limit increment can be calculated for any predicted time other than the current time. The maximum available driving force at the vehicle speed is determined based on the vehicle speed upper limit value of the previous time. The maximum available driving force is jointly determined by the driving torque upper limit value and the current vehicle speed. The maximum available driving force is subtracted from the road resistance corresponding to the vehicle speed upper limit value of the previous time to obtain the net driving force at that speed. The net driving force is divided by the vehicle mass to obtain the acceleration upper limit increment.

[0300] In this embodiment, the acceleration lower limit decrease value can be calculated based on the vehicle speed lower limit, braking force upper limit, and road resistance of the previous time. The braking force upper limit is added to the road resistance corresponding to the vehicle speed lower limit of the previous time to obtain the total braking force, and the total braking force is divided by the vehicle mass to obtain the acceleration lower limit decrease value.

[0301] In this embodiment, the upper limit of vehicle speed at the previous time step and the increment of the upper limit of acceleration at the current time step can be added together to obtain the upper limit of vehicle speed at the current time step. The upper limit of vehicle speed at the previous time step is added to the product of the increment of the upper limit of acceleration at the current time step and the sampling time interval to obtain the upper limit of vehicle speed at the current time step, which reflects the highest speed that the vehicle can reach under the action of maximum driving force.

[0302] In this embodiment, the lower limit of vehicle speed at the predicted time can be obtained by subtracting the lower limit of acceleration at the predicted time from the lower limit of vehicle speed at the previous time. The lower limit of vehicle speed at the previous time is then subtracted from the product of the lower limit of acceleration at the current time and the sampling time interval to obtain the lower limit of vehicle speed at the current predicted time, which reflects the minimum speed that the vehicle can reach under maximum braking force.

[0303] In practical implementation, the upper and lower limits of vehicle speed at each prediction time point within the prediction time domain can be determined in the following way: First, the vehicle's longitudinal dynamics satisfy:

[0304] Where M represents the total vehicle mass, Indicates the vehicle speed at the next moment. Indicates the vehicle speed at the current moment. Indicates the sampling time interval. This represents the longitudinal resultant force at the wheel end at the current moment. This represents the road resistance at the current vehicle speed. This equation describes how, during the longitudinal movement of the vehicle, the difference between the longitudinal resultant force at the wheel end and the road resistance determines the magnitude of the vehicle's acceleration. The acceleration multiplied by the total vehicle mass equals the resultant force, reflecting the vehicle's mechanical equilibrium in the longitudinal direction.

[0305] Road resistance consists of rolling resistance and air resistance:

[0306] Indicates road resistance. The value represents the rolling resistance coefficient, M represents the total vehicle mass, g represents the acceleration due to gravity, and ρ represents the air density. Let A represent the air resistance coefficient, V represent the vehicle's frontal area, and v represent the current vehicle speed. This equation describes the composition of road resistance experienced by a vehicle during travel. Road resistance consists of two parts: rolling resistance and air resistance. Rolling resistance is proportional to the product of the vehicle's mass and gravitational acceleration, while air resistance is proportional to the square of the vehicle speed, reflecting the physical law that air resistance increases with vehicle speed.

[0307] Under the current power mode, the driving and braking capability parameters are determined using a calibration table. In driving mode, the maximum maximum wheel-end driving torque is 4500 N·m, the maximum maximum driving power is 120 kW, and the maximum braking force is 0 N. In combined braking mode, the maximum wheel-end driving torque is 0 N·m, the maximum driving power is 0 kW, and the maximum braking force is 4000 N. In braking mode, the maximum wheel-end driving torque is 0 N·m, the maximum driving power is 0 kW, and the maximum braking force is 9000 N. These values ​​are illustrative examples; in practical applications, they can be obtained in real-time by looking up tables or interpolating based on the vehicle platform, gear position, battery state of charge, temperature, torque limit status, and braking system capabilities.

[0308] The maximum available driving force at vehicle speed v is defined as follows:

[0309] in, This indicates the maximum available driving force at the current vehicle speed and under the current power mode. This indicates the upper limit of the maximum driving torque at the wheel end under the current power mode. Indicates the equivalent radius of the wheel. This indicates the maximum drive power limit under the current power mode, and v represents the current vehicle speed. This represents the zero constant. The formula determines the maximum usable driving force by taking the smaller value between the driving force under torque limitation and the driving force under power limitation, reflecting the physical constraint that the vehicle's driving force is simultaneously limited by the maximum output torque and the maximum output power.

[0310] Defined in power mode Maximum available braking force:

[0311] This indicates the maximum available braking force under the current driving mode. Indicates the current power mode The maximum braking force limit under the current power mode is defined as the maximum available braking force. This formula directly uses the maximum braking force limit corresponding to the current power mode as the maximum available braking force, reflecting the physical constraint that the vehicle's braking force is limited by the maximum capacity of the braking system.

[0312] From longitudinal dynamics, we can obtain at time... The upper bound of the maximum available acceleration and the lower bound of the minimum available acceleration are respectively:

[0313]

[0314] This represents the upper bound of the maximum available acceleration at the current time k. This indicates the maximum available driving force at the current vehicle speed and under the current power mode. M represents the road resistance at the current vehicle speed, and M represents the total vehicle mass. This represents the lower bound of the minimum available acceleration at the current moment. This indicates the maximum available braking force under the current driving mode. This indicates the road resistance at the current vehicle speed.

[0315] Thus, the available acceleration range at the current time k is obtained:

[0316] Let the prediction step size be H, and define the upper and lower bound sequences of the velocity reachability domain. and The algorithm initializes the system and recursively calculates the next boundary for each step within the predicted step size range. When calculating the driving force, the current upper bound vehicle speed is used to calculate the maximum available driving force. This recursive method uses the current upper bound vehicle speed to calculate the available driving force when recursively calculating the upper boundary of the vehicle speed, which is a conservative boundary calculation method. Higher vehicle speeds result in greater air resistance, and under power constraints, the available driving force decreases with increasing vehicle speed. Using the upper bound vehicle speed avoids overestimating the maximum achievable speed for the next step, ensuring that the calculated speed achievable domain does not exceed the actual capabilities of the vehicle's powertrain. Substituting the boundary values ​​into the formulas for the upper bound of the maximum available acceleration and the lower bound of the minimum available acceleration, we obtain:

[0317]

[0318] Furthermore, the recursive approach to the speed-reachable domain boundary can be:

[0319] in, This represents the upper bound of the vehicle speed at time k+h+1. This represents the upper bound of the vehicle speed at time k+h. Let represent the upper bound of the maximum available acceleration at time k+h, and Δt represent the sampling time interval. This formula adds the upper bound of the vehicle speed at the previous time to the upper bound of the maximum available acceleration at the previous time multiplied by the sampling time interval, and obtains the upper bound of the vehicle speed at the current time. The larger value between 0 and this result is taken to ensure that the vehicle speed is not negative.

[0320]

[0321] in, This represents the lower bound of the vehicle speed at time k+h+1. This represents the lower bound of the vehicle speed at time k+h. Let represent the lower bound of the minimum available acceleration at time k+h, and Δt represent the sampling time interval. This formula adds the lower bound of the vehicle speed at the previous time to the result of multiplying the lower bound of the minimum available acceleration at the previous time by the sampling time interval, and takes the larger value between zero and the result to ensure that the vehicle speed is not negative.

[0322] The final predicted velocity range in the time domain is:

[0323] in, This represents the achievable velocity range within the predicted time domain, starting from the current moment. Let H represent the closed interval formed by the lower and upper limits of the vehicle speed at the (k+h)th prediction time, and let H represent the prediction step size. This speed reachability channel is composed of the upper and lower limits of the vehicle speed at each prediction time within the prediction time domain. It is used to subsequently impose physical feasibility constraints on the predicted vehicle speed to ensure that the prediction result does not exceed the actual capability range of the vehicle's power system.

[0324] This application determines the upper limit of driving torque and the upper limit of braking force based on the power mode, uses the current vehicle speed as the upper and lower limits of vehicle speed at the starting moment, obtains the road resistance, and calculates the upper limit increase of acceleration and the lower limit decrease of acceleration for each prediction moment based on the upper and lower limits of vehicle speed, the upper limit of driving torque, the upper limit of braking force, and the road resistance at the previous moment. Then, the upper and lower limits of vehicle speed at each prediction moment are obtained by summing or subtracting them. This solves the technical problem of the lack of speed reachability calculation caused by the fact that the existing prediction method does not consider the physical constraints of the vehicle power system, and provides an accurate speed reachability range for verifying the physical feasibility of predicting vehicle speed.

[0325] Optionally, the method further includes the following steps: S1141, For any prediction time, calculate the first difference between the original predicted vehicle speed value and the upper limit of vehicle speed, and the second difference between the original predicted vehicle speed value and the lower limit of vehicle speed. S1142, Determine the vehicle speed correction deviation at the predicted time based on the first difference and the second difference; S1143, Determine the prediction confidence of the predicted vehicle speed sequence based on the vehicle speed residual change, the cumulative change of the pattern, and the vehicle speed correction deviation at each prediction time. S1144 uses the predicted vehicle speed sequence and prediction confidence as the prediction result of the vehicle speed sequence.

[0326] In this embodiment, a first difference between the original predicted vehicle speed and the upper limit of vehicle speed, and a second difference between the original predicted vehicle speed and the lower limit of vehicle speed, can be calculated for any prediction time. The original predicted vehicle speed is a prediction result that has not been verified by physical constraints. The upper limit of vehicle speed and the lower limit of vehicle speed constitute the speed reachable range at that prediction time.

[0327] The first difference is obtained by subtracting the upper limit of the vehicle speed from the original predicted vehicle speed value, and the second difference is obtained by subtracting the lower limit of the vehicle speed from the original predicted vehicle speed value. The two differences are used to quantify the degree to which the original prediction result exceeds or falls below the physical boundary.

[0328] In this embodiment, the vehicle speed correction deviation at the prediction time can be determined based on a first difference and a second difference. The vehicle speed correction deviation is used to quantify the degree to which the original predicted vehicle speed deviates from the physically reachable range. When the original predicted vehicle speed is between the upper and lower speed limits, the vehicle speed correction deviation is zero. When the original predicted vehicle speed exceeds the upper speed limit, the vehicle speed correction deviation is determined by the first difference. When the original predicted vehicle speed is lower than the lower speed limit, the vehicle speed correction deviation is determined by the second difference.

[0329] In the embodiments of this application, the prediction confidence of the predicted vehicle speed sequence can be determined based on the vehicle speed residual change, the pattern cumulative change, and the vehicle speed correction deviation at each prediction time.

[0330] The vehicle speed residual variability reflects the degree of anomalousness of the current prediction error relative to historical fluctuation levels. The pattern cumulative variability reflects the drastic change in the dynamic pattern. The vehicle speed correction deviation reflects the degree to which the original prediction result deviates from the physically reachable range. These three quantities are integrated to obtain a comprehensive confidence index, used to characterize the overall reliability of the predicted vehicle speed sequence.

[0331] In the embodiments of this application, the predicted vehicle speed sequence and the prediction confidence can be used together as the prediction result of the vehicle speed sequence.

[0332] The predicted vehicle speed sequence is a sequence of future vehicle speed predictions verified by physical constraints, while the prediction confidence level is a quantitative assessment of the reliability of this sequence. Together, they constitute a complete prediction result, providing vehicle speed prediction data and its reliability information for subsequent higher-level control decisions.

[0333] In practice, the prediction confidence level can be calculated in the following way: Construct the average projected default value in the prediction time domain:

[0334] in, H represents the average projected default value within the prediction time domain, and H represents the prediction step size. This represents the projection default at prediction time k+h. The formula sums the projection defaults at each prediction time within the prediction time domain and divides them by the prediction step size to obtain the average projection default, which is used to quantify the degree to which the original prediction sequence deviates from the physical reachable range.

[0335] The formula for normalized projective default intensity is as follows:

[0336] in, Indicates the normalized projective default strength. This represents the average projected default rate. This represents the normalized scale constant for projection default. This formula divides the average projection default by the preset scale constant and restricts the result to the range of zero to one, obtaining the normalized projection default intensity, which is used for subsequent confidence calculation.

[0337] In one embodiment, if the vehicle speed is measured in meters per second, the projection default normalization scale constant can be taken as 0.5 meters per second. If the vehicle speed is measured in kilometers per hour, the projection default normalization scale constant can be taken as 2 kilometers per hour. When the average projection default reaches this scale constant, the normalized projection default intensity approaches 1.

[0338] The formula for calculating prediction confidence is as follows:

[0339] in, Indicates the prediction confidence level. Indicates the residual strength weight. Indicates the degree of change in vehicle speed residual. Indicates the intensity weight of pattern change. Indicates the cumulative degree of change of the pattern. Indicates the projected default strength weight. This represents the normalized projected default intensity. The formula multiplies the vehicle speed residual variation, the mode cumulative variation, and the normalized projected default intensity by their respective weighting coefficients, then sums them up. The negative of this sum is taken as the exponent of the natural constant to obtain the prediction confidence level.

[0340] The residual strength weight, mode change strength weight, and projected default strength weight are all positive coefficients greater than zero. In one embodiment, the three weight coefficients can be set to 1.0, 0.8, and 1.2, respectively, so that the impact of projected default strength on confidence is slightly higher than that of residual strength and mode cumulative change.

[0341] The prediction confidence level is not only displayed as an output but also used to determine the validity of the prediction results and for closed-loop self-correction control. In one embodiment, the confidence threshold is set to 0.6. When the prediction confidence level is greater than or equal to the confidence threshold, the current prediction sequence is considered reliable and directly output to the energy management, torque coordination, or noise, vibration, and roughness control modules. When the prediction confidence level is less than the confidence threshold, the current prediction is considered to have high uncertainty, a low confidence flag is output, and the update intensity of the current operating condition combination slice is increased in the next round of online updates, or the weight of the prediction sequence in the downstream control module is reduced. The downstream module can set the weight of the vehicle speed prediction sequence to the prediction confidence level and set the weight of the conservative backup strategy to one minus the prediction confidence level.

[0342] This application determines the vehicle speed correction deviation by calculating the difference between the original predicted vehicle speed value and the upper and lower limits of the vehicle speed at each prediction time. Then, it integrates the vehicle speed residual change, the mode cumulative change, and the vehicle speed correction deviation to jointly determine the prediction confidence. The predicted vehicle speed sequence and the prediction confidence are output together as the prediction result. This solves the technical problem that existing methods cannot quantify the reliability of the prediction results, provides vehicle speed prediction data and its reliability information for subsequent control decisions, and improves the acceptability of the prediction results.

[0343] Reference Figure 4 The diagram shows a range-extended vehicle powertrain system according to an embodiment of this application.

[0344] like Figure 4 The engine and alternator are mechanically connected to form a range extender, which converts the chemical energy in the fuel tank into electrical energy. The alternating current output from the generator is converted into direct current by a rectifier and then sent to the battery pack for storage or directly to the drive motor.

[0345] The battery pack, acting as an energy storage device, can be charged by the range extender and also provide power to the drive motor. The drive motor converts electrical energy into mechanical energy to propel the vehicle.

[0346] The controller receives status signals from various components and coordinates the start-stop of the range extender, the charging and discharging power of the battery pack, and the output torque of the drive motor according to the vehicle's energy management strategy, so as to achieve efficient allocation and utilization of the vehicle's energy.

[0347] The reduction gear, located between the drive motor and the wheel, converts the high-speed, low-torque output of the drive motor into low-speed, high-torque output to match the wheel's speed and torque requirements. Simultaneously, during vehicle braking or coasting, the drive motor acts as a generator, transferring the mechanical energy from the wheel to the drive motor via the reduction gear, thus achieving energy recovery. The reduction gear ensures that the drive motor always operates within its efficient speed range, improving vehicle power transmission efficiency and driving smoothness.

[0348] Reference Figure 5 The diagram shows a logical schematic of a vehicle control method provided in an embodiment of this application.

[0349] The CAN data acquisition layer acquires vehicle speed signals, acceleration signals, and braking signals from the vehicle bus and transmits the vehicle speed signals to the vehicle speed prediction module.

[0350] The vehicle speed prediction module takes historical window data at the sampling time as input, constructs the dynamic system mode variables and driving intention hidden states online, and obtains discrete states by discrete encoding the speed increment sequence. Under the conditions of dynamic mode and driving intention, it learns and updates the state transition probabilities online to form a conditional transition tensor.

[0351] The vehicle speed prediction module includes seven steps: S1 Residual Closed-Loop Pre-Update. After obtaining the actual vehicle speed, the one-step prediction residual from the previous round to the current moment is calculated, and the residual statistics are updated using exponential moving average estimation. The residual statistics are used for forgetting factor adjustment and self-correction strength adjustment in this round of online updates to improve adaptability to changes in operating conditions.

[0352] S2 Mode Change Point Detection and Mode Variable Construction. Feature vectors are constructed based on the state and constraint-related signals of the dynamical system. Change point statistics are generated using the concept of cumulative summation. When triggering conditions are met, a mode change point is determined, and the dynamical system mode variables are updated accordingly. These updates are used to index the mode dimension of the conditional transition tensor.

[0353] S3 Powertrain Constraint Modeling and Speed ​​Reachability Calculation. Under the current power mode, based on the power limit, torque limit, braking capacity limit, and road resistance model, the available acceleration range and corresponding speed reachability are calculated for each step, and the speed reachability channel in the predicted time domain is obtained through recursion.

[0354] S4 Inference of Driving Intent and Hidden State of Driving Conditions. Intent features are constructed based on vehicle speed signals, acceleration signals, and braking signals. Threshold rules and logical constraints are used to infer the hidden state of driving intent, which is used to index the intent dimension of the conditional transfer tensor to distinguish the different vehicle speed change patterns under conditions such as acceleration, cruising, coasting, and braking.

[0355] S5 State Discreteness and Markov Property Test. Continuous vehicle speed sequences are encoded into discrete state sequences. Discrete states are obtained through binning mapping based on speed increments. Markov property tests are performed on subsets of data containing power mode and driving intention conditions to determine the modeling order and discrete granularity.

[0356] S6 Conditional Transition Tensor Construction and Online Update. Transition counts are statistically analyzed and transition probabilities are estimated under different power modes and driving intentions to form a conditional transition tensor. A forgetting factor is introduced to enable online updates, allowing the model to gradually forget older data while learning new data. The forgetting factor is adaptively adjusted based on residual statistics and change point intensity, achieving rapid response to operating condition changes.

[0357] S7 multi-step prediction, dynamic constraint projection correction, and confidence output. Under the current dynamic mode and driving intention conditions, the corresponding transition probability matrix is ​​selected and multi-step distribution recursion is performed to obtain the original predicted vehicle speed sequence. The original predicted sequence is progressively projected into the speed reachability domain to obtain a predicted vehicle speed sequence that satisfies physical realizability, and the projection default amount is calculated. The confidence index is constructed and output by fusing residual uncertainty, projection default amount, and change point intensity.

[0358] When the residual exceeds a threshold, a self-correction mechanism is triggered to correct the conditional transition tensor or related parameters. The correction of the transition probability should satisfy non-negativity and normalization constraints to ensure the rationality of the transition probability.

[0359] The predicted vehicle speed sequence output by the vehicle speed prediction module is transmitted to the vehicle control layer. The vehicle control layer generates motor torque control commands and trajectory speed actuator control commands based on the predicted vehicle speed sequence. The actuator control commands are applied to the vehicle powertrain system. When the residual exceeds a threshold, a self-correction mechanism is triggered to correct the conditional transition tensor or related parameters. The actual operating status of the vehicle powertrain system is fed back to the vehicle speed prediction module through the CAN data acquisition layer, forming a closed-loop control.

[0360] Reference Figure 6 The diagram shows a structural schematic of a vehicle control device according to an embodiment of this application. The vehicle includes a power system, and the device includes: The parameter acquisition module 601 is used to acquire the vehicle's power state parameters and current speed at the current moment, and determine the vehicle's power mode, cumulative mode change degree, and driver's driving intention based on the power state parameters and current speed. The residual variation determination module 602 is used to obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual variation value based on the current vehicle speed and the historical vehicle speed prediction value. The initial transition probability acquisition module 603 is used to acquire the initial state transition probability corresponding to the power mode and the driving intention from a pre-constructed set of state transition probabilities; The target transition probability acquisition module 604 is used to adjust the initial state transition probability according to the cumulative change degree of the mode and the change degree of the vehicle speed residual to obtain the target state transition probability; The vehicle speed prediction module 605 is used to determine the future vehicle speed prediction value at each prediction time in the prediction time domain based on the current vehicle speed and the target state transition probability. The vehicle speed sequence prediction module 606 is used to combine the future vehicle speed prediction values ​​at each prediction time according to the time sequence of the prediction time to obtain the predicted vehicle speed sequence in the prediction time domain. The vehicle control module 607 is used to generate control commands based on the predicted vehicle speed sequence and adjust the operating parameters of the power system based on the control commands.

[0361] In one optional embodiment of this application, the power state parameters include motor output torque, brake pedal opening, and accelerator pedal opening. The parameter acquisition module 601 includes: The power mode determination submodule is used to determine the power mode based on the motor output torque and the brake pedal opening. The power state characteristic determination submodule is used to determine the power state characteristics based on the motor output torque, the brake pedal opening and the current vehicle speed. The mode cumulative change degree determination submodule is used to determine the mode cumulative change degree based on the dynamic state characteristics; The driving intention determination submodule is used to determine the driver's driving intention based on the current vehicle speed, the brake pedal opening, and the accelerator pedal opening.

[0362] In one optional embodiment of this application, the power mode determination submodule includes: The threshold acquisition unit is used to acquire the preset torque threshold and braking mode opening threshold. The first mode determination unit is used to determine the power mode as a drive mode if the brake pedal opening is less than the brake mode opening threshold and the motor output torque is greater than or equal to the torque threshold. The second mode determination unit is used to determine the power mode as a composite braking mode if the brake pedal opening is less than the braking mode opening threshold and the motor output torque is less than or equal to the negative torque threshold. The third mode determination unit is used to determine the power mode as a braking mode if the brake pedal opening is greater than or equal to the braking mode opening threshold.

[0363] In one optional embodiment of this application, the mode cumulative change determination submodule includes: The feature offset determination unit is used to determine the feature offset based on the difference between the dynamic state feature and the baseline feature; wherein the baseline feature is obtained by exponential moving average of historical dynamic state features; The cumulative feature offset determination unit is used to positively accumulate the feature offset over time to obtain the cumulative feature offset. The initial mode cumulative change degree determination unit is used to determine the initial mode cumulative change degree based on the ratio of the cumulative feature offset to a preset change point determination threshold. The first mode cumulative change degree determination unit is used to determine the mode cumulative change degree if the initial mode cumulative change degree is greater than the preset mode cumulative change degree threshold. The second mode cumulative change degree determination unit is used to determine the mode cumulative change degree if the initial mode cumulative change degree is less than or equal to the mode cumulative change degree threshold.

[0364] In one optional embodiment of this application, the driving intent determination submodule includes: The intent threshold acquisition unit is used to acquire preset acceleration threshold, accelerator pedal opening threshold, and braking intent opening threshold; An acceleration determination unit is used to determine the current acceleration based on the current vehicle speed; An acceleration intent determination unit is configured to determine the driving intent as an acceleration intent if the accelerator pedal opening is greater than or equal to the accelerator pedal opening threshold, the brake pedal opening is less than the brake intent opening threshold, and the current acceleration is greater than or equal to the acceleration threshold. A braking intention determination unit is used to determine the driving intention as a braking intention if the brake pedal opening is greater than or equal to the braking intention opening threshold. The cruise intention determination unit is configured to determine the driving intention as a cruise intention if the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the absolute value of the current acceleration is less than the acceleration threshold. The coasting intention determination unit is used to determine the driving intention as a coasting intention if the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is less than or equal to the negative acceleration threshold.

[0365] In one optional embodiment of this application, the residual variation determination module 602 includes: The current residual determination submodule is used to take the absolute value of the difference between the current vehicle speed and the historical vehicle speed prediction value as the current residual. The residual standard deviation determination submodule is used to perform sliding statistical processing on the current residual to obtain the residual standard deviation; The initial residual variation determination submodule is used to determine the initial residual variation based on the ratio of the current residual to the standard deviation of the residual; The first residual change determination submodule is used to take the residual intensity threshold as the vehicle speed residual change if the initial residual change is greater than the preset residual intensity threshold. The second residual variation determination submodule is used to take the initial residual variation as the vehicle speed residual variation if the initial residual variation is less than or equal to the residual strength threshold.

[0366] In an optional embodiment of this application, the apparatus further includes a state transition probability set construction module, configured to construct the state transition probability set in the following manner: Obtain historical vehicle speed sequences, which include historical vehicle speeds corresponding to different historical moments. Obtain the dynamic state parameters of the vehicle at each of the aforementioned historical moments; For any of the aforementioned historical moments, the power mode of the vehicle and the driver's driving intention corresponding to the historical moment are determined based on the power state parameters and historical vehicle speed corresponding to the historical moment. A speed increment sequence is determined based on the historical vehicle speed sequence; the speed increment sequence includes speed increments corresponding to different historical moments. Obtain several preset discrete increment intervals; wherein each discrete increment interval corresponds to a discrete increment state; The discrete increment intervals to which the velocity increments at different historical moments belong are determined respectively, and the discrete increment states corresponding to the discrete increment intervals are combined in the order of the historical moments to obtain a discrete increment state sequence. The discrete incremental state corresponding to any historical moment in the discrete incremental state sequence is associated with and saved with the vehicle's power mode and the driver's driving intention at the corresponding historical moment. Take any of the aforementioned power modes and any of the aforementioned driving intentions as a set of operating conditions, find the discrete incremental states that match the operating condition combination from the discrete incremental state sequence, and combine them in the order of the historical moments to obtain the discrete incremental state subsequence corresponding to the operating condition combination. For any of the aforementioned operating condition combinations, a first-order Markov property test is performed on the discrete incremental state subsequence corresponding to the operating condition combination. If the test passes, the set of state transition probabilities under the working condition combination is determined based on the discrete incremental state subsequence. The state transition probability set is obtained by combining the subsets of state transition probabilities under each of the aforementioned operating conditions.

[0367] In an optional embodiment of this application, the state transition probability set construction module further includes a state transition probability subset determination submodule, used for: In the discrete incremental state subsequence corresponding to the working condition combination, for each adjacent time pair, the discrete incremental state of the previous time is taken as the source discrete incremental state, and the discrete incremental state of the next time is taken as the target discrete incremental state. For any of the source discrete incremental states, determine the number of state transitions for the next discrete incremental state from the discrete incremental state subsequence as the target discrete incremental state; Based on the number of state transitions, determine the transition probability of the discrete incremental state from the source discrete incremental state to the target discrete incremental state under the given working condition combination. The transition probabilities from each source discrete incremental state to each target discrete incremental state under the working condition combination are combined to obtain the state transition probability subset corresponding to the working condition combination.

[0368] In one optional embodiment of this application, the initial transition probability acquisition module 603 includes: The operating condition combination determination submodule is used to determine the current operating condition combination based on the power mode and the driving intention; The initial transition probability selection submodule is used to select from the state transition probability set all transition probabilities from any source discrete incremental state to any target discrete incremental state under the current working condition combination, as the initial state transition probabilities.

[0369] In one optional embodiment of this application, the target transition probability acquisition module 604 includes: The parameter acquisition submodule is used to obtain the preset lower limit value of the forgetting factor, the upper limit value of the forgetting factor, the baseline forgetting value, the residual strength weight, and the change point strength weight. The first product calculation submodule is used to calculate the first product value of the residual strength weight and the vehicle speed residual change degree; The second product calculation submodule is used to calculate the second product value of the change point intensity weight and the cumulative change degree of the mode; The initial forgetting factor determination submodule is used to add the baseline forgetting value, the first product value, and the second product value to obtain the initial forgetting factor value; The first forgetting factor determination submodule is used to take the lower limit of the forgetting factor as the target forgetting factor value if the initial forgetting factor value is less than the lower limit of the forgetting factor value. The second forgetting factor determination submodule is used to take the upper limit of the forgetting factor as the target forgetting factor value if the initial forgetting factor value is greater than the upper limit of the forgetting factor value. The third forgetting factor determination submodule is used to take the initial forgetting factor value as the target forgetting factor value if the initial forgetting factor value is greater than the lower limit of the forgetting factor and less than the upper limit of the forgetting factor. The exponentially weighted update submodule is used to perform an exponentially weighted update on the initial state transition probability based on the target forgetting factor value, so as to obtain the target state transition probability.

[0370] In one optional embodiment of this application, the vehicle speed prediction module 605 includes: The speed increment determination submodule is used to determine the current speed increment based on the difference between the current vehicle speed and the vehicle speed at the previous moment. The original vehicle speed prediction submodule is used to determine the original vehicle speed prediction value at each prediction time based on the current speed increment and the target state transition probability. The vehicle speed limit acquisition submodule is used to acquire the upper and lower speed limits of the vehicle at each prediction time within the prediction time domain. The first vehicle speed value determination submodule is used to determine the vehicle speed upper limit value as the future vehicle speed prediction value at any given prediction time if the original vehicle speed prediction value is greater than or equal to the vehicle speed upper limit value. The second vehicle speed value determination submodule is used to determine the lower limit of vehicle speed as the future vehicle speed prediction value at the prediction time if the original vehicle speed prediction value is less than or equal to the lower limit of vehicle speed. The third vehicle speed value determination submodule is used to determine the original vehicle speed prediction value as the future vehicle speed prediction value at the prediction time if the original vehicle speed prediction value is greater than the lower limit of vehicle speed and less than the upper limit of vehicle speed.

[0371] In one optional embodiment of this application, the original vehicle speed prediction submodule includes: A discrete state determination unit is used to determine the discrete increment interval corresponding to the current speed based on the current speed increment, and to obtain the current discrete increment state corresponding to the current speed. The probability distribution recursion unit is used to recursively obtain the discrete incremental state probability distribution data at each prediction time in the prediction time domain based on the current discrete incremental state and the target state transition probability. The velocity increment expectation calculation unit is used to calculate the velocity increment expectation value for the corresponding prediction time based on the discrete increment state probability distribution data for each prediction time. The original vehicle speed accumulation unit is used to accumulate the original vehicle speed prediction value for each prediction time based on the current vehicle speed and the expected speed increment value for each prediction time.

[0372] In one optional embodiment of this application, the vehicle speed limit acquisition submodule includes: A torque braking force determination unit is used to determine the upper limit value of the driving torque and the upper limit value of the braking force corresponding to the power mode according to the power mode; The starting speed limit determination unit is used to use the current speed as the upper speed limit and lower speed limit at the current moment; wherein, the current moment is the starting moment of the prediction time domain; A road resistance acquisition unit is used to acquire the road resistance value of the road where the vehicle is currently located. An acceleration upper limit increment calculation unit is used to calculate the acceleration upper limit increment for any predicted time other than the current time, based on the vehicle speed upper limit value, the driving torque upper limit value, and the road resistance value of the previous time of the predicted time. The acceleration lower limit decrement calculation unit is used to calculate the acceleration lower limit decrement value at the prediction time based on the vehicle speed lower limit value, the braking force upper limit value, and the road resistance value at the previous time of the prediction time. The vehicle speed limit calculation unit is used to add the vehicle speed limit value of the previous time before the predicted time and the acceleration limit increment value of the predicted time to obtain the vehicle speed limit value of the predicted time. The vehicle speed lower limit calculation unit is used to subtract the vehicle speed lower limit value of the previous time from the acceleration lower limit decrease value of the prediction time to obtain the vehicle speed lower limit value of the prediction time.

[0373] In one optional embodiment of this application, the apparatus further includes: The difference calculation module is used to calculate, for any given prediction time, a first difference between the original predicted vehicle speed value and the upper limit of vehicle speed, and a second difference between the original predicted vehicle speed value and the lower limit of vehicle speed. The correction deviation determination module is used to determine the vehicle speed correction deviation at the predicted time based on the first difference and the second difference; The confidence determination module is used to determine the prediction confidence of the predicted vehicle speed sequence based on the vehicle speed residual change, the pattern cumulative change, and the vehicle speed correction deviation at each prediction time. The prediction result output module is used to use the predicted vehicle speed sequence and the prediction confidence as the prediction result of the vehicle speed sequence.

[0374] One embodiment of this application also provides a vehicle that may include a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the method described above.

[0375] An embodiment of this application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the method described above.

[0376] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0377] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0378] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0379] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0380] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0381] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other modifications and updates to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all modifications and updates falling within the scope of the embodiments of the present application.

[0382] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the aforementioned element.

[0383] The above provides a detailed description of a vehicle control method, device, vehicle, and medium. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A vehicle control method, characterized in that, The vehicle includes a powertrain system, and the method includes: The vehicle's power state parameters and current speed are obtained at the current moment. Based on the power state parameters and current speed, the vehicle's power mode, cumulative mode change degree, and driver's driving intention are determined. Obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual change degree based on the current vehicle speed and the historical vehicle speed prediction value; Obtain the initial state transition probabilities corresponding to the power mode and the driving intention from a pre-constructed set of state transition probabilities; The initial state transition probability is adjusted based on the cumulative change of the mode and the change of the vehicle speed residual to obtain the target state transition probability; The predicted future vehicle speed at each prediction time point within the prediction time domain is determined based on the current vehicle speed and the target state transition probability. The predicted vehicle speed values ​​at each of the predicted times are combined in chronological order to obtain the predicted vehicle speed sequence in the prediction time domain. Control commands are generated based on the predicted vehicle speed sequence, and the operating parameters of the power system are adjusted according to the control commands.

2. The method according to claim 1, characterized in that, The power state parameters include motor output torque, brake pedal opening, and accelerator pedal opening; The steps of determining the vehicle's power mode, cumulative mode change, and driver's driving intention based on the power state parameters and the current vehicle speed include: The power mode is determined based on the motor output torque and the brake pedal opening. The power state characteristics are determined based on the motor output torque, brake pedal opening, and current vehicle speed. The cumulative degree of change of the mode is determined based on the dynamic state characteristics; The driver's driving intention is determined based on the current vehicle speed, the brake pedal opening, and the accelerator pedal opening.

3. The method according to claim 2, characterized in that, The steps for determining the power mode based on the motor output torque and brake pedal opening include: Obtain the preset torque threshold and braking mode opening threshold; If the brake pedal opening is less than the braking mode opening threshold and the motor output torque is greater than or equal to the torque threshold, then the power mode is determined to be the drive mode. If the brake pedal opening is less than the braking mode opening threshold and the motor output torque is less than or equal to the negative torque threshold, then the power mode is determined to be a composite braking mode. If the brake pedal opening is greater than or equal to the braking mode opening threshold, then the power mode is determined to be a braking mode.

4. The method according to claim 2, characterized in that, The step of determining the cumulative degree of change of the mode based on the dynamic state characteristics includes: The feature offset is determined based on the difference between the dynamic state characteristics and the baseline characteristics; wherein, the baseline characteristics are obtained by exponential moving average of historical dynamic state characteristics; The feature offset is positively accumulated over time to obtain the cumulative feature offset; The initial mode cumulative change degree is determined based on the ratio of the cumulative feature offset to the preset change point determination threshold. If the initial cumulative change degree of the pattern is greater than the preset cumulative change degree threshold of the pattern, then the cumulative change degree threshold of the pattern is used as the cumulative change degree of the pattern. If the initial cumulative change of the pattern is less than or equal to the cumulative change threshold of the pattern, then the initial cumulative change of the pattern is taken as the cumulative change of the pattern.

5. The method according to claim 2, characterized in that, The step of determining the driver's driving intention based on the current vehicle speed, the brake pedal opening, and the accelerator pedal opening includes: Obtain preset acceleration threshold, accelerator pedal opening threshold, and brake intention opening threshold; Determine the current acceleration based on the current vehicle speed; If the accelerator pedal opening is greater than or equal to the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is greater than or equal to the acceleration threshold, then the driving intention is determined to be an acceleration intention. If the brake pedal opening is greater than or equal to the braking intention opening threshold, then the driving intention is determined to be a braking intention; If the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the absolute value of the current acceleration is less than the acceleration threshold, then the driving intention is determined to be a cruise intention. If the accelerator pedal opening is less than the accelerator pedal opening threshold, the brake pedal opening is less than the brake intention opening threshold, and the current acceleration is less than or equal to the negative acceleration threshold, then the driving intention is determined to be a coasting intention.

6. The method according to claim 1, characterized in that, The step of determining the degree of change of vehicle speed residual based on the current vehicle speed and the historical vehicle speed prediction value includes: The absolute value of the difference between the current vehicle speed and the historical vehicle speed prediction value is used as the current residual. The standard deviation of the residuals is obtained by performing sliding statistical processing on the current residuals; The initial residual variation is determined based on the ratio of the current residual to the standard deviation of the residual; If the initial residual change is greater than the preset residual strength threshold, then the residual strength threshold is used as the vehicle speed residual change. If the initial residual variation is less than or equal to the residual strength threshold, then the initial residual variation is taken as the vehicle speed residual variation.

7. The method according to claim 1, characterized in that, The set of state transition probabilities is constructed in the following manner: Obtain historical vehicle speed sequences, which include historical vehicle speeds corresponding to different historical moments. Obtain the dynamic state parameters of the vehicle at each of the aforementioned historical moments; For any of the aforementioned historical moments, the power mode of the vehicle and the driver's driving intention corresponding to the historical moment are determined based on the power state parameters and historical vehicle speed corresponding to the historical moment. A speed increment sequence is determined based on the historical vehicle speed sequence; the speed increment sequence includes speed increments corresponding to different historical moments. Obtain several preset discrete increment intervals; wherein each discrete increment interval corresponds to a discrete increment state; The discrete increment intervals to which the velocity increments at different historical moments belong are determined respectively, and the discrete increment states corresponding to the discrete increment intervals are combined in the order of the historical moments to obtain a discrete increment state sequence. The discrete incremental state corresponding to any historical moment in the discrete incremental state sequence is associated with and saved with the vehicle's power mode and the driver's driving intention at the corresponding historical moment. Take any of the aforementioned power modes and any of the aforementioned driving intentions as a set of operating conditions, find the discrete incremental states that match the operating condition combination from the discrete incremental state sequence, and combine them in the order of the historical moments to obtain the discrete incremental state subsequence corresponding to the operating condition combination. For any of the aforementioned operating condition combinations, a first-order Markov property test is performed on the discrete incremental state subsequence corresponding to the operating condition combination. If the test passes, the set of state transition probabilities under the working condition combination is determined based on the discrete incremental state subsequence. The state transition probability set is obtained by combining the subsets of state transition probabilities under each of the aforementioned operating conditions.

8. The method according to claim 7, characterized in that, The step of determining the subset of state transition probabilities under the working condition combination based on the discrete incremental state subsequence includes: In the discrete incremental state subsequence corresponding to the working condition combination, for each adjacent time pair, the discrete incremental state of the previous time is taken as the source discrete incremental state, and the discrete incremental state of the next time is taken as the target discrete incremental state. For any of the source discrete incremental states, determine the number of state transitions for the next discrete incremental state from the discrete incremental state subsequence as the target discrete incremental state; Based on the number of state transitions, determine the transition probability of the discrete incremental state from the source discrete incremental state to the target discrete incremental state under the given working condition combination. The transition probabilities from each source discrete incremental state to each target discrete incremental state under the working condition combination are combined to obtain the state transition probability subset corresponding to the working condition combination.

9. The method according to claim 1, characterized in that, The step of obtaining the initial state transition probabilities corresponding to the power mode and the driving intention from a pre-constructed set of state transition probabilities includes: The current operating condition combination is determined based on the power mode and the driving intention; Select from the set of state transition probabilities all transition probabilities from any source discrete incremental state to any target discrete incremental state under the current operating condition combination, and use them as the initial state transition probabilities.

10. The method according to claim 9, characterized in that, The step of adjusting the initial state transition probability based on the cumulative change of the mode and the change of the vehicle speed residual to obtain the target state transition probability includes: Obtain the preset lower limit of the forgetting factor, upper limit of the forgetting factor, baseline forgetting value, residual strength weight, and change point strength weight; Calculate the first product of the residual strength weight and the vehicle speed residual variability; Calculate the second product of the change point intensity weight and the cumulative change degree of the mode; The initial forgetting factor value is obtained by adding the baseline forgetting value, the first product value, and the second product value. If the initial forgetting factor value is less than the lower limit of the forgetting factor value, then the lower limit of the forgetting factor value is taken as the target forgetting factor value; If the initial forgetting factor value is greater than the upper limit of the forgetting factor value, then the upper limit of the forgetting factor value is taken as the target forgetting factor value; If the initial forgetting factor value is greater than the lower limit of the forgetting factor and less than the upper limit of the forgetting factor, then the initial forgetting factor value is taken as the target forgetting factor value. Based on the target forgetting factor value, the initial state transition probability is updated exponentially weighted to obtain the target state transition probability.

11. The method according to claim 10, characterized in that, The steps for determining the predicted future vehicle speed at each prediction time within the prediction time domain based on the current vehicle speed and the target state transition probability include: The current speed increment is determined based on the difference between the current vehicle speed and the vehicle speed at the previous moment; The original vehicle speed prediction value for each prediction time is determined based on the current speed increment and the target state transition probability; Obtain the upper limit and lower limit of vehicle speed at each prediction time within the prediction time domain; For any of the predicted times, if the original vehicle speed prediction value is greater than or equal to the vehicle speed limit value, then the vehicle speed limit value is determined as the future vehicle speed prediction value for the predicted time. If the original vehicle speed prediction value is less than or equal to the vehicle speed lower limit value, then the vehicle speed lower limit value is determined as the future vehicle speed prediction value at the prediction time. If the original vehicle speed prediction value is greater than the lower limit of vehicle speed and less than the upper limit of vehicle speed, then the original vehicle speed prediction value is determined as the future vehicle speed prediction value at the prediction time.

12. The method according to claim 11, characterized in that, The step of determining the original vehicle speed prediction value at each prediction time based on the current speed increment and the target state transition probability includes: Based on the current speed increment, determine the discrete increment interval corresponding to the current speed, and obtain the current discrete increment state corresponding to the current speed; Based on the current discrete incremental state and the target state transition probability, the discrete incremental state probability distribution data at each prediction time in the prediction time domain is recursively obtained; Based on the discrete incremental state probability distribution data at each of the predicted times, calculate the expected value of the velocity increment at the corresponding predicted time. Based on the current vehicle speed and the expected speed increment at each of the predicted times, the original predicted vehicle speed value at each of the predicted times is accumulated.

13. The method according to claim 11, characterized in that, The steps for obtaining the upper and lower limits of vehicle speed at each prediction time within the prediction time domain include: The upper limit values ​​of the driving torque and braking force corresponding to the driving mode are determined based on the driving mode. The current vehicle speed is used as the upper and lower speed limits for the current moment; wherein, the current moment is the start moment of the prediction time domain; Obtain the road resistance value of the road where the vehicle is currently located; For any predicted time other than the current time, the upper limit value of acceleration at the predicted time is calculated based on the upper limit value of vehicle speed, the upper limit value of driving torque, and the road resistance at the previous time. Based on the vehicle speed lower limit, braking force upper limit, and road resistance of the previous time, calculate the acceleration lower limit decrease value at the time of prediction; The upper limit of vehicle speed at the predicted time is obtained by adding the upper limit of acceleration at the predicted time to the upper limit of vehicle speed at the previous time. The lower limit of vehicle speed at the predicted time is obtained by subtracting the lower limit of acceleration at the previous time from the lower limit of vehicle speed at the predicted time.

14. The method according to claim 11, characterized in that, The method further includes the following steps: For any of the predicted times, calculate the first difference between the original predicted vehicle speed and the upper limit of vehicle speed, and the second difference between the original predicted vehicle speed and the lower limit of vehicle speed. Based on the first difference and the second difference, the vehicle speed correction deviation at the predicted time is determined; The prediction confidence of the predicted vehicle speed sequence is determined based on the vehicle speed residual change, the pattern cumulative change, and the vehicle speed correction deviation at each prediction time. The predicted vehicle speed sequence and the prediction confidence level are used as the prediction result of the vehicle speed sequence.

15. A vehicle control device, characterized in that, The vehicle includes a power system, the device comprising: The parameter acquisition module is used to acquire the vehicle's power state parameters and current speed at the current moment, and determine the vehicle's power mode, cumulative mode change degree, and driver's driving intention based on the power state parameters and current speed. The residual variation determination module is used to obtain the historical vehicle speed prediction value corresponding to the current moment, and determine the vehicle speed residual variation value based on the current vehicle speed and the historical vehicle speed prediction value. The initial transition probability acquisition module is used to acquire the initial state transition probability corresponding to the power mode and the driving intention from a pre-built set of state transition probabilities; The target transition probability acquisition module is used to adjust the initial state transition probability according to the cumulative change degree of the mode and the change degree of the vehicle speed residual to obtain the target state transition probability; The vehicle speed prediction module is used to determine the future vehicle speed prediction value at each prediction time in the prediction time domain based on the current vehicle speed and the target state transition probability. The vehicle speed sequence prediction module is used to combine the future vehicle speed prediction values ​​at each prediction time according to the time sequence of the prediction time to obtain the predicted vehicle speed sequence in the prediction time domain. The vehicle control module is used to generate control commands based on the predicted vehicle speed sequence and adjust the operating parameters of the power system according to the control commands.

16. A vehicle, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the method as described in any one of claims 1-14.

17. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the method as described in any one of claims 1-14.

Images