A multi-in-one electric drive assembly system and control method

Abstract

The application discloses a kind of multi-in-one electric drive assembly systems and control methods, it is related to new energy vehicle power system control technical field.The system includes engine, motor, power battery, clutch, vehicle transmission mechanism and electronic control system, the electronic control system includes reinforcement learning decision module, mode control module, execution control module, safety control module and strategy management module.The system is based on operating state information to generate power control instruction and drive mode instruction, realize the collaborative control of engine and motor power output and drive mode, support pure electric drive, series drive and parallel drive mode switching.The method can be according to vehicle state to execute multi-source power joint control, and have parameter setting and safety takeover ability.The application improves the energy efficiency, adaptability and operating stability of drive control system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of new energy vehicle power system control technology, and particularly relates to a multi-in-one electric drive assembly system and a control method. BACKGROUND

[0002] With the rapid development of new energy vehicles, the performance of the electric drive system becomes a core factor affecting the power performance, economy and intelligent level of the whole vehicle. In order to improve the power response efficiency, prolong the endurance and adapt to complex traffic conditions, hybrid power and multi-mode drive system gradually become one of the mainstream technical paths.

[0003] The existing multi-in-one electric drive assembly system generally integrates an engine, an electric motor, a power battery, a clutch and a transmission mechanism, and supports multiple drive modes such as pure electricity, series connection and parallel connection. However, such systems generally rely on control strategies constructed based on fixed rules or experience logic, and have poor adaptability to complex operating environments and dynamic conditions. The main technical problems are as follows: lack of adaptive control capability: traditional control strategies are difficult to achieve optimal power distribution and mode switching among multiple drive modes with high uncertainty and strong nonlinearity; energy management optimization is insufficient: fixed strategies are difficult to balance engine efficiency, battery SOC safety, electric motor response and vehicle energy consumption in real time, resulting in poor performance in some operating states; control path is rigid and lacks fault tolerance mechanism: when sensing abnormalities, strategy failure or actuator deviation occurs, there is a lack of flexible takeover mechanism, and system safety is difficult to guarantee; poor strategy deployment and migration capability: the existing control system responds slowly to changes in operating environment, and the strategy switching lags behind, and cannot achieve continuous control performance maintenance and optimization.

[0004] Therefore, there is an urgent need for an electronic control system architecture that can jointly perceive, coordinate strategies, adaptively adjust and have fault tolerance capability in complex states, especially lacking an engineering system solution that introduces deep reinforcement learning into the field of vehicle drive mode control. SUMMARY

[0005] The present application aims to overcome the technical deficiencies of the existing multi-in-one electric drive system control strategy, such as rigidity, poor adaptability, low energy utilization efficiency and lack of safety fault tolerance mechanism, and proposes a multi-in-one electric drive assembly system and control method to achieve adaptive control and optimization management of core drive units such as engines, electric motors and clutches in multiple drive modes, thereby improving the power performance, energy economy and operating safety of the whole vehicle.

[0006] To achieve the above purpose, the present application realizes the technical scheme as follows:

[0007] On the one hand, the present application provides a multi-in-one electric drive assembly system for realizing multiple drive mode switching and power coordination control of a vehicle, comprising:

[0008] An engine, an electric motor, a power battery, a clutch and a vehicle transmission mechanism, wherein:

[0009] The engine is selectively connected or decoupled with the vehicle transmission mechanism through the clutch, the electric motor is connected to the vehicle transmission mechanism for driving the vehicle, and the power battery is electrically connected with the electric motor for providing driving power or recovering braking energy;

[0010] Further comprising an electronic control system for cooperatively controlling the engine, the electric motor and the clutch during vehicle operation, the electronic control system comprising:

[0011] A reinforcement learning decision module for generating control instructions based on vehicle state information, the control instructions including power control instructions for controlling power output of the engine and the electric motor, and driving mode instructions for controlling connection state of the clutch and start-stop state of the engine;

[0012] A mode control module for controlling connection or disconnection state of the clutch and start or stop state of the engine according to the driving mode instructions, so as to switch operation state of the vehicle between pure electric driving, series driving and parallel driving modes;

[0013] An execution control module for controlling operation of the engine and the electric motor according to the power control instructions, so as to drive target torque and speed of the vehicle output;

[0014] A safety control module for judging whether abnormal control conditions are met based on vehicle state information and execution feedback of the control instructions, and implementing preset safety control strategies including output limitation, mode switching or control takeover when the conditions are met;

[0015] A strategy management module for adjusting parameters of the safety control strategies or switching control strategy models when operating environment parameters change, so as to maintain control performance of the electronic control system under different operating conditions.

[0016] Further improvement of the present application is that the reinforcement learning decision module comprises:

[0017] A state input processing unit for collecting and processing vehicle operating state information including but not limited to vehicle load, driving demand, battery state of charge SOC, engine speed, electric motor efficiency and current driving mode, for constructing a unified state vector;

[0018] The engine control agent unit is a control strategy network trained based on a deep reinforcement learning algorithm, takes the state vector as input, and outputs a power control instruction for controlling the power of the engine, and the optimization target includes fuel consumption rate minimization and output response performance optimization.

[0019] The motor control agent unit is a control strategy network trained based on a deep reinforcement learning algorithm, takes the unified state vector as input, and outputs a power control instruction for controlling the power of the motor, and the optimization target includes energy use efficiency improvement and battery safety maintenance.

[0020] The drive mode decision unit is used for generating a drive mode instruction for controlling the clutch connection state and the engine start-stop state based on the unified state vector and the agent output result.

[0021] The agent cooperative optimization unit is used for establishing a shared state interface and a reward coupling mechanism between the engine control agent unit and the motor control agent unit, and coordinating the consistency of the control instructions through joint optimization strategy output.

[0022] Further improvements of the application are that the motor control agent unit comprises:

[0023] The state sharing subunit is used for providing the unified state vector to the engine control agent unit and the motor control agent unit as common input respectively.

[0024] The reward fusion subunit is used for performing weighted calculation on the instant reward values of the engine control agent unit and the motor control agent unit according to a set reward coupling factor ρ, and generating a joint reward signal, wherein: when ρ=0, no reward sharing is performed; when ρ>0, partial reward sharing is performed; and when ρ is a specific non-zero value, joint optimization calculation is performed.

[0025] The conflict coordination subunit is used for adjusting the power control instruction or the drive mode instruction output by the two agent units when there is inconsistency, based on a preset priority, a weight function or a fusion rule.

[0026] The fusion strategy output subunit is used for outputting the power control instruction and the drive mode instruction processed by the conflict coordination subunit to the drive mode control module and the execution control module respectively.

[0027] Further improvements of the application are that the drive mode decision unit comprises:

[0028] The state discrimination subunit is used to identify the operating scenario based on the key parameters in the state vector. The key parameters include the vehicle's current speed, acceleration demand, battery state of charge, engine start / stop status, current power request, and system efficiency evaluation indicators.

[0029] The mode switching determination subunit is used to make a selection judgment among the three driving modes of pure electric drive, series drive and parallel drive based on the operation scenario identification result and target optimization strategy. The judgment process is implemented based on the strategy function and is equipped with a hysteresis mechanism to avoid frequent switching.

[0030] The clutch control command generation subunit is used to generate control commands for controlling the clutch engagement state based on the mode determination result, so as to realize the mechanical connection configuration required by the selected drive mode.

[0031] An engine start-stop control command generation subunit is used to generate the engine start-stop control signal based on the drive mode determination result, so as to enable the engine to engage or disengage from the power system as needed.

[0032] The safety redundancy determination subunit is used to monitor the execution result of the drive mode switching command in real time. If an abnormal mode switching, execution failure or system security risk is detected, a preset fault mode or degradation control strategy is triggered.

[0033] A further improvement of the present invention is that the execution control module includes a motor control unit for controlling the torque output of the first motor, and the motor control unit adopts a field-oriented control structure, including:

[0034] The speed outer loop adjustment unit takes the target speed ω as input. ref With actual rotational speed ω act The difference is output as the target torque current i. q,ref It is adjusted by a proportional-integral controller;

[0035] The current inner loop regulation unit takes the target current i as its input. q,ref With actual current i q,act The deviation between them is output as a voltage control signal or PWM duty cycle to drive the three-phase inverter;

[0036] The control structure uses a dq coordinate transformation method to transform the three-phase stationary coordinate system current to a rotating coordinate system;

[0037] The outer speed control unit and the inner current control unit form a cascaded PI control structure, which respectively completes the speed command calculation and the current control command adjustment.

[0038] The further improvement of the present application is to enhance the operation stability and regulation accuracy of the motor control unit in the execution control module, further comprising:

[0039] A parameter setting unit is configured to set the proportional gain and integral gain parameters K p , K i of the speed outer loop regulation unit and the current inner loop regulation unit in the field-oriented control structure offline, and the setting process adopts a non-dominated sorting genetic algorithm, and the optimization objective of the non-dominated sorting genetic algorithm is two functions as follows:

[0040]

[0041] Wherein, T m (t) is the actual torque of the first motor; is the average value of the torque in the optimization period T0, which is used to calculate the steady-state torque fluctuation index;

[0042] A torque feedback regulation unit is configured to collect the output torque T m (t) of the first motor in real time during the operation of the vehicle, and compare it with the preset torque fluctuation threshold value:

[0043] When the fluctuation amplitude exceeds the threshold value, the feedback regulation unit adjusts the target current i q,ref based on the current deviation value, or corrects the gain of the controller parameters K p , K i ; dt represents the integral with respect to time t; J1 is used to evaluate the speed tracking error integral of the control system in the time interval [0, T o ]; and J2 is used to measure the pulsation variance of the motor output torque in the interval.

[0044] The further improvement of the present application is that the safety control module comprises:

[0045] An abnormality determination unit is configured to obtain the vehicle speed, the state of charge SOC of the battery, the motor temperature and the working current in the vehicle operation state, and set safety threshold values for the above variables respectively; when the detection value exceeds the safety threshold value, an abnormal state determination signal is generated;

[0046] A control takeover unit is configured to replace the execution control module of the reinforcement learning decision module after receiving the abnormal state determination signal, and perform at least one of the following control operations:

[0047] A torque limiting subunit is configured to output a motor torque limiting instruction to the execution control module;

[0048] A clutch disconnection subunit is configured to output an engine clutch disconnection control instruction to the drive mode control module.

[0049] The PID control switching subunit is configured to call a preset PID regulator-based power control logic and send an output signal to the execution control module.

[0050] The further improvement of the present application is that the policy management module comprises:

[0051] A backup policy library configured with a plurality of control policy networks trained offline, each policy network corresponding to a different combination of vehicle operating environment parameters;

[0052] A policy switching judgment unit configured to judge whether a policy switching condition is met based on the battery state of charge (SOC), battery capacity change and environmental temperature parameters monitored during vehicle operation, and generate a policy switching control signal when the condition is met;

[0053] A policy switching execution unit configured to load a target policy network from the backup policy library and replace the current main policy network according to the policy switching control signal, and continuously adjust the control parameters of the original main policy network within a limited range during the replacement process.

[0054] On the other hand, the present application provides a multi-in-one electric drive assembly control method, which applies a multi-in-one electric drive assembly system according to any one of the preceding claims, and the method comprises the following steps:

[0055] Step 1: Collecting vehicle operating state information, including vehicle speed, acceleration request, battery state of charge (SOC), engine speed, motor efficiency parameters and current drive mode state;

[0056] Step 2: Based on the operating state information, generating a control instruction through reinforcement learning decision policy, including engine power instruction, motor power instruction and drive mode switching instruction;

[0057] Step 3: According to the drive mode switching instruction, controlling the engagement or disengagement state of the clutch and the start or stop state of the engine, to realize the switching between pure electric drive, series drive and parallel drive modes of the vehicle;

[0058] Step 4: According to the engine power instruction and motor power instruction, controlling the engine throttle opening and motor current vector respectively, to realize the output of vehicle target torque and speed;

[0059] Step 5: Monitoring key parameters during vehicle operation, including vehicle speed, battery SOC, motor temperature and current, and judging whether there is an abnormal control condition according to the control instruction feedback, and if so, triggering a preset safety control strategy;

[0060] Step 6: Adjust the control parameters of the safety control strategy or switch the preset control strategy model when detecting the change of the operating environment parameter, so as to ensure the control performance and stability of the electronic control system under different operating conditions.

[0061] The beneficial effects of the present application are: through the configuration of the integrated engine, motor, clutch and vehicle transmission mechanism, and the unified coordination of the electronic control system, flexible switching between pure electric driving, series driving and parallel driving modes is realized, thereby significantly improving the adaptability of the system to complex driving conditions and the continuity of power output. The system introduces a reinforcement learning decision module, which generates a composite control instruction including power control and driving mode switching based on vehicle operating state information, and can cope with dynamic optimization problems under complex environments such as nonlinearity and multivariable, overcoming the technical shortcomings of slow response and poor generalization ability of traditional control strategies. The electronic control system adopts modular design, with reinforcement learning decision module, mode control module, execution control module, safety control module and strategy management module, each module has clear responsibilities and close logical association, which is beneficial to the function division and engineering deployment of the control path. By coordinating the power output of the engine and motor, and combining real-time driving mode selection, the system can reduce fuel consumption and electricity consumption while ensuring power demand, significantly improving the energy utilization efficiency of the whole vehicle. At the same time, the safety control module can realize real-time determination of abnormal state based on key operating parameters, and trigger output limitation, control takeover and other fault tolerance strategies when necessary, thereby enhancing the safety and stability of the vehicle control; the strategy management module supports switching to a backup strategy or fine-tuning the main strategy parameters when the operating environment changes (such as battery capacity attenuation and external temperature fluctuation), thereby ensuring the continuous and reliable operation of the system under diversified conditions. BRIEF DESCRIPTION OF DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0063] Among them:

[0064] Figure 1 The system modularization diagram of the present application;

[0065] Figure 2 The method flowchart of the present application. DETAILED DESCRIPTION

[0066] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0067] like Figure 1 As shown, this is an embodiment of the present invention, which provides an all-in-one electric drive system for realizing multiple drive mode switching and power coordination control of a vehicle, including:

[0068] Engine, electric motor, power battery, clutch, and vehicle transmission mechanism, among which:

[0069] The engine can be selectively connected to or decoupled from the vehicle transmission mechanism via a clutch, the electric motor is connected to the vehicle transmission mechanism to drive the vehicle, and the power battery is electrically connected to the electric motor to provide driving power or recover braking energy.

[0070] It also includes an electronic control system for coordinated control of the engine, electric motor, and clutch during vehicle operation, the electronic control system comprising:

[0071] (1) Reinforcement Learning Decision Module

[0072] The control commands are used to generate control instructions based on vehicle status information. The control commands include power control instructions for controlling the power output of the engine and electric motor, and drive mode instructions for controlling the clutch engagement state and the engine start / stop state.

[0073] The reinforcement learning decision module includes:

[0074] The status input processing unit is used to collect and process vehicle operating status information, including but not limited to vehicle load, drive demand, battery state of charge (SOC), engine speed, electric motor efficiency, and current drive mode, in order to construct a unified state vector.

[0075] The engine control intelligent agent unit is a control strategy network trained based on a deep reinforcement learning algorithm. It takes the state vector as input and outputs a power control command for controlling the engine power. Its optimization objectives include minimizing fuel consumption rate and optimizing output response performance.

[0076] The motor control agent unit is a control strategy network trained based on a deep reinforcement learning algorithm, takes the unified state vector as input, and outputs a power control instruction for controlling the power of the motor, and the optimization target includes energy use efficiency improvement and battery safety maintenance.

[0077] The drive mode decision unit generates a drive mode instruction for controlling the clutch connection state and the engine start-stop state based on the unified state vector and the agent output result.

[0078] The agent cooperative optimization unit establishes a shared state interface and a reward coupling mechanism between the engine control agent unit and the motor control agent unit, and coordinates the consistency of the control instructions through joint optimization strategy output.

[0079] In one specific embodiment, the motor control agent unit includes:

[0080] The state sharing subunit provides the unified state vector to the engine control agent unit and the motor control agent unit as common input.

[0081] The reward fusion subunit performs weighted calculation on the immediate reward values of the engine control agent unit and the motor control agent unit according to a set reward coupling factor ρ, and generates a joint reward signal, wherein: when ρ=0, no reward sharing is performed; when ρ>0, partial reward sharing is performed; and when ρ is a specific non-zero value, joint optimization calculation is performed.

[0082] The conflict coordination subunit adjusts the control instructions based on a preset priority, a weight function, or a fusion rule when the power control instructions or the drive mode instructions output by the two agent units are inconsistent.

[0083] The fusion strategy output subunit outputs the power control instruction and the drive mode instruction processed by the conflict coordination subunit to the drive mode control module and the execution control module, respectively.

[0084] In one specific embodiment, the drive mode decision unit includes:

[0085] The state discrimination subunit performs running scene recognition based on key parameters in the state vector, including the current vehicle speed, acceleration demand, battery state of charge, engine start-stop state, current power request, and system efficiency evaluation index.

[0086] A mode switching determination subunit is configured to make an optimal determination among the three driving modes of pure electric driving, series driving and parallel driving according to the running scene recognition result and the target optimization strategy, and the determination process is implemented based on a strategy function or a mode switching map, and a hysteresis mechanism is provided to avoid frequent switching;

[0087] A clutch control instruction generation subunit is configured to generate a control instruction for controlling the clutch engagement state according to the mode determination result, so as to achieve the mechanical connection configuration required by the selected driving mode;

[0088] An engine start-stop control instruction generation subunit is configured to generate a start-stop control signal of the engine according to the driving mode determination result, so as to realize the on-demand intervention or withdrawal of the engine into or out of the power system;

[0089] A safety redundancy determination subunit is configured to monitor the execution result of the driving mode switching instruction in real time, and if a mode switching abnormality, an execution failure or a system safety risk is detected, a preset fault mode or a degradation control strategy is triggered.

[0090] Although the "reinforcement learning decision module" in the embodiment does not introduce a new component that exceeds the general elements of the existing control system in structure, its control logic organization mode, agent cooperation mechanism and decision output path all have strong systematic innovation features.

[0091] Firstly, the module normalizes the multi-dimensional state information through a "state input processing unit" to form a unified state vector, which is used to drive multiple control strategy agents and realize consistent coupling in the input space. Compared with the traditional way of modeling the motor and engine control paths separately, the embodiment realizes the fusion of the horizontal control dimension and improves the response consistency of the model to the changes of multi-source state variables.

[0092] Secondly, the embodiment adopts a "double-agent control structure", which sets the engine and the motor as independent control agents, respectively, and outputs control strategies based on independent training of respective reinforcement learning networks, and introduces a quantitative adjustable incentive sharing mechanism through a "reward fusion subunit" to dynamically adjust the optimization target proportion of the two according to the vehicle running task priority. This "soft coupling-adjustable fusion" method is relatively rare in existing multi-drive path control systems, and provides a new engineering implementation path for solving the dynamic game between energy efficiency and response performance.

[0093] Further, when the power control instruction and the driving mode instruction conflict, the embodiment is provided with a "conflict coordination subunit", which introduces preset priority and fusion rules to balance the conflicting instructions, ensuring that the final output behavior has functional continuity and decision stability. Compared with the existing method of relying only on fixed priority or fuzzy logic to handle conflicts, the present scheme clearly defines the conflict resolution path through structured modules, facilitating strategy deployment and expansion.

[0094] In addition, through the "fusion strategy output subunit", the embodiment uniformly coordinates the power distribution results and the driving mode selection results before executing the instruction distribution, enabling the engine, motor and clutch to be decoupled and recoupled in the control path, taking into account system response speed and control loop consistency.

[0095] In summary, the embodiment does not expand the traditional system in terms of hardware level or number of structural units, but optimizes and integrates the control strategy organization structure and path coordination mechanism, especially the mechanism of coordinated decision output of multiple control objects under the reinforcement learning framework, which has significant non-obviousness, can effectively solve the problems of poor dynamic environment adaptability and weak consistency of conflicting strategies in existing rule-driven systems, and has clear technical progress significance.

[0096] (2) Mode control module

[0097] for controlling the engagement or disengagement state of the clutch and the start or stop state of the engine according to the driving mode instruction, to switch the running state of the vehicle between pure electric driving, series driving and parallel driving modes;

[0098] The mode control module described in the embodiment, although similar to the control module of the existing hybrid power driving system in terms of hardware structure and functional target, mainly functions to realize switching between different driving modes and execution state management, but in terms of control trigger path organization, decision response logic clarity, and cooperation mechanism with reinforcement learning strategy output, it constitutes a system implementation with unique characteristics.

[0099] (3) Execution control module

[0100] for controlling the engine and the motor to operate according to the power control instruction to drive the vehicle to output target torque and speed;

[0101] The execution control module includes a motor control unit for controlling the torque output of the first motor, which adopts a field-oriented control structure, including:

[0102] a speed outer loop regulation unit, inputting target speed ω ref and actual speed ω actThe difference between the target torque current i q,ref is adjusted by a proportional-integral controller.

[0103] The current inner loop regulation unit inputs the target current i q,ref and the deviation between the actual current i q,act , and outputs a voltage control signal or a PWM duty cycle for driving the three-phase inverter.

[0104] The control structure adopts a d-q coordinate transformation method to transform the three-phase stationary coordinate system current to a rotating coordinate system.

[0105] The speed outer loop regulation unit and the current inner loop regulation unit constitute an inner-outer loop cascade PI control structure, which respectively complete the speed command calculation and the current control command adjustment.

[0106] In one embodiment, to enhance the operation stability and adjustment accuracy of the motor control unit in the execution control module, further comprising:

[0107] A parameter setting unit is used to set the proportional gain and integral gain parameters K p and K i of the speed outer loop regulation unit and the current inner loop regulation unit in the field-oriented control structure offline, and the setting process adopts a non-dominated sorting genetic algorithm, and the optimization objective of the non-dominated sorting genetic algorithm is the following two functions:

[0108]

[0109] Wherein: T m (t) is the actual torque of the first motor; is the average value of the torque in the optimization period T0, which is used to calculate the steady-state torque fluctuation index;

[0110] A torque feedback regulation unit is used to collect the output torque T m (t) of the first motor in real time during vehicle operation, and compare it with the preset torque fluctuation threshold value:

[0111] When the fluctuation amplitude exceeds the threshold value, the feedback regulation unit adjusts the target current i q,ref based on the current deviation value, or corrects the gain of the controller parameters K p and K i ; dt represents the integral with respect to time t; J1 is used to evaluate the speed tracking error integral of the control system in the time interval [0, T o ]; and J2 is used to measure the pulsation variance of the motor output torque in the interval.

[0112] The execution control module described in this embodiment adopts the mature technical path of field-oriented control strategy, but its systematic innovative design in structure configuration, parameter setting method and dynamic adjustment mechanism significantly improves the engineering applicability and control robustness of traditional motor control scheme.

[0113] Firstly, this embodiment constructs a double-loop control structure with speed outer loop and current inner loop in series. The speed loop is used to adjust the deviation between the target speed and the actual speed, and outputs the target current through the external proportional integral controller. The current loop controls the inverter output based on the current deviation, and finally realizes fine control of the motor torque. Compared with the existing control strategy mainly using single closed loop or fixed parameter control, this structure realizes control target decomposition across time scales and double-layer error correction mechanism.

[0114] Secondly, in order to ensure the stability of the controller in a wide range of operating conditions of the vehicle, this embodiment adopts a parameter setting method based on multi-objective optimization. By using intelligent optimization algorithm offline to jointly adjust the proportional gain and integral gain parameters in the controller, and considering the speed tracking accuracy and motor output torque stability of the control system at the same time, the parameter selection that takes into account the power response and driving stability is realized. This multi-objective evaluation and optimization method breaks through the traditional mode of relying on engineering experience or single performance index to set parameters, and reflects the practical landing of system performance trade-off thinking in the driving control strategy.

[0115] Furthermore, aiming at the output fluctuation problem of the motor in the dynamic running process, this embodiment further introduces a torque feedback adjustment mechanism. This mechanism continuously monitors the output torque of the motor during vehicle operation and compares it with the preset fluctuation threshold. When the fluctuation amplitude exceeds the safe range, the system automatically triggers the fine tuning mechanism to immediately correct the target current or controller parameters, thereby effectively suppressing the output instability phenomenon caused by road disturbance, load mutation, etc. This mechanism not only improves the disturbance adaptability of the electric drive system, but also ensures the power output continuity and safety of the whole vehicle in the actual driving process.

[0116] The innovation of this embodiment is not in the breakthrough of control principle, but in the organization method, feedback path design and optimization process structure of control strategy in actual engineering. The complete control system formed can significantly improve the robustness, self-adaptability and energy efficiency of the whole vehicle electric drive control, and has strong technical progress and practical application value.

[0117] (4) Safety control module

[0118] For the execution feedback based on vehicle state information and the control instruction, it is judged whether the abnormal control condition is met, and when the condition is met, the preset safety control strategy is implemented, including output limitation, mode switching or control takeover.

[0119] The safety control module comprises:

[0120] An abnormality determination unit is configured to acquire vehicle speed, battery state of charge (SOC), motor temperature and working current in a vehicle operating state, and set safety thresholds for the above variables respectively; when a detection value exceeds the safety threshold, an abnormality state determination signal is generated;

[0121] A control takeover unit is configured to replace the reinforcement learning decision module to control the execution control module after receiving the abnormality state determination signal, and perform at least one of the following control operations:

[0122] A torque limiting subunit is configured to output an electric motor torque limiting instruction to the execution control module;

[0123] A clutch disconnection subunit is configured to output an engine clutch disconnection control instruction to the drive mode control module;

[0124] A PID control switching subunit is configured to call a preset power control logic based on a PID regulator, and send an output signal to the execution control module.

[0125] (5) Strategy management module

[0126] The strategy management module is configured to adjust parameters of the safety control strategy or switch control strategy models when operating environment parameters change, so as to maintain control performance of the electronic control system under different operating conditions.

[0127] The strategy management module comprises:

[0128] A backup strategy library is configured with a plurality of control strategy networks obtained by offline training, each strategy network corresponding to a different combination of vehicle operating environment parameters;

[0129] A strategy switching determination unit is configured to determine whether a strategy switching condition is met based on monitored battery state of charge (SOC), battery capacity change and environmental temperature parameters during vehicle operation, and generate a strategy switching control signal when the condition is met;

[0130] A strategy switching execution unit is configured to load a target strategy network from the backup strategy library and replace a current main strategy network according to the strategy switching control signal, and continuously adjust control parameters of the original main strategy network within a limited range during the replacement process.

[0131] As Figure 2 shown, another embodiment of the present application provides a multi-in-one electric drive assembly control method, which applies any one of the multi-in-one electric drive assembly systems described above, and the method comprises the following steps:

[0132] Step 1: Collect the running state information of the vehicle, including vehicle speed, acceleration request, battery state of charge SOC, engine speed, motor efficiency parameter and current driving mode state;

[0133] Step 2: Based on the running state information, generate control instructions through reinforcement learning decision strategy, including engine power instruction, motor power instruction and driving mode switching instruction;

[0134] Step 3: According to the driving mode switching instruction, control the engagement or disengagement state of the clutch and the start or stop state of the engine, realize the switching between pure electric drive, series drive and parallel drive mode of the vehicle;

[0135] Step 4: According to the engine power instruction and motor power instruction, control the engine throttle opening and motor current vector respectively, realize the output of vehicle target torque and speed;

[0136] Step 5: Monitor the key parameters in the running process of the vehicle, including vehicle speed, battery SOC, motor temperature, current, and according to the control instruction, execute feedback to judge whether there is abnormal control condition, if there is, trigger the preset safety control strategy;

[0137] Step 6: When detecting the change of running environment parameters, adjust the control parameters of safety control strategy, or switch the pre-set control strategy model, to ensure the control performance and stability of the electronic control system under different running conditions.

[0138] In summary, the present application proposes an electronic control system solution scheme which integrates deep reinforcement learning control, modular structure design and adaptive management mechanism, aiming at the technical problems of existing multi-in-one electric drive system, such as inflexible driving mode switching, low power distribution efficiency, poor adaptive ability and rigid control strategy. Through the dynamic optimization control of engine and motor power output by reinforcement learning agent, as well as the intelligent judgment and switching of driving mode, the global trade-off between power performance and energy efficiency utilization is realized, and the response speed and environmental adaptability of the system are improved; by setting safety control module and strategy management module, the control stability and fault tolerance ability in the running process are ensured, meeting the long-term running demand under complex working conditions. The system structure of the present application is complete, the control logic is clear, has good engineering realization and scalability, can be widely applied to hybrid electric vehicle, extended range electric vehicle and high performance new energy vehicle and other application scenarios, has significant practical value and industrialization prospect.

[0139] In the description of the application, reference can be made to terms such as "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" etc. It is intended that there are at least one embodiment or example described in connection with the particular feature, structure, material or characteristic described in connection with the embodiment or example. Moreover, the described particular feature, structure, material or characteristic can be combined in any suitable manner in one or more embodiments or examples. Furthermore, the skilled person can combine and combine features of different embodiments or examples described in the specification and characteristics of different embodiments or examples, without mutual contradiction.

[0140] Any process or method descriptions or blocks in flow charts described herein and elsewhere in this specification can be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps in the process. As well, the preferred embodiments of this application can be implemented in hardware, software, or both hardware and software.

[0141] The above description is only specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of various changes or replacements within the technical scope disclosed in the present application, and these should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-in-one electric drive assembly system for realizing multiple driving mode switching and power coordination control of a vehicle, comprising: an engine, an electric motor, a power battery, a clutch and a vehicle transmission mechanism, wherein: the engine is selectively connected or decoupled with the vehicle transmission mechanism through the clutch, the electric motor is connected to the vehicle transmission mechanism for driving the vehicle, and the power battery is electrically connected with the electric motor for providing driving electric energy or recovering braking energy; characterized in that, further comprising an electronic control system for cooperatively controlling the engine, the electric motor and the clutch during vehicle operation, the electronic control system comprising: a reinforcement learning decision module for generating control instructions based on vehicle state information, the control instructions including power control instructions for controlling power output of the engine and the electric motor, and driving mode instructions for controlling connection state of the clutch and start-stop state of the engine; an electric motor control agent unit being a control strategy network trained based on a deep reinforcement learning algorithm, taking a unified state vector as input and outputting power control instructions for controlling power of the electric motor, optimization objectives including energy use efficiency improvement and battery safety maintenance; the electric motor control agent unit comprising: The reward fusion subunit is configured to calculate a joint reward signal according to a set reward coupling factor The instant reward values of the engine control intelligent agent unit and the motor control intelligent agent unit are weighted to generate a joint reward signal, wherein: when no reward sharing is performed; when partial reward sharing is performed; and when the joint optimization calculation is performed. a conflict coordination sub-unit for adjusting control instructions based on preset priority, weight function or fusion rule when power control instructions or driving mode instructions output by the engine control agent unit and the electric motor control agent unit respectively are inconsistent; a mode control module for controlling connection or disconnection state of the clutch and start or stop state of the engine according to the driving mode instructions, so as to switch running state of the vehicle between pure electric driving, series driving and parallel driving modes; an execution control module for controlling the engine and the electric motor to run according to the power control instructions, so as to drive the vehicle to output target torque and speed; The parameter setting unit is used for setting the proportional gain and integral gain parameters of the speed outer loop adjusting unit and the current inner loop adjusting unit in the magnetic field oriented control structure , The offline setting is performed by using a non-dominated sorting genetic algorithm. a safety control module for judging whether abnormal control conditions are met based on vehicle state information and execution feedback of the control instructions, and implementing preset safety control strategies including output limitation, mode switching or control takeover when the conditions are met; a control takeover unit for replacing the reinforcement learning decision module to control the execution control module after receiving an abnormal state judgment signal; a strategy management module for adjusting parameters of the safety control strategies or switching control strategy models when running environment parameters change, so as to maintain control performance of the electronic control system under different running conditions; and a strategy switching judgment unit for judging whether strategy switching conditions are met based on battery state of charge (SOC), battery capacity change and environmental temperature parameters monitored during vehicle operation, and generating a strategy switching control signal when the conditions are met.

2. A multi-in-one electric drive assembly system according to claim 1, wherein, the reinforcement learning decision module comprising: a state input processing unit for collecting and processing vehicle running state information including but not limited to vehicle load, driving demand, battery state of charge (SOC), engine speed, electric motor efficiency and current driving mode, for constructing a unified state vector. The engine control agent unit is a control strategy network trained based on a deep reinforcement learning algorithm, takes a state vector as input, and outputs a power control instruction for controlling the power of the engine, and the optimization target includes minimizing fuel consumption rate and optimizing output response performance; The driving mode decision unit is configured to generate driving mode instructions for controlling the clutch connection state and the engine start-stop state based on the unified state vector and the agent output result; The agent cooperative optimization unit is configured to establish a shared state interface and a reward coupling mechanism between the engine control agent unit and the motor control agent unit, and to coordinate the consistency of control instructions through joint optimization of strategy output.

3. A multi-in-one electric drive assembly system according to claim 1, wherein, The motor control agent unit includes: A state sharing subunit configured to provide the unified state vector to the engine control agent unit and the motor control agent unit as common input; A fusion strategy output subunit configured to output the power control instruction and the driving mode instruction processed by the conflict coordination subunit to the driving mode control module and the execution control module, respectively.

4. The all-in-one electric drive assembly system of claim 1, wherein, The driving mode decision unit includes: A state discrimination subunit configured to identify the operating scenario based on key parameters in the state vector, including the current vehicle speed, acceleration demand, battery state of charge, engine start-stop state, current power request, and system efficiency evaluation index; A mode switching determination subunit configured to determine the optimal driving mode among pure electric driving, series driving, and parallel driving based on the operating scenario identification result and the target optimization strategy, and the determination process is based on a strategy function and is provided with a hysteresis mechanism to avoid frequent switching; A clutch control instruction generation subunit configured to generate a control instruction for controlling the clutch engagement state based on the mode determination result; An engine start-stop control instruction generation subunit configured to generate an engine start-stop control signal based on the driving mode determination result; A safety redundancy determination subunit configured to monitor the execution result of the driving mode switching instruction in real time, and if mode switching abnormalities, execution failures, or system safety risks are detected, a pre-set fault mode or degradation control strategy is triggered.

5. A multi-in-one electric drive assembly system according to claim 4, wherein, The execution control module includes a motor control unit configured to control the torque output of the motor, and the motor control unit adopts a field-oriented control structure, including: Speed outer loop regulation unit, input is target speed Difference with actual speed Output is target torque current Regulated by proportional-integral controller Current inner loop regulating unit, input is target current deviation between actual current output is voltage control signal or PWM duty cycle, used to drive three-phase inverter; The control structure adopts The coordinate transformation mode transforms three-phase static coordinate system current to a rotating coordinate system. The speed outer loop regulation unit and the current inner loop regulation unit constitute an inner-outer loop cascade PI control structure, which respectively completes speed command calculation and current control instruction adjustment.

6. A multi-in-one electric drive assembly system according to claim 5, wherein, The non-dominated sorting genetic algorithm optimization target is the following two functions: ； wherein: is the actual torque of the first electric motor; is the average value of the torque over the optimization period for calculating the steady-state torque fluctuation index; a torque feedback adjusting unit, configured to collect the output torque of the first motor in real time during operation of the vehicle and compare with a preset torque fluctuation threshold The feedback adjustment unit fine-tunes the target current based on the current deviation value when the fluctuation amplitude exceeds the threshold value , or gain correction is made to the controller parameters , ; represents an integral with respect to time ; for evaluating the speed tracking error integral of the control system over a time interval ; for measuring the variance of the pulsation of the motor output torque over this interval.

7. A multi-in-one electric drive assembly system according to claim 1, wherein, The safety control module includes: An abnormality determination unit configured to obtain the vehicle speed, battery state of charge SOC, motor temperature, and working current in the vehicle operating state, and set safety thresholds for the above variables; when the detected value exceeds the safety threshold, an abnormal state determination signal is generated; A control takeover unit configured to replace the reinforcement learning decision module to control the execution control module and perform at least one of the following control operations after receiving the abnormal state determination signal: a torque limit subunit configured to output a motor torque limit instruction to the execution control module; a clutch disconnection subunit configured to output an engine clutch disconnection control instruction to the drive mode control module; a PID control switching subunit configured to call a preset PID regulator-based power control logic and send an output signal to the execution control module.

8. The all-in-one electric drive assembly system of claim 1, wherein, The strategy management module comprises: a backup strategy library configured with a plurality of control strategy networks obtained through offline training, each strategy network corresponding to a different combination of vehicle operating environment parameters; a strategy switching execution unit configured to load a target strategy network from the backup strategy library and replace the current main strategy network according to the strategy switching control signal, and to continuously adjust the control parameters of the original main strategy network within a limited range during the replacement process.

9. A method of controlling a multi-in-one electric drive assembly system as claimed in any one of claims 1-8, characterized in that, The method comprises the following steps: Step 1: Collecting vehicle operating state information, including vehicle speed, acceleration request, battery state of charge (SOC), engine speed, motor efficiency parameters, and current drive mode state; Step 2: Based on the operating state information, generating control instructions through reinforcement learning decision-making strategy, including engine power instruction, motor power instruction, and drive mode switching instruction; Step 3: According to the drive mode switching instruction, controlling the engagement or disengagement state of the clutch and the start or stop state of the engine, realizing the switching between pure electric drive, series drive, and parallel drive modes; Step 4: According to the engine power instruction and motor power instruction, controlling the engine throttle opening and motor current vector respectively, realizing the output of vehicle target torque and speed; Step 5: Monitoring key parameters during vehicle operation, including vehicle speed, battery SOC, motor temperature, and current, and according to the control instruction, executing feedback to determine whether there is an abnormal control condition, if so, triggering a preset safety control strategy; Step 6: When detecting changes in operating environment parameters, adjusting the control parameters of the safety control strategy, or switching the pre-set control strategy model, to ensure the control performance and stability of the electronic control system under different operating conditions.

Images