Throttle valve intelligent control method and system based on Internet of Things

Abstract

The invention relates to the technical field of Internet of Things and automobile engine control, and discloses an intelligent throttle valve control method and system based on the Internet of Things, and the method comprises the steps: obtaining and processing sensor data, and obtaining a real-time driving state recognition result; determining a computing resource allocation demand according to the real-time driving state recognition result, and adjusting the power of an edge processor according to the computing resource allocation demand to obtain an adjusted power level; according to the adjusted power level, fusing real-time data, calculating a throttle valve opening degree driving signal and generating a throttle valve actuator instruction; the execution response time of the instruction is monitored, the power is adjusted again according to the response time, and finally balanced throttle valve control signals are obtained; and determining an updated driving state recognition result according to the finally balanced throttle valve control signal. According to the method, a feed-forward and feedback combined double-layer dynamic power regulation mechanism is constructed, so that the problem that control precision and energy efficiency are difficult to consider under complex working conditions in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the fields of Internet of Things (IoT) technology and automotive engine control technology, and in particular to an IoT-based intelligent throttle control method and system. Background Technology

[0002] Currently, the Internet of Things (IoT) technology is being used more and more widely in the field of automotive engine control. Through real-time data acquisition and processing, it can significantly improve vehicle performance and energy efficiency, and is an important component of advanced combustion optimization technology.

[0003] In a current technology, engine throttle control, a crucial element affecting combustion efficiency, often relies on edge processors for computation and signal generation. However, its power regulation and computational resource allocation strategies are relatively fixed. This approach struggles to achieve a precise balance between power consumption and computational performance across different driving scenarios. For example, during rapid acceleration, if the processor fails to increase power promptly, computational delays may cause throttle opening response lag, impacting vehicle power output. Conversely, excessive power at idle can lead to unnecessary energy consumption. Therefore, existing methods, unable to dynamically adjust resources based on operating conditions, struggle to achieve both control precision and efficiency when rapidly switching driving states.

[0004] In summary, existing technologies suffer from the technical problem of failing to balance the precision and efficiency of throttle control. Summary of the Invention

[0005] This invention provides an intelligent throttle control method and system based on the Internet of Things, in order to solve the technical problem that it is difficult to balance the accuracy and efficiency of throttle control in the prior art.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a throttle intelligent control method based on the Internet of Things, comprising: Acquire sensor data and process it to obtain real-time driving status recognition results; Based on the real-time driving status recognition results, the computational resource allocation requirements are determined using preset threshold rules; Based on the computing resource allocation requirements, the power of the edge processor is adjusted to obtain the adjusted power level; Based on the adjusted power level and by integrating real-time monitoring data, the throttle opening drive signal is calculated, and the throttle actuator command is generated. The execution response time of the throttle actuator command is monitored, and the power of the edge processor is adjusted again based on the execution response time to obtain a final balanced throttle control signal; Based on the final balanced throttle control signal, the latest sensor data is preprocessed and classified to determine the updated driving state recognition result.

[0007] Preferably, the step of acquiring and processing sensor data to obtain real-time driving state recognition results includes: Acquire raw data streams of engine speed and accelerator pedal position; The original data stream is timestamped to obtain a synchronized data stream; The synchronous data stream is filtered and denoised to obtain a smooth data sequence; The data in the smoothed data sequence are normalized and weighted to obtain weighted feature values, which are then used as the real-time driving state recognition result.

[0008] Preferably, determining the computational resource allocation requirements based on the real-time driving state recognition results using a preset threshold rule includes: If the real-time driving status recognition result exceeds the preset dynamic scene threshold, an indication that computing resource allocation needs to be increased is generated as the computing resource allocation requirement.

[0009] Preferably, adjusting the edge processor power according to the computing resource allocation requirements to obtain the adjusted power level includes: The voltage and frequency parameters of the edge processor are adjusted using a dynamic voltage-frequency scaling method to obtain the adjusted voltage-frequency parameters. Based on the adjusted voltage frequency parameters, the power output of the edge processor is reconfigured to obtain the adjusted power level.

[0010] Preferably, the step of calculating the throttle opening drive signal and generating throttle actuator commands based on the adjusted power level and real-time monitoring data includes: Obtain the current energy consumption index from the adjusted power level; If the current energy consumption index is lower than the preset energy consumption threshold, then the engine speed data and accelerator pedal position data are acquired and fused to obtain the driving intensity index. The driving intensity index is input into a pre-trained logistic regression model to obtain the probability value of resource enhancement needs; Based on the comparison between the probability value and the preset scheduling threshold, one of the multiple preset scheduling schemes is selected as the optimized resource scheduling scheme. Based on the optimized resource scheduling scheme, and using a pre-established multi-dimensional mapping table that defines the relationship between sensor data and throttle opening, the throttle opening drive signal is calculated. The throttle opening drive signal is modulated to obtain the throttle actuator command.

[0011] Preferably, the step of monitoring the execution response time of the throttle actuator command and readjusting the edge processor power based on the execution response time to obtain a final balanced throttle control signal includes: The execution response time of the throttle actuator command is continuously monitored to obtain a response time dataset; If any response time in the response time dataset exceeds a preset response time threshold, the power output is recalculated to obtain a new power target value, which is then used as the first power output dataset. Adjust the performance level of the edge processor based on the first power output dataset; Under the adjusted performance level, the initial throttle control signal is recalculated based on the latest real-time sensor data; The initial throttle control signal is fused with the actual throttle opening feedback value from the previous moment to generate the final balanced throttle control signal.

[0012] Preferably, the step of preprocessing and classifying the latest sensor data based on the final balanced throttle control signal to determine the updated driving state recognition result includes: Based on the final balanced throttle control signal, acquire sensor data; The sensor data is subjected to signal noise verification to obtain the verification result; Based on the verification results, the sensor data is subjected to conditional filtering to obtain purified sensor data; The purified sensor data is normalized to obtain the data to be classified. The data to be classified is input into a pre-established driving state model for classification processing to obtain the updated driving state recognition result.

[0013] Secondly, the present invention provides an Internet of Things-based intelligent throttle control system, comprising: The status recognition module is used to acquire and process sensor data to obtain real-time driving status recognition results; The requirement determination module is used to determine the computing resource allocation requirements based on the real-time driving status recognition results and using preset threshold rules. The power adjustment module is used to adjust the power of the edge processor according to the computing resource allocation requirements to obtain the adjusted power level; The instruction generation module is used to calculate the throttle opening drive signal and generate throttle actuator instructions based on the adjusted power level and real-time monitoring data. The feedback control module is used to monitor the execution response time of the throttle actuator command and adjust the edge processor power again according to the execution response time to obtain the final balanced throttle control signal. The status update module is used to preprocess and classify the latest sensor data based on the final balanced throttle control signal to determine the updated driving status recognition result.

[0014] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the Internet of Things-based intelligent throttle control method described in any one of the above.

[0015] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the IoT-based intelligent throttle control method described above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention integrates multi-dimensional sensor data such as engine speed and accelerator pedal position, and uses signal filtering to remove noise, which can accurately identify the real-time driving status of the vehicle under different operating conditions such as idling, acceleration or cruising. This multi-dimensional and accurate perception process solves the problem that the existing technology relies on only a single feedback and cannot fully grasp the dynamic operating conditions, and provides a scientific and reliable data foundation for subsequent intelligent control.

[0017] (2) By constructing a feedforward control path of "state recognition - demand judgment - power adjustment", this invention can adjust the performance of the edge processor in advance in high dynamic scenarios by using a dynamic voltage and frequency scaling method. This predictive resource allocation mechanism derives and solves the response lag problem caused by fixed control parameters in existing technologies, thereby improving the active adjustment capability and response speed of the control system.

[0018] (3) This invention establishes a closed-loop feedback adjustment mechanism based on the actuator instruction response time, and performs a secondary dynamic adjustment of the processor power after the initial adjustment, ultimately generating a balanced control signal. This method transforms the control strategy from passive open-loop execution to active closed-loop optimization, ensuring the control accuracy and energy consumption balance of the system under complex operating conditions, and improving the operating efficiency of the entire throttle control system. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of a throttle intelligent control method based on the Internet of Things provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of a throttle intelligent control system based on the Internet of Things provided in the second embodiment of the present invention. Detailed Implementation

[0020] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1 The first embodiment of the present invention provides a throttle intelligent control method based on the Internet of Things, comprising the following steps: S11: Acquire sensor data and process it to obtain real-time driving status recognition results; S12, Based on the real-time driving state recognition results, determine the computing resource allocation requirements using preset threshold rules; S13, adjust the edge processor power according to the computing resource allocation requirements to obtain the adjusted power level; S14, based on the adjusted power level and by integrating real-time monitoring data, calculate the throttle opening drive signal and generate throttle actuator instructions; S15, monitor the execution response time of the throttle actuator command, and adjust the edge processor power again according to the execution response time to obtain the final balanced throttle control signal; S16, based on the final balanced throttle control signal, preprocess and classify the latest sensor data to determine the updated driving state recognition result.

[0022] In step S11, sensor data is acquired and processed to obtain real-time driving state recognition results, including: Acquire raw data streams of engine speed and accelerator pedal position; The original data stream is timestamped to obtain a synchronized data stream; The synchronous data stream is filtered and denoised to obtain a smooth data sequence; The data in the smoothed data sequence are normalized and weighted to obtain weighted feature values, which are then used as the real-time driving state recognition result.

[0023] In one implementation, the raw data stream is formed by collecting raw data of engine speed and raw data of accelerator pedal position, respectively, through a speed sensor deployed on the engine and a position sensor installed on the accelerator pedal.

[0024] It should be noted that due to differences in sensor hardware, the timestamps of each data point in the raw data stream may have millisecond-level deviations. To ensure that subsequent analysis is based on a consistent time reference, the system performs timestamp alignment processing on the raw data stream, merging the speed data and pedal data according to a unified time axis to obtain the synchronized data stream.

[0025] It is worth noting that, in order to eliminate high-frequency noise generated by engine vibration or signal interference in the synchronization data stream, the system uses a Kalman filter algorithm to filter the synchronization data stream.

[0026] For example, raw engine speed data may fluctuate by ±50 RPM due to vibration. After Kalman filtering, the fluctuation can be reduced to ±10 RPM, resulting in more stable data. In another implementation, a low-pass filtering method can be combined to further eliminate high-frequency interference and obtain the smoothed data sequence.

[0027] In one implementation, before data fusion, the system first normalizes the engine speed and accelerator pedal position data in the smoothed data sequence, mapping the values ​​of different physical units to a unified dimensionless interval; then, by weighting the normalized engine speed and accelerator pedal position, a weighted feature value is obtained.

[0028] It should be noted that the normalization process can be implemented using a min-max normalization method, that is, the original data is linearly mapped to the [0,1] interval based on preset maximum and minimum values. The maximum and minimum values ​​are predetermined based on the design specifications and physical operating limits of the engine or sensor.

[0029] It should be noted that the weight values ​​used in this weighted calculation are determined based on offline optimization of a historical dataset containing various known driving states (e.g., idling, smooth driving, rapid acceleration). For example, this optimization process includes: setting multiple candidate weight combinations, applying these weight combinations to calculate feature values ​​and classify states on the historical dataset, and calculating the classification accuracy under each weight group; finally, selecting the weight group that achieves the highest classification accuracy (e.g., engine speed weight 0.6, accelerator pedal position weight 0.4) as the final configuration.

[0030] It is worth noting that the calculation process of the weighted feature value is as follows: multiply the normalized engine speed value by its corresponding speed weight to obtain the first product; multiply the normalized accelerator pedal position value by its corresponding pedal position weight to obtain the second product; add the first product and the second product to obtain the weighted feature value.

[0031] For example, assuming the preset maximum engine speed is 8000 RPM and the preset minimum is 0 RPM, and the preset maximum accelerator pedal opening is 100% and the preset minimum is 0%, if the current engine speed is 6000 RPM (normalized value 0.75) and the pedal opening is 30% (normalized value 0.3), then the calculated dimensionless weighted feature value is 0.75 × 0.6 + 0.3 × 0.4 = 0.57. This calculated weighted feature value is used as the real-time driving state recognition result.

[0032] In step S12, based on the real-time driving state recognition result, the computational resource allocation requirements are determined using a preset threshold rule, including: If the real-time driving status recognition result exceeds the preset dynamic scene threshold, an indication that computing resource allocation needs to be increased is generated as the computing resource allocation requirement.

[0033] In one implementation, the real-time driving state recognition result is compared with a preset dynamic scene threshold. If the value exceeds the dynamic scene threshold, it indicates that the current driving scene is highly dynamic, and the system will then generate an indication that computing resource allocation needs to be increased; this indication is the computing resource allocation requirement.

[0034] It should be noted that the determination of the dynamic scene threshold is based on statistical analysis of real-time driving state recognition results data under a large number of historical high-dynamic driving scenarios (such as rapid acceleration and emergency avoidance). This analysis process includes calculating the probability density distribution of recognition results under all historical high-dynamic scenarios and selecting the quantile that can cover the 95% confidence interval as the preset dynamic scene threshold.

[0035] In step S13, the edge processor power is adjusted according to the computing resource allocation requirements to obtain the adjusted power level, including: The voltage and frequency parameters of the edge processor are adjusted using a dynamic voltage-frequency scaling method to obtain the adjusted voltage-frequency parameters. Based on the adjusted voltage frequency parameters, the power output of the edge processor is reconfigured to obtain the adjusted power level.

[0036] In one implementation, when the system receives the computing resource allocation request generated in S12, it will employ a Dynamic Voltage-Frequency Scaling (DVFS) method to adjust the core voltage and operating frequency parameters of the edge processor. It should be noted that Dynamic Voltage-Frequency Scaling is a control technique used to dynamically balance processor power consumption and computing performance. This method relies on a preset voltage-frequency mapping table, which defines multiple stable and reliable performance levels, each corresponding to a specific voltage and frequency combination.

[0037] It is worth noting that the voltage-frequency mapping table is usually provided by the edge processor manufacturer or determined through experimental calibration such as stress testing and power consumption analysis of the processor, ensuring the system's operational stability at each performance level.

[0038] For example, the voltage-frequency mapping table can define a "standard performance level" (corresponding to 1.2V voltage and 1GHz frequency) and a "high-performance level" (corresponding to 1.4V voltage and 1.5GHz frequency). When an instruction is received to increase the allocation of computing resources, the system selects the "high-performance level" from the mapping table and controls the processor to switch to running at 1.4V voltage and 1.5GHz frequency, thus obtaining the adjusted voltage-frequency parameters.

[0039] In one implementation, after obtaining the adjusted voltage frequency parameters, the system will reconfigure the power output of the edge processor through the power management module based on the parameters to obtain the adjusted power level, which can support faster sensor data processing and decision response.

[0040] In step S14, based on the adjusted power level and by integrating real-time monitoring data, the throttle opening drive signal is calculated, and a throttle actuator command is generated, including: Obtain the current energy consumption index from the adjusted power level; If the current energy consumption index is lower than the preset energy consumption threshold, then the engine speed data and accelerator pedal position data are acquired and fused to obtain the driving intensity index. The driving intensity index is input into a pre-trained logistic regression model to obtain the probability value of resource enhancement needs; Based on the comparison between the probability value and the preset scheduling threshold, one of the multiple preset scheduling schemes is selected as the optimized resource scheduling scheme. Based on the optimized resource scheduling scheme, and using a pre-established multi-dimensional mapping table that defines the relationship between sensor data and throttle opening, the throttle opening drive signal is calculated. The throttle opening drive signal is modulated to obtain the throttle actuator command.

[0041] In one implementation, the system uses a real-time monitoring module to collect the actual power consumption data of the edge processor from the adjusted power level obtained in S13 using an energy consumption sensor deployed on the power supply line of the edge processor. This data is the current energy consumption index.

[0042] It should be noted that the system compares the current energy consumption index with a preset energy consumption threshold. If the current energy consumption index is lower than the threshold, it indicates that the processor is currently in an energy-saving state, and the system will trigger subsequent resource scheduling optimization processes. The preset energy consumption threshold is obtained based on cluster analysis of the processor's historical power consumption data. This determination process includes: collecting continuous power consumption data points of the processor under various historical load conditions; using the K-means clustering algorithm to divide these data points into at least two clusters, corresponding to a "low-power state cluster" and a "high-power state cluster," respectively; and determining the boundary value between these two clusters as the preset energy consumption threshold. This threshold represents the critical point for the processor to transition from an energy-saving state to a higher-performance state.

[0043] It is worth noting that after triggering the optimization process, the system will again acquire engine speed data and accelerator pedal position data, and use the same normalization method and weight values ​​as in step S11 to perform weighted summation on the data to calculate a comprehensive driving intensity index. Subsequently, the system uses a pre-trained logistic regression model to analyze the driving intensity index and determine whether resources need to be increased.

[0044] It should be noted that the structure of the logistic regression model includes an input layer (receiving the driving intensity index), a linear layer with weights and bias parameters, and an output layer with a sigmoid activation function. Its training process includes: acquiring a training dataset containing a large number of historical "driving intensity indices" and corresponding "resource requirement labels" (0 representing standard requirements, 1 representing high-performance requirements); using supervised learning with binary cross-entropy as the loss function, iteratively training the model's weights and bias parameters through optimization algorithms such as gradient descent; when the model's loss value on the independent validation set decreases by less than a preset convergence threshold over several consecutive periods, the model is considered converged, and training is complete. It is worth noting that the convergence threshold is determined through grid search experiments. The experimental procedure includes: predefining a set of candidate convergence thresholds, such as {1e-3, 1e-4, 1e-5, 1e-6}; for each candidate threshold, using that threshold as the convergence criterion, performing multiple cross-validation training sessions on the training dataset; recording the average training time required for the model to converge at each candidate threshold, and the average classification accuracy of the model on the validation set after convergence; and selecting from all candidate thresholds the threshold that allows the model to achieve the highest classification accuracy within an acceptable training time (e.g., 1e-5) as the final preset convergence threshold.

[0045] In one implementation, the system inputs the real-time calculated driving intensity index into the logistic regression model, and the model outputs a probability value representing the demand for resource enhancement. The system compares this probability value with a preset scheduling threshold. The scheduling threshold is determined through ROC curve analysis on the model validation set, and its selection principle is to choose the probability value that minimizes the false positive rate while ensuring a true positive rate of not less than 98%.

[0046] It is worth noting that if the probability value exceeds the scheduling threshold, the system selects a preset "high-performance" resource scheduling scheme; otherwise, it selects a "standard" resource scheduling scheme. The parameters of these preset schemes are determined through a systematic performance calibration experiment.

[0047] It should be noted that the calibration experiment process includes: defining key performance indicators (KPIs) and their target thresholds for different driving modes (e.g., standard and high performance). For example, in high performance mode, the end-to-end latency of throttle control must be less than 5 milliseconds. Simulated loads are run using benchmark testing programs under different combinations of resource parameters (e.g., different CPU priorities and sensor sampling frequencies), and the actual performance indicators (e.g., average latency) and power consumption for each combination are recorded. The resource parameter combination that achieves the lowest power consumption while meeting the performance requirements of each mode is selected and fixed as the corresponding preset scheme. These schemes are stored in the form of configuration files or parameter tables, which explicitly define the allocation strategy of various resources under different schemes. For example, the standard scheme might configure the CPU priority as 5 and the sensor sampling frequency as 100Hz; while the high performance scheme might configure the CPU priority as 10 and the sensor sampling frequency as 500Hz. The selected scheme is the optimized resource scheduling scheme.

[0048] In one implementation, the system calculates the target throttle opening based on the optimized resource scheduling scheme. This calculation is based on a multi-dimensional mapping table pre-calibrated through engine bench testing. The calibration process includes: mounting the engine on a dynamometer and traversing all stable operating points within a test matrix containing multiple engine speed and accelerator pedal position combinations; at each operating point, traversing a preset throttle opening search range with a preset small step size (e.g., 0.1 degrees) and recording the engine output torque at each opening in real time; determining the opening point where the output torque reaches its maximum value as the optimal target opening value for that operating condition; and recording the correspondence between all operating points (inputs: engine speed, accelerator pedal position) and the optimal target opening (output) to form a mapping table. Simultaneously, independent mapping table regions are calibrated for different modes such as standard and high-performance. It should be noted that the preset throttle opening search range is set based on the physical travel limit of the throttle actuator and the effective response range determined by the engine bench test. The preset microstep (e.g., 0.1 degrees) is selected based on the minimum control resolution of the throttle actuator.

[0049] It should be noted that, to adapt to the control requirements of the actuator, the system can use pulse width modulation (PWM) to modulate the drive signal. This modulation controls the signal strength by adjusting the pulse width of the signal.

[0050] For example, if the final calculated target opening is 40.2%, the signal generator can generate a corresponding 2.5V voltage signal. Subsequently, the pulse width modulation module modulates this signal with a 50% duty cycle to generate the final drive signal. The drive signal is then processed by the actuator control module. This processing includes encapsulating parameters such as the duty cycle (50%) of the drive signal into a digital message conforming to the Controller Area Network (CAN) bus protocol. This digital message is the throttle actuator command.

[0051] In step S15, the execution response time of the throttle actuator command is monitored, and the edge processor power is adjusted again based on the execution response time to obtain a final balanced throttle control signal, including: The execution response time of the throttle actuator command is continuously monitored to obtain a response time dataset; If any response time in the response time dataset exceeds a preset response time threshold, the power output is recalculated to obtain a new power target value, which is then used as the first power output dataset. Adjust the performance level of the edge processor based on the first power output dataset; Under the adjusted performance level, the initial throttle control signal is recalculated based on the latest real-time sensor data; The initial throttle control signal is fused with the actual throttle opening feedback value from the previous moment to generate the final balanced throttle control signal.

[0052] In one implementation, the system continuously monitors the throttle actuator command issued by S14 using a high-precision timer. This monitoring process records the time difference between the command issuance and the completion of the actuator's physical action, and stores this time difference as the response time in a response time dataset.

[0053] It should be noted that the system compares the response times recorded in the response time dataset with a preset response time threshold. If the response time exceeds the threshold, it indicates that the current power allocation may not be able to meet the real-time load demand (for example, the engine load suddenly increases when the vehicle is going uphill), and the system then determines a power output adjustment requirement.

[0054] It is worth noting that the preset response time threshold is determined based on statistical analysis of historical response time data of the system under different loads. This process includes: collecting the response time distribution of the system under multiple known load levels; analyzing the average response time and standard deviation under each load level; and selecting a critical value that can distinguish between normal response and significantly delayed response (e.g., the average value plus three times the standard deviation) as the preset response time threshold.

[0055] In one implementation, after determining the power output adjustment requirement, the system employs a power allocation module to recalculate the power output using an incremental PI controller algorithm. It should be noted that this algorithm is a well-known technique for dynamic error correction in industrial control. Its calculation process is as follows: the excess response time is used as the error input; the power adjustment increment is calculated according to the following exemplary discrete PI control formula: ; in, Representative in the present The power adjustment increment calculated at any given time; This represents a preset scaling factor; Represents the preset integral coefficient; Representative in the present The error in time is the "actual response time" minus the "preset response time threshold"; Represents the previous moment ( Error in time (time).

[0056] The current power output is added to the power adjustment increment to obtain a new power target value, which is the first power output dataset. It should be noted that the proportional coefficient Kp and integral coefficient Ki are predetermined based on the dynamic characteristics of the controlled system using experimental methods known in the art, such as the Ziegler-Nichols tuning method.

[0057] For example, if the current power level of the edge processor is 5W, but the response time exceeds the limit, the incremental PI controller calculates the power adjustment increment based on the error as +0.5W, then the system determines the new power target value as 5.5W, which is the first power output dataset.

[0058] It should be noted that the system first applies the first power output dataset (i.e., the new power target value) to the edge processor, and then adjusts the processor's computing performance to match the new target value through power management techniques such as Dynamic Voltage Frequency Scaling (DVFS).

[0059] At this enhanced performance level, the system recalculates a new, ideal throttle opening command based on real-time feedback data (such as the current engine speed and the driver's accelerator pedal request) and using the multidimensional mapping table described in S14.

[0060] To ensure a smooth transition of the control signal and avoid vehicle swaying caused by sudden changes in commands, the system further integrates the newly calculated ideal throttle opening command with the actual throttle opening value fed back from the previous control cycle using a data fusion algorithm to generate the final balanced throttle control signal. This fusion process can employ a weighted average method, calculated by multiplying the newly calculated ideal command and the feedback actual opening value by their respective weights and then summing the results. The weight values ​​are determined based on offline sensitivity analysis experiments. These experiments measure the impact of varying weight ratios on the stability and response speed of the final control signal in a simulation environment; features with higher impact are assigned higher weights. For example, the weight of the newly calculated command can be set to 0.7, and the weight of the historical actual opening feedback can be set to 0.3, thereby ensuring smooth control while maintaining a fast response.

[0061] In step S16, based on the final balanced throttle control signal, the latest sensor data is preprocessed and classified to determine the updated driving state recognition result, including: Based on the final balanced throttle control signal, acquire sensor data; The sensor data is subjected to signal noise verification to obtain the verification result; Based on the verification results, the sensor data is subjected to conditional filtering to obtain purified sensor data; The purified sensor data is normalized to obtain the data to be classified. The data to be classified is input into a pre-established driving state model for classification processing to obtain the updated driving state recognition result.

[0062] In one implementation, after the final balanced throttle control signal is generated and sent out in S15, the system will again obtain the latest real-time sensor data through the sensor network, which reflects the system state after the control action is executed.

[0063] It should be noted that the system performs signal-noise verification on the latest acquired sensor data. This verification process includes calculating the standard deviation of the data sequence over a period of time, and using this standard deviation as a quantification indicator of signal-noise to obtain the verification result.

[0064] In one implementation, the system compares the verification result with a preset signal-noise threshold. If the verification result exceeds the threshold, conditional filtering is performed on the sensor data; otherwise, the original sensor data is directly used as the data to be classified.

[0065] It is worth noting that the determination of the signal noise threshold is based on statistical analysis of a large amount of historical sensor data under normal operating conditions. This analysis process includes calculating the probability density distribution of signal noise indicators under all historical normal operating conditions, and selecting the upper limit value that can cover the 95% confidence interval as the preset signal noise threshold.

[0066] It should be noted that the conditional filtering process can be implemented using a Kalman filter. This filter, through prediction and update steps, combines historical data and current measurements to smooth data containing excessive noise, generating a more stable dataset with lower noise levels. This dataset is the data to be classified.

[0067] It should be noted that before inputting the data to be classified into the driving state model, the system first normalizes it; then, the normalized data is input into a pre-established driving state model, which performs classification processing to obtain the updated driving state recognition result.

[0068] It should be noted that the normalization process is implemented using the same min-max normalization method as in step S11. That is, based on the preset maximum and minimum values, the data after conditional filtering is linearly mapped to the [0,1] interval to ensure that the scale of the data input to the model is completely consistent with the scale of the data used during model training.

[0069] It is worth noting that the driving state model is a classification model built based on the Support Vector Machine (SVM) algorithm. Its training process is as follows: A training dataset is obtained containing a large amount of normalized historical sensor data (as input features) and corresponding real driving state labels (such as "idling," "cruising," etc., as output) annotated by experts; the SVM model is trained using supervised learning by finding a hyperplane that maximizes the margin between samples of different classes; training terminates when the model's classification accuracy on the independent validation set improves by less than a preset convergence threshold (e.g., 0.01%) over several consecutive training epochs. It should be noted that this convergence threshold is obtained through a grid search experiment. The experimental procedure includes: predefining a set of candidate convergence thresholds, such as {0.1%, 0.01%, 0.001%}; for each candidate threshold, using that threshold as the stopping condition for training, performing multiple cross-validation training sessions on the training dataset; recording the average training time required for the model to converge at each candidate threshold, and the average classification accuracy of the model on the validation set after convergence; and selecting from all candidate thresholds the threshold that allows the model to achieve the highest classification accuracy within an acceptable training time (e.g., 0.01%) as the final preset convergence threshold.

[0070] It should be noted that the key hyperparameters of this model, such as the choice of kernel function (e.g., radial basis function (RBF) kernel) and its corresponding parameters (e.g., C and gamma), were determined through a systematic grid search and cross-validation experiment. The experimental process included: defining a candidate value grid for each hyperparameter to be optimized (e.g., C and gamma); for each parameter combination in the grid, training and evaluating the model using k-fold cross-validation on the training dataset; recording the average classification accuracy of each parameter combination across all validation folds; and selecting the parameter combination that achieves the highest average classification accuracy for training the final SVM model.

[0071] In practical use, the system takes the latest data to be classified (after normalization) as input, and the trained SVM model outputs a final driving state recognition result after closed-loop correction.

[0072] To facilitate understanding of the present invention, some preferred embodiments of the present invention will be described in further detail below.

[0073] In one implementation, the method of the present invention can be applied to an intelligent connected commercial vehicle equipped with an edge computing unit to achieve precise and energy-saving control of the engine throttle valve. A complete workflow example is as follows: Step 1: The vehicle drives smoothly on urban roads, and the system continuously acquires raw data streams of engine speed (e.g., 2000 RPM) and accelerator pedal position (e.g., 20%). After timestamp alignment, Kalman filtering, and min-max normalization, normalized engine speed values ​​of 0.25 and pedal position values ​​of 0.2 are obtained. The system calculates a weighted feature value of 0.23 based on preset weights and uses this value as the real-time driving status recognition result.

[0074] Step two: The system compares the identification result 0.23 with preset state type thresholds (e.g., idling threshold 0.2, cruise threshold 0.5) and classifies it as "cruise state". At the same time, since the value does not exceed the preset dynamic scenario threshold (e.g., 0.6), the system determines that there is no need to increase computing resources and generates an instruction to maintain the current resource allocation.

[0075] Step 3: Since no resource upgrade is required, the edge processor maintains a standard energy-saving power level (e.g., 1.2V, 1GHz). The energy consumption index obtained by the system from this power level is lower than the preset energy consumption threshold, indicating that there is surplus computing power. The optimization process is then triggered, selecting the "standard" resource scheduling scheme, and calculating the corresponding throttle opening command based on the mapping table calibrated on the engine test bench.

[0076] Step four: An obstacle appears in front of the vehicle, and the driver urgently presses the accelerator to overtake, the pedal position quickly changing to 90%. The new feature value calculated by the system in step one jumps to 0.75. Step two then determines that this is a high-dynamic scenario and generates an instruction to increase resources.

[0077] Step five: Upon receiving the instruction, the power regulation module immediately adjusts the processor to the "high-performance level" (e.g., 1.4V, 1.5GHz) via DVFS. However, due to the sharp increase in engine load, the throttle command issued by the instruction generation module, monitored by the feedback control module, has an actual execution response time of 0.12 seconds, exceeding the preset 0.1-second response threshold. The feedback control module is then triggered, calculating the power adjustment increment using an incremental PI controller algorithm, further boosting the power, and integrating real-time feedback data to generate a final balanced throttle control signal, ensuring the throttle responds as quickly as possible.

[0078] Step six: After the overtaking maneuver is completed, the state update module updates the vehicle's driving state to "smooth driving" by inputting the latest sensor data, which has undergone noise verification and normalization, into the pre-trained SVM model.

[0079] In another implementation, the system deployment architecture of this invention can be configured differently. For example, in a large fleet management scenario, the sensor data can be uploaded to a cloud server in real time via an onboard 5G module. All complex model training, state recognition, and power adjustment decisions are completed in the cloud, and the calculated final throttle control signal is then sent to the vehicle's local controller for execution via the network.

[0080] In summary, this invention discloses an intelligent throttle control method and system based on the Internet of Things (IoT). The method includes: acquiring and processing sensor data to obtain a real-time driving state recognition result; determining the computational resource allocation requirement based on the real-time driving state recognition result using a preset threshold rule; adjusting the edge processor power according to the computational resource allocation requirement to obtain an adjusted power level; calculating the throttle opening drive signal based on the adjusted power level and fusing real-time monitoring data, and generating a throttle actuator command; monitoring the execution response time of the command and readjusting the power again to obtain a final balanced throttle control signal; and finally determining an updated driving state recognition result based on the final balanced throttle control signal. This invention achieves an intelligent balance between computational resources and system response in throttle control by constructing a complete technical process from multi-dimensional data perception, a two-layer dynamic power adjustment combining feedforward and feedback, to the final closed-loop state update, thus solving the problem of existing technologies struggling to balance control accuracy and energy efficiency under complex dynamic conditions.

[0081] Reference Figure 2 The second embodiment of the present invention provides an Internet of Things-based intelligent throttle control system, comprising: The status recognition module is used to acquire and process sensor data to obtain real-time driving status recognition results; The requirement determination module is used to determine the computing resource allocation requirements based on the real-time driving status recognition results and using preset threshold rules. The power adjustment module is used to adjust the power of the edge processor according to the computing resource allocation requirements to obtain the adjusted power level; The instruction generation module is used to calculate the throttle opening drive signal and generate throttle actuator instructions based on the adjusted power level and real-time monitoring data. The feedback control module is used to monitor the execution response time of the throttle actuator command and adjust the edge processor power again according to the execution response time to obtain the final balanced throttle control signal. The status update module is used to preprocess and classify the latest sensor data based on the final balanced throttle control signal to determine the updated driving status recognition result.

[0082] It should be noted that the IoT-based intelligent throttle control system provided in this embodiment of the invention is used to execute all the process steps of the IoT-based intelligent throttle control method described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0083] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an IoT-based intelligent throttle control program. When the processor executes the computer program, it implements the steps in the various IoT-based intelligent throttle control method embodiments described above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the status recognition module.

[0084] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0085] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0086] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0087] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0088] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0089] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0090] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A throttle intelligent control method based on the Internet of Things, characterized in that, include: Acquire sensor data and process it to obtain real-time driving status recognition results; Based on the real-time driving status recognition results, the computational resource allocation requirements are determined using preset threshold rules; Based on the computing resource allocation requirements, the power of the edge processor is adjusted to obtain the adjusted power level; Based on the adjusted power level and by integrating real-time monitoring data, the throttle opening drive signal is calculated, and the throttle actuator command is generated. The execution response time of the throttle actuator command is monitored, and the power of the edge processor is adjusted again based on the execution response time to obtain a final balanced throttle control signal; Based on the final balanced throttle control signal, the latest sensor data is preprocessed and classified to determine the updated driving state recognition result.

2. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The process of acquiring and processing sensor data to obtain real-time driving state recognition results includes: Acquire raw data streams of engine speed and accelerator pedal position; The original data stream is timestamped to obtain a synchronized data stream; The synchronous data stream is filtered and denoised to obtain a smooth data sequence; The data in the smoothed data sequence are normalized and weighted to obtain weighted feature values, which are then used as the real-time driving state recognition result.

3. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The step of determining the computational resource allocation requirements based on the real-time driving state recognition results using preset threshold rules includes: If the real-time driving status recognition result exceeds the preset dynamic scene threshold, an indication that computing resource allocation needs to be increased is generated as the computing resource allocation requirement.

4. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The step of adjusting the edge processor power according to the computing resource allocation requirements to obtain the adjusted power level includes: The voltage and frequency parameters of the edge processor are adjusted using a dynamic voltage-frequency scaling method to obtain the adjusted voltage-frequency parameters. Based on the adjusted voltage frequency parameters, the power output of the edge processor is reconfigured to obtain the adjusted power level.

5. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The step of calculating the throttle opening drive signal based on the adjusted power level and integrating real-time monitoring data, and generating throttle actuator commands, includes: Obtain the current energy consumption index from the adjusted power level; If the current energy consumption index is lower than the preset energy consumption threshold, then the engine speed data and accelerator pedal position data are acquired and fused to obtain the driving intensity index. The driving intensity index is input into a pre-trained logistic regression model to obtain the probability value of resource enhancement needs; Based on the comparison between the probability value and the preset scheduling threshold, one of the multiple preset scheduling schemes is selected as the optimized resource scheduling scheme. Based on the optimized resource scheduling scheme, and using a pre-established multi-dimensional mapping table that defines the relationship between sensor data and throttle opening, the throttle opening drive signal is calculated. The throttle opening drive signal is modulated to obtain the throttle actuator command.

6. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The process of monitoring the execution response time of the throttle actuator command and readjusting the edge processor power based on the execution response time to obtain a final balanced throttle control signal includes: The execution response time of the throttle actuator command is continuously monitored to obtain a response time dataset; If any response time in the response time dataset exceeds a preset response time threshold, the power output is recalculated to obtain a new power target value, which is then used as the first power output dataset. Adjust the performance level of the edge processor based on the first power output dataset; Under the adjusted performance level, the initial throttle control signal is recalculated based on the latest real-time sensor data; The initial throttle control signal is fused with the actual throttle opening feedback value from the previous moment to generate the final balanced throttle control signal.

7. The IoT-based intelligent throttle control method according to claim 1, characterized in that, The step of preprocessing and classifying the latest sensor data based on the final balanced throttle control signal to determine the updated driving state recognition result includes: Based on the final balanced throttle control signal, acquire sensor data; The sensor data is subjected to signal noise verification to obtain the verification result; Based on the verification results, the sensor data is subjected to conditional filtering to obtain purified sensor data; The purified sensor data is normalized to obtain the data to be classified. The data to be classified is input into a pre-established driving state model for classification processing to obtain the updated driving state recognition result.

8. A throttle intelligent control system based on the Internet of Things, characterized in that, include: The status recognition module is used to acquire and process sensor data to obtain real-time driving status recognition results; The requirement determination module is used to determine the computing resource allocation requirements based on the real-time driving status recognition results and using preset threshold rules. The power adjustment module is used to adjust the power of the edge processor according to the computing resource allocation requirements to obtain the adjusted power level; The instruction generation module is used to calculate the throttle opening drive signal and generate throttle actuator instructions based on the adjusted power level and real-time monitoring data. The feedback control module is used to monitor the execution response time of the throttle actuator command and adjust the edge processor power again according to the execution response time to obtain the final balanced throttle control signal. The status update module is used to preprocess and classify the latest sensor data based on the final balanced throttle control signal to determine the updated driving status recognition result.

Images