Battery energy coupling optimization method and system, storage medium and computer

Abstract

The invention provides a battery energy coupling optimization method and system, a storage medium and a computer, and the optimization method comprises the steps: obtaining the current driving data of a vehicle, the driving data at least comprising vehicle driving data, battery state data and driving environment data; preprocessing the driving data, and inputting the preprocessed driving data into a feature extraction model to carry out analysis feature extraction; the importance of the extracted analysis features is sorted based on a self-attention mechanism, weight adjustment is carried out on the analysis features according to the sorting result of the importance, and the importance of the analysis features represents the influence degree of the analysis features on an energy management decision; and inputting the analysis characteristics after weight adjustment into a decision tree model for energy coupling analysis so as to adjust energy output and recovery of the battery according to the current driving state. The battery energy coupling optimization method provided by the invention is high in energy recovery and flexible in adjustment mode.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicles, and particularly relates to a battery energy coupling optimization method, system, storage medium and computer. Background Art

[0002] In recent years, the energy management technology of new energy vehicles has made remarkable progress. In particular, how to efficiently manage the energy output and recovery of power batteries in electric vehicles has become one of the key technologies for improving the vehicle's driving range and performance. The energy coupling optimization strategy has gradually become the mainstream research direction.

[0003] However, the current mainstream energy recovery systems usually only focus on energy recovery itself, ignoring the collaborative optimization between the power battery and braking energy, resulting in low energy utilization efficiency, difficult to achieve optimal energy distribution under complex working conditions, and causing energy waste. In addition, the decision-making mechanism of the energy management system is relatively fixed, unable to quickly adjust the energy distribution and recovery strategy under different driving conditions, and unable to self-adjust according to the vehicle usage situation, restricting the potential of the system. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide a battery energy coupling optimization method, system, and storage medium to solve the technical problems existing in the prior art.

[0005] The present invention proposes a battery energy coupling optimization method, including:

[0006] Obtain the current driving data of the vehicle, where the driving data at least includes vehicle driving data, battery state data, and driving environment data;

[0007] Preprocess the driving data, input the preprocessed driving data into a feature extraction model for feature extraction, and correspondingly obtain the analysis features related to energy management in the vehicle driving data, battery state data, and driving environment data;

[0008] Rank the importance of the extracted analysis features based on the self-attention mechanism, and adjust the weights of the analysis features according to the ranking result of the importance, where the importance of the analysis feature represents the influence degree of the analysis feature on the energy management decision;

[0009] Input the analysis features with adjusted weights into a decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state.

[0010] Preferably, the step of preprocessing the driving data includes:

[0011] Perform filtering processing on the driving data of the vehicle based on the Kalman filter algorithm to remove the noise data in the driving data;

[0012] The driving data after filtering is filled based on the KNN interpolation algorithm to supplement the missing data in the driving data and ensure data integrity;

[0013] The filled data is normalized to unify the driving data with different dimensions.

[0014] Preferably, the step of ranking the importance of each extracted analysis feature based on the self-attention mechanism includes:

[0015] Construct an original feature set based on a number of analysis features;

[0016] The original feature set is divided into several subsets according to the different values of each analysis feature within the time series, and each subset serves as a branch node of the original feature set;

[0017] Calculate the first information entropy of the original feature set and the second information entropy corresponding to the values of the analysis features included in each branch node;

[0018] Obtain the information gain of each analysis feature in the original feature set according to the first information entropy and the second information entropy, and determine the corresponding importance according to the information gain corresponding to each analysis feature and then sort them.

[0019] Preferably, the expression of the information gain is:

[0020]

[0021] In the formula, Gain(T,i) is the information gain of the i-th analysis feature in the original feature set, T=(t1,t2,…,t n ,) is the analysis feature included in the original feature set, Entr(T) is the first information entropy, T j is the analysis feature included in the j-th branch node, J is the number of branch nodes, is the second information entropy of the value of the i-th analysis feature on the j-th branch node;

[0022]

[0023] In the formula, n is the number of analysis features included in the original feature set, p i is the proportion of the i-th analysis feature in the original feature set;

[0024]

[0025] In the formula, M is the number of values of the analysis feature included in the j-th branch node, p’ i is the proportion of the value of the i-th analysis feature on the j-th branch node.

[0026] Preferably, the step of inputting the analysis features with adjusted weights into a decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state includes:

[0027] Establish a multi-input multi-output fuzzy system, where the input of the fuzzy system is the weights corresponding to the respective analysis features, and the output of the fuzzy system is the power for controlling the energy output and recovery of the battery;

[0028] Adopt a Z-shaped membership function as the membership function between the input and output of the fuzzy system;

[0029] Establish a rule base of the fuzzy system according to the number of subsets of the input and output of the fuzzy system;

[0030] Perform fuzzy solution based on the established rule base to adjust the energy output and recovery power of the battery according to the weights of the respective analysis features.

[0031] Preferably, before performing energy coupling analysis, the battery energy coupling optimization method further includes: training the decision tree model, and the step of training the decision tree model includes:

[0032] Collect the driving data of several vehicles under different driving conditions, and perform feature extraction after preprocessing the collected driving data;

[0033] Screen the key features that have the greatest impact on energy management decisions under different driving conditions, and divide the obtained key features into a data set to obtain a training set and a test set;

[0034] Select the CART decision tree algorithm to construct an initial decision tree model, and import the training set into the initial decision tree model;

[0035] Select the key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as the node splitting criterion, and dynamically adjust the hyperparameters in the node splitting process based on the K-fold cross-validation algorithm;

[0036] Set the maximum depth of the decision tree model, and use the maximum depth of the decision tree model as the condition for completing the training of the decision tree model;

[0037] Input the test set into the trained decision tree model, and verify the trained decision tree model based on a preset evaluation index.

[0038] The present invention also proposes a battery energy coupling optimization system, including;

[0039] An acquisition module, configured to acquire the current driving data of a vehicle, where the driving data at least includes vehicle driving data, battery state data, and driving environment data;

[0040] An extraction module, configured to preprocess the driving data, input the preprocessed driving data into a feature extraction model for feature extraction, and correspondingly obtain analysis features related to energy management in the vehicle driving data, battery state data, and driving environment data;

[0041] A sorting module, configured to sort the importance of the extracted analysis features based on the self-attention mechanism, and adjust the weights of the analysis features according to the sorting result of the importance, where the importance of the analysis feature represents the influence degree of the analysis feature on the energy management decision;

[0042] An adjustment module, configured to input the analysis features with adjusted weights into a decision tree model for energy coupling analysis, so as to adjust the energy output and recovery of the battery according to the current driving state.

[0043] Preferably, the battery energy coupling optimization system further includes:

[0044] A training module, configured to train the decision tree model, and the steps of training the decision tree model include:

[0045] Collect the driving data of a number of vehicles under different driving conditions, preprocess the collected driving data and then perform feature extraction;

[0046] Screen the key features that have the greatest impact on the energy management decision under different driving conditions, divide the obtained key features into a data set to obtain a training set and a test set;

[0047] Select the CART decision tree algorithm to construct an initial decision tree model, and import the training set into the initial decision tree model;

[0048] Select the key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as the node splitting criterion, and dynamically adjust the hyperparameters in the node splitting process based on the K-fold cross-validation algorithm;

[0049] Set the maximum depth of the decision tree model, and use the maximum depth of the decision tree model as the condition for completing the training of the decision tree model;

[0050] Input the test set into the trained decision tree model, and verify the trained decision tree model based on a preset evaluation index.

[0051] The present invention also proposes a storage medium, on which a computer program is stored, and when the program is executed by a processor, the above battery energy coupling optimization method is implemented.

[0052] The present invention also provides a computer, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned battery energy coupling optimization method is implemented.

[0053] The beneficial effects of the present invention compared with the prior art are as follows: The battery energy coupling optimization method provided by this application first obtains the current driving data of the vehicle. Among them, the driving data at least includes vehicle driving data, battery state data, and driving environment data. The obtained driving data comprehensively considers the influences of multiple factors such as vehicle driving factors, battery state factors, and driving environment factors. Then, the data is preprocessed to improve the data quality for subsequent analysis, and the analysis features related to energy management in the data are extracted to ensure that the model pays attention to key factors and reduces the amount of data processing. Based on self-attention, the extracted features are analyzed and weighted to automatically identify and prioritize the key influencing factors that affect energy management and sort them, improving the accuracy of data processing. The features with adjusted weights are input into a decision tree model for energy coupling analysis. On the basis of comprehensively considering the influences of various factors on the energy management strategy, according to the real-time state of the vehicle, the energy output and recovery of the battery are dynamically adjusted to achieve the best balance between the two. The energy management decision is flexible, and the recovery utilization rate is high, which is suitable for large-scale promotion.

[0054] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flowchart of the battery energy coupling optimization method in Embodiment 1 of the present invention;

[0056] Figure 2 It is a structural block diagram of the computer in Embodiment 4 of the present invention.

[0057] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. SPECIFIC EMBODIMENTS

[0058] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the present invention herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0060] Embodiment 1

[0061] Please refer to Figure 1 , which shows the battery energy coupling optimization method in the first embodiment of the present invention. The battery energy coupling optimization method specifically includes steps S10 to S40:

[0062] S10. Obtain the current driving data of the vehicle, where the driving data at least includes vehicle driving data, battery state data, and driving environment data;

[0063] Optionally, first collect data through various sensors on the vehicle. The collected data includes but is not limited to: driving data, driving environment data, battery state data, etc.; the driving data covers the speed, acceleration, and driving mode of the vehicle, etc.; the driving environment data includes road surface gradient and flatness, etc.; the battery data includes battery charge (SOC), temperature, and state of health (SOH), etc.

[0064] S20. Preprocess the driving data, input the preprocessed driving data into a feature extraction model for feature extraction, and correspondingly obtain analysis features related to energy management in the vehicle driving data, battery state data, and driving environment data;

[0065] Optionally, the collected data may be affected by noise interference or have missing values. To improve the accuracy of the model, we clean the original data. Data cleaning ensures the quality of the input data, and feature extraction ensures that the model pays attention to key factors; specifically, the data cleaning steps include:

[0066] Perform filtering processing on the driving data of the vehicle based on the Kalman filter algorithm to remove the noise data in the driving data;

[0067] Perform filling processing on the filtered driving data based on the KNN interpolation algorithm to supplement the missing data in the driving data and ensure data integrity;

[0068] Perform normalization processing on the filled data to unify the driving data with different dimensions.

[0069] Filtering processing can remove the noise in sensor data to ensure the accuracy of the data; the KNN (k-Nearest Neighbor) interpolation algorithm can fill in the missing data points to ensure the integrity of the data set; normalizing the data can unify data with different dimensions into the same range to avoid model training imbalance caused by dimensional problems during the training process.

[0070] After data cleaning, we need to extract important features related to energy management. This process is based on different driving conditions and battery states, aiming to select features that can effectively affect energy management decisions; such as vehicle driving speed and acceleration, which directly affect the energy consumption and recovery efficiency of the vehicle; the SOC (State Of Charge) of the battery, battery temperature, SOH (State of Health), which reflect the remaining power and health status of the battery and determine the upper limit of energy recovery; road gradient and flatness, which affect the energy consumption and recovery effect.

[0071] S30, rank the importance of the extracted analysis features based on the self-attention mechanism, and adjust the weights of the analysis features according to the ranking results of the importance, where the importance of the analysis feature represents the degree of influence of the analysis feature on the energy management decision;

[0072] Optionally, the step of ranking the importance of the extracted analysis features based on the self-attention mechanism includes:

[0073] Construct an original feature set based on several analysis features;

[0074] Divide the original feature set into several subsets according to different values of each analysis feature within the time series, and each subset serves as a branch node of the original feature set;

[0075] Calculate the first information entropy of the original feature set and the second information entropy corresponding to the analysis feature values included in each branch node;

[0076] Obtain the information gain of each analysis feature in the original feature set according to the first information entropy and the second information entropy, and determine the corresponding importance according to the information gain corresponding to each analysis feature and then rank them.

[0077] The expression of the information gain is:

[0078]

[0079] Wherein, Gain(T, i) is the information gain of the i-th analysis feature in the original feature set, T=(t1, t2, …, t2) is the analysis feature included in the original feature set, Entr(T) is the first information entropy, and T j is the analysis feature included in the j-th branch node, and J is the number of branch nodes. is the second information entropy of the value of the i-th analysis feature on the j-th branch node;

[0080]

[0081] Wherein, n is the number of analysis features included in the original feature set, and p i is the proportion of the i-th analysis feature in the original feature set;

[0082]

[0083] Wherein, M is the number of values of the analysis feature included in the j-th branch node, and p’ i is the proportion of the value of the i-th analysis feature on the j-th branch node.

[0084] Optionally, in this embodiment, the self-attention mechanism is introduced, which can help the system automatically identify and prioritize the key influencing factors affecting energy management and sort them. The self-attention mechanism calculates the importance of each analysis feature under the current working condition. For example, at high load, the battery temperature may be the most important monitoring object; while during long-term braking, the system will give priority to the energy recovery efficiency. In this embodiment, the self-attention mechanism includes an information gain algorithm to screen and sort the features affecting energy management, and considers vehicle speed, acceleration, road gradient, battery state, etc. for the features affecting energy management, providing a collaborative basis for subsequent battery output and recovery, and improving the energy utilization rate.

[0085] In specific implementation, branch nodes are constructed according to the values of each analysis feature at different times. For example, if the feature values at ten moments are selected, the original feature set is divided into ten branch nodes, and each branch node contains the values of each analysis feature at the corresponding moment. The information entropy of each analysis feature in each branch node is solved, and then the information entropy is summed according to the number of branch nodes. Finally, it is combined with the information entropy of the original feature set to solve the information gain corresponding to the analysis feature. The information gain represents the classification ability of a feature for data samples; the importance corresponding to each analysis feature is determined through the information gain; through the self-attention mechanism, the system can automatically adjust the attention degree to different factors to ensure the optimization of energy management.

[0086] S40. Input the analyzed features with adjusted weights into the decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state.

[0087] Before performing the energy coupling analysis, the battery energy coupling optimization method further includes: training the decision tree model, and the steps of training the decision tree model include:

[0088] Collect the driving data of several vehicles under different driving conditions, and perform feature extraction after preprocessing the collected driving data.

[0089] Screen the key features that have the greatest impact on the energy management decision under different driving conditions, divide the obtained key features into a data set to obtain a training set and a test set.

[0090] Select the CART decision tree algorithm to construct an initial decision tree model, and import the training set into the initial decision tree model.

[0091] Select the key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as the node splitting criterion, and dynamically adjust the hyperparameters in the node splitting process based on the K-fold cross-validation algorithm.

[0092] Set the maximum depth of the decision tree model, and use the maximum depth of the decision tree model as the condition for completing the training of the decision tree model.

[0093] Input the test set into the trained decision tree model, and verify the trained decision tree model based on the preset evaluation metrics.

[0094] Optionally, in this embodiment, the process of collecting driving data and preprocessing and feature extracting the driving data is similar to the steps of S20, which will not be elaborated here; in this embodiment, the key features that have the greatest impact on energy management decisions under different driving conditions can be screened based on the principal component analysis method. 80% of the data is used as the training set, and 20% of the data is used as the test set; the CART (Classification and Regression Trees) decision tree algorithm is used to train the data. The decision tree makes decisions through a series of "if-else" rules, and the Gini index is used as the node splitting criterion, which can help the decision tree select the feature that can most improve the purity; based on the K-fold cross-validation algorithm, the hyperparameters can be adjusted multiple times during the training process to ensure that the model has good prediction ability on different data sets. To avoid overfitting, the maximum depth or minimum number of samples of the decision tree can be set as the condition for the completion of training to ensure the generalization ability of the model. The trained decision tree model can be verified through the test set, and evaluation metrics can be set to evaluate the effect of the decision tree model in controlling the battery energy output and recovery. The evaluation metrics can be: accuracy, precision, recall, F1 score, etc., to ensure that it performs excellently in predicting and optimizing energy management. Optionally, when the evaluation fails, the system will also give feedback according to the effect of each energy management decision and update the training data, so that the model can be continuously optimized during the actual driving process and the overall energy management efficiency can be improved.

[0095] Optionally, in this embodiment, the step of inputting the analysis features with adjusted weights into the decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state includes:

[0096] Establish a multi-input multi-output fuzzy system, where the input of the fuzzy system is the weights corresponding to each analysis feature, and the output of the fuzzy system is the power for controlling the battery energy output and recovery;

[0097] Adopt the Z-shaped membership function as the membership function between the input and output of the fuzzy system;

[0098] Establish the rule base of the fuzzy system according to the number of subsets of the input and output of the fuzzy system;

[0099] Perform fuzzy solution based on the established rule base to adjust the energy output and recovery power of the battery according to the weights of each analysis feature.

[0100] In this embodiment, energy coupling analysis is performed based on a fuzzy algorithm in the decision tree model. The membership function between the input and output selects the Z-shaped membership function in the fuzzy algorithm. The graph of the Z-shaped membership function is "Z"-shaped, with a relatively steep slope, which can quickly adjust the output, ensure the system's quick response, avoid delay, and can quickly adjust the energy distribution and recovery strategy under different driving conditions. Through energy coupling analysis, the decision-making mechanism of the battery energy management system can intelligently adjust the coordinated work of battery energy output and recovery according to different driving scenarios, road conditions, and battery conditions, and dynamically adjust the power of battery energy output and recovery to improve the overall energy management efficiency.

[0101] Optionally, adjusting the energy output and recovery scenarios of the battery according to the weights of each analysis feature includes:

[0102] Speed and acceleration: Determine the current energy demand of the vehicle. When driving at high speed or accelerating, by predicting the vehicle's demand and giving priority to the needs of battery health and power output, at this time, the proportion of energy output will increase to provide sufficient power support while reducing energy recovery; when driving at low speed or decelerating, the vehicle's power demand is low, and the proportion of energy recovery is enhanced to maximize the energy recovery efficiency.

[0103] Gradient: When cruising at medium speed, the gradient is the main factor to consider. When going uphill, the vehicle needs more energy output to overcome gravity, and the system will give priority to mobilizing the battery output; when going downhill, the system will use regenerative braking energy to recover energy.

[0104] Battery state (SOC, temperature, etc.): The health state of the battery directly affects the energy management strategy. When the battery temperature is too high or the SOC is too low, the system will adjust the proportion of energy output and recovery, appropriately reduce the recovery power, prevent damage to the battery due to overcharging, and ensure that the battery always operates within a safe range; when the SOC is relatively high, the system will increase the recovery intensity to improve the recovery efficiency. By optimizing the battery energy output and recovery, while maximizing the energy recovery efficiency, the health of the battery is protected; further, the system can also use a model to predict the battery's service life and adjust the energy recovery strategy according to the prediction results to ensure that the battery can maintain the best working state for a long time.

[0105] In summary, the battery energy coupling optimization method provided by the present application first obtains the current driving data of the vehicle. Among them, the driving data at least includes vehicle driving data, battery state data, and driving environment data. The obtained driving data comprehensively considers the influences of multiple factors such as vehicle driving factors, battery state factors, and driving environment factors. Then, the data is preprocessed to improve the data quality for subsequent analysis, and the analysis features related to energy management in the data are extracted to ensure that the model pays attention to the key factors and reduces the data processing volume. Based on self-attention, the extracted features are analyzed and weighted to automatically identify and prioritize the key influencing factors that affect energy management and sort them, improving the accuracy of data processing. The features with adjusted weights are input into the decision tree model for energy coupling analysis. On the basis of comprehensively considering the influences of various factors on the energy management strategy, according to the real-time state of the vehicle, the energy output and recovery of the battery are dynamically adjusted to achieve the best balance between the two. The energy management decision is flexible and the recovery utilization rate is high.

[0106] Embodiment 2

[0107] This embodiment provides a battery energy coupling optimization system, including:

[0108] An acquisition module, configured to acquire the current driving data of the vehicle, where the driving data at least includes vehicle driving data, battery state data, and driving environment data;

[0109] An extraction module, configured to preprocess the driving data, input the preprocessed driving data into a feature extraction model for feature extraction, and correspondingly obtain the analysis features related to energy management in the vehicle driving data, battery state data, and driving environment data;

[0110] A sorting module, configured to sort the importance of the extracted analysis features based on the self-attention mechanism, and adjust the weights of the analysis features according to the sorting result of the importance, where the importance of the analysis feature represents the influence degree of the analysis feature on the energy management decision;

[0111] An adjustment module, configured to input the analysis features with adjusted weights into the decision tree model for energy coupling analysis, so as to adjust the energy output and recovery of the battery according to the current driving state.

[0112] Preferably, the step of preprocessing the driving data includes:

[0113] Filter the driving data of the vehicle based on the Kalman filter algorithm to remove the noise data in the driving data;

[0114] Perform filling processing on the filtered driving data based on the KNN interpolation algorithm to supplement the missing data in the driving data and ensure data integrity;

[0115] Normalize the data after filling to unify the driving data with different dimensions.

[0116] Preferably, the step of sorting the importance of each extracted analysis feature based on the self-attention mechanism includes:

[0117] Construct an original feature set based on a number of analysis features;

[0118] Divide the original feature set into several subsets according to the different values of each analysis feature within the time series, and each subset serves as a branch node of the original feature set;

[0119] Calculate the first information entropy of the original feature set and the second information entropy corresponding to the values of the analysis features included in each branch node;

[0120] Obtain the information gain of each analysis feature in the original feature set according to the first information entropy and the second information entropy, and determine the corresponding importance according to the information gain corresponding to each analysis feature and then sort them.

[0121] Preferably, the expression of the information gain is:

[0122]

[0123] In the formula, Gain(T,i) is the information gain of the i-th analysis feature in the original feature set, T=(t1,t2,…,t n ,) is the analysis feature included in the original feature set, Entr(T) is the first information entropy, T j is the analysis feature included in the j-th branch node, J is the number of branch nodes, is the second information entropy of the value of the i-th analysis feature on the j-th branch node;

[0124]

[0125] In the formula, n is the number of analysis features included in the original feature set, p i is the proportion of the i-th analysis feature in the original feature set;

[0126]

[0127] In the formula, M is the number of values of the analysis feature included in the j-th branch node, p’ i is the proportion of the value of the i-th analysis feature on the j-th branch node.

[0128] Preferably, the step of inputting the analysis features with adjusted weights into the decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state includes:

[0129] A multi-input multi-output fuzzy system is established, where the input of the fuzzy system is the weight corresponding to each analysis feature, and the output of the fuzzy system is the power for controlling the battery energy output and recovery.

[0130] The Z-shaped membership function is used as the membership function between the input and output of the fuzzy system.

[0131] The rule base of the fuzzy system is established according to the number of subsets of the input and output of the fuzzy system.

[0132] Fuzzy solution is carried out based on the established rule base to adjust the energy output and recovery power of the battery according to the weights of each analysis feature.

[0133] Preferably, before performing the energy coupling analysis, the battery energy coupling optimization method further includes: training a decision tree model, and the steps of training the decision tree model include:

[0134] Collect the driving data of several vehicles under different driving conditions, and perform feature extraction after preprocessing the collected driving data.

[0135] Screen the key features that have the greatest impact on the energy management decision under different driving conditions, and divide the obtained key features into a data set to obtain a training set and a test set.

[0136] Select the CART decision tree algorithm to construct an initial decision tree model, and import the training set into the initial decision tree model.

[0137] Select the key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as the node splitting criterion, and dynamically adjust the hyperparameters in the node splitting process based on the K-fold cross-validation algorithm.

[0138] Set the maximum depth of the decision tree model, and use the maximum depth of the decision tree model as the condition for completing the training of the decision tree model.

[0139] Input the test set into the trained decision tree model, and verify the trained decision tree model based on the preset evaluation index.

[0140] Embodiment III

[0141] This embodiment proposes a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the battery energy coupling optimization method as described above.

[0142] Embodiment IV

[0143] The present invention also proposes a computer, please refer to Figure 2, as shown is a computer in an embodiment of the present invention, including a memory 10, a processor 20, and a computer program 30 stored on the memory 10 and executable on the processor 20. When the processor 20 executes the computer program 30, the above-mentioned battery energy coupling optimization method is implemented.

[0144] Among them, the memory 10 includes at least one type of storage medium, and the storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, magnetic disk, optical disc, etc. The memory 10 can be an internal storage unit of the computer in some embodiments, such as the hard disk of the computer. The memory 10 can also be an external storage device in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 10 can also include both the internal storage unit of the computer and the external storage device. The memory 10 can be used not only to store application software installed on the computer and various types of data, but also to temporarily store data that has been output or will be output.

[0145] Among them, the processor 20 can be an Electronic Control Unit (ECU, also known as a vehicle computer), a Central Processing Unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments, and is used to run the program code stored in the memory 10 or process data, such as executing an access restriction program, etc.

[0146] It should be noted that Figure 2 the shown structure does not constitute a limitation on the computer. In other embodiments, the computer may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0147] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device, or equipment (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, device, or equipment), or in combination with these instruction execution systems, devices, or equipment. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, device, or equipment.

[0148] More specific examples (a non-exhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0149] It should be understood that the various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0150] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the various technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as falling within the scope described in this specification.

[0151] The above-described embodiments merely represent several implementation manners of the present application. Their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A battery energy coupling optimization method, characterized in that: include: Acquiring current driving data of the vehicle, wherein the driving data at least includes vehicle driving data, battery status data, and driving environment data; Preprocessing the driving data, inputting the preprocessed driving data into a feature extraction model for feature extraction, and correspondingly obtaining analysis features related to energy management in the vehicle driving data, battery status data, and driving environment data; The importance of each extracted analysis feature is ranked based on the self-attention mechanism, and the weight of each analysis feature is adjusted according to the ranking result of importance, where the importance of the analysis feature indicates the influence of the analysis feature on the energy management decision; The weighted analysis features are input into the decision tree model for energy coupling analysis to adjust the battery’s energy output and recovery according to the current driving status.

2. The battery energy coupling optimization method according to claim 1, characterized in that: The step of preprocessing the driving data comprises: The vehicle driving data is filtered based on the Kalman filter algorithm to remove the noise data of the driving data; The driving data after filtering is filled based on the KNN interpolation algorithm to supplement the missing data in the driving data and ensure data integrity; The padded data are normalized to unify the driving data of different dimensions.

3. The battery energy coupling optimization method according to claim 1, characterized in that: The step of ranking the importance of each extracted analysis feature based on the self-attention mechanism includes: Constructing an original feature set based on several analysis features; Dividing the original feature set into a plurality of subsets according to different values ​​of each analysis feature in the time series, each subset being a branch node of the original feature set; Calculate the first information entropy of the original feature set and the second information entropy corresponding to the analysis feature values ​​contained in each branch node; The information gain of each analysis feature in the original feature set is obtained according to the first information entropy and the second information entropy, and the corresponding importance is determined according to the information gain corresponding to each analysis feature, and then sorted.

4. The battery energy coupling optimization method according to claim 3, characterized in that: The expression of the information gain is: Where Gain(T,i) is the information gain of the i-th analysis feature in the original feature set, T = (t1, t2,…, t n ,) is the analysis feature contained in the original feature set, Entr(T) is the first information entropy, T j is the analysis feature contained in the jth branch node, J is the number of branch nodes, is the second information entropy of the i-th analysis feature value at the j-th branch node; Where n is the number of analysis features contained in the original feature set, p i is the proportion of the i-th analysis feature in the original feature set; Where M is the number of analysis feature values ​​contained in the j-th branch node, p' i is the proportion of the i-th analysis feature value on the j-th branch node.

5. The battery energy coupling optimization method according to claim 1, characterized in that: The step of inputting the weight-adjusted analysis features into the decision tree model for energy coupling analysis to adjust the energy output and recovery of the battery according to the current driving state includes: Establish a multi-input and multi-output fuzzy system, where the input of the fuzzy system is the weight corresponding to each analysis feature, and the output of the fuzzy system is the power for controlling the battery energy output and recovery; A Z-shaped membership function is used as the membership function between the input of the fuzzy system and the output of the fuzzy system; A rule base of the fuzzy system is established according to the subset number of the input of the fuzzy system and the output of the fuzzy system; Fuzzy solving is performed based on the established rule base to adjust the energy output and recovered power of the battery according to the weights of each analysis feature.

6. The battery energy coupling optimization method according to claim 1, characterized in that: Before performing energy coupling analysis, the battery energy coupling optimization method further includes: training a decision tree model, and the step of training the decision tree model includes: Collecting driving data of several vehicles under different driving conditions, and extracting features after preprocessing the collected driving data; Filter the key features that have the greatest impact on energy management decisions under different driving conditions, and divide the data set into training sets and test sets. Selecting the CART decision tree algorithm to build an initial decision tree model, and importing the training set into the initial decision tree model; Select key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as a node splitting criterion, and dynamically adjust hyperparameters in the node splitting process based on a K-fold cross validation algorithm; Set the maximum depth of the decision tree model and use the maximum depth of the decision tree model as the training completion condition of the decision tree model; The test set is input into the trained decision tree model, and the trained decision tree model is verified based on the preset evaluation indicators.

7. A battery energy coupling optimization system, characterized in that: include; An acquisition module, used to acquire current driving data of the vehicle, wherein the driving data at least includes vehicle driving data, battery status data and driving environment data; An extraction module, used to pre-process the driving data, input the pre-processed driving data into a feature extraction model for feature extraction, and obtain analysis features related to energy management in the vehicle driving data, battery status data, and driving environment data; A sorting module is used to sort the importance of each extracted analysis feature based on the self-attention mechanism, and adjust the weight of each analysis feature according to the ranking result of importance, wherein the importance of the analysis feature indicates the influence of the analysis feature on the energy management decision; The adjustment module is used to input the analysis features after weight adjustment into the decision tree model for energy coupling analysis, so as to adjust the energy output and recovery of the battery according to the current driving state.

8. The battery energy coupling optimization system according to claim 7, characterized in that: The battery energy coupling optimization system also includes: The training module is used to train the decision tree model. The steps of training the decision tree model include: Collecting driving data of several vehicles under different driving conditions, and extracting features after preprocessing the collected driving data; Filter the key features that have the greatest impact on energy management decisions under different driving conditions, and divide the data set into training sets and test sets. Selecting the CART decision tree algorithm to build an initial decision tree model, and importing the training set into the initial decision tree model; Select key features in the training set to perform node splitting in the initial decision tree model, use the Gini index as a node splitting criterion, and dynamically adjust hyperparameters in the node splitting process based on a K-fold cross validation algorithm; Set the maximum depth of the decision tree model and use the maximum depth of the decision tree model as the training completion condition of the decision tree model; The test set is input into the trained decision tree model, and the trained decision tree model is verified based on the preset evaluation indicators.

9. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the battery energy coupling optimization method as described in any one of claims 1 to 6 is implemented.

10. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the battery energy coupling optimization method according to any one of claims 1 to 6 is implemented.

Images