Battery energy-saving control system and method with regenerative braking energy recovery function

Abstract

The invention discloses a battery energy-saving control system and method with a regenerative braking energy recovery function, and belongs to the technical field of battery management. The battery energy-saving control system comprises a sensing module, a data processing module, a central control module and an execution module; the sensing module is used for continuously collecting multi-source information of sensing parameters of the electric vehicle under different working conditions; the data processing module is used for preprocessing the multi-source information and extracting key features; the central control module is used for matching an optimal sensing parameter value from a preset sensing parameter library and dynamically adjusting sensing parameters, so that the problem of insufficient personalized adaptation is effectively solved, and a corresponding control instruction is generated; meanwhile, the perception parameter library is updated, and perception parameter matching information in the perception parameter library is continuously perfected; the execution module is used for executing regenerative braking and energy storage operation according to the control instruction, conducting real-time monitoring in the execution process and generating execution feedback data.

Description

Technical Field

[0001] The invention relates to a battery energy-saving control system and method with a regenerative braking energy recovery function, and belongs to the technical field of battery management. Background Art

[0002] In the modern urban transportation system, electric vehicles, as a means of transportation in modern life, are popular among people because of their convenience and speed. They not only provide an economical solution for short-distance travel, but also effectively alleviate urban traffic congestion and improve overall traffic mobility.

[0003] The existing Chinese patent application with publication number CN117485340A discloses a braking energy recovery method, a vehicle and a storage medium, including: in a vehicle braking process, respectively determining the braking energy recovery rate and the whole vehicle energy saving rate corresponding to at least one time period, wherein, when the number of time periods is at least two, at least two time periods are continuous and the duration of any two time periods is equal; determining the average braking energy recovery rate according to the braking energy recovery rate corresponding to each time period; determining the average whole vehicle energy saving rate according to the whole vehicle energy saving rate corresponding to each time period; determining the braking energy recovery level according to the average braking energy recovery rate and the average whole vehicle energy saving rate, wherein the braking energy recovery level is used to characterize the braking energy recovery effect.

[0004] Although the existing technology can solve the problem of judging the current vehicle energy recovery status only by the braking energy recovery rate by dividing the braking energy recovery level, it does not consider the compatibility issues faced by regenerative braking energy recovery in practical applications, especially the voltage and power incompatibility and brake coordination in vehicles of different models or years. It also ignores the problem of insufficient personalized adaptation, and cannot meet the needs of some users for strong energy recovery to save energy, nor can it meet the preferences of users who prefer traditional braking feeling. It has major defects in user experience and is difficult to take into account the diverse needs of users. Therefore, the present application provides a battery energy-saving control system and method with regenerative braking energy recovery function. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a battery energy-saving control system and method with regenerative braking energy recovery function, which realizes adaptation to vehicles of different models and years by dynamically adjusting perception parameters and establishing a continuously optimized perception parameter library. At the same time, the regenerative braking strategy is flexibly adjusted according to the user's driving habits and needs, meeting the user's diverse needs for energy saving and braking feeling, and effectively improving the user experience.

[0006] To achieve the above object, the present invention provides the following technical solutions: A battery energy-saving control system with a regenerative braking energy recovery function, comprising: a sensing module, a data processing module, a central control module and an execution module; The sensing module is used to continuously collect multi-source information of sensing parameters of the electric vehicle under different working conditions; The data processing module is used to pre-process the multi-source information and extract key features; The central control module is used to match the best perception parameter value from a preset perception parameter library, dynamically adjust the perception parameter, and generate corresponding control instructions; at the same time, the perception parameter library is updated according to the continuously accumulated multi-source information; The execution module is used to execute regenerative braking and energy storage operations according to the control instructions, perform real-time monitoring during the execution process, and generate execution feedback data.

[0007] Furthermore, the specific steps of multi-source information preprocessing include: Compute perception samples The mean and standard deviation , and sample points The corresponding Z-score value ;in, For the perception sample Middle Sample points, , is the sample size; Set the initial deviation threshold to , determine the sample point Is it a preliminary outlier? , the sample point Not a preliminary abnormal point; if , the sample point is the initial abnormal point; Constructing a preliminary anomaly set , and calculate the initial outliers The relative Z-score value ;in, For the initial anomaly set Middle An initial anomaly, , is the number of preliminary outliers; Set the relative threshold to , determine the preliminary abnormal point Is it an extreme value? , the initial abnormal point Not an extreme value; if , the initial abnormal point is an extreme value; The number of statistical extreme values ​​is , and calculate the anomaly correction factor ; Define the correction interval as , remove the sample points outside the correction interval in the perception sample, and generate the corrected perception sample as ; Integrate the corrected perception samples of several perception parameters to generate a multidimensional data set , and perform standardization processing and data unified format conversion to generate a multidimensional data matrix .

[0008] Furthermore, the specific steps of extracting key features include: Calculate the multidimensional data matrix The covariance matrix of ; For the covariance matrix Perform eigenvalue decomposition and obtain the eigenvalue and the corresponding eigenvector , , , ; Calculate eigenvalues The corresponding principal component contribution rate , and calculate the cumulative contribution rate , select the cumulative contribution rate Reaching the preset cumulative threshold Before principal components and generate the principal component matrix ; The multidimensional data matrix With the principal component matrix Multiply to generate the principal component perception matrix ; Calculate the principal component perception matrix The row distances are calculated and the row distance matrix is ​​generated. ; The principal component perception matrix Decompose into cluster sets , according to the row distance matrix The new cluster set obtained by merging the two closest clusters , the process of merging clusters is repeated until the preset number of clusters is reached , get the key feature set .

[0009] Furthermore, the specific steps of matching perception parameters include: According to the key feature set Real-time collection of key characteristic data sets of electric vehicles during operation ; Calculate key characteristic data set and perception parameter library The Euclidean distance of each perceptual parameter vector in is selected, and the perceptual parameter vector corresponding to the minimum Euclidean distance is selected. , as the best matching perception parameter value; Key feature data set Perform standardized design and generate standard feature vectors .

[0010] Furthermore, the specific steps of real-time monitoring of electric vehicles include: The standard feature vector As the initial reference value, define the initial threshold of each perception parameter ; Get the key feature data set of the previous moment , calculate the difference between the current moment and the previous moment perception parameter value ; like , the perception parameters have not changed significantly, return to obtain the actual running data at the next moment, and re-match the perception parameters; if , the perception parameters change significantly, and the perception parameters are adjusted dynamically.

[0011] Furthermore, the specific steps of dynamically adjusting the perception parameters include: When the vehicle acceleration is detected Exceeding the rapid acceleration threshold When and energy storage charging current ; When the vehicle acceleration is detected Below the rapid deceleration threshold When the regenerative braking intensity is increased ; When the battery level Below the battery threshold When the energy storage charging current is increased , while reducing the intensity of regenerative braking ; Generate the corresponding control instruction vector according to the dynamically adjusted perception parameters .

[0012] Furthermore, the specific steps of updating the perception parameter library include: The perception parameter library is optimized using the gradient descent algorithm. Let the perception parameter vector is the parameter to be optimized, and the loss function is ; Calculating the loss function and gradient , and update the perception parameter vector , until the maximum iteration threshold is reached Then stop the iteration; is the current perception parameter vector; The perception parameter library is updated according to the perception parameter vector obtained after stopping the iteration.

[0013] Furthermore, the execution module includes a regenerative braking unit, an energy storage unit and a battery monitoring unit; The regenerative braking unit is used to receive control instructions from the central control module, adjust the braking force, and convert the kinetic energy of the vehicle into electrical energy during the vehicle braking process; The energy storage unit is used to store the electric energy recovered by regenerative braking in the battery according to the charging parameters set by the control instruction; The battery monitoring unit is used to monitor the working status of regenerative braking and energy storage in real time and generate feedback data.

[0014] A battery energy-saving control method with a regenerative braking energy recovery function, comprising: After the vehicle is started, multi-source information of the electric vehicle's sensing parameters under different working conditions is continuously collected, the multi-source data is preprocessed, and key features are extracted; Match the best perception parameter value from the preset perception parameter library, monitor the actual working state of the electric vehicle in real time during operation, and generate corresponding control instructions; Perform regenerative braking and energy storage operations according to control instructions, monitor the working status of the execution module in real time and feedback data.

[0015] Beneficial effects of the present invention: Through high-precision sensors, multi-dimensional information on vehicle operation is fully collected, and key features are extracted using principal component analysis and clustering algorithms, revealing the characteristics of vehicle equipment and user driving style; combined with the current vehicle status and the perception parameter library, a dynamic adjustment mechanism is used to achieve standardized design and adaptive optimization of perception parameters, ensuring the effectiveness and personalized adaptation of regenerative braking and energy storage strategies in different driving scenarios and vehicle conditions; the compatibility and adaptability of the system are improved, and through continuous learning and parameter library updates, the matching of perception parameters is continuously optimized, so that the system can respond more accurately to the needs of various types of electric vehicles; at the same time, through real-time monitoring of execution feedback and optimization of control instructions accordingly, the response speed and control accuracy of the system are further enhanced, and through continuous self-optimization, the system's adaptability to different vehicles and user needs is improved, effectively solving the problem of personalized adaptation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a structural diagram of a battery energy-saving control system with a regenerative braking energy recovery function; Figure 2 Flowchart for preprocessing multi-source information; Figure 3 Flowchart for extracting key features; Figure 4 A flow chart for generating control instructions; Figure 5 The present invention is a flow chart of a battery energy-saving control method with a regenerative braking energy recovery function. DETAILED DESCRIPTION

[0017] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0018] Example 1

[0019] refer to Figures 1 to 4 As shown, this embodiment introduces a battery energy-saving control system with a regenerative braking energy recovery function, including: a perception module, a data processing module, a central control module and an execution module; The perception module is used to utilize high-precision sensors equipped in the electric vehicle, such as electrical parameter sensors, braking parameter sensors, and driving behavior sensors, to continuously collect multi-source information of the sensing parameters of the electric vehicle under different working conditions, and transmit the multi-source information to the data processing module; wherein, the electrical parameter sensor is used to obtain the voltage and current of the battery of the electric vehicle and the input and output voltage and current of the motor in real time, and accurately obtain the operating status of the electrical equipment, including voltage sensors and current sensors; the braking parameter sensor is used to obtain the pressure change and travel information of the brake in real time, and sense the braking action and braking intensity of the electric vehicle, including pressure sensors and displacement sensors; the driving behavior sensor is used to obtain the acceleration and speed change of the electric vehicle in real time during driving; the electric vehicles in this embodiment include electric two-wheeled vehicles and three-wheeled vehicles; The data processing module is used to receive the multi-source information transmitted by the perception module in real time, and pre-process the multi-source information, such as removing noise and outliers, standardizing, converting data of different types and magnitudes into a unified format, and using principal component analysis and clustering algorithms to extract key features from the pre-processed multi-source information, such as the power fluctuation characteristics of electrical equipment, the response characteristics of braking equipment, and the characteristic vector of user driving habits. By analyzing these key features, the characteristics of the electrical equipment and braking equipment of the electric vehicle and the user's driving style preference are identified; at the same time, a multi-source database is established to classify and store the processed multi-source information to facilitate the query and retrospective analysis of historical data, and the data is regularly updated and maintained to ensure the accuracy and timeliness of the data; The central control module is used to use the key features extracted by the data processing module to match the best perception parameter values ​​from the preset perception parameter library to achieve standardized design of perception parameters, and to monitor the actual working status of the electric vehicle in real time during operation, capture the changes in the perception parameters, adaptively and dynamically adjust the perception parameters, and generate corresponding control instructions to send to the execution module. Among them, the dynamic adjustment mechanism flexibly changes the strategies of regenerative braking and energy storage according to different driving scenarios and vehicle states through real-time perception, effectively solving the problem of insufficient personalized adaptation; at the same time, the perception parameter library is updated according to the continuously accumulated multi-source information, and the best perception parameter values ​​of electric vehicles of different models and years under various working conditions are analyzed, and the perception parameter matching information in the perception parameter library is continuously improved, so that the system can dynamically adjust the perception parameters according to the characteristics of different electric vehicles, thereby improving the compatibility and adaptability to different vehicles, and ensuring that the regenerative braking energy recovery function can be better realized on different electric vehicles; The execution module is used to receive control instructions from the central control unit, and execute regenerative braking and energy storage operations according to the control instructions. It performs real-time monitoring during the execution process, generates execution feedback data, and transmits the execution feedback data to the data processing module and the central control module to update and optimize the control instructions, thereby forming a closed-loop control. The feedback data includes the actual working status of the execution module, such as the battery charging status and the actual effect of regenerative braking, which is helpful for understanding whether the control instructions issued previously have achieved the expected results. The system operation is continuously optimized through feedback data, which can better adapt to the personalized needs of different vehicles and users, and effectively solve the problem of insufficient personalized adaptation.

[0020] Furthermore, the specific steps of multi-source information preprocessing include: There are several perception parameters in the multi-source information transmitted by the perception module, such as electrical parameters, braking parameters, and driving parameters. Take a specific perception parameter as an example for preprocessing, select a specific perception parameter from the multi-source information, and arrange its data in order to form a perception sample ,in, is the sample size, For the perception sample Middle Sample points, ; Compute perception samples The mean and standard deviation , and the Z-score value is used to measure the degree of deviation of each sample point from the mean, and calculate the sample point The corresponding Z-score value , the expression is as follows: ; ; ; Set the initial deviation threshold to , and the perception samples Each sample point in is judged abnormally to determine the sample point Is it a preliminary outlier? , sample point If the deviation from the mean is within an acceptable range, the sample point is determined Not a preliminary abnormal point; if , sample point If the deviation from the mean is large, the sample point is determined is the initial abnormal point; Obtaining perception samples that have been judged to be abnormal Initial outlier points in , and construct a preliminary outlier set ,in, Initial exception set The number of preliminary outliers in For the initial exception set Middle An initial anomaly, ; For each preliminary anomaly point in the preliminary anomaly set , calculate the relative Z-score value , reflects the relative size of the deviation of the preliminary abnormal point from the mean of all sample points. On the basis of the preliminary abnormal point, the extreme values ​​with very large deviation from all sample points are further screened out to more accurately identify abnormal data. The expression is as follows: ; Set the relative threshold to , determine the initial abnormal points Is it an extreme value? , initial outlier point Although the deviation from the mean is large, relative to all sample points, its degree of deviation is still within a certain range, so it is determined as a preliminary outlier. Not an extreme value; if , initial outlier point Compared with all sample points, the deviation from the mean is very large, so the initial outlier point is determined. is an extreme value; The number of statistical extreme values ​​is , and calculate the anomaly correction factor , the expression is as follows: ; In the formula, is a constant used to adjust the width of the correction interval so as to determine the correction interval of the abnormal point later; According to the definition of abnormal correction coefficient, the correction interval of abnormal points is: , traverse the perception samples , define the sample points that fall outside the correction interval as outliers, remove the outliers, and generate the corrected perception samples as , and recalculate the mean and standard deviation ,in, is the number of samples after anomaly correction; Integrate the corrected perception samples of several perception parameters to generate a multidimensional data set , and use the linear normalization method to transform the multidimensional data set The data in the data are scaled to the range of [0, 1] to eliminate the dimension and magnitude differences of the data, making different types of data comparable. At the same time, data of different types and magnitudes are organized according to a preset format, such as unifying numerical data into fixed-precision floating-point numbers and timestamps into a specific date and time format to generate a multidimensional data matrix. ;in, , Both The matrix of is the number of samples after integration, is the number of perception parameter types.

[0021] Furthermore, the specific steps of extracting key features include: Computing multidimensional data matrices The covariance matrix of , which is used to measure the correlation between different perceptual parameters, and is a The matrix expression is as follows: ; Covariance matrix Perform eigenvalue decomposition and obtain the eigenvalue and the corresponding eigenvector , , …, , where the eigenvalue represents the variance of each principal component and the eigenvector represents the direction of the principal component. , and satisfies , ; Calculate eigenvalues The corresponding principal component contribution rate , and calculate the cumulative contribution rate , select the principal component containing most of the data information through the contribution rate to achieve dimensionality reduction, and select the cumulative contribution rate Reaching the preset cumulative threshold Before The corresponding eigenvalue is , the eigenvector is , , …, , and generate the principal component matrix ,for The matrix of ; the expression is as follows: ; ; In the formula, select , and after each calculation With cumulative threshold Compare to determine the current Whether the cumulative threshold is reached. When the cumulative threshold is reached, stop calculating the cumulative contribution rate and output the last obtained value, and ; Multidimensional data matrix With the principal component matrix Multiply and project onto the selected principal component to generate the principal component perception matrix ,in, is a A matrix with each column representing a principal component. Through principal component analysis, the original high-dimensional data is converted into low-dimensional principal component data, retaining most of the data information, reducing the data dimension, and improving the calculation efficiency; Calculate the principal component perception matrix The row distance is used to measure the distance between different samples in the principal component data, prepare for clustering, and generate the row distance matrix , is a The matrix of is as follows: ; In the formula, Principal component perception matrix Middle Row and The distance between rows, and Respectively Row and The first elements; The principal component perception matrix Each sample group in is regarded as a separate cluster, as the initial state of agglomerative clustering, the current cluster set , is the number of current clusters, according to the row distance matrix The new cluster set obtained by merging the two closest clusters , the process of merging clusters is repeated until the preset number of clusters is reached , get the key feature set ,Through agglomerative clustering, similar samples are clustered into one class to obtain representative key features ,which facilitates the classification and analysis of the data.

[0022] Furthermore, the specific steps of matching perception parameters include: According to the key feature set obtained by preprocessing Real-time collection of key characteristic data sets of electric vehicles during operation ; Using the Euclidean distance algorithm, calculate the key feature data set and perception parameter library The similarity of each perception parameter vector in is calculated, and the perception parameter that best matches the current vehicle operation state is found from the perception parameter library. The smaller the Euclidean distance, the more similar the two vectors are. From all the calculated Euclidean distances, the perception parameter vector corresponding to the minimum Euclidean distance is selected. , as the best matching perception parameter value; the expression is as follows: ; In the formula, is the first Group-aware parameter vector, , Perception Parameter Library The number of perceptual parameter vectors in , Key feature data set Middle Key feature values, is the first The first Key feature values; Key feature data set Perform standardized design, comprehensively consider the optimal perception parameters and the original parameters of the current vehicle, and obtain the standard feature vector , the expression is as follows: ; In the formula, is the weight coefficient, and its value range is ,when When it is close to 1, the optimal perception parameter is adopted. When it is close to 0, the original parameters of the current vehicle are retained; in this embodiment, .

[0023] Furthermore, the specific steps of real-time monitoring of electric vehicles include: The standard feature vector As the initial reference value, define the initial threshold of each perception parameter and normal range; all key features are considered as a whole at this time; Get the key feature data set of the previous moment , calculate the difference between the current moment and the previous moment perception parameter value , understand the change of parameters through the difference, and then judge whether the perception parameters have changed significantly; like , the perception parameters have not changed significantly, return to obtain the actual running data at the next moment, and re-match the perception parameters; if , the perception parameters change significantly, and the perception parameters are adjusted dynamically.

[0024] Furthermore, the specific steps of dynamically adjusting the perception parameters include: When the vehicle acceleration is detected Exceeding the rapid acceleration threshold In order to meet the power requirements during rapid acceleration and take energy recovery into consideration, the regenerative braking intensity is adjusted. and energy storage charging current , the expression is as follows: ; ; In the formula, is the current regenerative braking intensity, is the current energy storage charging current, , is the adjustment factor; When the vehicle acceleration is detected Below the rapid deceleration threshold When the regenerative braking intensity is increased to improve the energy recovery efficiency, the expression is as follows: ; In the formula, is the adjustment factor; When the battery level Below the battery threshold When the energy storage charging current is increased, the impact of the regenerative braking intensity on vehicle driving is appropriately reduced to quickly replenish the battery power. The expression is as follows: ; ; In the formula, , is the adjustment factor; Generate the corresponding control instruction vector according to the dynamically adjusted perception parameters , such as regenerative braking intensity control instructions, energy storage charging current control instructions, and sending control instruction vectors to the execution module; wherein, For the Control instructions, To control the number of instructions.

[0025] Furthermore, the specific steps of updating the perception parameter library include: Continuously accumulate multi-source information of electric vehicles and corresponding actual operating effect data, which include various status parameters of the vehicle, control instructions and actual working status of the execution module feedback (such as the actual charge of the battery, the actual energy recovery of regenerative braking); define performance evaluation indicators to measure the operating effect of the system under different working conditions, assuming that the energy recovery efficiency is , the expression is as follows: ; In the formula, The energy recovered by regenerative braking is is the total energy lost during vehicle braking; The perception parameter library is optimized using the gradient descent algorithm. Let the perception parameter vector is the parameter to be optimized, and the loss function is , maximize the energy recovery efficiency by minimizing the loss function , when the loss function is minimized, the energy recovery efficiency is solved The maximum value of and the perception parameter vector corresponding to the maximum value of energy recovery efficiency; Calculating the loss function and gradient , and update the perception parameter vector , until the maximum iteration threshold is reached Then stop iterating; the expression is as follows: ; In the formula, is the current perception parameter vector, is the updated perception parameter vector, is the learning rate, which controls the step size of parameter update at each iteration. The loss function is The gradient at According to the perception parameter vector obtained after stopping the iteration, the records of corresponding vehicle models, years and working conditions in the perception parameter library are updated; at the same time, the perception parameter library is cleaned and maintained regularly, and invalid or outdated records are deleted to ensure the accuracy and effectiveness of the parameter library.

[0026] Furthermore, the execution module includes a regenerative braking unit, an energy storage unit, and a battery monitoring unit; The regenerative braking unit is used to receive control instructions from the central control module, adjust the braking force, and convert the vehicle's kinetic energy into electrical energy during the vehicle braking process to achieve efficient energy recovery, effectively reduce energy loss during vehicle braking, and improve energy utilization efficiency; The energy storage unit is used to store the electric energy recovered by regenerative braking into the battery according to the charging parameters set by the control command; The battery monitoring unit is used to monitor the working status of regenerative braking and energy storage in real time, and generate feedback data, which is fed back to the central control module and data processing module to provide a basis for system optimization and adjustment.

[0027] Example 2

[0028] See also Figure 5 Another embodiment provided by the present invention is a battery energy-saving control method with a regenerative braking energy recovery function, comprising the following steps: After the vehicle is started, it continuously collects multi-source information of the electric vehicle's sensory parameters under different working conditions, such as electrical parameters, braking parameters, and driving parameters, pre-processes the multi-source data, and uses principal component analysis and clustering algorithms to extract key features; According to the extracted key features, the optimal perception parameter value is matched from the preset perception parameter library, and the actual working state of the electric vehicle during operation is monitored in real time to generate corresponding control instructions; Execute regenerative braking and energy storage operations according to control instructions, monitor the working status of the execution module in real time and feedback data to form a closed-loop control.

[0029] Furthermore, the specific steps of multi-source information preprocessing include: Compute perception samples The mean and standard deviation , and sample points The corresponding Z-score value , where the perception sample , is the sample size; Set the initial deviation threshold to , determine the sample point Is it a preliminary outlier? , then determine the sample point Not a preliminary abnormal point; if , then determine the sample point is the initial abnormal point; Constructing a preliminary anomaly set , calculate the relative Z-score value ; Set the relative threshold to , determine the initial abnormal points Is it an extreme value? , then determine the initial abnormal point Not an extreme value; if , then determine the initial abnormal point is an extreme value; Count the number of extreme values , and calculate the anomaly correction factor ; Define the correction interval as , traverse the perception samples , define the sample points that fall outside the correction interval as outliers, remove the outliers, and generate the corrected perception samples as , and recalculate the mean and standard deviation , Integrate the corrected perception samples of several perception parameters to generate a multidimensional data set , and perform standardization processing and data unified format conversion to generate a multidimensional data matrix .

[0030] Furthermore, the specific steps of extracting key features include: Computing multidimensional data matrices The covariance matrix of , and perform eigenvalue decomposition to obtain the eigenvalue and the corresponding eigenvector , , …, ; Calculate eigenvalues The corresponding principal component contribution rate , select the cumulative contribution rate Reaching the preset cumulative threshold Before principal components and generate the principal component matrix ; Multidimensional data matrix With the principal component matrix Multiply to generate the principal component perception matrix ; Calculate the principal component perception matrix The row distances are calculated and the row distance matrix is ​​generated. ; The principal component perception matrix Each sample group in is regarded as a separate cluster, which is used as the initial state of agglomerative clustering to generate a cluster set , according to the row distance matrix The new cluster set obtained by merging the two closest clusters , the process of merging clusters is repeated until the preset number of clusters is reached , get the key feature set .

[0031] Furthermore, the specific steps of matching perception parameters include: According to the key feature set obtained by preprocessing Real-time collection of key characteristic data sets of electric vehicles during operation ; Calculate key characteristic data set and perception parameter library The similarity of each perception parameter vector in the perception parameter library is used to find the perception parameter that best matches the current vehicle operation state. ; Taking into account the optimal perception parameters and the original parameters of the current vehicle, the key feature data set Perform standardized design and obtain the standard feature vector .

[0032] Furthermore, the specific steps of real-time monitoring of electric vehicles include: The standard feature vector As the initial reference value, define the initial threshold of each perception parameter and normal range; Get the key feature data set of the previous moment , calculate the difference between the current moment and the previous moment perception parameter value ; like , the perception parameters have not changed significantly, return to obtain the actual running data at the next moment, and re-match the perception parameters; if , the perception parameters change significantly, and the perception parameters are adjusted dynamically.

[0033] Furthermore, the specific steps of dynamically adjusting the perception parameters include: When the vehicle acceleration is detected Exceeding the rapid acceleration threshold In order to meet the power requirements during rapid acceleration and take energy recovery into consideration, the regenerative braking intensity is adjusted. and energy storage charging current , the expression is as follows: ; ; In the formula, is the current regenerative braking intensity, is the current energy storage charging current, , is the adjustment factor; When the vehicle acceleration is detected Below the rapid deceleration threshold When the regenerative braking intensity is increased to improve the energy recovery efficiency, the expression is as follows: ; In the formula, is the adjustment factor; When the battery level Below the battery threshold When the energy storage charging current is increased, the impact of the regenerative braking intensity on vehicle driving is appropriately reduced to quickly replenish the battery power. The expression is as follows: ;

[0034] ; In the formula, , is the adjustment factor; Generate the corresponding control instruction vector according to the dynamically adjusted perception parameters , such as regenerative braking intensity control instructions, energy storage charging current control instructions, and sending control instruction vectors to the execution module; wherein, For the Control instructions, To control the number of instructions.

[0035] Furthermore, the specific steps of updating the perception parameter library include: The perception parameter library is optimized using the gradient descent algorithm. Let the perception parameter vector is the parameter to be optimized, and the loss function is , maximize the energy recovery efficiency by minimizing the loss function ; Calculating the loss function and gradient , and update the perception parameter vector , until the maximum iteration threshold is reached Then the iteration stops; the expression is as follows: ; In the formula, is the current perception parameter vector, is the updated perception parameter vector, is the learning rate, which controls the step size of parameter update at each iteration. The loss function is The gradient at The perception parameter library is updated according to the perception parameter vector obtained after stopping the iteration.

[0036] In summary, the present invention collects multi-source information in real time, and performs preprocessing and feature extraction; based on the extracted features, the best parameters are matched from a preset perception parameter library to achieve standardized design, and through real-time monitoring and dynamic adjustment mechanisms, it can flexibly adapt to different driving scenarios and vehicle conditions, effectively solving the problem of personalized adaptation; as data accumulates, the perception parameter library is continuously optimized to improve compatibility and adaptability to different vehicle models; regenerative braking and energy storage operations are performed according to control instructions, and feedback data is generated by real-time monitoring, and through a closed-loop control mechanism, the operating effect is continuously optimized to better adapt to the personalized needs of different vehicles and users, and achieve efficient regenerative braking energy recovery.

[0037] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A battery energy-saving control system with a regenerative braking energy recovery function, characterized in that: include: Perception module, data processing module, central control module and execution module; The sensing module is used to continuously collect multi-source information of sensing parameters of the electric vehicle under different working conditions; The data processing module is used to pre-process the multi-source information and extract key features; The central control module is used to match the best perception parameter value from a preset perception parameter library, dynamically adjust the perception parameter, and generate corresponding control instructions; at the same time, the perception parameter library is updated according to the continuously accumulated multi-source information; The execution module is used to execute regenerative braking and energy storage operations according to the control instructions, perform real-time monitoring during the execution process, and generate execution feedback data.

2. The battery energy-saving control system with regenerative braking energy recovery function according to claim 1, characterized in that: The specific steps of multi-source information preprocessing include: Compute perception samples The mean and standard deviation , and sample points The corresponding Z-score value ;in, For the perception sample Middle Sample points, , is the sample size; Set the initial deviation threshold to , determine the sample point Is it a preliminary outlier? , the sample point Not a preliminary outlier point; if , the sample point is the initial abnormal point; Building a preliminary exception set , and calculate the initial outliers The relative Z-score value ;in, For the initial anomaly set Middle An initial anomaly, , is the number of preliminary outliers; Set the relative threshold to , determine the preliminary abnormal point Is it an extreme value? , the initial abnormal point Not an extreme value; if , the initial abnormal point is an extreme value; The number of statistical extreme values ​​is , and calculate the anomaly correction factor ; Define the correction interval as , remove the sample points outside the correction interval in the perception sample, and generate the corrected perception sample as ; Integrate the corrected perception samples of several perception parameters to generate a multidimensional data set , and perform standardization processing and data unified format conversion to generate a multidimensional data matrix .

3. The battery energy-saving control system with regenerative braking energy recovery function according to claim 2, characterized in that: The specific steps of extracting key features include: Calculate the multidimensional data matrix The covariance matrix of ; For the covariance matrix Perform eigenvalue decomposition and obtain the eigenvalue and the corresponding eigenvector , , , ; Calculate eigenvalues The corresponding principal component contribution rate , and calculate the cumulative contribution rate , select the cumulative contribution rate Reaching the preset cumulative threshold Before principal components and generate the principal component matrix ; The multidimensional data matrix With the principal component matrix Multiply to generate the principal component perception matrix ; Calculate the principal component perception matrix The row distances are calculated and the row distance matrix is ​​generated. ; The principal component perception matrix Decompose into cluster sets , according to the row distance matrix The new cluster set obtained by merging the two closest clusters , the process of merging clusters is repeated until the preset number of clusters is reached , get the key feature set .

4. The battery energy-saving control system with regenerative braking energy recovery function according to claim 3 is characterized in that: The specific steps of matching perception parameters include: According to the key feature set Real-time collection of key characteristic data sets of electric vehicles during operation ; Calculate key characteristic data set and perception parameter library The Euclidean distance of each perceptual parameter vector in is selected, and the perceptual parameter vector corresponding to the minimum Euclidean distance is selected. , as the best matching perception parameter value; Key feature data set Perform standardized design and generate standard feature vectors .

5. The battery energy-saving control system with regenerative braking energy recovery function according to claim 4, characterized in that: The specific steps for real-time monitoring of electric vehicles include: The standard feature vector As the initial reference value, define the initial threshold of each perception parameter ; Get the key feature data set of the previous moment , calculate the difference between the current moment and the previous moment perception parameter value ; like , the perception parameters have not changed significantly, return to obtain the actual running data at the next moment, and re-match the perception parameters; if , the perception parameters change significantly, and the perception parameters are adjusted dynamically.

6. The battery energy-saving control system with regenerative braking energy recovery function according to claim 5, characterized in that: The specific steps of dynamically adjusting perception parameters include: When the vehicle acceleration is detected Exceeding the rapid acceleration threshold When and energy storage charging current ; When the vehicle acceleration is detected Below the rapid deceleration threshold Increase the regenerative braking intensity ; When the battery level Below the battery threshold When the energy storage charging current is increased , while reducing the intensity of regenerative braking ; Generate the corresponding control instruction vector according to the dynamically adjusted perception parameters .

7. The battery energy-saving control system with regenerative braking energy recovery function according to claim 6, characterized in that: The specific steps for updating the perception parameter library include: The perception parameter library is optimized using the gradient descent algorithm. Let the perception parameter vector is the parameter to be optimized, and the loss function is ; Calculating the loss function and gradient , and update the perception parameter vector , until the maximum iteration threshold is reached Then stop the iteration; is the current perception parameter vector; The perception parameter library is updated according to the perception parameter vector obtained after stopping the iteration.

8. The battery energy-saving control system with regenerative braking energy recovery function according to claim 7, characterized in that: The execution module includes a regenerative braking unit, an energy storage unit and a battery monitoring unit; The regenerative braking unit is used to receive control instructions from the central control module, adjust the braking force, and convert the kinetic energy of the vehicle into electrical energy during the vehicle braking process; The energy storage unit is used to store the electric energy recovered by regenerative braking in the battery according to the charging parameters set by the control instruction; The battery monitoring unit is used to monitor the working status of regenerative braking and energy storage in real time and generate feedback data.

9. A battery energy-saving control method with a regenerative braking energy recovery function, which is implemented based on a battery energy-saving control system with a regenerative braking energy recovery function as claimed in any one of claims 1 to 8, characterized in that: include: After the vehicle is started, multi-source information of the electric vehicle's sensing parameters under different working conditions is continuously collected, the multi-source data is preprocessed, and key features are extracted; Match the best perception parameter value from the preset perception parameter library, monitor the actual working state of the electric vehicle in real time during operation, and generate corresponding control instructions; Perform regenerative braking and energy storage operations according to control instructions, monitor the working status of the execution module in real time and feedback data.

Images