Automobile power battery health state monitoring method based on cloud computing

Abstract

The invention discloses an automobile power battery health state monitoring method based on cloud computing, and relates to the technical field of automobile power battery health state monitoring, and the method comprises the steps: collecting automobile power battery data, and carrying out the preprocessing; inputting the preprocessed data into a convolutional neural network (CNN) model to extract a static feature vector, and inputting the preprocessed data into a recurrent neural network (RNN) model to extract a dynamic feature vector; based on the static feature vector and the dynamic feature vector, a weighted fusion technology is adopted to generate an automobile power battery health feature vector; a multi-layer perceptron (MLP) is adopted to construct an automobile power battery health state evaluation model, the automobile power battery health feature vector is input into the automobile power battery health state evaluation model, and an automobile power battery health state evaluation result is output; intelligent monitoring of the health state of the automobile power battery is achieved through systematic steps, and it is ensured that the whole process from data collection to fault early warning is efficient.

Description

Technical Field

[0001] The present invention relates to the technical field of health state monitoring of automotive power batteries, and particularly to a method for monitoring the health state of automotive power batteries based on cloud computing. Background Art

[0002] With the rapid development of new energy vehicles, the health condition of automotive power batteries has become a key factor affecting the service life, safety and maintenance cost of the whole vehicle. In order to ensure the safe operation of the batteries, the technology for monitoring the health state of power batteries has gradually become a research hotspot in the industry. In recent years, significant progress has been made in the application of machine learning and deep learning technologies in battery monitoring. By collecting the operating data of battery voltage, current and temperature in real time and using convolutional neural network (CNN) and recurrent neural network (RNN) models for feature extraction, the health state of the battery can be effectively evaluated. At the same time, the wide application of cloud computing technology provides strong support for the storage and processing of battery data, promotes the intelligence of battery management system (BMS), enables remote monitoring and intelligent decision-making of distributed batteries, and improves reliability and efficiency.

[0003] However, there are still some deficiencies in the existing technology for battery health assessment. First, static features and dynamic features usually need to be extracted through different models respectively, resulting in poor data fusion effect and limited assessment accuracy. Second, many existing assessment models rely on a single feature extraction method and lack adaptability under complex working conditions, and cannot fully cope with the changes in the health state of the battery. Therefore, how to effectively fuse static and dynamic features and combine multi-layer perceptron (MLP) and support vector machine (SVM) methods to improve the assessment accuracy is still a difficult problem to be solved by technology. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for monitoring the health state of automotive power batteries based on cloud computing to solve the problems of insufficient fusion of static and dynamic features and low accuracy of health assessment in the existing technology.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for monitoring the health state of automotive power batteries based on cloud computing, which includes: Collecting automotive power battery data and performing preprocessing; Inputting the preprocessed automotive power battery data into a convolutional neural network (CNN) model to extract static feature vectors, and at the same time, inputting the preprocessed automotive power battery data into a recurrent neural network (RNN) model to extract dynamic feature vectors; Based on the static feature vector and the dynamic feature vector, a weighted fusion technique is adopted to generate the health feature vector of the automotive power battery; A multi-layer perceptron (MLP) is used to construct an assessment model for the health state of the automotive power battery. The health feature vector of the automotive power battery is input into the assessment model for the health state of the automotive power battery, and the assessment result of the health state of the automotive power battery is output; Based on the assessment result of the health state of the automotive power battery, a support vector machine (SVM) is used to classify the fault modes; Based on the classification result of the fault modes, a fault warning is triggered and a handling suggestion is generated.

[0007] As a preferred solution of the method for monitoring the health state of the automotive power battery based on cloud computing according to the present invention, wherein: the collection of the automotive power battery data and the preprocessing are specifically carried out as follows: The automotive power battery data is collected through the vehicle-built sensors; The vehicle-built sensors refer to the voltage sensor, current sensor, temperature sensor, SOC sensor and SOH sensor of the automotive power battery; The automotive power battery data refers to the voltage, current, temperature, charge state and health state of the automotive power battery; For the collected automotive power battery data, the missing automotive power battery data is processed by the interpolation filling method; The filled automotive power battery data is denoised by using the Kalman filter; The denoised automotive power battery data is normalized by using the Z-score normalization method.

[0008] As a preferred solution of the method for monitoring the health state of the automotive power battery based on cloud computing according to the present invention, wherein: the input of the preprocessed automotive power battery data into the convolutional neural network (CNN) model to extract the static feature vector is specifically carried out as follows: The normalized automotive power battery data is input into the convolutional neural network (CNN) model, and the local static features of different scales of the automotive power battery data are extracted through convolutional operations to obtain the convolutional feature map ; The convolutional feature map is subjected to dimensionality reduction and max-pooling operations through the pooling layer to reduce the dimension of the feature map and obtain the pooled feature map ; After the pooled feature map is flattened into a one-dimensional vector, it is transformed through the fully connected layer incorporating the residual connection mechanism to generate the static feature vector of the automotive power battery data, and the expression is: ; wherein, Represents the static feature vector of the automotive power battery data, Represents the weight matrix of the fully connected layer, Represents the flattened three-dimensional feature map after pooling into a one-dimensional vector, Represents the ReLU activation function of the fully connected layer, Represents the bias term of the fully connected layer.

[0009] As a preferred solution of the method for monitoring the health status of automotive power batteries based on cloud computing according to the present invention, wherein: the step of inputting the preprocessed automotive power battery data into the recurrent neural network RNN model to extract the dynamic feature vector is as follows: Input the standardized automotive power battery data into the recurrent neural network RNN model; Through the time series structure of the recurrent neural network RNN model, capture the dynamic features of the health status of the automotive power battery changing over time, and generate the memory state at each time step ; Use the weighted average pooling method to perform weighted aggregation on each memory state to obtain the dynamic feature vector of the automotive power battery. The expression is: ; Wherein, Represents the dynamic feature vector of the automotive power battery data, Represents the total length of the time series, Represents the Weight at the moment.

[0010] As a preferred solution of the method for monitoring the health status of automotive power batteries based on cloud computing according to the present invention, wherein: based on the static feature vector and the dynamic feature vector, adopt the weighted fusion technology to generate the health feature vector of the automotive power battery. The specific steps are as follows: Combine the static feature vector extracted by the convolutional neural network CNN model and the dynamic feature vector extracted by the recurrent neural network RNN model using the weighted fusion technology to obtain the health feature vector of the automotive power battery. The expression is: ; Wherein, Represents the health feature vector of the automotive power battery at the moment, Represents the static feature vector of the automotive power battery data extracted at the moment, Represents the moment, Represents the weighting coefficient of the static feature vector of the automotive power battery data at the moment, Represents the weighted coefficient of the dynamic feature vector of the automotive power battery data at the moment.

[0011] As a preferred solution of the method for monitoring the health status of automotive power batteries based on cloud computing according to the present invention, wherein: the multi-layer perceptron MLP is used to construct an evaluation model for the health status of automotive power batteries, and the health feature vector of the automotive power battery is input into the evaluation model for the health status of the automotive power battery to output an evaluation result for the health status of the automotive power battery. The specific steps are as follows: Select the multi-layer perceptron MLP as the basis for the evaluation model of the health status of automotive power batteries; Input the health feature vector of the automotive power battery into the evaluation model for the health status of the automotive power battery to obtain an evaluation result for the battery health status. The expression is: ; wherein, is the evaluation result of the health status of the automotive power battery at the moment, represents the Sigmoid activation function used in the output layer, represents the number of hidden layers, represents the output weight of the th hidden layer, represents the ReLU activation function used in the hidden layer, represents the th bias term of the hidden layer, represents the bias term of the output layer.

[0012] As a preferred solution of the method for monitoring the health status of automotive power batteries based on cloud computing according to the present invention, wherein: based on the evaluation result of the health status of the automotive power battery, the support vector machine SVM is used to classify the fault modes. The specific steps are as follows: Input the evaluation result of the health status of the automotive power battery into the support vector machine SVM model to obtain a classification result for the fault modes. The expression is: ; wherein, represents the th fault mode of the automotive power battery at the moment, represents the sign function, represents the weight vector of the support vector machine SVM, represents the bias term of the support vector machine SVM; The classification result of the fault modes is the no-fault mode, degradation mode, and fault mode of the automotive power battery; According to the classification result of the fault modes, set the classification threshold for the fault modes ; When > If so, the automotive power battery mode is classified as a fault - free mode; When = If so, the automotive power battery mode is classified as a degradation mode; When < If so, the automotive power battery mode is classified as a fault mode.

[0013] As a preferred solution of the method for monitoring the health status of an automotive power battery based on cloud computing according to the present invention, wherein: based on the fault mode classification result, a fault warning is triggered and a processing suggestion is generated. The specific steps are as follows: According to the fault mode classification result, the threshold range for triggering a fault warning is set as ; Wherein, represents the lowest lower limit for triggering a fault warning, represents the highest upper limit for triggering a fault warning; When ≥ It indicates that the automotive power battery is in a fault - free mode, and there is no need to trigger a warning. Then the state of the automotive power battery is normal and no maintenance check is required; When ≤ < It indicates that the automotive power battery is in a degradation mode, and a mild warning is triggered. Then the health status of the automotive power battery is in the degradation stage, and maintenance checks are carried out according to the plan; When < It indicates that the automotive power battery is in a fault mode, and a severe warning is triggered. Then the health status of the automotive power battery has seriously declined. Immediately stop using the vehicle and contact a professional for repair.

[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for monitoring the health status of an automotive power battery based on cloud computing as described in the first aspect of the present invention is implemented.

[0015] In a third aspect, the present invention provides a computer - readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by the processor, any step of the method for monitoring the health status of an automotive power battery based on cloud computing as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: The present invention realizes the intelligent monitoring of the health state of automotive power batteries through systematic steps, ensuring the high efficiency of the entire process from data collection to fault warning; First, through the preprocessing of the original data, the data quality for subsequent analysis is guaranteed; Second, the CNN and RNN models are used to extract static and dynamic feature vectors respectively, and the weighted fusion technology is combined to generate health feature vectors, improving the accuracy of fault prediction. The MLP is used to construct an evaluation model to convert complex features into intuitive health scores, facilitating users to understand the battery state; Furthermore, SVM is used to classify fault modes, improving the speed of fault diagnosis and supporting rapid response; Finally, warnings are triggered according to the fault modes and treatment suggestions are put forward to ensure that measures are taken before the problem expands, reducing the maintenance cost. Generally speaking, the present invention not only optimizes the performance of electric vehicles, improves the user experience, but also realizes a win-win situation of safety, efficiency and economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of the method for monitoring the health state of automotive power batteries based on cloud computing in Embodiment 1; Figure 2 It is a schematic diagram of the evaluation result of the health state of automotive power batteries in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the above objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0020] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0022] Embodiment 1, refer to Figure 1 andFigure 2 , which is the first embodiment of the present invention. This embodiment provides a method for monitoring the health state of automotive power batteries based on cloud computing, including the following steps: S1. Collect automotive power battery data and perform preprocessing; Collect automotive power battery data through in-vehicle sensors; The in-vehicle sensors refer to automotive power battery voltage sensors, current sensors, temperature sensors, SOC sensors, and SOH sensors; The automotive power battery data refers to the voltage, current, temperature, charge state, and health state of the automotive power battery; For the collected automotive power battery data, interpolate and fill in the missing automotive power battery data; Use Kalman filtering to denoise the filled automotive power battery data; The denoised automotive power battery data is standardized using the Z-score normalization method; Through cleaning, denoising, and standardization, ensure the accuracy and stability of the input data, eliminate outliers and noise interference, provide reliable basic data for subsequent feature extraction and health state assessment, thus better reflecting the true health status of the battery and improving the accuracy and robustness of model prediction.

[0023] S2. Input the preprocessed automotive power battery data into a convolutional neural network CNN model to extract static feature vectors. At the same time, input the preprocessed automotive power battery data into a recurrent neural network RNN model to extract dynamic feature vectors; Input the standardized automotive power battery data into a convolutional neural network CNN model, and extract local static features of different scales of the automotive power battery data through convolutional operations to obtain a convolutional feature map, with the expression: ; Among them, represents the convolutional feature map, represents the dimension of the standardized data, represents the size of the convolutional kernel, and respectively represent the row and column position indexes in the convolutional operation, represents the element in the standardized data, located at the th row and the th column, represents the convolutional kernel; Reduce the dimension of the convolutional feature map through the pooling layer and perform max-pooling operations to reduce the dimension of the feature map and obtain a pooled feature map, with the expression: ; Among them, represents the feature map after pooling, and respectively represent the row and column position indices within the pooling window, represents the size of the pooling window, represents the element value at the th row and th column in the feature map after convolution, means that for all positions within the pooling window, the maximum value within the window is selected as the output of the pooling operation; After flattening the feature map after pooling into a one-dimensional vector, it is transformed through a fully connected layer incorporating a residual connection mechanism to generate a static feature vector of the automotive power battery data, and the expression is: ; Among them, represents the static feature vector of the automotive power battery data, represents the weight matrix of the fully connected layer, represents flattening the three-dimensional feature map after pooling into a one-dimensional vector, represents the ReLU activation function of the fully connected layer, represents the bias term of the fully connected layer; By using a convolutional layer to identify local change trends in battery data to discover potential abnormal fluctuations, and combining a pooling layer to reduce computational complexity while retaining key information, and then leveraging the residual connection mechanism to enhance the training effect and expressive power of feature extraction, it ensures the stable and efficient operation of the deep network in health assessment; Input the standardized automotive power battery data into a recurrent neural network (RNN) model; Through the time series structure of the recurrent neural network (RNN) model, capture the dynamic features of the health state of the automotive power battery changing over time, and generate the memory state at each time step. The expression is: ; Among them, represents the memory state at the th moment, represents the standardized automotive power battery data input at the th moment, represents the activation function, represents the weight matrix of the hidden layer, represents the weight matrix of the previous hidden state, represents the memory state of the previous moment, represents the bias term; Use the weighted average pooling method to perform weighted aggregation on each memory state to obtain the dynamic feature vector of the automotive power battery, and the expression is: ; Among them, represents the dynamic feature vector of the automotive power battery data, represents the total length of the time series, represents the weight at the th moment; The recurrent neural network (RNN) model can effectively capture the changing trends accumulated over time during the battery usage process, timely detect potential signs of gradual battery performance degradation and sudden failures, and provide keen insights for predictive maintenance.

[0024] S3. Based on the static feature vector and the dynamic feature vector, adopt the weighted fusion technology to generate the health feature vector of the automotive power battery; Combine the static feature vector extracted by the convolutional neural network (CNN) model and the dynamic feature vector extracted by the recurrent neural network (RNN) model using the weighted fusion technology to obtain the health feature vector of the automotive power battery, and the expression is: ; Among them, represents the health feature vector of the automotive power battery at the th moment, represents the static feature vector of the automotive power battery data extracted at the th moment, represents the dynamic feature vector of the automotive power battery data extracted at the th moment, represents the weighting coefficient of the static feature vector of the automotive power battery data at the th moment, represents the weighting coefficient of the dynamic feature vector of the automotive power battery data at the th moment; By weighted fusion of the static feature vector and the dynamic feature vector, the expressive ability of the health feature vector of the automotive power battery is enhanced, which not only more comprehensively reflects the actual health state of the battery, improves the prediction ability of the convolutional neural network (CNN) model and the recurrent neural network (RNN) model and the accuracy and reliability of fault prediction, especially in the aspect of early fault mode recognition, it can identify potential problems earlier and more accurately.

[0025] S4. Use a multi-layer perceptron (MLP) to construct a health state evaluation model for the automotive power battery, input the health feature vector of the automotive power battery into the health state evaluation model of the automotive power battery, and output the health state evaluation result of the automotive power battery; Select the multi-layer perceptron (MLP) as the basis of the health state evaluation model for the automotive power battery; Input the health feature vector of the automotive power battery into the health state assessment model of the automotive power battery to obtain the health state assessment result, and the expression is: ; Among them, is the health state assessment result of the automotive power battery at the th moment, represents the Sigmoid activation function used in the output layer, represents the number of hidden layers, represents the output weight of the th hidden layer, represents the ReLU activation function used in the hidden layer, represents the bias term of the th hidden layer, represents the bias term of the output layer; Through its carefully designed hidden layer structure and activation function, the multi-layer perceptron MLP model can effectively map multi-dimensional health features to a quantified health score, so as to accurately and generally reflect the complex health state such as the battery, providing an intuitive and reliable basis for decision-making.

[0026] S5. Based on the health state assessment result of the automotive power battery, use the support vector machine SVM to classify the fault mode; Input the health state assessment result of the automotive power battery into the support vector machine SVM model to obtain the fault mode classification result, and the expression is: ; Among them, represents the fault mode of the automotive power battery at the th moment, represents the sign function, represents the weight vector of the support vector machine SVM, represents the bias term of the support vector machine SVM; Set the fault mode classification threshold . The fault mode classification threshold of the present invention is 0; When > , then the automotive power battery mode is classified as a non-fault mode; When = , then the automotive power battery mode is classified as a degradation mode; When < , then the automotive power battery mode is classified as a fault mode; By accurately classifying battery fault modes, not only the real-time monitoring of the battery health status is achieved, but also key data support is provided for fault warning and maintenance, thereby reducing the need for manual intervention, significantly improving the automation level and decision-making efficiency.

[0027] S6. Based on the classification results of fault modes, trigger a fault warning and generate a handling suggestion; According to the classification results of fault modes, set the threshold range for triggering a fault warning as In the present invention, according to the evaluation results of the battery health status under different historical fault modes, the threshold range for triggering a fault warning is set as ; Among them, represents the lowest lower limit for triggering a fault warning, represents the highest upper limit for triggering a fault warning; When ≥ , it indicates that the automotive power battery is in a fault-free mode and there is no need to trigger a warning, so the state of the automotive power battery is normal and no maintenance inspection is required; When ≤ < , it indicates that the automotive power battery is in a degradation mode and a mild warning is triggered, so the health status of the automotive power battery is in the degradation stage and maintenance inspection is carried out according to the plan; When < , it indicates that the automotive power battery is in a fault mode and a severe warning is triggered, so the health status of the automotive power battery drops severely. Immediately stop using the vehicle and contact a professional for repair; By automatically triggering a fault warning through threshold judgment, it ensures a timely response when the battery health status is abnormal, avoids greater fault problems, and provides personalized handling suggestions to achieve precise maintenance to extend the battery life and reduce the maintenance cost.

[0028] This embodiment also provides a computer device, which is applicable to the situation of the method for monitoring the health status of an automotive power battery based on cloud computing, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for monitoring the health status of an automotive power battery based on cloud computing as proposed in the above embodiment; The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus; wherein, the processor of the computer device is used to provide computing and control capabilities; the memory of the computer device includes a non-volatile storage medium and an internal memory; the non-volatile storage medium stores an operating system and computer programs; the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium; the communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication) or other technologies; the display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.; This embodiment further provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method for monitoring the health status of an automotive power battery based on cloud computing as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, a magnetic disk or an optical disc; In summary, the present invention realizes the intelligent monitoring of the health state of automotive power batteries through systematic steps, ensuring the high efficiency of the entire process from data collection to fault warning. First, through the preprocessing of the original data, the data quality for subsequent analysis is guaranteed. Second, the CNN and RNN models are used to extract static and dynamic feature vectors respectively, and a weighted fusion technology is combined to generate a health feature vector, improving the accuracy of fault prediction. The MLP is used to construct an evaluation model to convert complex features into intuitive health scores, facilitating users to understand the battery state. Furthermore, the SVM is used for classifying fault modes, improving the fault diagnosis speed and supporting rapid response. Finally, warnings are triggered according to the fault modes and treatment suggestions are put forward to ensure that measures are taken before the problem expands, reducing the maintenance cost. Generally speaking, the present invention not only optimizes the performance of electric vehicles, improves the user experience, but also realizes a win-win situation of safety, efficiency and economy.

[0029] Example 2. Referring to Table 1, this is the second embodiment of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a method for monitoring the health state of automotive power batteries based on cloud computing are given. In order to verify the innovation and advantages of the method for monitoring the health state of automotive power batteries based on cloud computing in terms of fault mode classification and health state evaluation, the following experiment was designed. The experimental scenario was set as the monitoring of the health state and fault warning of the power batteries of a certain electric vehicle fleet. 50 new energy electric vehicles were selected, and the method of the present invention (experimental group) and the existing technology method (control group) were used to monitor the health state of the power batteries respectively.

[0030] Specific implementation process: 1. Data collection and preprocessing: The key data of the power battery of each vehicle is collected through built-in sensors, including voltage, current, temperature, SOC (state of charge) and SOH (health state). The sampling frequency is set to 1 Hz and recorded continuously for 7 days. The collected data is cleaned, missing values are processed, and the data integrity is repaired using the interpolation filling method. After data cleaning, the Kalman filtering method is used to remove measurement noise, and the data is normalized by the Z-score normalization method to ensure that the input data has a unified scale.

[0031] 2. Feature extraction: The experimental group inputs the preprocessed data into the convolutional neural network CNN and the recurrent neural network RNN respectively to extract static and dynamic feature vectors. The CNN extracts local static features through convolutional and pooling operations, and generates static feature vectors through the fully connected layer of the residual connection mechanism. The RNN captures dynamic features through time series and generates dynamic feature vectors. The static feature vector and the dynamic feature vector are fused by weighted fusion technology and applied to the formula to generate a health feature vector.

[0032] 3. Health status assessment and fault classification: The health feature vector is input into the health status assessment model constructed by the multi-layer perceptron MLP, and the battery health status score is output. The formula is applied ; The control group directly uses a linear regression model to calculate the health status score; The experimental group classifies fault modes based on the health status score through the support vector machine SVM, generating three classification results: no fault, degradation, and fault. The formula is applied and triggers corresponding early warnings and treatment suggestions.

[0033] 4. Fault early warning and treatment suggestions: Set the real fault mode data as a reference benchmark, and compare the accuracy rate, recall rate, F1 score, and delay detection time parameters of the two methods; Perform a correlation analysis on the health status score and the actual battery SOH value to verify the accuracy of the score.

[0034] Specifically, as shown in Table 1 below: Table 1 Comparative experiment table of the health status monitoring method of automotive power batteries based on cloud computing and the existing technology It can be seen from the table data that the method of the present invention is significantly superior to the existing technology in terms of fault mode classification, health status assessment, and data processing efficiency; 1. Fault mode classification performance: The fault mode classification accuracy rate of the experimental group reaches 96.8%. Compared with 85.2% of the control group, it has increased by 13.6%. At the same time, the recall rate and F1 score have increased by 12.8% and 14.5% respectively, indicating that the method of the present invention can more accurately identify the fault mode of the battery and effectively reduce the missed detection rate.

[0035] 2. Health status score accuracy: The correlation between the health status score of the experimental group and the actual SOH value reaches 0.93, which is significantly higher than 0.78 of the control group. This indicates that the multi-layer perceptron health status assessment model constructed by the method of the present invention can more accurately reflect the health level of the battery.

[0036] 3. Fault detection and early warning efficiency: The average delay time of the fault detection of the method of the present invention is only 2.1 seconds. Compared with 7.8 seconds of the control group, the delay time is reduced by 73.1%. This indicates that the method of the present invention has more advantages in timely identifying battery faults and triggering early warnings, and can effectively improve the safety of vehicle operation.

[0037] 4. Data processing efficiency: The method of the present invention only requires 0.45 seconds for single data processing. Compared with 1.32 seconds of the control group, the efficiency is increased by 65.9%. This benefits from the efficient feature extraction and weighted fusion technology of the convolutional neural network and the recurrent neural network in the present invention.

[0038] 5. Early warning trigger accuracy: The accuracy rate of early warning trigger in the experimental group is 97.2%. Compared with 84.6% of the control group, it is increased by 14.9%. This indicates that the method of the present invention can more accurately judge the health status of the battery and give reasonable treatment suggestions.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for monitoring the health status of a vehicle power battery based on cloud computing, characterized in that: include, Collect automotive power battery data and perform pre-processing; The preprocessed automobile power battery data is input into the convolutional neural network (CNN) model to extract static feature vectors. At the same time, the preprocessed automobile power battery data is input into the recurrent neural network (RNN) model to extract dynamic feature vectors. Based on static feature vectors and dynamic feature vectors, weighted fusion technology is used to generate the health feature vector of the vehicle power battery; A multi-layer perceptron (MLP) is used to build a health status assessment model for a vehicle power battery. The health feature vector of the vehicle power battery is input into the health status assessment model, and the health status assessment result of the vehicle power battery is output. Based on the health status assessment results of the vehicle power battery, the support vector machine (SVM) is used to classify the failure mode; Based on the failure mode classification results, fault warnings are triggered and processing suggestions are generated.

2. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 1, characterized in that: The specific steps of collecting automobile power battery data and preprocessing are as follows: Collect vehicle power battery data through vehicle built-in sensors; The vehicle built-in sensors refer to the vehicle power battery voltage sensor, current sensor, temperature sensor, SOC sensor and SOH sensor; The vehicle power battery data refers to the voltage, current, temperature, charging state and health state of the vehicle power battery; For the collected automobile power battery data, the missing automobile power battery data is processed by interpolation filling method; The padded car power battery data is de-noised using Kalman filtering; The denoised automotive battery data is standardized using the Z-score standardization method.

3. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 2 is characterized in that: The pre-processed automobile power battery data is input into the convolutional neural network (CNN) model to extract static feature vectors, and the specific steps are as follows: The standardized vehicle power battery data is input into the convolutional neural network (CNN) model, and the local static features of the vehicle power battery data at different scales are extracted through convolution operations to obtain the convolutional feature map. ; The convolutional feature map is passed through the pooling layer for dimensionality reduction and maximum pooling operations to reduce the dimension of the feature map and obtain the pooled feature map ; After flattening the pooled feature map into a one-dimensional vector, it is transformed through a fully connected layer that incorporates a residual connection mechanism to generate a static feature vector of the vehicle power battery data, expressed as: ； in, Represents the static feature vector of the vehicle power battery data, represents the weight matrix of the fully connected layer, It means that the three-dimensional feature map after pooling is flattened into a one-dimensional vector. represents the ReLU activation function of the fully connected layer, Represents the bias term of the fully connected layer.

4. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 3 is characterized in that: The pre-processed automobile power battery data is input into the recurrent neural network RNN ​​model to extract the dynamic feature vector, and the specific steps are as follows: Input the standardized vehicle power battery data into the recurrent neural network RNN ​​model; Through the time series structure of the recurrent neural network (RNN) model, the dynamic characteristics of the change of the health status of the vehicle power battery over time are captured, and the memory state of each time step is generated. ; The weighted average pooling method is used to perform weighted aggregation on each memory state to obtain the dynamic feature vector of the vehicle power battery, which is expressed as: ； in, The dynamic feature vector representing the vehicle power battery data, represents the total length of the time series, Indicates The weight of the moment.

5. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 4 is characterized in that: The method uses weighted fusion technology based on static feature vectors and dynamic feature vectors to generate a healthy feature vector of a vehicle power battery. The specific steps are as follows: The static feature vector extracted by the convolutional neural network CNN model and the dynamic feature vector extracted by the recurrent neural network RNN ​​model are combined using weighted fusion technology to obtain the healthy feature vector of the vehicle power battery, which is expressed as: ； in, Indicates the car power battery The health feature vector at each moment, Indicates The static feature vector of the vehicle power battery data extracted at all times, Indicates The dynamic feature vector of the vehicle power battery data extracted at every moment, Indicates The weight coefficient of the static feature vector of the vehicle power battery data at the moment, Indicates The weighting coefficient of the dynamic feature vector of the vehicle power battery data at each moment.

6. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 5 is characterized in that: The multi-layer perceptron MLP is used to construct a vehicle power battery health status assessment model, the vehicle power battery health feature vector is input into the vehicle power battery health status assessment model, and the vehicle power battery health status assessment result is output. The specific steps are: Select Multi-layer Perceptron MLP as the basis of the vehicle power battery health status assessment model; The vehicle power battery health feature vector is input into the vehicle power battery health status assessment model to obtain the battery health status assessment result, which is expressed as: ； in, For automotive power batteries The health status assessment results at all times, Represents the Sigmoid activation function used in the output layer, represents the number of hidden layers, Indicates The output weights of the hidden layers, ReLU activation function used in the hidden layer. Indicates The bias term of the hidden layer, Represents the bias term of the output layer.

7. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 6 is characterized in that: Based on the vehicle power battery health status assessment result, the support vector machine SVM is used to classify the failure mode. The specific steps are as follows: The vehicle power battery health status assessment results are input into the support vector machine (SVM) model to obtain the failure mode classification results, which are expressed as follows: ； in, Indicates The failure mode of the car power battery at all times, represents the symbolic function, represents the weight vector of the support vector machine SVM, Represents the bias term of the support vector machine SVM; According to the failure mode classification results, set the failure mode classification threshold ; when > , then the vehicle power battery mode is classified as a fault-free mode; when = , then the vehicle power battery mode is classified as a decay mode; when < , then the vehicle power battery mode is classified as a fault mode.

8. The method for monitoring the health status of a vehicle power battery based on cloud computing according to claim 7 is characterized in that: The specific steps of triggering a fault warning and generating a processing suggestion based on the fault mode classification result are as follows: According to the fault mode classification results, the threshold range for triggering fault warning is set as ; in, Indicates the minimum limit for triggering fault warning. Indicates the upper limit of triggering fault warning; when ≥ When the battery is in a fault-free mode, no early warning needs to be triggered, and the battery is in a normal state without maintenance or inspection. when ≤ < When the battery is in decline mode, a mild warning is triggered, indicating that the battery is in decline and maintenance inspection is required as planned. when < When the car power battery is in fault mode and a serious warning is triggered, the health status of the car power battery is seriously deteriorated. Stop using the vehicle immediately and contact a professional for maintenance.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the cloud computing-based automotive power battery health status monitoring method described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cloud computing-based automotive power battery health status monitoring method described in any one of claims 1 to 8 are implemented.

Images