Intelligent management method and device for lithium battery of two-wheel electric vehicle

Abstract

The application discloses a kind of two-wheel electric vehicle lithium battery intelligent management method and device, the method includes: obtaining the health state SOH data of lithium battery;According to the health state SOH data, data preprocessing is carried out, and state evaluation input data is obtained;According to the state evaluation input data, SOH evaluation model construction is carried out, and health state characteristic parameter is obtained;According to the health state characteristic parameter, loss function is established and model parameter is optimized, and the SOH evaluation model of minimum loss function value is obtained;According to the SOH evaluation model and RUL relationship model, remaining life prediction calculation is carried out, and the remaining useful life RUL of lithium battery is obtained;According to the remaining useful life RUL, combined with real-time driving data, dynamic health state quantitative analysis is carried out, lithium battery aging early warning signal is generated, and the aging early warning signal is sent to user end.This method can realize real-time accurate lithium battery aging early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium batteries, and in particular to an intelligent management method and device for lithium batteries of a two-wheeled electric vehicle. Background Art

[0002] With rising environmental awareness and the rapid development of the electric vehicle market, lithium batteries are gradually replacing traditional lead-acid batteries as the core power source for two-wheeled electric vehicles. Lithium batteries offer a long service life and high energy density, but due to the influence of various factors during use, such as temperature, state of charge, and the number of charge and discharge cycles, their state of health (SOH) gradually deteriorates, ultimately leading to performance degradation or even failure. Therefore, for electric vehicle users, accurately predicting the remaining useful life (RUL) of lithium batteries and real-time monitoring of their health status have become urgent needs. Early warning of battery aging can help users perform battery replacement or maintenance in a timely manner, avoiding electric vehicle failures caused by excessive battery aging and ensuring safe and convenient riding.

[0003] In existing technologies, battery health is assessed through monitoring methods based on parameters such as battery voltage, current, temperature, and state of charge (SOC). By monitoring the battery's charge and discharge processes and combining them with certain mathematical models, existing technologies can preliminarily calculate the state of health (SOH) of lithium batteries and predict the remaining useful life (RUL) based on the relationship between the battery's health and the remaining useful life (RUL).

[0004] In summary, existing technologies do not dynamically adjust driving data and battery health status in real time, and lack timely and accurate aging warning functions. They are unable to issue battery aging warnings to users in advance, resulting in potential battery failure risks. Summary of the Invention

[0005] The present invention provides an intelligent management method and device for lithium batteries of a two-wheeled electric vehicle, so as to realize real-time and accurate lithium battery aging warning.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides an intelligent management method for lithium batteries of two-wheeled electric vehicles, comprising:

[0007] Obtain the health status SOH data of lithium batteries;

[0008] Performing data preprocessing based on the health status SOH data to obtain status assessment input data;

[0009] Based on the state assessment input data, a SOH assessment model is constructed to obtain health state characteristic parameters;

[0010] According to the health status characteristic parameters, a loss function is established and model parameters are optimized to obtain a SOH evaluation model that minimizes the loss function value;

[0011] According to the SOH evaluation model and the RUL relationship model, a remaining life prediction calculation is performed to obtain the remaining service life RUL of the lithium battery;

[0012] According to the remaining service life RUL, combined with real-time driving data, a dynamic health status quantitative analysis is performed to generate a lithium battery aging warning signal, and the aging warning signal is sent to the user end.

[0013] As an optional implementation manner, obtaining the state of health (SOH) data of the lithium battery includes:

[0014] According to the data acquisition device, multi-source data is collected synchronously to obtain the voltage, current, temperature and state of charge (SOC) of the lithium battery;

[0015] Perform noise filtering and outlier elimination processing based on the voltage, the current, the temperature, and the state of charge (SOC) to obtain state of health (SOH) data of the lithium battery;

[0016] The calculation formula for noise filtering and outlier removal is as follows:

[0017]

[0018] in, for The data after processing, is the original data sequence, is the sliding window weight coefficient, is the window length.

[0019] As an optional implementation manner, performing data preprocessing based on the health status SOH data to obtain status assessment input data includes:

[0020] Performing data standardization processing on the health status SOH data to obtain standard status SOH data;

[0021] Extracting features based on the standard state SOH data to obtain health status features of the lithium battery;

[0022] According to the health status characteristics, feature vectorization processing is performed to obtain a status assessment input vector.

[0023] As an optional implementation, the SOH assessment model is constructed based on the state assessment input data to obtain health state characteristic parameters, including:

[0024] According to the state evaluation input data, extracting time domain features through a convolutional neural network to generate a time domain feature vector;

[0025] Extracting frequency domain features through a convolutional neural network based on the state evaluation input data to generate a frequency domain feature vector;

[0026] Performing feature fusion based on the time domain feature vector and the frequency domain feature vector to obtain a spatiotemporal feature vector;

[0027] The spatiotemporal feature vector is input into a bidirectional long short-term memory network to capture the timing dependency of the charge and discharge cycles and obtain health state feature parameters.

[0028] As an optional implementation, the hidden state update formula according to the bidirectional long short-term memory network is as follows:

[0029]

[0030] in, Represents vector concatenation operation, express Input features at all times, Represents the hidden state of the forward LSTM; represents the hidden state of the reverse LSTM; express The hidden state of the moment; and Represent the forward LSTM and backward LSTM networks respectively.

[0031] As an optional implementation, establishing a loss function and optimizing model parameters based on the health status characteristic parameters to obtain an SOH evaluation model that minimizes the loss function value includes:

[0032] Performing a composite loss calculation based on the health status characteristic parameters to obtain a composite loss value;

[0033] Optimizing model parameters according to the composite loss value to obtain optimized model weights;

[0034] Dynamically adjust the learning rate according to the model weight to obtain a dynamic learning rate during training;

[0035] Model training and verification are performed according to the dynamic learning rate to obtain an SOH evaluation model that minimizes the loss function value.

[0036] As an optional implementation, the calculation formula for the composite loss is as follows:

[0037]

[0038] in, and is the weight coefficient, Indicates the variance of the prediction results; represents the calculated composite loss value; M is the total number of health status characteristic parameters, For the The predicted health status SOH value of samples, For the The true health status SOH value of the samples.

[0039] As an optional implementation manner, performing a remaining life prediction calculation based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery includes:

[0040] According to the SOH evaluation model, the health status characteristic parameters are input to perform SOH value evaluation to obtain the SOH evaluation value of the lithium battery;

[0041] Calculating the remaining service life (RUL) of the lithium battery based on the SOH evaluation value and the RUL relationship model to obtain the remaining service life (RUL) of the lithium battery;

[0042] The calculation formula for the remaining useful life RUL is as follows:

[0043]

[0044] Among them, RUL represents the remaining service life of the lithium battery; Indicates the battery's SOH evaluation value; T indicates the battery's usage time; Indicates the number of charge and discharge cycles of the battery; 、 、 and Represents the regression coefficient obtained through training.

[0045] As an optional implementation, the method of performing a dynamic health status quantitative analysis based on the remaining service life RUL in combination with real-time driving data, generating a lithium battery aging warning signal, and sending the aging warning signal to a user terminal includes:

[0046] Based on the remaining service life RUL, combined with real-time driving data, a dynamic health status quantitative analysis is performed to obtain a battery health status value;

[0047] Determining the degree of aging based on the battery health status value and a preset aging threshold;

[0048] When the battery health status value is greater than a preset aging threshold, a lithium battery aging warning signal is generated and the aging warning signal is sent to a user terminal;

[0049] The real-time driving data includes driving distance data, battery load data, temperature data, and riding speed data.

[0050] In a second aspect, the present invention provides an intelligent management device for lithium batteries of a two-wheeled electric vehicle, comprising:

[0051] Data acquisition module, used to obtain the health status SOH data of lithium batteries;

[0052] A data preprocessing module is used to perform data preprocessing based on the health status SOH data to obtain status assessment input data;

[0053] A model building module is used to build a SOH assessment model based on the state assessment input data to obtain health state characteristic parameters;

[0054] A model optimization module is used to establish a loss function and optimize model parameters according to the health status characteristic parameters to obtain an SOH evaluation model that minimizes the loss function value;

[0055] A life prediction module is used to perform a remaining life prediction calculation based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery;

[0056] The early warning module is used to perform dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generate a lithium battery aging early warning signal, and send the aging early warning signal to the user end.

[0057] In a third aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned intelligent management methods for lithium batteries of two-wheeled electric vehicles.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The application provides a kind of two-wheel electric vehicle lithium battery intelligent management method, comprising: obtaining the health state SOH data of lithium battery;According to the health state SOH data, data preprocessing is carried out, and state evaluation input data is obtained;According to the state evaluation input data, SOH evaluation model construction is carried out, and health state characteristic parameter is obtained;According to the health state characteristic parameter, loss function is established and model parameter is optimized, and the SOH evaluation model of minimum loss function value is obtained;According to the SOH evaluation model and RUL relationship model, remaining life prediction calculation is carried out, and the remaining useful life RUL of lithium battery is obtained;According to the remaining useful life RUL, combined with real-time driving data, dynamic health state quantitative analysis is carried out, lithium battery aging early warning signal is generated, and the aging early warning signal is sent to user end.

[0060] In the application, by obtaining the health state SOH data of lithium battery and carrying out data preprocessing, state evaluation input data is generated, then health state feature extraction is carried out through SOH evaluation model and model parameters are optimized, to obtain the SOH evaluation model of minimum loss function value. Then, according to the evaluation model and the remaining useful life (RUL) relationship model, the prediction calculation of the remaining useful life RUL of lithium battery is carried out. At the same time, combined with real-time driving data, dynamic health state quantitative analysis is carried out, and lithium battery aging early warning signal is generated, which is sent to user end. The application can monitor the health state of battery in real time, accurately predict the remaining useful life of battery, and timely issue aging early warning according to the dynamic change of battery, to avoid the failure caused by excessive aging of battery. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 It is a kind of two-wheel electric vehicle lithium battery intelligent management method flow chart provided by the embodiment of the application;

[0062] Figure 2 It is a kind of two-wheel electric vehicle lithium battery intelligent management device structure schematic diagram provided by the embodiment of the application. DETAILED DESCRIPTION

[0063] With rising environmental awareness and the rapid development of the electric vehicle market, lithium batteries are gradually replacing traditional lead-acid batteries as the core power source for two-wheeled electric vehicles. Lithium batteries offer a long service life and high energy density, but due to the influence of various factors during use, such as temperature, state of charge, and the number of charge and discharge cycles, their state of health (SOH) gradually deteriorates, ultimately leading to performance degradation or even failure. Therefore, for electric vehicle users, accurately predicting the remaining useful life (RUL) of lithium batteries and real-time monitoring of their health status have become urgent needs. Early warning of battery aging can help users perform battery replacement or maintenance in a timely manner, avoiding electric vehicle failures caused by excessive battery aging and ensuring safe and convenient riding.

[0064] In existing technologies, battery health is assessed through monitoring methods based on parameters such as battery voltage, current, temperature, and state of charge (SOC). By monitoring the battery's charge and discharge processes and combining them with certain mathematical models, existing technologies can preliminarily calculate the state of health (SOH) of lithium batteries and predict the remaining useful life (RUL) based on the relationship between the battery's health and the remaining useful life (RUL).

[0065] In summary, existing technologies do not dynamically adjust driving data and battery health status in real time, and lack timely and accurate aging warning functions. They are unable to issue battery aging warnings to users in advance, resulting in potential battery failure risks.

[0066] In order to solve the above problems, the present invention provides an intelligent management method and device for lithium batteries of a two-wheeled electric vehicle, so as to achieve real-time and accurate lithium battery aging warning.

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] Reference Figure 1 The first embodiment of the present invention provides an intelligent management method for lithium batteries of a two-wheeled electric vehicle, comprising the following steps:

[0069] S11, obtaining the health status SOH data of the lithium battery;

[0070] S12, performing data preprocessing based on the health status SOH data to obtain status assessment input data;

[0071] S13, constructing a SOH assessment model based on the state assessment input data to obtain health state characteristic parameters;

[0072] S14, establishing a loss function and optimizing model parameters based on the health status characteristic parameters to obtain a SOH evaluation model that minimizes the loss function value;

[0073] S15, performing a remaining life prediction calculation based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery;

[0074] S16, performing a dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generating a lithium battery aging warning signal, and sending the aging warning signal to a user terminal.

[0075] In step S11, obtaining the state of health (SOH) data of the lithium battery includes:

[0076] According to the data acquisition device, multi-source data is collected synchronously to obtain the voltage, current, temperature and state of charge (SOC) of the lithium battery;

[0077] Perform noise filtering and outlier elimination processing based on the voltage, the current, the temperature, and the state of charge (SOC) to obtain state of health (SOH) data of the lithium battery;

[0078] The calculation formula for noise filtering and outlier removal is as follows:

[0079]

[0080] in, for The data after processing, is the original data sequence, is the sliding window weight coefficient, is the window length.

[0081] It should be noted that data acquisition and processing refers to extracting reliable input data from raw state-of-health (SOH) data through noise filtering and outlier removal. This process plays a vital role in battery health management, providing high-quality data support for subsequent health assessments and remaining useful life (RUL) predictions. In this embodiment of the present invention, data acquisition and processing is based on battery voltage, current, temperature, and state-of-charge (SOC) data. Noise filtering is performed using a sliding window weighted averaging method, and an outlier detection algorithm is used to remove illogical data.

[0082] It's worth noting that noise filtering involves processing raw data using a sliding window weighted averaging method to smooth battery voltage, current, temperature, and state of charge (SOC) data and remove high-frequency noise. By weighted averaging each data point, the sliding window stabilizes the processed data and avoids errors caused by transient fluctuations. Outlier removal involves detecting and removing outliers in the battery health data by setting a threshold or employing statistical methods. By removing unreasonable extreme values ​​or unexpected anomalies, the data used for subsequent processing is more reliable and effective. In this embodiment of the present invention, the standard deviation method is used. Of course, other methods, such as box plots, can also be used for outlier removal, which is not limited by the present invention. Through the above steps, data acquisition and processing can effectively improve the quality of SOH data, providing more accurate and stable data support for subsequent SOH assessment model training and remaining useful life (RUL) prediction, ensuring that the battery management system can accurately predict the battery's health and remaining useful life in real time.

[0083] In step S12, data preprocessing is performed based on the health status SOH data to obtain status assessment input data, including:

[0084] Performing data standardization processing on the health status SOH data to obtain standard status SOH data;

[0085] Extracting features based on the standard state SOH data to obtain health status features of the lithium battery;

[0086] According to the health status characteristics, feature vectorization processing is performed to obtain a status assessment input vector.

[0087] It should be noted that data preprocessing involves standardizing, extracting features, and quantizing lithium battery state-of-health (SOH) data to extract high-quality input data for status assessment. This process plays a crucial role in lithium battery health management, providing reliable data support for subsequent health assessments and remaining useful life (RUL) predictions. In embodiments of the present invention, data preprocessing includes steps such as data standardization, feature extraction, and feature quantization. In lithium battery management systems, data preprocessing can significantly improve data quality and ensure the accuracy and stability of subsequent assessments.

[0088] It's worth noting that data normalization involves standardizing the raw State of Health (SOH) data so that all feature data is processed at the same scale. This process converts the data to a zero-mean, unit-variance format, eliminating dimensional differences between features and ensuring accuracy and fairness during subsequent assessment model training. Feature extraction involves extracting key features that reflect the lithium battery's state of health from the standardized SOH data. By performing statistical calculations (including maximum, minimum, mean, and variance) on parameters such as battery voltage, current, temperature, and state of charge (SOC), feature information useful for health assessment is extracted. These extracted features provide important input data for subsequent model training. Feature vectorization converts the extracted SOH features into a unified feature vector format for processing by the subsequent state assessment model. By organizing multiple feature data into a vector format, this process ensures that each battery's state of health assessment can be analyzed and predicted based on a standardized and unified set of input data. Through the above steps, data preprocessing operations can effectively improve the quality of battery health status data, ensuring that in the subsequent health status assessment and remaining useful life (RUL) prediction process, the model can accurately calculate based on high-quality input data, thereby providing strong data support for real-time monitoring and management of batteries.

[0089] In step S13, the SOH assessment model is constructed based on the state assessment input data to obtain health state characteristic parameters, including:

[0090] According to the state evaluation input data, extracting time domain features through a convolutional neural network to generate a time domain feature vector;

[0091] Extracting frequency domain features through a convolutional neural network based on the state evaluation input data to generate a frequency domain feature vector;

[0092] Performing feature fusion based on the time domain feature vector and the frequency domain feature vector to obtain a spatiotemporal feature vector;

[0093] The spatiotemporal feature vector is input into a bidirectional long short-term memory network to capture the timing dependency of the charge and discharge cycles and obtain health state feature parameters.

[0094] It should be noted that the extraction of health status characteristic parameters refers to the extraction of lithium battery health status characteristics from status assessment input data using a deep learning algorithm. This operation plays an important role in lithium battery health assessment, providing key feature data for subsequent health status assessment and remaining useful life (RUL) prediction. In an embodiment of the present invention, the health status feature extraction operation is based on battery health status data and is implemented through steps such as a convolutional neural network (CNN) and a bidirectional long short-term memory network (Bi-LSTM). In lithium battery health management, the use of deep learning algorithms can significantly improve the accuracy and stability of health status assessments, ensuring the accuracy of subsequent processing.

[0095] It's worth noting that time-domain feature extraction involves using a convolutional neural network (CNN) to process time-series data such as battery voltage, current, and temperature in the health assessment input data to extract time-domain features of the battery's health status. Through multi-layer convolution and pooling operations, convolutional neural networks effectively capture both local and global features in the data, providing high-quality input data for subsequent feature fusion. Frequency-domain feature extraction involves performing a frequency-domain transformation on the health assessment input data using Fourier transforms or other frequency-domain analysis methods to extract frequency-domain features of the battery's health status. Frequency-domain features reveal the battery's behavior at different frequencies, helping to identify potential aging patterns and providing more comprehensive health assessment information. Feature fusion involves fusing time-domain and frequency-domain feature vectors to generate a spatiotemporal feature vector. By fusing these two types of features, we can comprehensively consider the battery's behavior in both the time and frequency domains, resulting in a more comprehensive health status profile. This spatiotemporal feature vector is then input into a bidirectional long short-term memory (Bi-LSTM) network to capture the temporal dependencies during the charge and discharge cycles. Bi-LSTM processes time-varying sequence data on battery health status, capturing long- and short-term dependencies in charge and discharge cycles. This allows for further optimization of health feature extraction to yield accurate health feature parameters. Through these steps, health feature extraction generates highly accurate health feature parameters, providing reliable support for subsequent SOH assessments and RUL predictions, ensuring the accuracy and stability of battery health assessments.

[0096] The hidden state update formula of the bidirectional long short-term memory network is as follows:

[0097]

[0098] in, Represents vector concatenation operation, express Input features at all times, Represents the hidden state of the forward LSTM; represents the hidden state of the reverse LSTM; express The hidden state of the moment; and Represent the forward LSTM and backward LSTM networks respectively.

[0099] It should be noted that the hidden state update operation involves time-series modeling of battery health status data using a bidirectional long short-term memory (Bi-LSTM) network, thereby capturing the temporal dependencies during the battery's charge and discharge cycles. This operation plays a crucial role in lithium battery health assessment, providing critical temporal information for subsequent health status feature extraction and remaining useful life (RUL) prediction. In this embodiment of the present invention, the Bi-LSTM network, through the combined action of forward LSTM and backward LSTM, effectively captures both long-term and short-term dependencies in battery data, enhancing the ability to predict battery health status.

[0100] The forward LSTM operation involves processing the current moment's input features with the previous moment's hidden state through the LSTM network to update the current moment's hidden state. The forward LSTM captures long-term data dependencies and transfers past battery health information to better understand the current moment's battery state. The reverse LSTM operation involves processing the current moment's input features with the next moment's hidden state through the ReLSTM (reverse LSTM) network to update the current moment's reverse hidden state. The reverse LSTM captures dependencies from future moments to the current moment, further enhancing the modeling capabilities of time series data. The hidden state concatenation operation concatenates the hidden states of the forward and reverse LSTMs to obtain the current moment's spatiotemporal feature vector. This concatenation operation integrates information from the forward and reverse LSTMs, enabling the model to simultaneously consider past and future temporal dependencies, thereby providing more accurate battery health characteristics. By inputting this concatenated spatiotemporal feature vector into a bidirectional long short-term memory (Bi-LSTM) network, the model can capture the temporal dependencies of charge and discharge cycles from both forward and reverse perspectives, generating accurate health characteristic parameters. This operation enables the model to process complex time series data and more accurately predict the battery health status.

[0101] In step S14, the loss function is established and the model parameters are optimized according to the health status characteristic parameters to obtain the SOH evaluation model that minimizes the loss function value, including:

[0102] Performing a composite loss calculation based on the health status characteristic parameters to obtain a composite loss value;

[0103] Optimizing model parameters according to the composite loss value to obtain optimized model weights;

[0104] Dynamically adjust the learning rate according to the model weight to obtain a dynamic learning rate during training;

[0105] Model training and verification are performed according to the dynamic learning rate to obtain an SOH evaluation model that minimizes the loss function value.

[0106] It should be noted that loss function optimization involves optimizing the performance of the SOH assessment model from the battery's state of health characteristic parameters through compound loss calculation and dynamic learning rate adjustment. This operation plays a crucial role in lithium battery health assessment, providing an accurate model for subsequent state of health prediction and remaining useful life (RUL) prediction. In an embodiment of the present invention, loss function optimization is implemented through steps such as compound loss function, dynamic learning rate adjustment, and backpropagation. In lithium battery management systems, the use of this optimization algorithm can significantly improve the accuracy and stability of the model, ensuring the efficiency and reliability of subsequent health assessments.

[0107] The composite loss calculation operation refers to calculating a composite loss value by weighted summation according to the health state characteristic parameters. The composite loss value combines multiple loss functions such as mean square error (MSE) and variance (Var) to measure the gap between the model prediction value and the true value, and considers the stability of the prediction result. Through this operation, the model can not only accurately predict the health state, but also stably output the result. The model parameter optimization operation refers to adjusting the parameters of the model by an optimization algorithm according to the composite loss value to minimize the loss function value. In the embodiment of the present application, the optimization algorithm used is the Adam optimization algorithm, of course, according to the actual application scene and user demand, the present application can also use other algorithms such as stochastic gradient descent algorithm, the present application does not limit this. The learning rate dynamic adjustment operation refers to dynamically adjusting the learning rate of the model according to the change of the loss function in the training process. The dynamic adjustment of the learning rate can help the model avoid falling into a local optimal solution and speed up the convergence of the model. In the embodiment of the present application, the dynamic adjustment algorithm used is the adaptive learning rate algorithm, of course, according to the actual application scene and user demand, the present application can also use other algorithms such as learning rate decay strategy, the present application does not limit this. The model training and verification operation refers to training and verifying the SOH evaluation model according to the optimized model weight and dynamic learning rate. Through multiple iterations on the training set, the training loss is calculated and the generalization ability of the model is evaluated through the validation set, and finally the SOH evaluation model with the minimum loss function value is obtained. This operation ensures the accuracy and reliability of the model under various data conditions. In the embodiment of the present application, the number of iterations is 500, of course, according to the actual application scene and user demand, other iteration numbers such as 300, 1000 can also be used, the present application does not limit this.

[0108] The calculation formula of the composite loss calculation is as follows:

[0109]

[0110] Among them, and are weight coefficients, represents the variance of the prediction result; represents the calculated composite loss value; M is the total number of health state characteristic parameters, is the predicted health state SOH value of the i th sample, is the true health state SOH value of the i th sample.

[0111] ​​It should be noted that the composite loss calculation operation refers to the calculation of a loss function using a weighted summation method that combines the mean squared error (MSE) and the variance of the prediction results. This operation plays a vital role in lithium battery state of health assessment, ensuring model accuracy while reducing the instability of the prediction results. In embodiments of the present invention, the composite loss calculation operation combines the difference between the predicted and true SOH values, as well as the variance of the prediction results, to achieve more accurate and stable SOH assessment model optimization.

[0112] It is worth noting that the mean square error calculation operation refers to calculating the square of the difference between the predicted health state SOH value and the actual health state SOH value and taking the average to obtain the size of the error. This operation can measure the deviation between the model prediction result and the actual health state, helping the model to optimize and reduce the prediction error. The variance calculation operation refers to measuring the degree of fluctuation of the prediction result by calculating the variance of the predicted value. The variance reflects the consistency of the model output results. A lower variance means that the model's prediction results are more stable and adaptable to changes in different situations. The weighted loss function calculation operation refers to obtaining a composite loss value by weightedly combining the mean square error and variance. Through the weight coefficient and , adjusting the influence of mean squared error and variance on the total loss, ensuring that the model optimizes both prediction accuracy and result stability. Through these steps, the composite loss calculation operation can balance model accuracy and stability, ensuring the optimization effect of the health state (SOH) assessment model during training, providing reliable loss function support for subsequent model training and optimization, and thus improving the accuracy and stability of the battery health management system.

[0113] In step S15, the remaining service life prediction calculation is performed based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery, including:

[0114] According to the SOH evaluation model, the health status characteristic parameters are input to perform SOH value evaluation to obtain the SOH evaluation value of the lithium battery;

[0115] Calculating the remaining service life (RUL) of the lithium battery based on the SOH evaluation value and the RUL relationship model to obtain the remaining service life (RUL) of the lithium battery;

[0116] The calculation formula for the remaining useful life RUL is as follows:

[0117]

[0118] Among them, RUL represents the remaining service life of the lithium battery; Indicates the battery's SOH evaluation value; T indicates the battery's usage time; Indicates the number of charge and discharge cycles of the battery; 、 、 and Represents the regression coefficient obtained through training.

[0119] It should be noted that the remaining useful life (RUL) calculation operation refers to the combination of the SOH assessment model and the RUL relationship model to predict the remaining useful life from the health status characteristics of the battery. This operation plays an important role in lithium battery health management and can provide accurate time warnings for subsequent battery maintenance and replacement decisions. In this embodiment of the present invention, the remaining useful life calculation combines the SOH assessment value with the battery usage time (T) and the number of charge and discharge cycles (cycles) to obtain an accurate RUL prediction value through a regression model.

[0120] Among them, the SOH evaluation operation refers to inputting the health status characteristic parameters into the SOH evaluation model to obtain the battery's SOH evaluation value. This evaluation value represents the current health status of the battery, can reflect the degree of battery degradation, and provide basic data for the remaining service life prediction. Based on the battery's SOH evaluation value and the established RUL relationship model, the remaining service life (RUL) is calculated. The RUL calculation is performed by taking the SOH value, battery usage time (T) and the number of charge and discharge cycles (cycles) as inputs, and combining the coefficients in the regression model ( 、 、 and ), predicting the remaining service life of the battery. The regression model optimization operation is to optimize the regression coefficient ( 、 、 and ), making the RUL prediction closer to the actual data. By minimizing the prediction error, the model can continuously adjust the coefficients to improve the accuracy and reliability of the remaining useful life prediction. Through the above steps, the remaining useful life (RUL) calculation operation can accurately predict the remaining useful life of the battery by combining the battery's health status, usage time, and number of charge and discharge cycles. This provides critical data support for the battery management system, helping users make timely battery maintenance or replacement decisions, and ensuring the safe and efficient operation of electric vehicles.

[0121] In step S16, the method of performing a dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generating a lithium battery aging warning signal, and sending the aging warning signal to the user terminal includes:

[0122] Based on the remaining service life RUL, combined with real-time driving data, a dynamic health status quantitative analysis is performed to obtain a battery health status value;

[0123] Determining the degree of aging based on the battery health status value and a preset aging threshold;

[0124] When the battery health status value is greater than a preset aging threshold, a lithium battery aging warning signal is generated and the aging warning signal is sent to a user terminal;

[0125] The real-time driving data includes driving distance data, battery load data, temperature data, and riding speed data.

[0126] It should be noted that the dynamic health status quantitative analysis operation refers to the real-time assessment of the health status of the lithium battery by combining the remaining service life RUL of the battery and real-time driving data, and generating a corresponding aging warning signal. This operation plays an important role in lithium battery management. It can provide early warning of the battery aging process and help users perform battery maintenance or replacement in a timely manner. In an embodiment of the present invention, the dynamic health status quantitative analysis operation uses real-time driving data, including parameters such as driving distance, battery load, temperature and riding speed, to comprehensively evaluate the battery health status, thereby generating a battery aging warning signal.

[0127] The health status calculation operation involves performing a dynamic quantitative analysis of the battery's health status based on the battery's remaining useful life (RUL) in combination with real-time driving data (such as driving distance, battery load, temperature, and riding speed) to obtain a battery health status value. By comprehensively considering the battery's operating conditions and current health status, it can more accurately reflect the battery's actual usage status. The aging degree determination operation involves determining the battery's aging degree based on the battery's health status value and a preset aging threshold. By comparing the value with the preset threshold, it is determined whether the battery has entered an aging state. If the health status value is lower than the threshold, it indicates that the battery has begun to age and requires maintenance or replacement. In this embodiment of the present invention, the preset aging threshold is 0.8. Of course, it can also be set to other values ​​such as 0.7 or 0.9 depending on the actual application scenario and user needs, and this is not limited by the present invention. When the battery health status value exceeds the preset aging threshold, a lithium battery aging warning signal is generated and transmitted to the user terminal. The warning signal notifies the user that the battery is about to reach a critical lifespan value, prompting the user to take appropriate measures as soon as possible. This warning signal is transmitted to the user terminal (such as a smartphone or in-vehicle system) via a communication network, providing real-time and effective battery maintenance reminders. Through the above steps, the dynamic health status quantitative analysis operation can monitor the battery health status in real time and generate aging warning signals in a timely manner based on real-time driving data, helping users predict battery aging trends and ensure the continued safe operation of the battery.

[0128] Reference Figure 2The second embodiment of the present invention provides an intelligent management device for lithium batteries of a two-wheeled electric vehicle, comprising:

[0129] Data acquisition module, used to obtain the health status SOH data of lithium batteries;

[0130] A data preprocessing module is used to perform data preprocessing based on the health status SOH data to obtain status assessment input data;

[0131] A model building module is used to build a SOH assessment model based on the state assessment input data to obtain health state characteristic parameters;

[0132] A model optimization module is used to establish a loss function and optimize model parameters according to the health status characteristic parameters to obtain an SOH evaluation model that minimizes the loss function value;

[0133] A life prediction module is used to perform a remaining life prediction calculation based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery;

[0134] The early warning module is used to perform dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generate a lithium battery aging early warning signal, and send the aging early warning signal to the user end.

[0135] It should be noted that the intelligent management device for lithium batteries of two-wheeled electric vehicles provided in an embodiment of the present invention is used to execute all the process steps of the intelligent management method for lithium batteries of two-wheeled electric vehicles in the above embodiment. The working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0136] In summary, the present invention obtains lithium battery state of health (SOH) data and performs data preprocessing to generate state assessment input data. The SOH assessment model then extracts state of health features and optimizes model parameters to obtain an SOH assessment model that minimizes the loss function value. Next, based on this assessment model and the remaining useful life (RUL) relationship model, the lithium battery's RUL is predicted and calculated. Simultaneously, dynamic state of health quantitative analysis is performed in conjunction with real-time driving data, and a lithium battery aging warning signal is generated and transmitted to the user. The present invention can monitor the battery's health status in real time, accurately predict the battery's remaining useful life, and promptly issue aging warnings based on the battery's dynamic changes, thereby avoiding battery failures caused by excessive aging.

[0137] The embodiment of the present invention further provides a terminal device. The terminal device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an intelligent management program for lithium batteries of two-wheeled electric vehicles. When the processor executes the computer program, the steps of the above-mentioned intelligent management method for lithium batteries of two-wheeled electric vehicles are implemented, such as Figure 1 Alternatively, the processor implements the functions of the modules / units in the above-mentioned system embodiments when executing the computer program.

[0138] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0139] The terminal device may be a computing device such as a desktop computer, laptop, PDA, or smart tablet. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that the aforementioned components are merely examples of terminal devices and do not constitute a limitation of the terminal device. The terminal device may include more or fewer components than those described above, or a combination of certain components or different components. For example, the terminal device may also include input / output devices, network access devices, buses, and the like.

[0140] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the terminal device and connects various parts of the entire terminal device using various interfaces and lines.

[0141] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the terminal device by running or executing the computer program and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), etc.; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a nonvolatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0142] The modules / units integrated in the terminal device, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can realize the steps of the above-mentioned various method embodiments when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or system, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium, etc. that can carry the computer program code. It should be noted that the contents included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electric carrier signals and telecommunication signals.

[0143] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0144] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An intelligent management method for lithium batteries of two-wheeled electric vehicles, characterized in that: include: Obtain the health status SOH data of lithium batteries; Performing data preprocessing based on the health status SOH data to obtain status assessment input data; Based on the state assessment input data, a SOH assessment model is constructed to obtain health state characteristic parameters; According to the health status characteristic parameters, a loss function is established and model parameters are optimized to obtain a SOH evaluation model that minimizes the loss function value; According to the SOH evaluation model and the RUL relationship model, a remaining life prediction calculation is performed to obtain the remaining service life RUL of the lithium battery; Based on the remaining service life RUL, combined with real-time driving data, a dynamic health status quantitative analysis is performed to generate a lithium battery aging warning signal, and the aging warning signal is sent to the user end; The step of obtaining the state of health (SOH) data of the lithium battery includes: According to the data acquisition device, multi-source data is collected synchronously to obtain the voltage, current, temperature and state of charge (SOC) of the lithium battery; Perform noise filtering and outlier elimination processing based on the voltage, the current, the temperature, and the state of charge (SOC) to obtain state of health (SOH) data of the lithium battery; The calculation formula for noise filtering and outlier removal is as follows: in, for The data after processing at any time, is the original data sequence, is the sliding window weight coefficient, is the window length; The SOH assessment model is constructed based on the state assessment input data to obtain health state characteristic parameters, including: According to the state evaluation input data, extracting time domain features through a convolutional neural network to generate a time domain feature vector; Extracting frequency domain features through a convolutional neural network based on the state evaluation input data to generate a frequency domain feature vector; Performing feature fusion based on the time domain feature vector and the frequency domain feature vector to obtain a spatiotemporal feature vector; Inputting the spatiotemporal feature vector into a bidirectional long short-term memory network to capture the timing dependency of the charge and discharge cycles and obtain health state feature parameters; The hidden state update formula of the bidirectional long short-term memory network is as follows: in, Represents vector concatenation operation, express Input features at all times, Represents the hidden state of the forward LSTM; represents the hidden state of the reverse LSTM; express The hidden state of the moment; and Represent the forward LSTM and reverse LSTM networks respectively; The method of establishing a loss function and optimizing model parameters based on the health status characteristic parameters to obtain an SOH evaluation model that minimizes the loss function value includes: Performing a composite loss calculation based on the health status characteristic parameters to obtain a composite loss value; Optimizing model parameters according to the composite loss value to obtain optimized model weights; Dynamically adjust the learning rate according to the model weight to obtain a dynamic learning rate during training; Perform model training and verification according to the dynamic learning rate to obtain a SOH evaluation model that minimizes the loss function value; The calculation formula for the composite loss is as follows: in, and is the weight coefficient, Indicates the variance of the prediction results; represents the calculated composite loss value; M is the total number of health status characteristic parameters, For the The predicted health status SOH value of samples, For the The true health status SOH value of the samples.

2. The intelligent management method of lithium batteries for two-wheeled electric vehicles according to claim 1, characterized in that: The step of performing data preprocessing based on the state of health (SOH) data to obtain state assessment input data includes: Performing data standardization processing on the health status SOH data to obtain standard status SOH data; Extracting features based on the standard state SOH data to obtain health status features of the lithium battery; According to the health status characteristics, feature vectorization processing is performed to obtain a status assessment input vector.

3. The intelligent management method of lithium batteries for two-wheeled electric vehicles according to claim 1, characterized in that: The remaining service life prediction calculation is performed based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery, including: According to the SOH evaluation model, the health status characteristic parameters are input to perform SOH value evaluation to obtain the SOH evaluation value of the lithium battery; Calculating the remaining service life (RUL) of the lithium battery based on the SOH evaluation value and the RUL relationship model to obtain the remaining service life (RUL) of the lithium battery; The calculation formula for the remaining useful life RUL is as follows: Among them, RUL represents the remaining service life of the lithium battery; Indicates the battery's SOH evaluation value; T indicates the battery's usage time; Indicates the number of charge and discharge cycles of the battery; 、 、 and Represents the regression coefficient obtained through training.

4. The intelligent management method of lithium batteries for two-wheeled electric vehicles according to claim 1, characterized in that: The method of performing a dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generating a lithium battery aging warning signal, and sending the aging warning signal to the user terminal includes: Based on the remaining service life RUL, combined with real-time driving data, a dynamic health status quantitative analysis is performed to obtain a battery health status value; Determining the degree of aging based on the battery health status value and a preset aging threshold; When the battery health status value is greater than a preset aging threshold, a lithium battery aging warning signal is generated and the aging warning signal is sent to a user terminal; The real-time driving data includes driving distance data, battery load data, temperature data, and riding speed data.

5. An intelligent management device for lithium batteries of two-wheeled electric vehicles, characterized in that: Used to implement the method according to any one of claims 1 to 4, comprising: Data acquisition module, used to obtain the health status SOH data of lithium batteries; A data preprocessing module is used to perform data preprocessing based on the health status SOH data to obtain status assessment input data; A model building module is used to build a SOH assessment model based on the state assessment input data to obtain health state characteristic parameters; A model optimization module is used to establish a loss function and optimize model parameters according to the health status characteristic parameters to obtain an SOH evaluation model that minimizes the loss function value; A life prediction module is used to perform a remaining life prediction calculation based on the SOH evaluation model and the RUL relationship model to obtain the remaining service life RUL of the lithium battery; The early warning module is used to perform dynamic health status quantitative analysis based on the remaining service life RUL and real-time driving data, generate a lithium battery aging early warning signal, and send the aging early warning signal to the user end.

Images