CN116774058B - Battery life prediction method, device, equipment and storage medium - Google Patents
Battery life prediction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN116774058B CN116774058B CN202311044415.4A CN202311044415A CN116774058B CN 116774058 B CN116774058 B CN 116774058B CN 202311044415 A CN202311044415 A CN 202311044415A CN 116774058 B CN116774058 B CN 116774058B
- Authority
- CN
- China
- Prior art keywords
- battery
- data
- target
- life prediction
- life
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Secondary Cells (AREA)
Abstract
The invention relates to the technical field of batteries and discloses a battery life prediction method, device, equipment and storage medium. The battery life prediction method comprises the following steps: acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data; and drawing a first life prediction curve by combining the first life prediction data based on a preset first life prediction function, and drawing a second life prediction curve by combining the second life prediction data based on a preset second life prediction function. The invention can improve the prediction accuracy and reliability of the service life of the battery, thereby providing better use experience for battery manufacturers and users and saving maintenance cost.
Description
Technical Field
The present invention relates to the field of battery technologies, and in particular, to a method, an apparatus, a device, and a storage medium for predicting battery life.
Background
With the rapid development of electronic products and electric vehicles, there is an increasing demand for battery performance and life. How to accurately measure and predict the life of a battery is a problem that is in need of solution in the industry. The existing battery life prediction method is mainly based on cycle times or capacity fading, but prediction accuracy is limited due to the lack of the capability of monitoring the working state of a battery in real time. To solve this problem, battery life prediction based on a Battery Management System (BMS) is becoming a research focus. However, the existing battery life prediction method still has a certain technical problem. The main technical problems in the prior art include:
data acquisition is incomplete: the existing BMS prediction method mainly focuses on information such as voltage, current and temperature of a battery, and other important internal factors such as battery chemical reaction are ignored, so that prediction accuracy can be affected.
The battery life prediction method is not advanced enough: the existing battery life prediction method mostly adopts a simple regression model, so that the prediction precision is insufficient, and the actual application requirements cannot be met. In practical applications, more advanced battery life prediction methods are required to more accurately predict battery life.
Accordingly, there is a need to provide a battery life prediction method to solve the above-mentioned technical problems.
Disclosure of Invention
The present invention provides a battery life prediction method, apparatus, device and storage medium for explaining the above-mentioned technical problems.
The first aspect of the present invention provides a battery life prediction method, comprising:
acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
drawing a first life prediction curve by combining the first life prediction data based on a preset first life prediction function, and drawing a second life prediction curve by combining the second life prediction data based on a preset second life prediction function; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
Performing stitching processing on the first target vector and the second target vector to obtain stitching vectors, and inputting the stitching vectors into a preset convolutional neural network model to extract local feature information to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
acquiring an actual running state of a target battery, carrying out weight analysis on the predicted life parameter of the battery in combination with the actual running state of the target battery to generate a first weight, carrying out weight analysis on the predicted life parameter of the battery in combination with a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of a predicted result, and carrying out target optimization on the predicted life parameter of the battery according to the weight combination to obtain a life predicted result of the target battery.
Optionally, in a first implementation manner of the first aspect of the present invention, the collecting, by each preset sensor, state data of a target battery to obtain battery state data corresponding to the target battery, and performing data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data, where the method includes:
acquiring state data of a target battery through each preset sensor, preprocessing the state data of the target battery acquired from each sensor, and converting the preprocessed state data into a uniform data format to obtain battery state data corresponding to the target battery;
extracting key features of the battery state data through a preset feature extraction algorithm, and screening the battery state data according to different key features to obtain key feature data related to life prediction;
classifying the key feature data related to life prediction based on a preset supervised learning model to obtain first life prediction data and second life prediction data, wherein the supervised learning model is obtained through SVM algorithm training.
Optionally, in a second implementation manner of the first aspect of the present invention, the first life prediction curve is drawn based on a preset first life prediction function and combined with the first life prediction data, and the second life prediction curve is drawn based on a preset second life prediction function and combined with the second life prediction data; and calculating a plurality of first vector values in the first life prediction curve, and calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first and second target vectors based on a preset box diagram rule, comprising:
inputting the first life prediction data into a first prediction function, drawing a first life prediction curve based on a preset visual curve model, extracting characteristic values from the first life prediction curve, calculating a plurality of first vector values according to the characteristic values, and calculating a first target vector based on a preset box diagram rule;
and inputting the second life prediction data into a second prediction function, drawing a second life prediction curve based on a preset visual curve model, extracting characteristic values from the second life prediction curve, calculating a plurality of second vector values according to the characteristic values, and calculating a second target vector based on a preset box diagram rule.
Optionally, in a third implementation manner of the first aspect of the present invention, the stitching processing is performed on the first target vector and the second target vector to obtain a stitched vector, and the stitched vector is input into a preset convolutional neural network model to perform local feature information extraction, so as to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain battery predicted life parameters, wherein the method comprises the following steps:
splicing the first target vector and the second target vector according to a preset rule to obtain a spliced vector, inputting the spliced vector into a preset convolutional neural network model, and extracting local feature information of the spliced vector through a multi-scale convolutional layer of the convolutional neural network model to obtain target feature data;
and inputting the target characteristic data into a multi-layer stack layer of the long and short memory network model based on a preset long and short memory network model for correlation analysis processing to obtain the predicted life parameter of the battery.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the obtaining an actual running state of the target battery, performing weight analysis on the predicted lifetime parameter of the battery in combination with the actual running state of the target battery to generate a first weight, performing weight analysis on the predicted lifetime parameter of the battery in combination with a historical battery data sample to obtain a second weight, performing weight balance optimization on the first weight and the second weight, and performing target optimization on the predicted lifetime parameter of the battery according to the weight combination to obtain a lifetime prediction result of the target battery, where the weight combination has the smallest error of the prediction result, includes:
Acquiring actual running state data of a target battery, wherein the actual running state data comprise working temperature, charge and discharge times and circulation capacity loss;
carrying out weight analysis on predicted life parameters of the battery by combining actual running state data of the target battery to generate a first weight;
acquiring a historical battery data sample, and carrying out weight analysis on the predicted life parameter of the battery by combining the historical battery data sample to obtain a second weight;
weighting cross-validation or genetic algorithm is adopted to carry out weighing optimization between the first weight and the second weight, so as to obtain a weight combination with the minimum error of the predicted result;
and carrying out target optimization on the predicted life parameters of the battery according to the weight combination with the minimum error of the obtained predicted result, and obtaining the life predicted result of the target battery.
A second aspect of the present invention provides a battery life prediction apparatus including:
the acquisition module is used for acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
The first processing module is used for drawing a first life prediction curve based on a preset first life prediction function and combining the first life prediction data, and drawing a second life prediction curve based on a preset second life prediction function and combining the second life prediction data; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
the second processing module is used for performing splicing processing on the first target vector and the second target vector to obtain a spliced vector, inputting the spliced vector into a preset convolutional neural network model to extract local feature information, and obtaining target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
The weight analysis module is used for acquiring the actual running state of the target battery, carrying out weight analysis on the battery predicted life parameter by combining the actual running state of the target battery to generate a first weight, carrying out weight analysis on the battery predicted life parameter by combining a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of the predicted result, and carrying out target optimization on the battery predicted life parameter according to the weight combination to obtain the life predicted result of the target battery.
A third aspect of the present invention provides a battery life prediction apparatus comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the battery life prediction device to perform the battery life prediction method described above.
A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein that, when executed on a computer, cause the computer to perform the above-described battery life prediction method.
In the technical scheme provided by the invention, the beneficial effects are as follows: according to the battery life prediction method, the device, the equipment and the storage medium, the state data of the target battery are collected through the preset sensor, and the first life prediction curve and the second life prediction curve are drawn by means of the prediction function so as to calculate the corresponding target vector. And (5) performing splicing processing, local characteristic information extraction and association analysis to obtain the predicted life parameter of the battery. And combining the actual running state of the target battery, carrying out weight analysis on the historical battery data sample, and carrying out weight combination to obtain a life prediction result of the target battery. The prediction accuracy and reliability of the service life of the battery can be improved, so that better use experience and maintenance cost saving are provided for battery manufacturers and users.
Drawings
FIG. 1 is a schematic diagram of an embodiment of a battery life prediction method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an embodiment of a battery life prediction apparatus according to the present invention.
Detailed Description
The embodiment of the invention provides a battery life prediction method, a device, equipment and a storage medium. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, a specific flow of an embodiment of the present invention is described below with reference to fig. 1, where an embodiment of a battery life prediction method in an embodiment of the present invention includes:
step 101, acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
it is to be understood that the execution subject of the present invention may be a battery life prediction device, and may also be a terminal or a server, which is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example.
Specifically, this process mainly includes the following steps:
data are collected by the sensor: first, state-related data of the battery is monitored and collected in real time by various sensors (e.g., voltage, current, temperature, etc. sensors) preset on the target battery.
Data preprocessing: the collected raw data may have noise or abnormal values, and the quality of the data needs to be improved through data preprocessing and cleaning. Common methods include null removal, outlier detection and processing, data smoothing, and the like.
Data screening: and screening the processed data according to a preset parameter range and sampling frequency, removing irrelevant or redundant data, and retaining state data closely related to the service life of the battery.
Data classification and labeling: the battery state data is classified and divided into different parts, such as first life prediction data (early life stage) and second life prediction data (middle and later life stage), and the data are labeled correspondingly so as to facilitate subsequent analysis and modeling.
Characteristic engineering: feature extraction and feature selection are performed on the screened and classified data, and features related to battery life prediction are creatively designed, for example: overall discharge capacity, battery internal resistance change, rate of temperature rise, etc.
And (3) establishing a model: and respectively training corresponding prediction models according to the first life prediction data and the second life prediction data, such as: linear regression, neural networks, support vector machines, etc.
Model fusion and optimization: and combining a plurality of prediction models to perform model fusion so as to enable a final prediction result to be more accurate. Meanwhile, the model performance is optimized by means of adjusting model parameters, feature selection and the like.
And (3) predicting and monitoring in real time: and applying the trained model to the real-time battery state data, continuously predicting the future life of the battery, and updating the prediction result in real time.
102, drawing a first life prediction curve by combining the first life prediction data based on a preset first life prediction function, and drawing a second life prediction curve by combining the second life prediction data based on a preset second life prediction function; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
specifically, this process can be divided into the following detailed steps:
applying a predictive function: and using the trained first life prediction model and second life prediction model as a first life prediction function and a second life prediction function, and finding corresponding first life prediction data and second life prediction data in the real-time battery state data.
And (3) drawing a prediction curve: and drawing a first life prediction curve according to the first life prediction function and the first life prediction data to reflect the performance change of the battery in an early life stage. And likewise, drawing a second life prediction curve by combining the second life prediction function and the second life prediction data, and reflecting the performance change of the battery in the middle-late life stage.
Calculating vector values: a plurality of key data points are selected on the first life prediction curve and the second life prediction curve, and a first vector value and a second vector value of the points are calculated respectively. The first and second vector values may be gradients, slopes, or curve-fitting parameters, etc.
Creating a box diagram: and according to a preset box diagram rule, carrying out statistical analysis on the calculated first vector value and second vector value, and drawing a box diagram. This may more intuitively represent the performance differences of the battery at different life stages.
The innovation content is as follows: in the process, innovation points such as selecting a specific prediction function according to an actual application scene to improve prediction accuracy can be introduced; introducing customized indexes or algorithms when calculating vector values; and expanding the box diagram, and drawing a mini-version or interactive box diagram and the like.
Deriving a target vector: and generating a corresponding first target vector and a corresponding second target vector according to the calculated first vector value and the second vector value, and taking the first target vector and the second target vector as the final result of battery life prediction. These target vectors can help identify differences in battery performance and formulate corresponding charge-discharge strategies, pre-warnings, battery replacement plans, and the like.
Step 103, performing stitching processing on the first target vector and the second target vector to obtain stitching vectors, and inputting the stitching vectors into a preset convolutional neural network model to extract local feature information to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
specifically, this process can be divided into the following detailed steps:
vector splicing, namely splicing the first target vector and the second target vector to form a unified spliced vector so as to contain information of the battery in different life stages.
And the convolutional neural network model is used for inputting the spliced vector into a preset convolutional neural network model to extract local characteristic information. The convolutional neural network model performs layer-by-layer processing through structures such as a convolutional layer, an activation function, a pooling layer and the like to obtain target characteristic data, and highlights key characteristics of battery performance change. The model may be pre-trained by a convolutional neural network algorithm.
Innovative optimization, namely introducing innovation points in the design or training stage of a convolutional neural network model, such as adding a attention mechanism to enhance the learning and expression of key features, or adjusting network parameters and depth to adapt to different scene requirements.
And the long and short memory network model is used for inputting the target characteristic data into the preset long and short memory network model for carrying out association analysis processing so as to capture the dependency relationship of the battery performance on the time sequence. The model is composed of long and short memory units and has the capability of processing long sequences. And (5) pre-training through a long and short memory network algorithm to obtain the association analysis processing capacity.
The output of the long and short memory network model represents the predicted life parameter of the battery, and the overall trend and potential evolution trend of the battery performance can be comprehensively reflected.
The importance of considering the characteristics of different time points by combining a attention mechanism is added in the application of the long and short memory network model, or the two-way long and short memory network is used for capturing the two-way information of the input sequence.
104, acquiring an actual running state of a target battery, carrying out weight analysis on the predicted life parameter of the battery by combining the actual running state of the target battery to generate a first weight, carrying out weight analysis on the predicted life parameter of the battery by combining a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of the predicted result, and carrying out target optimization on the predicted life parameter of the battery according to the weight combination to obtain a life predicted result of the target battery.
Specifically, this process can be divided into the following detailed steps:
actual running state acquisition: the actual running state of the target battery, such as voltage, current, temperature and the like, is collected and monitored in real time.
Weight analysis: and carrying out weight analysis on the predicted life parameter of the battery by combining the actual running state of the target battery to generate a first weight. And meanwhile, combining historical battery data samples, and carrying out weight analysis on the predicted life parameters of the battery to obtain a second weight.
Weight optimization algorithm: a weight optimization algorithm, such as a genetic algorithm, particle swarm optimization, etc., is creatively introduced to perform weighing optimization on the first weight and the second weight, and find a weight combination with the smallest prediction result error.
Weight trade-off optimization: and carrying out weighing optimization on the first weight and the second weight by combining a weight optimization algorithm to obtain a weight combination with the minimum error so as to reduce the deviation of a prediction result.
Target optimization: and carrying out target optimization on the predicted life parameter of the battery according to the obtained weight combination. The first weight and the second weight may be blended in a linear combination to obtain a more stable and reliable life prediction parameter.
In this process, the online learning technique enables the model to adaptively update the weights. And the knowledge of other battery types can be migrated to the target battery prediction by adopting migration learning so as to improve the prediction accuracy.
Life prediction results: and carrying out target optimization on the predicted life parameters of the battery through weight combination to obtain a final life prediction result of the target battery.
In the embodiment of the invention, the beneficial effects are as follows: according to the battery life prediction method provided by the invention, the state data of the target battery is acquired through the preset sensor, and the first life prediction curve and the second life prediction curve are drawn by means of the prediction function so as to calculate the corresponding target vector. And (5) performing splicing processing, local characteristic information extraction and association analysis to obtain the predicted life parameter of the battery. And combining the actual running state of the target battery, carrying out weight analysis on the historical battery data sample, and carrying out weight combination to obtain a life prediction result of the target battery. The prediction accuracy and reliability of the service life of the battery can be improved, so that better use experience and maintenance cost saving are provided for battery manufacturers and users.
Another embodiment of the battery life prediction method in the embodiment of the present invention includes:
Acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and performing data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data, wherein the method comprises the following steps of:
acquiring state data of a target battery through each preset sensor, preprocessing the state data of the target battery acquired from each sensor, and converting the preprocessed state data into a uniform data format to obtain battery state data corresponding to the target battery;
extracting key features of the battery state data through a preset feature extraction algorithm, and screening the battery state data according to different key features to obtain key feature data related to life prediction;
classifying the key feature data related to life prediction based on a preset supervised learning model to obtain first life prediction data and second life prediction data, wherein the supervised learning model is obtained through SVM algorithm training.
In the embodiment of the invention, the beneficial effects are as follows: first, the status data collected from the various sensors is made more uniform and normative by preprocessing and data format conversion. Thus, the subsequent state data analysis and processing process can be simplified, and the data processing efficiency can be improved. Second, during data screening, a preset feature extraction algorithm will help extract key features related to life prediction from battery state data. Thus, not only can the data volume be reduced, but also we can be helped to evaluate the battery life more accurately, and the potential problem can be found in advance. And thirdly, a supervised learning model obtained based on Support Vector Machine (SVM) algorithm training provides a highly accurate and reliable basis for whole life prediction. The algorithm has good generalization capability and higher prediction accuracy, so that the battery life prediction is more reliable. Finally, after knowing the first life prediction data and the second life prediction data of the battery, the user can effectively plan the replacement, maintenance, management and other works of the battery, thereby reducing the operation and maintenance cost and improving the service life of the battery.
Another embodiment of the battery life prediction method in the embodiment of the present invention includes:
the first life prediction function is based on a preset first life prediction function, a first life prediction curve is drawn by combining the first life prediction data, and a second life prediction curve is drawn by combining the second life prediction data on the basis of a preset second life prediction function; and calculating a plurality of first vector values in the first life prediction curve, and calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first and second target vectors based on a preset box diagram rule, comprising:
inputting the first life prediction data into a first prediction function, drawing a first life prediction curve based on a preset visual curve model, extracting characteristic values from the first life prediction curve, calculating a plurality of first vector values according to the characteristic values, and calculating a first target vector based on a preset box diagram rule;
and inputting the second life prediction data into a second prediction function, drawing a second life prediction curve based on a preset visual curve model, extracting characteristic values from the second life prediction curve, calculating a plurality of second vector values according to the characteristic values, and calculating a second target vector based on a preset box diagram rule.
In the embodiment of the invention, the beneficial effects are as follows: firstly, a preset life prediction function is adopted, and a life prediction curve is drawn by a visual curve model in combination with life prediction data. This makes the prediction result more intuitive, and the user can more conveniently observe the life expectancy change of the battery, thereby making and adjusting the battery maintenance plan more reasonably. And secondly, calculating vector values in the first life prediction curve and the second life prediction curve, and generating a target vector by using a preset box diagram rule, so that a user can more accurately evaluate the health condition of the battery. In this way, the user can reasonably arrange the replacement, maintenance or management of the battery, and ensure the reliability and continuous operation of the system. Again, the provision of the visualization curves and the target vectors provides a more comprehensive data support for the user. When predicting the service life of the battery, the user can evaluate the overall performance and the health condition of the battery, and further take corresponding measures to improve the service life of the battery and reduce the cost caused by untimely replacement of the battery. In addition, the user can conveniently input data, calculate and generate a prediction curve, and a target vector. The working efficiency of the user is greatly improved, and the time and the operation and maintenance cost are saved.
Another embodiment of the battery life prediction method in the embodiment of the present invention includes:
the first target vector and the second target vector are spliced to obtain spliced vectors, and the spliced vectors are input into a preset convolutional neural network model to extract local feature information, so that target feature data are obtained; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain battery predicted life parameters, wherein the method comprises the following steps:
splicing the first target vector and the second target vector according to a preset rule to obtain a spliced vector, inputting the spliced vector into a preset convolutional neural network model, and extracting local feature information of the spliced vector through a multi-scale convolutional layer of the convolutional neural network model to obtain target feature data;
and inputting the target characteristic data into a multi-layer stack layer of the long and short memory network model based on a preset long and short memory network model for correlation analysis processing to obtain the predicted life parameter of the battery.
In the embodiment of the invention, the beneficial effects are as follows: firstly, when the first target vector and the second target vector are processed, the characteristics of the first target vector and the second target vector can be combined by adopting a splicing processing method, so that richer data is provided for subsequent characteristic information extraction. The accuracy of battery life prediction can be improved, and more powerful support is provided for maintenance and management. And secondly, the local characteristic information of the spliced vector is extracted by applying the convolutional neural network model, so that the battery state information can be more effectively mined, and a more accurate basis is provided for battery life prediction. Meanwhile, the CNN model has good adaptability, can process a large amount of multidimensional data, and further ensures the reliability of a prediction result. And thirdly, performing association analysis processing on the target characteristic data by using the long and short memory network model, so that the data processing capacity of the model can be enhanced. The long-short memory network model overcomes the long-term dependence problem existing in the traditional Recurrent Neural Network (RNN), so that the model can capture the data dependence in a long time range. Therefore, by means of the LSTM, the calculation of the battery life prediction parameters is more accurate, and the reasonable maintenance strategy is facilitated to be formulated by a user. In summary, the embodiment of the invention combines the advanced algorithms such as CNN and LSTM, improves the accuracy of battery life prediction, and avoids the premature or too late replacement of the battery, thereby reducing the operation and maintenance cost. In addition, the technical means also provides a large amount of valuable information, which is helpful for users to early warn potential problems in time and formulate targeted maintenance measures. This will push the life of the battery to be extended, contributing power to achieve the goals of energy conservation, environmental protection, and sustainable development.
Another embodiment of the battery life prediction method in the embodiment of the present invention includes:
the method comprises the steps of obtaining an actual running state of a target battery, carrying out weight analysis on a predicted life parameter of the battery in combination with the actual running state of the target battery to generate a first weight, carrying out weight analysis on the predicted life parameter of the battery in combination with a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of a predicted result, carrying out target optimization on the predicted life parameter of the battery according to the weight combination to obtain a life predicted result of the target battery, and comprising the following steps:
acquiring actual running state data of a target battery, wherein the actual running state data comprise working temperature, charge and discharge times and circulation capacity loss;
carrying out weight analysis on predicted life parameters of the battery by combining actual running state data of the target battery to generate a first weight;
acquiring a historical battery data sample, and carrying out weight analysis on the predicted life parameter of the battery by combining the historical battery data sample to obtain a second weight;
weighting cross-validation or genetic algorithm is adopted to carry out weighing optimization between the first weight and the second weight, so as to obtain a weight combination with the minimum error of the predicted result;
And carrying out target optimization on the predicted life parameters of the battery according to the weight combination with the minimum error of the obtained predicted result, and obtaining the life predicted result of the target battery.
In the embodiment of the invention, the beneficial effects are as follows: firstly, when the actual running state of the target battery is considered, the technical means can better understand the actual use condition of the battery, so that more accurate weight analysis is carried out on the predicted parameters of the battery. By considering factors such as working temperature, charge and discharge times, circulation capacity loss and the like, the device can realize closer butt joint with the practical application environment and improve the reliability of a prediction result. And secondly, when a historical battery data sample is introduced to perform weight analysis, the performance data of similar batteries can be referenced, so that the model has a reference value. The method is favorable for further improving the accuracy of the prediction result, so that the consistency of the battery life prediction and the actual situation is obviously improved, and the weighting cross-validation or genetic algorithm is adopted for carrying out the balance optimization, so that the method is an innovative technical processing method. By carrying out balance optimization between the first weight and the second weight, the error of the predicted result is minimized, and the accuracy of the predicted life parameter of the battery in practical application is further improved.
The battery life prediction method in the embodiment of the present invention is described above, and the battery life prediction apparatus in the embodiment of the present invention is described below, referring to fig. 2, the battery life prediction apparatus 1 in the embodiment of the present invention includes:
the acquisition module 11 is configured to acquire state data of a target battery through preset sensors, obtain battery state data corresponding to the target battery, and perform data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
a first processing module 12, configured to draw a first life prediction curve based on a preset first life prediction function, in combination with the first life prediction data, and draw a second life prediction curve based on a preset second life prediction function, in combination with the second life prediction data; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
The second processing module 13 is configured to perform a stitching process on the first target vector and the second target vector to obtain a stitched vector, and input the stitched vector into a preset convolutional neural network model to perform local feature information extraction, so as to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
the weight analysis module 14 is configured to obtain an actual operation state of the target battery, perform weight analysis on the predicted lifetime parameter of the battery in combination with the actual operation state of the target battery, generate a first weight, perform weight analysis on the predicted lifetime parameter of the battery in combination with the historical battery data sample, obtain a second weight, perform weight balance optimization on the first weight and the second weight, obtain a weight combination with the smallest error of the predicted result, and perform target optimization on the predicted lifetime parameter of the battery according to the weight combination, so as to obtain a lifetime predicted result of the target battery.
The present invention also provides a battery life prediction apparatus including a memory and a processor, the memory storing computer readable instructions that, when executed by the processor, cause the processor to perform the steps of the battery life prediction method in the above embodiments.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, or a volatile computer readable storage medium, having stored therein instructions that, when executed on a computer, cause the computer to perform the steps of the battery life prediction method.
The beneficial effects are that: according to the battery life prediction method, the device, the equipment and the storage medium, the state data of the target battery are collected through the preset sensor, and the first life prediction curve and the second life prediction curve are drawn by means of the prediction function so as to calculate the corresponding target vector. And (5) performing splicing processing, local characteristic information extraction and association analysis to obtain the predicted life parameter of the battery. And combining the actual running state of the target battery, carrying out weight analysis on the historical battery data sample, and carrying out weight combination to obtain a life prediction result of the target battery. The prediction accuracy and reliability of the service life of the battery can be improved, so that better use experience and maintenance cost saving are provided for battery manufacturers and users.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (randomaccess memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. A battery life prediction method, comprising:
acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
drawing a first life prediction curve by combining the first life prediction data based on a preset first life prediction function, and drawing a second life prediction curve by combining the second life prediction data based on a preset second life prediction function; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
Performing stitching processing on the first target vector and the second target vector to obtain stitching vectors, and inputting the stitching vectors into a preset convolutional neural network model to extract local feature information to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
acquiring an actual running state of a target battery, carrying out weight analysis on the predicted life parameter of the battery in combination with the actual running state of the target battery to generate a first weight, carrying out weight analysis on the predicted life parameter of the battery in combination with a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of a predicted result, and carrying out target optimization on the predicted life parameter of the battery according to the weight combination to obtain a life predicted result of the target battery.
2. The method according to claim 1, wherein the acquiring, by each preset sensor, state data of a target battery to obtain battery state data corresponding to the target battery, and performing data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data, includes:
acquiring state data of a target battery through each preset sensor, preprocessing the state data of the target battery acquired from each sensor, and converting the preprocessed state data into a uniform data format to obtain battery state data corresponding to the target battery;
extracting key features of the battery state data through a preset feature extraction algorithm, and screening the battery state data according to different key features to obtain key feature data related to life prediction;
classifying the key feature data related to life prediction based on a preset supervised learning model to obtain first life prediction data and second life prediction data, wherein the supervised learning model is obtained through SVM algorithm training.
3. The method of claim 1, wherein the drawing a first life prediction curve based on a preset first life prediction function in combination with the first life prediction data, and drawing a second life prediction curve based on a preset second life prediction function in combination with the second life prediction data; and calculating a plurality of first vector values in the first life prediction curve, and calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first and second target vectors based on a preset box diagram rule, comprising:
inputting the first life prediction data into a first prediction function, drawing a first life prediction curve based on a preset visual curve model, extracting characteristic values from the first life prediction curve, calculating a plurality of first vector values according to the characteristic values, and calculating a first target vector based on a preset box diagram rule;
and inputting the second life prediction data into a second prediction function, drawing a second life prediction curve based on a preset visual curve model, extracting characteristic values from the second life prediction curve, calculating a plurality of second vector values according to the characteristic values, and calculating a second target vector based on a preset box diagram rule.
4. The method according to claim 1, wherein the first target vector and the second target vector are spliced to obtain a spliced vector, and the spliced vector is input into a preset convolutional neural network model to extract local feature information, so as to obtain target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain battery predicted life parameters, wherein the method comprises the following steps:
splicing the first target vector and the second target vector according to a preset rule to obtain a spliced vector, inputting the spliced vector into a preset convolutional neural network model, and extracting local feature information of the spliced vector through a multi-scale convolutional layer of the convolutional neural network model to obtain target feature data;
and inputting the target characteristic data into a multi-layer stack layer of the long and short memory network model based on a preset long and short memory network model for correlation analysis processing to obtain the predicted life parameter of the battery.
5. The method of claim 1, wherein the obtaining the actual operating state of the target battery, combining the actual operating state of the target battery, performing weight analysis on the predicted battery life parameter to generate a first weight, combining a historical battery data sample, performing weight analysis on the predicted battery life parameter to obtain a second weight, performing weight balance optimization on the first weight and the second weight, obtaining a weight combination with the smallest error of the predicted result, performing target optimization on the predicted battery life parameter according to the weight combination, and obtaining a predicted life result of the target battery, comprises:
Acquiring actual running state data of a target battery, wherein the actual running state data comprise working temperature, charge and discharge times and circulation capacity loss;
carrying out weight analysis on predicted life parameters of the battery by combining actual running state data of the target battery to generate a first weight;
acquiring a historical battery data sample, and carrying out weight analysis on the predicted life parameter of the battery by combining the historical battery data sample to obtain a second weight;
weighting cross-validation or genetic algorithm is adopted to carry out weighing optimization between the first weight and the second weight, so as to obtain a weight combination with the minimum error of the predicted result;
and carrying out target optimization on the predicted life parameters of the battery according to the weight combination with the minimum error of the obtained predicted result, and obtaining the life predicted result of the target battery.
6. A battery life prediction apparatus, characterized by comprising:
the acquisition module is used for acquiring state data of a target battery through preset sensors to obtain battery state data corresponding to the target battery, and carrying out data screening and data classification on the battery state data to obtain first life prediction data and second life prediction data in the battery state data;
The first processing module is used for drawing a first life prediction curve based on a preset first life prediction function and combining the first life prediction data, and drawing a second life prediction curve based on a preset second life prediction function and combining the second life prediction data; calculating a plurality of first vector values in the first life prediction curve, calculating a plurality of second vector values in the second life prediction curve, and generating corresponding first target vectors and second target vectors based on a preset box diagram rule;
the second processing module is used for performing splicing processing on the first target vector and the second target vector to obtain a spliced vector, inputting the spliced vector into a preset convolutional neural network model to extract local feature information, and obtaining target feature data; inputting the target characteristic data into a preset long and short memory network model for correlation analysis processing to obtain a battery predicted life parameter; the convolutional neural network model is obtained through a convolutional neural network algorithm pre-training, and the long and short memory network model is obtained through a long and short memory network algorithm pre-training;
The weight analysis module is used for acquiring the actual running state of the target battery, carrying out weight analysis on the battery predicted life parameter by combining the actual running state of the target battery to generate a first weight, carrying out weight analysis on the battery predicted life parameter by combining a historical battery data sample to obtain a second weight, carrying out weight balance optimization on the first weight and the second weight to obtain a weight combination with the minimum error of the predicted result, and carrying out target optimization on the battery predicted life parameter according to the weight combination to obtain the life predicted result of the target battery.
7. A battery life prediction apparatus, characterized by comprising: a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invokes the instructions in the memory to cause the battery life prediction device to perform the battery life prediction method of any of claims 1-5.
8. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the battery life prediction method of any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311044415.4A CN116774058B (en) | 2023-08-18 | 2023-08-18 | Battery life prediction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311044415.4A CN116774058B (en) | 2023-08-18 | 2023-08-18 | Battery life prediction method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116774058A CN116774058A (en) | 2023-09-19 |
| CN116774058B true CN116774058B (en) | 2023-10-20 |
Family
ID=87986201
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311044415.4A Active CN116774058B (en) | 2023-08-18 | 2023-08-18 | Battery life prediction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116774058B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102024999A (en) * | 2010-11-16 | 2011-04-20 | 上海交通大学 | Electric car running power management system |
| CN110188920A (en) * | 2019-04-26 | 2019-08-30 | 华中科技大学 | A kind of lithium battery method for predicting residual useful life |
| CN112348185A (en) * | 2020-11-04 | 2021-02-09 | 上海海事大学 | Lithium battery residual life prediction method based on variational modal decomposition and integrated depth model |
| CN112666480A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery life prediction method based on charging process characteristic attention |
| WO2022047204A1 (en) * | 2020-08-30 | 2022-03-03 | Hewlett-Packard Development Company, L.P. | Battery life predictions using machine learning models |
| CN114636933A (en) * | 2022-05-23 | 2022-06-17 | 长沙矿冶研究院有限责任公司 | Composite neural network-based retired power battery capacity detection method and system |
| CN114636932A (en) * | 2022-03-14 | 2022-06-17 | 华中科技大学 | Method and system for predicting remaining service life of battery |
| CN114660497A (en) * | 2022-03-17 | 2022-06-24 | 哈尔滨理工大学 | Lithium ion battery service life prediction method aiming at capacity regeneration phenomenon |
| CN115015769A (en) * | 2022-08-10 | 2022-09-06 | 深圳市杰成镍钴新能源科技有限公司 | Power battery residual value estimation method, device, equipment and medium |
| CN116298902A (en) * | 2022-12-20 | 2023-06-23 | 福建星云电子股份有限公司 | Lithium battery aging prediction method and system based on multitask learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7400115B2 (en) * | 2006-02-09 | 2008-07-15 | Lg Chem, Ltd. | System, method, and article of manufacture for determining an estimated combined battery state-parameter vector |
| KR20230003848A (en) * | 2021-06-30 | 2023-01-06 | 현대자동차주식회사 | Apparatus and method for estimating state of battery |
-
2023
- 2023-08-18 CN CN202311044415.4A patent/CN116774058B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102024999A (en) * | 2010-11-16 | 2011-04-20 | 上海交通大学 | Electric car running power management system |
| CN110188920A (en) * | 2019-04-26 | 2019-08-30 | 华中科技大学 | A kind of lithium battery method for predicting residual useful life |
| WO2022047204A1 (en) * | 2020-08-30 | 2022-03-03 | Hewlett-Packard Development Company, L.P. | Battery life predictions using machine learning models |
| CN112348185A (en) * | 2020-11-04 | 2021-02-09 | 上海海事大学 | Lithium battery residual life prediction method based on variational modal decomposition and integrated depth model |
| CN112666480A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery life prediction method based on charging process characteristic attention |
| CN114636932A (en) * | 2022-03-14 | 2022-06-17 | 华中科技大学 | Method and system for predicting remaining service life of battery |
| CN114660497A (en) * | 2022-03-17 | 2022-06-24 | 哈尔滨理工大学 | Lithium ion battery service life prediction method aiming at capacity regeneration phenomenon |
| CN114636933A (en) * | 2022-05-23 | 2022-06-17 | 长沙矿冶研究院有限责任公司 | Composite neural network-based retired power battery capacity detection method and system |
| CN115015769A (en) * | 2022-08-10 | 2022-09-06 | 深圳市杰成镍钴新能源科技有限公司 | Power battery residual value estimation method, device, equipment and medium |
| CN116298902A (en) * | 2022-12-20 | 2023-06-23 | 福建星云电子股份有限公司 | Lithium battery aging prediction method and system based on multitask learning |
Non-Patent Citations (3)
| Title |
|---|
| Chuan Li et al..Deep feature extraction in lifetime prognostics of lithium-ion batteries: Advances, challenges and perspectives.Renewable and Sustainable Energy Reviews .2023,第184卷第1-13. * |
| 王正 ; .基于机器学习的新能源汽车电池剩余寿命预测.机械与电子.2019,(第12期),第11-17页. * |
| 王畅.数据驱动的锂离子电池剩余使用寿命预测及快充参数优化.中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑.2021,(第08期),C042-494. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116774058A (en) | 2023-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Transfer learning for battery smarter state estimation and ageing prognostics: Recent progress, challenges, and prospects | |
| Chen et al. | An overview of data-driven battery health estimation technology for battery management system | |
| Zhao et al. | Battery prognostics and health management from a machine learning perspective | |
| JP7497432B2 (en) | Battery performance prediction | |
| Xu et al. | Data-driven-aided strategies in battery lifecycle management: prediction, monitoring, and optimization | |
| Zou et al. | Advancements in Artificial Neural Networks for health management of energy storage lithium-ion batteries: A comprehensive review | |
| Giobergia et al. | Mining sensor data for predictive maintenance in the automotive industry | |
| CN116466672B (en) | Data center machine room parameter regulation and control method based on artificial intelligence and related device | |
| CN117543791B (en) | Power supply detection method, device, equipment and storage medium for power supply | |
| Zhang et al. | A pattern-driven stochastic degradation model for the prediction of remaining useful life of rechargeable batteries | |
| CN117154263A (en) | Lithium battery cascade utilization charging and discharging system and control method | |
| CN117977028B (en) | Intelligent charging method and system based on vehicle battery monitoring | |
| Liu et al. | State of health estimation of lithium-ion batteries based on multi-feature extraction and temporal convolutional network | |
| Zou et al. | State of Health prediction of lithium-ion batteries based on temporal degeneration feature extraction with Deep Cycle Attention Network | |
| Zhao et al. | A deep learning approach incorporating attention mechanism and transfer learning for lithium-ion battery lifespan prediction | |
| Jafari et al. | Accurate remaining useful life estimation of lithium-ion batteries in electric vehicles based on a measurable feature-based approach with explainable AI | |
| CN116774058B (en) | Battery life prediction method, device, equipment and storage medium | |
| Zhang et al. | A novel supercapacitor degradation prediction using a 1D convolutional neural network and improved informer model | |
| Stoyanova et al. | State of health estimation of Lithium-ion batteries: A review of data-driven methods | |
| Kumar et al. | Estimating Remaining Useful Life and State of Health of EV Lithium Ion Batteries Using Sequential CNNs | |
| Muthuvinayagam et al. | Comprehensive Analysis on Power Electronic Systems in Relation with Machine Learning | |
| CN118549839B (en) | Battery early life prediction method based on relative position matrix and residual error network | |
| CN117439146B (en) | Data analysis control method and system for charging pile | |
| CN117631546B (en) | Intelligent biomass gasification carbonization method and system based on carbon emission | |
| Altinpulluk et al. | Federated battery diagnosis and prognosis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |