Abnormality detection method and system for energy storage battery based on optical fiber sensor

Abstract

The invention relates to the technical field of energy storage battery safety detection, in particular to an energy storage battery anomaly detection method and system based on an optical fiber sensor. The method comprises the steps of collecting working data of a target energy storage battery in real time; constructing a stress model of the target energy storage battery, solving a surface strain value at each time point, and generating a surface strain curve graph of the target energy storage battery; and performing anomaly detection on the target energy storage battery according to the surface strain curve graph. According to the method, the real-time strain, the strain in the unit time period and the predicted strain in the charging and discharging process of the target energy storage battery are compared with the surface strain curve graph and are compared with the preset threshold values, so that anomaly detection is carried out on the energy storage battery. Compared with a traditional detection mode, the method has higher real-time performance, sensitivity and comprehensive performance evaluation capability. Through a mode of combining a plurality of detection means, the accuracy and reliability of anomaly detection are improved, and a powerful guarantee is provided for long-term safe operation of the battery.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage battery safety detection, and particularly to an abnormal detection method for energy storage batteries based on optical fiber sensors and an abnormal detection system for energy storage batteries based on optical fiber sensors. Background Art

[0002] Energy storage batteries can store and release electric power, which helps to balance the fluctuations in power supply and demand. During the low power demand period, the excess electric energy is stored; during the peak period, the stored electric energy is released, thereby reducing the pressure on the power grid and improving the stability and reliability of the power system. It can also promote the large-scale application of renewable energy. For example, renewable energy sources such as solar energy and wind energy have the characteristics of intermittency and instability. Energy storage batteries can store the electric energy generated by these energy sources and provide power when there is no sunlight or insufficient wind, enabling renewable energy to be incorporated into the power grid more stably and promoting the optimization and transformation of the energy structure. It can provide independent power support for areas with insufficient power supply such as remote areas and islands, improving the autonomy and reliability of energy supply.

[0003] As a key device for energy storage, the normal operation of energy storage batteries is crucial for the stable and sustainable development of the energy field. However, in the application process of energy storage batteries, it has always been a thorny problem to detect battery abnormalities. Traditional detection methods often have many limitations. For example, by monitoring the voltage and current of the battery to judge its state, this method can only reflect the macroscopic electrical characteristics of the battery and is difficult to detect the subtle changes and potential problems inside the battery. Moreover, the detection of voltage and current is easily interfered by the external environment and load changes, resulting in inaccurate detection results. Using chemical analysis methods to detect the composition and concentration changes of the battery electrolyte can provide some information about the battery state, but this method is complex to operate, requires the battery to be disassembled and sampled, which is not only time-consuming and laborious, but also destructive and cannot achieve real-time on-line detection. In addition, relying on temperature sensors to monitor the temperature change of the battery to judge abnormalities also has certain defects. The change in temperature is often the result of battery abnormalities rather than an early warning signal. When the temperature rises significantly, the battery may already be in a relatively serious abnormal state and early prevention cannot be achieved. Summary of the Invention

[0004] The present invention provides an abnormal detection method and system for energy storage batteries based on optical fiber sensors to solve the defects existing in the prior art.

[0005] On the one hand, the present invention provides an abnormal detection method for energy storage batteries based on optical fiber sensors, including: Real-time collect the working data of the target energy storage battery, and preprocess the working data to obtain preprocessed data.

[0006] Construct a stress model of the target energy storage battery based on the preprocessed data, solve the surface strain values at each time point, and generate a surface strain curve graph of the target energy storage battery.

[0007] Based on the surface strain curve graph, perform anomaly detection on the target energy storage battery, including real-time strain value deviation detection, integral deviation detection of the battery surface strain curve segment within a unit time period, and predicted surface strain value deviation detection based on a machine learning prediction model.

[0008] According to an energy storage battery anomaly detection method based on an optical fiber sensor provided by the present invention, the working data includes performance parameters, state parameters, and environmental temperature data. The performance parameters include the charging current, discharging current, material parameters, elastic modulus, internal resistance, and capacity of the target energy storage battery. The state parameters include the temperature, number of cycles, charging time, and discharging time of the target energy storage battery.

[0009] According to an energy storage battery anomaly detection method based on an optical fiber sensor provided by the present invention, the process of preprocessing the working data includes: Use the linear interpolation method to process the missing values in the working data.

[0010] Set the outlier range through a box plot, delete the outliers with confirmed measurement errors, and retain and mark the true extreme values.

[0011] Perform standardization processing on the working data, and use the standardized data as the preprocessed data.

[0012] According to an energy storage battery anomaly detection method based on an optical fiber sensor provided by the present invention, the process of constructing a stress model of the target energy storage battery and solving the surface strain values at each time point includes: Construct a Mooney-Rivlin material constitutive model according to the material parameters of the target storage battery.

[0013] Derive the stress tensor according to the material constitutive model.

[0014] Based on the stress tensor, integrate the stress model based on the current, voltage, and temperature changes of the target energy storage battery.

[0015] Obtain the elastic matrix based on the material stiffness according to the elastic modulus of the target energy storage battery material, and combine it with the stress model to solve the surface strain values at each time point.

[0016] According to an energy storage battery anomaly detection method based on an optical fiber sensor provided by the present invention, the process of generating a surface strain curve graph of the target energy storage battery includes: Calculate the corresponding strain value for each time point to form a data pair.

[0017] Plot all the calculated strain values against the corresponding time values in a two-dimensional coordinate system, with the abscissa being time and the ordinate being the strain value, to obtain the surface strain curve of the target energy storage battery during charge and discharge.

[0018] According to an energy storage battery anomaly detection method based on fiber optic sensors provided by the present invention, the process of real-time strain value deviation detection includes: Measure the real-time surface strain value of the target energy storage battery, calculate the first deviation value between the real-time surface strain value and the surface strain data at the corresponding moment in the surface strain curve graph. If the first deviation value exceeds the preset first threshold, it is determined that the target energy storage battery is abnormal.

[0019] According to an energy storage battery anomaly detection method based on fiber optic sensors provided by the present invention, the process of integral deviation detection of the battery surface strain curve segment within a unit time period includes: Set a unit time period, and extract the graph strain value data sequence of the surface strain curve graph of the target energy storage battery within this time period.

[0020] Record the real-time surface strain value data sequence within this time period.

[0021] Calculate the integral deviation between the graph strain value data sequence and the real-time surface strain value data sequence.

[0022] Compare the integral deviation with the preset second threshold. If the integral deviation exceeds the preset second threshold, it is determined that the target energy storage battery is abnormal.

[0023] According to an energy storage battery anomaly detection method based on fiber optic sensors provided by the present invention, the process of predicted strain value deviation detection based on a machine learning prediction model includes: Construct a prediction model based on a long short-term memory network. Input performance parameters, state parameters, ambient temperature data, and the current surface strain value, and output the predicted surface strain value of the target energy storage battery at the next moment. Calculate the second deviation value between the predicted surface strain value and the surface strain data at the corresponding next moment in the surface strain curve graph. If the second deviation value exceeds the preset first threshold, it is determined that the target energy storage battery is abnormal.

[0024] According to an energy storage battery anomaly detection method based on fiber optic sensors provided by the present invention, the process of constructing a prediction model based on a long short-term memory network includes: Collect the historical working data of the target energy storage battery and the historical surface strain data corresponding to the historical working data at each moment. The historical working data includes historical performance parameters, historical state parameters, and historical ambient temperature data.

[0025] Preprocess the historical working data and historical surface strain data to obtain historical preprocessed data. Sort the historical preprocessed data by timestamp and divide it into a training set and a test set.

[0026] Set the prediction model structure, including an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to receive input data, the LSTM layer is used to capture the dynamic features of the input data in the time series. The fully connected layer is used to convert the dynamic features into prediction values. The output layer is used to output the predicted surface strain value of the target energy storage battery at the next moment.

[0027] Use the historical performance parameters, historical state parameters, historical ambient temperature data, and historical surface strain data at the first moment in the historical preprocessed data as inputs, and use the historical surface strain data at the second moment as the output. Train the prediction model and evaluate the accuracy of the prediction model on the test set until the accuracy meets the preset accuracy threshold, and retain the corresponding model parameters.

[0028] On the other hand, the present invention also provides an energy storage battery anomaly detection system based on an optical fiber sensor, including: A data acquisition module for real-time collecting the working data of the target energy storage battery.

[0029] A strain curve generation module for constructing a stress model of the target energy storage battery according to the working parameters, solving the surface strain value at each time point, and generating a surface strain curve graph of the target energy storage battery.

[0030] A real-time anomaly detection module for calculating the deviation value between the real-time surface strain value and the surface strain data at the corresponding moment in the surface strain curve graph, and performing anomaly detection on the target energy storage battery.

[0031] A fluctuation anomaly detection module for calculating the integral deviation between the real-time surface strain curve segment of the battery within a unit time period and the surface strain curve segment within the corresponding time period in the surface strain curve graph, and performing anomaly detection on the target energy storage battery.

[0032] A prediction anomaly detection module for constructing a prediction model based on a long short-term memory network, inputting the working parameters and the current surface strain value, outputting the predicted surface strain value of the target energy storage battery at the next moment, calculating the deviation value between the predicted surface strain value and the surface strain data at the corresponding next moment in the surface strain curve graph, and performing anomaly detection on the target energy storage battery.

[0033] An abnormal detection method and system for energy storage batteries based on fiber optic sensors provided by the present invention collect the working data of the target energy storage battery in real time, generate a surface strain curve graph of the energy storage battery according to the working data, and perform abnormal detection on the energy storage battery through the surface strain data during the charge and discharge process of the energy storage battery. Compared with the detection of parameters such as current and voltage, it can deeply understand the behavior mode of energy storage materials, can directly reflect the deformation characteristics of battery materials during operation, and can infer the internal microscopic physical state and performance of the battery through strain data, having irreplaceable advantages in detecting material performance degradation, fault occurrence, etc. At the same time, the sensitivity of strain sensing is relatively high, and it can capture small changes in materials at an early stage. Compared with the changes in general voltage and current levels, strain-based abnormal detection can detect abnormal behaviors of the battery earlier, take timely measures to prevent major accidents, and thus greatly improve the safety of the battery. By constructing a prediction model based on a long short-term memory network, comprehensive analysis can be carried out by combining multi-dimensional factors such as performance, state, and environment, enabling strain detection to provide a more comprehensive and accurate battery health assessment under the interaction of multiple factors, and then optimizing the operation management of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0035] Figure 1 is a schematic flowchart of an abnormal detection method for energy storage batteries based on fiber optic sensors provided by an embodiment of the present invention; Figure 2 is a schematic structural diagram of an abnormal detection system for energy storage batteries based on fiber optic sensors provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0037] The following will describe Figure 1 - Figure 2 an abnormal detection method and system for energy storage batteries based on fiber optic sensors of the present invention.

[0038] Figure 1It is a schematic structural diagram of an abnormal detection method for energy storage batteries based on optical fiber sensors provided by an embodiment of the present invention.

[0039] As Figure 1 shown, an abnormal detection method and system for energy storage batteries based on optical fiber sensors provided by an embodiment of the present invention, the execution subject can be an abnormal detection method for energy storage batteries based on optical fiber sensors, including: Real-time collect the working data of the target energy storage battery, the working data includes performance parameters, state parameters and ambient temperature data, and preprocess the working data to obtain preprocessed data.

[0040] The performance parameters include the charging current, discharging current, material parameters, elastic modulus, internal resistance and capacity of the target energy storage battery. The state parameters include the temperature, cycle times, charging time and discharging time of the target energy storage battery.

[0041] The process of preprocessing the working data includes: Adopt the linear interpolation method to process the missing values in the working data, and the process includes: Utilize two adjacent known values (x 1 , y 1 and x 2 , y 2 ) in the working data to linearly calculate the value of the unknown point between these two values. Let the x value of the unknown point be x 0 , then according to the linear relationship formula:

[0042] In the formula, represents the filled value.

[0043] For each missing value, interpolate and supplement it according to the known values before and after it to maintain the continuity of the time series.

[0044] Set the outlier range through the box plot, delete the outliers that confirm measurement errors, and retain and mark the true extreme values. The process includes: Use quartiles (Q 1 , Q 3 ) and the interquartile range (IQR = Q 3 -Q 1 ) to deduce the outlier range, and define the upper and lower limits as:

[0045]

[0046] In the formula, LB represents the lower limit of normal values, and UB represents the upper limit of normal values.

[0047] Standardize the working data and use the standardized data as preprocessed data.

[0048] In this embodiment, the material parameters include the viscoelastic properties and chemical composition of the material. The elastic modulus is a physical quantity characterizing the elastic rigidity of the material. The internal resistance of the battery is an important parameter affecting its performance and will change during operation. The capacity represents the charging ability of the battery and reflects the performance state of the battery. The internal temperature of the battery directly affects the chemical reaction rate and performance stability of the battery. The number of charge-discharge cycles of the battery characterizes its aging degree and performance state. The external environmental temperature affects the operating environment of the battery. By applying the linear interpolation method, the continuity of the time-series data is ensured, thus avoiding analysis deviations caused by missing data and improving the integrity of the data. The box plot outlier detection method effectively identifies measurement errors and avoids the impact of outliers on subsequent data analysis and modeling.

[0049] Construct a stress model of the target energy storage battery based on the preprocessed data, solve the surface strain value at each time point, and generate a surface strain curve graph of the target energy storage battery.

[0050] The process of constructing a stress model of the target energy storage battery and solving the surface strain value at each time point includes: Construct a Mooney-Rivlin material constitutive model based on the material parameters of the target storage battery, and the formula is expressed as:

[0051] In the formula, W represents the strain energy density, and represent material parameters.

[0052] represents the first invariant of the deformation matrix, represents the second invariant of the deformation matrix, and the calculation formula is expressed as:

[0053]

[0054] In the formula, F represents the deformation matrix.

[0055] Derive the stress tensor according to the material constitutive model. By differentiating the strain energy density W, the expression of the stress tensor can be obtained:

[0056] In the formula, represents the stress tensor, represents the rate of change of the strain energy density when the material undergoes small deformation, indicating the response ability of the battery material to deformation.

[0057] Based on the stress tensor, a stress model integrating the current, voltage, and temperature changes of the target energy storage battery is formulated as follows:

[0058] In the formula, represents the internal stress of the target energy storage battery at time t, represents the current stress contribution coefficient, represents the current at time t, represents the voltage stress contribution coefficient, represents the voltage at time t, represents the internal temperature of the target energy storage battery at time t, represents the ambient temperature, represents the temperature stress contribution coefficient.

[0059] Based on the elastic modulus of the target energy storage battery material, an elastic matrix based on material stiffness is obtained. Combining with the stress model, the surface strain value at each time point is solved, which is formulated as follows:

[0060] In the formula, represents the surface strain value of the target energy storage battery, represents the stress model, D represents the elastic matrix describing the stiffness of the battery material, reflecting the rigid characteristics of the battery material in each direction, and is formulated as follows:

[0061] In the formula, E represents the elastic modulus of the battery material, and v represents the Poisson's ratio.

[0062] The process of generating the surface strain curve of the target energy storage battery includes: For each time point t, the corresponding strain value ϵ(t) is calculated to form a data pair [t, ϵ(t)].

[0063] All the calculated strain values and the corresponding time values are plotted in a two-dimensional coordinate system, with the abscissa being the time t and the ordinate being the strain value ϵ(t), to obtain the surface strain curve of the target energy storage battery during charge and discharge.

[0064] In this embodiment, by plotting the strain curve graph, the change trend of the surface strain of the battery during charge and discharge can be visually observed. By comparing the strain values at different time points, the performance of the battery material under different working conditions can be judged. The strain curve can help quickly identify abnormal phenomena, such as abnormal high strain values or significant strain fluctuations, etc. These abnormal signals may indicate structural damage or performance deterioration of the battery. At the same time, the collection of long-term data and the plotting of the strain curve can help analyze the long-term stability and life law of the battery performance. By observing whether the strain value is stable, fluctuating or increasing over time, it can provide a decision-making basis for the maintenance and update of the battery. By plotting the strain curves under different charge and discharge states, the strain response characteristics of the battery under different working conditions can be analyzed, and the charge and discharge performance of the battery can be studied more deeply, so as to help improve the design or usage strategy.

[0065] According to the surface strain curve graph, abnormal detection is performed on the target energy storage battery, including real-time strain value deviation detection, integral deviation detection of the battery surface strain curve segment within a unit time period, and predicted surface strain value deviation detection based on a machine learning prediction model.

[0066] The process of real-time strain value deviation detection includes: Measure the real-time surface strain value of the target energy storage battery, calculate the first deviation value between the real-time surface strain value and the surface strain data at the corresponding moment in the surface strain curve graph. If the first deviation value exceeds the preset first threshold, it is determined that the target energy storage battery is abnormal.

[0067] In this embodiment, an optical fiber sensor is used to monitor the surface strain value of the energy storage battery in real time. The specific steps include: Read the signal of the optical fiber sensor at a preset period to obtain the surface strain value ϵ real (t) at the current moment.

[0068] Record the data and exclude the errors caused by external interference to make the real-time strain value as accurate as possible.

[0069] According to the real-time collected time t, find the corresponding theoretical strain value ϵ curve (t) from the strain curve.

[0070] Calculate the difference between the real-time strain value and the theoretical strain value to obtain the first deviation value, which is expressed by the formula:

[0071] In the formula, represents the first deviation value.

[0072] The method for setting the preset first threshold is to calculate the standard deviation based on the historically collected strain data, and use the mean plus a certain multiple of the standard deviation as the threshold for anomaly detection. The process includes: Collect historical strain value data over a period of time. The set of historical data is {ϵ historical (t 1 ), ϵ historical (t 2 ), …, ϵ historical (t n )}.

[0073] Calculate the mean of the historical strain data. The formula is expressed as:

[0074] Calculate the standard deviation. The formula is expressed as:

[0075] In the formula, represents the mean, represents the standard deviation, and n represents the sample size.

[0076] The formula for setting the preset first threshold is:

[0077] In the formula, represents the preset first threshold, represents the mean, k represents the empirical proportionality factor, usually taking values from 2 to 3 to cover most of the normal fluctuation range, represents the standard deviation.

[0078] The process of integral deviation detection for the battery surface strain curve segment within a unit time period includes: Set a unit time period [t 1 , t 2 , and extract the sequence of graph strain value data {ϵ(t 1 , t 2 )} within the time period [t 1 , t 1 + Δt), ϵ(t 1 + 2Δt), …, ϵ(t 2 )} from the target energy storage battery surface strain curve graph, where Δt represents the time step.

[0079] Record the real-time surface strain value data sequence {ϵ 1 , t 2 as {ϵ real (t 1 ), ϵ real (t 1 + Δt), ϵreal (t 1 +2Δt),…,ϵ real (t 2 )}。

[0080] Calculate the integral deviation between the graph strain value data sequence and the real-time surface strain value data sequence, which is expressed by the formula:

[0081] In the formula, represents the real-time strain data, represents the strain data in the strain curve graph.

[0082] Since it is impossible to perform analytical integration on continuous functions, the trapezoidal method is used for approximate calculation, which is expressed by the formula:

[0083] In the formula, n represents the number of step lengths within the time period and .

[0084] Compare the integral deviation with a preset second threshold. If the integral deviation exceeds the preset second threshold, it is determined that the target energy storage battery is abnormal.

[0085] The process of predicting strain value deviation detection based on a machine learning prediction model includes: Construct a prediction model based on a long short-term memory network. Input performance parameters, state parameters, ambient temperature data, and the current surface strain value, and output the predicted surface strain value of the target energy storage battery at the next moment. Calculate the second deviation value between the predicted surface strain value and the surface strain data corresponding to the next moment in the surface strain curve graph. If the second deviation value exceeds the preset first threshold, it is determined that the target energy storage battery is abnormal.

[0086] The process of constructing a prediction model based on a long short-term memory network includes: Collect the historical working data of the target energy storage battery and the historical surface strain data corresponding to the historical working data at each moment. The historical working data includes historical performance parameters, historical state parameters, and historical ambient temperature data.

[0087] Preprocess the historical working data and historical surface strain data to obtain historical preprocessed data. Sort the historical preprocessed data by timestamp and divide it into a training set and a test set.

[0088] Set the prediction model structure, including an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to receive input data, the LSTM layer is used to capture the dynamic features of the input data in the time series. The fully connected layer is used to convert the dynamic features into prediction values. The output layer is used to output the predicted surface strain value of the target energy storage battery at the next moment.

[0089] Use the historical performance parameters, historical state parameters, historical ambient temperature data, and historical surface strain data at the first moment in the historical preprocessed data as input, and use the historical surface strain data at the second moment as output to train the prediction model, and evaluate the accuracy of the prediction model on the test set until the accuracy meets the preset accuracy threshold, and retain the corresponding model parameters.

[0090] In summary, this embodiment provides an abnormal detection method for energy storage batteries based on fiber optic sensors. By collecting the working data of the target energy storage battery in real time, generating a surface strain curve graph of the energy storage battery according to the working data, and performing abnormal detection on the energy storage battery through the surface strain data during the charge and discharge process of the energy storage battery. Compared with the detection of parameters such as current and voltage, it can deeply understand the behavior mode of energy storage materials, can directly reflect the deformation characteristics of battery materials during operation, and can infer the internal microscopic physical state and performance of the battery through strain data, and has irreplaceable advantages in detecting material performance degradation, fault occurrence, etc. At the same time, the sensitivity of strain sensing is relatively high, and it can capture small changes in materials at an early stage. Compared with the changes in general voltage and current levels, strain-based abnormal detection can detect abnormal behaviors of the battery earlier, and take timely measures to prevent major accidents, thus greatly improving the safety of the battery. By constructing a prediction model based on a long short-term memory network, it is possible to perform comprehensive analysis by combining multi-dimensional factors such as performance, state, and environment, so that strain detection can provide a more comprehensive and accurate battery health assessment under the interaction of multiple factors, and then optimize the operation management of the battery.

[0091] Based on the same general inventive concept, the present invention also protects an abnormal detection system for energy storage batteries based on fiber optic sensors. Hereinafter, an abnormal detection system for energy storage batteries based on fiber optic sensors provided by the present invention will be described. The abnormal detection system for energy storage batteries based on fiber optic sensors described below can be correspondingly referred to the abnormal detection method for energy storage batteries based on fiber optic sensors described above.

[0092] Figure 2 It is a schematic flow chart of an abnormal detection system for energy storage batteries based on fiber optic sensors provided by an embodiment of the present invention.

[0093] As Figure 2As shown in the figure, the energy storage battery abnormal detection system based on fiber optic sensors includes a data acquisition module, a strain curve generation module, a real-time abnormal detection module, a fluctuation abnormal detection module, and a prediction abnormal detection module.

[0094] The data acquisition module is used to collect the working data of the target energy storage battery in real time.

[0095] The strain curve generation module is used to construct a stress model of the target energy storage battery according to the working data, solve the surface strain value at each time point, and generate a surface strain curve graph of the target energy storage battery.

[0096] The real-time abnormal detection module is used to calculate the deviation value between the real-time surface strain value and the surface strain data at the corresponding moment in the surface strain curve graph, and perform abnormal detection on the target energy storage battery.

[0097] The fluctuation abnormal detection module is used to calculate the integral deviation between the real-time surface strain curve segment of the battery within a unit time period and the surface strain curve segment within the corresponding time period in the surface strain curve graph, and perform abnormal detection on the target energy storage battery.

[0098] The prediction abnormal detection module is used to construct a prediction model based on a long short-term memory network, input the working parameters and the current surface strain value, output the predicted surface strain value of the target energy storage battery at the next moment, calculate the deviation value between the predicted surface strain value and the surface strain data at the corresponding next moment in the surface strain curve graph, and perform abnormal detection on the target energy storage battery.

[0099] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A method for detecting abnormality of an energy storage battery based on an optical fiber sensor, characterized in that: include: Collecting working data of the target energy storage battery in real time, and preprocessing the working data to obtain preprocessed data; Constructing a stress model of the target energy storage battery according to the preprocessed data, solving the surface strain value at each time point, and generating a surface strain curve graph of the target energy storage battery; According to the surface strain curve diagram, abnormality detection is performed on the target energy storage battery, including real-time strain value deviation detection, integral deviation detection of the battery surface strain curve segment within a unit time period, and predicted surface strain value deviation detection based on a machine learning prediction model.

2. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The working data includes performance parameters, state parameters and ambient temperature data. The performance parameters include the charging current, discharging current, material parameters and elastic modulus, internal resistance and capacity of the target energy storage battery; the state parameters include the temperature, number of cycles, charging time and discharging time of the target energy storage battery.

3. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The process of preprocessing the working data includes: Using linear interpolation to process missing values ​​in the working data; By setting the outlier range through the box plot, outliers that confirm measurement errors are deleted, and the real extreme values ​​are retained and marked; The working data is standardized and the standardized data is used as preprocessing data.

4. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The process of constructing the stress model of the target energy storage battery and solving the surface strain value at each time point includes: According to the material parameters of the target battery, the Mooney-Rivlin material constitutive model is constructed; deriving a stress tensor based on the material constitutive model; Based on the stress tensor, integrating a stress model based on current, voltage and temperature changes of the target energy storage battery; The elastic matrix based on the material stiffness is obtained according to the elastic modulus of the target energy storage battery material, and the surface strain value at each time point is solved in combination with the stress model.

5. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The process of generating the surface strain curve of the target energy storage battery includes: The corresponding strain value is calculated for each time point to form a data pair; All calculated strain values ​​and corresponding time values ​​are plotted in a two-dimensional coordinate system, with the horizontal axis being time and the vertical axis being strain value, to obtain the surface strain curve of the target energy storage battery during the charging and discharging process.

6. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The process of real-time strain value deviation detection includes: The real-time surface strain value of the target energy storage battery is measured, and a first deviation value between the real-time surface strain value and the surface strain data at the corresponding time in the surface strain curve diagram is calculated. If the first deviation value exceeds a preset first threshold value, it is determined that the target energy storage battery is abnormal.

7. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 1, characterized in that: The process of detecting the integral deviation of the battery surface strain curve segment within a unit time period includes: Set a unit time period, and extract the strain value data sequence of the surface strain curve of the target energy storage battery within the time period; A real-time surface strain value data sequence recorded within the time period; Calculating the integral deviation between the image strain value data sequence and the real-time surface strain value data sequence; The integral deviation is compared with a preset second threshold value, and if the integral deviation exceeds the preset second threshold value, it is determined that an abnormality exists in the target energy storage battery.

8. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 2, characterized in that: The process of predictive strain value deviation detection based on machine learning prediction model includes: A prediction model based on a long short-term memory network is constructed, the performance parameter, the state parameter, the ambient temperature data and the current surface strain value are input, the predicted surface strain value of the target energy storage battery at the next moment is output, and a second deviation value between the predicted surface strain value and the surface strain data corresponding to the next moment in the surface strain curve diagram is calculated. If the second deviation value exceeds a preset first threshold value, it is determined that the target energy storage battery is abnormal.

9. The method for detecting abnormality of an energy storage battery based on an optical fiber sensor according to claim 8, characterized in that: The process of building a prediction model based on LSTM networks includes: Collect historical working data of the target energy storage battery and historical surface strain data corresponding to the historical working data at each moment, wherein the historical working data includes historical performance parameters, historical state parameters and historical ambient temperature data; Preprocessing the historical working data and the historical surface strain data to obtain historical preprocessed data, sorting the historical preprocessed data by timestamp, and dividing the historical preprocessed data into a training set and a test set; The prediction model structure is set, including an input layer, an LSTM layer, a fully connected layer and an output layer; the input layer is used to receive input data, the LSTM layer is used to capture the dynamic characteristics of the input data in the time series; the fully connected layer is used to convert the dynamic characteristics into predicted values; the output layer is used to output the predicted surface strain value of the target energy storage battery at the next moment; The historical performance parameters, historical state parameters, historical ambient temperature data and historical surface strain data at the first moment in the historical preprocessing data are taken as input, and the historical surface strain data at the second moment is taken as output, the prediction model is trained, and the accuracy of the prediction model is evaluated on a test set until the accuracy meets a preset accuracy threshold, and the corresponding model parameters are retained.

10. An energy storage battery abnormality detection system based on optical fiber sensor, the energy storage battery abnormality detection system based on optical fiber sensor adopts an energy storage battery abnormality detection method based on optical fiber sensor as claimed in any one of claims 1 to 9, characterized in that: The energy storage battery abnormality detection system includes: Data acquisition module, used to collect working data of target energy storage batteries in real time; The strain curve generation module is used to construct a stress model of the target energy storage battery according to the working parameters, solve the surface strain value at each time point, and generate a surface strain curve of the target energy storage battery; A real-time anomaly detection module is used to calculate the deviation between the real-time surface strain value and the surface strain data at the corresponding time in the surface strain curve diagram, and perform anomaly detection on the target energy storage battery; The fluctuation anomaly detection module is used to calculate the integral deviation between the real-time surface strain curve segment of the battery in a unit time period and the surface strain curve segment in the corresponding time period in the surface strain curve diagram, and perform anomaly detection on the target energy storage battery; The prediction anomaly detection module is used to build a prediction model based on the long short-term memory network, input the working parameters and the current surface strain value, output the predicted surface strain value of the target energy storage battery at the next moment, calculate the deviation value between the predicted surface strain value and the surface strain data corresponding to the next moment in the surface strain curve diagram, and perform anomaly detection on the target energy storage battery.

Images