Battery health status prediction method

Abstract

This invention provides a method for predicting the health status of a battery, comprising: during the charging and discharging process of a battery, the degree of contraction or expansion at different locations on the anode plate of the battery varies at the same time, correspondingly, the strain at different locations on the anode plate varies at the same time. Based on this characteristic, a network model is constructed according to the strain at different locations on the anode plate; based on the corresponding network model, the current state of charge of the battery is predicted according to the strain and strain rate at different locations on the anode plate at the current time; and / or based on the corresponding network model, the health status of the battery is predicted according to the maximum strain and residual strain on the anode plate in each cycle up to the present. This invention can perform distributed, high-precision, and comprehensive prediction of the operating status of the battery anode plate without relying on electrochemical testing technology.

Description

Technical Field

[0001] This invention belongs to the field of fiber optic sensing, and specifically relates to a method for predicting battery health status. Background Technology

[0002] Traditional methods for predicting the state of health (SOH) of lithium-ion batteries rely on establishing electrochemical field models through theoretical calculations and measuring key electrical information to predict the battery state. Predicting battery SOH using only a single theoretical model is challenging due to the need to consider multiple nonlinear, coupled, and dynamic influencing factors, thus limiting the accuracy of the predictions. With the advancement of artificial intelligence, neural networks have been applied to estimate battery SOH. Accurate estimates of the battery's state of charge (SOC) and state of health (SOH) can be achieved using simple electrochemical parameters, such as voltage, current, and temperature during battery operation. However, these methods depend on data from specific locations within the battery, such as voltage, current, and surface temperature, which may not fully reflect the internal electrochemical reaction mechanisms. To date, no effective method has been found to achieve distributed, in-situ monitoring within lithium-ion batteries while simultaneously enabling high-precision prediction of battery SOH without relying on electrochemical testing techniques. Summary of the Invention

[0003] This invention provides a method for predicting the health status of batteries, in order to solve the problem that it is currently impossible to perform distributed, high-precision, and comprehensive prediction of the operating status of battery anode plates without relying on electrochemical testing technology.

[0004] According to a first aspect of the present invention, a method for predicting battery health status is provided, comprising:

[0005] Step S100: During the charging and discharging process of the battery, the degree of contraction or expansion at different positions on the anode plate of the battery is different at the same time. Correspondingly, the strain at different positions on the anode plate is different at the same time. Based on this characteristic, a network model is constructed according to the strain at different positions on the anode plate.

[0006] Step S200: Based on the corresponding network model, predict the current state of charge of the battery according to the strain and strain rate at different positions on the anode plate at the current time, and / or based on the corresponding network model, predict the health status of the battery according to the maximum strain and residual strain on the anode plate in each cycle so far, wherein the strain rate is the difference in strain per unit time at the corresponding position on the anode plate, the maximum strain is the maximum strain measured on the anode plate in the corresponding cycle minus the minimum strain, and the residual strain is the last strain measured on the anode plate in the corresponding cycle minus the first strain.

[0007] In one optional implementation, step S100 specifically includes: constructing a feedforward neural network, wherein the feedforward neural network includes an input layer, a hidden layer and an output layer connected in sequence, the input of the input layer is the strain and strain rate at different positions on the anode plate at the same time, and the output of the output layer is the current charge state of the battery;

[0008] Step S200 specifically includes: based on the feedforward neural network, predicting the current state of charge of the battery according to the strain and strain rate at different positions on the anode plate at the current moment.

[0009] In another alternative implementation, the feedforward neural network is constructed in step S100 according to the following steps:

[0010] The charging or discharging process of the battery during use is taken as a cycle. All strain variables and strain velocities at different positions on the anode plate in different cycles are used as training sets and input into the feedforward neural network for pre-training.

[0011] The expected state of charge of the battery output by the output layer is compared with the actual state of charge of the battery. Based on the comparison result, the parameters in the feedforward neural network are adjusted to obtain the optimal parameter set. The parameter set includes the number of hidden layers, the number of neurons in each hidden layer, the weight of each input in the input layer to each neuron in the first hidden layer, the weight of each neuron in each hidden layer to each neuron in the next hidden layer, and the weight of each neuron in the last hidden layer to each output in the output layer.

[0012] In another alternative implementation, the output quantity of the output layer is the current overall charge state of the battery or the current charge state at different corresponding locations on the battery.

[0013] In another optional implementation, step S100 specifically includes: constructing a long short-term memory network, wherein the long short-term memory network includes an input layer, a long short-term memory network layer, a fully connected layer and an output layer connected in sequence, the input of the input layer is the maximum strain and residual strain on the anode plate in each cycle up to the present, and the output of the output layer is the current health state of the battery; step S200 specifically includes: based on the long short-term memory network, predicting the health state of the battery according to the maximum strain and residual strain on the anode plate in the current cycle.

[0014] In another alternative implementation, the cycle period is the time period corresponding to each charging or discharging process during the use of the battery; the output layer of the long short-term memory network outputs a time-series signal without periodic changes, and the current health status of the battery is the current maximum battery capacity divided by the initial battery capacity.

[0015] In another alternative implementation, the strain at different locations on the anode plate is measured according to the following steps:

[0016] Step S1: The sensing optical fiber is covered on the upper surface of the anode plate of the battery in a serpentine manner in the first direction. The first optical fiber segment is the one that runs along the second direction after each bend of the sensing optical fiber. Each two adjacent first optical fiber segments are connected by a corresponding second optical fiber segment. The first direction is perpendicular to the second direction.

[0017] Step S2: For each first fiber segment, fix the two ends of the first fiber segment outside the anode plate to the protective shell below the anode plate respectively, so that the first fiber segment is subjected to corresponding prestress to ensure that the first fiber segment remains in a tensile state during the monitoring process.

[0018] Step S3: Connect one end of the sensing optical fiber to the strain measurement and analysis system, and perform a pressing test on both ends of each first optical fiber segment to calibrate the position of each first optical fiber segment.

[0019] Step S4: The strain measurement and analysis system determines the strain at each measurement point on the first fiber segment based on the scattering signal transmitted back in the reverse direction from each first fiber segment and the calibrated positions of the two ends of the first fiber segment; it reconstructs the two-dimensional strain field of the first fiber segment based on the strain at each measurement point on the first fiber segment; and it obtains the three-dimensional strain field of the anode plate composed of the two-dimensional strain fields of each first fiber segment based on the arrangement relationship between the first fiber segments.

[0020] In another alternative implementation, the protective shell is provided with the anode plate, the isolation layer and the cathode plate in sequence from top to bottom.

[0021] In another alternative implementation, the second fiber segment is laid on a protective housing beneath the anode plate. Step S4, before reconstructing the two-dimensional strain field of the first fiber segment based on the strain at each measurement point on the first fiber segment, further includes:

[0022] The strain measurement and analysis system determines the strain of each second fiber segment based on the backscattered signal transmitted from each second fiber segment and the calibrated positions of both ends of the first fiber segment; it determines the corresponding temperature of the second fiber segment based on the strain; it adds up the temperatures of each second fiber segment and takes the average value to obtain the average temperature of the battery; it converts the average temperature into the corresponding strain; and it subtracts the strain corresponding to the average temperature from the strain at each measurement point on the first fiber segment.

[0023] In another alternative implementation, each second fiber segment completely covers the protective shell on the corresponding side in the second direction, and the lengths of each second fiber segment are equal.

[0024] The beneficial effects of this invention are:

[0025] 1. This invention is based on the characteristic that the strain at different locations on the anode plate varies at the same time during battery charging and discharging. Sufficient data for building a network model can only be obtained by separately measuring the strain at different locations on the anode plate. Since the strain change process has a strict correspondence with the battery charging and discharging process, and the strain reflects the internal chemical reaction process of the battery to a certain extent, this invention correlates the strain, strain rate, and battery state of charge to construct a corresponding network model. This ensures the accuracy of predicting the current state of charge of the battery. Furthermore, the network model inputs the strain and strain rate at different locations on the anode plate at the same time, which reflects the overall strain of the anode plate. This invention uses strain rate to more comprehensively reflect the current state of charge of the battery, further improving the accuracy of predicting the current state of charge. It also correlates the maximum strain and residual strain on the anode plate within a cycle with the battery's health state, constructing a corresponding network model to ensure accurate prediction of the battery's current health state. Furthermore, the network model is input with the maximum strain and residual strain on the anode plate within each cycle up to the present, taking into account historical data from battery use, thus further improving the accuracy of predicting the battery's current health state. This invention can perform distributed, high-precision, and comprehensive prediction of the operating state of the battery anode plate without relying on electrochemical testing technology.

[0026] 2. When constructing the feedforward neural network, this invention considers the different stages of battery use, the charging and discharging process of the battery at different stages, and all strain variables and strain rates at different positions on the anode plate at different stages and in the corresponding charging or discharging states of the battery. Therefore, based on the optimal parameter set obtained during the construction process, this feedforward neural network can comprehensively and accurately predict the current charge state of the battery during the entire battery use process and in any charging or discharging state during battery use.

[0027] 3. This invention covers the anode plate of the battery with sensing optical fibers in both the first direction and the second direction perpendicular to the first direction, thus enabling monitoring of the strain of the entire anode plate. When deploying the first optical fiber segment directly above the anode plate, both ends of the first optical fiber segment are fixed to a protective shell below the anode plate, and prestress is applied to the first optical fiber segment to maintain tension throughout the monitoring process, allowing for accurate detection of anode plate deformation. Based on the calibrated positions of the two ends of each first optical fiber segment, the scattering signals transmitted back from the reverse direction by each first optical fiber segment, and the arrangement relationship between the first optical fiber segments, this invention can reconstruct the three-dimensional strain field of the anode plate. This invention embeds the sensing optical fiber inside the battery, achieving stress monitoring without affecting battery performance. Furthermore, this invention employs distributed monitoring, which, compared to single-point sensors, can acquire the strain field of the entire surface of the lithium battery anode material rather than strain information at a single point, resulting in higher reliability. This invention uses the reconstructed three-dimensional strain field of the anode plate as a basis for network model construction, further improving the accuracy of the model.

[0028] 4. This invention uses the sum of the strain at all measurement points on the second optical fiber segment as the strain of that segment. By establishing the relationship between the strain and temperature of the second optical fiber segment, the corresponding temperature of that segment is determined. Since the strain of the second optical fiber segment is relatively large, the accuracy of reflecting its temperature based on the strain of the second optical fiber segment is high. After determining the temperature of each second optical fiber segment, the temperatures of all segments are added together and then divided by the number of segments to obtain the average temperature of the battery. The accuracy of the average battery temperature obtained in this way is also high. This invention subtracts the strain caused by the average battery temperature from the strain detected on the first optical fiber segment to obtain the accurate strain caused by the anode plate deformation on the first optical fiber segment. This invention improves the accuracy of reconstructing the three-dimensional strain field of the anode plate by reducing strain. When eliminating the influence of temperature on the measurement of the corresponding variable, the invention establishes two types of temperature-related relationships. One type of temperature relationship is related to the total strain of the second fiber segment and the temperature of the area where the second fiber segment is located on the protective shells on both sides below the anode plate. The other type of temperature relationship is related to the average temperature of the protective shells on both sides below the anode plate and the strain of the first fiber segment on the anode plate caused by the average temperature. Both of these relationships establish the relationship between the fiber and the temperature of the protective shell. The relationship is simple to establish, and when the parameter in one of the relationships changes, only that relationship needs to be adjusted, without having to rebuild the entire relationship. Therefore, the temperature elimination is more flexible.

[0029] 5. This invention ensures that each second fiber segment completely covers the protective shell on the corresponding side in the second direction. Because the contact length between the second fiber segment and the protective shell is longer, it is easier for heat to reach thermal equilibrium quickly, reducing the temperature difference between the first and second fiber segments. This invention also ensures that the lengths of each second fiber segment are equal, so a unified standard can be used to determine the corresponding temperature of the second fiber segment based on the relationship between the strain of the second fiber segment and the temperature. Attached Figure Description

[0030] Figure 1 This is a flowchart of an embodiment of the battery health status prediction method of the present invention;

[0031] Figure 2 This is a schematic diagram of an embodiment of the feedforward neural network for predicting battery charge state according to the present invention;

[0032] Figure 3 This is a schematic diagram of an embodiment of the long short-term memory network of the present invention for predicting battery health status;

[0033] Figure 4 This is a diagram showing the layout of the sensing optical fiber of this invention;

[0034] Figure 5 This is a cross-sectional view of the sensor fiber of the present invention;

[0035] Figure 6 This is a schematic diagram of the three-dimensional strain field reconstruction of the anode plate of the present invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, and to make the above-mentioned objectives, features and advantages of the embodiments of the present invention more apparent and understandable, the technical solutions in the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0037] In the description of this invention, unless otherwise specified and limited, it should be noted that the term "connection" should be interpreted broadly. For example, it can be a mechanical connection or an electrical connection, or it can be a connection between two internal components. It can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above term according to the specific circumstances.

[0038] See Figure 1 This is a flowchart of an embodiment of the battery health status prediction method of the present invention. Combined with... Figure 2 and Figure 3 As shown, the method may include:

[0039] Step S100: During the charging and discharging process of the battery, the degree of contraction or expansion at different positions on the anode plate of the battery is different at the same time. Correspondingly, the strain at different positions on the anode plate is different at the same time. Based on this characteristic, a network model is constructed according to the strain at different positions on the anode plate.

[0040] In this embodiment, the battery can be a lithium battery. Under normal circumstances (except for cases of dead lithium, etc.), the strain change process of the anode plate has a strict correspondence with the charging and discharging process of the battery. When the battery is charging, the strain at all positions on the anode plate will decrease, but the degree of strain reduction at different positions at the same time may differ. When the battery is discharging, the strain at all positions on the anode plate will increase, and similarly, the degree of strain increase at different positions at the same time may also differ. This invention is based on this characteristic, and by measuring the strain at different positions on the anode plate separately, sufficient data can be obtained to establish the corresponding network model.

[0041] The network model constructed by this invention can include feedforward neural networks and long short-term memory networks, wherein, for example... Figure 2 As shown, the feedforward neural network may include an input layer, a hidden layer, and an output layer connected in sequence. The input to the input layer is the strain and strain rate at different positions on the anode plate at the same time. The output of the output layer is the current state of charge of the battery. The strain rate can be the difference in strain per unit time at the corresponding position on the anode plate (for example, the difference in strain at the corresponding position on the anode plate at the current time and the previous time). The output of the output layer can be the current overall state of charge of the battery or the current state of charge at different corresponding positions on the battery.

[0042] In step S100, the feedforward neural network can be constructed according to the following steps:

[0043] The charging and discharging process during the use of the battery is taken as the corresponding cycle (each charging process can be taken as one cycle and each discharging process as one cycle). All strain variables and strain velocities at different positions on the anode plate in different cycles are taken as training sets and input into the feedforward neural network for pre-training.

[0044] The predicted state of charge of the battery output by the output layer is compared with the actual state of charge of the battery. Based on the comparison result, the parameters in the feedforward neural network are adjusted to obtain the optimal parameter set. The parameter set includes the number of hidden layers, the number of neurons in each hidden layer, the weights of each input quantity in the input layer to each neuron in the first hidden layer, the weights of each neuron in each hidden layer to each neuron in the next hidden layer, and the weights of each neuron in the last hidden layer to each output quantity in the output layer. In one example, the hidden layer can be only two layers, each containing 16 neurons. The number of hidden layers and neurons can also be adjusted according to the actual situation.

[0045] This invention considers different stages of battery use, the charging and discharging process of the battery at different stages, and all strain variables and strain rates at different positions on the anode plate at different stages and in the corresponding charging or discharging states of the battery. Therefore, based on the optimal parameter set obtained during the construction process, this feedforward neural network can comprehensively and accurately predict the current charge state of the battery during the entire battery use process and in any charging or discharging state during battery use.

[0046] In addition, such as Figure 3 As shown, the Long Short-Term Memory (LSTM) network comprises an input layer, a time-dependent LSTM layer, a fully connected layer, and an output layer, connected sequentially. The input to the input layer is the maximum strain and residual strain on the anode plate within each cycle up to the present. The output of the output layer is the current health state of the battery. The cycle can be the time period corresponding to each charging or discharging process during battery use. The output of the LSTM network is a time-series signal without periodic changes, and the current health state of the battery can be calculated by dividing the current maximum battery capacity by the initial battery capacity.

[0047] In this embodiment, the maximum strain in each cycle is related to the battery's active material. As the battery is cycled, the active material is consumed and decreases, leading to a decrease in the maximum strain. Correspondingly, with the reduction of active material, the battery's lifespan also decreases. Furthermore, with battery cycling, ions embedded in the anode material cannot be completely deposited, resulting in dead lithium, which further leads to the continuous accumulation of residual strain, affecting the battery's lifespan. Therefore, this invention correlates the maximum strain and residual strain with the battery's health status, ensuring the accuracy of battery health status determination.

[0048] Step S200: Based on the corresponding network model, predict the current state of charge of the battery according to the strain and strain rate at different positions on the anode plate at the current time, and / or based on the corresponding network model, predict the health status of the battery according to the maximum strain and residual strain on the anode plate in each cycle so far, wherein the strain rate is the difference in strain per unit time at the corresponding position on the anode plate, the maximum strain is the maximum strain measured on the anode plate in the corresponding cycle minus the minimum strain, and the residual strain is the last strain measured on the anode plate in the corresponding cycle minus the first strain.

[0049] As can be seen from the above embodiments, this invention is based on the characteristic that the strain at different locations on the anode plate varies at the same time during battery charging and discharging. Sufficient data for building a network model can only be obtained by separately measuring the strain at different locations on the anode plate. Since the strain change process has a strict correspondence with the battery charging and discharging process, and the strain reflects the internal chemical reaction process of the battery to a certain extent, this invention correlates the strain, strain rate, and battery state of charge to construct a corresponding network model. This ensures the accuracy of predicting the current state of charge of the battery. Furthermore, the network model is input with the strain and strain rate at different locations on the anode plate at the same time, which can reflect the overall state of charge of the anode plate. This invention uses volumetric strain and strain rate to more comprehensively reflect the current state of charge of the battery, further improving the accuracy of predicting the current state of charge. It also correlates the maximum strain and residual strain on the anode plate within a cycle with the battery's health state, constructing a corresponding network model to ensure accurate prediction of the battery's current health state. Furthermore, the network model is input with the maximum strain and residual strain on the anode plate within each cycle up to the present, taking into account historical data from battery use, thus further improving the accuracy of predicting the battery's current health state. This invention can perform distributed, high-precision, and comprehensive prediction of the operating state of the battery anode plate without relying on electrochemical testing technology.

[0050] The above embodiments involve four parameters: strain, strain rate, maximum strain, and residual strain. When measuring the strain on the anode plate using a sensing fiber, the measured strain is affected not only by the contraction or expansion of the anode plate but also by the heat dissipated by the anode plate. Although the strain rate, maximum strain, and residual strain are all calculated by subtracting two strain values, different amounts of contraction or expansion correspond to different temperatures. Subtracting two strain values ​​with different values ​​does not eliminate the influence of temperature. Therefore, this invention designs the laying method of the sensing fiber on the anode plate. Combined with... Figures 4 to 6 As shown, the present invention measures the strain at different locations on the anode plate according to the following steps:

[0051] Step S1: The sensing optical fiber is covered on the upper surface of the anode plate 2 of the battery in a serpentine manner in the first direction, and the sensing optical fiber 1 is a first optical fiber segment 11 that bends and runs along the second direction each time. Each two adjacent first optical fiber segments 11 are connected by a corresponding second optical fiber segment 12. The first direction is perpendicular to the second direction.

[0052] In this embodiment, the first direction can be the length direction of the anode plate 2, and the second direction can be the width direction of the anode plate 2. The sensing optical fiber consists of a plurality of first optical fiber segments 11 arranged parallel to the second direction and spaced apart, and a plurality of second optical fiber segments 12 for connecting adjacent first optical fiber segments. The first optical fiber segments 12 completely cover the anode plate in the second direction (i.e., the width direction of the anode plate). The spacing between the parallel second optical fiber segments 12 is greater than 1 mm. The second optical fiber segments 12 are located on both sides of the anode plate 2, and the second optical fiber segments 12 on both sides can be staggered. The sensing optical fiber can be encapsulated in the protective shell of the battery along with the anode plate 2. The sensing optical fiber should have a high-temperature resistant coating layer, such as a polyimide coating or a gold coating, to ensure that the optical fiber under test will not break when the battery is encapsulated. When the battery is charged and discharged, ions in the battery will be released (e.g., lithium ions in a lithium battery). The anode plate 2 will undergo a chemical reaction, causing the anode plate to shrink or expand. This causes the first optical fiber segments covering the anode plate to be compressed or stretched along the optical fiber axis, generating axial strain in the optical fiber.

[0053] Step S2: For each first fiber segment 11, fix the two ends 13 of the first fiber segment 11 outside the anode plate 2 to the protective shell 3 below the anode plate 2 respectively (the fixing can be done by using tab adhesive, etc.), so that the first fiber segment 11 is subjected to corresponding prestress to ensure that the first fiber segment 11 always remains in a tensile state during the monitoring process.

[0054] In this embodiment, combined with Figure 5 As shown, the protective shell 3 is provided with the anode plate 2, the separator layer 4, and the cathode plate 5 sequentially from top to bottom, and another protective shell can also be provided on top of the anode plate 2. The battery can be a lithium battery, and the protective shell 3 can be an aluminum-plastic shell used to package the battery and prevent electrolyte leakage; the cathode plate 5 can be a lithium electrode; the separator layer can be a sheet of paper, firstly to ensure that the anode and cathode materials are separated and do not react, and secondly to ensure that the anode strain of the lithium battery is not interfered with by the cathode; the anode plate 3 can be replaced with various different materials, such as a mixed electrode of silicon suboxide (SiO) and graphite (C).

[0055] Step S3: Connect one end of the sensing optical fiber to the strain measurement and analysis system, and perform a pressing test on both ends 13 of each first optical fiber segment 11 to calibrate the position of each first optical fiber segment 11.

[0056] Step S4: The strain measurement and analysis system can determine the strain at each measurement point on the first fiber segment 11 based on the scattering signal transmitted back from each first fiber segment 11 in reverse and the calibrated positions of the two ends of the first fiber segment 11; reconstruct the two-dimensional strain field of the first fiber segment 11 based on the strain at each measurement point on the first fiber segment 11; and obtain the three-dimensional strain field of the anode plate 2 composed of the two-dimensional strain fields of each first fiber segment based on the arrangement relationship between the first fiber segments 11. In this embodiment, the strain measurement and analysis system may include an OFDR system (e.g., a phase-type optical frequency domain reflectometer φ-OFDR) and a data processing device. The OFDR system detects the axial strain generated on the sensing fiber based on the scattering signal transmitted back from the sensing fiber in reverse, and the data processing unit reconstructs the three-dimensional strain field of the battery based on the detected strain, thereby realizing distributed in-situ monitoring of the anode plate in the battery.

[0057] This invention covers the anode plate of the battery with sensing optical fibers in both a first direction and a second direction perpendicular to the first direction, thus enabling monitoring of the strain of the entire anode plate. When deploying the first optical fiber segment directly above the anode plate, both ends of the first optical fiber segment are fixed to a protective shell beneath the anode plate, and prestress is applied to the first optical fiber segment to maintain tension throughout the monitoring process, allowing for accurate detection of anode plate deformation. Based on the calibrated positions of the two ends of each first optical fiber segment, the scattering signals transmitted back from the reverse direction by each first optical fiber segment, and the arrangement relationship between the first optical fiber segments, this invention can reconstruct the three-dimensional strain field of the anode plate. By embedding the sensing optical fiber inside the battery, this invention achieves stress monitoring without affecting battery performance. Furthermore, this invention employs distributed monitoring, which, compared to single-point sensors, can acquire the strain field of the entire surface of the lithium battery anode material rather than strain information at a single point, resulting in higher reliability. This invention uses the reconstructed three-dimensional strain field of the anode plate as a basis for building a network model, further improving the accuracy of the model construction.

[0058] During battery charging and discharging, a chemical reaction occurs on the anode plate, causing deformation and generating heat. Therefore, the first fiber segment 11 in the sensing fiber will experience strain not only due to anode plate deformation but also due to temperature. To eliminate temperature interference, the second fiber segment can be laid on a protective housing beneath the anode plate. Before reconstructing the two-dimensional strain field of the first fiber segment based on the strain at various measurement points, step S4 may further include:

[0059] The strain measurement and analysis system determines the strain of each second fiber segment based on the backscattered signal transmitted from each second fiber segment and the calibrated positions of both ends of the first fiber segment; it determines the corresponding temperature of the second fiber segment based on the strain; it adds up the temperatures of each second fiber segment and takes the average value to obtain the average temperature of the battery; it converts the average temperature into the corresponding strain; and it subtracts the strain corresponding to the average temperature from the strain at each measurement point on the first fiber segment.

[0060] The strain of sensing optical fibers affected by temperature is usually small. If, when determining the average temperature of a battery, the strain at each measurement point on all second fiber segments is summed and then divided by the number of measurement points to obtain the average strain, and the average temperature is determined by establishing a relationship between the average strain and the average temperature, then the accuracy of reflecting the average temperature based on the average strain is not high due to the small average strain. This invention, however, uses the sum of the strain at all measurement points on the second fiber segment as the strain of that second fiber segment, and determines the corresponding temperature of the second fiber segment by establishing a relationship between the strain and temperature. Because the strain of the second fiber segment is large, the accuracy of reflecting its temperature based on the strain of the second fiber segment is high. After determining the temperature of each second fiber segment, the temperatures of each second fiber segment are summed and then divided by the number of second fiber segments to obtain the average temperature of the battery. The accuracy of the battery average temperature obtained in this way is also high. This invention subtracts the strain caused by the average battery temperature from the strain detected on the first fiber segment to obtain the accurate strain caused by the deformation of the anode plate on the first fiber segment, thereby improving the accuracy of the three-dimensional strain field reconstruction of the anode plate. Furthermore, when eliminating the influence of temperature on the measurement of the corresponding variable, this invention establishes two types of temperature-related correspondences. One type of temperature relationship is related to the total strain of the second optical fiber segment and the temperature of the area where the second optical fiber segment is located on the protective shells on both sides below the anode plate. The other type of temperature relationship is related to the average temperature of the protective shells on both sides below the anode plate and the strain of the first optical fiber segment on the anode plate caused by the average temperature. Both of these relationships establish the relationship between the optical fiber and the temperature of the protective shell. The relationship is simple to establish, and when the parameter in one of the relationships changes (such as the coverage area of ​​the corresponding area on the protective shell by the second optical fiber segment), only that relationship needs to be adjusted, without having to rebuild the entire relationship. Therefore, the temperature elimination is more flexible.

[0061] During battery charging and discharging, the anode plate generates heat. This heat can be transferred from the anode plate to the first sensing fiber segment and then to the second fiber segment. Simultaneously, it can be transferred to the protective shell for heat dissipation through the insulating layer and cathode layer. Since the second fiber segment is laid on the protective shell, a closed-loop heat dissipation path is formed, resulting in minimal temperature differences between the first and corresponding second fiber segments. Furthermore, this invention allows each second fiber segment 12 to completely cover the corresponding side of the protective shell 3 in the second direction. Because the contact length between the second fiber segment and the protective shell is longer, it facilitates faster thermal equilibrium and reduces the temperature difference between the first and second fiber segments. The lengths and shapes of all second fiber segments 12 can also be equal, allowing for a unified standard to be used to determine the corresponding temperature of the second fiber segment based on the relationship between the strain and temperature.

[0062] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0063] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined solely by the appended claims.

Claims

1. A battery state of health prediction method, characterized by, The method comprises the following steps: In step S100, the extent of shrinkage or expansion at different positions on the anode plate of the battery at the same time during the battery charging and discharging process is different, and correspondingly, the strain at different positions on the anode plate at the same time is different. Based on this characteristic, a network model is constructed according to the strain at different positions on the anode plate. In step S200, the current state of charge of the battery is predicted according to the strain and strain rate at different positions on the anode plate at the current time based on the corresponding network model, and / or the health state of the battery is predicted according to the maximum strain and residual strain on the anode plate in each cycle period so far based on the corresponding network model, wherein the strain rate is the difference in strain per unit time at the corresponding position on the anode plate, the maximum strain is the maximum strain minus the minimum strain on the anode plate measured in the corresponding cycle period, and the residual strain is the last strain minus the first strain measured on the anode plate in the corresponding cycle period. The strain at different positions on the anode plate is measured according to the following steps: In step S1, the sensing optical fiber covers the upper surface of the anode plate of the battery in a serpentine manner in a first direction, and each bending of the sensing optical fiber in a second direction is a first fiber segment, and each adjacent two first fiber segments are connected by a corresponding second fiber segment, and the first direction is perpendicular to the second direction. In step S2, for each first fiber segment, the two ends of the first fiber segment outside the anode plate are fixed to the protective shell under the anode plate respectively, so that the first fiber segment is subjected to a corresponding pre-stress to ensure that the first fiber segment always remains in a stretched state during monitoring. In step S3, one end of the sensing optical fiber is connected to a strain measurement and analysis system, and the two ends of each first fiber segment are pressed for position calibration. In step S4, the strain measurement and analysis system determines the strain at each measurement point on the first fiber segment according to the backscattered signals transmitted back by each first fiber segment and the calibrated positions of the two ends of the first fiber segment, reconstructs the two-dimensional strain field of the first fiber segment according to the strain at each measurement point on the first fiber segment, and obtains the three-dimensional strain field of the anode plate composed of the two-dimensional strain fields of each first fiber segment according to the layout relationship between each first fiber segment. The second fiber segment is laid on the protective shell under the anode plate, and before the two-dimensional strain field of the first fiber segment is reconstructed according to the strain at each measurement point on the first fiber segment in step S4, the strain measurement and analysis system further comprises the following steps: The strain measurement and analysis system determines the strain of each second fiber segment according to the backscattered signals transmitted back by each second fiber segment and the calibrated positions of the two ends of the first fiber segment, determines the temperature corresponding to each second fiber segment according to the strain of each second fiber segment, adds and averages the temperatures of each second fiber segment to obtain the average temperature of the battery, converts the average temperature into a corresponding strain, and subtracts the strain corresponding to the average temperature from the strain at each measurement point on the first fiber segment.

2. The battery state of health prediction method of claim 1, wherein, the step S100 specifically comprises: constructing a feedforward neural network, wherein the feedforward neural network comprises an input layer, a hidden layer and an output layer connected in sequence, the input quantities of the input layer are the strain and strain rate at different positions on the anode plate at the same time, and the output quantity of the output layer is the current state of charge of the battery; the step S200 specifically comprises: based on the feedforward neural network, predicting the current state of charge of the battery according to the strain and strain rate at different positions on the anode plate at the current time.

3. The battery state of health prediction method of claim 2, wherein, the feedforward neural network is constructed in the step S100 according to the following steps: the charging or discharging process in the use of the battery is taken as a cycle period respectively, all the strains and strain rates at different positions on the anode plate in different cycle periods are taken as a training set, and the training set is input into the feedforward neural network for pre-training; the predicted state of charge of the battery output by the output layer is compared with the actual state of charge of the battery, and each parameter in the feedforward neural network is adjusted according to the comparison result, so as to obtain an optimal parameter set; the parameter set comprises the number of layers of the hidden layer, the number of neurons of each layer of the hidden layer, the weight value of each input quantity in the input layer to each neuron in the first layer of the hidden layer, the weight value of each neuron in each layer of the hidden layer to each neuron in the next layer of the hidden layer, and the weight value of each neuron in the last layer of the hidden layer to each output quantity in the output layer.

4. The battery state of health prediction method according to claim 2 or 3, characterized by, the output quantity of the output layer is the current overall state of charge of the battery or the current state of charge at different corresponding positions on the battery.

5. The battery state of health prediction method of claim 1, wherein, the step S100 specifically comprises: constructing a long short-term memory network, wherein the long short-term memory network comprises an input layer, a long short-term memory network layer, a fully connected layer and an output layer connected in sequence, the input quantities of the input layer are the maximum strain and residual strain on the anode plate in each cycle period so far, and the output quantity of the output layer is the current state of health of the battery; the step S200 specifically comprises: based on the long short-term memory network, predicting the state of health of the battery according to the maximum strain and residual strain on the anode plate in the current cycle period.

6. The battery state of health prediction method of claim 5, wherein, the cycle period is a time period corresponding to each charging or discharging process in the use of the battery; the output of the output layer in the long short-term memory network is a time sequence signal without period change, and the current state of health of the battery is the maximum capacity of the current battery divided by the initial capacity of the battery.

7. The battery state of health prediction method of claim 1, wherein, the anode plate, the isolation layer and the cathode plate are sequentially arranged on the protection shell from top to bottom.

8. The battery state of health prediction method of claim 1, wherein, Each second optical fiber segment completely covers the protection shell on the corresponding side in the second direction, and the lengths of the second optical fiber segments are equal.

Images