Tab offset model generation method, device, equipment and storage medium

Abstract

The present invention relates to the technical field of battery cell winding control, and discloses a method, device, equipment and storage medium for generating a tab offset model, which is used to fit a target linear error parameter and a plurality of target nonlinear error parameters through a tab offset model, thereby improving the accuracy of tab offset prediction. The tab offset model generation method includes: constructing an initial tab offset model according to the target linear error parameter, a plurality of target nonlinear error parameters and a tab angle offset corresponding to each battery cell layer, and initializing the model parameters of the initial tab offset model; training the initial tab offset model according to the target linear error parameter, the plurality of target nonlinear error parameters and the tab angle offset corresponding to each battery cell layer, and obtaining a target tab offset model.

Description

Technical Field

[0001] The present invention relates to the technical field of battery cell winding control, and in particular to a method, device, equipment and storage medium for generating a tab offset model. Background Art

[0002] Multi-electrode batteries can effectively improve the high-rate discharge characteristics and safety of lithium-ion batteries. Figure 1 Take the cylindrical tab battery cell of FIG. 1 as an example. The cylindrical tab battery cell is as follows. Figure 1 (a) The plurality of positive electrode tabs 11 and the plurality of negative electrode tabs 12 are symmetrically distributed, and the tabs on the same side of the standard battery cell are aligned after winding. Figure 1 (b) As shown, however, the standard battery cell has very strict control over the process flow. During the winding process of the lithium battery cell with the same side tab, the alignment of the tabs on the same side is difficult to ensure due to various linear and nonlinear influencing factors, and it is easy to appear after winding. Figure 1 (c) The same-side tab offset situation results in the scrapping of multi-tab batteries.

[0003] Most of them control the accuracy of roller coating in the front-end process, reduce material thickness fluctuations, focus on improving the accuracy and performance of winding molding equipment, reduce tension fluctuations and correction amplitude fluctuations, and strictly control the temperature and humidity of material transportation, storage and production. However, the above methods cannot completely eliminate fluctuation errors, resulting in a low yield rate of multi-pole battery cells.

[0004] Since the influencing factors of the same-side tab alignment include multiple linear factors and multiple nonlinear factors, the types of influencing factors of the same-side tab alignment are complex and highly coupled. When the winding equipment is unstable, the multiple nonlinear factors of each battery cell layer are not constants, and an accurate functional relationship cannot be established through mathematical deduction. The accuracy of the tab offset determined by the traditional mathematical expression is low. Summary of the invention

[0005] The present invention provides a method, device, equipment and storage medium for generating a tab offset model, which are used to fit a target linear error parameter and multiple target nonlinear error parameters through the tab offset model, thereby improving the accuracy of tab offset prediction.

[0006] A first aspect of the present invention provides a method for generating a tab offset model, comprising: acquiring multiple sample cell data, each sample cell data comprising multiple winding parameters, multiple sensing sequence data and a tab angle offset corresponding to each cell layer; preprocessing each sample cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each cell layer; constructing an initial tab offset model according to the target linear error parameter corresponding to each cell layer, the multiple target nonlinear error parameters and the tab angle offset, and initializing the model parameters of the initial tab offset model; training the initial tab offset model according to the target linear error parameter corresponding to each cell layer, the multiple target nonlinear error parameters and the tab angle offset to obtain a target tab offset model.

[0007] The second aspect of the present invention provides a device for generating a pole lug offset model, comprising: an acquisition module for acquiring multiple sample battery cell data, each sample battery cell data including multiple winding parameters, multiple sensing sequence data and a pole lug angle offset corresponding to each battery cell layer; a preprocessing module for preprocessing each sample battery cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer; a construction module for constructing an initial pole lug offset model according to the target linear error parameter corresponding to each battery cell layer, the multiple target nonlinear error parameters and the pole lug angle offset, and initializing the model parameters of the initial pole lug offset model; a training module for training the initial pole lug offset model according to the target linear error parameter corresponding to each battery cell layer, the multiple target nonlinear error parameters and the pole lug angle offset to obtain a target pole lug offset model.

[0008] The third aspect of the present invention provides a tab offset model generation device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the tab offset model generation device executes the above-mentioned tab offset model generation method.

[0009] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for generating a tab offset model.

[0010] In the technical solution provided by the present invention, a target pole lug offset model is constructed and trained by using target linear error parameters, multiple target nonlinear error parameters and the pole lug angle offset corresponding to each battery core layer, thereby achieving the fitting of linear influencing factors and nonlinear influencing factors and improving the accuracy of the target pole lug offset in predicting the pole lug offset. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Schematic diagram of the distribution of pole ears of a cylindrical battery cell in an embodiment of the present invention;

[0012] Figure 2 Schematic diagram of an integrated system for cell winding processing in an embodiment of the present invention;

[0013] Figure 3 A schematic diagram of an embodiment of a method for generating a tab offset model in an embodiment of the present invention;

[0014] Figure 4 A schematic diagram of another embodiment of a method for generating a tab offset model in an embodiment of the present invention;

[0015] Figure 5 A schematic diagram of a neural network structure in an embodiment of the present invention;

[0016] Figure 6 A schematic diagram of an embodiment of a device for generating a tab offset model in an embodiment of the present invention;

[0017] Figure 7 It is a schematic diagram of another embodiment of a device for generating a tab offset model in an embodiment of the present invention;

[0018] Figure 8 It is a schematic diagram of an embodiment of a device for generating a tab offset model in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention provides a method, device, equipment and storage medium for generating a pole ear offset model, which are used to solve the technical problem that the influencing factors of the same-side pole ear alignment include multiple linear factors and multiple nonlinear factors, the types of influencing factors of the same-side pole ear alignment are complex and highly coupled, and when the winding equipment is unstable, it is impossible to establish an accurate mathematical model through mathematical deduction, and the compensation amount output by the traditional model has low accuracy.

[0020] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or system that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or systems.

[0021] It is understandable that the execution subject of the present invention can be a cell manufacturing and forming device such as a cutting and winding machine, or a cell winding processing integrated system, or a server, which is not limited here. The embodiment of the present invention is described by taking the cell winding processing integrated system as the execution subject as an example.

[0022] The same-side tabs shown in this embodiment refer to the positive electrode tabs on the same side of a multi-tab battery cell, or the positive electrode tabs on the same side of a multi-tab battery cell. The method for adjusting the alignment of the tabs on both sides is similar. This embodiment takes the influencing factors of the alignment of multiple tabs on the negative electrode sheet as an example for explanation.

[0023] The battery cell winding processing integrated system of this embodiment refers to an integrated control system for die-cutting, winding and other integrated processing of cylindrical or square battery cells. Please refer to Figure 2 , schematic diagram of the unwinding, thickness measurement, die-cutting, deviation correction and winding of the negative electrode sheet in the cylindrical battery cell winding processing integrated system.

[0024] The cylindrical battery cell winding processing integrated system mainly consists of a negative electrode sheet unwinding mechanism 21 , a thickness measuring mechanism 22 , a laser die-cutting mechanism 23 , a buffer mechanism 24 , and a winding mechanism 25 .

[0025] The negative electrode sheet unwinding mechanism 21 includes a first unwinding device 211 and a first tension swing rod 212, wherein the first unwinding device 211 is used to automatically unwind the negative electrode sheet, and the first tension swing rod 212 is used to control the tension during the unwinding process.

[0026] The thickness measuring mechanism 22 includes a thickness gauge 221 and a first driving device 222, wherein the thickness gauge 221 is used to collect thickness sequence data of the negative electrode sheet, that is, the real-time thickness of the negative electrode sheet strip; the first driving device 222 is used to provide power for the movement of the negative electrode sheet strip;

[0027] The laser die-cutting mechanism 23 includes a laser cutting device 231 and a second driving device 232. The laser cutting device 231 is used to cut the tabs at the designated positions of the negative electrode sheet, the second driving device 232 is used to provide power for the movement of the negative electrode sheet, and the cache mechanism 24 is used to buffer and temporarily store the negative electrode sheet.

[0028] The winding mechanism 25 includes a third driving device 251, a second tension swing rod 252, a correction control device 253, and a winding device 254, wherein the third driving device 251 provides power for the movement of the negative electrode sheet material strip, the second tension swing rod 252 controls the tension during the winding process, the correction control device 253 is used to correct the negative electrode sheet material strip before winding, and the winding device 254 is used to wind the negative electrode sheet material strip 26, the upper diaphragm material strip 27, the positive electrode sheet material strip 28 and the lower diaphragm material strip 29.

[0029] The cylindrical battery cell winding processing integrated system based on this embodiment is used to process multi-tab battery cells, wherein the laser die-cutting mechanism 23 may have tab cutting errors, and the first unwinding device 211 may introduce negative electrode sheet feeding errors, and the winding device 254 may have linear errors such as winding needle size tolerances.

[0030] When the above-mentioned cylindrical battery cell winding processing integrated system is unstable, there will be tension fluctuations in the first tension swing rod 212 and the second tension swing rod 252, there will be correction fluctuations in the correction control device 253, and there will be thickness fluctuations in the thickness of the negative electrode strip. The above-mentioned fluctuations are not constant values, and it is impossible to establish an accurate functional relationship through mathematical deduction. The accuracy of the pole ear offset determined by the traditional mathematical expression is low.

[0031] It is understandable that the above-mentioned linear error and nonlinear error may also have other errors depending on the device composition and processing flow in the battery cell winding processing integrated system, and this embodiment does not make any specific limitation.

[0032] To solve the above problems, this embodiment provides an embodiment of a method for generating a tab offset model, referring to Figure 3 The specific process is described as follows:

[0033] 301. Acquire multiple sample battery cell data, each sample battery cell data including multiple winding parameters, multiple sensor sequence data, and a tab angle offset corresponding to each battery cell layer.

[0034] The battery cell winding processing integrated system obtains multiple sample battery cell data for constructing a training sample set. Each sample battery cell is a battery cell formed by a winding process with a negative electrode sheet, a positive electrode sheet, and upper and lower separators. Each sample battery cell includes multiple battery cell layers.

[0035] In this embodiment, the various winding parameters refer to the equipment parameters that affect the alignment of the pole ears in each mechanism and device in the battery cell winding processing integrated system, for example, the winding needle size tolerance corresponding to the winding needle in the winding device 254, that is, the difference between the maximum limit size and the minimum limit size of the winding needle; the feeding error corresponding to the positive and negative electrode sheet unwinding mechanisms, that is, the difference between the actual feeding length and the target feeding length; the pole ear cutting error corresponding to the positive and negative electrode sheet laser die-cutting devices, that is, the difference between the actual cutting position of the pole ear and the target cutting position. The specific values ​​can be obtained by reading the equipment parameters, or calibrated according to the actual conditions of each mechanism or device over a period of time. The influence of each of the above winding parameters on the pole ear offset of each battery cell layer is linear.

[0036] In this embodiment, the multiple sensor sequence data refer to the sensor sequence data that affect the alignment of the pole lugs collected by sensors of various mechanisms and devices in the battery cell winding processing integrated system, such as thickness sequence data, tension sequence data and correction sequence data. The influence of each of the above sensor sequence data on the pole lug offset of each battery cell layer is nonlinear.

[0037] The lug angle offset of this embodiment can be obtained by an angle sensor to obtain the lug angle offset of each cell layer in each sample cell. The lug angle offset is based on the innermost lug of each cell. The offset angle between any outer lug and the innermost lug is expressed as Figure 1 Take the angle offset of the outermost tab of a cylindrical battery cell as an example, where the angle between the perpendicular bisector of the innermost tab and the perpendicular bisector of the outermost tab, whose center is the corner point, is the angle offset of the outermost tab.

[0038] 302. Preprocess each sample battery cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer.

[0039] In this embodiment, multiple winding parameters corresponding to each battery cell layer are superimposed to simplify the network structure of the subsequent model, each type of sensing sequence data of each battery cell layer is discretized according to a preset degree of each battery cell layer, and the mean value of the fluctuation error of each sensing sequence data is solved to obtain multiple target nonlinear error parameters corresponding to each battery cell layer.

[0040] 303. Construct an initial tab offset model according to target linear error parameters, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer, and initialize model parameters of the initial tab offset model.

[0041] The initial tab offset model of this embodiment adopts a back propagation neural network (BPNN) model, and the initial tab offset model includes an input layer, a hidden layer, and an output layer.

[0042] By taking the target linear error parameters and the number of multiple target nonlinear error parameters as the input of the initial lug offset model, and taking the lug angle offset as the output of the initial lug offset model, the number of neuron nodes in the input layer and the output layer of the initial lug offset model can be determined. The number of neuron nodes in the output layer is 1. For example, if the multiple target nonlinear error parameters are thickness fluctuation error, tension fluctuation error and correction fluctuation error, a total of four target error parameters are input, and the number of neuron nodes in the input layer of the initial lug offset model is determined to be 4, and the number of neuron nodes in the hidden layer is determined according to actual conditions.

[0043] The model parameters of the above-mentioned initial pole ear offset model include the weight matrix and bias matrix between the front and rear two network layer structures, the activation function, the learning rate, the loss threshold and the iteration number threshold. The model parameters of the initial pole ear offset model are initialized to obtain the initial pole ear offset model.

[0044] 304. Train the initial tab offset model according to the target linear error parameter, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer to obtain a target tab offset model.

[0045] The target linear error parameters and multiple target nonlinear error parameters corresponding to each battery cell layer are determined as the input of the initial tab offset model, and the tab angle offset of the corresponding battery cell layer is determined as the output of the initial tab offset model. The initial tab offset model is trained on multiple sample battery cell data until the loss function converges to obtain the target tab offset model.

[0046] In an embodiment of the present invention, a target lug offset model is constructed and trained by using target linear error parameters, multiple target nonlinear error parameters, and the lug angle offset corresponding to each battery core layer, thereby achieving the fitting of linear influencing factors and nonlinear influencing factors and improving the accuracy of the target lug offset in predicting the lug offset.

[0047] See also Figure 4 Another embodiment of the method for generating a tab offset model in the embodiment of the present invention includes:

[0048] 401. Acquire multiple sample cell data, each sample cell data including multiple winding parameters, multiple sensor sequence data, and a tab angle offset corresponding to each cell layer.

[0049] 402. Preprocess each sample battery cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer.

[0050] The sum of multiple winding parameters of each sample battery cell data is determined as the target linear error parameter corresponding to each battery cell layer; the multiple sensing sequence data of each sample battery cell data are discretized according to the preset material strip length of each battery cell layer to obtain each nonlinear sensing sequence data segment corresponding to each battery cell layer; the fluctuation error of each nonlinear sensing sequence data segment corresponding to each battery cell layer is averaged to obtain multiple target nonlinear error parameters corresponding to each battery cell layer.

[0051] The various winding parameters of this embodiment include the winding needle size tolerance e r , Laser cutting error e c , pole piece feeding error e d , target linear error parameter eLin =e r +e c +e d ; Multiple target nonlinear error parameters including thickness fluctuation error parameter e Thi , correction fluctuation error parameter e Ali and tension fluctuation error parameter e Ten .

[0052] 403. Construct an initial tab offset model according to the target linear error parameter, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer, and initialize the model parameters of the initial tab offset model.

[0053] The number of network layers of the initial lug offset model and the number of nodes corresponding to each network layer are determined according to the target linear error parameter, multiple target nonlinear error parameters and the lug angle offset corresponding to each battery cell layer; the weight matrix and bias matrix between each network layer of the initial lug offset model are initialized, and the Sigmond function is determined as the activation function of the initial lug offset model to obtain the initial lug offset model.

[0054] Reference Figure 5 The neural network structure, in which the winding needle size tolerance e r , Laser cutting error e c , pole piece feeding error e d , target linear error parameter e Lin =e r +e c +e d ; Multiple target nonlinear error parameters including thickness fluctuation error parameter e Thi , correction fluctuation error parameter e Ali and tension fluctuation error parameter e Ten is the input of the initial tab offset model. The initial tab offset model has 4 input layer nodes, 4 hidden layer nodes, and 1 output layer node.

[0055] 404. Train the initial tab offset model according to the target linear error parameter, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer to obtain a target tab offset model.

[0056] The input layer of the initial lug offset model is input according to the target linear error parameters and multiple target nonlinear error parameters, and operations are performed according to the weight matrix and bias matrix between each network layer to obtain the output of the hidden layer, and the output of the hidden layer is mapped to the lug angle offset of the corresponding battery cell layer through the Sigmond function. The loss value is calculated according to the lug angle offset of each battery cell layer, the preset expected angle offset and the preset loss function; if the loss value is greater than the preset loss threshold, the initial lug offset model is iteratively updated through the loss function and the preset learning rate until the loss value is less than or equal to the loss threshold, so as to obtain the target lug offset model.

[0057] Specifically, the error matrix of each cell layer is e=[e Lin e Thi e Ali e Ten ] Input, initialize the first weight matrix W and the first bias matrix β of the input layer and the hidden layer.

[0058] Among them, the first weight matrix

[0059] The input of the hidden layer is To prevent the network learning from falling into the local area, the first bias matrix β is added.

[0060] The output of the hidden layer is h i =f(z i +β i ), in this embodiment, the number of neuron nodes in the output layer and the hidden layer is the same, i is the number of neuron nodes in the input layer i∈[1,2,3,4].

[0061] The input of the output layer is Among them, v i is the second weight of the i-th neuron node in the hidden layer, the second weight matrix V=[v1 v2 v3 v4], and the second bias parameter λ.

[0062] Output of the output layer

[0063] The loss function formula of this embodiment is as follows:

[0064]

[0065] in, is the desired angle offset, Δθ is the predicted angle offset of the output layer.

[0066] In a feasible implementation, if the loss value is greater than a preset loss threshold, the initial pole lug offset model is iteratively updated through a loss function and a preset learning rate until the loss value is less than or equal to the loss threshold, and the target pole lug offset model is obtained, including: if the loss value is greater than the preset loss threshold, the model parameters are updated through a loss function and a preset learning rate, wherein the model parameters include a first weight matrix, a first bias matrix, a second weight matrix and a second bias parameter, to obtain a candidate pole lug offset model; according to the target linear error parameter and a plurality of target nonlinear error parameters, the updated candidate pole lug offset model is inputted thereto, until the loss value is less than or equal to the loss threshold, to obtain the target pole lug offset model.

[0067] Specifically, based on the chain rule, the update formula for each model parameter can be obtained, where the update formula for the second bias parameter is as follows:

[0068]

[0069] in, is the updated second bias parameter, Δλ is the change of the second bias parameter, Δλ changes fastest along the direction of the negative gradient, and η is the learning rate;

[0070] The update formula for each second weight parameter in the second weight matrix V is as follows:

[0071]

[0072] in, is the second weight parameter of the i-th neuron node in the hidden layer after update, Δv i is the change in the second weight parameter of the i-th neuron node in the hidden layer, h i is the output value of the i-th neuron node in the hidden layer;

[0073] The update formula for each first bias parameter in the first bias matrix is ​​as follows:

[0074]

[0075] in, is the first bias parameter of the i-th neuron node in the hidden layer after update, Δb i is the change in the first bias parameter of the i-th neuron node in the hidden layer;

[0076] The update formula for each first weight parameter in the first weight matrix W is w 11 The update formula is as follows:

[0077]

[0078] in, is the updated first bias parameter, Δw 11 is the change of the first bias parameter.

[0079] 405. Input a plurality of target nonlinear error parameters and target linear error parameters of a target battery layer in a real-time winding process into a target tab offset model to obtain a predicted angle offset of the target battery layer.

[0080] The thickness fluctuation error parameters, correction fluctuation error parameters, tension fluctuation error parameters and target linear errors of the target battery cell layer in the real-time winding process are input into the output layer of the target tab offset model; the first weight matrix and the first bias matrix of the input layer and the hidden layer are calculated and input into the hidden layer; the second weight matrix and the second bias parameters of the hidden layer and the output layer are calculated and input into the output layer, and mapped through the preset activation function in the output layer to obtain the predicted angle offset of the target battery cell layer.

[0081] 406. Dynamically compensate initial laser die-cutting parameters of the target battery layer according to the predicted angle offset of the target battery layer.

[0082] The predicted angle offset of the target battery layer, the total strip thickness corresponding to the target battery layer, the preset winding needle radius and the preset compensation formula are used for calculation to obtain the tab spacing offset corresponding to the target battery layer; the initial laser die-cutting parameters are adjusted according to the tab spacing offset corresponding to the target battery layer to obtain the target laser die-cutting parameters.

[0083] The above compensation formula is:

[0084]

[0085] Where, ΔL j is the tab spacing offset corresponding to the jth cell layer, Δθ j is the predicted angle offset corresponding to the jth cell layer, δ j is the thickness corresponding to the jth battery layer, and r0 is the winding needle radius.

[0086] In an embodiment of the present invention, a target pole lug offset model is constructed and trained by using target linear error parameters, multiple target nonlinear error parameters, and the pole lug angle offset corresponding to each battery core layer, thereby achieving the fitting of linear influencing factors and nonlinear influencing factors and obtaining a target pole lug offset model. The laser die-cutting parameters are dynamically compensated through the target pole lug offset model, thereby improving the alignment of the pole lugs on the same side.

[0087] The above describes the method for generating a tab offset model in an embodiment of the present invention. The following describes the device for generating a tab offset model in an embodiment of the present invention. Figure 6In one embodiment of the present invention, a device for generating a tab offset model includes:

[0088] An acquisition module 601 is used to acquire a plurality of sample battery cell data, each of which includes a plurality of winding parameters, a plurality of sensing sequence data, and a tab angle offset corresponding to each battery cell layer;

[0089] A preprocessing module 602 is used to preprocess each sample battery cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer;

[0090] A construction module 603 is used to construct an initial tab offset model according to the target linear error parameter corresponding to each battery core layer, multiple target nonlinear error parameters and the tab angle offset, and initialize the model parameters of the initial tab offset model;

[0091] The training module 604 is used to train the initial tab offset model according to the target linear error parameter, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer to obtain the target tab offset model.

[0092] In an embodiment of the present invention, a target lug offset model is constructed and trained by using target linear error parameters, multiple target nonlinear error parameters, and the lug angle offset corresponding to each battery core layer, thereby achieving the fitting of linear influencing factors and nonlinear influencing factors and improving the accuracy of the target lug offset in predicting the lug offset.

[0093] See also Figure 7 Another embodiment of the device for generating a tab offset model in the embodiment of the present invention includes:

[0094] An acquisition module 601 is used to acquire a plurality of sample battery cell data, each of which includes a plurality of winding parameters, a plurality of sensing sequence data, and a tab angle offset corresponding to each battery cell layer;

[0095] A preprocessing module 602 is used to preprocess each sample battery cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer;

[0096] A construction module 603 is used to construct an initial tab offset model according to the target linear error parameter corresponding to each battery core layer, multiple target nonlinear error parameters and the tab angle offset, and initialize the model parameters of the initial tab offset model;

[0097] The training module 604 is used to train the initial tab offset model according to the target linear error parameter, multiple target nonlinear error parameters and tab angle offset corresponding to each battery core layer to obtain the target tab offset model.

[0098] Optionally, the preprocessing module 602 is specifically used to: determine the sum of multiple winding parameters of each sample battery cell data as a target linear error parameter corresponding to each battery cell layer;

[0099] Discretize the multiple sensing sequence data of each sample battery cell data according to the preset strip length of each battery cell layer to obtain each nonlinear sensing sequence data segment corresponding to each battery cell layer;

[0100] The fluctuation error of each nonlinear sensing sequence data segment corresponding to each battery layer is averaged to obtain multiple target nonlinear error parameters corresponding to each battery layer.

[0101] Optionally, the tab offset model generating device also includes: a prediction module 605, which is used to input a plurality of target nonlinear error parameters and target linear error parameters of the target battery layer in the real-time winding process into the target tab offset model to obtain a predicted angle offset of the target battery layer; and a compensation module 606, which is used to dynamically compensate for the initial laser die-cutting parameters of the target battery layer according to the predicted angle offset of the target battery layer.

[0102] Optionally, the prediction module 605 is specifically used to: input the thickness fluctuation error parameter, the correction fluctuation error parameter, the tension fluctuation error parameter and the target linear error of the target battery layer in the real-time winding process into the output layer of the target tab offset model;

[0103] Perform calculations based on the first weight matrix and the first bias matrix of the input layer and the hidden layer and input the matrix into the hidden layer;

[0104] The second weight matrix and the second bias parameter of the hidden layer and the output layer are calculated and input into the output layer, and mapped through the activation function preset in the output layer to obtain the predicted angle offset of the target battery cell layer.

[0105] The compensation module 606 is specifically used to: calculate according to the predicted angle offset of the target battery layer, the total strip thickness corresponding to the target battery layer, the preset winding needle radius and the preset compensation amount formula to obtain the tab spacing offset corresponding to the target battery layer;

[0106] The initial laser die-cutting parameters are adjusted according to the tab spacing offset corresponding to the target battery core layer to obtain the target laser die-cutting parameters.

[0107] Optionally, the construction module 603 is specifically used to: determine the number of network layers of the initial tab offset model and the number of nodes corresponding to each network layer according to the target linear error parameter, multiple target nonlinear error parameters and the tab angle offset corresponding to each battery core layer;

[0108] The weight matrix and bias matrix between each network layer of the initial pole lug offset model are initialized, and the Sigmond function is determined as the activation function of the initial pole lug offset model to obtain the initial pole lug offset model.

[0109] Optionally, the training module 604 is specifically used to: input the input layer of the initial tab offset model according to the target linear error parameter and the multiple target nonlinear error parameters, and perform calculations according to the weight matrix and the bias matrix between each network layer to obtain the output of the hidden layer;

[0110] The output of the hidden layer is mapped to the angle offset of the tab of the corresponding battery layer through the Sigmond function, and the loss value is calculated according to the angle offset of the tab of each battery layer, the preset expected angle offset and the preset loss function;

[0111] If the loss value is greater than the preset loss threshold, the initial tab offset model is iteratively updated using the loss function and the preset learning rate until the loss value is less than or equal to the loss threshold, thereby obtaining the target tab offset model.

[0112] In an embodiment of the present invention, a target pole lug offset model is constructed and trained by using target linear error parameters, multiple target nonlinear error parameters, and the pole lug angle offset corresponding to each battery core layer, thereby achieving the fitting of linear influencing factors and nonlinear influencing factors and obtaining a target pole lug offset model. The laser die-cutting parameters are dynamically compensated through the target pole lug offset model, thereby improving the alignment of the pole lugs on the same side.

[0113] above Figure 6 and Figure 7 The tab offset model generating device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The tab offset model generating device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0114] See also Figure 8 As shown, the tab offset model generating device includes a processor 800 and a memory 801 , wherein the memory 801 stores machine executable instructions that can be executed by the processor 800 , and the processor 800 executes the machine executable instructions to implement the above-mentioned tab offset model generating method.

[0115] Further, Figure 8 The tab offset model generation device shown further includes a bus 802 and a communication interface 803 , and the processor 800 , the communication interface 803 and the memory 801 are connected via the bus 802 .

[0116] The memory 801 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), for example, at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 803 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 802 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0117] The processor 800 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 800. The above processor 800 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present disclosure can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present disclosure can be directly embodied as a hardware decoding processor to be executed, or a combination of hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 801 , and the processor 800 reads the information in the memory 801 and completes the method steps of the above-mentioned embodiment in combination with its hardware.

[0118] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the method for generating a pole ear offset model.

[0119] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0120] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer system (which can be a personal computer, a server, or a network system, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0121] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a tab offset model, characterized in that: The method for generating a tab offset model comprises: Acquire multiple sample battery cell data, each sample battery cell data includes multiple winding parameters, multiple sensor sequence data and a tab angle offset corresponding to each battery cell layer; Preprocess each sample battery cell data to obtain the target linear error parameter and multiple target nonlinear error parameters corresponding to each battery cell layer; Constructing an initial tab offset model according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset, and initializing model parameters of the initial tab offset model; The initial tab offset model is trained according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset to obtain a target tab offset model.

2. The method for generating a tab offset model according to claim 1, characterized in that: The preprocessing of each sample cell data to obtain a target linear error parameter and a plurality of target nonlinear error parameters corresponding to each cell layer includes: The sum of multiple winding parameters of each sample battery cell data is determined as a target linear error parameter corresponding to each battery cell layer; Discretize the multiple sensing sequence data of each sample battery cell data according to the preset strip length of each battery cell layer to obtain each nonlinear sensing sequence data segment corresponding to each battery cell layer; The fluctuation error of each nonlinear sensing sequence data segment corresponding to each battery layer is averaged to obtain multiple target nonlinear error parameters corresponding to each battery layer.

3. The method for generating a tab offset model according to claim 1, characterized in that: After the initial tab offset model is trained according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset to obtain the target tab offset model, the method further includes: Inputting a plurality of target nonlinear error parameters and target linear error parameters of a target battery layer in a real-time winding process into the target tab offset model to obtain a predicted angular offset of the target battery layer; The initial laser die-cutting parameters of the target battery core layer are dynamically compensated according to the predicted angle offset of the target battery core layer.

4. The method for generating a tab offset model according to claim 3, characterized in that: The target tab offset model includes an input layer, a hidden layer and an output layer, the multiple target nonlinear error parameters include a thickness fluctuation error parameter, a correction fluctuation error parameter and a tension fluctuation error parameter, and the multiple target nonlinear error parameters and target linear error parameters of the target battery layer in the real-time winding process are input into the target tab offset model to obtain the predicted angle offset of the target battery layer, including: Inputting the thickness fluctuation error parameter of the target battery layer in the real-time winding process, the correction fluctuation error parameter, the tension fluctuation error parameter and the target linear error into the input layer of the target tab offset model; Perform calculations based on the first weight matrix and the first bias matrix of the input layer and the hidden layer and input the calculations into the hidden layer; The second weight matrix and the second bias parameter of the hidden layer and the output layer are calculated and input into the output layer, and mapped through the activation function preset in the output layer to obtain the predicted angle offset of the target battery cell layer.

5. The method for generating a tab offset model according to claim 3, characterized in that: The dynamically compensating the initial laser die-cutting parameters of the target battery core layer according to the predicted angle offset of the target battery core layer comprises: Calculating according to the predicted angle offset of the target battery layer, the total strip thickness corresponding to the target battery layer, the preset winding needle radius and the preset compensation amount formula, to obtain the tab spacing offset corresponding to the target battery layer; The initial laser die-cutting parameters are adjusted according to the tab spacing offset corresponding to the target battery core layer to obtain the target laser die-cutting parameters.

6. The method for generating a tab offset model according to any one of claims 1 to 5, characterized in that: The initial tab offset model is constructed according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset, and the model parameters of the initial tab offset model are initialized, including: Determine the number of network layers of the initial tab offset model and the number of nodes corresponding to each network layer according to the target linear error parameter, the multiple target nonlinear error parameters and the tab angle offset corresponding to each battery core layer; The weight matrix and the bias matrix between each network layer of the initial tab offset model are initialized, and the Sigmond function is determined as the activation function of the initial tab offset model to obtain the initial tab offset model.

7. The method for generating a tab offset model according to claim 6, characterized in that: The initial tab offset model is trained according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset to obtain the target tab offset model, including: Inputting the target linear error parameter and multiple target nonlinear error parameters into the input layer, and performing operations according to the weight matrix and the bias matrix between each network layer to obtain the output of the hidden layer; The output of the hidden layer is mapped to the angle offset of the tab of the corresponding battery layer through the Sigmond function, and the loss value is calculated according to the angle offset of the tab of each battery layer, the preset expected angle offset and the preset loss function; If the loss value is greater than a preset loss threshold, the initial tab offset model is iteratively updated using the loss function and a preset learning rate until the loss value is less than or equal to the loss threshold, thereby obtaining a target tab offset model.

8. A device for generating a tab offset model, characterized in that: The tab offset model generating device comprises: An acquisition module is used to acquire multiple sample battery cell data, each sample battery cell data includes multiple winding parameters, multiple sensing sequence data and a tab angle offset corresponding to each battery cell layer; A preprocessing module is used to preprocess each sample cell data to obtain a target linear error parameter and multiple target nonlinear error parameters corresponding to each cell layer; A construction module, used to construct an initial tab offset model according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset, and initialize the model parameters of the initial tab offset model; A training module is used to train the initial tab offset model according to the target linear error parameter corresponding to each battery core layer, the multiple target nonlinear error parameters and the tab angle offset to obtain a target tab offset model.

9. A device for generating a tab offset model, characterized in that: The tab offset model generating device comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the tab offset model generation device to execute the tab offset model generation method according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is read and executed, the method for generating a tab offset model as described in any one of claims 1 to 7 is executed.

Images