A power battery cascade utilization decision method and system based on scenario matching

Abstract

The present invention belongs to the technical field related to power battery recycling, and discloses a power battery cascade utilization decision method based on scenario matching, including: obtaining standard usage parameters for cascade utilization of power batteries in various practical application scenarios, as well as basic parameters and performance parameters of the power batteries; constructing a neural network classification model, and using the original data set for training; based on the trained neural network classification model, comparing the obtained power battery performance parameters with the standard usage parameters for cascade utilization of power batteries in various practical application scenarios, and then obtaining the decision result of cascade utilization of power batteries. The present invention also discloses a corresponding system. Through the present invention, not only can big data be processed more quickly and comprehensively, and decision-making efficiency and accuracy can be improved, but also the disassembly depth decision of various retired power batteries and the matching decision problems with various practical application scenarios can be effectively solved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to power battery recycling, and more specifically, relates to a power battery tiered utilization decision method and system based on scene matching. Background Art

[0002] With the continuous development of new energy electric vehicles, a large number of power batteries on the market have entered the scrapping stage. It is estimated that by 2022, the recycling volume of retired power batteries will be close to 45GW·h, and this only refers to the battery performance of the power battery falling to 80% or less of the original performance. Although it does not meet the use standards of electric vehicles, it can still be used in occasions with low requirements for power battery performance, that is, entering the stage of cascade utilization. Cascade utilization can not only alleviate the pressure of a large number of retired power batteries entering the recycling stage, but also effectively reduce the consumption of corresponding national resources and improve the efficiency of resource use.

[0003] At present, recycling companies mainly disassemble retired power batteries into single-body form for utilization. This not only fails to maximize the residual value of retired power batteries, resulting in energy waste, but also greatly increases disassembly costs and reduces economic efficiency. However, in the process of cascade utilization, making effective and accurate disassembly decisions for retired power batteries and matching appropriate application scenarios are the prerequisites for achieving cascade utilization, and are also the key restrictive factors for the cascade utilization of retired power batteries. Without the correct cascade utilization decision, many power batteries that originally have cascade utilization value will often be over-disassembled, resulting in unnecessary waste of resources.

[0004] Patent search found that some solutions have been proposed in the prior art for the problem of cascade utilization detection of power batteries. For example, CN102755966A discloses a power battery cascade utilization sorting and evaluation method, which uses industrial CT non-destructive testing and three-dimensional imaging technology to detect the internal structure of retired batteries and achieve three-dimensional imaging to determine the health status of the battery; CN112966208A discloses a multi-parameter impact screening method for cascade utilization of electric vehicle power batteries, which records and processes the first-layer screening data and the second-layer screening data, and assigns different weight coefficients to each influencing factor after unified calculation domain transformation, so as to quickly obtain the power battery cascade utilization screening results.

[0005] However, further research shows that the above-mentioned existing technologies still have the following deficiencies or defects: First, the above-mentioned methods focus on the appearance and internal inspection of power batteries, and there is not much research on subsequent data algorithm processing, especially no scientific classification model has been established to process big data more quickly and comprehensively, which leads to the need to further improve the detection efficiency and accuracy; secondly, the above-mentioned detection method does not take into account the factors of different complex application scenarios, and in fact the disassembled power batteries need to meet the standard parameters of various application scenarios as accurately as possible. Summary of the invention

[0006] In view of the above defects or needs of the prior art, the purpose of the present invention is to provide a method and system for decision-making on cascade utilization of power batteries based on scenario matching, wherein a neural network algorithm is used to establish a unified cascade utilization model by fully considering and combining the characteristic analysis of the cascade utilization effect of power batteries and their application scenarios. Accordingly, not only can big data be processed more quickly and comprehensively, and decision-making efficiency and accuracy can be improved, but also the basic parameters and performance parameters of the power battery can be matched to the standard parameters of various application scenarios to the greatest extent, thereby solving the problem of large-scale series and parallel use of power batteries. Therefore, it is particularly suitable for application scenarios such as industrial-grade power battery disassembly and recycling.

[0007] To achieve the above object, according to one aspect of the present invention, a power battery cascade utilization decision method based on scene matching is provided, characterized in that the method comprises:

[0008] Step (1): obtaining standard usage parameters of cascade utilization of power batteries in various practical application scenarios and forming a first data set;

[0009] Step (2): performing appearance inspection and capacity test on the power battery to determine whether the power battery is detachable; if the power battery is detachable, further testing to obtain basic parameters and performance parameters of the power battery to form a second data set;

[0010] Step (3): constructing a neural network multi-label scenario classification model, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels, and the classification model is trained using the original data set composed of the first and second data sets;

[0011] Step (4): For the power battery to be tested, its basic parameters and performance parameters are input into the trained classification model as feature sets, and the classification and ranking results of the power battery's suitable application scenarios are output, thereby obtaining the decision results of the power battery's tiered utilization.

[0012] As a further preference, the power battery includes a power battery pack, a power battery module or a power battery cell.

[0013] As a further preference, in step (1), the various application scenarios preferably include energy storage power stations, low-speed electric vehicles, power banks, etc., and the standard usage parameters refer to the usage restriction data of the power battery in various application scenarios, preferably including external dimensions, operating voltage, operating current, power, capacity, cycle life, etc.

[0014] As a further preference, in step (2), the standard for the appearance inspection is preferably as follows: for the power battery pack and power battery module, the appearance shall not have any deformation and damage visible to the naked eye such as bulging, corrosion, breakage, etc., and the inside and outside shall be intact; for the power battery cell, the appearance shall not have any deformation and damage visible to the naked eye such as leakage, corrosion, breakage, etc., and the outside shall be intact.

[0015] As a further preference, in step (2), the standard for capacity detection is preferably as follows: a charge and discharge test is performed to obtain the capacity range of the power battery; when the capacity of the power battery pack and the power battery module is above 20%, it is determined that the power battery pack and the power battery module are detachable; when the capacity of the power battery cell is above 20%, it is determined that the power battery cell is reassembled; when the capacity of the power battery is less than 20%, it is determined that the power battery is scrapped.

[0016] As further preferred, in step (3), the neural network multi-label scene classification model is trained in the following manner:

[0017] Obtain the basic parameters and performance parameters of the power battery and the standard parameters of the power battery in applicable scenarios; use the basic parameters and performance parameters of the power battery as input feature sets and applicable scenarios as input labels to train the neural network multi-label scenario classification model.

[0018] As a further preference, in step (3), a convolutional neural network is preferably used for training, wherein the convolutional layer as the core of the convolutional neural network is designed as follows:

[0019]

[0020] Among them, * represents convolution calculation, K is the number of input layer channels of the convolution layer, L is the number of output layer channels, I×J is the convolution kernel size, X k and Y I are the two-dimensional feature matrices of the kth input channel and the lth output channel, respectively, H kJ is the two-dimensional convolution kernel of the kth row and lth column.

[0021] As further preferred, in step (4), the decision on the tiered utilization of the power battery is preferably made in the following manner:

[0022] When the obtained power battery performance parameters are within the preset threshold range, it is determined that the power battery can be utilized in stages; when the obtained power battery performance parameters are less than the preset lower threshold, it is determined that the power battery pack or module enters the disassembly stage; and when the various performance parameters of the disassembled power battery cell are less than the preset lower threshold, it is determined that the power battery cell is scrapped.

[0023] As a further preference, the basic parameters of the power battery preferably include external dimensions, materials, operating temperature, operating voltage, operating current, etc.; the performance parameters of the power battery preferably include battery characteristic test parameters including voltage, internal resistance, impedance, specific heat capacity, self-discharge rate, state of charge, cell temperature, cycle life, etc.

[0024] According to another aspect of the present invention, a corresponding power battery cascade utilization decision system is also provided, characterized in that the system comprises:

[0025] A first data acquisition module, the first data acquisition module is used to obtain standard usage parameters of cascade utilization of power batteries in various actual application scenarios, and form a first data set;

[0026] a second data acquisition module, the second data acquisition module is used to perform appearance inspection and capacity detection on the power battery to determine whether the power battery is detachable; if the power battery is detachable, further detect and obtain basic parameters and performance parameters of the power battery to form a second data set;

[0027] A prediction model building module, the prediction model building module is used to build a neural network multi-label scene classification model, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels, in this way, the original data set composed of the first and second data sets is used to train the classification model;

[0028] The decision analysis module is used to input the basic parameters and performance parameters of the power battery to be tested as feature sets into the trained classification model, output the classification and sorting results of the power battery's suitable application scenarios, and then obtain the decision results of the power battery's tiered utilization.

[0029] As a further preference, the above system preferably also includes a data preprocessing module, which is used to perform preprocessing such as decentralization and homogenization on the original data set to form a standard data set.

[0030] In general, the above technical solution conceived by the present invention has the following beneficial effects compared with the prior art:

[0031] (1) The present invention redesigns the entire decision-making process and makes targeted improvements in multiple aspects such as key operating steps and algorithm mechanisms. Compared with the prior art of directly disassembling the cells for cascade utilization, the present method implements decision-making while disassembling, thereby achieving the maximum cascade utilization of power batteries (whole pack level, module level, cell level), reducing disassembly costs, and improving power battery utilization efficiency;

[0032] (2) In the face of a large amount of basic parameter and performance parameter data of retired power batteries and standard data for various application scenarios, the present invention establishes a tiered utilization classification model by selectively selecting machine learning algorithms such as neural networks. This not only can quickly and comprehensively process big data, make tiered utilization decisions for power batteries, and improve decision-making efficiency, but also, when making scenario matching decisions, by making the basic parameters and performance parameters of the power batteries meet the standard parameters for various application scenarios as much as possible, thereby effectively solving the problem of large-scale series and parallel use of power batteries;

[0033] (3) The present invention further optimizes the design of the training process of the convolutional neural network, and accordingly can adjust the specific parameters of the model according to the increasing historical data, thereby enhancing the robustness and generalization ability of the model, and improving the accuracy and effectiveness of decision analysis. Therefore, the model is particularly suitable for industrial-grade power battery disassembly and recycling and other applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is an overall flow chart for exemplarily illustrating the power battery tiered utilization decision method based on scenario matching of the present invention;

[0035] Figure 2 It is a schematic diagram of a matching decision scenario showing more specifically the basic parameters and performance parameters of a power battery and standard parameters used in various application scenarios according to a preferred embodiment of the present invention;

[0036] Figure 3 It is a schematic diagram of the components of a power battery cascade utilization decision system according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] Figure 1 is an overall flow chart for exemplarily illustrating the power battery tiered utilization decision method based on scenario matching of the present invention. Figure 2 It is a more specific diagram showing the matching decision scenario between the basic parameters and performance parameters of the power battery and the standard parameters used in various application scenarios. Figure 1 and Figure 2 To explain the present invention in more detail.

[0039] First, there is the step of acquiring the first data set.

[0040] In this step, standard usage parameters of cascade utilization of power batteries in various practical application scenarios are obtained to form a first data set.

[0041] More specifically, the various application scenarios preferably include energy storage power stations, low-speed electric vehicles, power banks, etc. The standard parameters for use refer to the usage restriction data of power batteries in various application scenarios, preferably including dimensions, operating voltage, operating current, power, capacity, cycle life, etc. The above data may also be other parameters known to those skilled in the art, which are not limited here.

[0042] Next, there is a step of acquiring the second data set.

[0043] In this step, the power battery is visually inspected and its capacity is tested to determine whether the power battery is detachable; if the power battery is detachable, further testing is performed to obtain basic parameters and performance parameters of the power battery to form a second data set.

[0044] More specifically, taking the main technical parameters of a power battery pack as an example, the power battery pack comes from a certain model of electric pickup truck, uses a ternary material soft-pack battery cell, has no heating system and cooling system, and has been in service on the electric pickup truck for 3 years before disassembly. Perform an appearance inspection on the power battery pack, including: check that the appearance of the power battery pack has no obvious deformation and damage such as bulging, corrosion, and breakage that is visible to the naked eye, the shell is intact, the internal modules are complete, and the connection arrangement is normal. Power battery packs that pass the appearance inspection proceed to the next step, and those that fail will be scrapped.

[0045] When the power battery pack passes the appearance inspection, the power battery pack is subjected to charge and discharge tests to obtain the capacity status of the power battery pack. For example, when the capacity of the power battery pack is not less than 20%, the power battery pack passes the capacity test and can proceed to the next step; when the capacity of the power battery pack is less than 20%, the power battery pack is scrapped.

[0046] When it is determined that the power battery pack is detachable, the basic parameters and performance parameters of the power battery pack are continuously detected and acquired. Exemplarily, the performance parameters of the power battery pack are battery characteristic test parameters including voltage, internal resistance, impedance, specific heat capacity, self-discharge rate, state of charge, cell temperature, cycle life and other related parameters known to those skilled in the art, which are not limited here. The basic parameters of the power battery pack include external dimensions, materials, operating temperature, operating voltage, operating current, etc., which are not limited here.

[0047] Next, it is the step of building a neural network multi-label scene classification model and training it.

[0048] In this step, a neural network multi-label scene classification model is first constructed, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels, and the classification model is trained using the original data set composed of the first and second data sets. Among them, the basic principles and construction process of the neural network multi-label scene classification model are well known to those skilled in the art, and will not be described in detail here.

[0049] In addition, according to a preferred embodiment of the present invention, in order to further improve the training accuracy and final effect of the classification model, in the present invention, combined with the actual application needs, the convolutional neural network is preferably used for training, wherein the convolutional layer as the core of the convolutional neural network is designed as follows:

[0050]

[0051] Among them, * represents convolution calculation, K is the number of input layer channels of the convolution layer, L is the number of output layer channels, I×J is the convolution kernel size, X k and Y I are the two-dimensional feature matrices of the kth input channel and the lth output channel, respectively, H kJ is the two-dimensional convolution kernel of the kth row and lth column.

[0052] In addition, according to another preferred embodiment of the present invention, preprocessing such as decentralization and homogenization can be performed on the original data set to form a standard data set.

[0053] It should be pointed out that the above

[0054] Finally, it is the decision-making step for the tiered utilization of power batteries.

[0055] In this step, for the power battery to be tested, its basic parameters and performance parameters are input into the trained classification model as feature sets, and the classification and ranking results of the power battery's suitable application scenarios are output, thereby obtaining the decision results of the power battery's tiered utilization.

[0056] More specifically, the thresholds of various performance parameters of the power battery pack can be preset first. When the various performance parameters of the power battery pack are within the preset threshold range, it is determined that the power battery pack can be utilized in stages. Among them, the threshold range of various performance parameters of the power battery pack is a range that limits the power battery pack without disassembly. Exemplarily, the remaining capacity of the power battery pack is 78% of the initial capacity, the internal resistance is 5mΩ, the weight is 380kg, the open circuit voltage is 343v, the number of charge and discharge cycles is 400 times, and the self-discharge ratio is 0.28. This parameter is input into the neural network multi-label scenario classification model. When the model outputs certain power energy storage systems, the standard parameters for the use of the power battery pack in these energy storage system application scenarios constitute the performance parameter threshold range of the power battery pack, that is, the various performance parameters of the power battery pack are within the preset threshold range, and the power battery pack can be utilized in stages.

[0057] At the same time, the basic parameters of the power battery pack can also be combined according to the following constraints: the materials of the power batteries reorganized in the same scenario are the same; the performance parameters of the reorganized power batteries are highly consistent and meet the scenario requirements; the basic parameters of the reorganized power batteries such as external dimensions, rated voltage, rated current, etc. meet the application scenario requirements, and the power storage system that best suits the power battery pack can be found.

[0058] Among them, when the performance parameter of the power battery pack is less than the preset threshold lower limit, that is, the model output has no application scenario suitable for the power battery pack, the power battery pack enters the disassembly stage. After being disassembled into four modules, the above steps are repeated for each module until all decision-making processes are completed.

[0059] Figure 3 It is a schematic diagram of the components of a power battery cascade utilization decision system according to a preferred embodiment of the present invention.

[0060] like Figure 3 As shown in , the system mainly includes a first data acquisition module, a second data acquisition module, a prediction model building module and a decision analysis module, and a data preprocessing module can also be set according to specific needs. The following will explain them one by one.

[0061] A first data acquisition module, the first data acquisition module is used to obtain standard usage parameters of cascade utilization of power batteries in various actual application scenarios, and form a first data set;

[0062] a second data acquisition module, the second data acquisition module is used to perform appearance inspection and capacity detection on the power battery to determine whether the power battery is detachable; if the power battery is detachable, further detect and obtain basic parameters and performance parameters of the power battery to form a second data set;

[0063] A prediction model building module, the prediction model building module is used to build a neural network multi-label scene classification model, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels, in this way, the original data set composed of the first and second data sets is used to train the classification model;

[0064] The decision analysis module is used to input the basic parameters and performance parameters of the power battery to be tested as feature sets into the trained classification model, output the classification and sorting results of the power battery's suitable application scenarios, and then obtain the decision results of the power battery's tiered utilization.

[0065] The data preprocessing module is used to perform preprocessing such as decentralization and homogenization on the original data set and form a standard data set.

[0066] In summary, the present invention has redesigned the entire decision-making process, and targetedly selected machine learning algorithms such as neural networks to establish a tiered utilization classification model, which can not only process big data more quickly and comprehensively, improve decision-making efficiency and accuracy, but also effectively solve the disassembly depth decision-making of various retired power batteries and the matching decision-making problems with various practical application scenarios.

[0067] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A power battery cascade utilization decision method based on scenario matching, characterized in that: The method includes: Step (1): obtaining standard usage parameters for cascade utilization of power batteries in various practical application scenarios and forming a first data set; wherein the various practical application scenarios include energy storage power stations, low-speed electric vehicles, and power banks, and the standard usage parameters refer to the usage restriction data of power batteries in various practical application scenarios, including external dimensions, operating voltage, operating current, power, capacity, and cycle life; Step (2): performing appearance inspection and capacity test on the power battery to determine whether the power battery is detachable; if the power battery is detachable, further testing to obtain basic parameters and performance parameters of the power battery to form a second data set; Step (3): Construct a neural network multi-label scenario classification model, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels. In this way, the original data set composed of the first and second data sets is used and the classification model is trained using a convolutional neural network, wherein the convolutional layer as the core of the convolutional neural network is designed as follows: Among them, * represents convolution calculation, K is the number of input layer channels of the convolution layer, I×J is the convolution kernel size, X k and Y I are the two-dimensional feature matrices of the kth input channel and the Ith output channel, respectively, and H kl is the two-dimensional convolution kernel of the kth row and lth column; Step (4): For the power battery to be tested, its basic parameters and performance parameters are input into the trained classification model as feature sets, and the classification and ranking results of the power battery's suitable application scenarios are output, thereby obtaining the decision results of the power battery's tiered utilization.

2. The method according to claim 1, characterized in that The power battery includes a power battery pack, a power battery module or a power battery cell.

3. The method according to claim 2, characterized in that In step (2), the standard of the appearance inspection is as follows: for the power battery pack and power battery module, the appearance shall not have any deformation and damage such as bulging, corrosion, breakage, etc. that are visible to the naked eye, and the inside and outside shall be intact; for the power battery cell, the appearance shall not have any deformation and damage such as leakage, corrosion, breakage, etc. that are visible to the naked eye, and the outside shall be intact.

4. The method according to claim 3, characterized in that In step (2), the standard of the capacity detection is as follows: a charge and discharge test is performed to obtain the capacity range of the power battery; when the capacity of the power battery pack and the power battery module is above 20%, it is determined that the power battery pack and the power battery module are detachable; when the capacity of the power battery cell is above 20%, it is determined that the power battery cell is reconfigurable; when the capacity of the power battery is less than 20%, it is determined that the power battery is scrapped.

5. The method according to any one of claims 1 to 4, characterized in that: In step (4), the decision on the tiered utilization of the power battery is made in the following manner: When the obtained power battery performance parameters are within the preset threshold range, it is determined that the power battery can be utilized in stages; when the obtained power battery performance parameters are less than the preset lower threshold, it is determined that the power battery pack or module enters the disassembly stage; and when the various performance parameters of the disassembled power battery cell are less than the preset lower threshold, it is determined that the power battery cell is scrapped.

6. The method according to claim 5, characterized in that The basic parameters of the power battery include external dimensions, materials, operating temperature, operating voltage, and operating current; the performance parameters of the power battery include battery characteristic test parameters including voltage, internal resistance, impedance, specific heat capacity, self-discharge rate, state of charge, cell temperature, and cycle life.

7. A power battery cascade utilization decision system based on scenario matching, characterized in that: The system includes: A first data acquisition module, which is used to obtain standard usage parameters of cascade utilization of power batteries in various actual application scenarios and form a first data set; wherein the various actual application scenarios include energy storage power stations, low-speed electric vehicles, and power banks, and the standard usage parameters refer to the usage restriction data of power batteries in various actual application scenarios, including external dimensions, operating voltage, operating current, power, capacity, and cycle life; a second data acquisition module, the second data acquisition module is used to perform appearance inspection and capacity detection on the power battery to determine whether the power battery is detachable; if the power battery is detachable, further detect and obtain basic parameters and performance parameters of the power battery to form a second data set; A prediction model building module is used to build a neural network multi-label scenario classification model, wherein the basic parameters and performance parameters of the power battery are used as input feature sets, and the applicable scenarios are used as input labels. In this way, the original data set composed of the first and second data sets is used and the classification model is trained using a convolutional neural network, wherein the convolutional layer as the core of the convolutional neural network is designed as follows: Among them, * represents convolution calculation, K is the number of input layer channels of the convolution layer, I×J is the convolution kernel size, X k and Y I are the two-dimensional feature matrices of the kth input channel and the Ith output channel, respectively, kl is the two-dimensional convolution kernel of the kth row and lth column; The decision analysis module is used to input the basic parameters and performance parameters of the power battery to be tested as feature sets into the trained classification model, output the classification and sorting results of the power battery's suitable application scenarios, and then obtain the decision results of the power battery's tiered utilization.

8. The system according to claim 7, characterized in that The above system also includes a data preprocessing module, which is used to perform decentralized and uniform preprocessing on the original data set and form a standard data set.

Images