Thermal battery pole piece coating assembly line based on modular design

Abstract

The invention provides a thermal battery pole piece coating assembly line based on modular design, and relates to the technical field of thermal electrode piece processing control, and multiple optimization algorithms are applied on the thermal electrode piece coating assembly line, so that the assembly line can be used for coating thermal electrode pieces according to different production requirements and conditions. Strategies in the aspects of module combination, production scheduling, energy management, material distribution and the like are automatically adjusted, intelligent production decision making is achieved, manual intervention is reduced, the automation level and accuracy of production are improved, different modules can be flexibly combined on an assembly line through modular design, the assembly line is suitable for coating production of thermal battery pole pieces of different specifications and types, and the production efficiency is improved. Meanwhile, the optimization algorithm can be adjusted according to different production tasks, so that the assembly line has higher adaptability and universality, can quickly respond to the change of market requirements, can accurately control the coating process and optimize the material delivery path, reduces the waste of raw materials and the loss of resources, improves the utilization efficiency of the resources, and reduces the production cost. And sustainable utilization of resources is realized.

Description

Technical Field

[0001] The invention relates to the technical field of hot electrode sheet processing control, in particular to a hot battery electrode sheet coating production line based on modular design. Background Art

[0002] According to a method for preparing a coated thermal battery composite electrode sheet disclosed in China with publication number "CN110112431B", the composite electrode sheet consists of a heating layer, a positive electrode layer, and a diaphragm layer. The preparation thereof first comprises laying a layer of fiber mesh on a conductive substrate, then coating the positive electrode layer, and placing an asbestos mesh on the surface of the positive electrode layer after drying, and then coating the diaphragm layer on the positive electrode layer on which the asbestos mesh is placed. After the diaphragm layer is dried, the heating layer is coated on the other side of the conductive substrate to obtain a composite sheet, and then high-temperature resistant and waterproof conductive glue is sprayed on both sides of the composite sheet, and the composite positive electrode layer is obtained after drying. During the preparation process of the composite electrode sheet of the present invention, the thickness of the composite electrode sheet can be adjusted according to the height of the scraper, so that the thickness of the composite electrode sheet can be accurately controlled, and can be used together with a thermal battery negative electrode LiB sheet with a controllable thickness to form an ultra-thin thermal battery monomer. The prepared composite electrode sheet has the functions of preventing moisture absorption and preventing the diaphragm from melting and overflowing, and has a high preparation rate and a simple process, which meets the requirements of engineering use.

[0003] The above patent documents and prior art have the following technical problems when used:

[0004] Problem 1: In the above documents and in the prior art, when coating the surface of the hot electrode sheet, the processing path of the entire assembly line is fixed, the processing procedures and steps are the same, and the hot electrode sheets with different coating requirements cannot be quickly adapted and adjusted and the production scheduling optimized, resulting in reduced overall processing efficiency and poor assembly line adaptability;

[0005] Question 2: In the above documents, when coating the surface of the hot electrode sheet, the coating flow and stability are generally set directly by the machine, and the coating thickness and temperature are adjusted and adapted intelligently. Moreover, the defects on the surface of the hot electrode sheet cannot be quickly identified after coating, resulting in uneven quality of the hot electrode sheet after coating and low pass rate;

[0006] Question three: In the above-mentioned documents and existing technologies, the entire production line for coating hot electrode sheets has a low level of intelligence and is unable to optimize energy consumption and supply on the production line, resulting in problems of excess or insufficient energy, affecting the processing process and causing waste of resources. Summary of the invention

[0007] Technical issues solved

[0008] In view of the shortcomings of the prior art, the present invention provides a thermal battery pole piece coating production line based on modular design, which solves the following problems:

[0009] 1. The production process of the assembly line is fixed, and it is impossible to quickly adapt and optimize the production schedule for multi-scenario processing, resulting in low processing efficiency and insufficient adaptability;

[0010] 2. The coating thickness and temperature cannot be adjusted during coating, and intelligent identification and monitoring cannot be performed after coating, resulting in uneven quality of the hot electrode after coating;

[0011] 3. Intelligent optimization and management of energy leads to the problem of energy waste.

[0012] Technical Solution

[0013] To achieve the above objectives, the present invention is implemented by the following technical solutions: a thermal battery pole piece coating production line based on modular design, the control process of the production line is as follows:

[0014] Sp1: Modular production scheduling, by setting up a genetic algorithm to simulate the biological evolution process, the optimal combination distribution and production scheduling plan for each module of the assembly line is carried out, and the module combination and production task sequence are encoded as chromosomes. Each chromosome represents a possible solution. The fitness function is used to evaluate the quality of the chromosome. According to the fitness function value, the chromosome with high fitness is selected to enter the next generation population;

[0015] Sp2: Coating energy management and optimization, optimize energy consumption and energy supply, use particle swarm optimization algorithm to optimize energy consumption, and control and predict energy supply optimization by establishing a dynamic model of assembly line energy consumption;

[0016] Sp2.1: The particle swarm optimization algorithm simulates the foraging behavior of bird flocks to find the optimal energy consumption solution and encodes the operating parameters of the assembly line equipment into the position vector of the particles. And randomly initialize the position and velocity of the particles According to the energy consumption function Calculate the energy consumption value of the equipment operation parameter combination represented by each particle, and each particle is based on its own historical optimal position. and the historical optimal position of the entire particle swarm Update its own speed and position according to the speed update formula and position update formula, and repeat the above steps until the termination condition of the particle swarm optimization algorithm is met. If the maximum number of iterations is reached or the energy consumption reaches the minimum threshold, the optimal equipment operating parameters obtained by optimization are applied to the assembly line;

[0017] Sp2.2: Establish a dynamic model of the energy consumption of the assembly line and determine the system matrix and and the parameters of the energy supply cost function According to the current system status and historical data, the energy demand in the future is predicted. In each control cycle, according to the energy supply cost, energy demand forecast and equipment operation constraints, the optimization problem is solved to determine the optimal energy supply strategy, and the optimized energy supply strategy is applied to the actual energy supply system;

[0018] Sp3: Material distribution path optimization, based on the optimal combination distribution and production scheduling scheme of each module given in Sp1, optimizes the material distribution path by simulating the behavior of ants in the process of searching for food, optimizes the material distribution path through the ant colony algorithm, and initializes the pheromone concentration τ at the nodes and paths of the assembly line ij (0), set the initial position and initial state of the ant, and each ant chooses a path according to the probability formula Starting from the starting node, the next node is selected and the delivery path is gradually constructed. During the movement of the ants, the paths and corresponding costs, including time, distance, etc., are recorded. When all ants complete a path search, the pheromone update formula τ is used. ij (t+n)=ρ τj (t)+Δτ ij , update the pheromone concentration on the path, calculate the cost of the path taken by each ant, calculate the pheromone increment based on the cost, repeat the above ant path search and pheromone update steps until the termination condition of the ant colony algorithm is met. If the maximum number of iterations is reached or a satisfactory distribution path is found, the optimal material distribution path is output and applied to the actual material distribution process;

[0019] Sp4: Coating flow control, establish the state space model of the coating liquid flow control system, determine the state equation and measurement equation, predict the system state and prediction error covariance at the next moment based on the state estimate value and system state equation at the previous moment, combine the measurement value of the flow sensor, calculate the updated state estimate value and error covariance through the Kalman gain, adjust the parameters of the flow controller based on the flow estimate value obtained by the Kalman filter, and accurately control the coating liquid flow;

[0020] Sp5: Defect recognition, build a deep convolutional neural network CNN for surface defect recognition of thermal battery pole pieces after Sp4 coating, use image acquisition equipment to obtain image data of the pole piece, preprocess the image, input the preprocessed image data into the deep convolutional neural network for training, optimize the network parameters through the back propagation algorithm, so that the network can learn the characteristics of different defect types. During the training process, the performance of the model is evaluated according to the cross entropy loss function, and the learning rate and other parameters are adjusted according to the evaluation results. During the production process of the assembly line, the pole piece image is collected in real time and input into the trained deep learning model. The model analyzes and judges the pole piece image, identifies whether there are defects, and outputs information such as defect type and location. According to the defect recognition results, the defective pole piece is removed or an alarm is issued.

[0021] Preferably, in the modular production scheduling, the chromosome is encoded with integers, and is provided with n modules and m production tasks. The chromosome is an integer sequence with a length of n+m, the first n digits represent the selection order of the modules, and the last m digits represent the execution order of the production tasks. The fitness function formula is as follows:

[0022]

[0023] in,

[0024] T p Represents the production cycle, which is determined by calculating the sum of the processing time and waiting time of each production task in the assembly line;

[0025] U e Indicates equipment utilization, calculated by the ratio of actual equipment operation time to total operation time;

[0026] Q r It indicates the qualified rate of product quality, which is obtained by counting the ratio of the number of qualified products to the total number of products;

[0027] w1, w2, and w3 are weight coefficients, which are determined according to production requirements.

[0028] Preferably, in the coating energy management and optimization, each particle represents a set of operating parameters of the assembly line equipment, and the particle position vector Represents the operating parameters of the device, including motor speed, heating temperature, and particle velocity vector The table determines the moving direction and speed of the particle in the search space. The objective function is the energy consumption function, which is used to measure the energy consumption of the equipment operation parameter combination represented by the particle. The energy consumption function can be expressed as:

[0029]

[0030] Among them, E j (x j ) is the jth device operating with parameter x j Energy consumption under

[0031] Each particle has its own historical optimal position. and the historical optimal position of the entire particle swarm To update its own speed and position, the speed update formula is:

[0032]

[0033] The position update formula is:

[0034]

[0035] Where w is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the range [0,1], and d = 1, 2, …, n, k is the number of iterations.

[0036] Preferably, in the coating energy management and optimization, when establishing a dynamic model of the energy consumption of the assembly line, a linear time-invariant model is used to describe the system state, and the formula is as follows:

[0037]

[0038] in, Indicates the system status, including the power level of energy storage devices, the operating status of the equipment, etc. represents the control input, including the amount of electricity purchased, the charging and discharging strategy of the energy storage device, and is the system matrix;

[0039] In each control cycle, the model predictive control algorithm predicts the energy demand in the future. The prediction horizonsN determines the length of the prediction. By predicting the system state and energy demand at the next N moments, the energy supply strategy is planned in advance.

[0040] Solve the optimization problem based on energy supply cost, energy demand forecast and equipment operation constraints. The optimization goal is to minimize energy cost:

[0041]

[0042] While satisfying system constraints and:

[0043] and

[0044] in, is the control variable vector, is the system state variable vector, and are the upper and lower limits of the control variable.

[0045] Preferably, in the material distribution path optimization, when the ant moves between nodes, it selects the next node according to the pheromone concentration and heuristic information on the path. Suppose the probability of the ant moving from node i to node j is

[0046]

[0047] Among them, k represents the kth ant, τ ij (t) is the pheromone concentration at the time edge, α is the pheromone importance factor, η ij (t) is the heuristic information, which is usually related to factors such as the distance or cost between nodes, β is the importance factor of the heuristic information, and allowed k is the set of next nodes that the kth ant can choose;

[0048] The pheromone update formula is:

[0049] τ ij (t+n)=ρτ ij (t)+Δτ ij

[0050] Among them, ρ is the pheromone volatility coefficient, which controls the attenuation degree of pheromone, Δτ ij is the pheromone increment on edge (i, j) in this iteration, which is related to the objective function value obtained by the ant passing through this path;

[0051] After each ant completes a path search, its path cost is calculated, and the pheromone increment is calculated based on the cost. The formula is as follows:

[0052]

[0053] Where m is the number of ants, is the pheromone increase left by the kth ant on the edge (i, j), which can be defined as

[0054]

[0055] Where Q is a constant, L k is the length of the path taken by the kth ant.

[0056] Preferably, the system state equation in the state space model of the coating liquid flow control system established in the coating flow control is:

[0057]

[0058] in, Q k Indicates the flow rate of the coating liquid, represents the flow rate change rate, represents the state transfer matrix, represents process noise;

[0059] The measurement equation is:

[0060]

[0061] in, Indicates the measured value of the flow sensor, represents the measurement noise;

[0062] The formula for predicting the system state at the next moment based on the system state equation is:

[0063]

[0064] The forecast error covariance formula is:

[0065]

[0066] in, is the estimated value of the state at the previous moment, is the error covariance at the previous moment, is the process noise covariance;

[0067] Update the state estimate based on the measurements:

[0068]

[0069] Error covariance:

[0070]

[0071] Among them, Kalman gain, given by Calculated, is the measurement noise covariance, is the identity matrix.

[0072] Preferably, in the coating flow control, the temperature value during the coating process is set to T set , collect the actual temperature T(t) during the coating process in real time, and calculate the temperature error e(t) = T set -T(t), according to the error e(t), the output control quantity is calculated by the PID controller, the formula is as follows:

[0073]

[0074] Among them, K i Indicates the integral coefficient, K d represents the differential coefficient, by adjusting K p , K i and K d value, so that the temperature can quickly and steadily reach the set value.

[0075] Preferably, in the defect recognition, the deep convolutional neural network includes multiple convolutional layers, pooling layers and fully connected layers. The convolutional layer is used to extract the features of the electrode image. The operation of the convolutional layer is expressed as:

[0076]

[0077] in, is the jth feature map of the lth layer, is the convolution kernel, is the bias term, f(·) is the activation function, M j is the set of input feature maps;

[0078] The pooling layer is used to reduce the resolution of the feature map and the amount of data while maintaining the invariance of the features. The fully connected layer maps the extracted features to different defect categories.

[0079] Preferably, in the defect recognition training process, a large number of defective and non-defective electrode images are prepared as training data, and the images are annotated to mark different defect types. The training data is input into the network, and the network parameters are optimized by the back propagation algorithm. The loss function adopts the cross entropy loss function:

[0080]

[0081] Where N is the number of samples, y i is the true label, is the probability of this type of defect predicted by the network;

[0082] During the training process, by adjusting parameters such as learning rate and optimizer, the network gradually converges and can accurately identify various defects on the electrode, including scratches, holes, and uneven coating.

[0083] Beneficial Effects

[0084] The present invention provides a thermal battery pole piece coating production line based on modular design. It has the following beneficial effects:

[0085] 1. The present invention sets a plurality of processing module structures for coating the surface of the hot electrode sheet on the hot electrode sheet coating production line, and optimizes the module combination and production scheduling through genetic algorithm, which can shorten the production cycle, accurately arrange the working order of each module and the execution order of production tasks, avoid unnecessary waiting time and equipment idleness, make the operation of the whole production line smoother, thereby improving the product output per unit time, and optimize the material distribution path based on the combination of module structure and production scheduling, corresponding to the module combination scheduling, which can greatly shorten the transportation time of materials on the production line, and materials can reach each processing link more quickly and accurately, reducing production interruptions caused by material delays, and further improving production efficiency.

[0086] 2. The present invention uses the Kalman filter algorithm to accurately control the flow rate of the coating liquid on the hot electrode sheet coating line to ensure the uniformity and consistency of the coating thickness during the coating process. The stable control of the flow rate can avoid the situation where the coating is too thick or too thin, and improve the performance and quality of the electrode. The PID control algorithm enables the coating temperature to reach the set value quickly and stably. The appropriate and stable temperature helps the coating material to better adhere to the electrode, improve the adhesion and uniformity of the coating, and reduce coating defects caused by temperature fluctuations, thereby improving product quality. The electrode defect recognition algorithm based on deep learning can accurately identify various defects on the electrode, such as scratches, holes, uneven coating, etc., and promptly discover and remove defective electrodes during the production process, avoiding defective products from flowing into subsequent processes, and ensuring the quality consistency and reliability of the final product.

[0087] 3. The present invention optimizes energy consumption and energy supply on the hot electrode sheet coating assembly line, and optimizes the operating parameters of the assembly line equipment using a particle swarm optimization algorithm, so that the equipment can operate with the lowest energy consumption while meeting production needs. By adjusting parameters such as motor speed and heating temperature, unnecessary energy waste is reduced, and the energy consumption of the entire assembly line is reduced. The model predictive control algorithm reasonably plans the energy supply strategy based on energy demand forecasts and equipment operation constraints, such as increasing energy storage when energy prices are low and using stored energy at peak times, thereby reducing energy costs. At the same time, by optimizing the amount of electricity purchased and the charging and discharging strategies of energy storage equipment, energy utilization efficiency is improved and energy consumption is further reduced.

[0088] 4. The present invention uses a variety of optimization algorithms on the thermal electrode sheet coating production line. The production line can automatically adjust the strategies of module combination, production scheduling, energy management and material distribution according to different production needs and conditions, realize intelligent production decision-making, reduce manual intervention, and improve the level of automation and accuracy of production. The modular design enables the production line to flexibly combine different modules to adapt to the coating production of thermal battery electrodes of different specifications and types. At the same time, the optimization algorithm can be adjusted according to different production tasks, so that the production line has stronger adaptability and versatility, and can quickly respond to changes in market demand. Accurate coating process control and material distribution path optimization reduce the waste of raw materials and resource loss, improve resource utilization efficiency, and achieve sustainable resource utilization, which is in line with the trend of social sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 It is a schematic diagram of the assembly line steps of the present invention;

[0090] Figure 2 A visualization diagram of the legacy algorithm of the present invention;

[0091] Figure 3 A visualization chart for optimizing the material distribution path of the present invention;

[0092] Figure 4 A visualization diagram of the Kalman waves for coating flow control of the present invention;

[0093] Figure 5 A visualization diagram of energy management and optimization for the present invention coating. DETAILED DESCRIPTION

[0094] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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. Specific embodiment one:

[0096] like Figure 1-5 As shown in the figure, the control process of the thermal battery pole piece coating production line based on modular design is as follows:

[0097] Sp1: Modular production scheduling, by setting up a genetic algorithm to simulate the biological evolution process, the optimal combination distribution and production scheduling plan for each module of the assembly line is carried out, and the module combination and production task sequence are encoded as chromosomes. Each chromosome represents a possible solution. The fitness function is used to evaluate the quality of the chromosome. According to the fitness function value, the chromosome with high fitness is selected to enter the next generation population. In modular production scheduling, chromosomes are encoded with integers. There are n modules and m production tasks. The chromosome is an integer sequence with a length of n+m. The first n digits represent the selection order of the modules, and the last m digits represent the execution order of the production tasks. The fitness function formula is as follows:

[0098]

[0099] in,

[0100] T p Represents the production cycle, which is determined by calculating the sum of the processing time and waiting time of each production task in the assembly line;

[0101] U e Indicates equipment utilization, calculated by the ratio of actual equipment operation time to total operation time;

[0102] Q r It indicates the qualified rate of product quality, which is obtained by counting the ratio of the number of qualified products to the total number of products;

[0103] w1, w2, and w3 are weight coefficients, which are determined according to production requirements;

[0104] Selection operation: The fitness value of each chromosome is calculated according to the fitness function. Chromosomes with high fitness are more likely to be selected into the next generation population, similar to the individuals that adapt to the environment in natural selection are more likely to survive and reproduce. The selected chromosomes are crossed, and some genes are exchanged at the randomly selected intersections to generate new chromosome combinations, which increases the diversity of the population and helps to explore more possible solutions. Some genes of the chromosomes are mutated with a certain probability, that is, the value of the gene is randomly changed. This can prevent the algorithm from falling into the local optimum and increase the possibility of finding the global optimal solution.

[0105] By optimizing module combinations and production scheduling, we can reduce waiting time and equipment idleness in the production cycle, make the assembly line run more smoothly, and thus increase product output per unit time. By comprehensively considering equipment utilization, we can reasonably allocate equipment resources, avoid excessive use of some equipment while other equipment is idle, and improve overall resource utilization efficiency. Incorporating product quality qualification rate into the optimization goal will help consider quality factors in production scheduling, ensure the stability and consistency of the production process, and thus improve product quality.

[0106] Sp2: Coating energy management and optimization, optimize energy consumption and energy supply, use particle swarm optimization algorithm to optimize energy consumption, and control and predict energy supply optimization by establishing a dynamic model of assembly line energy consumption;

[0107] Sp2.1: The particle swarm optimization algorithm simulates the foraging behavior of bird flocks to find the optimal energy consumption solution and encodes the operating parameters of the assembly line equipment into the position vector of the particles. And randomly initialize the position and velocity of the particles According to the energy consumption function Calculate the energy consumption value of the equipment operation parameter combination represented by each particle, and each particle is based on its own historical optimal position. and the historical optimal position of the entire particle swarm Update your own speed and position according to the speed update formula and position update formula, repeat the above steps until the termination condition of the particle swarm optimization algorithm is met. If the maximum number of iterations is reached or the energy consumption reaches the minimum threshold, the optimal equipment operating parameters obtained by optimization are applied to the assembly line. In the coating energy management and optimization, each particle represents a set of operating parameters of the assembly line equipment, and the particle position vector Represents the operating parameters of the equipment, including motor speed and heating temperature. These parameters directly affect energy consumption. By adjusting these parameters, we can find the optimal energy consumption solution. The velocity vector of the particle The table determines the moving direction and speed of the particle in the search space. The objective function is the energy consumption function, which is used to measure the energy consumption of the equipment operation parameter combination represented by the particle. The energy consumption function can be expressed as:

[0108]

[0109] Among them, E j (x j ) is the jth device operating with parameter x j By calculating the value of this function, the energy consumption of the parameter combination represented by each particle can be quantitatively evaluated;

[0110] Each particle has its own historical optimal position. and the historical optimal position of the entire particle swarm To update its own speed and position, the speed update formula is:

[0111]

[0112] The position update formula is:

[0113]

[0114] Where w is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the range [0,1], and d = 1, 2, ..., n, k is the number of iterations;

[0115] By optimizing the equipment operating parameters, the equipment can operate at the lowest energy consumption while meeting production needs, thereby reducing the energy consumption of the entire assembly line and saving energy costs. Reasonable adjustment of equipment operating parameters can help reduce equipment wear and aging, extend equipment service life, and improve equipment stability and reliability.

[0116] Sp2.2: Establish a dynamic model of the energy consumption of the assembly line and determine the system matrix and and the parameters of the energy supply cost function According to the current system status and historical data, the energy demand in the future is predicted. In each control cycle, according to the energy supply cost, energy demand forecast and equipment operation constraints, the optimization problem is solved to determine the optimal energy supply strategy. The optimized energy supply strategy is applied to the actual energy supply system. When the dynamic model of the assembly line energy consumption is established in the energy management and optimization, the linear time-invariant model is used to describe the system state. The formula is as follows:

[0117]

[0118] in, Indicates the system status, including the power level of energy storage devices, the operating status of the equipment, etc. represents the control input, including the amount of electricity purchased, the charging and discharging strategy of the energy storage device, and is the system matrix;

[0119] In each control cycle, the model predictive control algorithm predicts the energy demand in the future based on the current system status and historical data. The prediction horizonsN determines the length of the prediction. By predicting the system status and energy demand at the next N moments, the energy supply strategy is planned in advance. In each control cycle, according to the energy supply cost, energy demand forecast and equipment operation constraints, the optimization problem is solved to determine the optimal energy supply strategy, such as the power purchase amount and the charging and discharging strategy of the energy storage device, and it is applied to the actual energy supply system.

[0120] Solve the optimization problem based on energy supply cost, energy demand forecast and equipment operation constraints. The optimization goal is to minimize energy cost:

[0121]

[0122] While satisfying system constraints and:

[0123] and

[0124] in, is the control variable vector, is the system state variable vector, and are the upper and lower limits of the control variables, respectively;

[0125] By rationally planning energy supply strategies and managing electricity procurement and energy storage equipment according to energy price fluctuations and demand changes, we can reduce energy procurement costs, improve energy utilization efficiency, and enhance energy supply stability: Formulating energy supply strategies based on energy demand forecasts and equipment operation constraints can ensure that energy supply matches production demand, avoid energy shortages or surpluses, and improve the stability and reliability of energy supply.

[0126] Sp3: Material distribution path optimization, based on the optimal combination distribution and production scheduling scheme of each module given in Sp1, optimizes the material distribution path by simulating the behavior of ants in the process of searching for food, optimizes the material distribution path through the ant colony algorithm, and initializes the pheromone concentration τ at the nodes and paths of the assembly line ij (0), set the initial position and initial state of the ant, and each ant chooses a path according to the probability formula Starting from the starting node, the next node is selected and the delivery path is gradually constructed. During the movement of the ants, the paths and corresponding costs, including time, distance, etc., are recorded. When all ants complete a path search, the pheromone update formula τ is used. ij (t+n)=ρτ ij (t)+Δτ ij , update the pheromone concentration on the path, calculate the cost of the path taken by each ant, calculate the pheromone increment based on the cost, repeat the above ant path search and pheromone update steps until the termination condition of the ant colony algorithm is met. If the maximum number of iterations is reached or a satisfactory distribution path is found, the optimal material distribution path is output and applied to the actual material distribution process. In the material distribution path optimization, when the ant moves between nodes, it selects the next node based on the pheromone concentration and heuristic information on the path. Suppose the probability of an ant moving from node i to node j is

[0127]

[0128] Among them, k represents the kth ant, τ ij (t) is the pheromone concentration at the time edge, α is the pheromone importance factor, η ij(t) is the heuristic information, which is usually related to factors such as the distance or cost between nodes, β is the importance factor of the heuristic information, and allowed k is the set of next nodes that the kth ant can choose. The higher the pheromone concentration, the more likely the ant will choose this path. Heuristic information is usually related to factors such as the distance or cost between nodes. The shorter the distance or the lower the cost, the greater the value of the heuristic information, and the more likely the ant will choose this path.

[0129] The pheromone update formula is:

[0130] τ ij (t+n)=ρτ ij (t)+Δτ ij

[0131] Among them, ρ is the pheromone volatility coefficient, which controls the attenuation degree of pheromone, Δτ ij is the pheromone increment on edge (i, j) in this iteration, which is related to the objective function value obtained by the ant passing through this path;

[0132] After each ant completes a path search, its path cost is calculated, and the pheromone increment is calculated based on the cost. The formula is as follows:

[0133]

[0134] Where m is the number of ants, is the pheromone increase left by the kth ant on the edge (i, j), which can be defined as

[0135]

[0136] Where Q is a constant, L k is the length of the path taken by the kth ant;

[0137] After each ant completes a path search, it calculates its path cost and calculates the pheromone increment Δτ based on the cost. ij , and then sum up the pheromone increments on the paths passed by all ants to update the pheromone concentration.

[0138] By optimizing the material distribution path, reducing the transportation time and cost of materials on the assembly line, and improving the timeliness and accuracy of material distribution, the efficiency of the entire production process can be improved. The optimized distribution path can reduce the consumption of human, material and other resources, reduce distribution costs, and improve the economic benefits of the enterprise.

[0139] Sp4: Coating flow control, establish the state space model of the coating liquid flow control system, determine the state equation and measurement equation, predict the system state and prediction error covariance at the next moment based on the state estimate value and system state equation at the previous moment, combine the measurement value of the flow sensor, calculate the updated state estimate value and error covariance through the Kalman gain, adjust the parameters of the flow controller based on the flow estimate value obtained by the Kalman filter, and accurately control the coating liquid flow. The system state equation in the state space model of the coating liquid flow control system established in the coating flow control is:

[0140]

[0141] in, Q k Indicates the flow rate of the coating liquid, represents the flow rate change rate, represents the state transfer matrix, represents process noise;

[0142] The measurement equation is:

[0143]

[0144] in, Indicates the measured value of the flow sensor, represents the measurement noise;

[0145] The formula for predicting the system state at the next moment based on the system state equation is:

[0146]

[0147] The forecast error covariance formula is:

[0148]

[0149] in, is the estimated value of the state at the previous moment, is the error covariance at the previous moment, is the process noise covariance;

[0150] Update the state estimate based on the measurements:

[0151]

[0152] Error covariance:

[0153]

[0154] Among them, Kalman gain, given by Calculated, is the measurement noise covariance, is the identity matrix.

[0155] In coating flow control, the coating temperature is controlled by the PID controller. The output control amount is calculated according to the error between the set temperature and the actual temperature. The parameters of the heating equipment are adjusted to stabilize the temperature at the set value. The temperature value during the coating process is set to T set , collect the actual temperature T(t) during the coating process in real time, and calculate the temperature error e(t) = T set -T(t), according to the error e(t), the output control quantity is calculated by the PID controller, the formula is as follows:

[0156]

[0157] Among them, K i Indicates the integral coefficient, K d represents the differential coefficient, by adjusting K p , K i and K d value, so that the temperature can quickly and steadily reach the set value.

[0158] The Kalman filter algorithm is used to accurately control the coating liquid flow rate to ensure the uniformity and consistency of the coating thickness during the coating process. Stable control of the flow rate can avoid the situation where the coating is too thick or too thin, and improve the performance and quality of the electrode. The PID control algorithm enables the coating temperature to reach the set value quickly and stably. The appropriate and stable temperature helps the coating material to better adhere to the electrode, improve the adhesion and uniformity of the coating, and reduce coating defects caused by temperature fluctuations, thereby improving product quality. The coating liquid flow rate is accurately controlled to ensure the uniformity and consistency of the coating thickness during the coating process, improve the performance and quality of the electrode, stabilize the coating temperature, help the coating material to better adhere to the electrode, reduce coating defects caused by temperature fluctuations, and further improve product quality.

[0159] Sp5: Defect recognition, build a deep convolutional neural network CNN for surface defect recognition of thermal battery pole pieces after Sp4 coating, use image acquisition equipment to obtain image data of pole pieces, preprocess the image, including image enhancement, denoising, cropping, normalization and other operations to improve image quality and data consistency, input the preprocessed image data into the deep convolutional neural network for training, optimize the network parameters through the back propagation algorithm so that the network can learn the characteristics of different defect types, during the training process, evaluate the performance of the model according to the cross entropy loss function, and adjust the learning rate and other parameters according to the evaluation results, in the production process of the assembly line, collect pole piece images in real time and input them into the trained deep learning model, the model analyzes and judges the pole piece images, identifies whether there are defects, and outputs information such as defect type and location, and removes defective pole pieces or issues alarm prompts according to the defect recognition results.

[0160] In defect recognition, the deep convolutional neural network includes multiple convolutional layers, pooling layers and fully connected layers. The convolutional layer is used to extract the features of the electrode image. The operation of the convolutional layer is expressed as:

[0161]

[0162] in, is the jth feature map of the lth layer, is the convolution kernel, is the bias term, f(·) is the activation function, M j is the set of input feature maps;

[0163] The pooling layer is used to reduce the resolution of the feature map and the amount of data while maintaining the invariance of the features. The fully connected layer maps the extracted features to different defect categories.

[0164] During the defect recognition training process, a large number of defective and non-defective electrode images are prepared as training data, and the images are annotated to mark different defect types. The training data is input into the network, and the network parameters are optimized through the back propagation algorithm. The loss function adopts the cross entropy loss function:

[0165]

[0166] Where N is the number of samples, y i is the true label, is the probability of this type of defect predicted by the network;

[0167] During the training process, by adjusting parameters such as learning rate and optimizer, the network gradually converges and can accurately identify various defects on the electrode, including scratches, holes, and uneven coating.

[0168] The pole piece defect recognition algorithm based on deep learning can accurately identify various defects on the pole piece, such as scratches, holes, uneven coating, etc., and promptly discover and remove defective pole pieces during the production process, avoiding defective products from flowing into subsequent processes, ensuring the quality consistency and reliability of the final product, accurately identifying various defects on the pole piece, and promptly removing defective pole pieces, ensuring the quality consistency and reliability of the final product, reducing the generation of waste caused by defective products flowing into subsequent processes, and reducing raw material waste and production costs.

[0169] When the entire assembly line is processing and using the hot electrode sheets, the module combination and production scheduling plan of the assembly line are determined through Sp1 modular production scheduling. Then, during the coating process, on the one hand, Sp2 coating energy management and optimization are carried out, including energy consumption optimization and energy supply optimization, to reduce energy costs and ensure stable energy supply; on the other hand, Sp4 coating flow control is carried out to accurately control the coating liquid flow and coating temperature to ensure coating quality. At the same time, Sp3 material distribution path optimization is carried out according to the production scheduling plan to optimize the material distribution path, improve distribution efficiency and reduce costs. Finally, after the coating is completed, Sp5 defect recognition is used to detect whether there are defects in the electrode to ensure product quality. The various steps of the entire assembly line cooperate with each other, and through optimization algorithms and intelligent control, high-efficiency, high-quality, and low-energy consumption hot battery electrodes are achieved. Coating production, by using a variety of optimization algorithms on the thermal electrode coating assembly line, the assembly line can automatically adjust the module combination, production scheduling, energy management and material distribution strategies according to different production needs and conditions, realize intelligent production decision-making, reduce manual intervention, and improve the level of automation and accuracy of production. The modular design enables the assembly line to flexibly combine different modules to adapt to the coating production of thermal battery electrodes of different specifications and types. At the same time, the optimization algorithm can be adjusted according to different production tasks, making the assembly line more adaptable and versatile, and able to quickly respond to changes in market demand. Accurate coating process control and material distribution path optimization reduce the waste of raw materials and resource loss, improve resource utilization efficiency, and achieve sustainable resource utilization, which is in line with the trend of social sustainable development. Specific embodiment 2:

[0171] like Figure 1-5 As shown, according to the contents in the above specific embodiments, the following contents are further disclosed:

[0172] According to the above content and the existing standard hot electrode sheet coating line in terms of production efficiency, product quality, energy consumption and other performance differences, the advantages of the new design line are verified. The specific experimental contents are as follows:

[0173] Modular assembly line: built according to the design scheme in the above specific embodiment 1, including a coating module, a drying module, a conveying module, a detection module, etc., as well as corresponding control systems and algorithms;

[0174] Existing standard hot electrode sheet coating production line: select the common and representative standard hot electrode sheet coating production line on the market as the comparison object;

[0175] Experimental materials: Thermal battery pole piece materials, coating materials, etc. of the same specifications and quality

[0176] Experimental steps:

[0177] Production efficiency test: In the same period of time, start the "thermal battery electrode coating line based on modular design" and the existing standard assembly line respectively, record the number of qualified electrodes produced by each, and calculate the output per unit time, that is, production efficiency, production efficiency = number of qualified electrodes / production time.

[0178] Product quality testing: A certain number of pole pieces are randomly selected from the two production lines for quality inspection. The inspection indicators include coating thickness uniformity, coating adhesion, pole piece surface defects, etc. For coating thickness uniformity, a professional thickness gauge is used to measure at different positions of the pole piece and calculate the thickness standard deviation; coating adhesion is tested by an adhesion tester; pole piece surface defects are detected by combining manual visual inspection with an image recognition system, and the number of defects is counted.

[0179] Energy consumption test: During the operation of the two production lines, energy metering equipment was installed to record the energy consumption of the two production lines under the same production task, including electricity, heat, etc., and calculate the energy consumption per unit product. Energy consumption = total energy consumption / number of qualified electrodes;

[0180] Production scheduling flexibility test: Change the specifications and quantity of production tasks, observe the time and difficulty required to adjust the production schedule of the two production lines, record the time to reconfigure modules and adjust the production sequence, and the impact on production during the adjustment process

[0181] The experimental data are shown in Table 1 below:

[0182]

[0183] Table 1

[0184] It can be clearly seen from the above table that the "thermal battery pole piece coating production line based on modular design" has shown superior performance to the existing standard production line in all aspects:

[0185] In terms of production efficiency, the newly designed assembly line can produce 500 pole pieces per hour, 100 more than the existing standard assembly line. This is mainly due to the module combination and production scheduling optimized by genetic algorithms, which reduces the waiting time and equipment idleness in the production cycle, making the assembly line run more smoothly, thereby increasing the output per unit time.

[0186] In terms of coating thickness uniformity, the standard deviation of the new production line is only 0.05, which is much lower than the 0.12 of the existing standard production line. This means that the coating uniformity of the new production line is better. This is because the Kalman filter algorithm accurately controls the coating liquid flow rate, ensuring the uniformity of the coating thickness during the coating process.

[0187] In terms of coating adhesion, the new production line reaches 15N, which is higher than the 10N of the existing standard production line, indicating that the coating material adheres more firmly to the pole piece. Due to the precise coating process control, including stable coating temperature and precise coating liquid flow rate, the coating material adheres better to the pole piece.

[0188] In terms of the number of surface defects on the pole pieces, the new production line has only 2 defects per 100 pieces, while the existing standard production line has 8. The new production line has higher product quality. The pole piece defect recognition algorithm based on deep learning can accurately identify and remove defective pole pieces, thereby improving product quality.

[0189] In terms of energy consumption, the new production line has a unit product energy consumption of 0.5kW·h / piece, which is lower than the 0.8kW·h / piece of the existing standard production line, reflecting its energy-saving advantages. The particle swarm optimization algorithm optimizes energy consumption and the model predictive control optimizes energy supply, enabling the equipment to use energy more efficiently during operation and reducing energy consumption.

[0190] In terms of production schedule adjustment time, the new assembly line only takes 1 hour, while the existing standard assembly line takes 3 hours. The new assembly line performs well in production scheduling flexibility. The modular design enables the new assembly line to adjust module combinations and production sequences more quickly and flexibly to adapt to different production task requirements and reduce the impact of the adjustment process on production.

[0191] The above charts show intuitively the advantages of the thermal battery pole piece coating production line based on modular design compared with the existing standard production line in terms of production efficiency, product quality, energy consumption and production scheduling flexibility.

[0192] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprising a reference structure" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0193] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The thermal battery pole piece coating line based on modular design is characterized by: The control flow of the pipeline is as follows: Sp1: Modular production scheduling, by setting up a genetic algorithm to simulate the biological evolution process, the optimal combination distribution and production scheduling plan for each module of the assembly line is carried out, and the module combination and production task sequence are encoded as chromosomes. Each chromosome represents a possible solution. The fitness function is used to evaluate the quality of the chromosome. According to the fitness function value, the chromosome with high fitness is selected to enter the next generation population; Sp2: Coating energy management and optimization, optimize energy consumption and energy supply, use particle swarm optimization algorithm to optimize energy consumption, and control and predict energy supply optimization by establishing a dynamic model of assembly line energy consumption; Sp2.1: The particle swarm optimization algorithm simulates the foraging behavior of bird flocks to find the optimal energy consumption solution and encodes the operating parameters of the assembly line equipment into the position vector of the particles. And randomly initialize the position and velocity of the particles According to the energy consumption function Calculate the energy consumption value of the equipment operation parameter combination represented by each particle, and each particle is based on its own historical optimal position. and the historical optimal position of the entire particle swarm Update its own speed and position according to the speed update formula and position update formula, and repeat the above steps until the termination condition of the particle swarm optimization algorithm is met. If the maximum number of iterations is reached or the energy consumption reaches the minimum threshold, the optimal equipment operating parameters obtained by optimization are applied to the assembly line; Sp2.2: Establish a dynamic model of the energy consumption of the assembly line and determine the system matrix and and the parameters of the energy supply cost function According to the current system status and historical data, the energy demand in the future is predicted. In each control cycle, according to the energy supply cost, energy demand forecast and equipment operation constraints, the optimization problem is solved to determine the optimal energy supply strategy, and the optimized energy supply strategy is applied to the actual energy supply system; Sp3: Material distribution path optimization, based on the optimal combination distribution and production scheduling scheme of each module given in Sp1, optimizes the material distribution path by simulating the behavior of ants in the process of searching for food, optimizes the material distribution path through the ant colony algorithm, and initializes the pheromone concentration τ at the nodes and paths of the assembly line ij (0), set the initial position and initial state of the ant, and each ant chooses a path according to the probability formula Starting from the starting node, the next node is selected and the delivery path is gradually constructed. During the movement of the ants, the paths and corresponding costs, including time, distance, etc., are recorded. When all ants complete a path search, the pheromone update formula τ is used. ij (t+n)=ρτ ij (t)+Δτ ij , update the pheromone concentration on the path, calculate the cost of the path taken by each ant, calculate the pheromone increment based on the cost, repeat the above ant path search and pheromone update steps until the termination condition of the ant colony algorithm is met. If the maximum number of iterations is reached or a satisfactory distribution path is found, the optimal material distribution path is output and applied to the actual material distribution process; Sp4: Coating flow control, establish the state space model of the coating liquid flow control system, determine the state equation and measurement equation, predict the system state and prediction error covariance at the next moment based on the state estimate value and system state equation at the previous moment, combine the measurement value of the flow sensor, calculate the updated state estimate value and error covariance through the Kalman gain, adjust the parameters of the flow controller based on the flow estimate value obtained by the Kalman filter, and accurately control the coating liquid flow; Sp5: Defect recognition, build a deep convolutional neural network CNN for surface defect recognition of thermal battery pole pieces after Sp4 coating, use image acquisition equipment to obtain image data of the pole piece, preprocess the image, input the preprocessed image data into the deep convolutional neural network for training, optimize the network parameters through the back propagation algorithm, so that the network can learn the characteristics of different defect types. During the training process, the performance of the model is evaluated according to the cross entropy loss function, and the learning rate and other parameters are adjusted according to the evaluation results. During the production process of the assembly line, the pole piece image is collected in real time and input into the trained deep learning model. The model analyzes and judges the pole piece image, identifies whether there are defects, and outputs information such as defect type and location. According to the defect recognition results, the defective pole piece is removed or an alarm is issued.

2. The thermal battery pole piece coating production line based on modular design according to claim 1 is characterized in that: In the modular production scheduling, the chromosome is encoded with integers, and is provided with n modules and m production tasks. The chromosome is an integer sequence with a length of n+m. The first n digits represent the selection order of the modules, and the last m digits represent the execution order of the production tasks. The fitness function formula is as follows: in, T p Represents the production cycle, which is determined by calculating the sum of the processing time and waiting time of each production task in the assembly line; U e Indicates equipment utilization, calculated by the ratio of actual equipment operation time to total operation time; Q r It indicates the qualified rate of product quality, which is obtained by counting the ratio of the number of qualified products to the total number of products; w1, w2, and w3 are weight coefficients, which are determined according to production requirements.

3. The thermal battery pole piece coating line based on modular design according to claim 1, characterized in that: In the coating energy management and optimization, each particle represents a set of operating parameters of the assembly line equipment, and the particle position vector Represents the operating parameters of the device, including motor speed, heating temperature, and particle velocity vector The table determines the moving direction and speed of the particle in the search space. The objective function is the energy consumption function, which is used to measure the energy consumption of the equipment operation parameter combination represented by the particle. The energy consumption function can be expressed as: Among them, E j (x j ) is the jth device operating with parameter x j Energy consumption under Each particle has its own historical optimal position. and the historical optimal position of the entire particle swarm To update its own speed and position, the speed update formula is: The position update formula is: Where w is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers in the range [0,1], and d = 1, 2, …, n, k is the number of iterations.

4. The thermal battery pole piece coating production line based on modular design according to claim 1 is characterized in that: In the coating energy management and optimization, when establishing a dynamic model of the energy consumption of the assembly line, a linear time-invariant model is used to describe the system state, and the formula is as follows: in, Indicates the system status, including the power level of energy storage devices, the operating status of the equipment, etc. represents the control input, including the amount of electricity purchased, the charging and discharging strategy of the energy storage device, and is the system matrix; In each control cycle, the model predictive control algorithm predicts the energy demand in the future. The prediction horizonsN determines the length of the prediction. By predicting the system state and energy demand at the next N moments, the energy supply strategy is planned in advance. Solve the optimization problem based on energy supply cost, energy demand forecast and equipment operation constraints. The optimization goal is to minimize energy cost: While satisfying system constraints and: in, is the control variable vector, is the system state variable vector, and are the upper and lower limits of the control variable.

5. The thermal battery pole piece coating production line based on modular design according to claim 1, characterized in that: In the material distribution path optimization, when the ant moves between nodes, it selects the next node according to the pheromone concentration and heuristic information on the path. Suppose the probability of the ant moving from node i to node j is Among them, k represents the kth ant, τ ij (t) is the pheromone concentration at the time edge, α is the pheromone importance factor, η ij (t) is the heuristic information, which is usually related to factors such as the distance or cost between nodes, β is the importance factor of the heuristic information, and allowed k is the set of next nodes that the kth ant can choose; The pheromone update formula is: t ij (t+n)=ρτ ij (t)+Δτ ij Among them, ρ is the pheromone volatility coefficient, which controls the attenuation degree of pheromone, Δτ ij is the pheromone increment on edge (i, j) in this iteration, which is related to the objective function value obtained by the ant passing through this path; After each ant completes a path search, its path cost is calculated, and the pheromone increment is calculated based on the cost. The formula is as follows: Where m is the number of ants, is the pheromone increase left by the kth ant on the edge (i, j), which can be defined as Where Q is a constant, L k is the length of the path taken by the kth ant.

6. The thermal battery pole piece coating production line based on modular design according to claim 1 is characterized in that: The system state equation in the state space model of the coating liquid flow control system established in the coating flow control is: in, Q k Indicates the flow rate of the coating liquid, represents the flow rate change rate, represents the state transfer matrix, represents process noise; The measurement equation is: in, Indicates the measured value of the flow sensor, represents the measurement noise; The formula for predicting the system state at the next moment based on the system state equation is: The forecast error covariance formula is: in, is the estimated value of the state at the previous moment, is the error covariance at the previous moment, is the process noise covariance; Update the state estimate based on the measurements: Error covariance: Among them, Kalman gain, given by Calculated, is the measurement noise covariance, is the identity matrix.

7. The thermal battery pole piece coating production line based on modular design according to claim 1 is characterized in that: In the coating flow control, the temperature value during the coating process is set to T set , collect the actual temperature T(t) during the coating process in real time, and calculate the temperature error e(t) = T set -T(t), according to the error e(t), the output control quantity is calculated by the PID controller, the formula is as follows: Among them, K i Indicates the integral coefficient, K d represents the differential coefficient, by adjusting K p , K i and K d value, so that the temperature can quickly and steadily reach the set value.

8. The thermal battery pole piece coating production line based on modular design according to claim 1 is characterized in that: In the defect recognition, the deep convolutional neural network includes multiple convolutional layers, pooling layers and fully connected layers. The convolutional layer is used to extract the features of the electrode image. The operation of the convolutional layer is expressed as: in, is the jth feature map of the lth layer, is the convolution kernel, is the bias term, f(·) is the activation function, M j is the set of input feature maps; The pooling layer is used to reduce the resolution of the feature map and the amount of data while maintaining the invariance of the features. The fully connected layer maps the extracted features to different defect categories.

9. The thermal battery pole piece coating production line based on modular design according to claim 1, characterized in that: In the defect recognition training process, a large number of defective and non-defective electrode images are prepared as training data, and the images are annotated to mark different defect types. The training data is input into the network, and the network parameters are optimized by the back propagation algorithm. The loss function adopts the cross entropy loss function: Where N is the number of samples, y i is the true label, is the probability of this type of defect predicted by the network; During the training process, by adjusting parameters such as learning rate and optimizer, the network gradually converges and can accurately identify various defects on the electrode, including scratches, holes, and uneven coating.

Images