Battery diaphragm coating agent formula optimization and process parameter self-correction method and system

Abstract

The invention relates to the technical field of formula parameter optimization, in particular to a battery diaphragm coating agent formula optimization and process parameter self-correction method and system, and the method comprises the following steps: an obtaining step: obtaining historical data of coating agent production; acquiring performance index data of a plurality of coating agent formulas; a processing step: establishing a prediction model of the coating agent formula and the product performance based on the historical data; screening out an optimized formula combination according to the performance index data; based on the prediction model, predicting coating performance indexes under different process parameters; an output step: generating an optimized coating agent formula; according to the method, the self-correcting process parameters are output, and a large number of formula combinations can be rapidly evaluated in a virtual environment by establishing a prediction model of the coating agent formula and the product performance, so that the number of actual tests is greatly reduced. Not only are time and cost saved, but also the influence on the environment is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of formula parameter optimization, and more specifically to a method and system for formula optimization and process parameter self-correction of a battery separator coating agent. Background Art

[0002] Battery separator is one of the key components of lithium-ion batteries, and its performance directly affects the safety, cycle life and energy density of the battery. In order to improve the comprehensive performance of the separator, coating technology is widely used in the separator manufacturing process. However, with the rapid development of battery technology and the continuous increase in market demand, traditional coating agent formulation optimization and process parameter adjustment methods have been unable to meet the requirements of modern production.

[0003] At present, the optimization of battery separator coating formulations mainly relies on empirical rules and trial and error. This method is not only time-consuming and labor-intensive, but also difficult to find the best balance between multiple performance indicators. For example, improving the mechanical strength of the coating may lead to a decrease in ionic conductivity, and increasing the coating thickness may affect the energy density of the battery. In addition, due to differences in raw material batches and fluctuations in the production environment, even the same formulation may result in significant differences between different batches of products.

[0004] In terms of process parameter adjustment, existing technologies usually use fixed process parameters or simple PID control methods. These methods are difficult to adapt to complex production environments and changing raw material properties. For example, when the ambient temperature or humidity changes, fixed coating temperature and speed may cause fluctuations in coating quality. Although simple PID control can adjust parameters to a certain extent, it is difficult to handle the complex situations of multi-parameter coupling and nonlinear systems.

[0005] In addition, existing quality control methods mainly rely on offline detection and manual judgment, which makes it difficult to achieve real-time monitoring and rapid response. This not only increases production costs, but may also lead to the production of large quantities of substandard products.

[0006] In view of the above problems, there is an urgent need for a method and system that can optimize the coating agent formula and realize self-correction of process parameters. The method and system should be able to comprehensively consider multiple performance indicators and quickly find the optimal formula; be able to automatically adjust process parameters according to real-time production data; and be able to realize online quality monitoring and timely intervention. Summary of the invention

[0007] The battery separator coating agent formula optimization and process parameter self-correction method and system provided by the present invention are proposed to solve the above technical problems. The method acquires historical production data and performance index data, establishes a prediction model of coating agent formula and product performance, and realizes the optimization of the formula. At the same time, by real-time collection of process parameters and performance data, combined with a deep reinforcement learning algorithm, automatic correction of process parameters is realized. This method not only greatly improves the efficiency of formula optimization and process adjustment, but also significantly improves the stability of product quality.

[0008] The present invention provides a method for optimizing the formula of a battery separator coating agent and self-correcting process parameters, comprising:

[0009] The acquisition steps include:

[0010] Obtain historical data on coating production;

[0011] Obtain performance indicator data for multiple coating formulations;

[0012] Processing steps include:

[0013] Based on the historical data, a prediction model of coating agent formulation and product performance is established;

[0014] According to the performance index data, an optimized formula combination is screened out;

[0015] Based on the prediction model, predict the coating performance index under different process parameters;

[0016] Output steps include:

[0017] Generate optimized coating formulation;

[0018] Output self-correcting process parameters.

[0019] Preferably, the obtaining step specifically includes:

[0020] Collecting performance parameter data of the coated battery separator through an online quality inspection system, wherein the performance parameters include film rupture temperature, liquid absorption rate, and coating thickness data;

[0021] The real-time viscosity data of the coating agent is obtained through an online viscosity detection device.

[0022] Preferably, establishing a prediction model in the processing step specifically includes:

[0023] Performing data preprocessing on the historical data, including data noise reduction, deduplication, and standardization;

[0024] A deep neural network is used to establish a predictive model for the mapping relationship between coating agent formula and product performance.

[0025] Preferably, the screening and optimizing of the formula combination in the processing step specifically includes:

[0026] Convert coating formulation design into a multi-objective optimization problem;

[0027] The Bayesian optimization algorithm is used to solve the multi-objective optimization problem and obtain an optimized formula combination.

[0028] Preferably, the process also includes a process parameter self-correction step:

[0029] Real-time collection of process parameters and coating performance data on the coating line;

[0030] Establish process parameter adjustment strategy based on deep reinforcement learning algorithm;

[0031] According to the adjustment strategy, the coating temperature, coating speed and coating viscosity are dynamically adjusted.

[0032] Preferably, the deep reinforcement learning algorithm in the process parameter self-correction step specifically includes:

[0033] The real-time viscosity value and coating temperature are used as the state space;

[0034] The coating speed and ambient temperature are used as the action space;

[0035] The reward function is designed based on film rupture temperature, liquid absorption rate, and coating thickness.

[0036] Preferably, the coating thickness adaptive adjustment step is also included:

[0037] Real-time collection of the thickness of the diaphragm after coating and the bonding strength of the coating;

[0038] Calculate the coating thickness adjustment strength according to the thickness deviation of the diaphragm after coating and the thickness deviation of the base film;

[0039] The coating temperature, coating speed and coating viscosity are adjusted based on the coating thickness adjustment intensity.

[0040] Preferably, the coating thickness adjustment intensity calculation formula in the coating thickness adaptive adjustment step is:

[0041] t=k 1 *A+k 2 *B

[0042] Where A is the thickness deviation of the diaphragm after coating, B is the thickness deviation of the base film, t is the coating thickness adjustment strength, k 1 is the first coefficient, k 2 is the second coefficient.

[0043] As a preference, an online quality monitoring step is also included:

[0044] The composition of the coating agent is detected in real time by an online infrared spectrometer;

[0045] The viscosity of the coating agent is detected in real time by an online viscometer;

[0046] The thickness of the coated diaphragm and the bonding strength of the coating are tested by an online thickness detector and a bonding strength tester.

[0047] A battery separator coating agent formula optimization and process parameter self-correction system for executing the method includes:

[0048] A data acquisition module is used to obtain historical data of coating agent production and performance index data of multiple coating agent formulas;

[0049] A prediction modeling module, used to establish a prediction model of coating agent formulation and product performance based on the historical data;

[0050] A recipe optimization module, used to screen out an optimized recipe combination according to the performance indicator data;

[0051] Process parameter self-correction module, used to dynamically adjust process parameters based on deep reinforcement learning algorithm;

[0052] Coating thickness adjustment module, used to adjust coating thickness in real time;

[0053] Online quality monitoring module, used to monitor various indicators of the coating process in real time;

[0054] Among them, the system realizes data storage, processing and model training through a cloud platform.

[0055] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0056] The method and system of the present invention have the following significant technical effects:

[0057] First, by establishing a prediction model for coating agent formula and product performance, the present invention can quickly evaluate a large number of formula combinations in a virtual environment, greatly reducing the number of actual tests. This not only saves time and cost, but also reduces the impact on the environment. At the same time, because the model takes into account the mutual influence between multiple performance indicators, it can find the best balance point of each indicator, thereby improving the overall performance of the product.

[0058] Secondly, the process parameter self-correction function of the present invention can respond to changes in the production environment and raw material characteristics in real time. Through the deep reinforcement learning algorithm, the system can continuously learn and optimize the adjustment strategy, gradually improving the accuracy and efficiency of parameter adjustment. This adaptive ability makes the production process more stable and significantly reduces product quality fluctuations caused by improper process parameters.

[0059] Furthermore, the online quality monitoring function of the present invention realizes real-time monitoring of the entire process. Through real-time detection of multiple indicators such as infrared spectrum, viscosity, thickness, etc., the system can promptly detect potential quality problems and quickly take corrective measures. This greatly reduces the generation rate of defective products and improves production efficiency.

[0060] Finally, the method and system of the present invention realize the organic combination of formula optimization, process adjustment and quality control. These three aspects cooperate and promote each other to form a closed-loop intelligent manufacturing system. For example, quality monitoring data is not only used for real-time control, but also can be fed back to the formula optimization module to further improve the formula design. This holistic optimization method can continuously improve product quality and production efficiency.

[0061] In summary, the battery separator coating agent formula optimization and process parameter self-correction method and system provided by the present invention realizes all-round optimization of the coating process through intelligent algorithms and advanced online monitoring technology. This not only greatly improves the performance and quality stability of the battery separator, but also makes an important contribution to the technological progress of the lithium-ion battery industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 The present invention is a flow chart of the method.

[0063] Figure 2 It is a logic block diagram of the data acquisition module of the present invention.

[0064] Figure 3 It is a logical block diagram of the prediction modeling module and the recipe optimization module of the present invention.

[0065] Figure 4 It is a logic block diagram of the process parameter self-correction module and the coating thickness adjustment module of the present invention. DETAILED DESCRIPTION

[0066] Please refer to Figure 1-4 The present invention provides a method and system for optimizing the formula of a battery separator coating agent and self-correcting process parameters. The method comprises the following steps:

[0067] First, in the acquisition step, the method acquires historical data of coating agent production and performance index data of multiple coating agent formulations. Preferably, the historical data includes various parameters collected during the coating agent production process over a period of time (e.g., 3-6 months), such as raw material ratios, process parameters, product performance, etc. The performance index data includes key indicators such as film breaking temperature, liquid absorption rate, coating thickness, etc. corresponding to different formulations.

[0068] Next, in the processing step, the method establishes a prediction model of coating agent formula and product performance based on the acquired historical data. The prediction model adopts a deep neural network structure, which can capture the complex nonlinear relationship between formula and performance. In one embodiment of the present invention, the neural network includes 3-5 hidden layers, and the number of neurons in each layer is between 50-200. The input of the model is the distribution ratio of each group of coating agents and key process parameters, and the output is the predicted product performance index.

[0069] Then, the method screens out the optimized formula combination according to the performance index data. Preferably, this step adopts a multi-objective optimization algorithm, such as the NSGA-II algorithm, and simultaneously considers multiple performance indicators such as film rupture temperature and liquid absorption rate to obtain a set of Pareto optimal solutions. In another embodiment of the present invention, expert experience can be introduced to assign weights to different indicators, and the multi-objective problem can be converted into a single-objective optimization problem by using a weighted sum method.

[0070] Finally, based on the established prediction model, this method predicts the coating performance indicators under different process parameters. This step can be done by performing a large number of sampling in the process parameter space through methods such as Monte Carlo simulation to obtain the distribution of performance indicators.

[0071] In the output step, the method generates an optimized coating agent formula and outputs self-correcting process parameters. The optimized formula usually includes the mass percentages of 3-5 main components, such as 30-40% binder, 20-30% inorganic filler, 25-35% solvent, etc. The self-correcting process parameters include key parameters such as coating temperature, coating speed, coating amount, etc.

[0072] Furthermore, the acquisition step of the method of the present invention specifically includes: collecting performance parameter data of the battery diaphragm after coating through an online quality inspection system, and obtaining real-time viscosity data of the coating agent through an online viscosity detection device. Among them, the performance parameters include film rupture temperature, liquid absorption rate, and coating thickness data. Preferably, the film rupture temperature is measured by differential scanning calorimetry (DSC) with a measurement range of 80-200°C; the liquid absorption rate is determined by immersion method, generally between 100-300%; the coating thickness is measured by a precision thickness gauge, and the typical value is 1-5μm.

[0073] The collection of real-time viscosity data is crucial to ensure coating quality. The present invention uses an online rotational viscometer to monitor the viscosity of the coating agent in real time within the range of 0.1-100000 mPa·s. When the viscosity deviates from the set value by ±10%, the system will automatically alarm and start the adjustment procedure.

[0074] When building a prediction model in the processing step, the historical data needs to be preprocessed first. This includes the following aspects:

[0075] 1. Data denoising: Use methods such as wavelet transform to remove high-frequency noise.

[0076] 2. Deduplication: Delete duplicate data records and keep the one with the latest timestamp.

[0077] 3. Standardization: Normalize parameters of different dimensions, usually using the z-score standardization method:

[0078]

[0079] Among them, x' is the standardized value, x is the original value, μ is the mean, and σ is the standard deviation.

[0080] The preprocessed data will be used to train a deep neural network model. The model uses a multi-layer perceptron (MLP) structure and the activation function is ReLU:

[0081] f(x)=max(0,x)

[0082] The loss function of the model uses the mean square error (MSE):

[0083]

[0084] Where n is the number of samples, y i is the true value, is the predicted value.

[0085] The model was trained using the back-propagation algorithm and the Adam optimizer. The initial value of the learning rate was set to 0.001, and the learning rate decay strategy was adopted, decaying by 10% every 50 epochs.

[0086] The method of the present invention can not only optimize the coating agent formula, but also realize the automatic correction of process parameters, greatly improving the intelligent level of battery separator production and product quality stability. In the processing steps of the method of the present invention, screening and optimizing the formula combination is a key link. Specifically, this step converts the coating agent formula design into a multi-objective optimization problem. Preferably, the method considers the three main performance indicators of film rupture temperature, liquid absorption rate and coating thickness, and converts its performance indicators into a three-objective optimization problem:

[0087] minF(x)=[f 1 (x),f 2 (x),f 3 (x)]

[0088] Among them, x represents the proportion of each component in the coating formulation, f 1 (x), f 2 (x) and f 3 (x) represent the objective functions of film rupture temperature, liquid absorption rate and coating thickness, respectively.

[0089] In order to solve this multi-objective optimization problem, the present invention adopts the Bayesian optimization algorithm. The core idea of ​​the algorithm is to construct a probability model of the objective function and approximate the real Pareto optimal solution set by continuously sampling and updating the model. In a preferred embodiment of the present invention, Gaussian Process Regression (GPR) is used as the probability model:

[0090] f(x)~GP(m(x),k(x,x′))

[0091] Among them, m(x) is the mean function, k(x,x') is the covariance function. This method uses the commonly used Matérn5 / 2 kernel function as the covariance function:

[0092]

[0093] in, r is the distance metric, σ and l i is a hyperparameter.

[0094] In each iteration, the method selects the next sampling point by maximizing the expected improvement (EI):

[0095] EI(x)=E[max(f(x)-f(x + ),0)]

[0096] Among them, x + It is the best solution currently known.

[0097] In this way, the method of the present invention can quickly find a near-optimal coating agent formulation combination within a relatively small number of experiments.

[0098] Next, the present invention also includes a process parameter self-correction step. This step first collects the process parameters and coating performance data on the coating line in real time. Preferably, the collection frequency is 1-10 times per second to ensure that process fluctuations can be captured. The collected parameters include but are not limited to: coating temperature, coating speed, coating amount, drying temperature, etc.

[0099] Based on the collected data, the method uses a deep reinforcement learning algorithm to establish a process parameter adjustment strategy. Specifically, the present invention uses a deep Q-network (DQN) to achieve this goal. In one embodiment of the present invention, the structure of DQN is as follows:

[0100] 1. Input layer: Receives status information, including current process parameters and recent performance indicators.

[0101] 2. Hidden layer: 3-5 fully connected layers, 128-256 neurons in each layer, and the activation function is ReLU.

[0102] 3. Output layer: Q value corresponding to each possible action (such as increasing / decreasing coating temperature, speeding up / slowing down coating speed, etc.).

[0103] DQN training uses Experience Replay and Target Network technology to improve the stability of learning. The reward function is designed as follows:

[0104] R=w 1 ·(T target -|TT target |)+w 2 ·(1-|η-η target |)+w 3 ·(1-|hh target |)

[0105] Where T, η and h represent the film breaking temperature, liquid absorption rate and coating thickness respectively, the subscript target represents the target value, and w 1 、w 2 and w 3 is the weight coefficient.

[0106] According to the adjustment strategy obtained through training, this method can dynamically adjust the coating temperature, coating speed and coating viscosity. For example, when the coating thickness is detected to be low, the system may reduce the coating speed or increase the coating amount; when the film breaking temperature does not meet the standard, the coating temperature may be adjusted or the proportion of certain components in the coating formula may be changed.

[0107] In the method of the present invention, the specific implementation of the deep reinforcement learning algorithm includes the following key points:

[0108] First, the real-time viscosity value and coating temperature are used as the state space. Preferably, the viscosity value range is 100-10000 mPa·s, and the coating temperature range is 60-120° C. These two parameters are discretized into 10 levels, thereby forming a 10x10 state space.

[0109] Secondly, the coating speed and ambient temperature are used as the action space. The adjustment range of the coating speed is ±10%, and the adjustment range of the ambient temperature is ±5°C. Each parameter can be increased, decreased, or kept unchanged, so the action space has a total of 3x3=9 possible combinations.

[0110] Finally, a reward function is designed based on film breaking temperature, liquid absorption rate, and coating thickness. The specific form is as follows:

[0111]

[0112] Among them, α, β and γ are weight coefficients, σ T , σ η and σ h is the standard deviation parameter that controls the sensitivity of the reward function.

[0113] Through this design, the method of the present invention can continuously learn and optimize in the actual production process, and gradually improve the accuracy and efficiency of process parameter adjustment. The method of the present invention also includes a coating thickness adaptive adjustment step. In this step, the thickness of the diaphragm after coating and the bonding strength of the coating are first collected in real time. Preferably, the thickness is collected using a laser thickness gauge with an accuracy of ±0.1μm; the bonding strength is measured by a peel strength tester with a measurement range of 0-10N / cm.

[0114] According to the thickness deviation of the diaphragm after coating and the thickness deviation of the base film, the method calculates the coating thickness adjustment strength. Specifically, the calculation formula of the coating thickness adjustment strength t is:

[0115] t=k 1 ·A+k 2 ·B

[0116] Where A is the thickness deviation of the diaphragm after coating, B is the thickness deviation of the base film, and k 1 is the first coefficient, k 2 is the second coefficient. In a preferred embodiment of the present invention, k 1 The value range is 0.6-0.8, k 2 The value range is 0.2-0.4. This value can make the system more sensitive to the change of the thickness of the diaphragm after coating, while also taking into account the influence of the fluctuation of the base film thickness.

[0117] Based on the calculated coating thickness adjustment strength, the method further adjusts the coating temperature, coating speed and coating viscosity. The adjustment strategy is as follows:

[0118] Coating temperature adjustment:

[0119] ΔT=α·t

[0120] Coating speed adjustment:

[0121] Δv=-β·t

[0122] Paint viscosity adjustment:

[0123] Δη=γ·t

[0124] Among them, α, β and γ are adjustment coefficients, and their values ​​need to be determined based on actual production experience. Generally speaking, the value range of α is 0.5-2℃ / (μm·min), the value range of β is 0.1-0.5m / (min·μm), and the value range of γ is 10-50mPa·s / (μm·min).

[0125] Through this adaptive adjustment mechanism, the method of the present invention can quickly respond to changes in coating thickness and maintain the stability of product quality.

[0126] In order to further improve the monitoring capability of the production process, the method of the present invention also includes an online quality monitoring step. This step is mainly achieved through the following aspects:

[0127] First, the composition of the coating agent is detected in real time by an online infrared spectrometer. The present invention uses a Fourier transform infrared spectrometer (FTIR) with an operating wavelength range of 4000-400 cm^-1 and a resolution of up to 0.5 cm^-1. By analyzing the position and intensity of the characteristic peaks, the content changes of each component of the coating agent can be monitored in real time.

[0128] Secondly, the viscosity of the coating agent is detected in real time by an online viscometer. This method uses a rotary online viscometer with a measurement range of 1-100000 mPa·s and an accuracy of ±1%. The viscosity data is updated once per second, providing timely feedback for process parameter adjustments.

[0129] Finally, the thickness of the coated diaphragm and the bonding strength of the coating are tested by an online thickness detector and bonding strength tester. The thickness test uses the β-ray method with a measurement range of 0-50μm and an accuracy of ±0.1μm. The bonding strength test uses an online peel strength tester with a measurement range of 0-20N / cm and an accuracy of ±0.1N / cm.

[0130] These online monitoring data are not only used for real-time quality control, but also provide valuable basis for subsequent process optimization and formula adjustment.

[0131] The present invention also provides a battery diaphragm coating agent formula optimization and process parameter self-correction system corresponding to the above method. The system includes the following key modules:

[0132] The data acquisition module 1 is used to obtain the historical data of coating agent production and the performance index data of multiple coating agent formulas. The module includes multiple sensors and data acquisition cards, which can collect various parameters on the production line in real time.

[0133] Prediction modeling module 2 is used to establish a prediction model of coating agent formula and product performance based on the historical data. This module uses a deep learning framework, such as TensorFlow or PyTorch, to implement neural network training and reasoning.

[0134] The recipe optimization module 3 is used to select the optimal recipe combination according to the performance index data. This module implements a multi-objective optimization algorithm, such as NSGA-II or MOEA / D.

[0135] The process parameter self-correction module 4 is used to dynamically adjust the process parameters based on the deep reinforcement learning algorithm. This module includes a training submodule and an execution submodule, the former is responsible for offline training of the model, and the latter is responsible for online decision-making.

[0136] The coating thickness adjustment module 5 is used to adjust the coating thickness in real time. The module receives data from the online detection device and outputs control instructions according to a preset adjustment strategy.

[0137] The online quality monitoring module 6 is used to monitor various indicators of the coating process in real time. This module integrates a variety of online detection equipment, including infrared spectrometer, viscometer, thickness detector, etc.

[0138] It is worth noting that this system realizes data storage, processing and model training through the cloud platform. The cloud platform adopts a distributed architecture, including data storage layer, computing layer and application layer. The data storage layer uses a distributed file system and NoSQL database to ensure efficient storage and retrieval of large-scale data. The computing layer uses big data processing frameworks such as Spark to support batch processing and stream processing. The application layer provides a friendly user interface and API interface to facilitate operators to monitor and control the entire production process.

[0139] Through this system architecture, the present invention realizes the optimization of the battery separator coating agent formula and the automatic correction of process parameters, which greatly improves the production efficiency and the stability of product quality.

[0140] In order to verify the superiority of the battery separator coating agent formula optimization and process parameter self-correction method and system of the present invention, a series of comparative tests were conducted. The specific operation process, test method and result analysis of the embodiments and comparative examples will be described in detail below.

[0141] Example 1: Using the method of the present invention, historical production data for the past 6 months, including coating agent formula, process parameters and corresponding product performance indicators, were first collected. The data preprocessing module was used to reduce noise, remove duplicates and standardize these data. Subsequently, a five-layer deep neural network model was constructed using the predictive modeling module 2 to predict the relationship between the coating agent formula and product performance.

[0142] Based on the trained model, multi-objective optimization was performed using the formulation optimization module 3. The optimization objectives included increasing the film rupture temperature and the liquid absorption rate while maintaining an appropriate coating thickness. A set of Pareto optimal solutions was obtained through the Bayesian optimization algorithm, from which an optimal formulation that balanced various properties was selected.

[0143] During the production process, the process parameter self-correction module 4 and the coating thickness adjustment module 5 are enabled. These two modules are based on deep reinforcement learning algorithms and dynamically adjust the coating temperature, coating speed and coating viscosity according to the real-time collected data. At the same time, the online quality monitoring module 6 continuously monitors product quality and provides timely feedback for parameter adjustment.

[0144] Comparative Example 1: In the comparative test, the coating agent formula was optimized by the traditional orthogonal experimental method, and fixed process parameters were used for production. The formula optimization process took two weeks and a total of 27 sets of experiments were conducted. During the production process, the operator manually adjusted the process parameters based on experience and regular sampling results.

[0145] Testing standards and methods:

[0146] 1. Membrane rupture temperature: Differential scanning calorimetry (DSC) was used to test the film according to GB / T 33824-2017.

[0147] 2. Liquid absorption rate: measured in accordance with GB / T 33825-2017 standard using the n-hexane immersion method.

[0148] 3. Coating thickness: Use a precision thickness gauge and measure in accordance with GB / T 33832-2017 standard.

[0149] 4. Ionic conductivity: tested using the AC impedance method in accordance with GB / T 33828-2017 standard.

[0150] 5. Mechanical properties: Use a tensile testing machine to test in accordance with GB / T 33831-2017 standard.

[0151] The test results are shown in the following table:

[0152] Performance Indicators Example 1 Comparative Example 1 Percentage increase Membrane breaking temperature (℃) 165 152 8.55％ Liquid absorption rate (%) 187 165 13.33％ Coating thickness uniformity (%) 97.5 92.3 5.63％ Ionic conductivity (mS / cm) 1.23 1.05 17.14％ Tensile strength(MPa) 125 108 15.74％ Production qualification rate (%) 98.7 94.2 4.78％

[0153] It can be seen from the test results that the battery separator produced by the method of the present invention is superior to the traditional method in all performance indicators. It is particularly noteworthy that the membrane rupture temperature has increased by 8.55%, which means that the safety performance of the battery has been significantly improved. The liquid absorption rate has increased by 13.33%, which is conducive to the full infiltration of the electrolyte and improves the capacity and rate performance of the battery. The improvement in the uniformity of the coating thickness reflects the advantage of the method of the present invention in terms of process stability.

[0154] The ionic conductivity has increased by 17.14%, which is a very significant improvement, and it is directly related to the internal resistance and power performance of the battery. The increase in tensile strength enhances the mechanical stability of the separator, which is beneficial to improving the cycle life of the battery.

[0155] Finally, although the increase in production qualification rate seems small, considering the already high base, the 4.78% increase actually means that the defective rate has been reduced by nearly half, which is of great significance for reducing production costs and improving resource utilization.

[0156] These results fully demonstrate the excellent performance of the method of the present invention in optimizing coating agent formulations and automatically adjusting process parameters. Through intelligent algorithms and real-time feedback systems, the method can more accurately control the production process, improve product performance, and significantly improve production efficiency and stability. In particular, the present invention shows obvious advantages in indicators such as coating uniformity and ionic conductivity that are difficult to optimize simultaneously through traditional methods.

[0157] It is worth mentioning that the formulation optimization process in Example 1 only took 3 days, which is much shorter than the two weeks in the comparative example. This not only greatly shortens the product development cycle, but also reduces the consumption of raw materials and energy. At the same time, the automated parameter adjustment system reduces the burden on operators and makes the production process more stable and controllable.

[0158] In summary, the battery separator coating agent formula optimization and process parameter self-correction method and system of the present invention have shown significant advantages in improving product performance, shortening development cycle, improving production efficiency and other aspects, and have made important contributions to the advancement of battery separator production technology.

[0159] It should be noted 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, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing the formula of a battery separator coating agent and self-correcting process parameters, characterized in that: include: The acquisition steps include: Obtain historical data on coating production; Obtain performance indicator data for multiple coating formulations; Processing steps include: Based on the historical data, a prediction model of coating agent formulation and product performance is established; According to the performance index data, an optimized formula combination is screened out; Based on the prediction model, predict the coating performance index under different process parameters; Output steps include: Generate optimized coating formulation; Output self-correcting process parameters.

2. The method according to claim 1, characterized in that The acquisition step specifically includes: Collecting performance parameter data of the coated battery separator through an online quality inspection system, wherein the performance parameters include film rupture temperature, liquid absorption rate, and coating thickness data; The real-time viscosity data of the coating agent is obtained through an online viscosity detection device.

3. The method according to claim 1, characterized in that The process of establishing a prediction model specifically includes: Performing data preprocessing on the historical data, including data noise reduction, deduplication, and standardization; A deep neural network is used to establish a predictive model for the mapping relationship between coating agent formula and product performance.

4. The method according to claim 1, characterized in that: The screening and optimization of the formula combination in the processing step specifically includes: Convert coating formulation design into a multi-objective optimization problem; The Bayesian optimization algorithm is used to solve the multi-objective optimization problem and obtain an optimized formula combination.

5. The method according to claim 1, characterized in that: It also includes the process parameter self-correction step: Real-time collection of process parameters and coating performance data on the coating line; Establish process parameter adjustment strategy based on deep reinforcement learning algorithm; According to the adjustment strategy, the coating temperature, coating speed and coating viscosity are dynamically adjusted.

6. The method according to claim 5, characterized in that The deep reinforcement learning algorithm in the process parameter self-correction step specifically includes: The real-time viscosity value and coating temperature are used as the state space; The coating speed and ambient temperature are used as the action space; The reward function is designed based on film rupture temperature, liquid absorption rate, and coating thickness.

7. The method according to claim 1, characterized in that It also includes the coating thickness adaptive adjustment step: Real-time collection of the thickness of the diaphragm after coating and the bonding strength of the coating; Calculate the coating thickness adjustment strength according to the thickness deviation of the diaphragm after coating and the thickness deviation of the base film; The coating temperature, coating speed and coating viscosity are adjusted based on the coating thickness adjustment intensity.

8. The method according to claim 7, characterized in that The calculation formula for the coating thickness adjustment strength in the coating thickness adaptive adjustment step is: t=k1*A+k2*B Among them, A is the thickness deviation of the diaphragm after coating, B is the deviation of the base film thickness, t is the coating thickness adjustment strength, k1 is the first coefficient, and k2 is the second coefficient.

9. The method according to claim 1, characterized in that: Also includes online quality monitoring steps: The composition of the coating agent is detected in real time by an online infrared spectrometer; The viscosity of the coating agent is detected in real time by an online viscometer; The thickness of the coated diaphragm and the bonding strength of the coating are tested by an online thickness detector and a bonding strength tester.

10. A battery separator coating agent formula optimization and process parameter self-correction system for executing the method according to any one of claims 1 to 9, characterized in that: include: A data acquisition module is used to obtain historical data of coating agent production and performance index data of multiple coating agent formulas; A prediction modeling module, used to establish a prediction model of coating agent formulation and product performance based on the historical data; A recipe optimization module, used to screen out an optimized recipe combination according to the performance indicator data; Process parameter self-correction module, used to dynamically adjust process parameters based on deep reinforcement learning algorithm; Coating thickness adjustment module, used to adjust coating thickness in real time; Online quality monitoring module, used to monitor various indicators of the coating process in real time; Among them, the system realizes data storage, processing and model training through a cloud platform.

Images