A shunt protection layer coating process parameter optimization method and system
By acquiring the characteristics of the shunt resistor and initial process parameters, using machine learning models to predict the resistance change rate and protection parameter distribution, and performing non-standard compensation, the optimization strategy is dynamically adjusted, which solves the shortcomings of the resistor coating process parameter setting in the existing technology and improves the stability of the resistor's electrical performance and the protection effect.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FOSHAN HAOYUN ELECTRICAL APPLIANCE ACCESSORIES CO LTD
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-09
AI Technical Summary
The existing shunt coating process parameters rely on experience, making it difficult to adapt to non-standard resistors. The prediction model is singular and lacks reliability, resulting in resistance value drift and decreased measurement accuracy.
By acquiring the characteristics of the resistive element and initial process parameters, machine learning models are used to predict the resistance change rate and the distribution of protection parameters. Compensation is then performed using a non-standard resistance coefficient, and optimization strategies are dynamically adjusted and process parameters are iteratively optimized to ensure resistance stability and protection effectiveness.
It achieves a comprehensive consideration of resistance change rate and protection performance, improves the adaptability and prediction accuracy of parameter optimization, and ensures the stability and effective protection of the electrical performance of the resistor.
Smart Images
Figure CN122177308A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of electronic component manufacturing technology, specifically to a method and system for optimizing the coating process parameters of a shunt protector layer. Background Technology
[0002] As a precision measuring element, the resistive element of a shunt needs to be coated with a protective layer to prevent environmental damage. However, the coating process itself can interfere with the microstructure of the resistive element due to material stress and interface effects, causing unexpected drift in the resistance value and directly affecting the measurement accuracy.
[0003] In existing technologies, the determination of coating process parameters relies on operator experience or trial-and-error methods based on limited samples. Some solutions use fixed process formulations, which cannot adapt to the characteristic differences of resistive elements of different specifications and batches. When dealing with non-standard resistive elements, problems such as insufficient protection or excessive resistance variation easily occur. Although there have been attempts to introduce experimental design or basic simulation, they usually only focus on a single performance index, or evaluate protection performance in isolation, or roughly estimate the resistance change trend, failing to establish a complete model that can synergistically predict the distribution of resistance change rate and protection parameter distribution. In addition, existing technologies lack a quantitative evaluation of the reliability of prediction results, as well as a mechanism for dynamically adjusting optimization strategies based on this evaluation. As a result, the optimization process is often based on prediction data with insufficient credibility, and the effect of the obtained optimal parameters in practical applications fluctuates significantly. Ultimately, it is impossible to reliably maintain the electrical performance stability of the resistive element while ensuring protection. Summary of the Invention
[0004] This invention addresses the technical problems of existing coating process parameters relying on experience, difficulty in adapting to non-standard resistors, and the lack of reliable prediction models to guide the optimization process. It provides a method and system for optimizing the coating process parameters of a shunt protector layer.
[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for optimizing the coating process parameters of a shunt protector layer, comprising: Obtain the resistive characteristics of the resistor element inside the shunt and obtain the initial coating process parameters for coating. Based on the initial coating process parameters and resistor characteristics, shunt coating prediction is performed to obtain the first predicted resistance change rate distribution and the first predicted protection parameter distribution. Based on the characteristics of the resistive element, the non-standard resistance coefficient is analyzed and obtained. Non-standard compensation is performed on the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution. Combined with the first predicted protection parameter distribution, the first coating adaptability and the first confidence level are calculated. The initial coating process parameters are iteratively optimized. During the optimization process, the optimization strategy is set according to the confidence level until the optimization converges and the optimal coating process parameters are obtained. The distributor is then coated.
[0006] Secondly, the present invention provides a system for optimizing the coating process parameters of a shunt protector layer, comprising: The parameter acquisition module is used to acquire the resistive characteristics of the resistor element inside the shunt and to acquire the initial coating process parameters for coating. The coating prediction module is used to predict the shunt coating based on the initial coating process parameters and resistor characteristics, and to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution. The compensation and evaluation module is used to analyze and obtain the non-standard resistance coefficient based on the characteristics of the resistive body, perform non-standard compensation on the first predicted resistance change rate distribution, obtain the first compensated predicted resistance change rate distribution, and calculate the first coating adaptability and the first confidence level in combination with the first predicted protection parameter distribution. The optimization processing module is used to iteratively optimize the initial coating process parameters. During the optimization process, the optimization strategy is set according to the confidence level until the optimization converges, and the optimal coating process parameters are obtained to coat the splitter.
[0007] The beneficial effects of this invention are: Compared to existing technologies, this invention firstly utilizes a collaborative prediction model to simultaneously output the resistance change rate distribution and the protection parameter distribution, thereby achieving a comprehensive consideration of protection performance and resistance stability. Secondly, it introduces an analysis and compensation mechanism for non-standard resistance coefficients, effectively improving the adaptability and prediction accuracy of parameter optimization for non-standard or individually variably sized resistors. Thirdly, it creatively uses the confidence level of the prediction results as feedback and guidance for the optimization process, dynamically adjusting the optimization strategy based on the confidence level, enhancing the intelligence and robustness of the optimization process. Finally, through the aforementioned closed-loop optimization process, this invention can determine the optimal coating adaptability for specific shunt resistors with high confidence levels, ensuring effective protection while maximizing the stability of the resistor's electrical performance. Attached Figure Description
[0008] Figure 1 A flowchart illustrating a method for optimizing the coating process parameters of a shunt protector layer provided by the present invention; Figure 2 This is a schematic diagram of a system for optimizing the coating process parameters of a shunt protector layer provided by the present invention.
[0009] In the attached diagram, the components represented by each number are as follows: Parameter acquisition module 11, coating prediction module 12, compensation and evaluation module 13, optimization processing module 14. Detailed Implementation
[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0011] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0012] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.
[0013] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for optimizing the coating process parameters of a shunt protector layer, including: S10: Obtain the resistive characteristics of the resistor element inside the shunt and obtain the initial coating process parameters for coating. First, the resistivity characteristics of the resistive element inside the shunt are obtained. The resistive element inside the shunt refers to the core sensitive element inside the shunt that performs the function of current measurement. Its resistivity characteristics are a set of physical and electrical parameters that describe the inherent properties of the element, representing the specific specifications and state of the resistive element. This is used to accurately characterize the initial conditions and differences of individual resistive elements during the coating process optimization.
[0014] Secondly, the initial coating process parameters are obtained. These initial coating process parameters are the starting point for subsequent iterative search and optimization calculations, representing an initial scheme for the coating operation of the protective layer.
[0015] Specifically, the resistive characteristics of the resistor element within the shunt are obtained, and the initial coating process parameters for coating are acquired, including: Obtain the resistive characteristics of the resistor element inside the shunt, wherein the resistive characteristics include the resistor element specification parameters; Obtain the coating process parameter space for the distributor protective layer coating, and obtain the initial coating process parameters set within the coating process parameter space, wherein each coating process parameter includes coating position, coating liquid composition, coating amount and coating temperature.
[0016] First, the resistive characteristics of the resistors within the shunt are obtained. These resistive characteristics quantitatively describe the physical properties of the core resistive element in the shunt, with the core being the resistor specifications. Preferably, these specifications typically include specific indicators such as the resistor's geometric dimensions, material composition, nominal resistance value, and allowable tolerance range. Obtaining these resistive characteristics is fundamental to all subsequent analysis and optimization steps, providing precise input for evaluating the impact of the coating process on the electrical performance of a specific resistor.
[0017] Secondly, the coating process parameter space for the distributor protective layer is obtained. This space defines all adjustable process variables affecting the coating effect and their theoretical or empirical value ranges. This space is a multi-dimensional set, its dimensions composed of key process parameters, set by combining historical production data, material property studies, and process limit analysis. Then, from the defined space, a set of specific parameter values is selected or set as the starting point for optimization, i.e., the initial coating process parameters. Each set of specific parameters includes several key sub-items, including coating location, coating liquid composition, coating amount, and coating temperature.
[0018] Specifically, the coating location parameter defines the specific area or path on which the protective material is applied to the resistive surface. The coating liquid composition parameter specifies the chemical components constituting the protective liquid and their proportions. The coating amount parameter clarifies the mass or volume of protective material required per unit area or overall. The coating temperature parameter sets the environmental or material temperature conditions that need to be maintained during the coating process. Obtaining these initial coating process parameters establishes a clear initial state for subsequent predictive simulations and iterative optimization processes.
[0019] S20: Based on the initial coating process parameters and resistor characteristics, perform shunt coating prediction to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution; Specifically, based on the initial coating process parameters and resistive characteristics, shunt coating prediction is performed to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution, including: Call the splitter to coat the predictor; The initial coating process parameters and resistor characteristics are input into the shunt coating predictor, which outputs multiple first predicted resistance change rates and multiple first predicted protection parameters to obtain the distribution of the first predicted resistance change rate and the distribution of the first predicted protection parameters.
[0020] First, the pre-built and trained shunt coating predictor is invoked. This shunt coating predictor is a computational module based on a machine learning model, whose function is to establish a complex mapping relationship from input process parameters and resistive characteristics to post-coating performance indicators.
[0021] Specifically, the shunt coating predictor is pre-trained, and the training steps include: Based on the coating test data records of the shunt, collect the sample coating process parameter set and the sample resistivity characteristic set; Multiple resistance change amplitudes were collected for each group of samples under the coating process parameters and after coating of the resistive body characteristics. These were used as resistance change rate groups, and the sample resistance change rate group set was obtained by labeling. The corrosion protection parameters of the shunt were collected under the coating process parameters and after coating of the resistive body characteristics of each sample. These parameters were used as protection parameter groups. The sample protection parameter group set was obtained by labeling. Each protection parameter includes the proportion of the shunt corrosion rate reduction under the same environment. Based on machine learning, a coating resistance change prediction branch and a coating protection prediction branch are constructed; Using the sample coating process parameter set and sample resistivity feature set as inputs, and the sample resistance change rate set and sample protection parameter set as labels, respectively, supervised training tests are performed on the coating resistance change prediction branch and the coating protection prediction branch until the test converges.
[0022] First, based on the coating test data records of the shunt, two core datasets were systematically collected. The coating test data records are structured experimental archives accumulated during the coating process verification in the production and R&D of the shunt. These include detailed records of the process settings for each experiment, the physical specifications of the resistors used, and the performance indicators measured after the experiment, obtained through standard laboratory testing procedures and a data management system. The two core datasets are: first, a set of sample coating process parameters, where each sample coating process parameter represents a combination of process conditions used in a specific coating experiment; and second, a set of sample resistor characteristics, where each sample resistor characteristic corresponds to the characteristic parameters of the resistors participating in that experiment.
[0023] Secondly, to construct the prediction target, corresponding performance label data needs to be extracted from the coating test data records. Specifically, for each specific sample coating process parameter and its corresponding sample resistivity characteristics, a series of resistance change amplitude data are collected after coating through repeated measurements or tests on different samples, and organized into a resistance change rate group. The resistance change rate groups of all experimental groups together constitute the sample resistance change rate group set. At the same time, for the same process and resistivity conditions, the corrosion protection performance data exhibited by the shunt after coating, i.e., the shunt corrosion protection parameters, are collected. Specifically, each protection parameter includes the percentage reduction in corrosion rate of the shunt under the same environment, and the corrosion protection parameter from each test is organized into a protection parameter group. The protection parameter groups of all experimental groups then constitute the sample protection parameter group set.
[0024] Furthermore, after the data preparation is complete, a shunt coating predictor is constructed based on a machine learning framework. Specifically, this shunt coating predictor is designed to include two parallel functional branches: one branch is a coating resistance change prediction branch, specifically used to learn and predict the impact of the coating process on the resistance change rate; the other branch is a coating protection prediction branch, specifically used to learn and predict the impact of the coating process on the protection performance.
[0025] Specifically, the sample coating process parameter set and the sample resistivity feature set are used as common input features of the model. For the coating resistance change prediction branch, the sample resistance change rate set is used as its training label. The internal network parameters of this branch are adjusted using the backpropagation algorithm in conjunction with the gradient descent optimizer, so that the difference between the predicted distribution output by this branch and the true resistance change rate set distribution, such as the mean square error, is continuously reduced. For the coating protection prediction branch, the sample protection parameter set is used as its training label. The same optimization method is used to adjust the parameters of this branch so that its output distribution approximates the true protection parameter set distribution. The two branches are relatively independent during training, and the training process is continuously iterated until the test converges, at which point a shunt coating predictor with reliable prediction capabilities is obtained. The convergence condition is set according to the trend of the loss function value on the validation set. For example, when the validation loss decreases by less than 5‰ for 20 consecutive training cycles, or when the preset maximum number of training cycles, such as 1000, is reached, the training is considered to have converged.
[0026] For example, since the relationship between coating process parameters, resistivity characteristics, and post-coating resistance changes and protective performance exhibits a highly nonlinear and complex correlation, and ensemble learning models perform well in handling multi-dimensional inputs, predicting continuous target distributions, and suppressing overfitting, random forests, gradient boosting trees, or deep neural networks can be chosen as the basic algorithms for constructing the shunt coating predictor. Specifically, a deep neural network will be selected as an example for detailed description: Specifically, the architecture of this shunt coating predictor is a dual-branch parallel structure, mainly composed of a shared input layer, a feature abstraction layer, and independent prediction branch output layers. The shared input layer receives a concatenated vectorized input, which is composed of a normalized set of sample coating process parameters and a set of sample resistivity features, covering all key variables from materials to processes. The feature abstraction layer adopts a fully connected neural network structure with two hidden layers, with the number of hidden layer neurons set to twice and once the input dimension, respectively. Each layer uses the ReLU activation function to introduce nonlinear modeling capabilities, and a Dropout layer with a dropout rate of 0.3 is embedded after the first hidden layer to enhance the model's generalization ability and prevent excessive dependence on specific data patterns during training.
[0027] The prediction branch output layer contains two independent sub-networks. The first sub-network is the coating resistance change prediction branch, whose output layer is configured with ten neurons, corresponding to the main statistical features or decimals of the predicted resistance change rate distribution. The output is then processed through a Softplus activation function to ensure a positive prediction value. The second sub-network is the coating protection prediction branch, whose output layer also has ten neurons. The output is ultimately processed through a Sigmoid activation function, constraining the predicted corrosion rate reduction ratio to between zero and one.
[0028] During training, key hyperparameters included a learning rate of 0.0005, a maximum number of training epochs of 500, and a batch size of 32. The learning rate was set to balance optimization speed and training stability; the number of training epochs provided sufficient iterations for the model to learn fully; and the batch size was chosen to consider both computational efficiency and the stability of gradient estimation. The training process employed supervised learning. The input sample set, consisting of the sample coating process parameters and the sample resistivity features, along with two label sets—the sample resistance change rate set and the sample protection parameter set—were divided into training, validation, and test sets in an 8:1:1 ratio.
[0029] Furthermore, using the input samples from the training set as input and two corresponding label sets as supervision signals, the weight parameters of the two prediction branches are simultaneously but independently optimized through backpropagation and the Adam optimizer. The shared feature abstraction layer parameters are updated jointly by the gradients of the two branches. The training objective is to minimize the composite loss function, which consists of two weighted parts: one is the Wasserstein distance loss between the output of the coating resistance change prediction branch and the true resistance change rate distribution; the other is the mean square error loss between the output of the coating protection prediction branch and the true protection parameter distribution. The weights of the two parts are set to 5:5. The training process is monitored using a validation set, and the convergence condition is set to the composite loss value on the validation set decreasing by less than 0.5% over 20 consecutive training epochs. Training terminates when this condition is met. Ultimately, a shunt coating predictor with high accuracy and generalization ability is obtained. This shunt coating predictor can effectively model the complex influence of process parameters and resistive characteristics on the distribution of dual performance indicators after coating.
[0030] Furthermore, the initial coating process parameters and resistive characteristics obtained above are used as input data and fed into the trained shunt coating predictor. Upon receiving the input, the shunt coating predictor performs calculations and inferences based on the patterns it has learned internally. Considering the unavoidable random fluctuations and measurement errors in the actual coating process, it outputs multiple first predicted resistance change rates and multiple first predicted protection parameters, obtaining the distributions of the first predicted resistance change rates and the first predicted protection parameters.
[0031] Specifically, the output of multiple first predicted resistance change rates characterizes the range of changes in the resistance value of the resistive body that may result from the coating operation under given process parameters. Each resistance change rate value represents a possible change scenario, and the smaller the resistance change rate value, the better the resistance stability. The output of multiple first predicted protection parameters characterizes the corrosion protection effect that the coating layer may provide under the same process conditions. Specifically, it can be quantified as the proportion of reduction in corrosion rate, and the larger the value, the stronger the protection performance.
[0032] Finally, by statistically analyzing the multiple first predicted resistance change rates output, the distribution of the first predicted resistance change rate is obtained. Similarly, by statistically analyzing the multiple first predicted protection parameters output, the distribution of the first predicted protection parameters is obtained. The final distributions of the first predicted resistance change rate and the first predicted protection parameter distribution together constitute a comprehensive and quantitative predictive evaluation of the potential effect of the initial coating process parameter scheme.
[0033] S30: Based on the characteristics of the resistive element, analyze and obtain the non-standard resistance coefficient, perform non-standard compensation on the first predicted resistance change rate distribution, obtain the first compensated predicted resistance change rate distribution, and calculate the first coating adaptability and the first confidence level in combination with the first predicted protection parameter distribution. First, based on the characteristics of the resistive element, non-standard resistance coefficients are analyzed and obtained. Non-standard compensation is then applied to the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution, including: Obtain the characteristic set of resistor elements of the same family for shunts of the same type; Frequent mining of resistive features is performed within the same set of resistive features to obtain frequent resistive features. The deviation amplitude between the resistive element characteristics and the frequent resistive element characteristics is calculated and used as the resistance non-standard coefficient; Based on the non-standard resistance coefficient, a non-standard compensation coefficient is set, and a non-standard compensation calculation is performed on the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution.
[0034] First, obtain the characteristic set of resistors from the same family of shunts. A shunt of the same type refers to a product series with the same or similar design specifications, rated parameters, and expected application scenarios as the shunt to be coated. Obtaining the characteristic set of resistors from the same family of shunts reflects the characteristic range and concentration trend typically exhibited by this type of resistor in mass production. Specifically, it covers various specification parameters of resistors within a large number of similar shunts produced historically, such as the statistical distribution of geometric dimensions, typical material composition ratios, and the centralized fluctuation range of resistance values. This constitutes an objective benchmark data pool for assessing whether a single resistor is in a standard state or deviates from its standard.
[0035] Secondly, within the characteristic set of resistors of the same family, data mining algorithms are applied to analyze the frequency and combination patterns of all characteristic parameters, identifying the most representative or most common set of characteristic values, which are then determined as frequent resistor characteristics. These frequent resistor characteristics represent the standard or typical specifications of this type of resistor.
[0036] Specifically, frequent mining of resistive features is performed within the same set of resistive features to obtain frequent resistive features, including: Randomly select a first family of resistor features from the family of resistor features; Calculate the similarity between the features of the first family of resistive elements and the features of other family of resistive elements, and calculate the mean to obtain the first frequency coefficient; Continue calculating the frequency coefficients of other resistor features in the same family, and select the resistor features in the same family that have the maximum frequency coefficient as frequent resistor features.
[0037] First, the process begins by randomly selecting a feature vector from the set of features of resistors in the same family as a starting reference point. This vector is called the first feature of resistors in the same family. This first feature of resistors in the same family represents a specific resistor specification instance in the set of features of resistors in the same family.
[0038] Secondly, similarity calculation and frequency assessment are performed. Specifically, using the first family of resistive features as a benchmark, the similarity between it and every other resistive feature within the family of resistive features (excluding itself) is calculated. The similarity calculation can employ various metrics, such as standardized Euclidean distance or cosine similarity. The arithmetic mean of all calculated similarity values is then taken; this mean is the first frequency coefficient. Specifically, the value of this first frequency coefficient reflects the overall average closeness between the first family of resistive features and other features within the family of resistive features. A higher first frequency coefficient indicates that the first family of resistive features is more prevalent or representative within the family of resistive features.
[0039] Furthermore, the above calculation process is extended to each feature in the same family of resistive features. For each resistive feature in the same family of resistive features, the above operation is repeated: using itself as a reference, the similarity between it and all other features in the set is calculated and the mean is obtained, thereby calculating a corresponding frequency coefficient for each feature.
[0040] Finally, after calculating the frequency coefficients corresponding to all resistive features within the same family of resistive features, all frequency coefficients are compared, and the one with the largest value is selected. The specific resistive feature corresponding to this largest frequency coefficient is determined as the frequent resistive feature. This frequent resistive feature represents, from a statistical perspective, the feature vector with the highest average similarity to other features in the same family of resistive features, the most centered, and thus can be regarded as the most typical or common specification of this type of resistive. This frequent resistive feature will serve as the standard reference benchmark for subsequent calculation of non-standard resistance coefficients.
[0041] Furthermore, the characteristics of the current resistive element to be processed are compared with the previously obtained frequent resistive element characteristics, and the comprehensive deviation of the two in each characteristic dimension is quantitatively calculated. The calculated value is the resistance non-standard coefficient. Specifically, the resistance non-standard coefficient quantifies the degree of deviation of the current resistive element from the standard resistive element of the same family. The greater the deviation, the more significant the individual error introduced during the manufacturing process of the resistive element. This non-standard characteristic is usually more sensitive to the coating process and may amplify the change in resistance value after coating.
[0042] Finally, based on the calculated resistance non-standard coefficient, a corresponding non-standard compensation coefficient is set. Preferably, this non-standard compensation coefficient can be set to 1 minus the resistance non-standard coefficient, i.e., non-standard compensation coefficient = 1 - resistance non-standard coefficient. Then, this non-standard compensation coefficient is used to multiply and correct each predicted resistance change rate value in the first predicted resistance change rate distribution. In this way, all predicted resistance change rate values in the first predicted resistance change rate distribution are systematically scaled and adjusted, ultimately forming the first compensated predicted resistance change rate distribution. This compensation process aims to specifically calibrate the resistance change risk prediction given by the general prediction model according to the degree of non-standardization of the resistor itself, making the prediction results closer to the actual potential response of the specific resistor.
[0043] Furthermore, based on the first predicted protection parameter distribution, the first coating fitness and the first confidence level are calculated, including: The fitness of the first coating resistor is calculated based on the first compensation predicted resistance change rate distribution, wherein the first compensation predicted resistance change rate distribution is negatively correlated with the fitness of the first coating resistor. The first coating protection fitness is calculated based on the first predicted protection parameter distribution, wherein the first predicted protection parameter distribution and the first coating protection fitness are positively correlated. The first coating adaptability is calculated based on the first coating resistance adaptability and the first coating protection adaptability; A confidence analysis is performed on the distribution of the first compensation predicted resistance change rate and the distribution of the first predicted protection parameter to obtain the first confidence level.
[0044] First, the fitness of the first coated resistor is calculated. The first compensated predicted resistance change rate distribution contains multiple predicted resistance change rate values after non-standard compensation. According to the optimization objective, the smaller the resistance change rate value, the better the resistance stability. Therefore, the fitness of the first coated resistor is negatively correlated with the first compensated predicted resistance change rate distribution. Preferably, each predicted resistance change rate value in the first compensated predicted resistance change rate distribution can be transformed, for example, by subtracting the predicted resistance change rate value itself from 1, thus obtaining an intermediate value negatively correlated with the resistance change. The arithmetic mean of the transformed intermediate values is calculated, and this average value is defined as the fitness of the first coated resistor. The higher the fitness value of the first coated resistor, the better the overall expected performance of the resistor maintaining its resistance value stability under the current process parameter prediction.
[0045] Next, the first coating protection fitness is calculated. The first predicted protection parameter distribution contains multiple predicted protection parameter values, such as the corrosion rate reduction ratio. According to the optimization objective, a larger protection parameter value indicates a better protection effect. Therefore, the first coating protection fitness is positively correlated with the first predicted protection parameter distribution. Preferably, the arithmetic mean of all predicted protection parameter values in the first predicted protection parameter distribution can be directly calculated, and this mean is defined as the first coating protection fitness. The higher the first coating protection fitness, the stronger the overall expected corrosion protection effect provided by the coating layer under the current process parameter prediction.
[0046] Further, the first coating adaptability is calculated. Preferably, the first coating resistance adaptability and the first coating protection adaptability calculated above are combined using a preset weighting or fusion formula to obtain a single numerical index, namely the first coating adaptability. First coating adaptability = α First coating resistance adaptability +β The first coating protection adaptability score is defined as follows: α and β are weighting coefficients set according to the different priorities of resistance stability and protection performance in the product design. For example, when resistance stability is given higher importance, α can be set to 0.7 and β to 0.3. This first coating adaptability score aims to comprehensively evaluate the overall expected performance of the current coating process parameter combination in maintaining resistance stability and improving protection performance. A higher value indicates better overall performance of the initial coating process parameters.
[0047] Finally, a confidence analysis is performed to obtain a first confidence level, which aims to assess the reliability and consistency of the prediction results themselves.
[0048] Specifically, a confidence analysis is performed on the distribution of the first compensation predicted resistance change rate and the distribution of the first predicted protection parameter to obtain a first confidence level, including: Obtain the maximum and minimum values within the distribution of the first compensation prediction resistance change rate, and calculate the length of the compensation prediction resistance change rate interval. Obtain the maximum and minimum values within the distribution of the first predicted protection parameter, and calculate the length of the predicted protection parameter interval. Obtain the total interval length of the resistance change rate and the total interval length of the protection parameters. Combine the interval length of the compensation prediction resistance change rate and the interval length of the prediction protection parameters to calculate the first resistance dispersion coefficient and the first protection dispersion coefficient. Calculate the similarity and use it as the first confidence level.
[0049] First, from the first distribution of predicted resistance change rates, the maximum and minimum values of all predicted resistance change rates are extracted. The difference between the maximum and minimum values is the length of the predicted resistance change rate interval. This interval reflects the possible fluctuation range of the predicted resistance change under the current initial coating process parameters. Similarly, from the first distribution of predicted protection parameters, the maximum and minimum values of all predicted protection parameters are extracted, and their difference is calculated to obtain the length of the predicted protection parameter interval. This interval reflects the dispersion of the predicted protection effect.
[0050] Secondly, the total range length of the resistance change rate is obtained. Specifically, the total range length of the resistance change rate is the global theoretical boundary value of the relative change range of the resistance of this type of shunt resistor under all possible process conditions due to coating. It represents the physical or empirical limit of the fluctuation range of the resistance change rate index, and is obtained by summarizing and analyzing historical production records, extreme process test data, and derivation calculations based on material physical properties. Simultaneously, the total range length of the protection parameter is obtained. This total range length of the protection parameter represents the theoretical maximum and minimum possible values of the corrosion rate reduction ratio that the coating protective layer can achieve under standard corrosion environments, i.e., the theoretical limit range of the protection effectiveness. It is obtained by retrieving historical environmental test reports, extreme results of accelerated corrosion experiments, and analyzing coating material performance databases.
[0051] Furthermore, the ratio obtained by dividing the calculated predicted resistance change rate interval length by the total interval length of the resistance change rate is defined as the first resistance dispersion coefficient. This first resistance dispersion coefficient standardizes the dispersion of the current predicted distribution, and its value is between 0 and 1. A larger value indicates a higher uncertainty in the predicted resistance change. Similarly, the ratio obtained by dividing the predicted protection parameter interval length by the total interval length of the protection parameters is defined as the first protection dispersion coefficient. This first protection dispersion coefficient defines a standardized measure of the dispersion of the current predicted protection parameter distribution relative to its theoretical limit range. It represents the concentration and reliability level of the predicted coating protection effectiveness under given process parameters, and its value is also between 0 and 1. A higher value indicates a greater uncertainty in the predicted protection effect.
[0052] Finally, ideally, the effects of the same set of process parameters on resistance change and protection effect should have some inherent correlation, and the dispersion of their predictions should show a certain consistency. Therefore, the first confidence level can be obtained by calculating the similarity between the first resistance dispersion coefficient and the first protection dispersion coefficient. For example, the first confidence level = 1 - |first resistance dispersion coefficient - first protection dispersion coefficient|. The value of this first confidence level is between 0 and 1. The higher the value, the closer the relative dispersion of the first predicted resistance change rate distribution and the first predicted protection parameter distribution, the more consistent the shunt coating predictor's assessment of the dual effect of the initial coating process parameters, and the higher the reliability of the result; conversely, it indicates that the prediction may have a large deviation or uncertainty, and the reliability is low.
[0053] S40: Perform iterative optimization of the initial coating process parameters. During the optimization process, set the optimization strategy according to the confidence level until the optimization converges and the optimal coating process parameters are obtained. Then, coat the distributor.
[0054] Specifically, iterative optimization of the initial coating process parameters is performed. During the optimization process, an optimization strategy is set according to the confidence level until the optimization converges, obtaining the optimal coating process parameters, including: Within the coating process parameter space, the initial coating process parameters are adjusted to obtain new coating process parameters, and the coating adaptability and confidence level are analyzed. When the first preset number of optimizations is reached, the average confidence level during the optimization process is calculated to determine whether it is less than the preset confidence level. If not, optimization continues; if so, the confidence level deviation is calculated. Based on the confidence deviation range, the remaining number of optimization attempts is adjusted and increased, and optimization continues until the number of optimization attempts converges, thereby obtaining the optimal coating process parameters with the greatest coating adaptability. The remaining number of optimization attempts is obtained by calculating the difference between the converged optimization attempts and the first preset number of optimization attempts, and is determined and updated during optimization.
[0055] Finally, the initial coating process parameters are iteratively optimized, and the optimization strategy is dynamically set according to the confidence level during the optimization process until the optimization converges and the optimal coating process parameters are obtained.
[0056] Specifically, firstly, within a predefined coating process parameter space, the initial coating process parameters are adjusted to generate a new set of coating process parameters. This adjustment process can be driven by search algorithms such as gradient descent, genetic algorithms, or particle swarm optimization, achieved by randomly perturbing the current parameter point or purposefully moving it along the fitness improvement direction. For the generated new coating process parameters, the prediction, compensation, and evaluation process, identical to the previous steps, is repeated: the shunt coating predictor is invoked to obtain the predicted distribution, non-standard compensation and fitness calculations are performed, and the confidence level is analyzed to calculate the new coating fitness and confidence level corresponding to the new parameter set.
[0057] To systematically manage the optimization process, an overall iterative framework needs to be established, namely, setting a maximum permissible number of optimizations as the upper limit of the entire optimization process, such as 200 times. Simultaneously, an initial convergence optimization count is set as the basic optimization round target, such as 100 times. Furthermore, a periodic checkpoint is set, namely the first preset optimization count, for example, performing a policy evaluation every 10 optimizations.
[0058] When the accumulated number of optimizations reaches the first preset number of optimizations, for example, after completing 10 optimizations, the strategy adjustment judgment stage begins. At this time, the arithmetic mean of the confidence scores corresponding to all evaluated parameter groups is calculated since the last adjustment or since the start of optimization. This average confidence score is compared with a preset threshold, i.e., a preset confidence score. This preset confidence score can be set based on historical optimization experience, for example, 0.8. If the average confidence score is not less than the preset confidence score, it indicates that the overall reliability of the prediction results for the current search area is high, and the optimization process can continue as planned.
[0059] Furthermore, if the mean confidence level is less than the preset confidence level, it indicates that the prediction model has significant uncertainty within the current parameter space region. Continuing to blindly search may be inefficient or prone to getting trapped in local optima. In this case, it is necessary to calculate the confidence level deviation to quantify the degree of uncertainty. Specifically, the confidence level deviation can be obtained by calculating the standard deviation of the assessed confidence levels, or by subtracting the mean confidence level from the preset confidence level. Based on the calculated confidence level deviation, the remaining optimization iterations are adjusted and increased. For example, the larger the deviation, the more iterations are added. The specific increase can be determined by multiplying the deviation by a preset proportional coefficient. The remaining optimization iterations after the increase will serve as the new optimization target. The initial value of the remaining optimization iterations is the difference between the convergence optimization iterations and the first preset optimization iterations, and it is updated based on the increase after each strategy adjustment.
[0060] The optimization process continues thereafter, generating new coating process parameters and evaluating fitness and confidence levels. Whenever the cumulative number of optimizations reaches a multiple of a pre-set number (e.g., 20 or 30), the above strategy evaluation and adjustment process is repeated to determine whether the remaining optimization count needs to be adjusted based on the latest average confidence level. This process is repeated until the actual number of optimizations performed reaches the sum of the updated remaining optimization count and the number of optimizations already performed, or reaches the maximum permissible number of optimizations, at which point the optimization process converges.
[0061] Finally, among all the coating process parameter sets recorded throughout the optimization process, the set with the highest coating fitness value is selected as the final optimal coating process parameters. This mechanism ensures efficient searching in areas with high prediction confidence and increased exploration depth in areas with low confidence, thus balancing optimization efficiency and result reliability.
[0062] Finally, the shunt is coated according to the obtained optimal coating process parameters. The shunt achieves the best protection effect determined by the optimization algorithm while ensuring the stability of its resistive electrical performance, thus completing the closed-loop process of process parameter optimization and final implementation.
[0063] In summary, the embodiments of this application have at least the following technical effects: First, by constructing a collaborative prediction model, a comprehensive prediction of the distribution of resistance change rate and protection parameter distribution after coating is achieved, overcoming the limitations of existing technologies that only focus on a single performance index or conduct isolated evaluations. Second, a non-standard resistance coefficient is introduced and calculated, and based on this, targeted compensation is made to the general prediction results, improving the adaptability and prediction accuracy of the optimization method for non-standard or individually variably resistant materials, solving the problem that traditional fixed formulations or empirical parameters are difficult to adapt to personalized characteristics. Third, the confidence level of the prediction results is creatively proposed and calculated, and used as a dynamic feedback signal in the optimization process, enhancing the intelligence, robustness, and search efficiency of the entire optimization process, effectively avoiding getting trapped in local optima or premature convergence in low-confidence regions.
[0064] Ultimately, through the aforementioned closed-loop optimization process, the optimal combination of process parameters for achieving the best overall coating adaptability can be determined for a specific shunt resistor element with a high degree of confidence. This resolves the core contradiction in the protective layer coating process, which makes it difficult to balance protective performance and resistance stability, thereby improving the long-term reliability, performance consistency, and production quality of shunt products.
[0065] Example 2, as Figure 2As shown, based on the same inventive concept as the method for optimizing the coating process parameters of a shunt protector layer provided in Embodiment 1, this embodiment of the invention also provides a system for optimizing the coating process parameters of a shunt protector layer, comprising: The parameter acquisition module 11 is used to acquire the resistive characteristics of the resistive element in the shunt and to acquire the initial coating process parameters for coating. The coating prediction module 12 is used to predict the shunt coating based on the initial coating process parameters and resistor characteristics, and to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution. The compensation and evaluation module 13 is used to analyze and obtain the non-standard resistance coefficient based on the characteristics of the resistive body, perform non-standard compensation on the first predicted resistance change rate distribution, obtain the first compensated predicted resistance change rate distribution, and calculate the first coating adaptability and the first confidence level in combination with the first predicted protection parameter distribution. The optimization processing module 14 is used to iteratively optimize the initial coating process parameters. During the optimization process, the optimization strategy is set according to the confidence level until the optimization converges, and the optimal coating process parameters are obtained to coat the splitter.
[0066] Specifically, the parameter acquisition module 11 is used for: Obtain the resistive characteristics of the resistor element within the shunt and acquire the initial coating process parameters for coating, including: Obtain the resistive characteristics of the resistor element inside the shunt, wherein the resistive characteristics include the resistor element specification parameters; Obtain the coating process parameter space for the distributor protective layer coating, and obtain the initial coating process parameters set within the coating process parameter space, wherein each coating process parameter includes coating position, coating liquid composition, coating amount and coating temperature.
[0067] Specifically, the coating prediction module 12 is used for: Based on the initial coating process parameters and resistor characteristics, shunt coating prediction is performed to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution, including: Call the splitter to coat the predictor; The initial coating process parameters and resistor characteristics are input into the shunt coating predictor, which outputs multiple first predicted resistance change rates and multiple first predicted protection parameters to obtain the distribution of the first predicted resistance change rate and the distribution of the first predicted protection parameters.
[0068] Specifically, the shunt coating predictor is pre-trained, and the training steps include: Based on the coating test data records of the shunt, collect the sample coating process parameter set and the sample resistivity characteristic set; Multiple resistance change amplitudes were collected for each group of samples under the coating process parameters and after coating of the resistive body characteristics. These were used as resistance change rate groups, and the sample resistance change rate group set was obtained by labeling. The corrosion protection parameters of the shunt were collected under the coating process parameters and after coating of the resistive body characteristics of each sample. These parameters were used as protection parameter groups. The sample protection parameter group set was obtained by labeling. Each protection parameter includes the proportion of the shunt corrosion rate reduction under the same environment. Based on machine learning, a coating resistance change prediction branch and a coating protection prediction branch are constructed; Using the sample coating process parameter set and sample resistivity feature set as inputs, and the sample resistance change rate set and sample protection parameter set as labels, respectively, supervised training tests are performed on the coating resistance change prediction branch and the coating protection prediction branch until the test converges.
[0069] The compensation and evaluation module 13 is specifically used for: Based on the characteristics of the resistive element, non-standard resistance coefficients are analyzed and obtained. Non-standard compensation is then applied to the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution, including: Obtain the characteristic set of resistor elements of the same family for shunts of the same type; Frequent mining of resistive features is performed within the same set of resistive features to obtain frequent resistive features. The deviation amplitude between the resistive element characteristics and the frequent resistive element characteristics is calculated and used as the resistance non-standard coefficient; Based on the non-standard resistance coefficient, a non-standard compensation coefficient is set, and a non-standard compensation calculation is performed on the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution.
[0070] Specifically, frequent mining of resistive features is performed within the same set of resistive features to obtain frequent resistive features, including: Randomly select a first family of resistor features from the family of resistor features; Calculate the similarity between the features of the first family of resistive elements and the features of other family of resistive elements, and calculate the mean to obtain the first frequency coefficient; Continue calculating the frequency coefficients of other resistor features in the same family, and select the resistor features in the same family that have the maximum frequency coefficient as frequent resistor features.
[0071] Furthermore, based on the first predicted protection parameter distribution, the first coating fitness and the first confidence level are calculated, including: The fitness of the first coating resistor is calculated based on the first compensation predicted resistance change rate distribution, wherein the first compensation predicted resistance change rate distribution is negatively correlated with the fitness of the first coating resistor. The first coating protection fitness is calculated based on the first predicted protection parameter distribution, wherein the first predicted protection parameter distribution and the first coating protection fitness are positively correlated. The first coating adaptability is calculated based on the first coating resistance adaptability and the first coating protection adaptability; A confidence analysis is performed on the distribution of the first compensation predicted resistance change rate and the distribution of the first predicted protection parameter to obtain the first confidence level.
[0072] Specifically, a confidence analysis is performed on the distribution of the first compensation predicted resistance change rate and the distribution of the first predicted protection parameter to obtain a first confidence level, including: Obtain the maximum and minimum values within the distribution of the first compensation prediction resistance change rate, and calculate the length of the compensation prediction resistance change rate interval. Obtain the maximum and minimum values within the distribution of the first predicted protection parameter, and calculate the length of the predicted protection parameter interval. Obtain the total interval length of the resistance change rate and the total interval length of the protection parameters. Combine the interval length of the compensation prediction resistance change rate and the interval length of the prediction protection parameters to calculate the first resistance dispersion coefficient and the first protection dispersion coefficient. Calculate the similarity and use it as the first confidence level.
[0073] The optimization processing module 14 is specifically used for: Iterative optimization of the initial coating process parameters is performed. During the optimization process, an optimization strategy is set according to the confidence level until the optimization converges, obtaining the optimal coating process parameters, including: Within the coating process parameter space, the initial coating process parameters are adjusted to obtain new coating process parameters, and the coating adaptability and confidence level are analyzed. When the first preset number of optimizations is reached, the average confidence level during the optimization process is calculated to determine whether it is less than the preset confidence level. If not, optimization continues; if so, the confidence level deviation is calculated. Based on the confidence deviation range, the remaining number of optimization attempts is adjusted and increased, and optimization continues until the number of optimization attempts converges, thereby obtaining the optimal coating process parameters with the greatest coating adaptability. The remaining number of optimization attempts is obtained by calculating the difference between the converged optimization attempts and the first preset number of optimization attempts, and is determined and updated during optimization.
[0074] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.
[0075] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
[0076] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.
Claims
1. A method for optimizing the coating process parameters of a shunt protector layer, characterized in that, The method includes: Obtain the resistive characteristics of the resistor element inside the shunt and obtain the initial coating process parameters for coating. Based on the initial coating process parameters and resistor characteristics, shunt coating prediction is performed to obtain the first predicted resistance change rate distribution and the first predicted protection parameter distribution. Based on the characteristics of the resistive element, the non-standard resistance coefficient is analyzed and obtained. Non-standard compensation is performed on the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution. Combined with the first predicted protection parameter distribution, the first coating adaptability and the first confidence level are calculated. The initial coating process parameters are iteratively optimized. During the optimization process, the optimization strategy is set according to the confidence level until the optimization converges and the optimal coating process parameters are obtained. The distributor is then coated.
2. The method for optimizing the coating process parameters of the shunt protector layer according to claim 1, characterized in that, Obtain the resistive characteristics of the resistor element within the shunt and acquire the initial coating process parameters for coating, including: Obtain the resistive characteristics of the resistor element inside the shunt, wherein the resistive characteristics include the resistor element specification parameters; Obtain the coating process parameter space for the distributor protective layer coating, and obtain the initial coating process parameters set within the coating process parameter space, wherein each coating process parameter includes coating position, coating liquid composition, coating amount and coating temperature.
3. The method for optimizing the coating process parameters of the shunt protector layer according to claim 1, characterized in that, Based on the initial coating process parameters and resistor characteristics, shunt coating prediction is performed to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution, including: Call the splitter to coat the predictor; The initial coating process parameters and resistor characteristics are input into the shunt coating predictor, which outputs multiple first predicted resistance change rates and multiple first predicted protection parameters to obtain the distribution of the first predicted resistance change rate and the distribution of the first predicted protection parameters.
4. The method for optimizing the coating process parameters of the shunt protector layer according to claim 3, characterized in that, The shunt coating predictor is pre-trained, and the training steps include: Based on the coating test data records of the shunt, collect the sample coating process parameter set and the sample resistivity characteristic set; Multiple resistance change amplitudes were collected for each group of samples under the coating process parameters and after coating of the resistive body characteristics. These were used as resistance change rate groups, and the sample resistance change rate group set was obtained by labeling. The corrosion protection parameters of the shunt were collected under the coating process parameters and after coating of the resistive body characteristics of each sample. These parameters were used as protection parameter groups. The sample protection parameter group set was obtained by labeling. Each protection parameter includes the proportion of the shunt corrosion rate reduction under the same environment. Based on machine learning, a coating resistance change prediction branch and a coating protection prediction branch are constructed; Using the sample coating process parameter set and sample resistivity feature set as inputs, and the sample resistance change rate set and sample protection parameter set as labels, respectively, supervised training tests are performed on the coating resistance change prediction branch and the coating protection prediction branch until the test converges.
5. The method for optimizing the coating process parameters of the shunt protector layer according to claim 1, characterized in that, Based on the characteristics of the resistive element, non-standard resistance coefficients are analyzed and obtained. Non-standard compensation is then applied to the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution, including: Obtain the characteristic set of resistor elements of the same family for shunts of the same type; Frequent mining of resistive features is performed within the same set of resistive features to obtain frequent resistive features. The deviation amplitude between the resistive element characteristics and the frequent resistive element characteristics is calculated and used as the resistance non-standard coefficient; Based on the non-standard resistance coefficient, a non-standard compensation coefficient is set, and a non-standard compensation calculation is performed on the first predicted resistance change rate distribution to obtain the first compensated predicted resistance change rate distribution.
6. The method for optimizing the coating process parameters of the shunt protector layer according to claim 5, characterized in that, Within the same set of resistive element features, frequent mining of resistive element features is performed to obtain frequent resistive element features, including: Randomly select a first family of resistor features from the family of resistor features; Calculate the similarity between the features of the first family of resistive elements and the features of other family of resistive elements, and calculate the mean to obtain the first frequency coefficient; Continue calculating the frequency coefficients of other resistor features in the same family, and select the resistor features in the same family that have the maximum frequency coefficient as frequent resistor features.
7. The method for optimizing the coating process parameters of the shunt protector layer according to claim 1, characterized in that, Based on the first predicted protection parameter distribution, the first coating fitness and the first confidence level are calculated, including: The fitness of the first coating resistor is calculated based on the first compensation predicted resistance change rate distribution, wherein the first compensation predicted resistance change rate distribution is negatively correlated with the fitness of the first coating resistor. The first coating protection fitness is calculated based on the first predicted protection parameter distribution, wherein the first predicted protection parameter distribution and the first coating protection fitness are positively correlated. The first coating adaptability is calculated based on the first coating resistance adaptability and the first coating protection adaptability; A confidence analysis is performed on the distribution of the first compensation predicted resistance change rate and the distribution of the first predicted protection parameter to obtain the first confidence level.
8. The method for optimizing the coating process parameters of the shunt protector layer according to claim 7, characterized in that, A confidence analysis is performed on the distribution of the first predicted rate of change of compensation resistance and the distribution of the first predicted protection parameters to obtain a first confidence level, including: Obtain the maximum and minimum values within the distribution of the first compensation prediction resistance change rate, and calculate the length of the compensation prediction resistance change rate interval. Obtain the maximum and minimum values within the distribution of the first predicted protection parameter, and calculate the length of the predicted protection parameter interval. Obtain the total interval length of the resistance change rate and the total interval length of the protection parameters. Combine the interval length of the compensation prediction resistance change rate and the interval length of the prediction protection parameters to calculate the first resistance dispersion coefficient and the first protection dispersion coefficient. Calculate the similarity and use it as the first confidence level.
9. The method for optimizing the coating process parameters of the shunt protector layer according to claim 1, characterized in that, Iterative optimization of the initial coating process parameters is performed. During the optimization process, an optimization strategy is set according to the confidence level until the optimization converges, obtaining the optimal coating process parameters, including: Within the coating process parameter space, the initial coating process parameters are adjusted to obtain new coating process parameters, and the coating adaptability and confidence level are analyzed. When the first preset number of optimizations is reached, the average confidence level during the optimization process is calculated to determine whether it is less than the preset confidence level. If not, optimization continues; if so, the confidence level deviation is calculated. Based on the confidence deviation range, the remaining number of optimization attempts is adjusted and increased, and optimization continues until the number of optimization attempts converges, thereby obtaining the optimal coating process parameters with the greatest coating adaptability. The remaining number of optimization attempts is obtained by calculating the difference between the converged optimization attempts and the first preset number of optimization attempts, and is determined and updated during optimization.
10. A system for optimizing the coating process parameters of a shunt protector layer, characterized in that, A method for optimizing the coating process parameters of a shunt protector layer according to any one of claims 1-9 includes: The parameter acquisition module is used to acquire the resistive characteristics of the resistor element inside the shunt and to acquire the initial coating process parameters for coating. The coating prediction module is used to predict the shunt coating based on the initial coating process parameters and resistor characteristics, and to obtain a first predicted resistance change rate distribution and a first predicted protection parameter distribution. The compensation and evaluation module is used to analyze and obtain the non-standard resistance coefficient based on the characteristics of the resistive body, perform non-standard compensation on the first predicted resistance change rate distribution, obtain the first compensated predicted resistance change rate distribution, and calculate the first coating adaptability and the first confidence level in combination with the first predicted protection parameter distribution. The optimization processing module is used to iteratively optimize the initial coating process parameters. During the optimization process, the optimization strategy is set according to the confidence level until the optimization converges, and the optimal coating process parameters are obtained to coat the splitter.