Carbon-based conductive new materials industry brain management cloud platform

Abstract

The present invention relates to the field of industrial management technology, and specifically, to a brain management cloud platform for the carbon-based conductive new material industry. It includes: a data acquisition unit that collects production parameters by connecting to equipment for producing carbon-based conductive materials, and uploads the parameters to a cloud platform; a data storage management unit that processes and stores different types of data through distributed storage; a data analysis unit that uses a support vector machine model to distinguish the quality of produced carbon-based conductive materials, estimates the output based on a ridge regression model, and uses a particle swarm optimization algorithm to improve production efficiency, and introduces the influence function of porosity in the process of support vector machine model distinction. The present invention is designed to introduce a support vector machine model and optimize the kernel function for the characteristics of carbon-based conductive materials, especially considering the effects of porosity and surface gas adsorption on material conductivity, so that the platform can more accurately distinguish the quality of produced materials.

Description

Technical Field

[0001] The present invention relates to the field of industrial management technology, and in particular to a brain management cloud platform for the carbon-based conductive new material industry. Background Art

[0002] In the traditional production process of carbon-based conductive materials, there are a lot of problems such as insufficient data collection and in-depth analysis, which leads to the lack of scientific basis for the adjustment of production process parameters and difficulty in achieving high-efficiency, high-quality and stable production; in the production of carbon-based conductive materials, due to the complex material properties (such as porosity, surface gas adsorption, etc.), traditional methods are difficult to accurately distinguish product quality and cannot effectively predict material performance, which poses a challenge to the product's pass rate and consistency; there may be correlations between production conditions such as temperature, pressure, and humidity, and this correlation may lead to instability in ordinary least squares estimation, thereby affecting the accurate prediction of future production. This not only increases the difficulty of inventory management, but also makes it difficult for companies to flexibly adjust production plans according to market demand. Therefore, a brain management cloud platform for carbon-based conductive new materials industry is designed. Summary of the invention

[0003] The purpose of the present invention is to provide a brain management cloud platform for the carbon-based conductive new material industry to solve the problems of difficult quality control and classification of carbon-based conductive new materials and inaccurate output prediction raised in the above background technology.

[0004] To achieve the above objectives, the present invention aims to provide a brain management cloud platform for the carbon-based conductive new material industry, including:

[0005] A data acquisition unit, which collects production parameters by connecting to a device for producing carbon-based conductive materials and uploads the parameters to a cloud platform;

[0006] A data storage management unit, wherein the data storage management unit processes and stores different types of data through distributed storage;

[0007] A data analysis unit, wherein the data analysis unit uses a support vector machine model to distinguish the quality of produced carbon-based conductive materials, estimates the output based on a ridge regression model, uses a particle swarm optimization algorithm to improve production efficiency, and introduces an influence function of porosity in the process of support vector machine model distinction.

[0008] As a further improvement of this technical solution, the production parameters include pressure ,temperature and humidity .

[0009] As a further improvement of the technical solution, the data storage management unit includes a data storage module and a data management module;

[0010] Wherein, the data storage module collects the production parameters through the distributed storage data acquisition unit;

[0011] The data management module creates an index for each type of stored production parameter data and manages the production parameter data through the index.

[0012] As a further improvement of the technical solution, the data analysis unit includes an identification and classification module, a production prediction module, and an optimization management module;

[0013] Wherein, the identification and classification module classifies the properties of carbon-based conductive materials based on a support vector machine model;

[0014] The production prediction module establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material performance based on the results of identification and classification and the ridge regression model to predict future production;

[0015] The optimization management module uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction results.

[0016] As a further improvement of the technical solution, the identification and classification module classifies the characteristics of the carbon-based conductive material based on a support vector machine model, comprising the following steps:

[0017] S1.1、Press ,temperature and humidity Data as input features : , set the classification label of carbon-based conductive materials ;

[0018] S1.2. Pressure ,temperature and humidity Preprocess the data;

[0019] S1.3, the pressure ,temperature and humidity The data is divided into training set and test set;

[0020] S1.4, select the kernel function and adjust the key parameters of the support vector machine;

[0021] S1.5. Use the kernel function and training set data to train the support vector machine model, optimize the kernel function in view of the influence of porosity on the conductivity of the carbon-based conductive new material, and further optimize the kernel function in view of the influence of surface gas adsorption on the carbon-based conductive new material;

[0022] S1.6. After training, the support vector machine model is used to learn how to classify carbon-based conductive materials according to input features and output quality classification labels.

[0023] As a further improvement of the technical solution, in S1.5, the kernel function is:

[0024] ;

[0025] in, Represents input features; The feature vector representing the test sample; Represents the parameter that controls the width of the kernel function; Represents the similarity measure between the input features and the feature vector of the test sample;

[0026] In view of the fact that the conductivity of new carbon-based conductive materials is affected by porosity, the kernel function is optimized:

[0027] ;

[0028] in, Represents the similarity measure between the optimized input features and the feature vector of the test sample; represents the influence function of porosity; Represents input features The porosity of the sample; Indicates the porosity of the test sample;

[0029] In view of the influence of the surface gas adsorption on the new carbon-based conductive materials, the kernel function is further optimized:

[0030] ;

[0031] in, represents the similarity measure between the input features and the feature vector of the test sample after further optimization; represents the gas adsorption amount function; Indicates the amount of gas adsorption; Indicates the ambient temperature.

[0032] As a further improvement of the technical solution, in S1.6, the support vector machine model is:

[0033] ;

[0034] in, Indicates the quality classification label; The feature vector representing the test sample; represents the Lagrange multiplier; Represents the label of the training sample.

[0035] As a further improvement of the technical solution, the production prediction module establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material performance based on a ridge regression model to predict future production, including the following steps:

[0036] S2.1. Collect carbon-based conductive materials at different pressures ,temperature and humidity The performance data under the conditions will be pre-processed;

[0037] S2.2. Select ridge regression as the prediction model, initialize the ridge regression model, and set the hyperparameters;

[0038] S2.3. Convert the quality classification labels into binary features through one-hot encoding and convert the pressure ,temperature ,humidity and quality classification labels as input features ;

[0039] S2.4. Construct a loss function, optimize the loss function according to the fluctuation of the quality of the new carbon-based conductive material, and solve the regression coefficient through ridge regression To minimize the loss function:

[0040] ;

[0041] in, represents the input feature moment, is the output yield, is the regularization parameter, is the identity matrix;

[0042] S2.5, evaluate the trained ridge regression model by mean square error;

[0043] S2.6. After the evaluation is completed, the trained ridge regression model is used to predict the yield:

[0044] ;

[0045] in, represents the number of categories of quality classification labels, represents the one-hot encoded dimensional eigenvalue, represents the predicted output, Indicates new pressure, represents the new temperature, Indicates the new humidity.

[0046] As a further improvement of the technical solution, in S2.4, the loss function is:

[0047] ;

[0048] in, represents the objective function; represents the regression coefficient; Indicates actual output; represents the number of samples; represents the regularization parameter; Represents the model regression coefficients; represents the regression coefficient index; Indicates the sample index;

[0049] Aiming at the fluctuation of the quality of new carbon-based conductive materials, the loss function is optimized:

[0050] ;

[0051] in, Represents the optimized loss function; The constraint item representing the material quality; Represents the weight coefficient of the quality constraint.

[0052] As a further improvement of the technical solution, the optimization management module uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction result, including the following steps:

[0053] S3.1. Initialize the particle swarm. The initial position of each particle represents the process parameter combination. The position of each particle is a three-dimensional vector: , Represents particles The corresponding pressure, Represents particles The corresponding temperature, Represents particles The corresponding humidity;

[0054] S3.2. The initial velocity of each particle is a vector, which indicates the moving speed of the particle in each dimension: ,in, Represents particles The speed corresponding to the temperature, Represents particles The speed corresponding to the pressure, Represents particles Speed ​​corresponding to humidity;

[0055] S3.3. For each particle, calculate its fitness value : ;

[0056] S3.4. Update the speed and position of each particle:

[0057] S3.5. Each particle checks the current fitness value Is it better than its historical best value? If so, update the individual best position;

[0058] S3.6, update the global optimal position by comparing the fitness values ​​of all particles;

[0059] S3.7. After reaching the maximum number of iterations, the particle swarm optimization algorithm will converge to the optimal solution, which will maximize the output.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1. In the carbon-based conductive new material industry brain management cloud platform, by introducing the support vector machine model and optimizing the kernel function for the characteristics of carbon-based conductive materials, especially considering the influence of porosity and surface gas adsorption on material conductivity, the platform can more accurately distinguish the quality of production materials. This optimization not only enhances the model's ability to handle nonlinear relationships, but also improves the accuracy of classification and prediction. This helps to ensure the consistency and stability of product quality, reduce the defective rate, and thus improve overall production efficiency and market competitiveness.

[0062] 2. The brain management cloud platform for the carbon-based conductive new material industry uses a particle swarm optimization algorithm combined with a ridge regression model. The platform can dynamically adjust and optimize the production process while taking into account the complex relationships between production parameters, so as to maximize output and effectively utilize resources. At the same time, by optimizing the loss function and adding quality fluctuation constraints, it can ensure that each batch of materials produced has consistent performance in key performance indicators, thereby enhancing product reliability and predictability. This enables companies to quickly respond to market changes or changes in production conditions, adjust production strategies in a timely manner, reduce production costs, and improve economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is the overall flow chart of the present invention;

[0064] The meaning of each number in the figure is:

[0065] 1. Data acquisition unit; 2. Data storage management unit; 21. Data storage module; 22. Data management module; 3. Data analysis unit; 31. Identification and classification module; 32. Production forecasting module; 33. Optimization management module. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] See also Figure 1 As shown, it provides a brain management cloud platform for the carbon-based conductive new material industry, including:

[0068] The data acquisition unit 1 collects production parameters by connecting to the equipment for producing carbon-based conductive materials, and uploads the parameters to the cloud platform;

[0069] In this embodiment, the production parameters include pressure ,temperature and humidity .

[0070] The data storage management unit 2 processes and stores different types of data through distributed storage;

[0071] In this embodiment, the data storage management unit 2 includes a data storage module 21 and a data management module 22;

[0072] The data storage module 21 stores the production parameters collected by the data acquisition unit 1 in a distributed manner, and stores the production parameters collected by the data acquisition unit 1 in multiple nodes through the principle of distributed storage, so as to achieve efficient data management, access and reliability;

[0073] The data management module 22 establishes an index for each type of stored production parameter data (such as temperature, humidity, pressure, etc.) and manages the production parameter data through the index to facilitate efficient retrieval on the cloud platform. The data management module 22 establishes an index for each type of production parameter data and maps the data to a specific storage location to accelerate retrieval and management efficiency, similar to the way a book catalog locates content.

[0074] Data analysis unit 3 uses a support vector machine model to distinguish the quality of produced carbon-based conductive materials, estimates the output based on a ridge regression model, and uses a particle swarm optimization algorithm to improve production efficiency;

[0075] In this embodiment, the data analysis unit 3 includes an identification and classification module 31, a production prediction module 32, and an optimization management module 33;

[0076] Wherein, the identification and classification module 31 classifies the characteristics of the carbon-based conductive material based on the support vector machine model, for example, distinguishing between qualified products and defective products;

[0077] The production prediction module 32 establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material properties based on the results of identification and classification and the ridge regression model to predict future production;

[0078] The optimization management module 33 uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction results;

[0079] Among them, support vector machine (SVM) is a supervised learning model used for classification and regression analysis. It maximizes the interval between different categories by finding the optimal hyperplane. The support vector machine model (SVM) is good at processing high-dimensional data, which is especially important for carbon-based conductive materials with complex properties. Even if the production parameters and material properties involve multiple variables, SVM can effectively classify; by using kernel functions, support vector machines can achieve nonlinear classification boundaries in feature space. For carbon-based conductive materials, their performance is affected by many factors, such as temperature, pressure, humidity, and porosity, and the relationship between these factors may be nonlinear. The kernel method of SVM allows the model to capture this complex nonlinear relationship, thereby more accurately classifying the quality of the material; VM constructs a decision boundary by maximizing the edge, which means that it tries to find a boundary that can best distinguish different categories while being farthest away from all categories. This feature makes SVM more robust to noise and outliers, which helps to improve the reliability of classification;

[0080] The identification and classification module 31 classifies the characteristics of the carbon-based conductive material based on the support vector machine model, including the following steps:

[0081] S1.1、Press ,temperature and humidity Data as input features : , set the classification label of carbon-based conductive materials (e.g., good and bad products);

[0082] S1.2. Pressure ,temperature and humidity Data preprocessing, including missing value processing and normalization;

[0083] S1.3, the pressure ,temperature and humidity The data is divided into training set and test set;

[0084] S1.4. Select the kernel function and adjust the key parameters of the support vector machine, such as C (penalty coefficient) and (kernel function parameters);

[0085] S1.5. Use the kernel function and training set data to train the support vector machine model, optimize the kernel function in view of the influence of porosity on the conductivity of the carbon-based conductive new material, and further optimize the kernel function in view of the influence of surface gas adsorption on the carbon-based conductive new material;

[0086] Furthermore, the kernel function is:

[0087] ;

[0088] in, Represents input features; The feature vector representing the test sample; Represents the parameter that controls the width of the kernel function; It represents the similarity measure between the input feature and the feature vector of the test sample. It is the kernel function in the support vector machine and is used to calculate the similarity between two samples in the feature space.

[0089] In view of the fact that the conductivity of new carbon-based conductive materials is affected by porosity, the kernel function is optimized:

[0090] Porosity is a measure of the proportion of pores or voids in a material to the total volume. For new carbon-based conductive materials, porosity usually directly affects the conductivity. Too high a porosity will lead to changes in the conductive channels of the material, which may lead to a decrease in conductivity, because more voids mean a more complex current path and difficulty in charge transfer. The traditional support vector machine (SVM) kernel function only measures the similarity between samples based on the Euclidean distance of the input features, and does not consider the important influence of porosity, a specific physical parameter, on conductivity. Therefore, porosity is incorporated into the similarity calculation formula as an influencing factor to make the kernel function more in line with the material properties. By introducing the porosity influence function, the kernel function can more accurately distinguish between high-conductivity and low-conductivity material samples. The optimized kernel function can better reflect the complex relationship between conductivity and porosity, and improve the accuracy of classification and prediction.

[0091] ;

[0092] ;

[0093] in, Represents the similarity measure between the optimized input features and the feature vector of the test sample; represents the influence function of porosity; Represents input features The porosity of the sample; Indicates the porosity of the test sample;

[0094] In view of the influence of the surface gas adsorption on the new carbon-based conductive materials, the kernel function is further optimized:

[0095] The porous structure of carbon-based materials usually has a high specific surface area and can adsorb gas molecules in the surrounding environment. The adsorbed gas molecules may change the electron transfer path or surface charge distribution of the material, thereby affecting the conductivity; different types or quantities of adsorbed gas molecules may affect the conductivity in the following ways: changing the charge transfer on the surface of the material, forming a path that hinders the movement of electrons in the porous structure, and changing the chemical reactivity of the surface, thereby affecting the conduction behavior of electrons; the effect of gas adsorption on conductivity is usually nonlinear, so the optimized kernel function can better handle this complex relationship and avoid the limitations of traditional kernel functions;

[0096] ;

[0097] ;

[0098] in, represents the similarity measure between the input features and the feature vector of the test sample after further optimization; represents the gas adsorption amount function; Indicates the amount of gas adsorption, which indicates the number of adsorbed molecules per unit volume or unit surface, usually expressed as the number of adsorbed molecules or weight per unit area, reflecting the amount of gas adsorbed on the surface of the material; Indicates the ambient temperature; Indicates the maximum operating temperature of the material under the current experimental conditions, which is used to normalize the temperature parameters; represents the weight of the adsorption amount; Indicates the weight of the ambient temperature;

[0099] S1.6. After training, the support vector machine model is used to learn how to classify carbon-based conductive materials according to input features and output quality classification labels;

[0100] Furthermore, the support vector machine model is:

[0101] ;

[0102] in, Indicates the quality classification label; The feature vector representing the test sample represents the new sample data that needs to be predicted, including input parameters such as temperature, pressure, humidity, etc. Represents the Lagrange multiplier, which is the coefficient obtained during the support vector machine training process and reflects the influence of the training sample on the decision hyperplane (or regression plane); Represents the label of the training sample.

[0103] Among them, the ridge regression model is a statistical model that adds a regularization term (L2 norm penalty) to linear regression to prevent overfitting and improve the stability of prediction; ridge regression can effectively deal with the problem of multicollinearity between input variables by introducing regularization terms (i.e., penalty terms). In the production process of carbon-based conductive materials, there may be correlations between parameters such as temperature, pressure, and humidity, and this correlation may cause instability in ordinary least squares estimation. Ridge regression reduces this instability by shrinking coefficients, thereby providing more reliable predictions; since ridge regression adds regularization terms, it can limit model complexity to a certain extent and avoid overfitting of the model to training data. This helps to improve the generalization ability of the model on unseen data and ensure that the prediction of future output is more accurate and reliable; ridge regression can automatically evaluate the importance of different features and reflect their impact on the final output by adjusting the coefficients corresponding to each feature. For carbon-based conductive materials, this means that the model can learn which production conditions (such as temperature, pressure, and humidity) are most critical to output and performance based on historical data, thereby providing guidance for optimizing the production process;

[0104] The production prediction module 32 establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material performance based on the ridge regression model to predict future production, including the following steps:

[0105] S2.1. Collect carbon-based conductive materials at different pressures ,temperature and humidity The performance data under the conditions will be preprocessed, including outlier detection, normalization, etc.

[0106] S2.2. Select ridge regression as the prediction model, initialize the ridge regression model, and set hyperparameters (such as regularization strength );

[0107] S2.3. Convert the quality classification labels into binary features through one-hot encoding and convert the pressure ,temperature ,humidity and quality classification labels as input features ;

[0108] S2.4. Construct a loss function, optimize the loss function according to the fluctuation of the quality of the new carbon-based conductive material, and solve the regression coefficient through ridge regression To minimize the loss function:

[0109] ;

[0110] in, represents the input feature matrix (temperature, pressure, humidity), is the output yield, is the regularization parameter, is the identity matrix;

[0111] Furthermore, the loss function is:

[0112] ;

[0113] in, represents the objective function, which is a mathematical expression that measures the difference between the model's predictions and the actual data; represents the regression coefficient, which indicates the weight of each input variable; Indicates actual output; represents the number of samples; Represents the regularization parameter, which is used to control the strength of the regularization term and determine the degree of penalty for model complexity; Represents the model The regression coefficients of the model are variables used in machine learning models to map input features to predicted outputs. Each parameter corresponds to the weight of an input feature. These parameters are adjusted through optimization during the training process to minimize the loss function, thereby improving the model's predictive ability; represents the regression coefficient index; Indicates the sample index;

[0114] Aiming at the fluctuation of the quality of new carbon-based conductive materials, the loss function is optimized:

[0115] The quality of new carbon-based conductive materials may fluctuate during the production process due to a variety of factors (such as temperature, humidity, pressure, raw material ratio, production equipment status, etc.). These fluctuations will cause differences in conductivity, strength, stability, etc. between different batches of materials, affecting the quality consistency of the final product; by optimizing the loss function and introducing quality fluctuation constraints, it can ensure that each batch of materials produced has consistent quality performance in key performance indicators (such as conductivity, compressive strength, etc.), thereby enhancing product reliability and predictability. Optimizing the loss function can force the model to consider the stability of material quality during training and reduce differences between batches; by optimizing the loss function and adding quality control items, the model can automatically learn the laws of the production process, and then dynamically adjust production parameters to control fluctuations in the production process, thereby reducing the risks brought by unstable product quality. The optimized loss function can help the platform adjust the production process in real time and reduce uncertainty in the production process by forcing consideration of quality stability constraints;

[0116] ;

[0117] ;

[0118] in, Represents the optimized loss function; The constraint item representing the material quality; represents the weight coefficient of quality constraint; The number of dimensions representing conductivity; Indicates Observed values ​​of dimensional conductivity; Indicates The variance of the dimensional conductivity is used to measure the volatility of this indicator;

[0119] S2.5, evaluate the trained ridge regression model by mean square error;

[0120] S2.6. After the evaluation is completed, the trained ridge regression model is used to predict the yield:

[0121] ;

[0122] in, represents the number of categories of quality classification labels, represents the one-hot encoded dimensional eigenvalue, represents the predicted output, Indicates new pressure, represents the new temperature, Indicates the new humidity.

[0123] Among them, the particle swarm optimization algorithm (PSO) is a heuristic search method that imitates the group behavior of bird flocks or fish schools to find the optimal solution. This algorithm has strong global search capabilities and can find the global optimal solution or a solution close to the global optimal solution in a complex multi-dimensional space. It is suitable for the optimization of multiple process parameters involved in the production process of carbon-based conductive materials; the PSO algorithm can usually converge to a satisfactory solution quickly. For industrial processes that need to be adjusted in time to respond to market changes or changes in production conditions, it is very important to quickly obtain an optimization solution. Through rapid iteration, PSO can quickly provide a set of optimized process parameter combinations, thereby speeding up production adjustments; each particle in particle swarm optimization can be evaluated and updated independently, which is very suitable for parallel computing environments. On modern high-performance computing platforms, parallel processing can significantly reduce computing time and further improve optimization efficiency;

[0124] The optimization management module 33 uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction results, including the following steps:

[0125] S3.1. Initialize the particle swarm. The initial position of each particle represents the process parameter combination. The position of each particle is a three-dimensional vector: , Represents particles The corresponding pressure, Represents particles The corresponding temperature, Represents particles The corresponding humidity;

[0126] S3.2. The initial velocity of each particle is a vector, which indicates the moving speed of the particle in each dimension: ,in, Represents particles The speed corresponding to the temperature, Represents particles The speed corresponding to the pressure, Represents particles Speed ​​corresponding to humidity;

[0127] S3.3. For each particle, calculate its fitness value , that is, the objective function value (yield prediction): ,This fitness value reflects the quality of the current particle's corresponding process parameters, and the particle will adjust its position according to the fitness value;

[0128] S3.4. Update the speed and position of each particle:

[0129] Speed ​​Update:

[0130] ;

[0131] ;

[0132]

[0133] in, is the inertia weight, which controls the influence of the particle's historical velocity, , is the learning factor, which controls the particle’s dependence on its own best position and the global best position; , is a random number in the interval [0,1]. , , It is a particle The individual best position of , , is the best global optimal position among all particles;

[0134] Location Updates:

[0135] ;

[0136] ;

[0137] ;

[0138] S3.5. Each particle checks the current fitness value Is it better than its historical best value? If so, update the individual best position. , , ;

[0139] S3.6. Update the global optimal position by comparing the fitness values ​​of all particles , , ;

[0140] S3.7. After reaching the maximum number of iterations, the particle swarm optimization algorithm will converge to an optimal solution, which corresponds to the best combination of process parameters and will maximize the output.

[0141] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. The brain management cloud platform for carbon-based conductive new materials industry is characterized by: include: A data collection unit (1), wherein the data collection unit (1) collects production parameters by connecting to a device for producing carbon-based conductive materials, and uploads the parameters to a cloud platform; A data storage management unit (2), wherein the data storage management unit (2) processes and stores different types of data through distributed storage; A data analysis unit (3), wherein the data analysis unit (3) uses a support vector machine model to distinguish the quality of the produced carbon-based conductive material, estimates the output based on a ridge regression model, and uses a particle swarm optimization algorithm to improve production efficiency, and introduces an influence function of porosity in the process of distinguishing the support vector machine model; The method of using a support vector machine model to distinguish the quality of the produced carbon-based conductive materials includes training the support vector machine model using a kernel function and training set data, optimizing the kernel function in view of the fact that the conductivity of the carbon-based conductive new material is affected by the porosity, and further optimizing the kernel function in view of the fact that the carbon-based conductive new material is affected by the amount of surface gas adsorption; The kernel function is: ； in, Represents input features; The feature vector representing the test sample; Represents the parameter that controls the width of the kernel function; Represents the similarity measure between the input features and the feature vector of the test sample; In view of the fact that the conductivity of new carbon-based conductive materials is affected by porosity, the kernel function is optimized: ； ； in, Represents the similarity measure between the optimized input features and the feature vector of the test sample; represents the influence function of porosity; Represents input features The porosity of the sample; Indicates the porosity of the test sample; In view of the influence of the surface gas adsorption on the new carbon-based conductive materials, the kernel function is further optimized: ； in, represents the similarity measure between the input features and the feature vector of the test sample after further optimization; represents the gas adsorption amount function; Indicates the amount of gas adsorption; Indicates the ambient temperature.

2. The brain management cloud platform for the carbon-based conductive new material industry according to claim 1 is characterized by: The production parameters include pressure ,temperature and humidity .

3. The brain management cloud platform for the carbon-based conductive new material industry according to claim 2 is characterized by: The data storage management unit (2) comprises a data storage module (21) and a data management module (22); Wherein, the data storage module (21) stores the production parameters collected by the data acquisition unit (1) through distributed storage; The data management module (22) creates an index for each type of stored production parameter data, and manages the production parameter data through the index.

4. The brain management cloud platform for the carbon-based conductive new material industry according to claim 3 is characterized by: The data analysis unit (3) comprises an identification and classification module (31), a production prediction module (32), and an optimization management module (33); Wherein, the identification and classification module (31) classifies the characteristics of the carbon-based conductive material based on a support vector machine model; The production prediction module (32) establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material performance based on the results of identification and classification and a ridge regression model to predict future production; The optimization management module (33) uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction result.

5. The brain management cloud platform for the carbon-based conductive new material industry according to claim 4 is characterized by: The identification and classification module (31) classifies the characteristics of the carbon-based conductive material based on a support vector machine model, comprising the following steps: S1.1、Press ,temperature and humidity Data as input features : , set the classification label of carbon-based conductive materials ; S1.

2. Pressure ,temperature and humidity Preprocess the data; S1.3, the pressure ,temperature and humidity The data is divided into training set and test set; S1.4, select the kernel function and adjust the key parameters of the support vector machine; S1.

5. Use the kernel function and training set data to train the support vector machine model, optimize the kernel function in view of the influence of porosity on the conductivity of the carbon-based conductive new material, and further optimize the kernel function in view of the influence of surface gas adsorption on the carbon-based conductive new material; S1.

6. After training, the support vector machine model is used to learn how to classify carbon-based conductive materials according to input features and output quality classification labels.

6. The brain management cloud platform for the carbon-based conductive new material industry according to claim 5 is characterized by: In S1.6, the support vector machine model is: ； in, Indicates the quality classification label; The feature vector representing the test sample; represents the Lagrange multiplier; Represents the label of the training sample.

7. The brain management cloud platform for the carbon-based conductive new material industry according to claim 6 is characterized by: The production prediction module (32) establishes a relationship model between temperature, pressure, humidity and carbon-based conductive material performance based on a ridge regression model to predict future production, including the following steps: S2.

1. Collect carbon-based conductive materials at different pressures ,temperature and humidity The performance data under the conditions will be pre-processed; S2.

2. Select ridge regression as the prediction model, initialize the ridge regression model, and set the hyperparameters; S2.

3. Convert the quality classification labels into binary features through one-hot encoding and convert the pressure ,temperature ,humidity and quality classification labels as input features ; S2.

4. Construct a loss function, optimize the loss function according to the fluctuation of the quality of the new carbon-based conductive material, and solve the regression coefficient through ridge regression To minimize the loss function: ； in, represents the input feature moment, is the output yield, is the regularization parameter, is the identity matrix; S2.5, evaluate the trained ridge regression model by mean square error; S2.

6. After the evaluation is completed, the trained ridge regression model is used to predict the yield: ； in, represents the number of categories of quality classification labels, represents the one-hot encoded dimensional eigenvalue, represents the predicted output, Indicates new pressure, represents the new temperature, Indicates the new humidity.

8. The brain management cloud platform for the carbon-based conductive new material industry according to claim 7 is characterized by: In S2.4, the loss function is: ； in, represents the objective function; represents the regression coefficient; Indicates actual output; represents the number of samples; represents the regularization parameter; Represents the model regression coefficients; represents the regression coefficient index; Indicates the sample index; Aiming at the fluctuation of the quality of new carbon-based conductive materials, the loss function is optimized: ； in, Represents the optimized loss function; The constraint item representing the material quality; Represents the weight coefficient of the quality constraint.

9. The brain management cloud platform for the carbon-based conductive new material industry according to claim 8 is characterized by: The optimization management module (33) uses a particle swarm optimization algorithm to improve product efficiency based on the output prediction result, and includes the following steps: S3.

1. Initialize the particle swarm. The initial position of each particle represents the process parameter combination. The position of each particle is a three-dimensional vector: , Represents particles The corresponding pressure, Represents particles The corresponding temperature, Represents particles The corresponding humidity; S3.

2. The initial velocity of each particle is a vector, which indicates the moving speed of the particle in each dimension: ,in, Represents particles The speed corresponding to the temperature, Represents particles The speed corresponding to the pressure, Represents particles Speed ​​corresponding to humidity; S3.

3. For each particle, calculate its fitness value : ; S3.

4. Update the speed and position of each particle: S3.

5. Each particle checks the current fitness value Is it better than its historical best value? If so, update the individual best position; S3.6, update the global optimal position by comparing the fitness values ​​of all particles; S3.

7. After reaching the maximum number of iterations, the particle swarm optimization algorithm will converge to the optimal solution, which will maximize the output.

Images