Production monitoring method and system for non-toxic household disinfectant and storage medium

Abstract

The invention relates to the technical field of production monitoring, and discloses a production monitoring method and system for a non-toxic household disinfectant and a storage medium. The method comprises the following steps: acquiring a parameter configuration group of the household disinfectant, defining key indexes for evaluating the quality of the household disinfectant, measuring the key indexes of each production batch to obtain historical measurement data, and performing dimensionality reduction on the historical configuration data to obtain key parameters; establishing a performance prediction model and an anomaly analysis model, inputting the real-time configuration data of the key parameters in the current batch into the performance prediction model to obtain the batch type of the current batch, if the current batch is an abnormal batch, obtaining dangerous parameters from the key parameters based on the anomaly analysis model, and obtaining an anomaly reason based on the dangerous parameters; and constructing a deep reinforcement learning model, carrying out iterative optimization on the danger parameters, and inputting the optimized danger parameters into the performance prediction model until the current batch is determined to be a normal batch. The production efficiency of the household disinfectant is improved.

Description

Technical Field

[0001] The present application relates to the technical field of production monitoring, and in particular to a production monitoring method, system and storage medium for a non-toxic household disinfectant. Background Art

[0002] As people pay more attention to family hygiene and health, the market demand for non-toxic, non-irritating and environmentally friendly household disinfectants is growing. However, the production process of disinfectants is complicated, involving the precise proportions of multiple chemical components and the control of reaction conditions. Any slight deviation may lead to product quality problems and even produce toxic substances. Therefore, how to achieve accurate monitoring and control of the production process of non-toxic household disinfectants to ensure stable product quality and non-toxicity has become an urgent problem to be solved in the industry.

[0003] At present, there are few production monitoring methods for household disinfectants, but the production monitoring methods for other items mostly use machine learning and other algorithms. For example, the Chinese patent application with publication number CN118963303A discloses an intelligent monitoring method and system for milk powder production line based on artificial intelligence, which obtains the target production line equipment status data from the milk powder production line, identifies the target abnormal node reflecting the target abnormal monitoring project, and generates the corresponding first state path vector. Then, based on the production line equipment status data of the current and next equipment, the abnormal monitoring project status data reflecting the abnormal monitoring project is obtained, and the corresponding second state path vector is generated. After being fused with the first state path vector and output as the target state path vector, the abnormal indicator of the abnormal monitoring project is determined, thereby obtaining the linkage characteristic parameters of the abnormal indicator of the abnormal monitoring project in the next equipment status data, so as to determine the abnormal monitoring project node reflecting the abnormal state partition of the target abnormal monitoring project in the next equipment status data, and then predict the possible abnormality.

[0004] As another example, a Chinese patent application with publication number CN119004214A discloses a printing parameter correction method, system and medium for a solder paste printer, including obtaining historical operating parameters and historical environmental parameters of the solder paste printer within a preset time period before frequent drift printing parameters occur, and constructing a printing parameter drift prediction model based on the historical operating parameters and historical environmental parameters of the dangerous parameters; obtaining the real-time operating parameters of the solder paste printer and the real-time environmental parameters of the printing environment within a preset working time period, importing the real-time operating parameters of the dangerous parameters and the real-time environmental parameters into the printing parameter drift prediction model for prediction, and obtaining the drift warning printing parameters of the solder paste printer based on the prediction results; and generating a printing parameter correction plan in combination with the drift warning printing parameters of the solder paste printer.

[0005] However, the above-mentioned prior art has low efficiency in detecting abnormalities in the production process and optimizing parameters, and is not able to monitor the production process well. Summary of the invention

[0006] The present application provides a production monitoring method, system and storage medium for a non-toxic household disinfectant, which are used to improve the production efficiency of the non-toxic household disinfectant.

[0007] In a first aspect, the present application provides a production monitoring method for a non-toxic household disinfectant, the method comprising: Step S1: Acquire historical configuration data of household disinfectant during the production process, wherein the historical configuration data includes a variety of configuration parameters, define a combination of various configuration parameters of each production batch as a parameter configuration group, and define key indicators for evaluating the quality of the household disinfectant; Step S2: measuring the key indicators of each production batch of household disinfectant to obtain historical measurement data, wherein the historical measurement data and the corresponding parameter configuration group together constitute a first data set, normalizing and denoising the first data set, and reducing the dimension of the historical configuration data based on the first data set to obtain key parameters; Step S3: Establish a performance prediction model and an abnormality analysis model, obtain real-time configuration data of key parameters of the current batch of household disinfectant during the production process and input them into the performance prediction model, the performance prediction model outputs the batch type of the current batch, if the current batch is an abnormal batch, obtain dangerous parameters from the key parameters based on the abnormality analysis model, the key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormality causes are obtained based on the dangerous parameters; Step S4: construct a deep reinforcement learning model, iteratively optimize the risk parameters, and input the optimized risk parameters and the common parameters into the performance prediction model until the current batch is determined to be a normal batch.

[0008] In combination with the first aspect, in a first implementation of the first aspect of the present application, the historical configuration data is dimensionally reduced based on the first data set to obtain key parameters, including: Set a standard value for each configuration parameter, calculate the degree of match between the historical configuration data of each configuration parameter and the standard value, normalize the historical configuration data of the configuration parameter of each production batch and the degree of match, and estimate the quality of the household disinfectant of the i-th production batch based on the first formula to obtain a corresponding estimated value , the first formula is: , is the total number of configuration parameters, is the proportional constant of the kth configuration parameter in the ith production batch, is the matching degree between the historical configuration data of the kth configuration parameter in the i-th production batch and the standard value, is the kth configuration parameter in the i-th production batch, For each configuration parameter Estimated value the extent of contribution; If the estimated value is greater than the first threshold, the configuration parameter whose contribution degree is greater than the second threshold is obtained from all configuration parameters corresponding to the estimated value as the key parameter.

[0009] In combination with the first aspect, in a second implementation of the first aspect of the present application, calculating the degree of match between the historical configuration data of each configuration parameter and the standard value includes: The degree of matching between the historical configuration data of the kth configuration parameter in the i-th production batch and the standard value is calculated based on the second formula , the second formula is: ,in, is the actual value of the kth configuration parameter in the ith production batch, is the standard value of the kth configuration parameter.

[0010] In combination with the first aspect, in a third implementation method of the first aspect of the present application, the performance prediction model outputs the batch type of each production batch, including: the performance prediction model is a regression model, the regression model generates a predicted value for each key indicator, sets a standard range for each key indicator, and determines whether each predicted value is within the standard range. If so, the key indicator corresponding to the predicted value is defined as a reliable prediction indicator; if the total number of the key indicators in the production batch that are reliable prediction indicators is greater than or equal to a third threshold, the production batch is defined as a normal batch, otherwise it is defined as an abnormal batch.

[0011] In combination with the first aspect, in a fourth implementation manner of the first aspect of the present application, obtaining a risk parameter from the key parameter based on the abnormal cause, and obtaining the abnormal cause based on the risk parameter includes: Collecting simulation configuration data of key parameters corresponding to all abnormal causes, extracting multiple feature quantities from the simulation configuration data, and defining the abnormal causes and the feature quantities as a second data set; Create a first random forest model, input all the second data sets into the first random forest model for training, the first random forest model calculates the change in the Gini coefficient of each feature value in each decision tree before and after the split, takes the root node of each decision tree as the starting node, and determines whether the decision tree has a path with the feature value as the split node. If yes, define the decision tree as the first decision tree, calculate the sum of the changes in the Gini coefficients corresponding to all the first decision trees and the total number of the first decision trees, define the ratio of the sum of the changes in the Gini coefficients to the total number as the risk index of the feature value, arrange the risk indexes of all the feature values ​​corresponding to the abnormal causes in order, and define the key parameters corresponding to the feature values ​​ranked greater than the fourth threshold as risk parameters; Create a second random forest model, input the simulated configuration data corresponding to the dangerous parameters and the corresponding abnormal reasons into the second random forest model, the random forest model outputs the corresponding abnormal reasons, input the real-time configuration data into the second random forest model, and the second random forest model outputs the abnormal reasons for the current batch.

[0012] In combination with the first aspect, in a fifth implementation of the first aspect of the present application, iteratively optimizing the risk parameter includes: Combine multiple risk parameters to generate a state vector, define the adjusted value as the adjustment amount of the risk parameter, use the state vector as the state input of the deep reinforcement learning model, use the adjustment amount of the risk parameter as the action output of the deep reinforcement learning model, and set the reward function of the deep learning model as: ,in is the adjustment amount of the risk parameter taken when the state vector is s The reward value, m is the total number of key indicators, is the weight parameter of the i-th key indicator, The predicted value of the key indicator obtained by inputting the performance prediction model with the adjustment amount of the risk parameter as q under the state vector s, is the standard value of the i-th key indicator, To adjust the coefficient, the adjustment amount of the risk parameter in each abnormal batch is dynamically adjusted within a preset range until the reward function reaches a maximum value.

[0013] In combination with the first aspect, in a sixth implementation of the first aspect of the present application, dynamically adjusting the adjustment amount of the risk parameter in each abnormal batch within a preset range includes: Obtain the minimum standard for each key indicator and calculate the adjustment amount of the risk parameter at time t based on the third formula , the third formula is: ,in, is the preset optimal adjustment amount of the risk parameter at time t, is the weight coefficient of the optimal adjustment amount, is the risk parameter before adjustment, and the weight coefficient of the optimal adjustment amount is obtained based on the fourth formula, and the fourth formula is: ,in, is the initial weight, is the value of the optimal adjustment amount obtained based on the reinforcement learning algorithm, To obtain the value of the previous adjustment based on the reinforcement learning algorithm, For each key indicator, the minimum standard The default value is a positive number.

[0014] In a second aspect, the present application provides a production monitoring system for a non-toxic household disinfectant, the system comprising: An acquisition module, used to acquire historical configuration data of household disinfectant during the production process, wherein the historical configuration data includes a plurality of configuration parameters, a combination of various configuration parameters of each production batch is defined as a parameter configuration group, and key indicators for evaluating the quality of the household disinfectant are defined; A processing module, used to measure the key indicators of each production batch of household disinfectant to obtain historical measurement data, wherein the historical measurement data and the corresponding parameter configuration group together constitute a first data set, perform normalization and denoising on the first data set, and perform dimension reduction on the historical configuration data based on the first data set to obtain key parameters; A performance prediction and abnormality analysis module, which is used to establish a performance prediction model and an abnormality analysis model, obtain the real-time configuration data of key parameters of the current batch of household disinfectant during the production process and input them into the performance prediction model, the performance prediction model outputs the batch type of the current batch, and if the current batch is an abnormal batch, obtains the dangerous parameters from the key parameters based on the abnormality analysis model, and the key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormal reasons are obtained based on the dangerous parameters; The optimization module is used to construct a deep reinforcement learning model, iteratively optimize the risk parameters, and input the optimized risk parameters and the common parameters into the performance prediction model until the current batch is determined to be a normal batch.

[0015] A third aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for monitoring the production of a non-toxic household disinfectant.

[0016] In the technical solution provided by the present application, by collecting historical configuration data including multiple configuration parameters and defining parameter configuration groups, the specific conditions of each production batch can be traced, which is convenient for analysis and optimization, and the key indicators for evaluating the quality of household disinfectants are clearly defined, providing targets for subsequent quality control and model establishment; combining historical measurement data and parameter configuration groups to form a first data set, providing complete information for model training and analysis, and through normalization processing, ensuring that data of different dimensions and ranges can be compared under the same standard, improving the accuracy and generalization ability of the model, reducing data dimensions, extracting key parameters from configuration parameters, simplifying the model, and improving computational efficiency and prediction accuracy; the performance prediction model can output the type of the current batch in real time and quickly feedback the production status; the abnormal analysis model can identify abnormal batches and obtain dangerous parameters from key parameters, help quickly locate problems, obtain abnormal causes based on dangerous parameters, and provide a scientific basis for problem solving; the deep reinforcement learning model is used to automatically iteratively optimize dangerous parameters to reduce manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 A schematic diagram of an embodiment of a production monitoring method for a non-toxic household disinfectant in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of a production monitoring system for a non-toxic household disinfectant in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The present application embodiment provides a production monitoring method, system and storage medium for a non-toxic household disinfectant. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0020] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of a production monitoring method of a non-toxic household disinfectant in the embodiment of the present application includes: Step S1, obtaining historical configuration data of household disinfectant during the production process, the historical configuration data including a variety of configuration parameters, defining the combination of various configuration parameters of each production batch as a parameter configuration group, and defining key indicators for evaluating the quality of household disinfectant.

[0021] Specifically, historical configuration data of multiple production batches of household disinfectants in the production process are extracted from the production database, where the historical configuration data includes multiple configuration parameters, such as raw material ratio, production process parameters (such as temperature, pressure, stirring speed, reaction time, etc.), environmental parameters (such as humidity, cleanliness, etc.) and equipment status parameters (such as operation, shutdown, maintenance, etc.). The collected historical configuration data should cover multiple batches in different time periods and under different production conditions to ensure the representativeness and diversity of the data. The parameter configuration group includes, for example, raw material ratio, reaction temperature, reaction time, etc.

[0022] The key indicators for evaluating the quality of non-toxic household disinfectants include but are not limited to: bactericidal rate, pH value, viscosity and irritation. Each key indicator is quantified. For example, the bactericidal rate can be expressed by the degree of reduction in the number of surviving microorganisms after a certain period of time when a known number of specific microorganisms are mixed with the disinfectant.

[0023] Step S2: Measure the key indicators of each production batch of household disinfectant to obtain historical measurement data. The historical measurement data and the corresponding parameter configuration group together constitute a first data set. The first data set is normalized and denoised. Based on the first data set, the historical configuration data is reduced in dimension to obtain key parameters.

[0024] Specifically, at the end of each production batch, key indicators such as the sterilization rate, pH value, viscosity and irritation of the household disinfectant are accurately measured, and the historical measurement data obtained are integrated with the parameter configuration group of the batch (such as raw material ratio, temperature, time, etc.) to form a first data set. All parameters in the first data set are normalized, and random noise in the first data set is reduced by identifying and processing outliers and using smoothing techniques or filters.

[0025] Based on the first data set, the historical configuration data is reduced in dimension to obtain the key parameters, which is described in detail later. The key parameters are configuration parameters that have a certain impact on the key indicators of household disinfectants. By identifying the key parameters, we can better understand the risks and deviated configuration parameters in the production process, take measures in advance to reduce risks, and subsequently use the historical configuration data of key parameters to build a performance prediction model to predict the quality performance of future production batches, thereby improving work efficiency.

[0026] Step S3, establish a performance prediction model and an abnormality analysis model, obtain the real-time configuration data of the key parameters of the current batch of household disinfectant during the production process and input them into the performance prediction model, the performance prediction model outputs the batch type of the current batch, if the current batch is an abnormal batch, then based on the abnormality analysis model, obtain the dangerous parameters from the key parameters, the key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormal causes are obtained based on the dangerous parameters.

[0027] Specifically, the performance prediction model and the anomaly analysis model are both, for example, random forest models. The historical configuration data of key parameters are used as input data of the performance prediction model. The performance prediction model calculates the predicted value of each key indicator, and obtains the type of production batch based on the predicted value for output. The real-time configuration data of the key parameters of the current batch is input into the performance prediction model, and the model outputs the prediction result of the current batch. If the current batch is an abnormal batch, the anomaly analysis model is trained using the historical configuration data marked as abnormal. The model focuses on identifying dangerous parameters from key parameters. Based on the identified dangerous parameters and combined with the domain knowledge of the production process, the specific causes of the anomaly are analyzed and determined. Through these performance predictions and anomaly analyses, the production process of household disinfectants can be effectively monitored and optimized to ensure product quality, while abnormal situations in the production process can be discovered and handled in a timely manner.

[0028] Step S4: construct a deep reinforcement learning model, iteratively optimize the risk parameters, and input the optimized risk parameters and common parameters into the performance prediction model until the current batch is determined to be a normal batch.

[0029] Specifically, to construct a deep reinforcement learning model such as a deep Q-network, we first define an environment that simulates the production process of household disinfectant. The state space includes key parameters such as raw material ratio, temperature, time, etc. The action space consists of all dangerous parameters that may be adjusted. These parameters are considered to be variables that have a significant impact on the production process. In the simulated production environment, the model is trained by interacting with the environment.

[0030] The model learns by trying different actions and receiving rewards. The model gradually learns which actions (adjustment amounts of risk parameters) to take under different states can maximize the reward value, that is, optimize the production process. The risk parameters and other key parameters optimized by the deep reinforcement learning model are input into the performance prediction model. The performance prediction model quickly evaluates the impact of the optimized parameters on product quality and obtains the batch type of the current batch until the current batch is a normal batch.

[0031] In the technical solution provided by the present application, by collecting historical configuration data including multiple configuration parameters and defining parameter configuration groups, the specific conditions of each production batch can be traced, which is convenient for analysis and optimization, and the key indicators for evaluating the quality of household disinfectants are clearly defined, providing targets for subsequent quality control and model establishment; combining historical measurement data and parameter configuration groups to form a first data set, providing complete information for model training and analysis, and through normalization processing, ensuring that data of different dimensions and ranges can be compared under the same standard, improving the accuracy and generalization ability of the model, reducing data dimensions, extracting key parameters from configuration parameters, simplifying the model, and improving computational efficiency and prediction accuracy; the performance prediction model can output the type of the current batch in real time and quickly feedback the production status; the abnormal analysis model can identify abnormal batches and obtain dangerous parameters from key parameters, help quickly locate problems, obtain abnormal causes based on dangerous parameters, and provide a scientific basis for problem solving; the deep reinforcement learning model is used to automatically iteratively optimize dangerous parameters to reduce manual intervention.

[0032] In a specific embodiment, reducing the dimension of the historical configuration data based on the first data set to obtain the key parameters includes the following steps: Set the standard value of each configuration parameter, calculate the matching degree between the historical configuration data of each configuration parameter and the standard value, normalize the historical configuration data and matching degree of the configuration parameters of each production batch, and estimate the quality of the household disinfectant of the i-th production batch based on the first formula to obtain the corresponding estimated value , the first formula is: , is the total number of configuration parameters, is the proportional constant of the kth configuration parameter in the ith production batch, is the matching degree between the historical configuration data and the standard value of the kth configuration parameter in the i-th production batch, is the kth configuration parameter in the i-th production batch, For each configuration parameter Estimated value degree of contribution.

[0033] If the estimated value is greater than the first threshold, a configuration parameter whose contribution degree is greater than the second threshold is obtained from all configuration parameters corresponding to the estimated value as a key parameter.

[0034] Specifically, assume that the standard values ​​of the four configuration parameters in production batch 1: raw material A ratio, raw material B ratio, reaction temperature, and reaction time are 50%, 30%, 30°C, and 60min, respectively; Among them, the historical configuration data of the four configuration parameters in production batch 1 are: 55%, 28%, 32℃, 65min, and the minimum-maximum normalization method is used to normalize the historical configuration data. For example, the minimum and maximum values ​​of the raw material A ratio are 30% and 70%, the minimum and maximum values ​​of the raw material B ratio are: 10% and 50%, the minimum and maximum values ​​of the reaction temperature are: 10℃ and 50℃, and the minimum and maximum values ​​of the reaction time are: 40min and 80min, respectively. The normalized values ​​of the historical configuration data of the raw material A ratio, raw material B ratio, reaction temperature, and reaction time are , , , ,The matching degree of each configuration parameter in production batch 1 and the standard value are 0.81, -0.55, 0.54, and 0.7, respectively.

[0035] The first formula is used to calculate the estimated value V1 of production batch 1. The estimated value is a quantitative indicator of the performance or quality of household disinfectant, which can be used to evaluate whether the production batch meets the expected quality standards. The performance or quality of the disinfectant may be determined by multiple key indicators, such as bactericidal rate, pH value, viscosity, and irritation. Setting an overall quantitative indicator can combine these indicators of different dimensions to provide a comprehensive assessment and identify the configuration parameters that have a greater impact on the target variable from the production batches that meet the quality standards. Defined as the contribution degree, it is to ensure that the contribution is non-negative, even if The value is opposite to the standard value (i.e. is less than the standard value), its contribution is also positive, which reflects the degree to which the configuration parameter deviates from the standard value, rather than the direction of deviation, thereby more accurately identifying the key parameters.

[0036] V1= + + , where 0.3375, 0.165, 0.18, and 0.2333 are the contributions of the proportion of raw material A, the proportion of raw material B, the reaction temperature, and the reaction time to the quality of household disinfectant, respectively. If it is greater than the preset threshold (assuming it is 0.9), it means that the production batch is closer to the standard batch, so the corresponding key parameters can be obtained according to production batch 1, where the contribution of the proportion of raw material A, reaction time, proportion of raw material B and reaction temperature decreases in turn. Then the configuration parameters (ratio of raw material A, reaction time, proportion of raw material B) greater than 0.17 can be set as the key parameters of production batch 1. This type of key parameter is the configuration parameter that deviates from the standard value and needs to be analyzed.

[0037] In a specific embodiment, calculating the matching degree between the historical configuration data and the standard value of each configuration parameter specifically includes the following steps: Calculate the degree of match between the historical configuration data and the standard value of the kth configuration parameter in the i-th production batch based on the second formula , the second formula is: ,in, is the actual value of the kth configuration parameter in the ith production batch, is the standard value of the kth configuration parameter.

[0038] Specifically, the matching degrees of the ratio of raw material A, the ratio of raw material B, the reaction temperature, and the reaction time in production batch 1 are: , , , .

[0039] In a specific embodiment, the performance prediction model outputs the batch type of each production batch and specifically includes the following steps: The performance prediction model is a regression model. The regression model generates a predicted value for each key indicator, sets a standard range for each key indicator, and determines whether each predicted value is within the standard range. If so, the key indicator corresponding to the predicted value is defined as a reliable prediction indicator. If the total number of key indicators in a production batch that are reliable prediction indicators is greater than or equal to a third threshold, the production batch is defined as a normal batch, otherwise it is defined as an abnormal batch.

[0040] Specifically, assume that there are three key indicators: sterilization rate, pH value and viscosity, and their standard ranges are 99%-100%, 6-8 and 10-50 mPa·s, respectively. Use the trained performance prediction model to generate predicted values ​​for each key indicator, assuming they are 98%, 6 and 20 mPa·s, respectively. All of them are within the standard range. Then all key indicators are reliable prediction indicators. The total number is 2. The third threshold is set to 3 (the total number of key indicators). Because the total number of 2 is less than the third threshold of 3, the current batch is an abnormal batch. By using the regression model to predict key indicators, abnormal batches can be quickly identified and corrective measures can be taken in time to prevent unqualified products from entering the market. At the same time, the subjectivity and errors of manual judgment are reduced, and the automation and standardization of quality control are realized.

[0041] In a specific embodiment, the risk parameter is obtained from the key parameter based on the abnormal cause, and obtaining the abnormal cause based on the risk parameter includes the following steps: Collecting simulation configuration data of key parameters corresponding to all abnormal causes, extracting multiple feature quantities from the simulation configuration data, and defining the abnormal causes and the feature quantities as a second data set; Create a first random forest model, input all the second data sets into the first random forest model for training, the first random forest model calculates the change in the Gini coefficient of each feature quantity in each decision tree before and after the split, takes the root node of each decision tree as the starting node, and determines whether the decision tree has a path with the feature quantity as the split node. If yes, define the decision tree as the first decision tree, calculate the sum of the changes in the Gini coefficients corresponding to all first decision trees and the total number of first decision trees, define the ratio of the sum of the changes in the Gini coefficients to the total number as the risk index of the feature quantity, arrange the risk indexes of all feature quantities corresponding to the abnormal causes in order, and define the key parameters corresponding to the feature quantities ranked greater than the fourth threshold as risk parameters; Create a second random forest model, input the simulated configuration data corresponding to the dangerous parameters and the corresponding abnormal reasons into the second random forest model, the random forest model outputs the corresponding abnormal reasons, input the real-time configuration data into the second random forest model, and the second random forest model outputs the abnormal reasons for the current batch.

[0042] Specifically, in the production process, simulated anomalies are artificially introduced to simulate each predefined abnormal cause, such as abnormal cause A (raw material contamination: may be caused by impure raw materials themselves or improper storage conditions), and the corresponding key parameter data are raw material A ratio = 55%, raw material B ratio = 28%, temperature = 30°C, reaction time = 65 minutes; abnormal cause B (chemical reaction abnormality: may be caused by improper control of reaction temperature, pH value or reaction time), the corresponding key parameter data are raw material A ratio = 45%, raw material B ratio = 32%, temperature = 40°C, reaction time = 50 minutes, and feature quantities are extracted from the simulated configuration data, such as: raw material A ratio, raw material B ratio, reaction temperature, reaction time.

[0043] The first random forest model is a general model used to analyze and evaluate the importance of feature quantities and identify risk parameters from the feature quantities. Risk parameters refer to those key parameters that have the greatest impact on the causes of abnormalities.

[0044] A random forest is composed of multiple decision trees, each of which is an independent tree structure model. The change in the Gini coefficient indicates the reduction in the impurity of the data set after splitting on a certain feature. The larger the change, the greater the contribution of the feature to the model prediction. If feature A is used as a split node, it means that the feature contributes to the model prediction. Therefore, the sum of the changes in the Gini coefficients corresponding to all first decision trees (decision trees with feature A as split nodes) and the total number of first decision trees are calculated, which respectively represent the total contribution of feature A in all decision trees and the number of decision trees where the feature is used as a split node.

[0045] If only the sum of the changes in the Gini coefficient is used to define the risk index, the importance of the feature will directly depend on how many decision trees it is used in. Some feature quantities may be used in more decision trees due to the structure of the random forest model or the characteristics of the training data, but this does not mean that they contribute more to each decision tree. The present invention uses the ratio of the sum of the changes in the Gini coefficient to the total number (the average Gini coefficient change) as the risk index of each feature quantity, so that the importance of different feature quantities can be fairly compared, even if they are used at different frequencies in decision trees. For example, feature A is used in 5 decision trees, the total Gini coefficient change is 0.5, and the risk factor is 0.5 / 5=0.1; feature B is used in 10 decision trees, the total Gini coefficient change is 0.8, and the risk factor is 0.8 / 10=0.08. If only the total is used, feature B (0.8) will be higher than feature A (0.5), but this does not truly reflect their importance. By dividing by the total number, the average value of feature A (0.1) is higher than feature B (0.08), which more accurately reflects that feature A has a greater impact on the model.

[0046] Then, based on the above method, the risk indexes of all feature quantities corresponding to the abnormal causes are arranged in order, and the key parameters corresponding to the feature quantities with a ranking greater than the fourth threshold (which can be set to 3) are defined as risk parameters.

[0047] The main goal of creating the second random forest model is to use the risk parameters to predict the specific abnormal causes of the current production batch. Assuming that the risk parameters of abnormal cause A (raw material contamination) are the proportion of raw material A: 55% and the reaction time: 65 minutes, each simulation configuration data is associated with a specific abnormal cause. The simulation configuration data corresponding to the risk parameters and the corresponding abnormal causes are input into the second random forest model for training. The model learns the changing rules of the risk parameters under different abnormal causes, so that it can predict the abnormal causes corresponding to the new input data. The real-time monitoring data of the current production batch is input into the trained second random forest model. The model predicts the abnormal cause corresponding to the current batch based on the learned rules. For example, if the input real-time monitoring data is the proportion of raw material A: 54% and the reaction time: 66 minutes, the predicted abnormal cause may be raw material contamination.

[0048] In a specific embodiment, iterative optimization of risk parameters specifically includes the following steps: Combine multiple danger parameters to generate a state vector. The adjusted value is defined as the adjustment amount of the danger parameter. The state vector is used as the state input of the deep reinforcement learning model, and the adjustment amount of the danger parameter is used as the action output of the deep reinforcement learning model. The reward function of the deep learning model is set as: ,in is the adjustment amount of the risk parameter taken when the state vector is s The reward value, m is the total number of key indicators, is the weight parameter of the i-th key indicator, is the predicted value of the key indicator obtained by inputting the performance prediction model with the adjustment amount of the risk parameter as q under the state vector s, is the standard value of the i-th key indicator, To adjust the coefficient, the adjustment amount of the dangerous parameter in each abnormal batch is dynamically adjusted within a preset range until the reward function reaches the maximum value.

[0049] Specifically, determine the risk parameters that have the greatest impact on the production process, and combine the values ​​of these risk parameters into a vector, such as [Raw Material A Ratio, Reaction Time] = [54%, 66]. According to experience and experimental data, set the adjustment range of each risk parameter, such as the adjustment amount of the Raw Material A Ratio is ± %, Response time adjustment: ± min, taking the state vector as the input state of the deep reinforcement learning model, the model outputs the adjustment amount of each dangerous parameter and defines a reward function to evaluate the adjustment effect.

[0050] The first term in the reward function of the present invention is: It represents the deviation term, which is used to evaluate the deviation between the predicted value and the standard value of the key indicator. The predicted value is the predicted value of the key indicator obtained by inputting the adjusted risk parameter together with other key parameters into the performance prediction model. The weight parameter It reflects the importance of different key indicators. For non-toxic household disinfectants, sterilization rate and pH value are the most important key indicators, followed by important key indicators such as viscosity and stability. Assume that key indicator 1: sterilization rate, standard value (θ1): 99%, predicted value (ρ1(s,q)): 94%, weight parameter (ω1): 0.4, sterilization rate deviation term: - 0.4 * (94% - 99%) = - 0.4 * (-5%) = 0.02; key indicator 2: pH value, standard value (θ2): 7.0, predicted value (ρ2(s,q)): 6.5, weight parameter (ω2), pH value deviation term: 0.3, = - 0.3 * (6.5 - 7.0) = - 0.3 * (-0.5) = 0.015. According to the above method, the viscosity deviation term and stability deviation term are calculated to be 0.02 and 0.005 respectively, and the total deviation term = 0.02 + 0.015 + 0.02 + 0.005 = 0.06; Item 2: It represents the squared deviation term, which is the sum of the squares of the deviations between the predicted value and the standard value. It is used to penalize large deviations and encourage the model to minimize large errors when adjusting parameters. The square term is used to adjust the coefficient to balance the influence of the prediction deviation. The square term is used to punish large deviations, so that the model can adjust the parameters more accurately. After calculation, the square deviation terms of sterilization rate, pH value, viscosity and stability are -0.025, -0.025, -0.1 and -2.5 respectively. The total square deviation term is equal to (-0.025)+(-0.025)+(-0.1)+(-2.5)=-2.65, and the total reward is equal to 0.06-2.65=-2.59. Although the deviation term is positive, the total reward is still negative due to the existence of the square deviation term. This shows that the model needs to further adjust the parameters to reduce large errors and improve the prediction accuracy until the reward function reaches the maximum value. The purpose of this process is to make the value of the reward function as large as possible by continuously adjusting the risk parameters, that is, to reduce the deviation and square deviation as much as possible.

[0051] In a specific embodiment, dynamically adjusting the adjustment amount of the dangerous parameter in each abnormal batch within a preset range specifically includes the following steps: Obtain the minimum standard for each key indicator and calculate the adjustment amount of the risk parameter at time t based on the third formula , the third formula is: ,in, is the preset optimal adjustment amount of the risk parameter at time t, is the weight coefficient of the optimal adjustment amount, is the risk parameter before adjustment, and the weight coefficient of the optimal adjustment amount is obtained based on the fourth formula, which is: ,in, is the initial weight, is the value of the optimal adjustment amount obtained based on the reinforcement learning algorithm, To obtain the value of the previous adjustment based on the reinforcement learning algorithm, For each key indicator, the minimum standard The default value is a positive number.

[0052] Specifically, the minimum acceptable standard for each key indicator is determined for subsequent adjustment and evaluation. The third formula is used to calculate the adjustment amount q(t) of the risk parameter at the current moment, which combines the current optimal adjustment amount and previous adjustment , weighted average is performed through the weight coefficient φ(t), and weight coefficient A balance point can be found between the new and old adjustment amounts, which not only takes advantage of the current optimal adjustment amount but also does not completely abandon the previous adjustment experience, making the adjustment process smoother, avoiding drastic fluctuations in the adjustment amount, and ensuring the stability of the production process.

[0053] The fourth formula is used to calculate the weight coefficient of the optimal adjustment amount , which comprehensively considers the value of the current optimal adjustment and the previous adjustment, as well as the preset standard value and constant. b(t) represents the minimum standard of each key indicator, ensuring that the adjusted parameters are not lower than the minimum requirements and ensuring product quality. and They respectively represent the value of the current optimal adjustment amount and the previous adjustment amount. By comparing these two items, the pros and cons of the new and old adjustments can be evaluated, so as to reasonably allocate weights.

[0054] Through such calculations and settings, the third and fourth formulas can realize dynamic adjustment of hazardous parameters, optimize the production process, and improve product quality. This method combines reinforcement learning and weighted average technology, which can flexibly respond to various changes in the production process and provide strong support for intelligent manufacturing.

[0055] The above describes a production monitoring method for a non-toxic household disinfectant in an embodiment of the present application. The following describes a production monitoring system for a non-toxic household disinfectant in an embodiment of the present application. Figure 2 , an embodiment of a production monitoring system for a non-toxic household disinfectant in an embodiment of the present application includes: An acquisition module is used to acquire historical configuration data of household disinfectant during the production process. The historical configuration data includes multiple configuration parameters. The combination of various configuration parameters of each production batch is defined as a parameter configuration group, and key indicators for evaluating the quality of household disinfectant are defined; A processing module, used to measure the key indicators of each production batch of household disinfectant to obtain historical measurement data, the historical measurement data and the corresponding parameter configuration group together constitute a first data set, normalize and denoise the first data set, and reduce the dimension of the historical configuration data based on the first data set to obtain key parameters; The performance prediction and abnormality analysis module is used to establish a performance prediction model and an abnormality analysis model, obtain the real-time configuration data of key parameters of the current batch of household disinfectants during the production process and input them into the performance prediction model. The performance prediction model outputs the batch type of the current batch. If the current batch is an abnormal batch, the dangerous parameters are obtained from the key parameters based on the abnormality analysis model. The key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormal reasons are obtained based on the dangerous parameters. The optimization module is used to build a deep reinforcement learning model, iteratively optimize the dangerous parameters, and input the optimized dangerous parameters and normal parameters into the performance prediction model until the current batch is determined to be a normal batch.

[0056] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of a method for monitoring the production of a non-toxic household disinfectant.

[0057] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0058] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0059] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A production monitoring method for a non-toxic household disinfectant, characterized in that: The method comprises: Step S1: Acquire historical configuration data of household disinfectant during the production process, wherein the historical configuration data includes a variety of configuration parameters, define a combination of various configuration parameters of each production batch as a parameter configuration group, and define key indicators for evaluating the quality of the household disinfectant; Step S2: measuring the key indicators of each production batch of household disinfectant to obtain historical measurement data, wherein the historical measurement data and the corresponding parameter configuration group together constitute a first data set, normalizing and denoising the first data set, and reducing the dimension of the historical configuration data based on the first data set to obtain key parameters; Step S3: Establish a performance prediction model and an abnormality analysis model, obtain real-time configuration data of key parameters of the current batch of household disinfectant during the production process and input them into the performance prediction model, the performance prediction model outputs the batch type of the current batch, if the current batch is an abnormal batch, obtain dangerous parameters from the key parameters based on the abnormality analysis model, the key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormality causes are obtained based on the dangerous parameters; Step S4: construct a deep reinforcement learning model, iteratively optimize the risk parameters, and input the optimized risk parameters and the common parameters into the performance prediction model until the current batch is determined to be a normal batch.

2. The production monitoring method of a non-toxic household disinfectant according to claim 1, characterized in that: The historical configuration data is reduced in dimension based on the first data set to obtain key parameters, including: Set a standard value for each configuration parameter, calculate the degree of match between the historical configuration data of each configuration parameter and the standard value, normalize the historical configuration data of the configuration parameter of each production batch and the degree of match, and estimate the quality of the household disinfectant of the i-th production batch based on the first formula to obtain a corresponding estimated value , the first formula is: , is the total number of configuration parameters, is the proportional constant of the kth configuration parameter in the ith production batch, is the matching degree between the historical configuration data of the kth configuration parameter in the i-th production batch and the standard value, is the kth configuration parameter in the i-th production batch, For each configuration parameter Estimated value the extent of contribution; If the estimated value is greater than the first threshold, the configuration parameter whose contribution degree is greater than the second threshold is obtained from all configuration parameters corresponding to the estimated value as the key parameter.

3. The production monitoring method of a non-toxic household disinfectant according to claim 2, characterized in that: Calculating the matching degree between the historical configuration data of each configuration parameter and the standard value, including: The degree of matching between the historical configuration data of the kth configuration parameter in the i-th production batch and the standard value is calculated based on the second formula , the second formula is: ,in, is the actual value of the kth configuration parameter in the ith production batch, is the standard value of the kth configuration parameter.

4. The production monitoring method of a non-toxic household disinfectant according to claim 1, characterized in that: The performance prediction model outputs the batch type for each production batch, including: The performance prediction model is a regression model, which generates a predicted value for each key indicator, sets a standard range for each key indicator, and determines whether each predicted value is within the standard range. If so, the key indicator corresponding to the predicted value is defined as a reliable prediction indicator. If the total number of key indicators in the production batch that are reliable prediction indicators is greater than or equal to a third threshold, the production batch is defined as a normal batch, otherwise it is defined as an abnormal batch.

5. The production monitoring method of a non-toxic household disinfectant according to claim 4, characterized in that: Acquiring a risk parameter from the key parameter based on the abnormal cause, wherein acquiring the abnormal cause based on the risk parameter includes: Collecting simulation configuration data of key parameters corresponding to all abnormal causes, extracting multiple feature quantities from the simulation configuration data, and defining the abnormal causes and the feature quantities as a second data set; Create a first random forest model, input all the second data sets into the first random forest model for training, the first random forest model calculates the change in the Gini coefficient of each feature value in each decision tree before and after the split, takes the root node of each decision tree as the starting node, and determines whether the decision tree has a path with the feature value as the split node. If yes, define the decision tree as the first decision tree, calculate the sum of the changes in the Gini coefficients corresponding to all the first decision trees and the total number of the first decision trees, define the ratio of the sum of the changes in the Gini coefficients to the total number as the risk index of the feature value, arrange the risk indexes of all the feature values ​​corresponding to the abnormal causes in order, and define the key parameters corresponding to the feature values ​​ranked greater than the fourth threshold as risk parameters; Create a second random forest model, input the simulated configuration data corresponding to the dangerous parameters and the corresponding abnormal reasons into the second random forest model, the random forest model outputs the corresponding abnormal reasons, input the real-time configuration data into the second random forest model, and the second random forest model outputs the abnormal reasons for the current batch.

6. The production monitoring method of a non-toxic household disinfectant according to claim 1, characterized in that: Iterative optimization of risk parameters, including: Combine multiple risk parameters to generate a state vector, define the adjusted value as the adjustment amount of the risk parameter, use the state vector as the state input of the deep reinforcement learning model, use the adjustment amount of the risk parameter as the action output of the deep reinforcement learning model, and set the reward function of the deep learning model as: ,in is the adjustment amount of the risk parameter taken when the state vector is s The reward value, m is the total number of key indicators, is the weight parameter of the i-th key indicator, The predicted value of the key indicator obtained by inputting the performance prediction model with the adjustment amount of the risk parameter as q under the state vector s, is the standard value of the i-th key indicator, To adjust the coefficient, the adjustment amount of the risk parameter in each abnormal batch is dynamically adjusted within a preset range until the reward function reaches a maximum value.

7. The production monitoring method of a non-toxic household disinfectant according to claim 6, characterized in that: Dynamically adjust the adjustment amount of the dangerous parameters in each abnormal batch within the preset range, including: Obtain the minimum standard for each key indicator and calculate the adjustment amount of the risk parameter at time t based on the third formula , the third formula is: ,in, is the preset optimal adjustment amount of the risk parameter at time t, is the weight coefficient of the optimal adjustment amount, is the risk parameter before adjustment, and the weight coefficient of the optimal adjustment amount is obtained based on the fourth formula, and the fourth formula is: ,in, is the initial weight, is the value of the optimal adjustment amount obtained based on the reinforcement learning algorithm, To obtain the value of the previous adjustment based on the reinforcement learning algorithm, For each key indicator, the minimum standard The default value is a positive number.

8. A production monitoring system for non-toxic household disinfectant, used to implement a production monitoring method for non-toxic household disinfectant as claimed in any one of claims 1 to 7, characterized in that: The system comprises: An acquisition module, used to acquire historical configuration data of household disinfectant during the production process, wherein the historical configuration data includes a plurality of configuration parameters, a combination of various configuration parameters of each production batch is defined as a parameter configuration group, and key indicators for evaluating the quality of the household disinfectant are defined; A processing module, used to measure the key indicators of each production batch of household disinfectant to obtain historical measurement data, wherein the historical measurement data and the corresponding parameter configuration group together constitute a first data set, perform normalization and denoising on the first data set, and perform dimension reduction on the historical configuration data based on the first data set to obtain key parameters; A performance prediction and abnormality analysis module, which is used to establish a performance prediction model and an abnormality analysis model, obtain the real-time configuration data of key parameters of the current batch of household disinfectant during the production process and input them into the performance prediction model, the performance prediction model outputs the batch type of the current batch, and if the current batch is an abnormal batch, obtains the dangerous parameters from the key parameters based on the abnormality analysis model, and the key parameters other than the dangerous parameters are defined as ordinary parameters, and the abnormal reasons are obtained based on the dangerous parameters; The optimization module is used to construct a deep reinforcement learning model, iteratively optimize the risk parameters, and input the optimized risk parameters and the common parameters into the performance prediction model until the current batch is determined to be a normal batch.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, a production monitoring method for a non-toxic household disinfectant according to any one of claims 1 to 7 is implemented.

Images