Automatic optimization method and system for quality control strategy of experimental analysis

Abstract

The invention relates to the technical field of experimental quality control analysis, in particular to an automatic optimization method and system for a quality control strategy of experimental analysis. The method comprises the following steps: acquiring historical experiment quality control data of a control sample; performing clustering control mode control limit analysis based on the historical experiment quality control data to generate clustering control mode control limit data; according to the control limit data of the clustering control mode, performing experimental quality control leaf node region analysis of the clustering control mode on the historical experimental quality control data, and generating experimental quality control leaf node region data of the clustering control mode; designing an experiment quality control gain tree model based on the clustering control mode experiment quality control leaf node region data; collecting experiment monitoring data; transmitting experiment monitoring data to an experiment quality control gain tree model to analyze an intelligent experiment optimization control strategy; and executing the experiment quality control strategy automatic optimization operation according to the intelligent experiment optimization control strategy. According to the invention, efficient and intelligent experiment quality control strategy optimization is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of experimental quality control analysis, and in particular to an automatic optimization method and system for a quality control strategy of an experimental analysis. Background Art

[0002] With the development of science and technology, the accuracy of various analytical equipment and instruments in laboratories has been continuously improved, and the quality control requirements for experimental data have become increasingly higher, especially in the fields of experimental analysis such as biomedicine, environmental monitoring, and chemical analysis. The quality of experimental data directly affects the reliability and accuracy of experimental results. Laboratories are facing increasingly complex data flows and quality control (QC) requirements in data acquisition, processing, and analysis. Quality control is an important means to ensure the accuracy and reliability of experimental analysis results. Through the continuous development of artificial intelligence (AI) and machine learning (ML) technologies, data can be acquired and analyzed in real time during the experiment, potential quality problems can be automatically identified, and experimental conditions can be adjusted in time to optimize experimental results. However, the existing automatic optimization methods for quality control strategies of experimental analysis cannot effectively identify complex data patterns, and how to establish effective control modes and optimize experimental conditions and how to monitor and adjust in real time during the experiment are still technical difficulties, mainly focusing on a single aspect, such as monitoring of experimental data or optimization of control strategies, and it is difficult to achieve integrated and comprehensive intelligent quality control strategy optimization. Summary of the invention

[0003] Based on this, the present invention provides a method and system for automatically optimizing the quality control strategy of experimental analysis to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for automatically optimizing the quality control strategy of experimental analysis includes the following steps: Step S1: Acquire historical experimental quality control data of control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; Step S2: Perform cluster control mode experiment quality control deviation evaluation processing on cluster control mode experiment quality control data according to cluster control mode control limit data, and generate cluster control mode experiment quality control deviation data; perform cluster control mode experiment quality control influencing factor analysis processing based on cluster control mode experiment quality control deviation data, and generate cluster control mode experiment quality control influencing factor; perform cluster control mode experiment quality control leaf node area analysis according to cluster control mode experiment quality control influencing factor, and generate cluster control mode experiment quality control leaf node area data; Step S3: Designing the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generating the adjustment experimental quality control gain tree model; performing the optimal fitting training processing of the experimental quality control leaf node area on the adjustment experimental quality control gain tree model based on the historical experimental quality control data, and generating the experimental quality control gain tree model; Step S4: Use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; perform intelligent experimental optimization control strategy analysis on the experimental monitoring data based on the experimental quality control gain tree model to generate an intelligent experimental optimization control strategy; execute the experimental quality control strategy automation optimization operation according to the intelligent experimental optimization control strategy.

[0005] Further, step S1 includes the following steps: Step S11: Obtain historical experimental quality control data of control samples; Step S12: performing control mode recognition processing according to historical experimental quality control data to generate historical experimental control mode data; Step S13: performing cluster control mode experiment quality control analysis on the historical experiment quality control data based on the historical experiment control mode data to generate cluster control mode experiment quality control data; Step S14: extracting the quality control state characteristics of the cluster control mode experiment according to the cluster control mode experiment quality control data, and generating the quality control state characteristic data of the cluster control mode experiment; Step S15: Perform cluster control mode control limit analysis based on the cluster control mode experimental quality control state characteristic data to generate cluster control mode control limit data.

[0006] Further, step S13 includes the following steps: Step S131: performing cluster control mode analysis based on historical experimental control mode data to generate cluster experimental control mode data; Step S132: establishing a clustering experiment control mode matrix through clustering experiment control mode data; Step S133: Perform cluster control mode experiment quality control analysis by selecting historical experiment quality control data corresponding to cluster nodes in the cluster experiment control mode matrix to generate cluster control mode experiment quality control data.

[0007] Further, step S2 includes the following steps: Step S21: performing cluster control mode experiment quality control deviation evaluation processing on the cluster control mode experiment quality control data according to the cluster control mode control limit data to generate cluster control mode experiment quality control deviation data; Step S22: performing cluster control mode experiment control influencing parameter analysis based on cluster control mode experiment quality control data to generate cluster control mode experiment control influencing parameters; Step S23: performing cluster control mode experiment quality control influence characteristic analysis on cluster control mode experiment control influence parameters according to cluster control mode experiment quality control deviation data to obtain cluster control mode experiment quality control influence factors; Step S24: performing an experimental quality control leaf node region analysis of the cluster control model according to the cluster control model experimental quality control influencing factors, and generating cluster control model experimental quality control leaf node region data.

[0008] Further, step S24 includes the following steps: Step S241: designing the first experimental quality control leaf node region data according to the experimental quality control influencing factors of the cluster control mode; Step S242: setting an influencing factor adaptive partitioning window, and using the influencing factor adaptive partitioning window to perform experimental quality control sub-node region partitioning on the first experimental quality control leaf node region data to design the second experimental quality control leaf node region data, wherein the second experimental quality control leaf node region data is a sub-leaf node region of the first experimental quality control leaf node region data; Step S243: The second experiment quality control leaf node region data is used as the first experiment quality control leaf node region data and fed back to step S242 for iterative analysis and processing of the experiment quality control leaf node region, thereby obtaining the cluster control mode experiment quality control leaf node region data.

[0009] Further, step S3 includes the following steps: Step S31: Based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the cluster control mode, a mapping relationship between the experimental quality control influencing factors and the experimental quality control gains of the cluster control mode is established to generate an initial experimental quality control gain tree model, and the initial experimental quality control gain tree model is subjected to model fitting parameter limit adjustment processing through the control limit data of the cluster control mode to generate an adjusted experimental quality control gain tree model; Step S32: Based on the historical experimental quality control data, the optimal fitting training process of the experimental quality control leaf node region is performed on the adjusted experimental quality control gain tree model to generate the experimental quality control gain tree model.

[0010] Further, step S32 includes the following steps: Step S321: transferring the historical experimental quality control data to the adjusted experimental quality control gain tree model for experimental quality control benefit model training processing to generate experimental quality control benefit model training data; Step S322: performing correlation model training data analysis based on the experimental quality control benefit model training data to generate experimental quality control benefit correlation model training data; Step S323: performing experimental quality control correlation gradient improvement feature analysis based on the experimental quality control benefit correlation model training data to generate experimental quality control correlation gradient improvement feature data; Step S324: performing optimal fitting training processing on the experimental quality control leaf node region of the adjusted experimental quality control gain tree model through the experimental quality control associated gradient lifting feature data to generate the experimental quality control gain tree model.

[0011] Further, step S4 includes the following steps: Step S41: using a sensor monitoring device to perform real-time monitoring processing on the control sample during the experiment to obtain experimental monitoring data; Step S42: extracting real-time experiment quality control data according to the experiment monitoring data to generate real-time experiment quality control data, wherein the real-time experiment quality control data includes real-time experiment control data and real-time experiment quality data; Step S43: transmitting the real-time experiment quality data to the experiment quality control gain tree model to perform intelligent experiment optimization control strategy analysis and generate an intelligent experiment optimization control strategy; Step S44: Execute the automated optimization operation of the experimental quality control strategy according to the intelligent experimental optimization control strategy.

[0012] Further, step S43 includes the following steps: Step S431: performing an experiment quality optimization demand feature analysis on the real-time experiment quality data through the experiment quality control gain tree model to generate experiment quality optimization demand feature data; Step S432: Perform intelligent analysis of the experiment optimization control strategy on the real-time experiment control data according to the experiment quality optimization demand characteristic data to generate an intelligent experiment optimization control strategy.

[0013] This specification provides a system for automatically optimizing the quality control strategy of experimental analysis, which is used to execute the automatic optimization method of the quality control strategy of experimental analysis as described above. The automatic optimization of the quality control strategy of experimental analysis includes: The experimental control limit analysis module is used to obtain the historical experimental quality control data of the control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; The experimental quality control impact analysis module is used to evaluate the experimental quality control deviation of the cluster control mode based on the control limit data of the cluster control mode, and generate the experimental quality control deviation data of the cluster control mode; to analyze the experimental quality control impact factor of the cluster control mode based on the experimental quality control deviation data of the cluster control mode, and generate the experimental quality control impact factor of the cluster control mode; to analyze the experimental quality control leaf node area of ​​the cluster control mode based on the experimental quality control impact factor of the cluster control mode, and generate the experimental quality control leaf node area data of the cluster control mode; The experimental quality control gain tree model construction module is used to design the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generate the adjustment experimental quality control gain tree model; based on the historical experimental quality control data, the experimental quality control leaf node area of ​​the adjustment experimental quality control gain tree model is subjected to the optimal fitting training processing to generate the experimental quality control gain tree model; The intelligent optimization module of experimental quality control strategy is used to use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; based on the experimental quality control gain tree model, the intelligent experimental optimization control strategy is analyzed to generate an intelligent experimental optimization control strategy; and the experimental quality control strategy is automatically optimized according to the intelligent experimental optimization control strategy.

[0014] The beneficial effect of this application is that, by obtaining the historical experimental quality control data of the control samples, the necessary basic data is provided for the entire quality control strategy optimization process. These historical data record the quality control of various control samples in the experiment, reflecting the potential systematic problems and variation factors in the experiment. By collecting and analyzing historical experimental data, data support is provided for the subsequent optimization of quality control strategies, so that the control strategy can be accurately adjusted according to the real experimental data, reducing the dependence on assumptions and prediction models. Based on the historical experimental quality control data, control pattern recognition processing is performed to identify the potential control mode under different experimental conditions, and a stable and representative control mode can be extracted from a large amount of historical data, and the potential trend or problem in the data can be identified, thereby providing a clear direction for the subsequent optimization analysis, avoiding blind adjustment, and ensuring the scientificity and pertinence of the experimental control. Based on the historical experimental control mode data, clustering control mode experimental quality control analysis is performed, and the experimental data is effectively divided into multiple cluster groups with similar characteristics. Different clustering patterns represent different control states or characteristics in the experiment, which can reveal the regularity of data changes under different experimental conditions. Through cluster analysis, experimental data under different modes can be described and processed more accurately, the control scheme can be optimized, and more effective intervention can be carried out during the experiment. Experimental quality control state feature extraction is performed on the experimental quality control data of the cluster control mode, and the key features in each cluster mode are further extracted to help identify which experimental states have a decisive impact on the quality of the experimental results, accurately capture the key variables in the experiment, and avoid misjudgment or omission caused by over-reliance on a single indicator in traditional quality control methods. Based on the experimental quality control state feature data of the cluster control mode, the control limit analysis of the cluster control mode is performed, and the corresponding control limit is set for each control mode in the experiment. According to the experimental characteristics under different control modes, a scientific and reasonable control limit value can be formulated for each cluster mode, and the quality deviation during the experiment is ensured not to exceed the predetermined range. By accurately setting the control limit, not only can the experimental deviation be discovered in time, but also the negative impact of too loose or too strict control limits on the experimental results can be avoided. According to the control limit data of the cluster control mode, the deviation of the experimental quality control data of the cluster control mode is evaluated to evaluate whether the experimental data is abnormal, and effectively identify the quality problems in the data, such as whether the experimental data exceeds the predetermined control limit or whether there is abnormal fluctuation. By analyzing the experimental quality control parameters based on the experimental quality control data of the cluster control mode, it is possible to extract the key parameters that affect the experimental quality from different clustering modes, perform parameterized analysis on the quality control data under different experimental conditions, and find out which experimental parameters have a significant impact on the quality of the results, so as to optimize them in a targeted manner. By establishing a more sophisticated experimental quality control parameter model, researchers can adjust the experimental conditions in a targeted manner to avoid affecting the experimental results due to improper parameter settings, thereby improving the repeatability and accuracy of the experiment.According to the experimental quality control deviation data of the cluster control mode, the influence characteristics of the experimental quality control parameters are analyzed to further identify the key factors affecting the experimental results, effectively reveal the relationship between the variables under different experimental control modes, help researchers identify which factors have a greater impact on the experimental results, and provide a basis for further adjustments. By focusing on these influencing factors, unnecessary interference factors are eliminated to improve the stability and accuracy of the experiment. According to the experimental quality control influencing factors of the cluster control mode, the experimental quality control leaf node area analysis is carried out to achieve fine division and adjustment of the experimental quality control area. According to the different control modes and characteristics of the influencing factors of the experiment, the leaf node area of ​​the experimental quality control is accurately determined, providing a precise operating range for automatic adjustment, and the experimental data is more finely segmented. And a personalized control strategy is formulated for each leaf node area. Through this customized control area analysis, the experimental quality control strategy can be adjusted more flexibly, the experimental conditions can be optimized, and the accuracy and stability of the experimental results can be improved. Based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, the mapping relationship between the experimental quality control influencing factors and the experimental quality control gains is established. The gradient boosting decision tree algorithm can efficiently learn the nonlinear relationship between each influencing factor and the quality control gain from a large amount of experimental data, so that the system can capture the complex laws hidden behind the data. After generating the initial experimental quality control gain tree model, the model fitting parameters are adjusted in combination with the control limit data of the clustering control mode to further optimize the accuracy and robustness of the model. The parameters of the model can be automatically adjusted to ensure that the gain tree model is more in line with the distribution and control requirements of the actual experimental data, thereby providing a more accurate decision-making basis for the intelligent adjustment of experimental quality control. Based on the historical experimental quality control data, the optimal fitting training of the adjusted experimental quality control gain tree model is carried out. As the real experimental feedback, the historical experimental data can effectively provide the necessary training data for the gain tree model to help the model to be further adjusted and optimized. Through the optimal fitting training, the gain tree model can capture the subtle changes and relationships of various factors in the experiment on the basis of ensuring high accuracy, making the model more flexible and efficient when facing actual experimental data. The resulting experimental quality control gain tree model can provide more accurate and reliable decision support for the intelligent optimization control strategy during the experiment, ensuring that the quality control during the experiment can be adjusted and optimized in a timely manner, and more accurately predicting the changes in experimental results, greatly improving the automation and intelligence level of experimental quality control, and promoting the development of experimental operations in a more efficient and accurate direction. Sensor monitoring equipment is used to monitor the experimental process of control samples in real time to ensure dynamic tracking of experimental data. Through real-time data collection by sensor equipment, the status of the samples, environmental parameters and operating conditions are continuously monitored during the experiment, thereby providing instant feedback on key factors in the experimental process.According to the experimental monitoring data, the real-time experimental quality control data is extracted, including the real-time experimental control data and the real-time experimental quality data, so that the real-time monitoring data can be converted into actionable quality control information. The real-time experimental quality data is transmitted to the experimental quality control gain tree model for intelligent experimental optimization control strategy analysis, and the experimental optimization control strategy is generated. The intelligent analysis function of the experimental quality control gain tree model is used to deeply process the real-time quality data, automatically identify the optimization space in the experimental process, and generate a suitable control strategy. Through intelligent analysis, the system can combine historical data and real-time data to automatically adjust the experimental parameters or control conditions. Intelligent optimization not only reduces manual intervention and avoids errors caused by human judgment, but also can achieve precise control during the experiment, ensuring that the experimental results are more stable and efficient. According to the intelligent experimental optimization control strategy, the experimental quality control strategy is automatically optimized, which can simulate the execution of the experiment, verify the effects of different control strategies, optimize the strategies, and improve the automation level of the experimental strategy. Therefore, the automatic optimization method of the quality control strategy of the experimental analysis of the present invention can realize the effective recognition of complex data patterns, establish an effective control mode and optimize the experimental conditions and perform real-time monitoring and automatic adjustment during the experiment, etc., thereby realizing the optimization of the experimental quality control strategy through the automatic control of the monitoring of experimental data, and realizing integrated and comprehensive intelligent quality control strategy optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of the steps of an automatic optimization method for quality control strategy of experimental analysis of the present invention; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0016] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0017] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0018] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another. For example, a first unit may be referred to as a second unit, and similarly a second unit may be referred to as a first unit without departing from the scope of the exemplary embodiments. The term "and / or" used herein includes any and all combinations of one or more of the listed items of the same type.

[0019] To achieve this, please refer to Figure 1 to Figure 2 The present invention provides a method for automatically optimizing the quality control strategy of experimental analysis. In the embodiments of the present invention, please refer to Figure 1 FIG. 1 is a schematic diagram of a process flow of an automatic optimization method for a quality control strategy of an experimental analysis of the present invention, wherein the automatic optimization method for a quality control strategy of an experimental analysis comprises the following steps: Step S1: Acquire historical experimental quality control data of control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; In an embodiment of the present invention, historical experimental quality control data of control samples are obtained. In order to ensure the accuracy and completeness of the data, the experimenter extracts experimental quality control data containing detailed information such as sample batch number, experimental conditions, control data, quality standards, test values ​​and timestamps from the historical experimental database through the laboratory information management system (LIMS). These data include the measurement results of various control samples at different experimental stages, such as key control parameters such as the concentration, temperature, humidity, and pH value of the target substance. Control pattern recognition processing is performed based on the historical experimental quality control data, and data mining technology is applied to analyze the historical experimental quality control data to identify the typical patterns of control samples in different experimental stages, and experimental quality control data with similar characteristics are classified into the same category. Through control pattern recognition, the system can automatically detect data sets that meet certain specific characteristics, for example, the temperature is too high and the pH value fluctuates too much. Based on the identified historical experimental control pattern, the historical experimental quality control data is clustered and the control pattern experimental quality control analysis is performed, and a suitable clustering algorithm (such as K-means clustering algorithm) is selected to input the identified historical experimental control pattern data into the clustering model. By calculating the similarity of each data point, the model will divide the data into different clusters according to their similarity, representing the experimental quality control status corresponding to different experimental control modes. The generated cluster control mode experimental quality control data will reflect the quality control characteristics under different experimental states, providing data support for subsequent control analysis and adjustment. Experimental quality control state feature extraction is performed based on cluster control mode experimental quality control data. By extracting the key state features in each cluster, the inherent laws of the cluster data are further analyzed, and statistical analysis methods such as mean, standard deviation, kurtosis and skewness are used to perform descriptive statistics on the experimental quality control data in each cluster, extracting the state features that have a significant impact on the experimental results, and excluding those noise features that may not affect the experimental quality control results, so as to obtain the class control mode experimental quality control state features. These state feature data will be used as input for subsequent control limit analysis to help the system identify which features are critical to controlling the quality of the experiment. Based on the experimental quality control status characteristics of the cluster control mode, control limit analysis is performed. The extracted experimental quality control status characteristic data is input into the control limit analysis model for analysis and generation of control limit data for the cluster control mode. Based on statistical methods such as normal distribution analysis, the control limits of each cluster, such as upper and lower limits, are calculated. For each cluster, the average value and upper and lower limits of this type of data are calculated to determine the allowable deviation range under the cluster mode. For example, the temperature fluctuation range of a cluster may be set within the range of ±2°C. Exceeding this range is considered abnormal, and a set of control limit data is generated for each cluster mode.

[0020] Step S2: Perform cluster control mode experiment quality control deviation evaluation processing on cluster control mode experiment quality control data according to cluster control mode control limit data, and generate cluster control mode experiment quality control deviation data; perform cluster control mode experiment quality control influencing factor analysis processing based on cluster control mode experiment quality control deviation data, and generate cluster control mode experiment quality control influencing factor; perform cluster control mode experiment quality control leaf node area analysis according to cluster control mode experiment quality control influencing factor, and generate cluster control mode experiment quality control leaf node area data; In an embodiment of the present invention, cluster control mode experimental quality control data is evaluated for cluster control mode experimental quality control deviation according to cluster control mode control limit data, cluster control mode experimental quality control data is compared with these control limit data, the deviation between each data point and its corresponding control limit is calculated, and the difference calculation formula (i.e., each experimental data point minus the control limit value) is used to evaluate the deviation of all data points. If the deviation value exceeds the set threshold range, the system is marked as deviation abnormal data to obtain cluster control mode experimental quality control deviation data, which reflects the deviation between the experimental data and the control limit under each control mode, and provides a basis for subsequent data optimization. According to the cluster control mode experimental quality control data, cluster control mode experimental control influencing parameter analysis is performed, and the cluster control mode experimental quality control data contains multiple experimental data under different control modes, which are closely related to the experimental results. The system extracts the control parameters of the experimental process for these experimental data through a multivariate regression analysis method. Then, the correlation analysis method is used to evaluate the relationship between these variables and the experimental results, and the main parameters affecting the experimental results are determined. Through stepwise regression, principal component analysis (PCA) and other techniques, the parameters that have the greatest impact on the control mode are screened out, and cluster control mode experimental control influencing parameter data are generated. These parameter data will provide a basis for subsequent impact feature analysis and optimization decisions. According to the cluster control mode experimental quality control deviation data, the cluster control mode experimental control influencing parameters are analyzed for cluster control mode experimental quality control impact characteristics, and the deviation data is combined with the experimental control influencing parameters to perform impact feature analysis. The purpose of this analysis is to find out which control parameter changes have a strong correlation with the deviation data, and the correlation matrix analysis is used to calculate the correlation coefficient between each influencing parameter and the experimental data deviation. If the correlation coefficient is high, it means that the fluctuation of the control parameter during the experiment has a greater impact on the deviation. The decision tree algorithm, random forest or other machine learning methods are used to deeply analyze the degree of influence of each control parameter on the experimental result deviation, identify the quantitative information of the impact of the experimental control influencing parameters on quality control, and thus obtain the cluster control mode experimental quality control influencing factors. According to the clustering control mode experimental quality control influencing factors, the experimental quality control leaf node area of ​​the clustering control mode is analyzed. The clustering control mode experimental quality control influencing factors reflect which control parameters have a decisive influence on the experimental results. The preliminary leaf node area is set by the clustering control mode experimental quality control influencing factors, and the preliminary leaf node area is gradually divided by setting the influence factor adaptive division window. The influence factor adaptive division window considers that the setting can finely divide the leaf node area and use it as the next leaf node area. The leaf node area is iteratively analyzed until the preset convergence threshold is reached, and finally the clustering control mode experimental quality control leaf node area data is obtained.

[0021] Step S3: Designing the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generating the adjustment experimental quality control gain tree model; performing the optimal fitting training processing of the experimental quality control leaf node area on the adjustment experimental quality control gain tree model based on the historical experimental quality control data, and generating the experimental quality control gain tree model; In an embodiment of the present invention, a mapping relationship between experimental quality control influencing factors and experimental quality control gains is established based on a preset gradient boosting decision tree (GBDT) algorithm and cluster control mode experimental quality control leaf node area data, and the experimental quality control leaf node area data obtained from the cluster analysis are used. These data contain the influence of different control parameters (such as temperature, humidity, pressure, etc.) on the experimental quality in each control mode. The algorithm constructs multiple decision trees and gradually adjusts the weight of each tree so that the final model can accurately predict the experimental quality control gain. By setting appropriate input features (such as temperature, pressure and other control factors) and target variables (experimental quality control gain), a gradient boosting decision tree is used to model the relationship between these features and gains. During the construction of each tree, the gain of each decision node is repeatedly calculated through the cluster control mode experimental quality control leaf node area data, and the prediction accuracy of the model is improved by the weighted average method, and the initial experimental quality control gain tree model is established. In order to optimize the accuracy of the model, the cluster control mode control limit data is used to adjust the parameters of the initial experimental quality control gain tree model. The cluster control mode control limit data provides the system with the control parameter range or threshold allowed in each experimental mode, adjusts the parameters of the initial gain tree model, and further optimizes the split nodes and gain values ​​of the decision tree to obtain the adjusted experimental quality control gain tree model. The adjusted model can more accurately reflect the control mode in the experimental process and its impact on the experimental quality control results.

[0022] In an embodiment of the present invention, historical experimental quality control data are collected and sorted, including control parameters (such as temperature, humidity, pressure, etc.) of each experiment, experimental results (such as output, accuracy, qualified rate, etc.) and corresponding experimental quality control gains (such as qualified rate, experimental error rate, etc.). These data are used as training sets. The historical experimental data are cleaned and processed to remove outliers and missing values ​​to ensure the accuracy and consistency of the data. Each experimental result in the historical data and its corresponding control conditions (such as temperature, humidity, etc.) are used as input features, and the experimental gain (such as qualified rate, efficiency, etc.) is used as the target output. The relationship between the input feature and the target output will be used to train the model. The adjusted experimental quality control gain tree model is used for training. According to the historical experimental data and the initial parameters of the adjusted experimental quality control gain tree model, the system starts the training process to optimize the various parameters in the model to ensure the best fit. Part of the data in the historical experimental quality control data (usually 70% to 80%) is used for the training process, and the remaining data is used to verify the prediction effect of the model. The adjusted experimental quality control gain tree model (i.e., the tree model that contains the control limit data of the cluster control mode and the optimal parameter settings) is trained using the gradient boosting decision tree algorithm (GBDT). Each time the model is trained, the prediction error is gradually corrected to finally obtain the optimized gain tree. During the training process, GBDT improves the performance of the model by calculating the error of each tree, gradually reduces the training error, and finally achieves the best fit. During the training process, the model gradually adjusts the division of each leaf node area according to the data in the training set. Each leaf node area represents a set of control conditions with similar experimental quality control gains (for example, experimental conditions under a certain temperature and humidity). The system ensures that each leaf node contains the optimal experimental conditions and their corresponding experimental gains by adjusting the splitting rules of each leaf node. At this point, the system performs the following operations: Based on the data in the training set, the system optimizes the division of the experimental quality control leaf node area. At each split, the system selects the optimal segmentation feature based on the distribution of the data and the change trend of the gain. For example, a leaf node can represent an experimental condition at a specific temperature, while another leaf node represents the experimental results under different humidity conditions. The optimization process makes decisions by selecting features that minimize the error and maximize the gain. In each leaf node area, the system calculates the average experimental quality control gain of the area as the predicted output of the node. If the experimental gain value fluctuates greatly in some areas, the model will be adjusted according to the training data to ensure that the output of each leaf node is as stable as possible. In the optimal fitting process, the division method of the leaf node area is cross-validated to further verify the fitting effect of the model and ensure its performance on the validation set is optimal. The system avoids overfitting or underfitting by adjusting parameters such as the learning rate, the depth of the tree, and the number of features of the split node. Standard evaluation indicators (such as mean square error MSE, R², accuracy, etc.) are used to evaluate the fitting effect of the model.If the performance of the model on the validation set meets the predetermined standard, the model is considered to have achieved optimal fit. After the training is completed, the system outputs the final experimental quality control gain tree model. The model can predict the experimental quality control gain value based on the experimental control conditions and can be used for subsequent experimental optimization control strategies.

[0023] Step S4: Use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; perform intelligent experimental optimization control strategy analysis on the experimental monitoring data based on the experimental quality control gain tree model to generate an intelligent experimental optimization control strategy; execute the experimental quality control strategy automation optimization operation according to the intelligent experimental optimization control strategy.

[0024] The real-time monitoring of the experimental process is completed through a group of sensors installed on the experimental equipment. The types of sensors include monitoring sensors, temperature sensors, humidity sensors, pressure sensors, and reaction rate sensors. The sampling frequency of each sensor is set according to the experimental requirements. For example, the sampling frequency of temperature and humidity sensors can be set to once per second, while the sampling frequency of reaction rate sensors can be set to once per minute to ensure the accuracy and timeliness of various data during the experiment. The collected sensor data is processed in real time by the data acquisition equipment, and the data is transmitted to the data processing platform in real time to obtain the experimental monitoring data. Based on the real-time monitoring data, data extraction and processing are performed. The real-time experimental quality control data mainly includes two types of information: real-time experimental control data and real-time experimental quality data. Experimental control data includes various control parameters used in the experiment, such as temperature, humidity, pressure, reaction time, etc. Experimental quality data refers to the quality indicators obtained by monitoring the reaction process, sample characteristics, etc., including information such as reaction rate, product concentration, sample color change, etc. The quality indicators of the experiment will be reflected in the experimental data in real time, which is convenient for real-time evaluation of the progress and quality of the experiment. The extracted control data and quality data are integrated according to the timestamp to form a complete real-time experimental quality control data set. The real-time experimental quality data is input into the pre-trained experimental quality control gain tree model, and the real-time generated real-time experimental quality data will be transmitted to the experimental quality control gain tree model through the data interface. The model has been trained based on historical experimental data and experimental optimization strategies, and can perform real-time decision analysis based on the input data. For example, real-time temperature, humidity, reaction rate and other data will be input into the model as input, and the model will use these input data for reasoning and generate optimized control strategies. After the data is input, the experimental quality control gain tree model analyzes the real-time data and evaluates the impact of experimental control parameters on experimental quality. According to the current experimental state, the model will calculate the control strategy that can maximize the experimental benefits. Based on the analysis results, the experimental quality control gain tree model will output an optimization control strategy for the current experimental state, that is, the intelligent experimental optimization control strategy, which will include the adjustment range of experimental parameters (such as temperature, reaction time, etc.) and the recommended values ​​after adjustment. This control strategy not only considers the experimental quality, but also predicts the improvement of experimental benefits based on historical data, thereby ensuring the effectiveness of the optimization decision. Based on the intelligent experimental optimization control strategy, the automated optimization operation of the experimental quality control strategy is implemented, which involves adjusting the experimental operation according to the optimization strategy and monitoring the implementation effect. The experimental quality control system adjusts the operation of the experimental equipment according to the intelligent optimization control strategy. For example, if the strategy recommends increasing the heating temperature in the temperature control system, the temperature control system will automatically adjust the working state of the heating equipment according to the strategy instructions. Other control systems, such as reaction time, pressure, etc., will also be automatically adjusted according to the generated optimization strategy to ensure that the experimental conditions are consistent with the optimal control strategy.During the execution of the strategy, the system will continuously monitor the experimental status, including key parameters such as temperature, humidity, and reaction rate. If a control parameter fails to reach the predetermined target, the system will adjust and output new control instructions in real time to ensure that the experimental process is always carried out under optimal control. During the execution of the experiment, new experimental data (such as temperature changes, reaction rates, etc.) will be fed back to the control system to evaluate the effect of the optimization strategy. If the current strategy fails to improve the experimental benefits as expected, the system will automatically adjust the control strategy for further optimization.

[0025] Further, step S1 includes the following steps: Step S11: Obtain historical experimental quality control data of control samples; In an embodiment of the present invention, historical experimental quality control data of control samples are obtained. In order to ensure the accuracy and completeness of the data, the experimenter extracts the experimental quality control data containing detailed information such as sample batch number, experimental conditions, control data, quality standards, test values ​​and timestamps from the historical experimental database through the laboratory information management system (LIMS). These data include the measurement results of various control samples at different experimental stages, such as key control parameters such as the concentration, temperature, humidity, and pH value of the target substance. During the data acquisition process, the system will perform data cleaning according to the preset standardized format, remove outliers, and ensure the integrity of the data. Through the data import process, all historical experimental quality control data are stored in a structured form in the database, providing a reliable data basis for subsequent data analysis.

[0026] Step S12: performing control mode recognition processing according to historical experimental quality control data to generate historical experimental control mode data; In an embodiment of the present invention, control pattern recognition processing is performed based on historical experimental quality control data, and data mining technology is applied to analyze the historical experimental quality control data to identify the typical patterns of control samples in different experimental stages, and experimental quality control data with similar characteristics are classified into the same category. Through control pattern recognition, the system can automatically detect data sets that meet certain specific characteristics, such as excessive temperature and excessive pH fluctuations. The output of this process is a control pattern data set, which includes characteristic values ​​of various control patterns, such as the maximum fluctuation range, data distribution, deviation, etc. within the pattern, thereby providing key data support for subsequent cluster analysis.

[0027] Step S13: performing cluster control mode experiment quality control analysis on the historical experiment quality control data based on the historical experiment control mode data to generate cluster control mode experiment quality control data; In an embodiment of the present invention, a cluster control mode experimental quality control analysis is performed on the historical experimental quality control data based on the historical experimental control mode data, and a suitable clustering algorithm (such as a K-means clustering algorithm) is selected to input the identified historical experimental control mode data into a clustering model. By calculating the similarity of each data point, the model will divide the data into different clusters according to the similarity of the data, respectively representing the experimental quality control states corresponding to different experimental control modes. For example, a cluster subset represents a normal control state during the experiment, and another corresponding cluster subset represents a state with a large deviation during the experiment. During the clustering process, the system ensures that the results of each cluster have a high internal consistency and a low external difference by analyzing the internal differences of each cluster. Ultimately, the generated cluster control mode experimental quality control data will reflect the quality control characteristics under different experimental states, providing data support for subsequent control analysis and adjustment.

[0028] Step S14: extracting the quality control state characteristics of the cluster control mode experiment according to the cluster control mode experiment quality control data, and generating the quality control state characteristic data of the cluster control mode experiment; In an embodiment of the present invention, experimental quality control state feature extraction is performed based on experimental quality control data of cluster control mode. By extracting the key state features in each cluster, the inherent laws of cluster data are further analyzed, and statistical analysis methods such as mean, standard deviation, kurtosis and skewness are used to perform descriptive statistics on the experimental quality control data in each cluster. Based on these statistical features, the experimental state features of each cluster are extracted, such as fluctuation range, central trend, deviation, etc. The core of this process is to extract the state features that have a significant impact on the experimental results and exclude those noise features that may not affect the experimental quality control results. For example, in a certain cluster, it is found that the fluctuation of pH value has the greatest impact on the experimental results, so the fluctuation range of pH value will be used as an important experimental quality control state feature. These state feature data will be used as input for subsequent control limit analysis to help the system identify which features are crucial for controlling the quality of the experiment.

[0029] Step S15: Perform cluster control mode control limit analysis based on the cluster control mode experimental quality control state characteristic data to generate cluster control mode control limit data.

[0030] In an embodiment of the present invention, control limit analysis is performed based on the experimental quality control state characteristic data of the cluster control mode, the extracted experimental quality control state characteristic data is input into the control limit analysis model, and the control limit data of the cluster control mode is analyzed and generated, and the control limit of each cluster is calculated according to statistical methods such as normal distribution analysis, such as upper and lower limits. These control limits are calculated based on the characteristic values ​​of the experimental quality control data to ensure that the experimental data within the control range can maintain the stability of quality. For each cluster, the average value and the upper and lower limits of this type of data are calculated to determine the allowable deviation range under the cluster mode. For example, the temperature fluctuation range of a cluster may be set within the range of ±2°C, and it is considered abnormal if it exceeds this range. Ultimately, the system will generate a set of control limit data for each cluster mode, which will serve as the basis for subsequent optimization of the control strategy. Through this analysis, it is ensured that the quality control data in the experiment is always maintained within a reasonable control range, and a scientific basis is provided for the automated quality control strategy.

[0031] Further, step S13 includes the following steps: Step S131: performing cluster control mode analysis based on historical experimental control mode data to generate cluster experimental control mode data; In an embodiment of the present invention, cluster control mode analysis is performed according to historical experimental control mode data, and the historical experimental control mode data includes data of multiple experimental cycles, which involve changes in various control parameters during the experiment, such as temperature, humidity, pH value, etc. In order to realize cluster analysis, the system adopts K-means clustering algorithm. This algorithm divides the historical experimental control mode data into several classes according to the similarity of the control parameters, and each class represents a different experimental control mode. In actual operation, the data is first standardized to ensure that the data are compared under the same dimension. Then, the system selects a suitable number of clusters k, and distributes the data to the closest class by calculating the distance between each data point and the center point of each class. The result of cluster analysis is the grouping of multiple control modes, and the experimental data in these groups have high similarity, while the difference between the groups is large, so that different experimental control modes can be accurately distinguished. The key to this process is to ensure that the clustering results have good internal consistency and low external difference, and provide effective data support for subsequent quality control analysis.

[0032] Step S132: establishing a clustering experiment control mode matrix through clustering experiment control mode data; In an embodiment of the present invention, a clustering experiment control mode matrix is ​​established by clustering the experimental control mode data, and a control mode matrix is ​​generated according to the experimental data contained in each cluster. Each row of the matrix represents a clustering experimental control mode, and each column represents the characteristic value of different experimental control parameters in the cluster. For example, in a temperature control mode cluster, a column of the matrix represents the mean of the temperature parameter, and another column represents the standard deviation of the temperature parameter. The system generates a multidimensional control mode matrix by calculating the statistical characteristics (such as mean, standard deviation, maximum and minimum) of all experimental data points in each cluster. The matrix not only reflects the similarities and differences between different clusters, but also provides a clear data framework for subsequent analysis. Through this matrix, the system can quickly identify the changing trends of various control parameters under different control modes.

[0033] Step S133: Perform cluster control mode experiment quality control analysis by selecting historical experiment quality control data corresponding to cluster nodes in the cluster experiment control mode matrix to generate cluster control mode experiment quality control data.

[0034] In the embodiment of the present invention, the historical experimental quality control data corresponding to the clustering node in the clustering experiment control mode matrix is ​​selected to perform clustering control mode experiment quality control analysis, and the historical experimental data corresponding to each clustering node is identified according to the established clustering experiment control mode matrix. For example, a certain clustering node may correspond to a "normal control mode" cluster, which contains temperature, humidity and pressure data in multiple experiments. The system will extract the data set corresponding to the clustering node from the historical experimental quality control data, and these data sets will reflect the experimental state under this mode. Then, using these data, the clustering control mode is further analyzed to evaluate the quality control effect of each control mode. By comparing the stability, fluctuation range and influence of the experimental results under different control modes, the system generates clustering control mode experimental quality control data. These data reflect the experimental deviation and quality control effect under different control modes, which can help identify which control modes are the most stable in actual operation and which may have large deviations or instability. The generated clustering control mode experimental quality control data will serve as the basis for automatic optimization of the experimental quality control strategy, and support subsequent optimization decisions and control strategy adjustments.

[0035] Further, as an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S2 in FIG. 1 , in this embodiment, step S2 includes the following steps: Step S21: performing cluster control mode experiment quality control deviation evaluation processing on the cluster control mode experiment quality control data according to the cluster control mode control limit data to generate cluster control mode experiment quality control deviation data; In an embodiment of the present invention, the cluster control mode experimental quality control data is evaluated for cluster control mode experimental quality control deviation according to the cluster control mode control limit data. The control limit data includes the statistical boundary values ​​under each cluster control mode, such as the upper limit, the lower limit, the warning line, etc., which are used to determine whether the experimental results are within the allowed control range. The cluster control mode experimental quality control data is compared with these control limit data, the deviation between each data point and its corresponding control limit is calculated, and the difference calculation formula (i.e., each experimental data point minus the control limit value) is used to evaluate the deviation of all data points. If the deviation value exceeds the set threshold range, the system is marked as abnormal deviation data. During the evaluation process, the system can also use statistical methods, such as standard deviation, deviation variance, etc., to further analyze the deviation distribution of the experimental data to ensure the accuracy of the deviation evaluation. Finally, through this evaluation process, the system generates cluster control mode experimental quality control deviation data, which reflects the deviation between the experimental data and the control limit under each control mode, and provides a basis for subsequent data optimization.

[0036] Step S22: performing cluster control mode experiment control influencing parameter analysis based on cluster control mode experiment quality control data to generate cluster control mode experiment control influencing parameters; In an embodiment of the present invention, cluster control mode experimental control influencing parameter analysis is performed based on cluster control mode experimental quality control data, and the cluster control mode experimental quality control data contains multiple experimental data under different control modes, which are closely related to the experimental results. The system uses a multiple regression analysis method to extract the control parameters of the experimental process for these experimental data, and extracts control features for the experimental quality control data of each cluster, and extracts key control variables in the experiment, such as temperature, humidity, experimental time, etc. Then, the correlation analysis method is used to evaluate the relationship between these variables and the experimental results, and determine the main parameters affecting the experimental results. Through stepwise regression, principal component analysis (PCA) and other techniques, the parameters that have the greatest impact on the control mode are screened out, and cluster control mode experimental control influencing parameter data are generated. These parameter data will provide a basis for subsequent impact feature analysis and optimization decisions.

[0037] Step S23: performing cluster control mode experiment quality control influence characteristic analysis on cluster control mode experiment control influence parameters according to cluster control mode experiment quality control deviation data to obtain cluster control mode experiment quality control influence factors; In an embodiment of the present invention, cluster control mode experimental quality control influence characteristic analysis is performed on cluster control mode experimental control influencing parameters according to cluster control mode experimental quality control deviation data, and the system combines the deviation data with the experimental control influencing parameters to perform influence characteristic analysis. The purpose of this analysis is to find out which control parameter changes have a strong correlation with the deviation data, and use correlation matrix analysis to calculate the correlation coefficient between each influencing parameter and the experimental data deviation. If the correlation coefficient is high, it means that the fluctuation of the control parameter during the experiment has a greater impact on the deviation. A decision tree algorithm, random forest or other machine learning methods are used to deeply analyze the degree of influence of each control parameter on the experimental result deviation, and identify the quantitative information of the impact of the experimental control influencing parameters on quality control, thereby obtaining the cluster control mode experimental quality control influencing factor, and providing accurate control variables for subsequent optimization.

[0038] Step S24: performing an experimental quality control leaf node region analysis of the cluster control model according to the cluster control model experimental quality control influencing factors, and generating cluster control model experimental quality control leaf node region data.

[0039] In an embodiment of the present invention, the experimental quality control leaf node area analysis of the clustering control mode is performed according to the clustering control mode experimental quality control influencing factor. The clustering control mode experimental quality control influencing factor reflects which control parameters have a decisive influence on the experimental results. The preliminary leaf node area is set by the clustering control mode experimental quality control influencing factor, and the preliminary leaf node area is gradually divided by setting the influence factor adaptive division window. The influence factor adaptive division window considers that the setting can finely divide the leaf node area and use it as the next leaf node area, so as to iteratively analyze the leaf node area until the preset convergence threshold is reached, and finally the clustering control mode experimental quality control leaf node area data is obtained.

[0040] Further, step S24 includes the following steps: Step S241: designing the first experimental quality control leaf node region data according to the experimental quality control influencing factors of the cluster control mode; In the embodiment of the present invention, the cluster control mode experimental quality control influencing factors have been determined through the above steps, and these factors represent the key control parameters that affect the quality control results during the experiment, such as temperature, pressure, pH value, etc. These influencing factors will be used to set the division rules of the experimental quality control leaf node area, and the preliminary area of ​​the leaf node will be set first according to the importance of the influencing factors and the range of changes during the experiment. For example, if temperature is a key influencing factor, the system will set the preliminary division of the experimental quality control leaf node according to different temperature intervals, and divide the experimental data into multiple areas. Each area represents the experimental control state at a temperature within a specific range, and these states reflect the experimental quality under different temperature conditions. Then, the system classifies the experimental data according to the threshold setting of the influencing factor to ensure that the data under different control modes can be correctly mapped to the corresponding leaf node area. Finally, the system designs the first experimental quality control leaf node area data through this process, and these data provide a preliminary regional division basis for subsequent fine adjustment and optimization.

[0041] Step S242: setting an influencing factor adaptive partitioning window, and using the influencing factor adaptive partitioning window to perform experimental quality control sub-node region partitioning on the first experimental quality control leaf node region data to design the second experimental quality control leaf node region data, wherein the second experimental quality control leaf node region data is a sub-leaf node region of the first experimental quality control leaf node region data; In an embodiment of the present invention, an influencing factor adaptive partitioning window is set, and the influencing factor adaptive partitioning window is used to perform experimental quality control sub-node regional division on the first experimental quality control leaf node regional data, and an influencing factor adaptive partitioning window is defined based on the first experimental quality control leaf node regional data. The function of this window is to automatically adjust the fineness of regional division according to different values ​​of the influencing factor. For example, some influencing factors may change less within a specific range, and fluctuate more in other ranges. The system will analyze the changes in these influencing factors and set the size of the adaptive partitioning window so that the region can be divided more finely within the range of larger fluctuations, while maintaining a larger partition interval within the range of smaller fluctuations. The fineness of the division is automatically adjusted according to the specific distribution of the influencing factors, and the data with influencing factor values ​​falling in different intervals are divided into multiple sub-node regions. These sub-node regions are more refined than the first experimental quality control leaf node regional data, and can more accurately reflect the different quality control states under the experimental control mode. For example, if the temperature fluctuation in a certain control mode is large, the system will further refine the temperature range into multiple sub-node areas, each area corresponding to a different temperature range and corresponding experimental quality control data, so as to design the second experimental quality control leaf node area data, where the second experimental quality control leaf node area data is the sub-leaf node area of ​​the first experimental quality control leaf node area data, and these data have higher accuracy and stronger distinguishing ability.

[0042] Step S243: The second experiment quality control leaf node region data is used as the first experiment quality control leaf node region data and fed back to step S242 for iterative analysis and processing of the experiment quality control leaf node region, thereby obtaining the cluster control mode experiment quality control leaf node region data.

[0043] In an embodiment of the present invention, the second experimental quality control leaf node area data is used as the first experimental quality control leaf node area data and fed back to step S242 for iterative analysis and processing of the experimental quality control leaf node area, thereby obtaining the cluster control mode experimental quality control leaf node area data. First, the system obtains the second experimental quality control leaf node area data through step S242, and these data provide a more detailed division of the experimental quality control area. Next, the system feeds these data back to step S242 as input for iterative analysis and processing. During the iteration process, the system will further check whether the division of each sub-node area is appropriate, and adjust the parameters of the adaptive division window according to the actual distribution of the experimental data. For example, the division of some sub-node areas may be too detailed, resulting in no significant difference between adjacent areas, and the system will optimize the division by merging these similar areas. For areas with large fluctuations or changes, the system will further refine the division window to ensure that the experimental data in each area can accurately reflect the changes in the control mode. In each round of iteration, the system will continuously evaluate the division effect of the leaf node area until the optimal area division accuracy is achieved. In this process, the system uses the preset cluster control mode experimental quality control leaf node area convergence threshold to evaluate the results of each round of iteration to ensure the accuracy and operability of the experimental quality control leaf node area data. Finally, after multiple rounds of iterations, the final cluster control mode experimental quality control leaf node area data is obtained. These data can accurately reflect the experimental quality control status under each control mode and provide a basis for subsequent intelligent optimization and automatic adjustment.

[0044] Further, step S3 includes the following steps: Step S31: Based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the cluster control mode, a mapping relationship between the experimental quality control influencing factors and the experimental quality control gains of the cluster control mode is established to generate an initial experimental quality control gain tree model, and the initial experimental quality control gain tree model is subjected to model fitting parameter limit adjustment processing through the control limit data of the cluster control mode to generate an adjusted experimental quality control gain tree model; In an embodiment of the present invention, a mapping relationship between experimental quality control influencing factors and experimental quality control gains is established based on a preset gradient boosting decision tree (GBDT) algorithm and clustering control mode experimental quality control leaf node area data, and the experimental quality control leaf node area data obtained from cluster analysis are used. These data contain the influence of different control parameters (such as temperature, humidity, pressure, etc.) on the experimental quality during the experiment under each control mode. The gradient boosting decision tree algorithm is adopted, which is a powerful machine learning algorithm based on the idea of ​​ensemble learning. The tree model parameters in each clustering mode are designed by clustering control mode experimental quality control leaf node area data, such as a learning rate of 0.05. The learning rate controls the contribution of each tree to the model. A lower learning rate helps to improve the generalization ability of the model. The number of trees is 100, indicating that 100 decision trees will be constructed in the model. The number of trees is set to 100 to ensure sufficient complexity to capture the relationship in the experimental data. The maximum depth is 6, and the maximum depth of each tree is set to 6 to avoid overfitting and ensure that the model can be generalized under different experimental modes. The minimum number of sample splits is 20, which is used to control the splitting rules of the tree. It is set to 20 to prevent some too small nodes from having an excessive impact on the results. The minimum number of sample leaf nodes is 10, which ensures that each leaf node contains a sufficient number of samples, further reducing the possibility of overfitting. The algorithm constructs multiple decision trees and gradually adjusts the weight of each tree so that the final model can accurately predict the experimental quality control gain. By setting appropriate input features (such as temperature, pressure and other control factors) and target variables (experimental quality control gain), the gradient boosting decision tree is used to model the relationship between these features and gains. During the construction of each tree, the gain of each decision node is repeatedly calculated through the experimental quality control leaf node area data of the cluster control mode, and the prediction accuracy of the model is improved by the weighted average method to establish the initial experimental quality control gain tree model. In order to optimize the accuracy of the model, the control limit data of the cluster control mode is used to adjust the parameters of the initial experimental quality control gain tree model. The control limit data of the cluster control mode provides the system with the control parameter range or threshold allowed in each experimental mode. These control limit data help the model understand which experimental control states are "normal" or "abnormal". On this basis, the system adjusts the parameters of the initial gain tree model and further optimizes the split nodes and gain values ​​of the decision tree to obtain the adjusted experimental quality control gain tree model. The adjusted model can more accurately reflect the control mode in the experimental process and its impact on the experimental quality control results.

[0045] Step S32: Based on the historical experimental quality control data, the optimal fitting training process of the experimental quality control leaf node region is performed on the adjusted experimental quality control gain tree model to generate the experimental quality control gain tree model.

[0046] In an embodiment of the present invention, historical experimental quality control data are collected and sorted, including control parameters (such as temperature, humidity, pressure, etc.) of each experiment, experimental results (such as output, accuracy, qualified rate, etc.) and corresponding experimental quality control gains (such as qualified rate, experimental error rate, etc.). These data are used as training sets. The historical experimental data are cleaned and processed to remove outliers and missing values ​​to ensure the accuracy and consistency of the data. Each experimental result in the historical data and its corresponding control conditions (such as temperature, humidity, etc.) are used as input features, and the experimental gain (such as qualified rate, efficiency, etc.) is used as the target output. The relationship between the input feature and the target output will be used to train the model. The adjusted experimental quality control gain tree model is used for training. According to the historical experimental data and the initial parameters of the adjusted experimental quality control gain tree model, the system starts the training process to optimize the various parameters in the model to ensure the best fit. Part of the data in the historical experimental quality control data (usually 70% to 80%) is used for the training process, and the remaining data is used to verify the prediction effect of the model. The adjusted experimental quality control gain tree model (i.e., the tree model that contains the control limit data of the cluster control mode and the optimal parameter settings) is trained using the gradient boosting decision tree algorithm (GBDT). Each time the model is trained, the prediction error is gradually corrected to finally obtain the optimized gain tree. During the training process, GBDT improves the performance of the model by calculating the error of each tree, gradually reduces the training error, and finally achieves the best fit. During the training process, the model gradually adjusts the division of each leaf node area according to the data in the training set. Each leaf node area represents a set of control conditions with similar experimental quality control gains (for example, experimental conditions under a certain temperature and humidity). The system ensures that each leaf node contains the optimal experimental conditions and their corresponding experimental gains by adjusting the splitting rules of each leaf node. At this point, the system performs the following operations: Based on the data in the training set, the system optimizes the division of the experimental quality control leaf node area. At each split, the system selects the optimal segmentation feature based on the distribution of the data and the change trend of the gain. For example, a leaf node can represent an experimental condition at a specific temperature, while another leaf node represents the experimental results under different humidity conditions. The optimization process makes decisions by selecting features that minimize the error and maximize the gain. In each leaf node area, the system calculates the average experimental quality control gain of the area as the predicted output of the node. If the experimental gain value fluctuates greatly in some areas, the model will be adjusted according to the training data to ensure that the output of each leaf node is as stable as possible. In the optimal fitting process, the division method of the leaf node area is cross-validated to further verify the fitting effect of the model and ensure its performance on the validation set is optimal. The system avoids overfitting or underfitting by adjusting parameters such as the learning rate, the depth of the tree, and the number of features of the split node. Standard evaluation indicators (such as mean square error MSE, R², accuracy, etc.) are used to evaluate the fitting effect of the model.If the performance of the model on the validation set meets the predetermined standard, the model is considered to have achieved optimal fit. After the training is completed, the system outputs the final experimental quality control gain tree model. The model can predict the experimental quality control gain value based on the experimental control conditions and can be used for subsequent experimental optimization control strategies.

[0047] Further, step S32 includes the following steps: Step S321: transferring the historical experimental quality control data to the adjusted experimental quality control gain tree model for experimental quality control benefit model training processing to generate experimental quality control benefit model training data; In an embodiment of the present invention, historical experimental data is transmitted to the adjustment experiment quality control gain tree model for training. The goal of the adjustment experiment quality control gain tree model is to optimize the data accuracy of the prediction between the control parameters and the experimental benefits by training historical data. The input of the model includes various experimental control parameters, and the output is the experimental benefit index. During the training process, the model will analyze the relationship between the control parameters and the experimental results based on historical data. By using regression algorithms, decision trees and other techniques, the adjustment experiment quality control gain tree model continuously adjusts its parameters and gradually finds the key factors that affect the experimental benefits. After the training is completed, the experimental quality control benefit model training data generated by the adjustment experiment quality control gain tree model will include the effect prediction results of each experimental configuration and its corresponding benefit score.

[0048] Step S322: performing correlation model training data analysis based on the experimental quality control benefit model training data to generate experimental quality control benefit correlation model training data; In an embodiment of the present invention, the association model training data analysis is performed based on the experimental quality control benefit model training data, the nonlinear relationship between different experimental control parameters and experimental benefit indicators in the experimental quality control benefit model training data is evaluated using Spearman rank correlation, and the association rule mining method (such as Apriori algorithm or FP-growth algorithm) is used to reveal the correlation and influence between the control parameters. For example, the experimental data may show that when the temperature is higher than a certain threshold, the experimental success rate under certain specific humidity ranges is higher. In this way, the system can identify the potential association rules between experimental control conditions and benefits, generate association model training data, including each control parameter combination and its corresponding benefit performance, and analyze to find out the applicable scope of these rules under different experimental conditions, and use feature selection methods (such as LASSO, decision tree, information gain, etc.) to perform feature screening to find out the control factors that have the greatest impact on the experimental benefits. At the same time, dimensionality reduction techniques (such as PCA) are used to reduce the data dimension, further improve the efficiency of model analysis, and finally generate experimental quality control benefit association model training data.

[0049] Step S323: performing experimental quality control correlation gradient improvement feature analysis based on the experimental quality control benefit correlation model training data to generate experimental quality control correlation gradient improvement feature data; In an embodiment of the present invention, based on the experimental quality control benefit association model training data, the experimental quality control gain tree model is used to perform feature analysis. The experimental quality control gain tree model is adjusted to effectively discover the complex relationship between different experimental control parameters and experimental benefits through the model characteristics of the gradient boosting decision tree, and identify the global gradient boosting important features of the experimental quality control of the experimental quality control benefit association model training data. The contribution of each control feature of the experimental quality control benefit association model training data to the global experimental benefit is gradually learned. Each feature represents the weight of the influence of different control conditions on the experimental benefit, and a feature importance ranking is automatically generated to reveal the degree of influence of each control parameter on the global experimental benefit, identify the most influential features, and generate experimental quality control associated gradient boosting feature data. These data will provide input for model optimization in subsequent steps, help build a more accurate experimental quality control gain tree model, and the experimental quality control associated gradient boosting feature data includes the weight of each control parameter, the importance score, and their specific impact on the experimental benefit.

[0050] Step S324: performing optimal fitting training processing on the experimental quality control leaf node region of the adjusted experimental quality control gain tree model through the experimental quality control associated gradient lifting feature data to generate the experimental quality control gain tree model.

[0051] In an embodiment of the present invention, based on the experimental quality control associated gradient boosting feature data, the experimental quality control gain tree model is continuously optimized to complete the optimal fitting training process of the experimental quality control leaf node region. The goal of adjusting the experimental quality control gain tree model is to further refine the division of the leaf node region through the associated feature data to ensure that the model can accurately predict the benefits under different experimental conditions, and adjust the leaf node splitting rules of the model according to the gradient boosting feature data. Each leaf node represents a group of similar experimental control conditions and their corresponding experimental benefits. Through repeated training and adjustment, the model will learn the changing law of benefits under different control conditions, thereby generating a leaf node region with the best fitting effect. For example, assuming that under a certain specific control condition, the qualified rate of the experiment performs best, the system will optimize and adjust the leaf node corresponding to the control condition to ensure that the output result of the node is as accurate as possible. By continuously optimizing the division of the leaf nodes, it is used to find the optimal fitting value in the region, and finally generate a highly accurate experimental quality control gain tree model, which can predict the benefits and optimization direction of the experiment according to the input experimental control conditions.

[0052] Further, step S4 includes the following steps: Step S41: using a sensor monitoring device to perform real-time monitoring processing on the control sample during the experiment to obtain experimental monitoring data; In an embodiment of the present invention, the real-time monitoring of the experimental process is completed by a group of sensors installed on the experimental equipment, and the types of sensors include monitoring sensors, temperature sensors, humidity sensors, pressure sensors and reaction rate sensors, etc. The sensor system needs to be closely integrated with the experimental device, and can continuously collect and record experimental data during the experiment. For example, during the experiment, it is necessary to monitor the temperature and humidity conditions of the sample in real time through the temperature sensor and the humidity sensor, and collect the rate data of the chemical reaction in real time through the reaction rate sensor. The sampling frequency of each sensor is set according to the experimental requirements, for example, the sampling frequency of the temperature and humidity sensor can be set to once per second, and the sampling frequency of the reaction rate sensor can be set to once per minute, to ensure the accuracy and timeliness of various data during the experiment. The collected sensor data is processed in real time by the data acquisition device, and the data is transmitted to the data processing platform in real time to obtain the experimental monitoring data. For example, the temperature change data monitored by the temperature sensor will be transmitted to the central data processing system in real time for storage and analysis.

[0053] Step S42: extracting real-time experiment quality control data according to the experiment monitoring data to generate real-time experiment quality control data, wherein the real-time experiment quality control data includes real-time experiment control data and real-time experiment quality data; In an embodiment of the present invention, data extraction and processing are performed based on real-time monitoring data, and the instant experiment quality control data mainly includes two types of information: instant experiment control data and instant experiment quality data. Experiment control data includes various control parameters used in the experiment, such as temperature, humidity, pressure, reaction time, etc. By extracting the current values ​​of these control parameters from the real-time monitoring data, the instant experiment control data is constructed, and the temperature value monitored in real time by the temperature sensor and the humidity value provided by the humidity sensor are integrated into an instant experiment control data set, indicating the specific control parameters of the experiment in progress. Experiment quality data refers to the quality indicators obtained by monitoring the reaction process, sample characteristics, etc., including information such as reaction rate, product concentration, and sample color change. For example, the real-time reaction rate obtained by the reaction rate sensor and the product concentration obtained by chemical analysis will be extracted and integrated into an instant experiment quality data set. In this way, the quality indicators of the experiment will be reflected in the experimental data in real time, which is convenient for real-time evaluation of the progress and quality of the experiment. The extracted control data and quality data are integrated according to the timestamp to form a complete instant experiment quality control data set.

[0054] Step S43: transmitting the real-time experiment quality data to the experiment quality control gain tree model to perform intelligent experiment optimization control strategy analysis and generate an intelligent experiment optimization control strategy; In an embodiment of the present invention, the real-time experimental quality data is input into a pre-trained experimental quality control gain tree model, and the real-time experimental quality data generated in real time will be transmitted to the experimental quality control gain tree model through a data interface. The model has been trained based on historical experimental data and experimental optimization strategies, and can perform real-time decision analysis based on the input data. For example, real-time temperature, humidity, reaction rate and other data will be input into the model as input, and the model uses these input data for reasoning to generate an optimized control strategy. After the data is input, the experimental quality control gain tree model analyzes the real-time data and evaluates the impact of the experimental control parameters on the experimental quality. According to the current experimental state, the model will calculate a control strategy that can maximize the experimental benefits. For example, if the model analysis finds that the current temperature is low, resulting in a slow reaction rate, the model may output a control strategy that recommends increasing the temperature to accelerate the reaction process, thereby improving the efficiency of the experiment. Based on the analysis results, the experimental quality control gain tree model will output an optimization control strategy for the current experimental state, namely the intelligent experiment optimization control strategy, which will conduct real-time experiments, including the adjustment range of changing experimental parameters (such as temperature, reaction time, etc.), as well as the recommended values ​​after adjustment. This control strategy not only considers the experimental quality, but also predicts the improvement of experimental benefits based on historical data, thereby ensuring the effectiveness of the optimization decision.

[0055] Step S44: Execute the automated optimization operation of the experimental quality control strategy according to the intelligent experimental optimization control strategy.

[0056] In an embodiment of the present invention, based on the intelligent experiment optimization control strategy, the automated optimization operation of the experimental quality control strategy is implemented, which involves adjusting the experimental operation according to the optimization strategy and monitoring the implementation effect. The experimental quality control system adjusts the operation of the experimental equipment according to the intelligent optimization control strategy. For example, if the strategy recommends increasing the heating temperature in the temperature control system, the temperature control system will automatically adjust the working state of the heating equipment according to the strategy instructions. Other control systems, such as reaction time, pressure, etc., will also be automatically adjusted according to the generated optimization strategy to ensure that the experimental conditions are consistent with the optimal control strategy. During the execution of the strategy, the system will continue to monitor the experimental status, including key parameters such as temperature, humidity, and reaction rate. If a control parameter fails to reach the predetermined target, the system will adjust and output a new control instruction in real time to ensure that the experimental process is always carried out under optimal control. During the execution of the experiment, new experimental data (such as temperature changes, reaction rates, etc.) will be fed back to the control system to evaluate the effect of the optimization strategy. If the current strategy fails to improve the experimental benefits as expected, the system will automatically adjust the control strategy for further optimization.

[0057] Further, step S43 includes the following steps: Step S431: performing an experiment quality optimization demand feature analysis on the real-time experiment quality data through the experiment quality control gain tree model to generate experiment quality optimization demand feature data; In an embodiment of the present invention, the real-time experimental quality data collected in real time is input into the trained experimental quality control gain tree model for quality optimization demand analysis. The goal of this process is to analyze the potential problems in the experimental quality data and determine the specific characteristics of the experimental optimization requirements. The real-time experimental quality data includes key quality indicators in the experiment, such as reaction rate, product concentration, physical properties of the sample (such as color, particle size, etc.) and other quality characteristics related to the experimental results. These data are extracted from the real-time monitoring system and input into the gain tree model after being standardized. During the transmission process, the input data needs to be further optimized by the feature selection algorithm to ensure that the input quality data has sufficient representativeness and accuracy for the model to perform effective analysis. In the gain tree model, each leaf node represents a control decision process of the experiment. Through the analysis of the real-time experimental quality data, the gain tree model will calculate the relationship between the data and the experimental quality, and evaluate the potential demand for quality optimization under the current experimental state. For example, if the product concentration of the experimental reaction is low, the model will analyze its relationship with control parameters such as temperature and reaction time, and determine whether these control parameters need to be adjusted to improve the reaction efficiency. The gain tree model generates corresponding demand feature data based on the input quality data. The feature data will reveal the key control points for quality improvement in the current state of the experiment. For example, a low reaction rate may be due to too low temperature or insufficient reaction time. After the gain tree model completes the analysis of the data, the model output contains feature data for the experimental quality optimization requirements. This data includes specific demand characteristics of quality indicators, such as temperature control range, time extension amplitude, etc. The feature data will be transmitted to the subsequent steps as a basis for further generating intelligent optimization control strategies.

[0058] Step S432: Perform intelligent analysis of the experiment optimization control strategy on the real-time experiment control data according to the experiment quality optimization demand characteristic data to generate an intelligent experiment optimization control strategy.

[0059] In an embodiment of the present invention, the experimental quality optimization demand feature data is combined with the real-time experimental control data to perform intelligent analysis on the experimental optimization control strategy, and the experimental quality optimization demand feature data and the real-time experimental control data will be integrated into an input data set. The experimental quality optimization demand feature data provides detailed information on the current quality problems of the experiment, while the real-time experimental control data reflects the current operating status of the experiment (such as temperature, humidity, reaction rate, etc.). After combining these two types of data, a complete set of input data is formed and transmitted to the intelligent optimization analysis system. Machine learning algorithms such as gradient boosting decision tree (GBDT) are used to perform intelligent analysis on the integrated data. The core of this process is to generate corresponding control decisions based on the quality optimization demand feature data by analyzing the current experimental control data. For example, the model analysis finds that the concentration of the reaction product is lower than expected. Combined with control data such as temperature and reaction time, the system will determine whether it is necessary to adjust the temperature or extend the reaction time to increase the product concentration. In the intelligent analysis process, the model learns and identifies specific control rules through historical experimental data, and generates experimental optimization control strategies based on these rules. The optimized experimental control strategy will be generated, specifically including adjustment suggestions for control parameters such as temperature, reaction time, and pressure. For example, if the model finds that the temperature is not high enough to support the increase in reaction rate, the generated control strategy will recommend increasing the temperature and extending the reaction time to promote the reaction process. In addition, the intelligent optimization control strategy will also take into account the constraints of the experiment, such as the maximum heating capacity of the equipment and the volume of the reaction vessel, to ensure that the strategy can not only improve the quality of the experiment, but also be implemented within the operational range.

[0060] This specification provides a system for automatically optimizing the quality control strategy of experimental analysis, which is used to execute the automatic optimization method of the quality control strategy of experimental analysis as described above. The automatic optimization of the quality control strategy of experimental analysis includes: The experimental control limit analysis module is used to obtain the historical experimental quality control data of the control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; The experimental quality control impact analysis module is used to evaluate the experimental quality control deviation of the cluster control mode based on the control limit data of the cluster control mode, and generate the experimental quality control deviation data of the cluster control mode; to analyze the experimental quality control impact factor of the cluster control mode based on the experimental quality control deviation data of the cluster control mode, and generate the experimental quality control impact factor of the cluster control mode; to analyze the experimental quality control leaf node area of ​​the cluster control mode based on the experimental quality control impact factor of the cluster control mode, and generate the experimental quality control leaf node area data of the cluster control mode; The experimental quality control gain tree model construction module is used to design the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generate the adjustment experimental quality control gain tree model; based on the historical experimental quality control data, the experimental quality control leaf node area of ​​the adjustment experimental quality control gain tree model is subjected to the optimal fitting training processing to generate the experimental quality control gain tree model; The intelligent optimization module of experimental quality control strategy is used to use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; based on the experimental quality control gain tree model, the intelligent experimental optimization control strategy is analyzed to generate an intelligent experimental optimization control strategy; and the experimental quality control strategy is automatically optimized according to the intelligent experimental optimization control strategy.

[0061] The beneficial effect of this application is that, by obtaining the historical experimental quality control data of the control samples, the necessary basic data is provided for the entire quality control strategy optimization process. These historical data record the quality control of various control samples in the experiment, reflecting the potential systematic problems and variation factors in the experiment. By collecting and analyzing historical experimental data, data support is provided for the subsequent optimization of quality control strategies, so that the control strategy can be accurately adjusted according to the real experimental data, reducing the dependence on assumptions and prediction models. Based on the historical experimental quality control data, control pattern recognition processing is performed to identify the potential control mode under different experimental conditions, and a stable and representative control mode can be extracted from a large amount of historical data, and the potential trend or problem in the data can be identified, thereby providing a clear direction for the subsequent optimization analysis, avoiding blind adjustment, and ensuring the scientificity and pertinence of the experimental control. Based on the historical experimental control mode data, clustering control mode experimental quality control analysis is performed, and the experimental data is effectively divided into multiple cluster groups with similar characteristics. Different clustering patterns represent different control states or characteristics in the experiment, which can reveal the regularity of data changes under different experimental conditions. Through cluster analysis, experimental data under different modes can be described and processed more accurately, the control scheme can be optimized, and more effective intervention can be carried out during the experiment. Experimental quality control state feature extraction is performed on the experimental quality control data of the cluster control mode, and the key features in each cluster mode are further extracted to help identify which experimental states have a decisive impact on the quality of the experimental results, accurately capture the key variables in the experiment, and avoid misjudgment or omission caused by over-reliance on a single indicator in traditional quality control methods. Based on the experimental quality control state feature data of the cluster control mode, the control limit analysis of the cluster control mode is performed, and the corresponding control limit is set for each control mode in the experiment. According to the experimental characteristics under different control modes, a scientific and reasonable control limit value can be formulated for each cluster mode, and the quality deviation during the experiment is ensured not to exceed the predetermined range. By accurately setting the control limit, not only can the experimental deviation be discovered in time, but also the negative impact of too loose or too strict control limits on the experimental results can be avoided. According to the control limit data of the cluster control mode, the deviation of the experimental quality control data of the cluster control mode is evaluated to evaluate whether the experimental data is abnormal, and effectively identify the quality problems in the data, such as whether the experimental data exceeds the predetermined control limit or whether there is abnormal fluctuation. By analyzing the experimental quality control parameters based on the experimental quality control data of the cluster control mode, it is possible to extract the key parameters that affect the experimental quality from different clustering modes, perform parameterized analysis on the quality control data under different experimental conditions, and find out which experimental parameters have a significant impact on the quality of the results, so as to optimize them in a targeted manner. By establishing a more sophisticated experimental quality control parameter model, researchers can adjust the experimental conditions in a targeted manner to avoid affecting the experimental results due to improper parameter settings, thereby improving the repeatability and accuracy of the experiment.According to the experimental quality control deviation data of the cluster control mode, the influence characteristics of the experimental quality control parameters are analyzed to further identify the key factors affecting the experimental results, effectively reveal the relationship between the variables under different experimental control modes, help researchers identify which factors have a greater impact on the experimental results, and provide a basis for further adjustments. By focusing on these influencing factors, unnecessary interference factors are eliminated to improve the stability and accuracy of the experiment. According to the experimental quality control influencing factors of the cluster control mode, the experimental quality control leaf node area analysis is carried out to achieve fine division and adjustment of the experimental quality control area. According to the different control modes and characteristics of the influencing factors of the experiment, the leaf node area of ​​the experimental quality control is accurately determined, providing a precise operating range for automatic adjustment, and the experimental data is more finely segmented. And a personalized control strategy is formulated for each leaf node area. Through this customized control area analysis, the experimental quality control strategy can be adjusted more flexibly, the experimental conditions can be optimized, and the accuracy and stability of the experimental results can be improved. Based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, the mapping relationship between the experimental quality control influencing factors and the experimental quality control gains is established. The gradient boosting decision tree algorithm can efficiently learn the nonlinear relationship between each influencing factor and the quality control gain from a large amount of experimental data, so that the system can capture the complex laws hidden behind the data. After generating the initial experimental quality control gain tree model, the model fitting parameters are adjusted in combination with the control limit data of the clustering control mode to further optimize the accuracy and robustness of the model. The parameters of the model can be automatically adjusted to ensure that the gain tree model is more in line with the distribution and control requirements of the actual experimental data, thereby providing a more accurate decision-making basis for the intelligent adjustment of experimental quality control. Based on the historical experimental quality control data, the optimal fitting training of the adjusted experimental quality control gain tree model is carried out. As the real experimental feedback, the historical experimental data can effectively provide the necessary training data for the gain tree model to help the model to be further adjusted and optimized. Through the optimal fitting training, the gain tree model can capture the subtle changes and relationships of various factors in the experiment on the basis of ensuring high accuracy, making the model more flexible and efficient when facing actual experimental data. The resulting experimental quality control gain tree model can provide more accurate and reliable decision support for the intelligent optimization control strategy during the experiment, ensuring that the quality control during the experiment can be adjusted and optimized in a timely manner, and more accurately predicting the changes in experimental results, greatly improving the automation and intelligence level of experimental quality control, and promoting the development of experimental operations in a more efficient and accurate direction. Sensor monitoring equipment is used to monitor the experimental process of control samples in real time to ensure dynamic tracking of experimental data. Through real-time data collection by sensor equipment, the status of the samples, environmental parameters and operating conditions are continuously monitored during the experiment, thereby providing instant feedback on key factors in the experimental process.According to the experimental monitoring data, the real-time experimental quality control data is extracted, including the real-time experimental control data and the real-time experimental quality data, so that the real-time monitoring data can be converted into actionable quality control information. The real-time experimental quality data is transmitted to the experimental quality control gain tree model for intelligent experimental optimization control strategy analysis, and the experimental optimization control strategy is generated. The intelligent analysis function of the experimental quality control gain tree model is used to deeply process the real-time quality data, automatically identify the optimization space in the experimental process, and generate a suitable control strategy. Through intelligent analysis, the system can combine historical data and real-time data to automatically adjust the experimental parameters or control conditions. Intelligent optimization not only reduces manual intervention and avoids errors caused by human judgment, but also can achieve precise control during the experiment, ensuring that the experimental results are more stable and efficient. According to the intelligent experimental optimization control strategy, the experimental quality control strategy is automatically optimized, which can simulate the execution of the experiment, verify the effects of different control strategies, optimize the strategies, and improve the automation level of the experimental strategy. Therefore, the automatic optimization method of the quality control strategy of the experimental analysis of the present invention can realize the effective recognition of complex data patterns, establish an effective control mode and optimize the experimental conditions and perform real-time monitoring and automatic adjustment during the experiment, etc., thereby realizing the optimization of the experimental quality control strategy through the automatic control of the monitoring of experimental data, and realizing integrated and comprehensive intelligent quality control strategy optimization.

[0062] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0063] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for automatically optimizing the quality control strategy of experimental analysis, characterized in that: The following steps are involved: Step S1: Acquire historical experimental quality control data of control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; Step S2: performing cluster control mode experiment quality control deviation evaluation processing on cluster control mode experiment quality control data according to cluster control mode control limit data to generate cluster control mode experiment quality control deviation data; performing cluster control mode experiment quality control influencing factor analysis processing based on cluster control mode experiment quality control deviation data to generate cluster control mode experiment quality control influencing factor; According to the cluster control mode experimental quality control influencing factors, the cluster control mode experimental quality control leaf node area analysis is performed to generate cluster control mode experimental quality control leaf node area data; Step S3: Designing the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generating the adjustment experimental quality control gain tree model; performing the optimal fitting training processing of the experimental quality control leaf node area on the adjustment experimental quality control gain tree model based on the historical experimental quality control data, and generating the experimental quality control gain tree model; Step S4: Use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; perform intelligent experimental optimization control strategy analysis on the experimental monitoring data based on the experimental quality control gain tree model to generate an intelligent experimental optimization control strategy; execute the experimental quality control strategy automation optimization operation according to the intelligent experimental optimization control strategy.

2. The method for automatic optimization of quality control strategy for experimental analysis according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Obtain historical experimental quality control data of control samples; Step S12: performing control mode recognition processing according to historical experimental quality control data to generate historical experimental control mode data; Step S13: performing cluster control mode experiment quality control analysis on the historical experiment quality control data based on the historical experiment control mode data to generate cluster control mode experiment quality control data; Step S14: extracting the quality control state characteristics of the cluster control mode experiment according to the cluster control mode experiment quality control data, and generating the quality control state characteristic data of the cluster control mode experiment; Step S15: Perform cluster control mode control limit analysis based on the cluster control mode experimental quality control state characteristic data to generate cluster control mode control limit data.

3. The method for automatic optimization of quality control strategy for experimental analysis according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: performing cluster control mode analysis based on historical experimental control mode data to generate cluster experimental control mode data; Step S132: establishing a clustering experiment control mode matrix through clustering experiment control mode data; Step S133: Perform cluster control mode experiment quality control analysis by selecting historical experiment quality control data corresponding to cluster nodes in the cluster experiment control mode matrix to generate cluster control mode experiment quality control data.

4. The method for automatic optimization of quality control strategy for experimental analysis according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing cluster control mode experiment quality control deviation evaluation processing on the cluster control mode experiment quality control data according to the cluster control mode control limit data to generate cluster control mode experiment quality control deviation data; Step S22: performing cluster control mode experiment control influencing parameter analysis based on cluster control mode experiment quality control data to generate cluster control mode experiment control influencing parameters; Step S23: performing cluster control mode experiment quality control influence characteristic analysis on cluster control mode experiment control influence parameters according to cluster control mode experiment quality control deviation data to obtain cluster control mode experiment quality control influence factors; Step S24: performing an experimental quality control leaf node region analysis of the cluster control model according to the cluster control model experimental quality control influencing factors, and generating cluster control model experimental quality control leaf node region data.

5. The method for automatic optimization of quality control strategy for experimental analysis according to claim 4, characterized in that: Step S24 includes the following steps: Step S241: designing the first experimental quality control leaf node region data according to the experimental quality control influencing factors of the cluster control mode; Step S242: setting an influencing factor adaptive partitioning window, and using the influencing factor adaptive partitioning window to perform experimental quality control sub-node region partitioning on the first experimental quality control leaf node region data to design the second experimental quality control leaf node region data, wherein the second experimental quality control leaf node region data is a sub-leaf node region of the first experimental quality control leaf node region data; Step S243: The second experiment quality control leaf node region data is used as the first experiment quality control leaf node region data and fed back to step S242 for iterative analysis and processing of the experiment quality control leaf node region, thereby obtaining the cluster control mode experiment quality control leaf node region data.

6. The method for automatic optimization of quality control strategy for experimental analysis according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: Based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the cluster control mode, a mapping relationship between the experimental quality control influencing factors and the experimental quality control gains of the cluster control mode is established to generate an initial experimental quality control gain tree model, and the initial experimental quality control gain tree model is subjected to model fitting parameter limit adjustment processing through the control limit data of the cluster control mode to generate an adjusted experimental quality control gain tree model; Step S32: Based on the historical experimental quality control data, the optimal fitting training process of the experimental quality control leaf node region is performed on the adjusted experimental quality control gain tree model to generate the experimental quality control gain tree model.

7. The method for automatic optimization of quality control strategy for experimental analysis according to claim 6, characterized in that: Step S32 includes the following steps: Step S321: transferring the historical experimental quality control data to the adjusted experimental quality control gain tree model for experimental quality control benefit model training processing to generate experimental quality control benefit model training data; Step S322: performing correlation model training data analysis based on the experimental quality control benefit model training data to generate experimental quality control benefit correlation model training data; Step S323: performing experimental quality control correlation gradient improvement feature analysis based on the experimental quality control benefit correlation model training data to generate experimental quality control correlation gradient improvement feature data; Step S324: performing optimal fitting training processing on the experimental quality control leaf node region of the adjusted experimental quality control gain tree model through the experimental quality control associated gradient lifting feature data to generate the experimental quality control gain tree model.

8. The method for automatic optimization of quality control strategy for experimental analysis according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: using a sensor monitoring device to perform real-time monitoring processing on the control sample during the experiment to obtain experimental monitoring data; Step S42: extracting real-time experiment quality control data according to the experiment monitoring data to generate real-time experiment quality control data, wherein the real-time experiment quality control data includes real-time experiment control data and real-time experiment quality data; Step S43: transmitting the real-time experiment quality data to the experiment quality control gain tree model to perform intelligent experiment optimization control strategy analysis and generate an intelligent experiment optimization control strategy; Step S44: Execute the automated optimization operation of the experimental quality control strategy according to the intelligent experimental optimization control strategy.

9. The method for automatic optimization of quality control strategy for experimental analysis according to claim 8, characterized in that: Step S43 includes the following steps: Step S431: performing an experiment quality optimization demand feature analysis on the real-time experiment quality data through the experiment quality control gain tree model to generate experiment quality optimization demand feature data; Step S432: Perform intelligent analysis of the experiment optimization control strategy on the real-time experiment control data according to the experiment quality optimization demand characteristic data to generate an intelligent experiment optimization control strategy.

10. An automatic optimization system for quality control strategy of experimental analysis, characterized in that: The method for automatically optimizing the quality control strategy for performing the experimental analysis according to claim 1, wherein the automatic optimization of the quality control strategy for the experimental analysis comprises: The experimental control limit analysis module is used to obtain the historical experimental quality control data of the control samples; perform cluster control mode experimental quality control analysis based on the historical experimental quality control data to generate cluster control mode experimental quality control data; perform cluster control mode control limit analysis based on the cluster control mode experimental quality control data to generate cluster control mode control limit data; The experimental quality control impact analysis module is used to evaluate the experimental quality control deviation of the cluster control mode based on the control limit data of the cluster control mode, and generate the experimental quality control deviation data of the cluster control mode; to analyze the experimental quality control impact factor of the cluster control mode based on the experimental quality control deviation data of the cluster control mode, and generate the experimental quality control impact factor of the cluster control mode; to analyze the experimental quality control leaf node area of ​​the cluster control mode based on the experimental quality control impact factor of the cluster control mode, and generate the experimental quality control leaf node area data of the cluster control mode; The experimental quality control gain tree model construction module is used to design the experimental quality control gain decision tree model of the clustering control mode based on the preset gradient boosting decision tree algorithm and the experimental quality control leaf node area data of the clustering control mode, and generate the adjustment experimental quality control gain tree model; based on the historical experimental quality control data, the experimental quality control leaf node area of ​​the adjustment experimental quality control gain tree model is subjected to the optimal fitting training processing to generate the experimental quality control gain tree model; The intelligent optimization module of experimental quality control strategy is used to use sensor monitoring equipment to monitor and process the experimental process of control samples in real time to obtain experimental monitoring data; based on the experimental quality control gain tree model, the intelligent experimental optimization control strategy is analyzed to generate an intelligent experimental optimization control strategy; and the experimental quality control strategy is automatically optimized according to the intelligent experimental optimization control strategy.

Images