Processing material management system and method

Abstract

The invention provides a processing material management system and method, and relates to the field of material management. The system comprises a data acquisition module, a data analysis module and a material regulation and control module, the data acquisition module is used for collecting material processing information on a production line in real time; the data analysis module is used for analyzing the material processing information to obtain a material management and control scheme; and the material regulation and control module is used for managing and controlling the materials on the production line according to the material management and control scheme. Through the scheme, refined management of the processed materials can be realized.

Description

Technical Field

[0001] The present application relates to the field of material management, and in particular to a processing material management system and method. Background Art

[0002] With the rapid development of industrial manufacturing, the types and quantities of processed materials are increasing, and production processes are becoming increasingly complex. How to achieve refined management of materials and ensure production quality and efficiency has become a problem that manufacturing companies need to solve urgently. At present, material management on the production line mainly relies on manual experience and qualitative judgment. Operators manually mix various raw and auxiliary materials according to production plans, set equipment parameters based on experience, and then control product quality through sampling inspection. This traditional management model makes it difficult to achieve refined management of processed materials and improve production efficiency. Summary of the invention

[0003] The present application provides a processing material management system and method, which can realize refined management of processing materials.

[0004] In a first aspect, the present application provides a material management system, the system comprising: a data collection module, a data analysis module and a material control module; Data acquisition module, used to collect material processing information on the production line in real time; Data analysis module, used to analyze material processing information to obtain material control solutions; The material control module is used to control the materials on the production line according to the material control plan.

[0005] By adopting the above technical solutions, the automation and intelligence level of the production process has been significantly improved through data-driven, intelligent analysis and precise execution. The data acquisition module lays the data foundation for analysis and optimization, revealing the overall picture of production; the data analysis module mines the data value and guides production decisions with quantitative and optimized solutions; the material control module refines and implements the solutions and accurately controls the production process. The three modules are closely linked and indispensable, and together they build a solid framework for intelligent management.

[0006] In a second aspect of the present application, a processing material management method is provided, the method comprising: Collect material processing information on the production line in real time; Analyze material processing information to obtain material control plans; Control materials on the production line according to the material control plan.

[0007] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This application significantly improves the automation and intelligence level of the production process through data-driven, intelligent analysis and precise execution. The data acquisition module lays the data foundation for analysis and optimization, revealing the overall picture of production; the data analysis module mines the value of data and guides production decisions with quantitative and optimized solutions; the material control module refines and implements the solution and accurately controls the production process. The three modules are closely linked and indispensable, and together they build a solid framework for intelligent management.

[0008] 2. This application realizes the systematic analysis and high-quality control of material processing information by setting up a data analysis module including a data preprocessing unit, a statistical analysis unit and an optimization decision unit. Among them, the data preprocessing unit performs data cleaning and conversion processing on the material processing information, and eliminates the noise and inconsistency in the data by removing outliers, filling missing values, standardizing data formats and other operations, ensuring that subsequent analysis is based on high-quality target material processing information, and ensuring the reliability of the analysis results from the source. On this basis, the statistical analysis unit uses a variety of statistical methods and machine learning algorithms to conduct a full-scale mining of material processing information from multiple dimensions such as time, space, and attributes, revealing valuable insights such as correlation patterns, evolution trends, and abnormal behaviors contained in the data, so that subsequent optimization decisions can be based on a deep understanding of the laws of production operations.

[0009] 3. This application forms an intelligent material control decision-making method that combines data-driven, simulation-verified, and knowledge-guided by optimizing the multi-objective optimization unit, simulation optimization unit, and knowledge reasoning unit in the decision-making unit. Among them, the multi-objective optimization unit is based on a systematic understanding of the statistical analysis results, comprehensively considers multiple decision-making goals such as quality, cost, and efficiency, and uses multi-objective planning, heuristic search and other methods to quickly lock in a batch of preliminary material control plans that are expected to achieve target balance and meet constraints in a massive decision space. These plans not only adhere to the objective laws revealed by the data, but also take into account subjective decision-making preferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 An architectural diagram of a processing material management system provided in an embodiment of the present application; Figure 2 An architecture diagram of a data analysis module provided in an embodiment of the present application; Figure 3 An architectural diagram of a statistical analysis unit provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0012] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0013] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0014] In order to facilitate understanding of the method and system provided by the embodiments of the present application, before introducing the embodiments of the present application, the background of the embodiments of the present application is first introduced.

[0015] With the vigorous development of industrial manufacturing, the types and quantity of materials processed on the production line are becoming increasingly diverse, and the production process is becoming more and more complex. In this context, how to achieve refined management of materials, ensure stable production quality and improve production efficiency has become a major issue facing manufacturing companies.

[0016] At present, the material management of production lines in many companies still mainly relies on manual experience and qualitative judgment. Operators need to manually adjust the proportions of various raw materials and auxiliary materials according to the production plan, set the process parameters of the equipment based on their own experience, and then control product quality through sampling inspection. This traditional management model has many shortcomings.

[0017] First, it is difficult to adapt to the growing variety and quantity of materials. Each material has its own characteristics and optimal ratio, which is difficult for operators to fully grasp, and even more difficult to flexibly allocate in a complex production environment. Secondly, the empirical parameter setting lacks a quantitative basis, and the experience levels of different operators vary greatly, which may lead to inconsistent judgments, and it is difficult for product quality to stabilize within a reasonable range. Thirdly, the qualitative management method cannot make full use of production data, nor can it dynamically optimize and adjust the production process in real time.

[0018] Finally, the post-inspection quality control model has a large lag. When the problem is discovered, a large number of defective products have already been produced, and the losses are difficult to make up. This passive response method cannot prevent quality defects from the source, let alone actively optimize production formulas and process parameters.

[0019] In summary, the traditional material management model can no longer meet the high-quality and high-efficiency development requirements of modern manufacturing. In order to break through the shackles of empirical and qualitative management, realize refined material management and intelligent optimization of production processes, it is urgent to develop a new intelligent material management system. After the background introduction of the above content, those skilled in the art can understand the problems existing in the prior art. The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0020] Based on the above background technology, please refer to Figure 1 , Figure 1 The schematic diagram of the architecture of a processing material management system provided in the embodiment of the present application, the system can be implemented by a computer program, and can also be run as an independent tool application. Specifically, in the embodiment of the present application, the method can be applied on a server, but can also be applied to electronic devices such as a server. The processing material management system includes: a data acquisition module 1, a data analysis module 2 and a material control module 3; Data acquisition module, used to collect material processing information on the production line in real time; Specifically, the data acquisition module automatically and continuously collects various data such as material attribute parameters, processing equipment parameters, and process parameters by deploying various sensors and data acquisition devices on the production line. For example, sensors can measure the material's temperature, humidity, viscosity and other physical and chemical indicators in real time, collectors can record the equipment's speed, pressure, energy consumption and other operating parameters, and quality inspection devices can collect product quality data such as size, strength, and appearance.

[0021] Based on the above embodiment, as an optional embodiment, the data acquisition module 1 includes: a material identification unit 11, a quality data acquisition unit 12, a data integration unit 13 and an equipment status acquisition unit 14; A material identification unit 11, used to identify material information of materials on the production line; Specifically, the material identification unit automatically collects key information about materials on the production line through a variety of technical means. First, when the material enters the production line, the unit uses radio frequency identification (RFID) technology to read the electronic data on the material label to obtain basic information such as the material's name, model, batch, and supplier. For bulk materials, the unit also uses a laser radar scanner to scan the volume and surface morphology of the material and estimate its quantity and particle size distribution.

[0022] For some key attribute parameters, the material identification unit uses precise online detection devices for real-time collection. For example, in the material storage link, the unit uses online moisture meter and near-infrared spectrometer to quickly measure the moisture content and chemical composition of the material; in the material transportation link, the unit uses online viscometer and density meter to continuously monitor the rheological properties and uniformity of the material. These real-time collected physical parameters lay a data foundation for subsequent process optimization.

[0023] At the same time, the material identification unit will also record the source and destination information of each batch of materials, including supplier name, purchase time, storage batch, receiving workshop, material feeding time, etc., to form a full-cycle traceability chain of materials. This refined material identity management can not only track the source of materials for each batch of products and lock in problematic materials in a timely manner, but also analyze the production performance of different materials, optimize supplier selection and procurement strategies, and thus ensure material quality from the source.

[0024] The quality data collection unit 12 is used to collect the quality data of the material according to the material information; Specifically, the quality data collection unit uses a variety of online detection devices to selectively collect key quality indicators according to the characteristics of different materials. For example, for metal materials, the unit focuses on monitoring their chemical composition, mechanical properties, surface quality, etc.; for liquid materials, the unit focuses on detecting their viscosity, density, impurity content, etc.; for powder materials, the unit focuses on measuring their particle size distribution, fluidity, moisture content, etc. Regardless of the material, the unit can automatically match the optimal detection solution and comprehensively collect quality data.

[0025] The equipment status acquisition unit 13 is used to collect the operation data of the processing equipment; Specifically, the equipment status acquisition unit installs various sensors at key locations of the processing equipment to collect the equipment's operating parameters in real time. For example, speed sensors and torque sensors are installed on the spindle motor to monitor the spindle's speed fluctuations and load changes; vibration sensors and noise sensors are installed on the transmission components to monitor the dynamic performance of components such as gearboxes and bearings; data acquisition modules are set up in the control system to record the execution of instructions from PLCs and CNC systems; temperature, pressure, flow and other sensors are arranged at key process parameter measurement points to collect real-time values ​​of process parameters. Through all-round, multi-parameter state perception, the unit has a comprehensive grasp of the "health status" of the equipment.

[0026] The data integration unit 14 is used to integrate the quality data and the operation data to obtain the material processing information.

[0027] Specifically, the data integration unit uses big data processing technology to clean, fuse, and associate the massive data uploaded by the quality data collection unit and the equipment status collection unit, and finally form structured material processing information. First, the unit preprocesses the collected raw data, including removing outliers, correcting missing values, and unifying data formats to improve the accuracy and standardization of the data. Then, the unit matches and fuses data from different sources according to key fields such as timestamps and batch numbers, and establishes a mapping relationship between quality data and equipment status data.

[0028] Data analysis module, used to analyze material processing information to obtain material control solutions; Specifically, the data analysis module uses a variety of data mining and machine learning algorithms to conduct a comprehensive analysis of material processing information. First, the module uses descriptive statistical analysis to calculate the mean, variance, correlation coefficient and other statistical quantities of each parameter, characterize their distribution characteristics and mutual relationships, and gain a preliminary understanding of the production status. Then, the module uses regression analysis, principal component analysis and other methods to identify the key factors that have the greatest impact on product quality and production efficiency from a large number of parameters, and clarify the size and direction of the effects of each factor in order to optimize control in a targeted manner.

[0029] Based on the above embodiment, as an optional embodiment, please refer to Figure 2 , the data analysis module 2 includes: a data preprocessing unit 21, a statistical analysis unit 22 and an optimization decision unit 23; The data preprocessing unit 21 is used to perform data cleaning and conversion processing on the material processing information to obtain the target material processing information; Specifically, the data preprocessing unit uses a series of data processing algorithms to clean and convert material processing information in multiple dimensions. First, the unit will detect whether there are missing values ​​in the data records. For a small number of missing continuous variables, they can be filled by interpolation. For a large number of missing discrete variables, the record is directly deleted. At the same time, the unit will also use tools such as box plots to identify outliers and determine whether they are "noise" introduced by measurement errors or reflect some special production situations. The former needs to be eliminated or corrected, while the latter needs to be retained and analyzed in detail.

[0030] A statistical analysis unit 22, used for performing statistical analysis on the material processing information to obtain analysis results; Specifically, the statistical analysis unit uses descriptive statistics, inferential statistics, regression analysis, variance analysis, multivariate statistics and other methods to analyze the target material processing information from multiple angles and in an all-round manner, revealing the distribution characteristics and evolution laws of key production parameters. First, the unit quantifies the product quality level and production efficiency indicators through descriptive statistics, calculates the mean, standard deviation, quantile and other representative indicators of each parameter, and intuitively displays the overall appearance of the production status. When these indicators fluctuate abnormally, it often indicates that the production process deviates from normal working conditions. Secondly, the statistical analysis unit uses inferential statistical methods such as hypothesis testing to determine whether the effects of different process parameters and equipment status on product quality are statistically significant. For example, the t-test is used to compare whether there are significant differences in the quality means of different batches of products. If so, it is necessary to focus on checking the production records of the batch; the chi-square test is used to analyze the correlation between equipment failure and product defective rate. If there is a positive correlation, it is necessary to focus on optimizing the maintenance strategy of the equipment. These inferential analyses can identify the key influencing factors in the production process and provide a handle for product quality traceability and equipment management.

[0031] Thirdly, the statistical analysis unit widely uses regression analysis and variance analysis to quantitatively characterize the functional relationship between production parameters and quality indicators. For example, by establishing a linear model between raw material ratio, process parameters, equipment status and product performance through multivariate linear regression and analyzing their respective regression coefficients, the importance of each factor can be determined; variance analysis is used to explore the differences in product quality under different teams and different process routes, and then the best man-machine-material method is selected. These quantitative models reveal the laws of quality evolution from the perspective of multiple parameters and strong correlation, and are important tools for quality prediction and process optimization.

[0032] Finally, the statistical analysis unit will also use a combination of dimensionality reduction methods such as principal component analysis and factor analysis, and multivariate statistical methods such as cluster analysis and discriminant analysis to deeply explore the internal structure and distribution patterns of material processing information. Taking principal component analysis as an example, this method can condense the original multidimensional production parameters into a few comprehensive variables while retaining the main information, greatly simplifying the quality analysis model; taking cluster analysis as an example, this method can automatically divide massive production records into several homogeneous groups, with similar internal characteristics of each group and obvious differences between groups, which is convenient for cluster quality management.

[0033] Based on the above embodiment, as an optional embodiment, please refer to Figure 3 , the statistical analysis unit 22 includes: a descriptive statistics unit 221, a correlation analysis unit 222, a regression analysis unit 223 and an analysis result generation unit 224; A descriptive statistics unit 221, used to determine the material concentration trend and material dispersion degree of the processed materials according to the material processing information; Specifically, the descriptive statistics unit will first collect various original parameters of the processed materials from the industrial Internet of Things platform, such as the chemical composition, physical properties, incoming batches, etc. Then, the unit will conduct a central trend analysis on each material parameter, mainly using statistical indicators such as mean and median to measure the concentration of material indicators, and obtain a "portrait" of the overall quality of the material. At the same time, the unit will also calculate the standard deviation, coefficient of variation and other discrete quantities of each indicator to reveal the stability of material quality between different batches. During the analysis process, the unit will make full use of big data processing technology to ensure the timeliness and accuracy of the analysis results.

[0034] After the material quality "portrait" is generated, the descriptive statistics unit will compare these quantitative indicators with the historical statistical values ​​of the material, the industry benchmark values ​​of similar materials, etc., to intuitively reveal the advantages and disadvantages of the current material in terms of quality level and stability, and determine whether it can meet the requirements of downstream production processes and product quality. For materials whose quality level is lower than the threshold or whose fluctuations are significantly exceeded, the unit will issue a timely warning and prompt the relevant departments to take countermeasures, such as changing suppliers and optimizing the feeding process. By continuously tracking the dynamic changes in material quality, the unit builds a set of "descriptive statistical maps" of material quality, clearly depicting the quality distribution of various materials in different periods, and providing visual data support for material quality management.

[0035] A correlation analysis unit 222, used for analyzing the correlation between material processing information according to the material processing information; Specifically, the regression analysis unit will first extract the strongly correlated material processing indicators identified by the correlation analysis unit and use them as candidate independent variables. At the same time, the unit will also collect a series of key quality indicators representing product performance from the quality inspection feedback unit, such as strength, hardness, residual stress, etc., and use them as dependent variables. On this basis, the unit will use algorithms such as multivariate linear regression, partial least squares regression, and support vector regression to establish a function regression model between the candidate independent variables and each quality indicator, and obtain a series of regression equations describing quality characteristics. During the modeling process, the unit will use methods such as cross-validation to optimize model parameters and improve regression quality.

[0036] After the quality regression model is established, the regression analysis unit will comprehensively evaluate the goodness of fit and generalization ability of each model. On the one hand, the unit will calculate the complex correlation coefficient, root mean square error and other indicators of the regression model to quantitatively judge the degree of fit of the model to the known data; on the other hand, the unit will also use the newly collected material processing data to perform extrapolation tests on the model to examine its prediction effect on unknown data. For models with high goodness of fit and strong prediction ability, the unit will formally add them to the quality prediction model library; for under-fitting or over-fitting models, the unit will further optimize its feature selection and parameter estimation methods to continuously improve model performance. By systematically constructing and screening quality regression models, the unit forms a set of multi-quality characteristics and wide-coverage function model systems, mapping the discrete state of the production process into a continuous quality prediction surface, which can not only deduce the impact of each process parameter adjustment on product quality in real time, but also dynamically plan the optimal control path to reduce quality fluctuations.

[0037] A regression analysis unit 223 is used to establish a regression model between material processing information and material quality, and input the material processing information into the regression model to obtain quality quantitative parameters; Specifically, the regression analysis unit will first extract the real-time material processing information collected by the processing execution unit, including feed ratio, process parameter setting, processing time, etc., and clean and feature vectorize these multi-source information to form a structured sample input matrix. Then, the unit will obtain the product quality inspection results from the quality inspection unit and extract key quality indicators such as hardness, strength, and residual stress as dependent variables. On this basis, the unit comprehensively considers the product process characteristics and quality management requirements, selects algorithms such as multivariate linear regression and partial least squares regression, and constructs a function regression model between the candidate material processing parameters and each target quality indicator, and obtains a series of regression equations that use mathematical formulas to characterize the intrinsic relationship between "formula + process → quality". During the modeling process, the unit makes full use of cross-validation, adaptive parameter estimation and other technologies to improve the goodness of fit of the model.

[0038] After the quality regression model is established, the unit will iteratively optimize the generalization performance of each model. On the one hand, through feature selection methods such as principal component analysis, the unit screens out the key processing parameters that have the most significant impact on quality, ensuring the model's prediction ability while reducing the dimension; on the other hand, the unit continues to use newly collected process status data to extrapolate and verify the model performance, and dynamically adjusts the model structure and parameters to continuously improve its prediction effect on unknown data. For high-quality models that have passed optimization and screening, the unit will formally add them to the online quality prediction engine, receive material processing information input in real time, and dynamically output quality prediction results. By inputting production status data into the regression model, the unit can know the quantitative parameters of its quality level in advance before the product goes offline. Once the predicted value deviates from the target, it can issue an early warning and adjust the process in time to nip quality risks in the bud and achieve a quality management transition from post-detection to pre-prevention.

[0039] The analysis result generating unit 224 is used to generate analysis results according to the correlation between material processing information, material concentration trend, material dispersion degree and quality quantitative parameters.

[0040] Specifically, the analysis result generation unit will first extract the correlation matrix between the material processing information output by the correlation analysis unit, which quantitatively reflects the strength and direction of the mutual influence between the various process parameters. At the same time, the unit will also obtain the material ratio, the central tendency and the discrete degree index of the process parameters generated by the cluster analysis unit to characterize the consistency level of the material input composition of different batches and the stability of the production process control. In addition, the quality quantitative parameters output by the quality regression analysis unit will also be passed to the unit to measure the product quality status and its numerical response to the change of processing factors. On the basis of collecting the output of each analysis module, the unit comprehensively uses visualization charts, statistical inference and other methods to cross-compare and systematically interpret the collected multi-source heterogeneous analysis results to form a comprehensive and three-dimensional production process comprehensive analysis report. In the process of result integration, the unit pays special attention to exploring the internal logical relationship between quality quantitative parameters and other analysis indicators, interpreting the main influencing factors of quality fluctuations with the strength of correlation, revealing the material composition root behind quality consistency with the central tendency, reflecting the overtones of quality instability and imbalance of production control capabilities with the degree of discreteness, and deducing the causal chain of product quality formation from multiple angles, which has both breadth and depth.

[0041] The optimization decision unit 23 is used to generate a material control plan based on the analysis results.

[0042] Specifically, the optimization decision-making unit uses a combination of mechanism modeling, intelligent optimization, adaptive control and other technologies to solve the optimal control parameters under multiple objectives and multiple constraints. First, based on the results of statistical analysis, the unit establishes a mechanism model between material ratios, process parameters, product quality and production efficiency to characterize the quantitative relationship between various factors. Regression models are generally used to describe the linear correlation between various parameters, while for nonlinear complex systems, machine learning models such as neural networks are used to approximate their internal mechanisms. These mechanism models explain the evolution of the production process from the perspectives of physics, chemistry and even information theory, and are the theoretical basis for optimization decisions.

[0043] Based on the above embodiment, as an optional embodiment, the optimization decision unit includes: a multi-objective optimization unit, a simulation optimization unit and a knowledge reasoning unit; The multi-objective optimization unit is used to determine the optimization objectives based on the analysis results, and generate a preliminary material control plan based on the analysis results and the optimization objectives; Specifically, the multi-objective optimization unit first uses data mining methods such as correlation analysis and causal analysis, and makes full use of the quality data, equipment data, material data, etc. gathered by the quality management big data platform to deeply analyze the inherent laws and evolution trends of the production process, characterize the complex relationship between production factors and quality characteristics, identify potential influencing factors that significantly affect key quality characteristics, and reveal the causal mechanism of quality problems. On this basis, the unit fully draws on the ideas of quality management tools such as QFD and FMEA, weighs the input-output ratio of quality management activities from the dimensions of quality, cost, efficiency, etc., determines the multiple goals of production process optimization, and scientifically decomposes and dynamically prioritizes the goals, and clarifies the key points of stage-by-stage control. Therefore, the multi-objective optimization unit clarifies the influencing mechanism of the production process and the logical main line of quality management through data analysis and goal decomposition, and points out the direction for the systematic optimization of material control solutions.

[0044] On the basis of analyzing the causes of quality problems and establishing optimization goals, the unit comprehensively considers key influencing factors such as raw material composition and structure, production process parameters, and process control variables, and uses multi-objective planning, multi-attribute decision-making and other methods to search for the optimal balance point of multiple objectives in the feasible space of key influencing factors, solve the optimal parameter combination of the production process under the constraints of quality, cost, efficiency and other goals, and generate a phased optimization plan that takes into account both product quality improvement and production cost savings. Considering the time-varying characteristics of production factors, the unit also makes full use of time series analysis, probability prediction and other methods to characterize the dynamic change law of production factors, and optimizes the dynamic threshold and real-time adjustment strategy of the material control plan in a targeted manner, so as to achieve dynamic control and real-time optimization of the production process. It is worth mentioning that the preliminary material control plan generated by the unit is not achieved overnight, but is continuously iterated and optimized in the cycle of "analyzing problems-determining goals-generating plans-evaluating effects", weighing the dynamic balance of various goals in multiple rounds of plan evaluation and feedback, and pursuing the "optimal solution" under complex constraints.

[0045] A simulation optimization unit is used to simulate the material processing process according to the preliminary material control plan, obtain the simulated processing parameters, and optimize the preliminary material control plan according to the simulated processing parameters to obtain the simulated optimized material control plan; Specifically, the simulation optimization unit will first absorb the preliminary material control plan output by the multi-objective optimization unit, and comprehensively use physical modeling, data-driven modeling and other methods to build a set of production process simulation models covering raw material properties, equipment conditions, operating parameters, environmental parameters and other elements. During the model construction process, the unit, on the one hand, makes full use of mechanism knowledge and engineering experience to finely depict the internal mechanism of the production process and ensure the physical interpretability of the model; on the other hand, the unit also widely draws on actual production data, and uses data-driven methods such as machine learning to automatically calibrate and optimize the simulation model to improve the model's fidelity to the actual process. Through the integration of mechanism modeling and data modeling, the simulation optimization unit builds a set of high-fidelity simulation models that are close to reality, embody mechanisms, and are highly interpretable, providing a high-confidence digital test field for subsequent program evaluation and optimization. On this basis, the unit quantifies the preliminary material control plan into a set of specific model parameter inputs, simulates the production process after the implementation of the simulation plan through simulation operation, evaluates the dynamic response of key process parameters and product quality characteristics, and examines the expected effect of the plan.

[0046] After the preliminary results of the simulation evaluation are formed, the unit will iteratively carry out simulation-based solution optimization. On the one hand, through methods such as sensitivity analysis, the unit identifies the key control parameters that have the most significant impact on the quality target, further focuses on the key points of control, and simplifies the complexity of the optimization problem; on the other hand, the unit makes full use of the experimental design method to construct a set of test schemes in the feasible space of key parameters, quickly explores the quality benefits of each parameter combination through batch simulation operation, and solves the optimal parameter combination with the help of intelligent optimization algorithms based on the analysis of simulation results, forming a more refined and optimized control scheme. In the iterative exploration of simulation optimization, the unit makes full use of the virtual trial and error capability of the high-fidelity model, and continuously refines the preliminary scheme through the spiral rise of "trial and error-analysis-optimization-trial and error again", so that the scheme has been perfected before touching the real production system. It is worth mentioning that the simulation optimization unit is not satisfied with the local optimization of a single scheme in the process of scheme optimization, but focuses on generating a set of "combination punches" of control parameters, forming a set of control strategies based on different production scenarios, and realizing the flexible response of "one policy, multiple methods, changing according to needs", providing a rich plan library and alternative solution pool for the implementation of the scheme.

[0047] Based on the above embodiment, as an optional embodiment, the simulation optimization unit includes: a modeling unit, a simulation solving unit and a simulation optimization unit; A modeling unit is used to construct a simulation model reflecting the processing process according to the material processing technology; Specifically, the modeling unit adopts a hybrid modeling paradigm that combines mechanism modeling with data-driven modeling, making full use of physical mechanism knowledge and industrial big data to build a simulation model of the processing process from multiple angles and levels. On the one hand, the unit integrates multidisciplinary mechanism knowledge, and constructs a mechanism model that reflects the physical and chemical mechanism of the material processing process from the perspectives of raw material characteristics, reaction kinetics, heat transfer and mass transfer, etc., accurately describes the evolution of process parameters such as temperature, pressure, and flow, and quantitatively characterizes them using mathematical forms such as partial differential equations and constitutive equations. On the other hand, the modeling unit fully explores the value of industrial big data, extracts the statistical laws of the process flow from massive production data, constructs a data model of the processing process, describes the correlation pattern between process parameters and product indicators, and quantitatively fits them using data-driven models such as regression models and neural networks. In the hybrid modeling process, the modeling unit continuously strengthens the integration of the mechanism model and the data model. On the one hand, it uses the process knowledge provided by the mechanism model to guide the structural optimization of the data model, and on the other hand, it uses the correlation pattern learned by the data model to dynamically correct the constitutive parameters of the mechanism model, so as to achieve complementary gains of the two types of models and form a hybrid simulation model with both mechanism explanatory power and data accuracy.

[0048] A simulation solution unit is used to input the preliminary material control plan into the simulation model to obtain simulation processing parameters; Specifically, the simulation solution unit adopts a multiple solution mechanism combining direct problem solution and inverse problem optimization to solve the optimal process parameter combination in a multi-dimensional parameter space. On the one hand, the unit directly analyzes the feasibility of the scheme through direct problem solution, facing the raw material ratio given by the preliminary control scheme. Substitute the scheme parameters into the material processing simulation model constructed by the modeling unit to form a mathematically complete direct problem, and select numerical discretization methods such as finite element method and finite difference method according to the characteristics of the model to convert the complex model into a discrete form that is easy to calculate, and obtain simulation processing parameters such as temperature distribution and flow field curve through iterative calculation. Direct problem solution can quickly obtain the specific requirements for equipment working conditions and operating parameters under the given control scheme. On the other hand, the simulation solution unit also applies inverse problem optimization technology to solve the optimal process parameter combination under given process constraints and objective functions. According to actual needs such as production scheduling and energy consumption control, constraints on process parameters are formed, and corresponding objective functions are constructed for control targets such as output, quality, and cost. On this basis, multi-objective optimization solutions are carried out, and the optimal combination of process parameters is explored through evolutionary algorithms, heuristic searches, etc., to achieve balanced optimization of management and control objectives.

[0049] The optimization unit is used to optimize the preliminary material control plan according to the simulated processing parameters and the preset optimization goals to obtain the simulated optimized material control plan.

[0050] Specifically, the optimization unit adopts an integrated two-layer optimization mechanism to systematically optimize the preliminary control plan from the two levels of overall optimization and local optimization. The first layer is the "macro-overall optimization" based on global search. This level focuses on the top-level design of material control. On the basis of the quality, efficiency, cost and other performances revealed by the simulation processing parameters, combined with the preset optimization goals, the global optimization techniques such as evolutionary algorithms and response surface methods are used to explore the optimal combination in the macro-decision space such as raw material type selection and ratio setting. By predicting the performance of the objective function under each combination of optimization variables, the search direction is adjusted adaptively, and the approximate optimal overall optimization plan is quickly locked, laying the foundation for optimization at the macro level. The second layer is the "micro-local optimization" based on local search. This level focuses on the micro-execution of material control. Under the guidance of the overall optimization plan, the temperature, pressure and other process parameters are further optimized locally. On the one hand, the sensitivity between process parameters and optimization targets is analyzed by using numerical gradients, mutual information and other methods to identify the key parameters that have a greater impact on the optimization targets; on the other hand, local optimization algorithms such as sequential quadratic programming and interior point method are used to search for the optimal settings near the key parameters to achieve comprehensive optimization from macro to micro. Driven by the two-level optimization, various elements of the preliminary material control plan, such as raw material selection, proportioning method, and process parameters, have achieved an organic unity of global synergy and local efficiency improvement, and the overall operation quality has been significantly improved.

[0051] The knowledge reasoning unit is used to rationally verify the simulation-optimized material control plan according to empirical rules, obtain verification results, and constrain the simulation-optimized material control plan according to the verification results to obtain a material control plan.

[0052] Specifically, the knowledge reasoning unit first uses technologies such as ontology construction and knowledge extraction to extract implicit knowledge such as production laws and operating procedures in the minds of employees into structured knowledge rules, and builds a comprehensive knowledge base covering rules in multiple fields such as safety, process, and quality. On this basis, the unit matches the process parameters of the simulation optimization material control plan with the rule base to identify potential risk points in the plan that may violate the empirical rules. For example, when the kiln temperature of the optimization plan exceeds the permitted range of the regulations, the knowledge reasoning unit can discover this risk and determine that the plan does not conform to the empirical rules, thereby making a verification result that the optimization plan is unreasonable. While obtaining the verification result, the knowledge reasoning unit further matches and infers the temperature constraints that the optimization plan should meet, uses the temperature constraints as the basis for correcting the optimization plan, guides the simulation optimization unit to carry out plan re-optimization, and finally makes the optimization plan meet the rationality requirements of the empirical rules. In the process of constraint identification and reasoning, the unit makes full use of machine learning algorithms to accurately characterize production laws, and realizes the dynamic expansion of rule knowledge based on case reasoning. Through the iterative integration of machine learning and knowledge reasoning, it continuously strengthens the control of the rationality of the optimization plan.

[0053] Based on the above embodiment, as an optional embodiment, the knowledge reasoning unit includes: a knowledge acquisition unit, an ontology construction unit, a rule mining unit, a reasoning verification unit and a constraint unit; Knowledge acquisition unit, used to acquire material domain knowledge; Specifically, the knowledge acquisition unit uses crawler technology to capture theoretical knowledge such as material ratio, processing technology, quality standards, etc. on the Internet, and forms structured knowledge items after parsing, cleaning, filtering and other processing. At the same time, the unit uses natural language processing technology to perform semantic analysis and key information extraction on unstructured texts such as industry standards, operating procedures, and patent documents, further enriching the breadth and depth of theoretical knowledge. While acquiring explicit theoretical knowledge, the knowledge acquisition unit also focuses on collecting implicit experience knowledge. Through interviews with senior employees, employees are guided to express their accumulated practical skills and lessons learned in words, and key information in language descriptions is extracted using technologies such as speech recognition and text analysis, so as to transform employee experience into structured knowledge that can be understood by computers. In addition, the knowledge acquisition unit can also deeply mine employees' operational behavior data, analyze employees' excellent operating modes in production practice, and use algorithms such as sequence pattern mining and frequent item set mining to extract the behavioral laws contained therein, forming reusable and inheritable "best practice" knowledge.

[0054] Ontology construction unit, used to abstract material domain knowledge into a formalized ontology model; Specifically, the ontology construction unit adopts a construction strategy combining top-down and bottom-up, and uses both manual definition and automatic extraction to construct the material domain ontology. In the top-down construction, the knowledge engineer first refers to industry standards and authoritative literature to manually define the core concepts in the material field, such as "raw materials", "proportion", "process", "equipment", "quality", etc., and sorts out the hierarchical relationship between concepts to form the skeleton framework of the ontology model. On this basis, the knowledge engineer further refines the attribute characteristics of each core concept, defines the data type and constraints of the attribute value, and improves the semantic description ability of the ontology model. In the bottom-up construction, the ontology construction unit uses natural language processing technology to perform part-of-speech tagging, syntactic analysis, semantic role labeling and other processing on the material domain text, and automatically extracts the concepts, relationships, attributes and other knowledge elements contained in the text. Subsequently, the unit uses dictionary comparison, rule mapping and other technologies to map the extracted concepts, relationships, and attributes to the manually defined ontology framework to achieve bottom-up ontology filling. During the ontology construction process, knowledge engineers fully absorb feedback from domain experts, repeatedly review and revise the ontology model, and rely on authoritative domain knowledge to ensure the completeness and accuracy of the ontology model.

[0055] A rule mining unit is used to mine rules according to the ontology model and obtain empirical rules; Specifically, the rule mining unit uses an ontology-based association rule mining algorithm to complete the automatic extraction of rules under the guidance of the ontology. First, based on the concept system defined by the ontology model, the unit semantically annotates business data and employee experience cases to identify key elements such as raw materials, ratios, and processes in the cases. Then, the unit uses frequent pattern mining algorithms such as Apriori and FP-Growth to automatically discover frequent patterns of different elements, and evaluate the association strength of frequent patterns based on indicators such as confidence and support to form candidate association rules. On this basis, the rule mining unit makes full use of the domain knowledge contained in the ontology to optimize the quality of rule mining from multiple perspectives. On the one hand, the concept hierarchy defined by the ontology is used to semantically generalize the antecedent and consequent concepts of the rules to improve the scope of application of the rules; on the other hand, the constraints described by the ontology are used to automatically identify and filter out invalid rules that violate basic constraints such as physics and chemistry to improve the credibility of the rules. At the same time, the unit will submit the mined rules to field experts for review, and guide the experts to revise and improve the applicable boundaries, confidence strength and other contents of the rules based on practical experience, further improve the quality of the rules, and ultimately form a decision-making rule knowledge base that is both general and applicable.

[0056] Based on the above embodiment, as an optional embodiment, the rule mining unit includes: a feature selection unit, an association rule mining unit, a decision tree learning unit and a rule post-processing unit; The feature selection unit is used to select the most discriminative and relevant feature subsets from the ontology model; Specifically, the feature selection unit adopts a hybrid feature selection strategy that combines filtering and wrapping feature evaluation paradigms, taking into account the value of the feature itself and the overall utility of the feature subset, and selects the optimal feature combination from the ontology model. First, the unit adopts a filtering feature selection method to evaluate the importance of the feature according to its own statistical indicators to achieve the initial screening of the feature. On the one hand, the variance, information gain and other indicators are used to measure the discrimination ability of each feature, and the high-discrimination features that contribute more to sample discrimination are screened out; on the other hand, the Pearson correlation coefficient, mutual information and other indicators are used to measure the correlation between each feature and the control target, and the high-correlation features that are highly correlated with the target are screened out. Combining the two types of evaluation indicators, redundant and irrelevant features are eliminated to obtain the reduced candidate feature set. On this basis, the feature selection unit further adopts a wrapping feature selection method to embed feature selection into rule learning, evaluate the utility of feature subsets guided by rule quality, and achieve feature selection. Specifically, the decision tree learning unit is used to construct a rule learning model based on different feature subsets, and the classification or prediction performance of each model is evaluated by cross-validation and other methods, thereby obtaining the utility measurement of each feature subset. On this basis, we use optimization search technologies such as genetic algorithms and simulated annealing to find the optimal feature subset that combines model performance and feature simplicity to maximize overall utility. After layers of screening and selection, we finally get a set of key features that condenses the essence of data rules and closely follows the guidance of management and control needs.

[0057] An association rule mining unit, used to determine the frequent co-occurrence patterns between different features from the feature subset and generate association rules based on the frequent co-occurrence patterns; Specifically, the association rule mining unit adopts a "two-stage, multi-step" mining mechanism based on the Apriori idea to efficiently screen and extract frequent feature patterns and association rules from massive feature combinations. The first stage is "frequent pattern mining", which determines the frequent co-occurring feature combinations from the feature subset through iterative layer-by-layer search. First, according to the minimum support threshold, the frequency of occurrence of each single feature is found by scanning the feature matrix, and a frequent feature set of length 1 is obtained by screening; then, the candidate feature combination is generated by self-connection of the frequent feature set, and the frequent combination is determined by support counting to obtain a frequent feature set of length 2; on this basis, the frequent set generation and screening process is repeated, and the length of the frequent feature set is increased by one in each iteration until no longer frequent sets can be generated. Through layer-by-layer iterative search, frequent feature sets of different lengths are obtained at a low computational cost, and the frequent co-occurrence patterns of features at various granularities are fully characterized. The second stage is "association rule generation", which further screens strong association rules based on the frequent feature set. For each frequent feature set, all possible subset divisions are enumerated to form rule candidates in the form of "antecedent-consequent"; then, the confidence of each candidate rule is calculated, and strong association rules are obtained by filtering according to the minimum confidence threshold. Through rule enumeration and confidence evaluation based on frequent feature sets, the rule search space is greatly reduced, and high-confidence discriminant rules are quickly locked, realizing the improvement of rules from feature co-occurrence to conditional association. After frequent pattern mining and strong association rule screening, an association rule set with both prior support and posterior prediction power is finally formed.

[0058] A decision tree learning unit is used to generate a decision rule for a target concept according to the ontology model and form a decision tree according to the decision rule; Specifically, the decision tree learning unit continuously strengthens the domain adaptability and robustness of learning. Through domain adaptive dynamic learning technology, the judgment target, algorithm parameters and termination conditions are adaptively adjusted according to the type of control task and decision stage to achieve dynamic construction for different control scenarios; through knowledge-based incremental learning technology, new process knowledge is continuously absorbed during the decision tree construction process, and the online optimization and continuous evolution of the decision tree are achieved through knowledge-guided incremental node generation and judgment rule update; through cost-sensitive robust learning technology, the unequal costs of different types of judgment errors are considered in the construction target, and the robustness of the decision tree is improved by adaptively adjusting the sample weights and classification thresholds, so that robust decisions can still be obtained when the data is biased or the distribution drifts; through interpretable decision tree construction technology, feature predicates that are easy to interpret in engineering are selected to generate shallow and readable judgment rules, and rule visualization and knowledge distillation methods are used to enhance the understanding and trust of the decision tree. Through the above-mentioned system innovation design, the decision tree learning unit can fully integrate knowledge guidance and data-driven, build a domain-fitting, iteratively optimized, robust and reliable intelligent decision model, and fully support the intelligentization of the production control process.

[0059] The rule post-processing unit is used to perform rule mining on the ontology model according to the decision tree to obtain empirical rules.

[0060] Specifically, the rule post-processing unit adopts a rule improvement method based on ontology mapping, comprehensively considers the three dimensions of judgment rules, ontology knowledge, and control tasks, and constructs an empirical rule library that is both accurate and interpretable. First, the unit makes full use of ontology knowledge, maps the feature items in the judgment rules to ontology concepts through semantic matching, and coordinates the rule expression with the ontology terminology system; at the same time, the antecedent and consequent of the rule are expanded through knowledge association, and the context concept outside the judgment path is introduced to enhance the completeness and semantic richness of the rule. Secondly, the unit fully considers the characteristics of the control task and gives targeted post-processing to the rules in different decision-making scenarios. For process optimization rules, it focuses on integrating process mechanism knowledge, explaining the rules with causal logic, and introducing adjustable process parameters as rule consequents according to the optimization objectives; for abnormal diagnosis rules, it focuses on integrating fault case knowledge, adding diagnostic elements such as fault mode and abnormal location, and improving the scene pertinence and diagnostic granularity of the rules; for control decision rules, it focuses on integrating control logic knowledge, introducing control strategies such as feedback correction and prediction compensation into the rules, and improving the robustness and adaptability of the rules. Thirdly, the unit makes full use of the rule quality assessment system to evaluate the quality of rules from multiple dimensions such as accuracy, novelty, and interpretability, screens high-quality rules that contribute highly to control tasks and are easy to read in engineering, and removes duplicate rules in combination with the knowledge redundancy detection algorithm to form a concise and efficient experience rule library. After integration with ontology knowledge, task-oriented improvement, and multi-criteria evaluation and screening, we finally get control experience rules with rich connotations, standardized forms, and high quality.

[0061] The reasoning verification unit is used to rationalize and verify the simulation optimization material control plan according to empirical rules to obtain the verification results; Specifically, the reasoning verification unit adopts a rule-based reasoning verification strategy and uses the empirical rule library to perform logical verification on the simulation optimization material control plan. First, the unit extracts the features of the core elements such as raw material selection, ratio setting, and process parameters in the simulation optimization plan in turn, and automatically matches the relevant rules triggered by each element of the plan by comparing the rule antecedents in the empirical rule library. Then, based on the rule reasoning mechanism, the unit comprehensively analyzes the matching results of each rule, identifies the problems in the simulation optimization plan in terms of raw material applicability, ratio rationality, process feasibility, etc., and gives specific evidence for violating the empirical rules. While forming a list of problems, the reasoning verification unit also makes full use of the causal logic contained in the empirical rules to automatically deduce the impact and consequences of the problem, evaluate the severity of the problem, and clarify the priority order of the problem. During the reasoning verification process, the unit can also interact with domain experts, who will further confirm the suspected problems and provide practical rectification suggestions. The problem annotation and rectification ideas fed back by the experts are used to guide the rational improvement of the simulation optimization plan on the one hand, and also feed back to the update of the empirical rule library on the other hand, forming a virtuous cycle of reasoning verification and knowledge accumulation.

[0062] The constraint unit is used to constrain the simulation optimized material control plan according to the verification results to obtain the material control plan.

[0063] Specifically, the constraint unit adopts a dynamic constraint strategy based on verification results to transform the problems found by reasoning verification into quantifiable and operational constraints, and feeds back the constraints to the simulation optimization unit to re-carry out optimization search in the constraint space. First, the constraint unit conducts a comprehensive analysis of the problems given by reasoning verification and their severity, matches typical constraint templates according to the problem type, and automatically generates mathematical constraint forms that characterize the problems. Through constraint solidification, the raw material selection, ratio setting and other elements with problems in the simulation optimization scheme are given clear value restriction boundaries. At the same time, the constraint unit sets constraints in a graded manner according to the severity of the problem. For problems that seriously violate experience, hard constraints are set to directly prohibit the value range of the problem parameters; for minor problems to be verified, flexible constraints are set to allow parameter values ​​within a certain range. For those decision variables that are not covered by the reasoning verification problem, their relaxed constraints are maintained to provide a search space for simulation optimization. In addition, the constraint unit also supports user-defined personalized constraints. Users can manually set constraints for decision variables according to specific application scenarios and optimization demands, giving simulation optimization greater flexibility.

[0064] The material control module 3 is used to control the materials on the production line according to the material control plan.

[0065] Based on the above embodiment, as an optional embodiment, the material control module 3 includes: a material ratio control unit 31, a process parameter control unit 32 and a quality detection feedback unit 33; The material ratio control unit 31 is used to determine the optimal ratio according to the material control plan, and accurately control the proportion of various raw and auxiliary materials according to the optimal ratio; Specifically, after receiving the material control plan generated by the optimization decision-making unit, the material ratio control unit will extract the optimal material ratio parameters, mainly including the theoretical amount of each raw material and its allowable deviation. Based on these parameters, the unit will further break down and calculate the actual amount of each material in each batch and each time period to form a set of careful batching plans to ensure that at any time and in any batch, the actual amount of each material is strictly equal to the optimal value specified by the theoretical ratio. At the same time, the batching plan will also reserve a certain buffer to cope with the risk of fluctuations in material supply.

[0066] A process parameter control unit 32 is used to determine the process parameters according to the material control plan and adjust the processing parameters of the equipment according to the process parameters; Specifically, after receiving the material control plan generated by the optimization decision unit, the process parameter control unit will extract a complete set of optimal process parameters, covering the set values ​​and their allowable deviations for the entire production process, such as temperature curves, pressure ranges, time nodes, etc. Based on these parameters, the unit will further refine the process control instructions for each device and each batch to ensure that at any time, the actual working conditions of each device strictly match the optimized process parameters.

[0067] The quality inspection feedback unit 33 is used to perform quality inspection on the processed products according to the quality standards determined by the material control plan to obtain quality inspection results, and feed back the quality inspection results to the data analysis module.

[0068] Specifically, after receiving the material control plan generated by the optimization decision-making unit, the quality inspection feedback unit will extract detailed product quality standards from it, including various performance indicators and their tolerance requirements, such as strength, hardness, size, etc. Based on these standards, the unit will formulate a thorough quality inspection plan, clarify key elements such as inspection objects, inspection items, sampling frequency, and judgment rules, to ensure that the quality inspection process is complete and standardized, and the quality inspection results are objective and accurate. When formulating the quality inspection plan, the unit will also fully consider factors such as production rhythm and process characteristics, and strive to minimize interference with production while ensuring the intensity of quality inspection.

[0069] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0070] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A processing material management system, characterized in that: The system includes: a data acquisition module, a data analysis module and a material control module; The data acquisition module is used to collect material processing information on the production line in real time; The data analysis module is used to analyze the material processing information to obtain a material control plan; The material control module is used to control the materials on the production line according to the material control plan.

2. The system according to claim 1, wherein the data acquisition module comprises: Material identification unit, quality data collection unit, data integration unit and equipment status collection unit; The material identification unit is used to identify material information of materials on the production line; The quality data collection unit is used to collect the quality data of the material according to the material information; The equipment status acquisition unit is used to collect operating data of the processing equipment; The data integration unit is used to integrate the quality data and operation data to obtain material processing information.

3. The system according to claim 1, wherein the data analysis module comprises: Data preprocessing unit, statistical analysis unit and optimization decision unit; The data preprocessing unit is used to perform data cleaning and conversion processing on the material processing information to obtain target material processing information; The statistical analysis unit is used to perform statistical analysis on the material processing information to obtain analysis results; The optimization decision unit is used to generate the material control plan according to the analysis result.

4. The system according to claim 1, wherein the material control module comprises: Material ratio control unit, process parameter control unit and quality detection feedback unit; The material ratio control unit is used to determine the optimal ratio according to the material control plan, and accurately control the proportion of various raw and auxiliary materials according to the optimal ratio; The process parameter control unit is used to determine the process parameters according to the material control plan, and adjust the processing parameters of the equipment according to the process parameters; The quality inspection feedback unit is used to perform quality inspection on the processed products according to the quality standards determined by the material control plan to obtain quality inspection results, and feed back the quality inspection results to the data analysis module.

5. The system according to claim 3, wherein the statistical analysis unit comprises: Descriptive statistics unit, correlation analysis unit, regression analysis unit and analysis result generation unit; The descriptive statistics unit is used to determine the material concentration trend and material dispersion degree of the processed materials according to the material processing information; The correlation analysis unit is used to analyze the correlation between the material processing information according to the material processing information; The regression analysis unit is used to establish a regression model between the material processing information and the material quality, and input the material processing information into the regression model to obtain the quality quantification parameter; The analysis result generating unit is used to generate the analysis result according to the correlation between the material processing information, the material concentration trend, the material dispersion degree and the quality quantification parameter.

6. The system according to claim 3, wherein the optimization decision unit comprises: Multi-objective optimization unit, simulation optimization unit and knowledge reasoning unit; The multi-objective optimization unit is used to determine the optimization target according to the analysis result, and generate a preliminary material control plan according to the analysis result and the optimization target; The simulation optimization unit is used to simulate the material processing process according to the preliminary material control plan to obtain simulated processing parameters, and optimize the preliminary material control plan according to the simulated processing parameters to obtain a simulated optimized material control plan; The knowledge reasoning unit is used to perform rational verification on the simulation-optimized material control plan according to empirical rules to obtain verification results, and to constrain the simulation-optimized material control plan according to the verification results to obtain the material control plan.

7. The system according to claim 6, wherein the knowledge reasoning unit comprises: Knowledge acquisition unit, ontology construction unit, rule mining unit, reasoning verification unit and constraint unit; The knowledge acquisition unit is used to acquire material domain knowledge; The ontology construction unit is used to abstract the material domain knowledge into a formalized ontology model; The rule mining unit is used to perform rule mining according to the ontology model to obtain the empirical rules; The reasoning verification unit is used to perform rational verification on the simulation optimization material control plan according to the empirical rules to obtain a verification result; The constraint unit is used to constrain the simulation optimized material control plan according to the verification result to obtain the material control plan.

8. The system according to claim 6, wherein the simulation optimization unit comprises: Modeling unit, simulation solution unit and optimization unit; The modeling unit is used to construct a simulation model reflecting the processing process according to the material processing technology; The simulation solving unit is used to input the preliminary material control plan into the simulation model to obtain simulation processing parameters; The optimization unit is used to optimize the preliminary material control plan according to the simulation processing parameters and the preset optimization goals to obtain the simulation optimized material control plan.

9. The system according to claim 7, wherein the rule mining unit comprises: Feature selection unit, association rule mining unit, decision tree learning unit and rule post-processing unit; The feature selection unit is used to select the most discriminative and relevant feature subset from the ontology model; The association rule mining unit is used to determine the frequent co-occurrence pattern between different features from the feature subset, and generate association rules according to the frequent co-occurrence pattern; The decision tree learning unit is used to generate a decision rule of a target concept according to the ontology model, and form a decision tree according to the decision rule; The rule post-processing unit is used to perform rule mining on the ontology model according to the decision tree to obtain the empirical rules.

10. A processing material management method, characterized in that: The method comprises: Collect material processing information on the production line in real time; Analyze the material processing information to obtain a material control plan; The materials on the production line are controlled according to the material control plan.

Images