Process knowledge base updating method, device, terminal equipment and system

Abstract

The invention is suitable for the technical field of data processing, and provides a process knowledge base updating method, device, terminal equipment and system, and the method comprises the steps: collecting the operation data of each equipment corresponding to a process route used for producing a product during the production of the product, and collecting the first production data of the product during the production process; analyzing the operation data of each device and the first production data to obtain a process detection result of the product; optimizing the process route based on the process detection result to obtain an optimization result; acquiring second production data during continuous production based on the optimization result; and updating the process knowledge base based on the operation data of each device, the process detection result, the optimization result and the second production data. According to the method, the production data and the operation data can be collected in real time and analyzed to automatically update the process knowledge base, so that the hysteresis of traditional manual updating is avoided, and the timeliness and the accuracy of process knowledge are ensured.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data processing, and particularly relates to a process knowledge base updating method and device, a terminal device and a system. BACKGROUND

[0002] The development of intelligent manufacturing promotes the gradual realization of autonomous perception, intelligent interconnection, intelligent learning and autonomous decision-making of production systems. Process decision-making, as a bridge between design and manufacturing, is a key link of intelligent manufacturing. Taking LED display screens as an example, the production process is complex, involving chip packaging, module production, box assembly, detection and calibration and other links, and process parameters have a significant impact on product quality. Therefore, production needs to be combined with a process knowledge base. The process knowledge base is a database that stores and manages the knowledge required in the entire process design process.

[0003] However, existing process knowledge bases are usually stored in the form of documents or static databases, and are not comprehensive enough to meet actual needs. SUMMARY

[0004] The embodiments of the application provide a process knowledge base updating method, device, terminal device and system to solve the problem that the prior art is not comprehensive enough to meet actual needs.

[0005] In a first aspect, the embodiments of the application provide a process knowledge base updating method, comprising:

[0006] When producing a product, collecting operation data of each device corresponding to a process route used to produce the product, and collecting first production data of the product in the production process;

[0007] Analyzing the operation data of each device and the first production data to obtain a process detection result of the product;

[0008] Optimizing the process route based on the process detection result to obtain an optimization result;

[0009] Obtaining second production data when continuing production based on the optimization result;

[0010] Updating a process knowledge base based on the operation data of each device, the process detection result, the optimization result and the second production data.

[0011] The embodiments of the application have the beneficial effects that:

[0012] The embodiment of the application provides a process knowledge base updating method, which comprises the following steps: collecting running data of each device corresponding to a process route for producing a product and collecting first production data of the product in a production process when the product is produced; analyzing the running data of each device and the first production data to obtain a process detection result of the product; optimizing the process route based on the process detection result to obtain an optimization result; obtaining second production data when production is continued based on the optimization result; and updating a process knowledge base based on the running data of each device, the process detection result, the optimization result and the second production data. The application can collect production data and running data in real time and analyze the data to automatically update the process knowledge base, thereby avoiding the hysteresis of traditional manual updating and ensuring the timeliness and accuracy of process knowledge.

[0013] Optionally, the process detection result comprises a process analysis result and an abnormality analysis result, and the step of analyzing the running data of each device and the first production data to obtain the process detection result of the product comprises:

[0014] performing a data preprocessing operation on the running data of each device and the first production data to obtain effective data;

[0015] inputting the effective data into a machine learning model for processing to obtain the process analysis result;

[0016] inputting the effective data into an abnormality analysis model for processing to obtain the abnormality analysis result.

[0017] In the above embodiment, the effective data is input into the machine learning model, the implicit rules can be mined from the massive data, and thus accurate process detection results can be generated; meanwhile, the effective data is input into the abnormality analysis model, and thus abnormal conditions in the production process can be identified in real time, and accurate abnormality analysis results can be generated. Therefore, the method combines the process detection result and the abnormality analysis result, and thus process cognition can be improved, and high-quality content can be provided for updating of the process knowledge base.

[0018] Optionally, the machine learning model comprises a classification model, a regression model and a clustering model, the process analysis result comprises a defective type detection result, a process effect and a rule detection result of the product; and the step of inputting the effective data into the machine learning model for processing to obtain the process analysis result comprises:

[0019] inputting the effective data into the classification model for identification to obtain the defective type detection result;

[0020] inputting the effective data into the regression model for analysis to obtain the process effect;

[0021] Input the effective data into the clustering model for processing to obtain the rule detection result.

[0022] In the above embodiments, the classification model can be used to quickly and accurately classify the bad types, the regression model can be used to quantify the correlation between the process parameters and the production indexes to improve the accuracy of subsequent process parameter optimization, and the clustering model can be used to mine data clusters with similar characteristics from the effective data, thereby mining implicit process correlations to enrich the subsequent process knowledge base.

[0023] Optionally, the process route is optimized based on the process detection result to obtain an optimization result, including:

[0024] The process detection result is input into a rule engine for processing to obtain an adjustment instruction;

[0025] If the adjustment instruction is to adjust the process parameters, a parameter to be adjusted and a parameter adjustment method are determined;

[0026] The parameter to be adjusted is adjusted based on the parameter adjustment method to obtain a first parameter;

[0027] The first parameter is optimized based on an optimization algorithm to obtain the optimization result.

[0028] In the above embodiments, the progressive process of first generating an adjustment instruction based on a rule engine, then determining a parameter adjustment direction, then preliminarily adjusting the first parameter based on the parameter adjustment method, and finally accurately optimizing the first parameter based on an optimization algorithm, realizes the standardization, precision and efficiency of process optimization.

[0029] Optionally, the process knowledge base includes an abnormal case library, a process parameter library and a historical production database, and the process knowledge base is updated based on the running data of each device, the process detection result, the optimization result and the second production data, including:

[0030] The process detection result and the optimization result are analyzed to obtain an abnormal case and an abnormal adjustment method;

[0031] The second production data is analyzed to obtain an adjustment result of the abnormal adjustment method;

[0032] The abnormal case, the abnormal adjustment method and the adjustment result are stored in the abnormal case library;

[0033] The process parameter library is updated based on the optimization result;

[0034] The operation data of each device, the second production data, and the adjustment result are stored in the historical production database.

[0035] In the above embodiments, by refining abnormal cases and abnormal adjustment methods, and updating the abnormal case library in combination with the adjustment result, a complete knowledge chain including abnormal phenomena, coping schemes, and implementation effects is formed, and the subsequent abnormal processing efficiency is improved; based on the optimization result, the process parameter library is updated, which can ensure that the parameters in the process parameter library are always matched with the current production demand, device state, and raw material characteristics, and the device operation data, second production data, and adjustment result are stored in association, which can form a complete data chain covering the device state, production process, and optimization effect. Therefore, through the collaborative update of the abnormal case library, process parameter library, and historical production database, a knowledge base ecosystem is formed, which not only provides instant support for daily production, but also provides a data basis for long-term process optimization.

[0036] Optionally, the updating of the process parameter library based on the optimization result comprises:

[0037] analyzing the optimization result to determine an optimization parameter and an optimization method;

[0038] obtaining an initial parameter range of the optimization parameter from the process parameter library;

[0039] updating the initial parameter range based on the optimization method to obtain a standard parameter range.

[0040] In the above embodiments, in the process of updating the process parameter library based on the optimization result, the optimization result is first disassembled to obtain the optimization parameter and the optimization method, then the optimization parameter is matched with the initial parameter, and finally the range is accurately updated in combination with the optimization method. The fine operation realizes the upgrade of the process parameter library from static storage to dynamic adaptation, so that the process parameter library always maintains a dynamic optimal state, which can not only adapt to the change of production conditions in real time, but also provide stable and reliable parameter support for process design, production operation, and quality control.

[0041] Optionally, after the second production data based on the optimization result is obtained, the method further comprises:

[0042] analyzing the second production data and the optimization result to obtain an abnormal processing method;

[0043] updating the rule engine based on the abnormal processing method;

[0044] and / or,

[0045] detecting the second production data to obtain a data detection result;

[0046] updating the optimization algorithm based on the data detection result.

[0047] In the above embodiment, after obtaining the second production data, the double-path iteration of updating the rule engine by optimizing the abnormal handling mode based on the data and updating the optimization algorithm by optimizing the data detection model is implemented to realize self-evolution of the process optimization support system. Meanwhile, the dynamic updating mechanism enables the process optimization support system (i.e., the rule engine + the optimization algorithm) to have the ability of self-learning and self-improvement, which not only can solve the current production problem in time, but also can continuously adapt to future production changes to provide core technical support for long-term iteration of the process knowledge base.

[0048] In a second aspect, an embodiment of the present application provides a process knowledge base updating apparatus, comprising:

[0049] a collection unit, configured to collect running data of each device corresponding to a process route for producing a product and collect first production data of the product in a production process when the product is produced;

[0050] a first analysis unit, configured to analyze the running data of each device and the first production data to obtain a process detection result of the product;

[0051] a first optimization unit, configured to optimize the process route based on the process detection result to obtain an optimization result;

[0052] a first acquisition unit, configured to acquire second production data when production is continued based on the optimization result;

[0053] a first updating unit, configured to update a process knowledge base based on the running data of each device, the process detection result, the optimization result and the second production data.

[0054] In a third aspect, an embodiment of the present application provides a terminal device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the process knowledge base updating method in any one of the above first aspect.

[0055] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the process knowledge base updating method in any one of the above first aspect.

[0056] In a fifth aspect, an embodiment of the present application provides a computer program product, which, when executed on a terminal device, enables the terminal device to execute the process knowledge base updating method in any one of the above first aspect.

[0057] In a sixth aspect, an embodiment of the present application provides a process knowledge base updating system, comprising:

[0058] a data collection layer, configured to collect operation data of each device corresponding to a process route for producing a product and collect first production data of the product in a production process when the product is produced;

[0059] a data analysis layer, configured to analyze the operation data of each device and the first production data to obtain a process detection result of the product;

[0060] a data storage layer, configured to store process parameter ranges, historical data and abnormal cases of each production link of the product; the historical data comprises historical production data, historical operation data of each device and historical quality detection data;

[0061] a decision layer, configured to optimize the process route based on the process detection result to obtain an optimization result;

[0062] a feedback updating layer, configured to obtain second production data when production is continued based on the optimization result, and update a process knowledge base in the data storage layer based on the operation data of each device, the process detection result, the optimization result and the second production data. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0064] Figure 1 FIG. 1 is a structural schematic diagram of a process knowledge base updating system provided by an embodiment of the present application;

[0065] Figure 2 FIG. 2 is a working principle schematic diagram of a process knowledge base updating system provided by an embodiment of the present application;

[0066] Figure 3 FIG. 3 is an implementation flowchart of a process knowledge base updating method provided by an embodiment of the present application;

[0067] Figure 4 FIG. 4 is an implementation flowchart of a process knowledge base updating method provided by another embodiment of the present application;

[0068] Figure 5is an implementation flowchart of a process knowledge base updating method provided by another embodiment of the present application;

[0069] Figure 6 is a structural schematic diagram of a process knowledge base updating device provided by an embodiment of the present application.

[0070] Figure 7 is a structural schematic diagram of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0071] In actual applications, the development of intelligent manufacturing promotes the gradual realization of autonomous perception, intelligent interconnection, intelligent learning and autonomous decision-making of production systems. As a bridge between design and manufacturing, process decision-making is a key link of intelligent manufacturing. Taking LED display screens as an example, the production process of LED display screens is complex, involving chip packaging, module production, box assembly, detection and calibration and other links, and process parameters have a significant impact on product quality. In traditional production, process knowledge is mostly accumulated by manual experience and stored in the form of documents or static databases, that is, a process knowledge base is obtained. However, the existing process knowledge base has the following obvious defects:

[0072] Process knowledge updating lags behind, and cannot respond to production abnormalities in real time. In addition, abnormal handling relies on technical personnel to conduct on-site troubleshooting, which is time-consuming and has poor consistency. Real-time running data of equipment and production data of a manufacturing execution system (MES) are not effectively integrated, making it difficult to find optimization points of process knowledge. Historical abnormal cases and solutions do not form structured knowledge, and the reuse rate is low. At the same time, the process knowledge base has no self-evolution mechanism, and newly added process knowledge needs to be manually entered one by one, which has poor adaptability. In addition, the process knowledge base lacks closed-loop learning ability, and the data collection and knowledge updating are disconnected, which cannot be dynamically optimized based on real-time production results.

[0073] Therefore, an embodiment of the present application provides a process knowledge base updating system, which can collect and analyze production data and running data in real time to automatically update the process knowledge base, so as to solve the above defects.

[0074] Please refer to Figure 1 , Figure 1 Figure 1 is a structural schematic diagram of a process knowledge base updating system provided by an embodiment of the present application. Please refer to Figure 2 , Figure 2 is a working principle schematic diagram of a process knowledge base updating system provided by an embodiment of the present application.

[0075] As Figure 1 and Figure 2As shown, the process knowledge base updating system 1 comprises a data collection layer 10, a data analysis layer 20, a data storage layer 30, a decision layer 40, and a feedback updating layer 50. The data collection layer 10 is communicatively connected to the data analysis layer 20 and the decision layer 40, the feedback updating layer 50 is communicatively connected to the data storage layer 30 and the decision layer 40, and the data storage layer 30 is communicatively connected to the data analysis layer 20.

[0076] In an embodiment of the present application, the data collection layer 10 is configured to collect operation data of each device corresponding to a process route for producing a product and collect first production data of the product in a production process.

[0077] It should be noted that before producing the product, the decision layer 40 can obtain the process route of the product by using product design data, process data, quality data, the process knowledge base in the data storage layer 30, and a rule engine and an optimization algorithm provided by itself. The specific implementation process of obtaining the process route of the product can refer to an existing process route production method, which will not be described here.

[0078] In an embodiment of the present application, the operation data of each device includes but is not limited to operation parameters and operation state information. The operation parameters include but are not limited to soldering temperature, soldering time, device speed, pressure, etc. The operation state information includes but is not limited to normal operation, shutdown, fault alarm, etc.

[0079] The first production data includes but is not limited to production order information, product model, production quantity, production progress, and abnormal data (such as the number of defective products, defective types (such as dead lights, color differences, and virtual welding, etc.)) in a production process.

[0080] In an embodiment of the present application, the data collection layer 10 can include a device data collection module and an MES system data collection module.

[0081] In this embodiment, the device data collection module is configured to establish a connection with each device (such as a chip mounter and a reflow soldering furnace) corresponding to the process route of the product by using an industrial Internet of Things (IIoT) technology, and collect operation data of each device in real time.

[0082] The MES system data collection module is connected to an MES system to automatically obtain the first production data.

[0083] In an embodiment of the present application, the data analysis layer 20 is configured to analyze the operation data of each device and the first production data to obtain a process detection result of the product. The process detection result includes but is not limited to a process analysis result and an abnormality analysis result.

[0084] In an embodiment of the present application, the data analysis layer 20 can include a data preprocessing module, a machine learning model analysis module, and an anomaly analysis module.

[0085] In the present embodiment, the data preprocessing module is configured to clean the collected operation data and first production data of each device, remove noise data, and fill in missing values to ensure the accuracy and integrity of each data. Then, the data preprocessing module is configured to continue to perform standardization processing on each data after the above processing, unify the data format and unit, and obtain effective data for subsequent analysis and processing.

[0086] The machine learning model analysis module is configured to analyze the effective data after the above preprocessing by using multiple machine learning models to obtain a process analysis result.

[0087] For example, the multiple machine learning models can include but are not limited to a classification model, a regression model, and a clustering model. The classification model can identify and classify the types of product defects. The regression model can analyze the relationship between process parameters and product quality to evaluate the process effect. The clustering model can mine potential patterns and rules in the production data, i.e., obtain the process analysis result.

[0088] In practical applications, the classification model can include but is not limited to Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest, Gradient Boosting Tree (GBDT / XGBoost / LightGBM), and K-Nearest Neighbors (K-NN).

[0089] The regression model can include but is not limited to linear regression or ridge regression.

[0090] The clustering model can include but is not limited to K-Means, Hierarchical Clustering, or Density-Based Spatial Clustering of Applications with Noise (DBSCAN).

[0091] The anomaly analysis module is configured to identify and analyze the effective data based on historical data and a preset threshold to obtain an anomaly analysis result. The anomaly analysis result can include but is not limited to the link, time, influence range, and cause of the anomaly (such as unreasonable process parameter setting, equipment failure, raw material quality problem, etc.).

[0092] In the embodiments of the present application, the data storage layer 30 is configured to store process parameter ranges, historical data and abnormal cases of products at various production links; wherein the historical data includes historical production data, historical operation data of various devices and historical quality detection data. The historical quality detection data includes, but is not limited to, detection results of automated optical inspection (AOI) of products in a historical time period and system records of a manufacturing execution system (MES) and the like. The historical time period can be determined according to actual needs, which is not limited herein.

[0093] It should be noted that the data storage layer is provided with a process knowledge base. The process knowledge base can include a process parameter library, a historical production database and an abnormal case library.

[0094] The process parameter library is configured to store process parameter ranges of various production links in a process route, the historical production database is configured to store historical data, and the abnormal case library is configured to store abnormal cases.

[0095] In some possible embodiments, the data storage layer 30 can use a distributed database for data storage to meet the storage and rapid access requirements of a large amount of data, while ensuring the security and reliability of the data.

[0096] In the embodiments of the present application, the decision layer 40 is configured to optimize the process route based on the process detection results to obtain an optimization result.

[0097] In an embodiment of the present application, the decision layer 40 can include a rule engine module and an optimization algorithm module.

[0098] In the embodiments, the rule engine module is configured to make real-time judgments and decisions for the production process based on rules and historical data in the process knowledge base.

[0099] For example, when the data analysis layer 20 detects an abnormal situation or a process parameter deviating from a standard range, the rule engine module can issue corresponding instructions such as adjusting process parameters, stopping inspection and the like according to the preset rules through the rule engine.

[0100] The optimization algorithm module is configured to use optimization algorithms (such as genetic algorithms, particle swarm optimization algorithms and the like) to further optimize process parameters that need to be adjusted according to the process detection results output by the data analysis layer 20, to provide a more reasonable process parameter setting scheme for the production process, to optimize the process route, to obtain an optimization result, and to improve product quality and production efficiency.

[0101] In the embodiments of the present application, the feedback updating layer 50 is configured to acquire second production data based on the optimization result for continuing production, and update the process knowledge base in the data storage layer based on the operation data of each device, the process detection result, the optimization result, and the second production data.

[0102] In an embodiment of the present application, the feedback updating layer 50 can include a data feedback module, a knowledge base updating module, and a decision layer updating module.

[0103] In the embodiments, the data feedback module is configured to feed back the process detection result output by the data analysis layer 20, the optimization result output by the decision layer 40, the second production data, and the operation data of each device to the data storage layer 30 and the decision layer 40.

[0104] The knowledge base updating module is configured to automatically update the process knowledge base of the data storage layer according to the information fed back by the data feedback module. For example, when a new abnormal case and a solution are obtained based on the abnormal analysis result, the new abnormal case and the solution are added to the abnormal case library; the process parameter range in the process parameter library is updated based on the optimization result of the process parameter; and the second production data is stored in the historical production database.

[0105] The decision layer updating module is configured to update the rule engine and the optimization algorithm in the decision layer 40 according to the information fed back by the data feedback module. For example, when a new process rule or an abnormal processing rule is found, the new process rule or the abnormal processing rule is included in the rule engine; and the optimization algorithm is adjusted and improved based on the adjustment result of the second production data, so as to improve the optimization performance of the optimization algorithm.

[0106] In summary, the process knowledge base updating system provided by the embodiments of the present application has the following beneficial effects:

[0107] 1. The process knowledge base is automatically updated: through the closed-loop learning mechanism, the production data is collected and analyzed in real time, and the process knowledge base is automatically updated, which avoids the hysteresis of traditional manual updating and ensures the timeliness and accuracy of the process knowledge.

[0108] 2. The abnormal processing efficiency is improved: the abnormal data in the production process can be automatically identified and analyzed, the abnormal causes can be quickly located, and the corresponding solutions can be given based on the process knowledge base, which reduces the time and cost of manual investigation and improves the efficiency and accuracy of abnormal processing.

[0109] 3. The data resources are fully utilized: the operation data of each device and the production data of the MES system are integrated, and the value of the data is deeply mined through the machine learning model, which provides strong support for process optimization and decision making.

[0110] 4. Enhancing the reusability of process knowledge: The continuous enrichment and improvement of the exception case library enable the quick borrowing of historical solutions for similar new exceptions, improving the reusability of process knowledge and reducing the dependence on individual technical personnel's experience.

[0111] 5. Realizing continuous improvement of the system: Through the feedback update layer, the process knowledge base and the rule engine and optimization algorithm of the decision layer are continuously updated, enabling the system to continuously adapt to changes in the production process, realize continuous improvement, and improve the production quality and efficiency of products.

[0112] The following will take the LED display module dead light exception as an example, combined with the process knowledge base update system Figures 1 to 2 The working process of the process knowledge base update system will be described in detail:

[0113] I. Data acquisition layer:

[0114] The equipment data acquisition module collects real-time data such as the soldering temperature of the reflow soldering furnace at 275℃ (standard range: 240-260℃), soldering time of 3 seconds (standard range: 2-3 seconds), and the mounting pressure of the chip mounter at 0.3MPa (standard range: 0.2-0.25MPa).

[0115] The MES system data acquisition module obtains the production data of the batch of LED display modules, with a production quantity of 500 pieces, and finds that 30 pieces have dead light exceptions, with a defect rate of 6%.

[0116] II. Data analysis layer:

[0117] The data preprocessing module cleans and standardizes the collected data, removing invalid data.

[0118] The machine learning model analysis module identifies the defect type as dead light through classification algorithms, and uses regression algorithms to analyze that high soldering temperature and excessive mounting pressure have a strong correlation with dead light exceptions.

[0119] The exception analysis module combines historical data and standard ranges to determine that the main cause of the dead light exception is high soldering temperature leading to solder pad virtual welding and excessive mounting pressure damaging the chip.

[0120] III. Decision layer:

[0121] The rule engine issues adjustment instructions according to the rules in the process knowledge base, requiring the soldering temperature of the reflow soldering furnace to be reduced to 250℃ and the mounting pressure of the chip mounter to be adjusted to 0.23MPa.

[0122] The optimization algorithm further optimizes the process parameters adjusted by the rule engine, obtaining more accurate parameter setting suggestions.

[0123] IV. Feedback update layer:

[0124] The data feedback module feeds the abnormality analysis result, the above adjustment instruction and the optimization suggestion to the data storage layer and the decision layer.

[0125] The knowledge base updating module adds the dead lamp abnormality case (including abnormal phenomenon, cause, solution and treatment effect) to the abnormal case library, updates the standard parameter range of the reflow soldering and mounting link in the process parameter library according to the optimized process parameters, and stores the production data and the analysis result of the production data into the historical production database.

[0126] The decision layer updating module incorporates the new abnormality treatment rule into the rule engine, and adjusts and improves the optimization algorithm according to the application effect.

[0127] Through the above process, the process knowledge base updating system provided by the embodiment of the application realizes the self-evolution of the process knowledge base, and when similar dead lamp abnormality occurs again, the system can quickly call the historical case and the optimized process parameters to timely and effectively solve the problem.

[0128] Please refer to Figure 3 , Figure 3 is the implementation flowchart of the process knowledge base updating method provided by an embodiment of the application. In the embodiment of the application, the execution subject of the process knowledge base updating method is a terminal device. The terminal device includes but is not limited to: notebook computers, desktop computers and computer devices.

[0129] It should be noted that the terminal device is built-in with the above process knowledge base updating system.

[0130] As shown in Figure 3 , the process knowledge base updating method provided by an embodiment of the application can include S101-S105, which are described in detail as follows:

[0131] In S101, when producing a product, the running data of each device corresponding to the process route for producing the product is collected, and the first production data of the product in the production process is collected.

[0132] In one implementation manner of the embodiment of the application, the terminal device can collect the running data of each device in real time through each device corresponding to the process route for producing the product which is in communication connection with the terminal device. The communication connection manner includes but is not limited to wireless communication connection and wired communication connection. The manufacturing execution system for producing the product collects the first production data of the product in the production process in real time.

[0133] In another implementation form of the embodiment of the application, the terminal device can analyze the operation data of each device and the first production data in S102 to obtain the process detection result of the product.

[0134] In the embodiment of the application, after obtaining the operation data of each device and the first production data, the terminal device can input the operation data of each device and the first production data into the trained process detection model for processing to obtain the process detection result of the product.

[0135] It should be noted that the process detection model can be obtained by training a first deep learning model based on a preset sample set. Each sample data in the preset sample set includes sample data information (including the operation data of each device and the first production data) and a sample process detection result corresponding to the sample data information. When training the first deep learning model, the sample data information in each sample is used as the input of the first deep learning model, and the sample process detection result corresponding to the sample data information in each sample is used as the output of the first deep learning model. Through training, the first deep learning model can learn the corresponding relationship between all possible sample data information and sample process detection results, and the trained first deep learning model is used as the process detection model.

[0136] In an embodiment of the application, the process detection result includes a process analysis result and an abnormality analysis result. Therefore, in order to generate accurate process detection results and accurate abnormality analysis results, the terminal device can obtain the process detection result through steps S201-S203 as shown in the following. Figure 4

[0137] In S201, the operation data of each device and the first production data are subjected to a data preprocessing operation to obtain valid data.

[0138] In this embodiment, the terminal device can clean the collected operation data of each device and the first production data, remove noise data, and fill in missing values to ensure the accuracy and integrity of each data. Then, the terminal device continues to perform standardization processing on each data after the above processing to unify the data format and unit to obtain valid data for subsequent analysis and processing.

[0139] In S202, the valid data is input into a machine learning model for processing to obtain the process analysis result.

[0140] In this embodiment, the terminal device can input the valid data into a plurality of machine learning models for analysis to obtain the process analysis result.

[0141] ​It should be noted that the plurality of machine learning models can include, but are not limited to, a classification model, a regression model, and a clustering model. Among them, the classification model can identify and classify the bad types of the product; the regression model can analyze the relationship between the process parameters and the product quality, and evaluate the process effect; and the clustering model can mine the potential patterns and rules in the production data.

[0142] Therefore, in an embodiment of the present application, when the plurality of machine learning models include, but are not limited to, a classification model, a regression model, and a clustering model, and the process analysis result includes the bad type detection result of the product, the process effect, and the rule detection result, the terminal device can specifically implement step S202 according to the following steps, which are described in detail as follows:

[0143] inputting the effective data into the classification model for identification to obtain the bad type detection result;

[0144] inputting the effective data into the regression model for analysis to obtain the process effect;

[0145] inputting the effective data into the clustering model for processing to obtain the rule detection result.

[0146] In this embodiment, the terminal device can input the effective data into the classification model for identification to obtain the bad type detection result. Among them, the bad type detection result is used to describe the bad types (such as dead lights, color differences, and virtual welds) of the defective products existing in the product.

[0147] The terminal device can input the effective data into the regression model for analysis to analyze the relationship between the process parameters and the product quality, thereby evaluating the process effect.

[0148] The terminal device can input the effective data into the clustering model for processing to mine the potential patterns and rules in the production data, thereby obtaining the rule detection result.

[0149] In this embodiment, since the classification model can quickly identify the bad type in the new production data by learning the historical bad data, the effective data is input into the classification model for identification to obtain the bad type detection result, the automatic and accurate classification of product defects is realized, and the quality detection efficiency is improved; since the regression model can establish the mathematical relationship between the equipment operation parameters and the product quality / efficiency index, and output the quantitative result such as "the yield is improved by 1.2% when the welding line temperature is increased by 10℃(when the temperature is less than or equal to 240℃)", the effective data is input into the regression model for analysis to obtain the process effect, the correlation between the energy process parameters and the production index can be quantified, and the accurate process effect is obtained; since the clustering model does not need to preset a label, it can automatically cluster similar data into a class and discover potential patterns that are difficult for humans to detect, so the effective data is input into the clustering model for processing to obtain the rule detection result, the hidden process correlation rule can be mined, and the accurate rule detection result is obtained.

[0150] In S203, the effective data is input into the anomaly analysis model for processing to obtain the anomaly analysis result.

[0151] In this embodiment, the terminal device can input the effective data into the anomaly analysis model for identification and analysis to obtain the anomaly analysis result. The anomaly analysis result includes but is not limited to: the link, time, influence range of anomaly occurrence, and the cause of anomaly occurrence (such as unreasonable process parameter setting, equipment failure, raw material quality problem, etc.).

[0152] It should be noted that the anomaly analysis model can be obtained by training a third deep learning model pre-constructed based on a preset sample set. Each sample data in the preset sample set includes sample effective data and sample anomaly analysis result corresponding to the sample effective data. When training the third deep learning model pre-constructed, the sample effective data in each sample is taken as the input of the third deep learning model, and the sample anomaly analysis result corresponding to the sample effective data in each sample is taken as the output of the third deep learning model. Through training, the third deep learning model can learn the corresponding relationship between all possible sample effective data and sample anomaly analysis results, and the trained third deep learning model is taken as the anomaly analysis model.

[0153] In some possible embodiments, the anomaly analysis model can also be an isolation forest.

[0154] In this embodiment, the operation data and the first production data of each device are preprocessed to obtain effective data, which is the basis for ensuring the quality of subsequent analysis. Through preprocessing, invalid information can be eliminated, data format can be unified, abnormal values can be corrected, and the data input into the model can be ensured to be accurate and reliable, thereby providing high-quality input for subsequent model analysis, reducing the interference of garbage data on the results, and improving the analysis reliability from the source. Then, since the machine learning model can learn the rules in the historical production data to establish the mapping relationship between the device parameters and the product quality, the effective data is input into the machine learning model for processing to obtain the process analysis result, which can deeply mine the data correlation and realize the accurate evaluation of the process performance. Finally, since the abnormal analysis model can quickly identify abnormal points deviating from the normal range, the effective data is input into the abnormal analysis model for processing to obtain the abnormal analysis result, which can realize the accurate identification and positioning of production abnormalities and reduce the quality risk. It can be seen that in the method, the data preprocessing lays the data foundation for analysis, the machine learning model realizes the deep analysis of the process performance, the abnormal analysis model ensures the stability of the production process, and the three work together to make the process detection result not only comprehensively reflect the process state, but also accurately locate the problem.

[0155] In S103, the process route is optimized based on the process detection result to obtain an optimization result.

[0156] In the embodiment of the application, the terminal device can input the process detection result into the trained optimization model for processing to obtain an optimization result. The optimization result can include an optimization method and a specific optimization process. The optimization method includes, but is not limited to, adjustment and optimization of the operation parameters of the device, optimization of the process sequence of the process route, and optimization of the process parameters, etc. The specific optimization process is used to describe the specific optimization steps under the optimization method. For example, if the temperature of the reflow area of the reflow soldering furnace is too low, which leads to poor curing of the solder paste, the temperature is increased from 230°C to 240°C, the speed of the conveyor belt is reduced from 80 mm / min to 70 mm / min, and the curing time is ensured to be sufficient.

[0157] It should be noted that the optimization model can be obtained by training a pre-constructed second deep learning model based on a preset sample set. Each sample data in the preset sample set includes sample process detection results and sample optimization results corresponding to the sample process detection results. When training the pre-constructed second deep learning model, the sample process detection results in each sample are used as the input of the second deep learning model, and the sample optimization results corresponding to the sample process detection results in each sample are used as the output of the second deep learning model. Through training, the second deep learning model can learn the corresponding relationship between all possible sample process detection results and sample optimization results, and the trained second deep learning model is used as the optimization model.

[0158] In an embodiment of the present application, in order to realize the standardization, precision and high efficiency of process optimization, the terminal device can specifically implement step S103 according to the following steps, which are described in detail as follows:

[0159] The process detection result is input into a rule engine for processing to obtain an adjustment instruction.

[0160] If the adjustment instruction is to adjust the process parameters, the parameter to be adjusted and the parameter adjustment mode are determined.

[0161] The parameter to be adjusted is adjusted based on the parameter adjustment mode to obtain a first parameter.

[0162] The first parameter is optimized based on an optimization algorithm to obtain the optimization result.

[0163] In the embodiment, after obtaining the process detection result, the terminal device can convert the process detection result (including the defect type detection result, process effect, rule detection result and abnormality analysis result) into a rule matching field based on keyword extraction. For example: extract defect type = solder paste cavity (confidence 95%), process effect deviation rate = 3.2% (to be improved), abnormal parameter = reflow solder conveyor belt speed 85 mm / min (standard 70-80 mm / min).

[0164] Then, the terminal device can match the parsed field, i.e., the rule matching field, with each rule in the rule engine one by one; if multiple rules are triggered at the same time (such as solder paste cavity triggering a defect type corresponding rule and conveyor belt speed exceeding triggering a parameter exceeding adjustment rule), the rules are sorted according to the priority of abnormality level > defect proportion > effect deviation rate (high-level abnormality rule is preferred, such as high-level abnormality rule priority is higher than medium-level effect optimization rule).

[0165] Therefore, the terminal device can generate the adjustment instruction according to the above-mentioned rule of the highest priority. The adjustment instruction can include an instruction type (such as parameter adjustment / device calibration / stopover maintenance), an associated process, a target parameter, and an adjustment direction, and the like. For example, the instruction type is process parameter adjustment, the associated process is reflow soldering, the target parameter is the conveyor belt speed, and the adjustment direction is to reduce from 85 mm / min to 78 mm / min.

[0166] In some possible embodiments, the terminal device can check the rationality of the adjustment instruction (such as whether the adjusted parameter is within the range allowed by the device hardware, such as the maximum adjustable reflow soldering temperature of 260°C, to avoid the instruction exceeding the device capability), and output the final adjustment instruction after the check is passed.

[0167] In this embodiment, after obtaining the adjustment instruction, the terminal device can extract the instruction type from the adjustment instruction, and determine whether the adjustment instruction is to adjust the process parameter according to the instruction type.

[0168] When the terminal device detects that the adjustment instruction is to adjust the process parameter, the terminal device can determine the parameter to be adjusted from the target parameter included in the adjustment instruction, and determine the parameter adjustment mode from the adjustment direction in the adjustment instruction.

[0169] In this embodiment, after determining the parameter to be adjusted and the parameter adjustment mode, the terminal device can adjust the parameter to be adjusted based on the parameter adjustment mode to obtain an adjusted first parameter.

[0170] After obtaining the first parameter, the terminal device can optimize the first parameter based on an optimization algorithm to obtain an optimization result.

[0171] Specifically, the terminal device can determine a target function according to the actual demand of the product. Then, the terminal device can generate a plurality of initial parameter individuals randomly in the parameter range allowed by the device based on the first parameter, construct an initial parameter set, and obtain an initialization population. Then, the terminal device can evaluate each initial parameter individual in the initial parameter set based on the target function to obtain the performance of each initial parameter individual. Then, the terminal device can screen high-quality parameter individuals according to the performance of each initial parameter individual, and process the high-quality parameter individuals through selection, crossover, and mutation operations to generate a next generation population. The process is repeated to obtain an optimal solution set. Then, the terminal device can determine the optimal value in the optimal solution set as the optimization result.

[0172] In this embodiment, since the rule engine is built-in with the professional experience rules in product production, the process detection result is input to the rule engine for processing, the adjustment direction can be quickly output by matching the process detection result with the preset rules, and the accurate adjustment instruction can be obtained, so as to provide the standardized and automatic decision basis for process optimization; then, based on the output of the rule engine, that is, adjusting the process parameters, the accurate to-be-adjusted parameters and adjustment mode can be determined, so as to reduce the interference on irrelevant parameters, reduce the trial and error cost of parameter adjustment, and the standardized adjustment mode ensures the controllability of the optimization process, avoiding production fluctuation caused by parameter mutation; then, the to-be-adjusted parameters are adjusted based on the parameter adjustment mode, and the first parameters are obtained, the adjustment instruction can be converted into an executable preliminary parameter scheme, and a basis is provided for subsequent deep optimization; finally, since the optimization algorithm can search for a more optimal parameter combination on the basis of the first parameters and in combination with multi-dimensional constraint conditions, the first parameters are optimized based on the optimization algorithm to obtain the optimization result, the global optimization of the parameters can be realized, and the determination accuracy of the optimization result is improved.

[0173] In S104, second production data based on the optimization result for continuing production is obtained.

[0174] In the embodiment of the application, after obtaining the optimization result, the terminal device can optimize the process route of the product based on the optimization result to obtain an optimized process route. Then, the terminal device can produce the product again based on the optimized process route and obtain second production data based on the optimized process route for continuing production.

[0175] In some possible embodiments, the data acquisition layer in the process knowledge base updating system can obtain the second production data based on the optimized process route for continuing production in real time when the product is produced again based on the optimized process route.

[0176] In S105, the process knowledge base is updated based on the running data of each device, the process detection result, the optimization result, and the second production data.

[0177] In the embodiment of the application, after obtaining the running data of each device, the process detection result, the optimization result, and the second production data, the terminal device can update the data corresponding to each device in the process knowledge base based on the running data of each device, the process detection result, the optimization result, and the second production data.

[0178] In an embodiment of the present application, the process knowledge base includes an abnormal case library, a process parameter library, and a historical production database. Therefore, in order to improve the updating accuracy of the process knowledge base, a knowledge base ecology that supports and verifies each other is formed, which can provide instant support for daily production and data basis for long-term process optimization. The terminal device can implement step S105 by steps S301-S305 as shown in the figure, which are described in detail as follows. Figure 5

[0179] In S301, the process detection result and the optimization result are analyzed to obtain an abnormal case and an abnormal adjustment mode.

[0180] In the embodiment, the terminal device can extract abnormal detection information from the process detection result (such as an abnormal analysis result) and construct an abnormal case according to five elements of abnormal ID- abnormal type- associated process- abnormal parameter- influence range.

[0181] In some possible embodiments, the terminal device can input the process detection result into a trained abnormal detection model for processing to obtain the abnormal case. The abnormal detection model can be trained by a pre-constructed neural network model.

[0182] Meanwhile, the terminal device can analyze the optimization result to disassemble and obtain adjustment logic, and classify and arrange according to adjustment ID- associated abnormal ID- adjustment parameter- adjustment step- execution device to obtain an abnormal adjustment mode.

[0183] In some possible embodiments, the terminal device can input the optimization result into a trained mode detection model for processing to obtain the abnormal adjustment mode. The mode detection model can be trained by a pre-constructed neural network model.

[0184] In S302, the second production data is analyzed to obtain an adjustment result of the abnormal adjustment mode.

[0185] In S303, the abnormal case, the abnormal adjustment mode, and the adjustment result are stored in the abnormal case library.

[0186] In the embodiment, the terminal device can extract key indicators related to the above-mentioned abnormal case and abnormal adjustment mode from the second production data, and compare the differences of the key indicators before and after optimization to obtain the adjustment result of the abnormal adjustment mode.

[0187] Then, the terminal device can store the abnormal case, the abnormal adjustment mode, and the adjustment result in the abnormal case library.

[0188] ​In some possible embodiments, the terminal device can employ a relational database (such as MySQL) to build a table structure of the abnormal case library. The core fields can include:

[0189] basic information, such as abnormal ID, associated production batch, entry time, and data source;

[0190] abnormal details, such as abnormal type, associated process, abnormal parameter (including standard value / actual value), and impact range;

[0191] adjustment information, such as adjustment ID, adjustment parameter, adjustment step, and execution device;

[0192] effect information, such as adjustment result rating, core index improvement data, and second production data number.

[0193] In S304, the process parameter library is updated based on the optimization result.

[0194] In this embodiment, the terminal device can update the parameter range corresponding to the parameter related to the optimization result in the process parameter library based on the optimization result.

[0195] In an embodiment of the present application, in order to realize the upgrade of the process parameter library from static storage to dynamic adaptation, to make the process parameter library always maintain a dynamic optimal state, to adapt to the change of production conditions in real time, and to provide stable and reliable parameter support for process design, production operation, and quality control, the terminal device can specifically implement step S304 according to the following steps, which are described in detail as follows:

[0196] The optimization result is analyzed to determine the optimization parameter and the optimization mode;

[0197] An initial parameter range of the optimization parameter is obtained from the process parameter library;

[0198] The initial parameter range is updated based on the optimization mode to obtain a standard parameter range.

[0199] In this embodiment, the terminal device can extract the optimization parameter and the optimization mode from the optimization result.

[0200] In some possible embodiments, the terminal device can input the optimization result into a trained optimization detection model for processing to obtain the optimization parameter and the optimization mode. The optimization detection model can be trained by a pre-constructed neural network model.

[0201] Then, the terminal device can obtain an initial parameter range of the optimization parameter from the process parameter library, and update the initial parameter range based on the optimization mode to obtain a standard parameter range.

[0202] In some possible embodiments, when the optimization mode is numerical interval adjustment, the corresponding update mode can be interval upward / downward adjustment of the initial parameter range or interval expansion / reduction; when the optimization mode is numerical single-point optimization, the corresponding update mode can be direct replacement or conditional limitation; and when the optimization mode is scenario-based parameter addition, the corresponding update mode can be adding a new scenario parameter.

[0203] In this embodiment, since the optimization result can contain multi-dimensional information, key optimization parameters and corresponding optimization modes can be extracted through analysis of the optimization result; then, since the initial range stored in the process parameter library is a standard based on historical production experience, obtaining the initial parameter range of the optimization parameter from the process parameter library can provide a historical benchmark for subsequent parameter adjustment, ensuring the continuity and traceability of the updating process; finally, the initial parameter range is updated based on the optimization mode to obtain a standard parameter range, so that the process parameter library can reflect the latest process optimal state, improving the guiding value of the process parameter library.

[0204] In S305, the running data of each device, the second production data, and the adjustment result are stored in association in the historical production database.

[0205] In this embodiment, the terminal device can store the running data of each device, the second production data, and the adjustment result in association in the historical production database.

[0206] In this embodiment, since the abnormality analysis result in the process detection result is combined with the adjustment logic in the optimization result, a standardized abnormal case (containing abnormal phenomenon and influence range) and a corresponding abnormal adjustment mode can be formed, so that the core information of the production problem and the solution can be accurately refined through analysis of the process detection result and the optimization result, thereby obtaining an accurate abnormal case and an abnormal adjustment mode; then, since the actual effect of the adjustment mode can be quantitatively evaluated through the second production data after optimization, the adjustment result of the abnormal adjustment mode can be obtained through analysis of the second production data; then, the abnormal case, the abnormal adjustment mode, and the adjustment result are stored in association in the abnormal case library, thereby constructing a closed-loop knowledge system of problem, solution, and effect, ensuring the integrity of each abnormal case in the abnormal case library; meanwhile, since the process parameter library stores the standard process parameter range of each process, the process parameter library can be updated based on the optimization result to dynamically refresh the parameter standard, so that the production guidance is more suitable for the actual conditions such as the current device state and material characteristics, and the timeliness and optimality of the process benchmark parameter are also ensured; meanwhile, the running data of each device, the second production data, and the adjustment result are stored in association in the historical production database, thereby forming a complete production data chain and providing a large number of high-quality samples for data analysis and model iteration.

[0207] As can be seen from the above, the process knowledge base updating method provided by the embodiments of the present application can comprehensively obtain the basic information in the production process by collecting the operation data of each device corresponding to the process route for producing the product and collecting the first production data of the product in the production process. Meanwhile, the working state of the device can be understood in real time by collecting the operation data of the device, which provides a basis for subsequent analysis, and the actual performance of the product in the production process can be intuitively reflected by collecting the first production data of the product. Then, the operation data of each device and the first production data are analyzed, that is, the operation data of the device is combined with the production data of the product for analysis, which can accurately determine whether the product meets the process standard, so as to obtain an accurate process detection result of the product. Then, the process route can be adjusted in a targeted manner based on the process detection result, the process route is optimized, and an accurate optimization result is obtained. Then, the second production data when the production is continued based on the optimization result is obtained, which can intuitively determine whether the process optimization achieves the expected effect and provide new information feedback. Finally, the process knowledge base is updated based on the operation data of each device, the process detection result, the optimization result and the second production data, so that the data and results from multiple aspects are integrated into the process knowledge base, and the practicality and accuracy of the process knowledge base are improved. It can be seen that the process knowledge base can be automatically updated, the lag of traditional manual updating is avoided, and the timeliness and accuracy of the process knowledge are ensured.

[0208] In an embodiment of the present application, since the decision layer in the process knowledge base updating system includes a rule engine and an optimization algorithm, in order to enable the support system of process optimization (i.e. rule engine + optimization algorithm) to have the ability of self-learning and self-improvement, not only can the current production problem be solved immediately, but also can the future production changes be continuously adapted, which provides core technical support for long-term iteration of the process knowledge base. After step S104, the terminal device can specifically execute the following steps, which are described in detail as follows:

[0209] analyzing the second production data and the optimization result to obtain an abnormal processing mode;

[0210] updating the rule engine based on the abnormal processing mode;

[0211] and / or,

[0212] detecting the second production data to obtain a data detection result;

[0213] updating the optimization algorithm based on the data detection result.

[0214] In this embodiment, the terminal device can extract two types of key information, i.e., unmet expected indicators and newly occurring abnormalities, from the second production data, and determine the causes of the abnormalities in combination with the optimization result.

[0215] Specifically, the terminal device can compare the corresponding expected effect of the optimization result with the corresponding actual effect of the second production data to obtain unmet expected indicators. The terminal device can analyze newly added quality problems in the second production data (for example, after optimizing the reflow soldering parameters, chip pin oxidation occurs, which is not present in the process detection result before optimization), to obtain newly occurring abnormalities.

[0216] Subsequently, the terminal device can input the above unmet expected indicators, newly occurring abnormalities, and second production data into the trained abnormality processing mode detection model for processing to obtain an abnormality processing mode. The abnormality processing mode detection model is trained by a pre-constructed neural network model.

[0217] In this embodiment, the terminal device can convert the above abnormality processing mode into a newly added rule or a modified rule in the rule engine, and integrate it into the rule engine according to the data structure of trigger condition-execution instruction-priority, to complete the update of the rule engine.

[0218] In this embodiment, the terminal device can detect the second production data from four dimensions, i.e., integrity, accuracy, effectiveness, and correlation, to obtain a data detection result.

[0219] The integrity detection is to check whether the key indicators are complete (for example, whether the core data such as production efficiency and defective product rate is 100% collected without null values), and if the missing rate is greater than 1%, the missing data needs to be supplemented or supplemented by interpolation.

[0220] The accuracy detection is to compare the artificial re-inspection data with the automatically collected data (for example, the component displacement rate detected by AOI and the artificial sampling result), and if the deviation rate is greater than 0.2%, the automatic collection equipment needs to be calibrated (for example, the AOI optical parameters are adjusted).

[0221] The effectiveness detection is to eliminate abnormal values (for example, production efficiency of 0 pieces / hour caused by equipment failure) and retain data under normal production state (confirmed by equipment operation log without downtime and alarm record).

[0222] The correlation detection is to verify the logical correlation between the optimization parameters and the second production data (for example, whether the component displacement rate decreases synchronously after the placement accuracy is improved), and if the correlation is reversed (for example, the accuracy is improved but the displacement rate increases), the data collection link needs to be checked for interference (for example, the sensor position is offset).

[0223] In this embodiment, the terminal device can determine the update direction of the optimization algorithm according to the deviation between the data detection result and the actual effect-expected effect of the second production data.

[0224] It should be noted that the update direction can include target function weight adjustment, algorithm parameter optimization, and feature dimension supplement, etc.

[0225] Among them, the target function weight adjustment is that if a certain target function does not reach the expectation (such as the non-conforming product rate decreases but the energy consumption rises in the cost target function), the weight distribution can be adjusted. For example, the original target function is Min f3=non-conforming product rate*100+energy consumption*0.1; because the energy consumption rises from 120kW / h to 130kW / h (up 8.3%), it can be adjusted to Min f3=non-conforming product rate*100+energy consumption*0.2 (increase the weight of energy consumption to balance the non-conforming product and energy consumption).

[0226] The algorithm parameter optimization is that if the convergence speed of the optimization algorithm is slow (such as the NSGA-II algorithm still does not get a stable solution set after 50 iterations) or the solution set quality is low (such as only 1 in the Pareto optimal solution set meets the quality requirement), the algorithm core parameters (such as population size, crossover probability or mutation probability, etc.) can be adjusted.

[0227] The feature dimension supplement is that if the data detection finds that the parameter does not affect the effect index not included in the optimization algorithm, the parameter is added as the algorithm input feature to improve the mapping relationship between the parameter and the effect.

[0228] Then, the terminal device can update the optimization algorithm based on the above update direction.

[0229] It can be seen that the process knowledge base updating method provided in the embodiment can obtain accurate abnormal processing methods by analyzing the second production data and the optimization result, because the second production data is real feedback after the optimization result is implemented. Then, the rule engine is updated based on the abnormal processing methods, so that the rule engine can continuously accumulate new experience, and when the same type of abnormality occurs again in subsequent production, the rule engine can directly output the corresponding adjustment instruction without repeated analysis. Meanwhile, the data detection result is obtained by detecting the second production data, which provides verification data in a real scene for iteration of the optimization algorithm. Therefore, the optimization algorithm is updated based on the data detection result, which improves the generalization ability and prediction accuracy of the optimization algorithm and makes it more suitable for actual production requirements. It can be seen that the method iterates in two paths after obtaining the second production data, that is, analyzing data to optimize abnormal processing methods to update the rule engine and detecting data to optimize the model to update the optimization algorithm, so as to realize self-evolution of the process optimization support system. Meanwhile, the dynamic updating mechanism makes the process optimization support system (that is, the rule engine and the optimization algorithm) have the ability of self-learning and self-improvement, which not only can solve the current production problem immediately, but also can continuously adapt to future production changes to provide core technical support for long-term iteration of the process knowledge base.

[0230] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0231] Corresponding to the process knowledge base updating method described in the above embodiment, Figure 6 A structural schematic diagram of a process knowledge base updating device provided by an embodiment of the present application is shown, and only parts related to the embodiments of the present application are shown for ease of description. For details, refer to Figure 6 The process knowledge base updating device 600 includes an acquisition unit 61, a first analysis unit 62, a first optimization unit 63, a first acquisition unit 64, and a first updating unit 65. Wherein:

[0232] The acquisition unit 61 is configured to acquire running data of each device corresponding to a process route for producing a product when the product is produced, and acquire first production data of the product in the production process.

[0233] The first analysis unit 62 is configured to analyze the running data of each device and the first production data to obtain a process detection result of the product.

[0234] The first optimization unit 63 is configured to optimize the process route based on the process detection result to obtain an optimization result.

[0235] The first obtaining unit 64 is configured to obtain second production data obtained when production is continued based on the optimization result.

[0236] The first updating unit 65 is configured to update the process knowledge base based on the operation data of each device, the process detection result, the optimization result, and the second production data.

[0237] In an embodiment of the present application, the process detection result includes a process analysis result and an abnormality analysis result, and the first analysis unit 62 specifically includes a first processing unit, a second processing unit, and a third processing unit. Wherein:

[0238] The first processing unit is configured to perform a data preprocessing operation on the operation data of each device and the first production data to obtain effective data.

[0239] The second processing unit is configured to input the effective data into a machine learning model for processing to obtain the process analysis result.

[0240] The third processing unit is configured to input the effective data into an abnormality analysis model for processing to obtain the abnormality analysis result.

[0241] In an embodiment of the present application, the machine learning model includes a classification model, a regression model, and a clustering model, and the process analysis result includes a defect type detection result of the product, a process effect, and a regularity detection result. The second processing unit specifically includes an identification unit, a second analysis unit, and a fourth processing unit. Wherein:

[0242] The identification unit is configured to input the effective data into the classification model for identification to obtain the defect type detection result.

[0243] The second analysis unit is configured to input the effective data into the regression model for analysis to obtain the process effect.

[0244] The fourth processing unit is configured to input the effective data into the clustering model for processing to obtain the regularity detection result.

[0245] In an embodiment of the present application, the first optimization unit 63 specifically includes a fifth processing unit, a first determination unit, a first adjustment unit, and a second optimization unit. Wherein:

[0246] The fifth processing unit is configured to input the process detection result into a rule engine for processing to obtain an adjustment instruction.

[0247] The first determination unit is configured to determine a parameter to be adjusted and a parameter adjustment method if the adjustment instruction is to adjust a process parameter.

[0248] The first adjusting unit is configured to adjust the to-be-adjusted parameter based on the parameter adjustment mode, to obtain a first parameter.

[0249] The second optimizing unit is configured to optimize the first parameter based on an optimization algorithm, to obtain the optimization result.

[0250] In an embodiment of the present application, the process knowledge base includes an abnormal case library, a process parameter library, and a historical production database, and the first updating unit 65 specifically includes a third analyzing unit, a fourth analyzing unit, a first storing unit, a second updating unit, and a second storing unit. Wherein:

[0251] The third analyzing unit is configured to analyze the process detection result and the optimization result, to obtain an abnormal case and an abnormal adjustment mode.

[0252] The fourth analyzing unit is configured to analyze the second production data, to obtain an adjustment result of the abnormal adjustment mode.

[0253] The first storing unit is configured to store the abnormal case, the abnormal adjustment mode, and the adjustment result in the abnormal case library.

[0254] The second updating unit is configured to update the process parameter library based on the optimization result.

[0255] The second storing unit is configured to store the running data of each device, the second production data, and the adjustment result in the historical production database.

[0256] In an embodiment of the present application, the second updating unit specifically includes a fifth analyzing unit, a second acquiring unit, and a third updating unit. Wherein:

[0257] The fifth analyzing unit is configured to analyze the optimization result, to determine an optimization parameter and an optimization mode.

[0258] The second acquiring unit is configured to acquire an initial parameter range of the optimization parameter from the process parameter library.

[0259] The third updating unit is configured to update the initial parameter range based on the optimization mode, to obtain a standard parameter range.

[0260] In an embodiment of the present application, the process knowledge base updating device 600 further includes a sixth analyzing unit and a fourth updating unit, and / or a detecting unit and a fifth updating unit. Wherein:

[0261] The sixth analyzing unit is configured to analyze the second production data and the optimization result, to obtain an abnormal processing mode.

[0262] The fourth update unit is used to update the rule engine based on the exception handling method.

[0263] The detection unit is used to detect the second production data and obtain the data detection results.

[0264] The fifth update unit is used to update the optimization algorithm based on the data detection results.

[0265] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0266] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0267] Figure 7 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. Figure 7 As shown, the terminal device 7 of this embodiment includes: at least one processor 70 ( Figure 7 (Only one is shown in the diagram), memory 71, and computer program 72 stored in said memory 71 and executable on said at least one processor 70, wherein the processor 70 executes said computer program 72 to implement the steps in any of the above embodiments of the process knowledge base update method.

[0268] The terminal device 7 may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art will understand that... Figure 7 This is merely an example of terminal device 7 and does not constitute a limitation on terminal device 7. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0269] The processor 70 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0270] The memory 71 can be an internal storage unit of the terminal device 7 in some embodiments, for example, a memory of the terminal device 7. The memory 71 can also be an external storage device of the terminal device 7 in other embodiments, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, or the like. Further, the memory 71 can include both an internal storage unit and an external storage device of the terminal device 7. The memory 71 is used to store an operating system, an application program, a boot loader, data, and other programs, for example, program codes of the computer program, and the like. The memory 71 can also be used to temporarily store data that has been output or will be output.

[0271] The embodiments of the present application further provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps in the above-mentioned various method embodiments.

[0272] The embodiments of the present application provide a computer program product. When the computer program product is run on a terminal device, the terminal device is caused to implement the steps in the above-mentioned various method embodiments.

Claims

1. A process knowledge base updating method characterized by, Comprising: collecting running data of each device corresponding to a process route for producing a product and collecting first production data of the product in a production process when the product is produced; analyzing the running data of each device and the first production data to obtain a process detection result of the product; optimizing the process route based on the process detection result to obtain an optimization result; obtaining second production data when production is continued based on the optimization result; updating a process knowledge base based on the running data of each device, the process detection result, the optimization result and the second production data.

2. The process knowledge base updating method of claim 1, wherein, The process detection result includes process analysis result and abnormal analysis result, and the analysis of the running data of each device and the first production data to obtain the process detection result of the product includes: performing data preprocessing operation on the running data of each device and the first production data to obtain effective data; inputting the effective data into a machine learning model for processing to obtain the process analysis result; inputting the effective data into an abnormal analysis model for processing to obtain the abnormal analysis result.

3. The process knowledge base updating method of claim 2, wherein, The machine learning model includes a classification model, a regression model and a clustering model, and the process analysis result includes a bad type detection result, a process effect and a rule detection result of the product; and the inputting of the effective data into the machine learning model for processing to obtain the process analysis result includes: inputting the effective data into the classification model for identification to obtain the bad type detection result; inputting the effective data into the regression model for analysis to obtain the process effect; inputting the effective data into the clustering model for processing to obtain the rule detection result.

4. The process knowledge base updating method of claim 1, wherein, The optimization of the process route based on the process detection result to obtain the optimization result includes: inputting the process detection result into a rule engine for processing to obtain an adjustment instruction; if the adjustment instruction is to adjust a process parameter, determining a parameter to be adjusted and a parameter adjustment mode; adjusting the parameter to be adjusted based on the parameter adjustment mode to obtain a first parameter; optimizing the first parameter based on an optimization algorithm to obtain the optimization result.

5. The process knowledge base updating method of claim 1, wherein, The process knowledge base includes an abnormal case library, a process parameter library and a historical production database, and the updating of the process knowledge base based on the running data of each device, the process detection result, the optimization result and the second production data includes: analyzing the process detection result and the optimization result to obtain an abnormal case and an abnormal adjustment mode; analyzing the second production data to obtain an adjustment result of the abnormal adjustment mode; storing the abnormal case, the abnormal adjustment mode and the adjustment result in association to the abnormal case library; updating the process parameter library based on the optimization result; storing the running data of each device, the second production data and the adjustment result in association to the historical production database.

6. The process knowledge base updating method of claim 5, wherein, The updating of the process parameter library based on the optimization result includes: analyzing the optimization result to determine an optimization parameter and an optimization mode; obtaining an initial parameter range of the optimization parameter from the process parameter library; updating the initial parameter range based on the optimization mode to obtain a standard parameter range.

7. The process knowledge base updating method of any one of claims 1-6, wherein, After the second production data based on the optimization result is obtained, the method further comprises: analyzing the second production data and the optimization result to obtain an abnormal processing mode; updating the rule engine based on the abnormal processing mode; and / or, detecting the second production data to obtain a data detection result; updating the optimization algorithm based on the data detection result.

8. A process knowledge base updating apparatus characterized by comprising: The method comprises: a collection unit configured to collect operation data of each device corresponding to a process route for producing a product when the product is produced, and collect first production data of the product in a production process; a first analysis unit configured to analyze the operation data of each device and the first production data to obtain a process detection result of the product; a first optimization unit configured to optimize the process route based on the process detection result to obtain an optimization result; a first obtaining unit configured to obtain second production data based on the optimization result when production is continued; a first updating unit configured to update a process knowledge base based on the operation data of each device, the process detection result, the optimization result, and the second production data.

9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor implements the process knowledge base updating method according to any one of claims 1 to 7 when executing the computer program.

10. A process knowledge base updating system characterized by comprising: The method comprises: a data collection layer configured to collect operation data of each device corresponding to a process route for producing a product when the product is produced, and collect first production data of the product in a production process; a data analysis layer configured to analyze the operation data of each device and the first production data to obtain a process detection result of the product; a data storage layer configured to store process parameter ranges, historical data, and abnormal cases of the product in each production link; the historical data comprises historical production data, historical operation data of each device, and historical quality detection data; a decision layer configured to optimize the process route based on the process detection result to obtain an optimization result; a feedback updating layer configured to obtain second production data based on the optimization result when production is continued, and update a process knowledge base in the data storage layer based on the operation data of each device, the process detection result, the optimization result, and the second production data.

Images