Method and system for managing process for process optimization

Abstract

The present invention relates to a method and system for managing a process. The method for managing a process includes the steps of: receiving, by a data input unit, an input of process-related data including sensor data, abnormality information, and process information collected from a device or a sensor installed at a manufacturing site; generating, by an AI structure generation unit, a characteristic factor map by applying clustering using machine learning to the process-related data to identify, in a ranking order, a hierarchical relationship between sensor data for a specific result and a sensor affecting the specific result, during a process at the manufacturing site; generating, by an AI rule generation unit, a management rule by applying the clustering using the machine learning to the process-related data to analyze a safety section and a risk section for each sensor used in each process at the manufacturing site, and calculate a management range; and managing, by an integrated management unit, the process at the manufacturing site by using the characteristic factor map and the management rule. According to the present invention, a process can be managed by automatically generating a characteristic factor map for setting a management structure that enables optimized operation of process management on the basis of machine learning, and a management rule for calculating a management range.

Description

Process management method and system for process optimization

[0001] The present invention relates to a process management method and a system thereof, and more specifically, to a process management method and a system thereof capable of managing by creating a management structure and rules to enable process optimization.

[0002]

[0003] In general, process management refers to the activities of planning, controlling, and improving the flow of each stage in the production process of products or services to enhance efficiency and quality. Process management primarily encompasses quality control, production planning and scheduling, worker performance management, and cost management, and various techniques and tools are utilized.

[0004] Meanwhile, process management for process optimization seeks to find ways to reduce costs while minimizing resource waste, maximizing efficiency, and maintaining quality during the production process. This aims to improve productivity, reduce costs, and enhance quality, and involves continuous activities to analyze and improve every stage of the process.

[0005] To optimize process management, it is fundamentally necessary to conduct a current state analysis, which involves meticulously analyzing each stage of the production process to identify bottlenecks, unnecessary tasks, and resource waste, as well as a process analysis, which evaluates process efficiency by collecting data generated during production. In addition, real-time monitoring of the process, anomaly detection and predictive maintenance, quality control, automation, and smart process management are also required.

[0006] However, in small and medium-sized production plants lacking manpower or technology, it is difficult to even establish the management structure or scope that forms the basis of process management.

[0007] Therefore, in order to optimize process management, it is necessary to have a method to manage processes by systematizing various data collected from production processes at the manufacturing site and automatically generating management rules that can determine the management structure and scope.

[0008]

[0009] Accordingly, the objective of the present invention is to provide a process management method and a system that can manage a process by utilizing machine learning technology to automatically generate a characteristic cause diagram for setting a management structure and a management rule for calculating a management range.

[0010]

[0011] A process management method according to the present invention for achieving the above objective comprises: a step of receiving process-related data including sensor data, abnormal information, and process information collected from a device or sensor installed at a manufacturing site in a data input unit; a step of generating a characteristic factor diagram in which the hierarchical relationship of the sensor data regarding a specific result in the process of the manufacturing site and the sensors that influenced the specific result can be identified in a ranking order by applying clustering using machine learning to the process-related data in an AI structure generation unit; a step of generating a management rule in which the safety zone and the risk zone are analyzed for each sensor used in each process of the manufacturing site by applying clustering using machine learning to the process-related data in an AI rule generation unit to generate a management rule that can calculate a management range; and a step of managing the process of the manufacturing site using the characteristic factor diagram and the management rule in an integrated management unit.

[0012] The above ranking can be calculated using Randomforest importance, LinearRegressor coefficients, and Correlation coefficients regarding the relationship between the above specific result and the above sensor data value, and in the process of generating the above management rule, when analyzing the above safe zone and risk zone, Randomforest importance, LinearRegressor coefficients, Naive Bayes variable contribution, and Correlation coefficients can be used.

[0013] Additionally, the method may further include the step of providing a menu screen for setting a workspace, setting a work environment, setting work data, and setting parameters and analysis elements to be applied to the machine learning for generating the characteristic factor diagram, and the step of providing a menu screen for setting a workspace, setting a work environment, setting work data, and setting parameters and analysis elements to be applied to the machine learning for generating the management rule.

[0014] In addition, the method may further include the step of providing a menu screen for modifying the management rule or creating a new rule according to a user command, and the step of simulating the management rule using virtual data on the integrated server and modifying the management rule by reflecting the simulation results.

[0015] And, the process management system according to the present invention for achieving the above objective includes: a data input unit that receives process-related data including sensor data, abnormal information, and process information collected from a device or sensor installed at a manufacturing site; an AI structure generation unit that generates a characteristic factor diagram by applying clustering using machine learning to the process-related data, which can identify the hierarchical relationship of the sensor data regarding a specific result in the process of the manufacturing site and the sensors that influenced the specific result in a ranking order; an AI rule generation unit that generates a management rule capable of calculating a management range by analyzing safety zones and risk zones for each sensor used in each process of the manufacturing site by applying clustering using machine learning to the process-related data; and an integrated management unit that manages the process of the manufacturing site using the characteristic factor diagram and the management rule.

[0016] In addition, it may further include a data collector for collecting process-related data, a menu screen for generating the characteristic factor diagram, and a user terminal displaying a menu screen for generating the management rule.

[0017]

[0018] According to the present invention, a process can be managed by automatically generating a cause-and-effect diagram for setting a management structure that enables optimized operation of process management based on machine learning, and a management rule for calculating the management range. Through this, even in factories with limited manpower or technology, a management structure and a management range that form the basis of process management can be established, and the results can be used as basic data for multi-purpose analysis in factories with multi-product processes or multiple objectives. In addition, the results of the generated cause-and-effect diagram can be used for cross-validation of existing cause-and-effect diagrams or as a means of knowledge generation.

[0019] The generated management rules can be modified in advance through tests regarding normal operation and simulations using virtual sensor data. By utilizing the generated cause-and-effect diagrams and management rules, not only is process management more efficient, but quality enhancement can also be achieved through the introduction of an advanced smart production system. Furthermore, by continuously and dynamically monitoring the manufacturing environment in real time, it is possible to quickly identify and address the causes of abnormalities, and to secure the ability to respond immediately to problems occurring throughout the manufacturing site.

[0020]

[0021] FIG. 1 is a drawing referenced to explain an example of a system to which a process control method according to the present invention is applied.

[0022] FIG. 2 is a block diagram of the integrated management server in FIG. 1,

[0023] FIGS. 3 to 5 are flowcharts provided for describing a process control method according to an embodiment of the present invention, and

[0024] FIGS. 6 to 10 are drawings illustrating examples of screens provided through a process management system according to the present invention.

[0025]

[0026] In this specification, where it is stated that one component is "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, or that there may be other components in between. Other expressions describing the relationship between components, such as "between" or "neighboring to," and expressions such as one component "transmits" a signal to another component, should be interpreted in the same way.

[0027] The present invention will be described in more detail below with reference to the drawings.

[0028] Figure 1 shows an example of a system to which the process control method according to the present invention is applied.

[0029] Referring to FIG. 1, the system (100) may include a data collection device (130), a user terminal (150), and an integrated management server (200).

[0030] The data collection device (130) can collect various data related to the manufacturing status in real time from various sensors or devices installed at the manufacturing site. Examples of data related to the manufacturing status include environmental data such as temperature, humidity, dust, and VOC (Volatile Organic Composite), sensor data, device-related data, and process data.

[0031] Sensors installed at manufacturing sites include IoT sensors and smart sensors. IoT sensors collect data using IoT (Internet of Things) technology. Smart sensors are intelligent sensors that integrate nanotechnology or MEMS technology with general sensor technology, which detects physical and chemical information of objects being measured, enabling them to embed signal processing functions such as data processing, automatic calibration, self-diagnosis, decision-making, and communication.

[0032] A user terminal (150) is connected to an integrated management server (200) for communication, and can input user commands, data, and various information for process management at a manufacturing site into the integrated management server (200), and various menu screens or analysis screens for process management at a manufacturing site are displayed. The user terminal (150) may be connected to the integrated management server (200) through a communication network such as a mobile communication network, the internet, or a network combining these, and one or more user terminals (150) may be connected.

[0033] The integrated management server (200) can generate a management rule capable of calculating a characteristic factor diagram and a management range for setting a management structure by utilizing machine learning on data received from the data collection device (130) and information for process management at the manufacturing site.

[0034] The integrated management server (200) can check for abnormalities in the management rules by testing the management rules created using data that is measured and analyzed in real time from the data collection device (130), and can also simulate the management rules using virtual data such as virtual sensor data, and can modify the management rules by reflecting the results of such tests and simulations.

[0035] The integrated management server (200) can change management rules according to user settings and can also provide dynamic monitoring functions for the manufacturing site. The integrated management server (200) can provide data that measures and analyzes data transmitted from the data collection device (130), and related information that enables immediate response to specific situational changes, and provides a user-centric UI environment and menu screen including experts, and allows the user to view information accumulated in the integrated management server (200) through the user terminal (150).

[0036] The integrated management server (200) can be connected to a cloud system via a communication network, and through this, multiple integrated management servers (200) deployed at each of the unit factories or partner companies can be connected. In addition, with this configuration, the unit factories or partner companies can be managed integrally by the parent company.

[0037] Figure 2 is a block diagram of the integrated management server in Figure 1.

[0038] Referring to FIG. 2, the integrated management server (200) may include a data input unit (210), a data management unit (220), a screen providing unit (230), an AI structure generation unit (240), an AI rule generation unit (250), and an integrated management unit (260). When implemented in an actual application, these components may be configured such that two or more components are combined into one component as needed, or one component is subdivided into two or more components.

[0039] The data input unit (210) receives data for creating a characteristic cause diagram and generating management rules for setting the management structure. Data for creating the characteristic cause diagram and generating management rules can be received from a data collection device (130) and can also be received through a user terminal (150).

[0040] The data management unit (220) stores and manages data input through the data input unit (210) and also manages related files. The data management unit (220) may preprocess input data or previously secured data to generate virtual data, such as virtual sensor data, which can be used for testing or simulation in the AI ​​structure generation unit (240) or the AI ​​rule generation unit (250).

[0041] If a specific sensor value is missing from the input data, the data management unit (220) can train a prediction model using other sensor data it possesses and, through this, predict the missing sensor value to generate virtual sensor data.

[0042] The screen providing unit (230) provides the user terminal (150) with a menu screen for creating and managing rule-making diagrams of characteristic factors, as well as related program screens and analysis screens.

[0043] The AI ​​structure generation unit (240) generates a cause-and-effects diagram for setting up a management structure by applying clustering using machine learning to process-related data. The cause-and-effects diagram is also called a fishbone diagram and refers to a diagram that makes it easy to see at a glance how characteristics and causes, such as the results of work or problems, are related.

[0044] The characteristic factor diagram generated by the AI ​​structure generation unit (240) displays the hierarchical relationships of sensor data for a specific result in the process and the sensors that influenced the specific result in a ranking order according to importance, and through the structure of the visualized characteristic factor diagram, it is possible to know which sensors are related to each other and which sensor has the most influence on the result.

[0045] The AI ​​rule generation unit (250) generates a management rule that can calculate the optimal management range by analyzing safety and risk sections for each sensor used in the process based on machine learning, and manages the generated management rule.

[0046] Here, a rule refers to a set of rules that must be met regarding actions or conditions in a production process. These rules are necessary to maintain the efficiency and quality of the production process and define actions such as proceeding with work when specific conditions are met or stopping work if conditions are not met. For example, there are quality rules that stop production or require rework if product quality falls below standards, material rules that determine whether specific materials can be used for specific tasks, and time rules that require each stage of the process to be completed within a specific time.

[0047] The integrated management unit (260) can manage the manufacturing process according to user-set conditions, etc. by using the generated characteristic cause diagram and management rules. The integrated management unit (260) can set the operating conditions of the management rules according to conditions pre-set by the user, and can also perform dynamic monitoring of the manufacturing site by reflecting the management rules, etc.

[0048] FIGS. 3 to 5 are flowcharts provided to describe a process control method according to an embodiment of the present invention.

[0049] Referring to FIG. 3, the data input unit (210) receives process-related data including sensor data, abnormal information, and process information collected from a device or sensor installed at the manufacturing site (S300).

[0050] The AI ​​structure generation unit (240) applies clustering using machine learning to process-related data to generate a characteristic factor diagram that can identify the hierarchical relationship of sensor data regarding a specific result in a manufacturing process and the sensors that influenced the specific result in order of importance (S310). The generated characteristic factor diagram can be displayed on the screen according to the user's command.

[0051] The AI ​​rule generation unit (250) refers to the characteristic cause diagram and applies multi-stage clustering using machine learning to process-related data to analyze safety zones and danger zones for each sensor used in each process of the manufacturing site, and generates a management rule that can calculate the management range (S320).

[0052] The AI ​​rule generation unit (250) can test whether there is an abnormality in the management rule using data that is measured and analyzed in real time from the data collection device (130), and can simulate the management rule using virtual data such as virtual sensor data generated by the data management unit (220), and can also modify the management rule by reflecting the test and simulation results.

[0053] The integrated management department (260) manages the processes at the manufacturing site using the characteristic cause diagram and management rules (S330).

[0054] Through this process, cause-and-effect diagrams and control rules for process management at the manufacturing site can be generated and used for process management.

[0055] Figure 4 is a flowchart provided to explain the process of generating the characteristic cause diagram in Figure 3.

[0056] Referring to FIG. 4, the AI ​​structure generation unit (240) performs a preprocessing process on the input data (S311). Data preprocessing refers to the task of separating and processing data to suit a specific analysis.

[0057] For data preprocessing, methods such as Regression Tree Series Denoising, which removes noise based on results derived from Random Forest predictions using process data as ground truth labels; LPF and HPF utilizing simple time cycles; and Autoencoder Denoising, which trains algorithms using process data as ground truth labels and removes data showing significant differences by comparing extracted results with actual data, can be used.

[0058] Another method that can be used is the slide window time series prediction noise filter, STL (Seasonal and Trend decomposition using Loess), and similarity techniques (DWT) utilizing time series deep learning technologies such as DNN, RNN, and LSTM.

[0059] When using simple classification results, the median and mean values ​​can be placed in the missing values, and the information can be used as is without deleting the data. Additionally, linear interpolation, polynomial interpolation, and cubic interpolation can be used.

[0060] Among these methods, the optimal preprocessing method can be determined and applied through consultation with an expert, based on data characteristics, information, context, comparisons between data, and results before and after preprocessing.

[0061] Data normalization can also be performed after data preprocessing. Data normalization ensures that all data points are reflected with the same scale (importance), and this process can improve the training efficiency of machine learning. Methods that can be used for data normalization include Min-Max Normalization and Z-Score Normalization.

[0062] Next, the AI ​​structure generation unit (240) identifies sensor information by utilizing data separation and Fast Fourier Transform (FFT) to convert time domain signals into frequency domain signals, using information about the periodic patterns and frequencies of the data (S313).

[0063] Next, the AI ​​structure generation unit (240) performs hierarchical clustering to cluster data hierarchically and generates the structure of the characteristic factor diagram (S315).

[0064] After generating the structure, the AI ​​structure generation unit (240) calculates a ranking based on importance to determine which sensor has the most influence on a specific result using the identified sensor information (S317). The higher the ranking, the higher the importance. The ranking can be calculated by comprehensively using RandomForest importance, Linear Regressor coefficients, and correlation coefficients regarding the relationship between the specific result and the sensor data value.

[0065] Here, RandomForest importance is an indicator representing the influence of each feature (variable) on the prediction result when the model performs predictions. The RandomForest model is an ensemble model combining multiple decision trees, where each tree performs predictions by repeatedly splitting based on the characteristics of the data. Linear Regressor coefficients are values ​​representing the influence of each independent variable (feature) on the dependent variable (target variable) in a linear regression model. Linear regression is a model that finds a straight line that fits the data, and the slope and intercept of this line can be expressed as coefficients. Furthermore, the correlation coefficient is an indicator representing the strength and direction of the relationship between two variables. In other words, it indicates how strong the linear relationship between the two variables is, and whether that relationship is in a positive or negative direction.

[0066] The AI ​​structure generation unit (240) generates a characteristic factor diagram by performing hierarchical clustering after calculating the ranking and displaying the calculated ranking on the generated structure (S319).

[0067] Through this process, a characteristic factor diagram for process management of the AI ​​structure generation unit (240) is generated using machine learning-based clustering and data mining techniques.

[0068] Figure 5 is a flowchart provided for the explanation of the process of generating management rules in Figure 3.

[0069] Referring to FIG. 5, the AI ​​rule generation unit (250) calculates an abnormal situation by linking process data with time-series data, such as power data, which can continuously detect abnormalities at the manufacturing site (S321). Examples of abnormal situations include power abnormalities and quality abnormalities.

[0070] The AI ​​rule generation unit (250) performs multi-stage clustering on sensor data (S323). Multi-stage clustering refers to a method of clustering data by applying a clustering algorithm over several stages.

[0071] The multi-stage clustering process performs an initial clustering process as the first step. In this first step, the data is divided into approximate clusters overall. At this stage, simple clustering techniques such as the DBSCAN (Density Based Spatial Cluster of Applications with Noise) algorithm can be used.

[0072] The second step involves a segmentation process. In this second step, each cluster formed in the first step is clustered again to create finer clusters. Clustering techniques such as the K-means algorithm can be used at this stage. During this process, data within each cluster can be further subdivided, and this process can be repeated.

[0073] The final stage is the optimization and adjustment stage, where the characteristics or form of each cluster are reviewed and the clusters are modified or optimized as necessary. For example, some clusters may be too numerous or too few, so they can be adjusted.

[0074] Specifically, clustering is performed based on density or multiple centroids selected to determine their respective values, and distance-based clustering is used to differentiate each value into multiple levels. Re-clustering is then performed based on the defect rate, order, and value of the data within each cluster. Additionally, the ranges not included in the clusters are calculated, and the above method is repeated.

[0075] Then, the AI ​​rule generation unit (250) performs binarization to divide the clustered data clusters into two categories: safe and dangerous (S325). In the binarization process, the intervals can be set to levels 0 to N with high prediction accuracy using optimization techniques such as genetic algorithms, and the 0 to N levels can be divided into binarized values ​​according to the target standard notification value to distinguish the final notification form. Here, N is a natural number, and a higher value indicates an increase in the defect rate. Additionally, in the binarization process, binarization can be performed by reflecting the weight of RandomForest importance, Linear Regressor coefficients, Naive Bayes variable contribution, and correlation coefficients. Here, Naive Bayes variable contribution indicates how much each characteristic (variable) influences class prediction, and this contribution is calculated through conditional probability (the probability that each characteristic belongs to a specific class).

[0076] The AI ​​rule generation unit (250) undergoes a binarization process to generate the final management rule of the AI ​​rule generation unit (250) (S327).

[0077] Through this process, management rules can be generated, and using the generated rules, safety / risk status, sensor data value ranges, defect rates, etc., can be verified for each process section.

[0078] FIGS. 6 to 10 are drawings illustrating examples of screens provided in a process management system according to the present invention.

[0079] Figure 6 shows an example of a characteristic factor diagram for setting up a management structure and a screen for generating management rules.

[0080] As illustrated in FIG. 6, the screen (400) for creating the characteristic cause diagram and management rule displays a menu (401, 403) that allows selecting an item to work on and performing related actions.

[0081] In a menu like this (401, 403), after selecting the 'Project Management' item and clicking the 'New' button to enter and save project information, a new project window (405) is displayed.

[0082] At the bottom of the project window (405), an area (407, 409) is displayed that shows information related to the generated structure when the generation of the characteristic cause diagram is completed.

[0083] Figure 7 shows an example of a settings screen for generating a cause-and-effect diagram.

[0084] As illustrated in FIG. 7, the settings screen (420) displays a window (425) for entering settings to generate a characteristic factor diagram.

[0085] Here, in the 'Clustering Degree' field, you enter a number that determines how finely to cluster when creating the structure; the smaller the setting, the smaller the clustering becomes, and the smaller the clustering degree, the longer the time required.

[0086] In the ranking ratio settings, enter the RandomForest importance for Importance, the Linear Regressor coefficient for Sensitivity, and the correlation coefficient for Correlation, ensuring that the sum equals 100%. The ranking ratio settings are used to calculate rankings after the structure is created. For example, if correlation is considered important, the higher the correlation coefficient with a specific result, the higher the ranking.

[0087] Figure 8 shows an example of the result screen of generating a cause-and-effect diagram.

[0088] As illustrated in FIG. 8, the result screen (440) for generating the characteristic factor diagram displays a table (443) showing the ranking of sensor data and a characteristic factor diagram (445). The characteristic factor diagram (445) analyzes the sensor data factors until a specific result is derived, identifies the importance of the factors, and selects and displays the priority order.

[0089] The cause-and-effect diagram (445) is used to identify causes and effects, to identify problems in the early stages of the process, and to analyze predictions and results. Through the visualization of the cause-and-effect diagram based on production resource data (4M1E: Man, Machine, Material, Method, Environment), it is possible to identify the working environment of the process and analyze environmental factors related to the workers' work.

[0090] Figure 9 shows an example of a settings screen for creating management rules.

[0091] As illustrated in FIG. 9, in the settings screen (460) for creating management, the 'maximum cluster count setting' sets the maximum number of clusters. In the management calculation ratio setting, the importance is the RandomForest importance, the sensitivity is the Linear Regressor coefficient, the likelihood is the Naive Bayes variable, and the correlation is the correlation coefficient, and the input is entered so that the sum becomes 100%. The management rule generated may vary depending on the management calculation ratio. For example, if the sensitivity is set high, the management rule is generated by considering factors that linearly affect the result as more important.

[0092] Figure 10 shows an example of the management rule creation result screen.

[0093] As illustrated in FIG. 10, the management rule creation result screen (480) displays a screen (483) for selecting data, and a result screen (485) for items indicating safety / risk, etc. for the selected data is displayed. The user may modify the safety / risk section to change the management rule.

[0094] In addition, the process management system (100) can provide various screens for process management.

[0095] Meanwhile, the process control method and system according to the present invention are not limited to the configurations of the embodiments described above; rather, all or part of each embodiment may be selectively combined to allow for various modifications to the embodiments.

[0096] Furthermore, the present invention can be implemented as a computer program on a programmable computer. Such a computer may include a processor, a storage device, an input device, and an output device. To implement the contents described in the present invention, program code may be input via a mouse or keyboard input device. Such programs may be implemented in a high-level language or an object-oriented language. They may also be implemented as a computer system implemented in assembly or machine code.

[0097] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. Various modifications are possible by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention.

[0098]

[0099] 130 : Input section

[0100] 150 : User terminal

[0101] 200 : Integrated Management Server

[0102] 210 : Data Input Section

[0103] 220 : Data Management Department

[0104] 230 : Screen providing unit

[0105] 240 : AI Structure Generation Unit

[0106] 250 : AI Rule Generation Unit

[0107] 260 : Integrated Management Department

Claims

1. A step of receiving process-related data, including sensor data, abnormal information, and process information collected from devices or sensors installed at the manufacturing site, from a data input unit; A step of generating a factor-effect diagram in which the hierarchical relationships of the sensor data regarding a specific result in the process of the manufacturing site and the sensors that influenced the specific result can be identified in ranking order by applying clustering using machine learning to the process-related data in the AI ​​structure generation unit; A step of generating a management rule capable of calculating a management range by applying clustering using machine learning to the process-related data in the AI ​​rule generation unit to analyze safety zones and risk zones for each sensor used in each process of the manufacturing site; and A process management method comprising the step of managing the process of the manufacturing site using the above-mentioned characteristic cause diagram and the above-mentioned management rule in an integrated management department.

2. In Paragraph 1, A process control method characterized by calculating the above ranking using RandomForest importance, Linear Regressor coefficient, and correlation coefficient regarding the relationship between the above specific result and the above sensor data value.

3. In Paragraph 1, A process control method characterized by using RandomForest importance, Linear Regressor coefficients, Naive Bayes variable contribution, and correlation coefficients when analyzing the safety zone and risk zone during the process of generating the above-mentioned management rule.

4. In Paragraph 1, A process management method further comprising the step of providing a menu screen for setting a workspace, setting a work environment, setting work data, and setting parameters and analysis elements to be applied to machine learning for generating the above-mentioned characteristic cause diagram.

5. In Paragraph 1, A process management method further comprising the step of providing a menu screen for setting a workspace, setting a work environment, setting work data, and setting parameters and analysis elements to be applied to machine learning for generating the above management rule.

6. In Paragraph 1, A process management method further comprising the step of providing a menu screen for modifying the above management rule or creating a new rule according to a user command.

7. In Paragraph 1, A process management method further comprising the step of simulating the management rule using virtual data on the integrated server and modifying the management rule by reflecting the simulation results.

8. A data input unit that receives process-related data including sensor data, abnormal information, and process information collected from equipment or sensors installed at a manufacturing site; An AI structure generation unit that generates a factor-effect diagram by applying machine learning-based clustering to the above process-related data, which can identify the hierarchical relationships of the sensor data regarding a specific result in the process of the above manufacturing site and the sensors that influenced the said specific result in ranking order; An AI rule generation unit that generates a management rule capable of calculating a management range by applying clustering using machine learning to the process-related data to analyze safety zones and risk zones for each sensor used in each process of the manufacturing site; and A process management system comprising an integrated management unit that manages the process of the manufacturing site using the above-mentioned characteristic cause diagram and the above-mentioned management rule.

9. In Paragraph 8, A process management system further comprising a data collector for collecting the above-mentioned process-related data.

10. In Paragraph 8, A process management system further comprising a user terminal displaying a menu screen for generating the above-mentioned characteristic cause diagram and a menu screen for generating the above-mentioned management rule.

Images