Industrial production whole-process intelligent MES system supporting multi-protocol docking and AI collaboration

Abstract

The present application relates to the technical field of intelligent MES, and discloses an industrial production whole-process intelligent MES system supporting multi-protocol docking and AI cooperation, which comprises a protocol adaptation module, a model establishment module, a space construction module, a state selection module and a space updating module.The protocol adaptation module is used for constructing an industrial unified protocol adaptation engine, establishing data connection with production equipment according to the industrial unified protocol adaptation engine, and collecting historical industrial data corresponding to different production equipment; the historical industrial data of the production equipment is trained by using a deep learning algorithm, the running characteristics of each device and the process correlation characteristics between devices can be extracted at the same time, a unique device identifier is bound to each device model, and the device models are sorted in strict accordance with the production and processing sequence and the process connection logic, so that the device model sequence is completely consistent with the actual production process, and linkage control of the industrial production whole-process device model is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent MES technology, and more specifically, to an intelligent MES system that supports multi-protocol integration and AI collaboration for the entire industrial production process. Background Technology

[0002] Existing technologies related to industrial protocol adaptation, artificial intelligence collaborative control, and production execution systems in the field of intelligent MES aim to realize the digital management and control of industrial production equipment and the automated optimization of production processes. Their research and development purpose is to solve the problems of decentralized equipment management, excessive manual intervention, and low production efficiency in traditional industrial production.

[0003] In terms of industrial protocol adaptation, existing technologies are mostly single-protocol gateways or partial industrial protocol integration solutions, which cannot be compatible with mainstream general industrial protocols such as Modbus, OPCUA, Profinet, and EtherNet / IP at the same time. They have poor compatibility with access to old and non-standard production equipment, which can easily create data silos in the industrial production site and prevent the effective connection of equipment data throughout the entire production process.

[0004] Secondly, in terms of AI modeling and production control, existing AI technologies can only extract the single operating characteristics of the equipment itself when building production equipment models, without mining the process correlation characteristics between equipment. Furthermore, the equipment models lack the sorting logic that fits the actual production process, and the parameter control of production parameters lacks the parameter correlation constraints between the preceding and following processes. At the same time, the construction method of the qualified state space of the equipment is fixed and cannot adapt to the dynamic changes in the operation of the equipment during the production process.

[0005] Therefore, an intelligent MES system supporting multi-protocol interoperability and AI collaboration for the entire industrial production process is proposed. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent MES system that supports multi-protocol interoperability and AI collaboration for the entire industrial production process, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, an intelligent MES system supporting multi-protocol integration and AI collaboration for the entire industrial production process is provided, including a protocol adaptation module, a model building module, a space construction module, a state selection module, and a space update module.

[0008] The protocol adaptation module is used to build an industrial unified protocol adaptation engine, and establish a data connection with the production equipment based on the industrial unified protocol adaptation engine, thereby collecting historical industrial data corresponding to different production equipment.

[0009] The model building module is used to build a production process equipment model by combining AI with historical industrial data of production equipment, and sort the equipment models according to the production process.

[0010] The space construction module is used to filter qualified industrial data from historical industrial data, and at the same time, construct qualified state space for each equipment model based on the filtered qualified industrial data, and use the industrial data corresponding to the previous equipment model as the spatial center of the qualified state space.

[0011] The state selection module is used to obtain the latest industrial data of the first equipment model, and at the same time calculate the correlation probability between each piece of industrial data. Then, based on the latest industrial data and the correlation probability, the state is selected sequentially in the qualified state space of the subsequent equipment models to form a list of candidate equipment states. The production equipment is controlled based on the list of candidate equipment states.

[0012] The spatial update module is used to match the latest industrial data transmitted by each device model with the device candidate status list. When the latest industrial data does not match the device candidate status, the device candidate status list is updated with the latest industrial data. In the event that there is no qualified industrial status spatial update, the spatial extension is performed by combining qualified industrial data with AI.

[0013] As a further improvement to this technical solution, the protocol adaptation module integrates four common industrial protocols: Modbus, OPCUA, Profinet, and EtherNet / IP, to build an industrial unified protocol adaptation engine. Then, it automatically identifies the communication port and protocol type of the production equipment, and performs protocol conversion through the industrial unified protocol adaptation engine to convert the equipment's native protocol into a unified communication protocol, thereby establishing a bidirectional data transmission channel.

[0014] Historical industrial data from different production equipment is collected through a two-way data transmission channel.

[0015] As a further improvement to this technical solution, in the model building module, when building a production process equipment model by combining AI with historical industrial data of production equipment, deep learning is used to train the historical industrial data, extract the operating characteristics and process correlation characteristics of each piece of equipment, and each equipment model corresponds to a unique equipment identifier.

[0016] Following the order of production and processing, and based on the logical connection of production processes, the ordered equipment model sequence is completely consistent with the actual production process.

[0017] As a further improvement to this technical solution, the space construction module is equipped with a qualification standard, which corresponds to the range of equipment operating parameters required by the production process.

[0018] Historical industrial data was compared with the qualification criteria;

[0019] If historical industrial data exceeds the acceptance criteria, it is judged as abnormal data, and abnormal data that exceeds the parameter range, redundant data that is recorded repeatedly, and invalid data that cannot reflect the equipment status are removed.

[0020] Conversely, if the historical industrial data does not exceed the qualification criteria, it is judged as qualified data, and the qualified data is summarized as the qualified industrial data corresponding to the equipment model.

[0021] As a further improvement to this technical solution, in the space construction module, when constructing a qualified state space for each equipment model based on the selected qualified industrial data, the qualified state space takes the equipment operating parameters as the dimension, covers all parameter combinations for normal equipment operation, and takes the industrial data corresponding to the previous equipment model as the spatial center of the current qualified state space, so that the qualified state spaces of adjacent equipment models are linked.

[0022] As a further improvement to this technical solution, the state selection module uses a bidirectional data transmission channel established by the industrial unified protocol adaptation engine to collect the latest industrial data of the first equipment model in real time.

[0023] When calculating the correlation probability between various industrial data, an AI correlation analysis algorithm is used. Taking the latest industrial data of the first equipment model as the analysis object, the correlation coefficient between any two industrial parameters is calculated one by one. The correlation coefficient is converted into the correlation probability. The correlation probability ranges from 0 to 1. The closer the correlation probability is to 1, the greater the mutual influence between the corresponding two industrial data. The closer it is to 0, the smaller the mutual influence. This clarifies the correlation relationship between various data.

[0024] Using the latest industrial data of the first equipment model as a benchmark, and combining the correlation probabilities of various data, the operating state with the highest matching degree with the benchmark data in the qualified state space of the subsequent first equipment model is determined. Then, using this selected state as the new benchmark, the qualified state spaces of the remaining equipment models are matched and selected in turn. The matching and selection results are summarized to obtain a list of candidate equipment states.

[0025] As a further improvement to this technical solution, the state selection module converts the state commands in the candidate state list of equipment into control signals for equipment adaptation and recognition through the industrial unified protocol adaptation engine, and sends them to the corresponding production equipment one by one to realize the coordinated control of the production equipment.

[0026] As a further improvement to this technical solution, in the spatial update module, the unique identifier of each equipment model is extracted, and then the latest industrial data of each equipment model is compared with the qualified status of the corresponding identifier in the equipment candidate status list according to the identifier.

[0027] When the latest industrial data does not match the candidate status of the equipment, update the list of candidate status of the equipment using the latest industrial data.

[0028] Specifically, by sending the latest industrial data to the space construction module, using the latest industrial data of the production equipment as the space center, and then combining it with the equipment model required for subsequent production, the qualified state space is re-established, thereby completing the update of the original qualified state space.

[0029] When the latest industrial data matches the candidate status of the equipment, monitoring continues.

[0030] As a further improvement to this technical solution, in the space update module, when there is no qualified data to support the re-establishment of the qualified state space, the AI ​​algorithm is used to perform trend mining on the historically screened qualified industrial data, extract the changing patterns and fluctuation ranges of equipment operating parameters, and then, based on these changing patterns, the boundary of the original qualified state space is extended in a gradient.

[0031] Meanwhile, after using the extended qualified status for equipment control, the latest industrial data of subsequent production equipment is matched with abnormal data in historical industrial data. When the latest industrial data matches the abnormal data, the production of the product is determined to be a failure, and the gradient extension is stopped.

[0032] Conversely, gradient extension continues.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] 1. This intelligent MES system for the entire industrial production process, which supports multi-protocol integration and AI collaboration, integrates four common industrial protocols—Modbus, OPCUA, Profinet, and EtherNet / IP—through a protocol adaptation module to build a unified industrial protocol adaptation engine. This enables automatic identification of the communication ports and protocol types of production equipment, while simultaneously converting the equipment's native protocol to a unified communication protocol and establishing a bidirectional data transmission channel. This effectively breaks down the data silos of equipment in traditional industrial production sites, enabling unified data collection and standardized issuance of control commands for different brands and types of production equipment. It significantly improves the compatibility of equipment access in industrial sites and the real-time performance and stability of data transmission, solving the core problems of cumbersome access and poor adaptability of traditional multi-protocol adaptation technologies.

[0035] 2. In this intelligent MES system that supports multi-protocol integration and AI collaboration, deep learning algorithms are used to train historical industrial data of production equipment. It can simultaneously extract the operating characteristics of each piece of equipment and the process correlation characteristics between equipment. Each equipment model is bound to a unique equipment identifier, and the equipment models are sorted in strict accordance with the production and processing sequence and process connection logic. This ensures that the equipment model sequence is completely consistent with the actual production process, realizing the linkage control of equipment models throughout the entire industrial production process. This breaks the limitations of traditional AI modeling, which only targets a single equipment and does not extract process correlation characteristics. It lays a precise and realistic digital model foundation for the intelligent control of the entire industrial production process.

[0036] 3. In this intelligent MES system that supports multi-protocol integration and AI collaboration across the entire industrial production process, the space construction module sets qualification criteria corresponding to production process requirements. It accurately filters historical industrial data and removes abnormal, redundant, and invalid data. At the same time, it constructs a qualification state space for each equipment model based on equipment operating parameters, and uses the industrial data corresponding to the previous equipment model as the spatial center of the current qualification state space. This creates a strong correlation between the qualification state spaces of adjacent equipment models, ensuring the connection and continuity of equipment operating parameters between industrial production processes. It solves the problem of the traditional single qualification state space construction and the disconnect between parameters of preceding and subsequent processes, and provides a compliant range basis that fits the actual production for the selection of subsequent equipment states. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the structure of the intelligent MES system for the entire industrial production process that supports multi-protocol interoperability and AI collaboration according to the present invention;

[0038] Figure 2 This is a flowchart illustrating the protocol adaptation module of the present invention.

[0039] Figure 3 A flowchart illustrating the model building module of this invention;

[0040] Figure 4 This is a flowchart illustrating the spatial construction module of the present invention;

[0041] Figure 5 This is a flowchart illustrating the state selection module of the present invention;

[0042] Figure 6 This is a flowchart illustrating the spatial update module of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figures 1-6 As shown, the purpose of this embodiment is to provide an intelligent MES system for the entire industrial production process that supports multi-protocol integration and AI collaboration, including a protocol adaptation module, a model building module, a space construction module, a state selection module, and a space update module.

[0045] The protocol adaptation module is used to build an industrial unified protocol adaptation engine, and establish a data connection with the production equipment based on the industrial unified protocol adaptation engine, thereby collecting historical industrial data corresponding to different production equipment.

[0046] In the protocol adaptation module, four common industrial protocols, Modbus, OPCUA, Profinet, and EtherNet / IP, are integrated to build an industrial unified protocol adaptation engine. Then, the communication ports and protocol types of production equipment are automatically identified, and the protocol is converted through the industrial unified protocol adaptation engine to convert the native protocol of the equipment into a unified communication protocol and establish a two-way data transmission channel.

[0047] It integrates four common industrial protocols: Modbus, OPCUA, Profinet, and EtherNet / IP. It standardizes and encapsulates the message parsing rules, data frame structures, and communication interaction logic of each protocol. Through a bus scheduling mechanism, it integrates all protocol parsing components to form a unified scheduleable protocol processing kernel, thus completing the construction of an industrial unified protocol adaptation engine.

[0048] Send probe data frames to the industrial field communication bus, sequentially scan the communication ports of all online production equipment, receive the response feature information returned by the equipment, compare the response information with the engine's built-in protocol feature library, determine the communication port number and native protocol type of the equipment, and complete the automatic identification of equipment communication information.

[0049] Based on the identified native protocol type of the device, the corresponding protocol parsing component in the adaptation engine is called to disassemble the native communication message of the device, extract the device operation-related parameter data, and reorganize the native data of different formats according to the unified data encoding rules and data structure to generate the system's common unified communication protocol data, thus completing the standardization conversion of the protocol format.

[0050] Based on a unified communication protocol, downlinks are established for the engine to send control commands to the production equipment, and uplinks are established for the production equipment to upload running data to the engine. Communication handshake verification and transmission stability verification are performed on the uplinks and downlinks. After the link verification is passed, a bidirectional data transmission channel that can simultaneously send commands and upload data is formed.

[0051] Historical industrial data from different production equipment is collected through a two-way data transmission channel.

[0052] The model building module is used to build equipment models for the entire production process by combining AI with historical industrial data of production equipment, and to sort the equipment models according to the production process.

[0053] In the model building module, when building equipment models for the entire production process using AI combined with historical industrial data of production equipment, deep learning is used to train the historical industrial data, extracting the operational characteristics and process-related characteristics of each piece of equipment. Each equipment model corresponds to a unique equipment identifier. The steps are as follows:

[0054] The historical industrial data of each production equipment collected by the protocol adaptation module is called, and abnormal, redundant and invalid data are removed (the qualified industrial data screened by the spatial construction module is used). The data is normalized, the data units are unified, and the training set (80%), validation set (10%) and test set (10%) are divided to ensure data quality and lay the foundation for model training.

[0055] A hybrid deep learning model of CNN-LSTM was selected (taking into account the temporal correlation of industrial time series data and the spatial coupling of equipment parameters). An input layer, a feature extraction layer, a fully connected layer, and an output layer were constructed. The input layer receives preprocessed historical equipment data, and the output layer corresponds to the predicted results of equipment operating status.

[0056] The preprocessed training set data is input into the CNN-LSTM model. Hyperparameters such as the number of training iterations, learning rate, and batch size are set. With the goal of minimizing the model prediction error, the model parameters are updated through the backpropagation algorithm. The training is continuously iterated until the model converges (the validation set error tends to stabilize).

[0057] Through the model feature extraction layer, two types of core features are extracted: equipment operation features, which extract the inherent coupling rules of parameters such as equipment load, temperature, speed, and current to characterize the operating characteristics of a single piece of equipment; and process correlation features, which explore the parameter correlation rules between the current equipment and the equipment in the preceding and following processes to characterize the process connection characteristics between equipment.

[0058] For each trained single-device model, a unique device identifier (corresponding to the physical identifier of the production equipment) is bound, and the corresponding equipment type and process step are labeled to ensure that the model accurately corresponds to the physical equipment and production process. At the same time, the test set data is input into the calibrated single-device model to verify the model's feature extraction accuracy and operating status prediction accuracy. If the accuracy does not reach the preset threshold (optimal threshold ≥ 95%), the model hyperparameters are adjusted and retrained and optimized until the accuracy requirements of industrial production are met.

[0059] Following the order of production and processing, and based on the logical connection of production processes, the ordered equipment model sequence is completely consistent with the actual production process.

[0060] The process involves analyzing the sequence of steps in the actual production process, identifying the production equipment corresponding to each step, and marking the connection relationships between each step (e.g., step B can only proceed after step A is completed, and the corresponding equipment model B can only proceed after equipment model A is completed). This forms a process connection logic diagram. Then, a connection weight is assigned to each pair of adjacent steps (and their corresponding equipment). The weight value is set based on the tightness of the process connection and the degree of process influence (the connection weight of core steps is greater than that of auxiliary steps, and the weight value ranges from 1 to 10). The higher the weight, the greater the impact of the connection on the production process.

[0061] Based on the process connection logic diagram, a topological sorting algorithm is used, with the production and processing sequence as the core constraint and the process connection weights combined, to sort all single equipment models, ensuring that the sorted model sequence is completely consistent with the actual production process sequence and connection logic.

[0062] The space construction module is used to filter qualified industrial data from historical industrial data. At the same time, it constructs qualified state space for each equipment model based on the filtered qualified industrial data, and uses the industrial data corresponding to the previous equipment model as the spatial center of the qualified state space.

[0063] In the space construction module, a pass / fail criterion is set, which corresponds to the range of equipment operating parameters required by the production process.

[0064] Based on the production process documents, clarify the compliance range of each piece of equipment and each operating parameter (such as temperature, speed, current, pressure, etc.) to form a standardized qualification judgment standard. The standard must correspond one-to-one with the equipment model and production process, and be bound to the corresponding equipment identification to ensure the pertinence of the judgment standard (avoid the use of the same standard for different equipment).

[0065] Historical industrial data was compared with the qualification criteria;

[0066] The system calls the protocol adaptation module to collect all historical industrial data, classifies it by equipment identifier, compares each set of historical data for a single equipment with the corresponding qualification criteria for that equipment parameter by parameter, and records the comparison result (compliant / non-compliant) for each set of data.

[0067] If historical industrial data exceeds the acceptance criteria, it is judged as abnormal data, and abnormal data that exceeds the parameter range, redundant data that is recorded repeatedly, and invalid data that cannot reflect the equipment status are removed.

[0068] Conversely, if the historical industrial data does not exceed the qualification criteria, it is judged as qualified data, and the qualified data is summarized as the qualified industrial data corresponding to the equipment model.

[0069] In the spatial construction module, when constructing qualified state spaces for each equipment model based on the selected qualified industrial data, the qualified state space is based on the equipment operating parameters, covering all parameter combinations for normal equipment operation. Simultaneously, the industrial data corresponding to the previous equipment model is used as the spatial center of the current qualified state space, thus linking the qualified state spaces of adjacent equipment models. The steps are as follows:

[0070] Using equipment operating parameters as dimensions, each key operating parameter (such as temperature, speed, current, and pressure) is treated as an independent dimension to construct a multi-dimensional qualified state space. The number of dimensions is consistent with the number of key operating parameters of the equipment (e.g., if a piece of equipment has 3 key parameters, then a 3-dimensional qualified state space is constructed). This ensures that the space can fully cover the parameter dimensions for normal equipment operation. Then, statistical analysis is used to statistically analyze the qualified industrial data for each dimension, calculate the maximum, minimum, and mean values ​​of the parameters in that dimension, and determine the compliance boundary for each dimension. The combination of the boundaries of all dimensions forms the complete boundary of the qualified state space, ensuring that the space covers all parameter combinations for normal equipment operation.

[0071] Strictly follow the equipment model sorting in the production process, and use the average of qualified industrial data corresponding to the previous equipment model as the spatial center of the qualified state space of the current equipment model, so that the state space of the current equipment is strongly correlated with the operating state of the previous equipment, ensuring the continuity of the equipment state of the preceding and following processes.

[0072] Calculate the correlation between the current equipment qualified status space and the previous equipment qualified status space, ensure that the correlation reaches the preset threshold (0.7), verify the rationality of the connection between adjacent spaces, and avoid spatial disconnection (i.e., the normal operating parameters of the previous equipment cannot fall into the qualified status space of the subsequent equipment).

[0073] After successful verification, the qualified state space (including dimensions, boundaries, and spatial center) of each device model is fixed and bound to the corresponding device identifier, forming a full-process qualified state space system that covers all device models and has strong correlation between adjacent spaces, as shown in the following formula:

[0074]

[0075] in, The correlation between the current device and the qualified state space of the previous device (the value ranges from 0 to 1, and the closer it is to 1, the higher the correlation), and k is the number of dimensions of the qualified state space (i.e. the number of key operating parameters of the device). Let be the mean of the spatial centers in the i-th dimension of the current device. The mean of the spatial centers in the i-th dimension of the previous device. It represents the standard deviation of the qualified data in the i-th dimension of the previous device.

[0076] The state selection module is used to obtain the latest industrial data of the first equipment model, and at the same time to calculate the correlation probability between each piece of industrial data. Then, based on the latest industrial data and the correlation probability, the state is selected in the qualified state space of the subsequent equipment models in turn to form a list of candidate equipment states. The production equipment is controlled based on the list of candidate equipment states.

[0077] In the status selection module, the latest industrial data of the first equipment model is collected in real time through a two-way data transmission channel established by the industrial unified protocol adaptation engine.

[0078] When calculating the correlation probability between various industrial data, an AI correlation analysis algorithm is used. Taking the latest industrial data of the first equipment model as the analysis object, the correlation coefficient between any two industrial parameters is calculated one by one. The correlation coefficient is converted into the correlation probability. The correlation probability ranges from 0 to 1. The closer the correlation probability is to 1, the greater the mutual influence between the corresponding two industrial data. The closer it is to 0, the smaller the mutual influence. This clarifies the correlation relationship between various data.

[0079] All operational parameter dimensions are extracted from the latest industrial data of the first equipment model. The Pearson correlation coefficient algorithm is used as the AI ​​correlation analysis method to calculate the linear correlation between any two industrial parameters. The calculated correlation coefficients are then normalized and transformed into correlation probability values ​​within the range of 0 to 1. The influence of parameters is distinguished based on the correlation probability values, thus completing the quantitative calibration of the correlation relationships between all industrial data. The formula is as follows:

[0080]

[0081] in, Let be the correlation coefficient between parameter a and parameter b. , Let a be the i-th sampled value of parameters a and b. , Let a and b be the sampled mean values. The number of sampled data.

[0082]

[0083] in, Let be the association probability between parameters a and b. The range is 0-1, with larger values ​​indicating stronger inter-parameter influence.

[0084] Using the latest industrial data of the first equipment model as a benchmark, and combining the correlation probabilities of various data, the operating state with the highest matching degree with the benchmark data in the qualified state space of the subsequent first equipment model is determined. Then, using this selected state as the new benchmark, the qualified state spaces of the remaining equipment models are matched and selected in turn. The matching and selection results are summarized to obtain a list of candidate equipment states.

[0085] In the status selection module, the status commands in the candidate status list of equipment are converted into control signals for equipment adaptation and recognition through the industrial unified protocol adaptation engine, and then sent to the corresponding production equipment one by one to realize the collaborative control of the production equipment.

[0086] Read the target operating status parameters of each device in the candidate status list, and use the industrial unified protocol adaptation engine to reverse convert the system standard status commands into native protocol control signals that can be recognized by the corresponding production equipment. According to the unique identifier of the equipment, the control signals are sent to the corresponding production equipment one by one in a specific direction. Each production equipment executes actions synchronously according to the received control signals, realizing the collaborative closed-loop control of the entire production equipment.

[0087] The spatial update module is used to match the latest industrial data transmitted by each equipment model with the equipment candidate status list. When the latest industrial data does not match the equipment candidate status, the equipment candidate status list is updated with the latest industrial data. In the event that there is no qualified industrial status spatial update, the spatial extension is performed by combining qualified industrial data with AI.

[0088] In the spatial update module, the unique identifier of each equipment model is extracted, and then the latest industrial data of each equipment model is compared with the qualified status of the corresponding identifier in the equipment candidate status list based on the identifier.

[0089] When the latest industrial data does not match the candidate status of the equipment, update the list of candidate status of the equipment using the latest industrial data.

[0090] Specifically, by sending the latest industrial data to the space construction module, using the latest industrial data of the production equipment as the space center, and then combining it with the equipment model required for subsequent production, the qualified state space is re-established, thereby completing the update of the original qualified state space.

[0091] When the latest industrial data matches the candidate status of the equipment, monitoring continues.

[0092] In the space update module, when there is no qualified data to support the re-establishment of the qualified state space, the AI ​​algorithm is used to perform trend mining on the historically screened qualified industrial data, extract the changing patterns and fluctuation ranges of equipment operating parameters, and then, based on these changing patterns, the boundary of the original qualified state space is extended in a gradient.

[0093] The historical qualified industrial dataset that has been filtered in the spatial construction module is retrieved. The time series trend fitting algorithm is used to mine the change patterns of the historical qualified data, extract the long-term change trends of each operating parameter of the equipment with the production process, determine the upper and lower limits of normal parameter fluctuations, clarify the linkage relationship between parameters, and provide a stable data basis for gradient extension.

[0094] Based on the variation patterns obtained from historical qualified data mining, the parameter boundaries of the original qualified state space are gradually extended according to a fixed and uniform gradient step size. Each extension only adjusts a single gradient amplitude to ensure that the parameter changes are stable and do not exceed the equipment safety operation limits. During the extension process, the spatial correlation between the previous equipment model and the current equipment model is maintained, and the spatial center benchmark is not changed. After the gradient extension is completed, a new temporary qualified state space is generated.

[0095] Meanwhile, after using the extended qualified status for equipment control, the latest industrial data of subsequent production equipment is matched with abnormal data in historical industrial data. When the latest industrial data matches the abnormal data, the production of the product is determined to be a failure, and the gradient extension is stopped.

[0096] Conversely, gradient extension continues.

[0097] The equipment model sequence formed by the above sorting is the fixed process execution chain. Subsequent state selection, equipment control, and space updates are all performed within this fixed process execution chain. The subsequent equipment model refers to the equipment model of the subsequent process in the fixed process execution chain, rather than the equipment object reselected based on the association probability.

[0098] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent MES system supporting multi-protocol integration and AI collaboration across the entire industrial production process, characterized by: It includes a protocol adaptation module, a model building module, a space construction module, a state selection module, and a space update module; The protocol adaptation module is used to build an industrial unified protocol adaptation engine, and establish a data connection with the production equipment based on the industrial unified protocol adaptation engine, thereby collecting historical industrial data corresponding to different production equipment. The model building module is used to build a production process equipment model by combining AI with historical industrial data of production equipment, and sort the equipment models according to the production process. The space construction module is used to filter qualified industrial data from historical industrial data, and at the same time, construct qualified state space for each equipment model based on the filtered qualified industrial data, and use the industrial data corresponding to the previous equipment model as the spatial center of the qualified state space. The state selection module is used to obtain the latest industrial data of the first equipment model, and at the same time calculate the correlation probability between each piece of industrial data. Then, based on the latest industrial data and the correlation probability, the state is selected sequentially in the qualified state space of the subsequent equipment models to form a list of candidate equipment states. The production equipment is controlled based on the list of candidate equipment states. The spatial update module is used to match the latest industrial data transmitted by each device model with the device candidate status list. When the latest industrial data does not match the device candidate status, the device candidate status list is updated with the latest industrial data. In the event that there is no qualified industrial status spatial update, the spatial extension is performed by combining qualified industrial data with AI.

2. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: The protocol adaptation module integrates four common industrial protocols: Modbus, OPCUA, Profinet, and EtherNet / IP, to build an industrial unified protocol adaptation engine. Then, it automatically identifies the communication ports and protocol types of the production equipment and performs protocol conversion through the industrial unified protocol adaptation engine to convert the equipment's native protocol into a unified communication protocol and establish a two-way data transmission channel. Historical industrial data from different production equipment is collected through a two-way data transmission channel.

3. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the model building module, when building a production process equipment model by combining AI with historical industrial data of production equipment, deep learning is used to train the historical industrial data, extract the operating characteristics and process correlation characteristics of each piece of equipment, and each equipment model corresponds to a unique equipment identifier. Following the order of production and processing, and based on the logical connection of production processes, the ordered equipment model sequence is completely consistent with the actual production process.

4. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the space construction module, a qualification judgment standard is set, which corresponds to the range of equipment operating parameters required by the production process. Historical industrial data was compared with the qualification criteria; If historical industrial data exceeds the acceptance criteria, it is judged as abnormal data, and abnormal data that exceeds the parameter range, redundant data that is recorded repeatedly, and invalid data that cannot reflect the equipment status are removed. Conversely, if the historical industrial data does not exceed the qualification criteria, it is judged as qualified data, and the qualified data is summarized as the qualified industrial data corresponding to the equipment model.

5. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the space construction module, when constructing qualified state spaces for each equipment model based on the selected qualified industrial data, the qualified state space takes the equipment operating parameters as the dimension, covers all parameter combinations for normal equipment operation, and takes the industrial data corresponding to the previous equipment model as the spatial center of the current qualified state space, so that the qualified state spaces of adjacent equipment models are linked.

6. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the state selection module, the latest industrial data of the first equipment model is collected in real time through a two-way data transmission channel established by the industrial unified protocol adaptation engine. When calculating the correlation probability between various industrial data, an AI correlation analysis algorithm is used. Taking the latest industrial data of the first equipment model as the analysis object, the correlation coefficient between any two industrial parameters is calculated one by one. The correlation coefficient is converted into the correlation probability. The correlation probability ranges from 0 to 1. The closer the correlation probability is to 1, the greater the mutual influence between the corresponding two industrial data. The closer it is to 0, the smaller the mutual influence. This clarifies the correlation relationship between various data. Using the latest industrial data of the first equipment model as a benchmark, and combining the correlation probabilities of various data, the operating state with the highest matching degree with the benchmark data in the qualified state space of the subsequent first equipment model is determined. Then, using this selected state as the new benchmark, the qualified state spaces of the remaining equipment models are matched and selected in turn. The matching and selection results are summarized to obtain a list of candidate equipment states.

7. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the state selection module, the state commands in the candidate state list of equipment are converted into control signals for equipment adaptation and recognition through the industrial unified protocol adaptation engine, and then sent to the corresponding production equipment one by one to realize the coordinated control of the production equipment.

8. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 1, characterized in that: In the spatial update module, the unique identifier of each equipment model is extracted, and then the latest industrial data of each equipment model is compared with the qualified status of the corresponding identifier in the equipment candidate status list based on the identifier. When the latest industrial data does not match the candidate status of the equipment, update the list of candidate status of the equipment using the latest industrial data. Specifically, by sending the latest industrial data to the space construction module, using the latest industrial data of the production equipment as the space center, and then combining it with the equipment model required for subsequent production, the qualified state space is re-established, thereby completing the update of the original qualified state space. When the latest industrial data matches the candidate status of the equipment, monitoring continues.

9. The intelligent MES system for the entire industrial production process supporting multi-protocol integration and AI collaboration as described in claim 8, characterized in that: In the space update module, when there is no qualified data to support the re-establishment of the qualified state space, the AI ​​algorithm is used to perform trend mining on the historically screened qualified industrial data, extract the change pattern and fluctuation range of equipment operating parameters, and then based on the change pattern, the boundary of the original qualified state space is extended by gradient. Meanwhile, after using the extended qualified status for equipment control, the latest industrial data of subsequent production equipment is matched with abnormal data in historical industrial data. When the latest industrial data matches the abnormal data, the product production is determined to have failed, and the gradient extension is stopped. Conversely, gradient extension continues.

Images