Intelligent manufacturing control method, device and system

Abstract

The invention relates to an intelligent manufacturing control method, device and system. The method comprises the following steps: acquiring state monitoring data corresponding to industrial equipment; determining whether the industrial equipment is abnormal or not based on the state monitoring data; under the condition that the industrial equipment is abnormal, carrying out emergency stop control on the industrial equipment, and sending first priority data to a cloud platform, so that the cloud platform carries out emergency scheduling processing based on the first priority data; and under the condition that the industrial equipment is not abnormal, second priority data is sent to the cloud platform, so that the cloud platform performs long-term storage and deep analysis based on the second priority data, and the priority of the second priority data is lower than that of the first priority data. Therefore, the real-time performance of data processing and the depth performance of data analysis can be well considered, and the problem that the real-time performance of data processing and the depth performance of data analysis cannot be well considered in a traditional intelligent manufacturing control mode is effectively solved.

Description

Technical Field

[0001] This application relates to the field of intelligent manufacturing technology, and in particular to an intelligent manufacturing control method, device and system. Background Technology

[0002] Currently, traditional intelligent manufacturing control methods are typically implemented solely by edge terminals or solely by cloud platforms. While edge terminals can reduce latency issues caused by data transmission, they cannot perform in-depth analysis of multi-dimensional data. Similarly, cloud platforms can perform in-depth analysis of multi-dimensional data, but they cannot reduce latency issues caused by data transmission. Therefore, traditional intelligent manufacturing control methods cannot effectively balance the real-time performance of data processing with the depth of data analysis. Summary of the Invention

[0003] This application provides an intelligent manufacturing control method, device, and system to solve the problem that traditional intelligent manufacturing control methods cannot adequately balance the real-time performance of data processing and the depth of data analysis.

[0004] In a first aspect, embodiments of this application provide an intelligent manufacturing control method applied to an edge terminal, wherein the edge terminal is communicatively connected to industrial equipment and a cloud platform, and the method includes:

[0005] Obtain the status monitoring data corresponding to the industrial equipment;

[0006] Based on the status monitoring data, determine whether the industrial equipment is malfunctioning;

[0007] In the event of an abnormality in the industrial equipment, an emergency stop control is implemented on the industrial equipment, and first priority data is sent to the cloud platform so that the cloud platform can perform emergency dispatch processing based on the first priority data.

[0008] If there are no abnormalities in the industrial equipment, second priority data is sent to the cloud platform for long-term storage and in-depth analysis. The priority of the second priority data is lower than that of the first priority data.

[0009] Optionally, determining whether the industrial equipment is abnormal based on the status monitoring data includes:

[0010] The status monitoring data is preprocessed;

[0011] Perform time-domain and / or frequency-domain analysis on the preprocessed state monitoring data to obtain time-domain and / or frequency-domain characteristics;

[0012] An anomaly probability is obtained by inferring from the time-domain features and / or the frequency-domain features using a lightweight artificial intelligence model;

[0013] If the probability of an anomaly is greater than a preset threshold, it is determined that the industrial equipment is malfunctioning.

[0014] If the probability of an anomaly is less than or equal to the preset threshold, it is determined that the industrial equipment is not abnormal.

[0015] Optionally, the step of performing emergency stop control on the industrial equipment and sending first priority data to the cloud platform includes:

[0016] An emergency stop signal is generated and sent to the controller corresponding to the industrial equipment to perform emergency stop control on the industrial equipment; and,

[0017] The identification information of the industrial equipment, the timestamp confirming the anomaly, and the anomaly type are encapsulated into a message, and the message is sent to the cloud platform as the first priority data.

[0018] Optionally, sending second-priority data to the cloud platform includes:

[0019] The production process record data and status monitoring data of the industrial equipment are sent to the cloud platform as the second priority data.

[0020] Secondly, embodiments of this application also provide an intelligent manufacturing control method applied to a cloud platform, wherein the cloud platform is communicatively connected to at least one edge terminal, and the method includes:

[0021] The system receives first priority data and second priority data sent by each of the at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an anomaly in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no anomaly in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data.

[0022] Emergency scheduling is performed based on the first priority data, and long-term storage and in-depth analysis are performed based on the second priority data.

[0023] Optionally, the method further includes:

[0024] Receive model difference parameters sent by each edge terminal in the at least one edge terminal;

[0025] The model difference parameters sent by each edge terminal are federated through a federated learning mechanism to obtain the updated global model parameters.

[0026] The updated global model parameters are compressed and periodically distributed to each edge terminal.

[0027] Optionally, the method further includes:

[0028] By using a reinforcement learning engine and a preset reward function, the downtime, inventory cost, and energy consumption indicators of industrial equipment are optimized. The preset reward function is used to characterize the relationship between the reward value and the availability, inventory health, and energy consumption indicators of industrial equipment. The reinforcement learning engine is used to maximize the reward value of the preset reward function.

[0029] Thirdly, this application also provides an intelligent manufacturing control device applied to an edge terminal, wherein the edge terminal is communicatively connected to industrial equipment and a cloud platform, and the device includes:

[0030] The acquisition module is used to acquire the status monitoring data corresponding to the industrial equipment;

[0031] The determination module is used to determine whether there is an abnormality in the industrial equipment based on the status monitoring data;

[0032] The first sending module is used to perform emergency stop control on the industrial equipment in the event of an abnormality, and to send first priority data to the cloud platform so that the cloud platform can perform emergency scheduling processing based on the first priority data.

[0033] The second sending module is used to send second priority data to the cloud platform when there is no abnormality in the industrial equipment, so that the cloud platform can perform long-term storage and in-depth analysis based on the second priority data, wherein the priority of the second priority data is lower than the priority of the first priority data.

[0034] Fourthly, embodiments of this application also provide an intelligent manufacturing control device applied to a cloud platform, wherein the cloud platform is communicatively connected to at least one edge terminal, and the device includes:

[0035] The first receiving module is configured to receive first priority data and second priority data sent by each edge terminal in the at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an abnormality in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no abnormality in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data.

[0036] The analysis module is used to perform emergency scheduling processing based on the first priority data and to perform long-term storage and in-depth analysis based on the second priority data.

[0037] Fifthly, embodiments of this application also provide an intelligent manufacturing control system, the system comprising a cloud platform, at least one edge terminal communicatively connected to the cloud platform, and industrial equipment communicatively connected to each of the at least one edge terminal;

[0038] Wherein, each of the at least one edge terminal is used to execute the intelligent manufacturing control method described in the first aspect;

[0039] The cloud platform is used to execute the intelligent manufacturing control method described in the second aspect.

[0040] Compared with the prior art, the technical solution provided in this application has the following advantages: The method provided in this application acquires the status monitoring data corresponding to the industrial equipment; determines whether the industrial equipment is abnormal based on the status monitoring data; if the industrial equipment is abnormal, performs emergency stop control on the industrial equipment and sends first priority data to the cloud platform for emergency scheduling based on the first priority data; if the industrial equipment is not abnormal, sends second priority data to the cloud platform for long-term storage and in-depth analysis based on the second priority data, wherein the priority of the second priority data is lower than the priority of the first priority data. Through the above method, the edge terminal can monitor the status of the industrial equipment in real time, perform emergency stop control on the industrial equipment in real time when an abnormality occurs, and coordinate with the cloud platform for emergency scheduling; and also coordinate with the cloud platform for in-depth data analysis when the industrial equipment is not abnormal. This effectively balances the real-time performance of data processing with the depth of data analysis, thus solving the problem that traditional intelligent manufacturing control methods cannot adequately balance these two aspects. Attached Figure Description

[0041] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0044] Figure 1 A flowchart illustrating an intelligent manufacturing control method provided in an embodiment of this application;

[0045] Figure 2 A flowchart illustrating yet another intelligent manufacturing control method provided in this application embodiment;

[0046] Figure 3 This is a schematic diagram of the structure of an intelligent manufacturing control device provided in an embodiment of this application;

[0047] Figure 4 A schematic diagram of the structure of another intelligent manufacturing control device provided in the embodiments of this application;

[0048] Figure 5 This is a schematic diagram of the structure of an intelligent manufacturing control system provided in an embodiment of this application;

[0049] Figure 6 This is a schematic diagram of the structure of another intelligent manufacturing control system provided in the embodiments of this application;

[0050] Figure 7 A schematic diagram of a real-time data processing flow at the edge layer provided in an embodiment of this application;

[0051] Figure 8 This is a schematic diagram of a multi-objective collaborative optimization process at the cloud layer, provided as an embodiment of this application. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0053] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0054] To address the problem that traditional intelligent manufacturing control methods cannot adequately balance the real-time performance of data processing with the depth of data analysis, this application provides an intelligent manufacturing control method, device, and system that can effectively balance the real-time performance of data processing with the depth of data analysis.

[0055] See Figure 1 , Figure 1 This is a flowchart illustrating an intelligent manufacturing control method provided in an embodiment of this application. Figure 1 As shown, this intelligent manufacturing control method is applied to an edge terminal, which is communicatively connected to both industrial equipment and a cloud platform. The intelligent manufacturing control method may include the following steps:

[0056] Step S101: Obtain the status monitoring data corresponding to the industrial equipment.

[0057] Specifically, the intelligent manufacturing control method provided in this application embodiment can be applied to an edge terminal, which is communicatively connected to both industrial equipment and a cloud platform. The industrial equipment can vary depending on the actual application scenario and can be any manufacturing-related industrial equipment, such as robotic arms on a production line, automotive production equipment, or electrical appliance production equipment. The edge terminal can be a computing-capable terminal located near the industrial equipment, such as a personal computer, laptop, or smartphone. The cloud platform can be an independent node or a node in a cluster. In addition to communicating with one or more edge terminals, the cloud platform can also interface with systems such as Manufacturing Execution System (MES) and Enterprise Resource Planning (ERP) systems to obtain multi-dimensional data.

[0058] The aforementioned condition monitoring data refers to condition data related to industrial equipment, such as temperature, vibration, moisture content, power, and rotational speed. This condition monitoring data can vary depending on the application scenario. Various sensors can be used to acquire condition monitoring data; for example, temperature sensors can be used to collect the temperature of industrial equipment, vibration sensors can be used to collect the vibration of industrial equipment, and moisture content sensors can be used to collect the moisture content of industrial equipment.

[0059] Step S102: Based on the condition monitoring data, determine whether there are any abnormalities in the industrial equipment.

[0060] After acquiring status monitoring data, the edge terminal can analyze the data in real time to determine whether there are any abnormalities in the industrial equipment and perform different subsequent processing logic based on the determination results. During the analysis of the status monitoring data, analysis can be based on pre-trained models or preset judgment rules to determine the abnormal state of the industrial equipment according to each status monitoring data point.

[0061] Step S103: In the event of an abnormality in the industrial equipment, perform emergency stop control on the industrial equipment and send first priority data to the cloud platform so that the cloud platform can perform emergency dispatch processing based on the first priority data.

[0062] When industrial equipment malfunctions, the edge terminal can initiate an emergency stop and send first-priority data to the cloud platform. The cloud platform can then perform emergency scheduling based on this first-priority data, such as triggering warehouse replenishment or production adjustments. This first-priority data can be abnormal alarm data from the industrial equipment, which may include, but is not limited to, the equipment's identification information, the timestamp confirming the abnormality, and the type of abnormality.

[0063] Step S104: If there are no abnormalities in the industrial equipment, send the second priority data to the cloud platform so that the cloud platform can perform long-term storage and in-depth analysis based on the second priority data. The priority of the second priority data is lower than that of the first priority data.

[0064] When there are no abnormalities in the industrial equipment, the edge terminal can send second-priority data to the cloud platform. The cloud platform can then perform long-term storage and in-depth analysis based on this second-priority data. This second-priority data can include production process records and status monitoring data from lower-priority industrial equipment.

[0065] Through the above methods, edge terminals can monitor the status of industrial equipment in real time. When abnormalities occur in industrial equipment, they can perform emergency stop control in real time and coordinate with the cloud platform for emergency dispatching. When there are no abnormalities in industrial equipment, they can also coordinate with the cloud platform for in-depth data analysis. This can better balance the real-time performance of data processing and the depth of data analysis, effectively solving the problem that traditional intelligent manufacturing control methods cannot adequately balance the real-time performance of data processing and the depth of data analysis.

[0066] In an optional embodiment, step S102, determining whether there is an anomaly in the industrial equipment based on condition monitoring data, includes:

[0067] Preprocess the condition monitoring data;

[0068] Perform time-domain and / or frequency-domain analysis on the preprocessed state monitoring data to obtain time-domain and / or frequency-domain characteristics;

[0069] A lightweight artificial intelligence model is used to infer anomaly probabilities from time-domain and / or frequency-domain features.

[0070] If the probability of an anomaly exceeds a preset threshold, it is determined that the industrial equipment is malfunctioning.

[0071] If the probability of an anomaly is less than or equal to a preset threshold, it is determined that there is no anomaly in the industrial equipment.

[0072] Specifically, when determining whether industrial equipment is abnormal based on condition monitoring data, the condition monitoring data can be preprocessed first. For example, high-frequency noise can be eliminated by using a second-order Butterworth low-pass filter for the collected vibration signal, and smoothing can be performed by using a moving average filter for the collected temperature signal. After preprocessing the condition monitoring data, the preprocessed data can be divided into preset time windows (e.g., 200ms), and then time-domain and / or frequency-domain analysis can be performed on the data in each time window to obtain time-domain features (e.g., peak-to-peak value, root mean square value, etc.) and / or frequency-domain features (e.g., energy proportion in the 50-200Hz frequency band calculated based on 1024-point Fast Fourier Transform). Then, a lightweight artificial intelligence model is used to infer the time-domain and / or frequency-domain features to obtain the anomaly probability. This anomaly probability refers to the probability that the industrial equipment is abnormal. If the anomaly probability is greater than a preset threshold, If the probability of an anomaly is less than or equal to a preset threshold, then the industrial equipment is considered to be without an anomaly. This preset threshold can be set according to actual needs, such as 0.85. This lightweight AI model can be any predictive model, such as MobileMamba or DeepSeek. As an optional implementation, the lightweight AI model can be an improved ResNet-18 model, which uses channel pruning to remove 50% of redundant channels and utilizes TensorFlow. The Lite Converter tool implements 8-bit integer quantization, compressing the model size from 45MB to 3.8MB. As an alternative implementation, this lightweight AI model can be a lightweight MobileNetV3 model generated using knowledge distillation technology. The generation process of the MobileNetV3 model involves training a large teacher model (such as ResNet-50) on a cloud platform, and then using the "soft labels" (probability distributions) output by the teacher model to guide the training of the student model (i.e., MobileNetV3). This allows the student model to learn the generalization ability of the teacher model, thereby achieving model lightweighting while maintaining high accuracy. The model size is compressed to 9.6MB while maintaining 92% recognition accuracy.

[0073] In this way, edge terminals can determine whether there are any abnormalities in industrial equipment in real time and accurately based on status monitoring data, providing a basis for subsequent processing.

[0074] In an optional embodiment, the above steps of performing emergency stop control on the industrial equipment and sending first priority data to the cloud platform include:

[0075] Generate an emergency stop signal and send it to the corresponding controller of the industrial equipment to perform emergency stop control on the industrial equipment; and,

[0076] The identification information of the industrial equipment, the timestamp confirming the anomaly, and the anomaly type are encapsulated into a message, which is then sent to the cloud platform as the highest priority data.

[0077] Specifically, when an industrial device malfunctions, the edge terminal can generate an emergency stop signal (such as a high-level emergency stop signal) and send it to the corresponding controller, such as a programmable logic controller (PLC), to perform emergency stop control on the industrial device. The edge terminal can send the emergency stop signal to the controller through an industrial gateway, which enables bidirectional communication between the controller and the edge terminal.

[0078] Meanwhile, the edge terminal can also encapsulate the identification information of industrial equipment, the timestamp confirming the anomaly, and the anomaly type into a message, and send the message as the highest priority data to the cloud platform. This message can be in JSON format, facilitating the cloud platform's mapping of heterogeneous data (equipment status, inventory, order priority, etc.) to a unified format.

[0079] In this way, the edge terminal can perform emergency stop control on industrial equipment in real time when there is an anomaly, and coordinate with the cloud platform for emergency dispatch and processing.

[0080] In an optional embodiment, the above step of sending second priority data to the cloud platform includes:

[0081] The production process records and status monitoring data of industrial equipment are sent to the cloud platform as the second priority data.

[0082] Specifically, when there are no abnormalities in the industrial equipment, the edge terminal can send the production process record data and status monitoring data of the industrial equipment to the cloud platform as second-priority data. This second-priority data transmission can be achieved through the transport layer. This transport layer can establish a data channel using the Message Queuing Telemetry Transport (MQTT) 5.0 protocol, setting high-priority topics for transmitting equipment alarm data (i.e., first-priority data) and low-priority topics for transmitting production logs (i.e., second-priority data). The low-priority topics are mainly used to transmit non-real-time critical production process record data (such as output counts and routine operation logs). This data is asynchronously uploaded to the cloud platform for long-term storage and offline analysis when bandwidth allows, avoiding consuming the real-time alarm channel. The MQTT message server (i.e., MQTT broker) in the transport layer is responsible for message routing, caching (especially for low-priority data), and ensuring reliable transmission of critical messages, providing a unified data access point for the upper layer.

[0083] In this way, edge terminals can perform in-depth data analysis in conjunction with the cloud platform when there are no abnormalities in industrial equipment.

[0084] See Figure 2 , Figure 2 This is a flowchart illustrating another intelligent manufacturing control method provided in an embodiment of this application. Figure 2 As shown, this intelligent manufacturing control method is applied to a cloud platform, which is communicatively connected to at least one edge terminal. The intelligent manufacturing control method may include the following steps:

[0085] Step S201: Receive first priority data and second priority data sent by each edge terminal in at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an anomaly in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no anomaly in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data.

[0086] Specifically, the intelligent manufacturing control method provided in this application embodiment can be applied to a cloud platform, which is communicatively connected to at least one edge terminal. The cloud platform can be an independent node or a node in a cluster. In addition to communicating with one or more edge terminals, the cloud platform can also interface with systems such as factory production execution systems and enterprise resource planning systems to obtain multi-dimensional data. The edge terminals can be computing-capable devices located near industrial equipment, such as personal computers, laptops, and smartphones.

[0087] The cloud platform can receive first-priority data and second-priority data sent by each edge terminal in at least one edge terminal. The first-priority data is sent when the edge terminal detects an anomaly in the connected industrial equipment, and the second-priority data is sent when the edge terminal detects no anomaly in the connected industrial equipment. The first-priority data can be anomaly alarm data from the industrial equipment, which may include, but is not limited to, the industrial equipment's identification information, a timestamp confirming the anomaly, and the anomaly type. The second-priority data can be production process record data and status monitoring data from lower-priority industrial equipment.

[0088] Step S202: Perform emergency scheduling based on the first priority data, and perform long-term storage and in-depth analysis based on the second priority data.

[0089] After receiving the first-priority data and the second-priority data, the cloud platform can perform emergency scheduling based on the first-priority data, such as triggering warehouse replenishment or production scheduling adjustments; and it can also perform long-term storage and in-depth analysis based on the second-priority data.

[0090] Through the above methods, edge terminals can monitor the status of industrial equipment in real time. When abnormalities occur in industrial equipment, they can perform emergency stop control in real time and coordinate with the cloud platform for emergency dispatching. When there are no abnormalities in industrial equipment, they can also coordinate with the cloud platform for in-depth data analysis. This can better balance the real-time performance of data processing and the depth of data analysis, effectively solving the problem that traditional intelligent manufacturing control methods cannot adequately balance the real-time performance of data processing and the depth of data analysis.

[0091] In an optional embodiment, the method further includes:

[0092] Receive model difference parameters sent by each edge terminal in at least one edge terminal;

[0093] The model difference parameters sent by each edge terminal are federated through a federated learning mechanism to obtain the updated global model parameters.

[0094] The updated global model parameters are compressed and periodically distributed to each edge terminal.

[0095] Specifically, the aforementioned model difference parameters refer to the parameters that differ between the parameters updated after training the local model on the edge terminal and the initial global model parameters downloaded from the cloud platform.

[0096] The cloud platform receives model difference parameters from all edge terminals participating in federated learning, and aggregates them using a data-weighted average algorithm, calculated using the following formula:

[0097] ΔW_global=Σ(ΔW_i×N_i) / ΣN_i;

[0098] Where ΔW_global represents the aggregated global model parameters, ΔW_i represents the model difference parameters uploaded by the i-th edge terminal, and N_i represents the number of effective samples used by the i-th edge terminal for local model training within this aggregation cycle. This formula calculates the weighted average of the model difference parameters across all edge terminals. The model difference parameters are determined by the sample size N_i of each edge terminal; edge terminals with larger sample sizes contribute more to the global model update. The updated global model parameters can be compressed using differential coding compression and distributed to the edge terminals periodically (e.g., every 24 hours). The size of this compressed update package is approximately 15% of the full model size.

[0099] In this way, edge terminals only upload model difference parameters to the cloud platform, reducing bandwidth consumption while ensuring the accuracy of global model iteration. Furthermore, the cloud platform can integrate historical data from multiple factories based on a federated learning mechanism to build multi-dimensional production optimization models (such as capacity prediction and energy consumption optimization), and dynamically distribute updates to the edge terminals.

[0100] In an optional embodiment, the method further includes:

[0101] By using a reinforcement learning engine and a pre-defined reward function, the downtime, inventory costs, and energy consumption of industrial equipment are optimized. The pre-defined reward function is used to characterize the relationship between the reward value and the availability, inventory health, and energy consumption of industrial equipment, while the reinforcement learning engine is used to maximize the reward value of the pre-defined reward function.

[0102] Specifically, the aforementioned reinforcement learning engine can construct a multi-dimensional state space, including 32 features such as Overall Equipment Effectiveness (OEE), order priority coefficient, and raw material inventory level. The output action space covers equipment start / stop commands, production rate adjustments (±20%), and replenishment strategies. The preset reward function can be expressed by the following formula:

[0103] R_t=K1×(1-T_dowm / T_total)+K2×(1-C_inventory / C_max)-K3×P_energy / P_base;

[0104] Where R_t represents the instantaneous reward at time t, T_dowm represents the equipment downtime, T_total represents the total downtime, (1-T_dowm / T_total) represents the equipment availability, C_inventory represents the current inventory cost, C_max represents the maximum tolerable inventory cost, (1-C_inventory / C_max) reflects the inventory health, P_energy represents the current energy consumption, P_base represents the baseline energy consumption, and K1, K2 and K3 represent the weighting factors, respectively.

[0105] In this way, by designing a preset reward function, the reinforcement learning agent can be driven to maximize equipment uptime (reduce downtime), minimize inventory costs, and control energy consumption at a reasonable level when making decisions, thus comprehensively optimizing equipment downtime, inventory costs, and energy consumption indicators.

[0106] See Figure 3 , Figure 3 This is a schematic diagram of the structure of an intelligent manufacturing control device provided in an embodiment of this application. Figure 3 As shown, the intelligent manufacturing control device 300 is applied to an edge terminal, which is communicatively connected to both industrial equipment and a cloud platform. The intelligent manufacturing control device 300 includes:

[0107] The acquisition module 301 is used to acquire the status monitoring data corresponding to the industrial equipment;

[0108] The determination module 302 is used to determine whether there is an anomaly in the industrial equipment based on the condition monitoring data;

[0109] The first sending module 303 is used to perform emergency stop control on the industrial equipment in the event of an abnormality, and send first priority data to the cloud platform so that the cloud platform can perform emergency scheduling processing based on the first priority data.

[0110] The second sending module 304 is used to send second priority data to the cloud platform when there is no abnormality in the industrial equipment, so that the cloud platform can perform long-term storage and in-depth analysis based on the second priority data, wherein the priority of the second priority data is lower than the priority of the first priority data.

[0111] Furthermore, module 302 includes:

[0112] The preprocessing submodule is used to preprocess the status monitoring data;

[0113] The analysis submodule is used to perform time-domain and / or frequency-domain analysis on the preprocessed state monitoring data to obtain time-domain and / or frequency-domain features.

[0114] The inference submodule is used to infer anomaly probabilities from time-domain and / or frequency-domain features using a lightweight artificial intelligence model.

[0115] The first determination submodule is used to determine that there is an anomaly in the industrial equipment when the anomaly probability is greater than a preset threshold.

[0116] The second determination submodule is used to determine that there is no abnormality in the industrial equipment when the abnormality probability is less than or equal to a preset threshold.

[0117] Furthermore, the first transmitting module 303 includes:

[0118] The first transmitting submodule is used to generate an emergency stop signal and send the emergency stop signal to the controller corresponding to the industrial equipment to perform emergency stop control on the industrial equipment; and,

[0119] The second sending submodule is used to encapsulate the identification information of the industrial equipment, the timestamp confirming the existence of an anomaly, and the anomaly type into a message, and send the message as the first priority data to the cloud platform.

[0120] Furthermore, the second transmitting module 304 includes:

[0121] The third sending submodule is used to send the production process record data and status monitoring data of industrial equipment to the cloud platform as the second priority data.

[0122] It should be noted that the intelligent manufacturing control device 300 can achieve the aforementioned Figure 1 The intelligent manufacturing control method provided in the illustrated embodiments can achieve the same technical effect, and will not be described in detail here.

[0123] See Figure 4 , Figure 4 This is a schematic diagram of the structure of another intelligent manufacturing control device provided in an embodiment of this application. Figure 4 As shown, the intelligent manufacturing control device 400 is applied to a cloud platform, which is communicatively connected to at least one edge terminal. The intelligent manufacturing control device 400 includes:

[0124] The first receiving module 401 is used to receive first priority data and second priority data sent by each edge terminal in at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an abnormality in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no abnormality in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data.

[0125] Analysis module 402 is used for emergency scheduling processing based on first priority data and for long-term storage and in-depth analysis based on second priority data.

[0126] Furthermore, the intelligent manufacturing control device 400 also includes:

[0127] The second receiving module is used to receive model difference parameters sent by each edge terminal in at least one edge terminal;

[0128] The federated learning module is used to perform federated learning on the model difference parameters sent by each edge terminal to obtain the updated global model parameters.

[0129] The compression module is used to compress the updated global model parameters and periodically distribute the compressed global model parameters to each edge terminal.

[0130] Furthermore, the intelligent manufacturing control device 400 also includes:

[0131] The optimization module is used to optimize the downtime, inventory cost, and energy consumption indicators of industrial equipment using a reinforcement learning engine and a preset reward function. The preset reward function is used to characterize the relationship between the reward value and the availability, inventory health, and energy consumption indicators of industrial equipment, while the reinforcement learning engine is used to maximize the reward value of the preset reward function.

[0132] It should be noted that the intelligent manufacturing control device 400 can achieve the aforementioned Figure 2 The intelligent manufacturing control method provided in the illustrated embodiments can achieve the same technical effect, and will not be described in detail here.

[0133] See Figure 5 , Figure 5 This is a schematic diagram of the structure of an intelligent manufacturing control system provided in an embodiment of this application. Figure 5 As shown, the intelligent manufacturing control system 500 includes a cloud platform 501, at least one edge terminal 502 that is communicatively connected to the cloud platform 501, and industrial equipment 503 that is communicatively connected to each edge terminal in the at least one edge terminal 502.

[0134] Each edge terminal in at least one edge terminal 502 is used to perform the aforementioned implementation. Figure 1 The method embodiment shown provides an intelligent manufacturing control method;

[0135] Cloud platform 501 is used to execute the aforementioned implementation Figure 2 The method embodiment shown provides an intelligent manufacturing control method.

[0136] In a practical application, this intelligent manufacturing control system can achieve a dynamic balance between real-time control and global optimization through a hierarchical processing architecture. This intelligent manufacturing control system can adopt a four-layer architecture design, such as... Figure 6As shown in the diagram. The edge layer, deployed within the factory workshop, comprises edge terminals and industrial IoT gateways. The edge terminals can utilize the Raspberry Pi Compute Module 4 (CM4) hardware platform, equipped with a quad-core ARM Cortex-A72 processor, 4GB of Low Power Double Data Rate (LPDDR4) SDRAM, and an edge AI accelerator module. This module connects to temperature and vibration sensors and can be expanded to connect to other sensors (such as moisture sensors) to adapt to different application scenarios. Alternatively, the edge terminals can employ the NVIDIA Jetson AGX Orin module (NVIDIA's smallest, most powerful, and most energy-efficient AI supercomputer), utilizing its built-in 2048-core Graphics Processing Unit (GPU) to perform parallel processing of multimodal data (such as point cloud data collected by laser rangefinders and thermal imaging data collected by infrared thermal imaging devices). This edge terminal can also deploy a Field Programmable Gate Array (FPGA) to accelerate the vibration analysis model of Long Short-Term Memory (LSTM) networks, reducing inference latency to the 0.8ms level. The industrial IoT gateway can use the UC-8100 series, supporting Modbus TCP (a TCP / IP-based communication protocol) and the Open Platform Unified Architecture (OPC UA) protocol to achieve bidirectional communication between the Programmable Logic Controller (PLC) and the edge terminal. The industrial IoT gateway can also be extended to support the PROFINET (Professional Industrial Network) protocol, allowing direct access to Siemens S7 series PLC devices. The transport layer can use the Message Queuing Telemetry Transport (MQTT) 5.0 protocol to establish a data channel, setting high-priority topics for transmitting device alarm data and low-priority topics for transmitting production logs. Low-priority topics are primarily used to transmit non-real-time critical production process record data (such as production counts, routine operation logs, etc.). This data is asynchronously uploaded to the cloud for long-term storage and offline analysis when bandwidth allows, thus avoiding occupancy of the real-time alarm channel.The MQTT message server (i.e., MQTT broker) in the transport layer is responsible for message routing, caching (especially for low-priority data), and ensuring reliable transmission of critical messages, providing a unified data access point for the upper layers. The transport layer can also switch to a combination of Constrained Application Protocol (CoAP) and Quick UDP Internet Connections (QUIC). When the packet loss rate of the workshop wireless network exceeds 15%, forward error correction technology is automatically activated to ensure a critical command transmission success rate of ≥99.9%; alternatively, a hybrid network of Time-Sensitive Networking (TSN) and Ultra-Reliable Low Latency Communications (URLLC) can be used, with end-to-end transmission latency consistently below 3ms, meeting the real-time control requirements of high-precision equipment such as CNC machine tools. The cloud layer can build a distributed computing cluster based on Apache Hadoop 3.3.4, configured with 10 nodes (including 1 NameNode and 9 DataNodes). Each node is equipped with dual Intel Xeon Gold 6338 processors and 4 NVIDIA A100 GPUs, running a federated learning framework and a deep reinforcement learning optimization engine. The application layer interfaces with the factory's MES and ERP systems through a Representational State Transfer (RESTful) Application Programming Interface (API) to dynamically issue production scheduling instructions.

[0137] The real-time data processing flow at the edge layer is as follows: Figure 7As shown, to meet the real-time requirements of industrial equipment anomaly detection, the edge terminal executes a multi-stage data processing flow. After sensor data is transmitted to the edge terminal through the hardware interface, it first undergoes signal preprocessing: vibration signals are filtered by a second-order Butterworth low-pass filter to eliminate high-frequency noise, and temperature signals are filtered using a moving average. The preprocessed data is divided into 200ms time windows, and time-domain features (such as peak-to-peak value and root mean square value) and frequency-domain features (such as the energy proportion of the 50-200Hz frequency band calculated based on a 1024-point fast Fourier transform) are extracted. A lightweight artificial intelligence module deploys an improved ResNet-18 model, using channel pruning technology to remove 50% of redundant channels, and uses TensorFlowLite Converter to achieve 8-bit integer quantization, compressing the model size from 45MB to 3.8MB. When the model output anomaly probability exceeds a preset threshold (such as 0.85), a high-level emergency stop signal is immediately sent to the PLC controller via GPIO23 pin. Simultaneously, a JSON message containing the industrial equipment's identification information, a timestamp confirming the anomaly, and the anomaly type is generated and pushed to a high-priority MQTT topic.

[0138] Among them, the multi-objective collaborative optimization process in the cloud layer is as follows: Figure 8 As shown, the cloud layer integrates multi-factory data resources through federated learning to achieve global model optimization. Each edge terminal uploads the model difference parameters of the last fully connected layer every 30 minutes. The cloud aggregation server receives the model difference parameters from all edge terminals participating in federated learning and aggregates the parameters using a data-weighted average algorithm, calculated using the following formula:

[0139] ΔW_global=Σ(ΔW_i×N_i) / ΣN_i;

[0140] Where ΔW_global represents the aggregated global model parameters, ΔW_i represents the model difference parameters uploaded by the i-th edge terminal, and N_i represents the number of effective samples used by the i-th edge terminal for local model training within this aggregation cycle. This formula calculates the weighted average of the model difference parameters across all edge terminals. The model difference parameters are determined by the sample size N_i of each edge terminal; edge terminals with larger sample sizes contribute more to the global model update. The updated global model parameters can be compressed using differential coding compression and distributed to the edge terminals periodically (e.g., every 24 hours). The size of this compressed update package is approximately 15% of the full model size.

[0141] The reinforcement learning engine can construct a multi-dimensional state space, including 32 features such as Overall Equipment Effectiveness (OEE), order priority coefficient, and raw material inventory level. The output action space covers equipment start / stop commands, production rate adjustments (±20%), and replenishment strategies. The aforementioned preset reward function can be expressed by the following formula:

[0142] R_t=K1×(1-T_dowm / T_total)+K2×(1-C_inventory / C_max)-K3×P_energy / P_base;

[0143] Where R_t represents the instantaneous reward at time t, T_dowm represents the equipment downtime, T_total represents the total downtime, (1-T_dowm / T_total) represents the equipment availability, C_inventory represents the current inventory cost, C_max represents the maximum tolerable inventory cost, (1-C_inventory / C_max) reflects the inventory health, P_energy represents the current energy consumption, P_base represents the baseline energy consumption, and K1, K2 and K3 represent the weighting factors, respectively.

[0144] To address the data silo problem across multiple subsystems, this intelligent manufacturing control system also defines a standardized event coding system and response protocol. Critical events, including equipment overheating, inventory shortages, and order delivery date changes, are broadcast to the ` / edge / event` topic using an MQTT publish-subscribe model. Subscribers (such as the warehouse management system) must return a confirmation response within 200ms; timeout events trigger a backup communication link. All critical operation records are stored on a blockchain. Blockchain nodes are deployed independently (or partially physically co-located with Hadoop cluster nodes but logically isolated). A 256-bit Secure Hash Algorithm (SHA-256) is used to generate data hash values, which are synchronized to the cloud platform verification nodes every 5 minutes to ensure the immutability of operation records.

[0145] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent manufacturing control method provided in any of the foregoing method embodiments.

[0146] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0147] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0148] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0149] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A smart manufacturing control method, characterized in that, Applied to edge terminals, which are communicatively connected to industrial equipment and a cloud platform, the method includes: Obtain the status monitoring data corresponding to the industrial equipment; Based on the status monitoring data, determine whether the industrial equipment is malfunctioning; In the event of an abnormality in the industrial equipment, an emergency stop control is implemented on the industrial equipment, and first priority data is sent to the cloud platform so that the cloud platform can perform emergency dispatch processing based on the first priority data; If there are no abnormalities in the industrial equipment, second priority data is sent to the cloud platform for long-term storage and in-depth analysis. The priority of the second priority data is lower than that of the first priority data.

2. The method according to claim 1, characterized in that, The step of determining whether the industrial equipment is abnormal based on the status monitoring data includes: The status monitoring data is preprocessed; Perform time-domain and / or frequency-domain analysis on the preprocessed state monitoring data to obtain time-domain and / or frequency-domain characteristics; An anomaly probability is obtained by inferring from the time-domain features and / or the frequency-domain features using a lightweight artificial intelligence model; If the probability of an anomaly is greater than a preset threshold, it is determined that the industrial equipment is malfunctioning. If the probability of an anomaly is less than or equal to the preset threshold, it is determined that the industrial equipment is not abnormal.

3. The method according to claim 1, characterized in that, The step of performing emergency stop control on the industrial equipment and sending first priority data to the cloud platform includes: An emergency stop signal is generated and sent to the controller corresponding to the industrial equipment to perform emergency stop control on the industrial equipment; and, The identification information of the industrial equipment, the timestamp confirming the anomaly, and the anomaly type are encapsulated into a message, and the message is sent to the cloud platform as the first priority data.

4. The method according to claim 1, characterized in that, Sending second-priority data to the cloud platform includes: The production process record data and status monitoring data of the industrial equipment are sent to the cloud platform as the second priority data.

5. A smart manufacturing control method, characterized in that, The method is applied to a cloud platform, which is communicatively connected to at least one edge terminal, and includes: The system receives first priority data and second priority data sent by each of the at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an anomaly in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no anomaly in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data. Emergency scheduling is performed based on the first priority data, and long-term storage and in-depth analysis are performed based on the second priority data.

6. The method according to claim 5, characterized in that, The method further includes: Receive model difference parameters sent by each edge terminal in the at least one edge terminal; The model difference parameters sent by each edge terminal are federated through a federated learning mechanism to obtain the updated global model parameters. The updated global model parameters are compressed and periodically distributed to each edge terminal.

7. The method according to claim 5, characterized in that, The method further includes: By using a reinforcement learning engine and a preset reward function, the downtime, inventory cost, and energy consumption indicators of industrial equipment are optimized. The preset reward function is used to characterize the relationship between the reward value and the availability, inventory health, and energy consumption indicators of industrial equipment. The reinforcement learning engine is used to maximize the reward value of the preset reward function.

8. A smart manufacturing control device, characterized in that, The device is applied to edge terminals, which are communicatively connected to industrial equipment and a cloud platform, and includes: The acquisition module is used to acquire the status monitoring data corresponding to the industrial equipment; The determination module is used to determine whether there is an abnormality in the industrial equipment based on the status monitoring data; The first sending module is used to perform emergency stop control on the industrial equipment in the event of an abnormality, and to send first priority data to the cloud platform so that the cloud platform can perform emergency scheduling processing based on the first priority data. The second sending module is used to send second priority data to the cloud platform when there is no abnormality in the industrial equipment, so that the cloud platform can perform long-term storage and in-depth analysis based on the second priority data, wherein the priority of the second priority data is lower than the priority of the first priority data.

9. A smart manufacturing control device, characterized in that, The device is applied to a cloud platform, which is communicatively connected to at least one edge terminal, and includes: The first receiving module is configured to receive first priority data and second priority data sent by each edge terminal in the at least one edge terminal, wherein the first priority data is sent when the edge terminal detects an abnormality in the industrial equipment connected to it, and the second priority data is sent when the edge terminal detects that there is no abnormality in the industrial equipment connected to it, and the priority of the second priority data is lower than the priority of the first priority data. The analysis module is used to perform emergency scheduling processing based on the first priority data and to perform long-term storage and in-depth analysis based on the second priority data.

10. An intelligent manufacturing control system, characterized in that, The system includes a cloud platform, at least one edge terminal communicatively connected to the cloud platform, and industrial equipment communicatively connected to each of the at least one edge terminal. Wherein, each of the at least one edge terminal is used to execute the intelligent manufacturing control method according to any one of claims 1-4; The cloud platform is used to execute the intelligent manufacturing control method according to any one of claims 5-7.

Images