A digital and intelligent factory management and control platform system

Abstract

The application relates to an intelligent factory management and control platform system, and relates to the technical field of industrial automation.The system comprises a data acquisition module, which is used for collecting state data and production environment parameters of production equipment in real time through a plurality of Internet of Things sensors arranged on a production line.The state data comprises temperature, pressure and vibration frequency.In the application, firstly, massive raw data are subjected to localized real-time processing and feature extraction through an edge computing node, so that the load pressure of a network and a central server is significantly reduced, and millisecond-level rapid response to equipment emergency abnormalities is realized;secondly, an artificial intelligence analysis engine integrated in the central server is used for deep mining and predictive analysis of the simplified data, so that a leap from passive alarm to active prediction is realized, potential equipment faults and production abnormality trends are accurately identified;and thirdly, the system is further combined with intelligent visual monitoring, so that real-time compliance detection of personnel operation and material state is realized.

Description

Technical Field

[0001] This application relates to the field of industrial automation technology, and in particular to a digital intelligent factory management and control platform system. Background Technology

[0002] Modern manufacturing faces multiple challenges, including improving production efficiency, ensuring product quality, reducing operating costs, and responding quickly to market changes. Traditional production line monitoring systems often suffer from problems such as data silos, response delays, and insufficient analytical depth. While IoT technology can collect equipment data, the transmission and processing of massive amounts of raw data puts enormous pressure on network bandwidth and central servers. At the same time, anomalies in the production process, whether due to equipment status or human error, require faster and smarter identification and response mechanisms. Existing production planning also relies heavily on historical experience and static data, making it difficult to adapt to dynamic market changes and internal production conditions. Summary of the Invention

[0003] To address the aforementioned issues, this application provides a digital intelligent factory management platform system.

[0004] This application provides a digital intelligent factory management platform system, which adopts the following technical solution: A digital intelligent factory management and control platform system includes: The data acquisition module is used to collect real-time status data of production equipment and production environment parameters through multiple IoT sensors deployed on the production line. The status data includes temperature, pressure and vibration frequency. An edge computing node is communicatively connected to the data acquisition module. The edge computing node is used to perform preliminary processing on the acquired real-time data. The preliminary processing includes data filtering, outlier removal, and data compression. The preliminary processing is used to generate a simplified dataset. A central server, which is communicatively connected to the edge computing nodes, is used to receive the simplified dataset and run an artificial intelligence analysis engine to perform in-depth analysis on the simplified dataset in order to identify abnormal production trends. The control execution module is communicatively connected to the central server. The output of the control execution module is also communicatively connected to the actuators on the production line. The control execution module is used to generate control commands based on the analysis results output by the artificial intelligence analysis engine to automatically adjust the operating parameters of relevant production equipment or generate early warning information and push it to the management terminal.

[0005] As a preferred technical solution of this application, the edge computing node includes: The data preprocessing unit is used to clean and format real-time data by applying preset rules; A feature extraction unit is used to extract key feature indicators from the cleaned data. The local decision-making unit is used to compare the key feature indicators with preset thresholds. When an emergency anomaly is determined, it directly sends a primary control command to the control execution module and simultaneously marks the relevant data as high priority and sends it to the central server.

[0006] As a preferred technical solution of this application, the artificial intelligence analysis engine includes: A real-time monitoring sub-engine continuously analyzes the simplified dataset based on streaming data processing technology to detect anomalies that deviate from normal operating conditions in real time. The predictive analysis sub-engine is based on a machine learning model and predicts the potential failure risk of a specified device within a preset time period based on historical data and the current operating status. An optimization suggestion sub-engine generates optimization information, including parameter adjustment suggestions or maintenance plans, based on the analysis results.

[0007] As a preferred technical solution of this application, the output end of the central server is further communicatively connected to an intelligent visual monitoring module, the intelligent visual monitoring module being communicatively connected to the edge computing node, and the intelligent visual monitoring module comprising: An image acquisition unit, which is a high-definition industrial camera deployed at a key workstation, is used to continuously acquire video streams of the production operation process; An image recognition unit is communicatively connected to the image acquisition unit. The image recognition unit is used to run an AI image recognition algorithm to analyze the video stream in real time, so as to identify whether the operator's actions conform to the preset specifications and detect the existence and location of the production materials. An alarm triggering unit is communicatively connected to the image recognition unit. The alarm triggering unit is used to immediately trigger an audiovisual alarm and generate an alarm log containing timestamps and workstation information when non-compliant operations or material abnormalities are detected. The central server is used to perform correlation analysis between the alarm log and the real-time equipment data collected by the data acquisition module.

[0008] As a preferred technical solution of this application, the AI ​​image recognition algorithm is a pose recognition model trained by deep learning, and the image recognition unit is used to classify the severity level of the identified non-compliant operations. For non-compliant operations classified as high severity, the alarm triggering unit sends an instruction to the control execution module to suspend the operation of the relevant production line segment.

[0009] As a preferred technical solution of this application, the edge computing node is configured as follows: The edge computing node receives analysis results from the image recognition unit. When the analysis results indicate non-compliant operation and real-time device data from the data acquisition module indicates abnormal device operating parameters, the edge computing node is triggered to execute a predefined emergency intervention process.

[0010] As a preferred technical solution of this application, the output end of the central server is further communicatively connected to a production planning optimization module, the production planning optimization module comprising: A data integration interface, which is used to obtain historical sales data, market trend information and raw material supply data from market data sources; A prediction model unit is used to run a machine learning prediction algorithm. Based on the acquired data and the current production efficiency assessment provided by the artificial intelligence analysis engine, the prediction model unit predicts the product demand and optimal product mix for a specific future period. The planning generation unit is used to automatically generate detailed production plans and schedules based on the prediction results, and send them to the control execution module to adjust the production rhythm and product type of the production line.

[0011] As a preferred technical solution of this application, the machine learning prediction algorithm is a time series analysis model or a regression model, and the prediction model unit is further configured as follows: The system periodically receives real-time production data and equipment status data from the central server to dynamically update the parameters of the prediction model. Based on the updated model parameters, it re-predicts demand and adjusts the production schedule on a rolling basis.

[0012] In summary, this application includes at least one of the following beneficial technical effects of a digital intelligent factory management platform system: This application, firstly, utilizes edge computing nodes to perform localized real-time processing and feature extraction of massive amounts of raw data, significantly reducing the load on the network and central server, and achieving millisecond-level rapid response to emergency equipment anomalies. Secondly, the artificial intelligence analysis engine integrated into the central server performs in-depth mining and predictive analysis of the simplified data, achieving a leap from passive alarms to proactive prediction, accurately identifying potential equipment failures and production anomaly trends. Thirdly, the system further integrates intelligent visual monitoring, enabling real-time compliance detection of personnel operations and material status, forming a multi-dimensional collaborative safety protection network for people, machines, and materials. Finally, the production planning optimization module drives production planning from static experience-based decision-making to dynamic data-driven decision-making through internal and external data fusion and rolling optimization, enabling the production line to flexibly adapt to market fluctuations. Attached Figure Description

[0013] Figure 1 This is the system architecture diagram of the digital factory management platform of this application. Detailed Implementation

[0014] The following is in conjunction with the appendix Figure 1 This application will be described in further detail.

[0015] See Figure 1 A digital intelligent factory management and control platform system, characterized in that it includes: The data acquisition module is used to collect real-time status data of production equipment and production environment parameters through multiple IoT sensors deployed on the production line. The status data includes temperature, pressure and vibration frequency. The data acquisition module consists of a large number of high-precision industrial IoT sensors that are directly deployed at key monitoring points of production line equipment. For example, PT100 platinum resistance temperature sensors are tightly installed on the motor bearing housing or reactor wall to monitor temperature rise; piezoresistive pressure sensors are integrated into hydraulic station pipelines or pressure vessel interfaces to sense pressure fluctuations; and ICP accelerometers are attached to the housing of rotating equipment (such as pumps and fans) via magnetic bases to capture vibration signals. These sensors convert physical quantities into standard electrical signals (such as 4-20mA analog signals or direct digital signals) at a specific sampling frequency based on the physical characteristics of the measured parameters. The signals are transmitted in real time to the nearest edge computing node via wired (such as RS-485 bus, Ethernet) or wireless (such as Wi-Fi, LoRa) industrial network protocols. Before transmission, the data is encapsulated into a standard data packet containing a unique device ID, sensor type, high-precision timestamp, measurement value, and status code to ensure the traceability of the data source.

[0016] Edge computing nodes communicate with the data acquisition module and perform preliminary processing on the acquired real-time data. This preliminary processing includes data filtering, outlier removal, and data compression to generate a simplified dataset. Each edge computing node includes: a data preprocessing unit, which cleans and formats the real-time data using preset rules; a feature extraction unit, which extracts key feature indicators from the cleaned data; and a local decision-making unit, which compares the key feature indicators with preset thresholds. When an urgent anomaly is detected, the unit directly sends a primary control command to the control execution module and simultaneously marks the relevant data as high priority before sending it to the central server.

[0017] First, the data preprocessing unit starts immediately and applies preset rules to clean the data: including range filtering based on physical probability (such as removing obviously absurd temperature values), using statistical methods (such as Z-score) to identify and remove outliers caused by transient interference, and using techniques such as dead zone compression or rotating door algorithm to perform lossy compression on high-frequency data, which significantly reduces the amount of data while ensuring that key information is not lost. Finally, the data is formatted into a unified standard (such as JSON). Next, the feature extraction unit calculates key feature indicators from the cleaned data. For example, it calculates the sliding window mean and standard deviation for temperature and pressure data, and performs a fast Fourier transform on vibration data to extract the main frequency amplitude, thereby generating a concise dataset that characterizes the current state of the equipment. The local decision-making unit continuously compares the extracted key features (such as average temperature and vibration RMS value) with the preset safety thresholds within the node in real time. Once an emergency abnormal condition is determined to be met (such as continuous temperature exceeding the limit), the unit will bypass the cloud and directly send a primary control command (such as emergency stop) to the control execution module. At the same time, this event and related data will be marked as the highest priority and quickly reported to the central server through an independent channel, realizing the coordination between rapid local autonomy and timely cloud perception.

[0018] The central server communicates with the edge computing nodes. It receives a simplified dataset and runs an AI analysis engine to perform in-depth analysis to identify abnormal production trends. The AI ​​analysis engine includes: a real-time monitoring sub-engine, which continuously analyzes the simplified dataset using streaming data processing technology to detect anomalies deviating from normal operating conditions in real time; a predictive analysis sub-engine, which uses machine learning models to predict potential failure risks of specified equipment within a preset future timeframe based on historical data and current operating status; and an optimization suggestion sub-engine, which generates optimization information based on the analysis results, including parameter adjustment suggestions or maintenance plans.

[0019] The central server receives simplified datasets and high-priority event alarms uploaded from various edge computing nodes. These massive amounts of time-series data are stored in a high-performance database. The artificial intelligence analysis engine is then activated: the real-time monitoring sub-engine performs real-time calculations on the continuously flowing data stream based on a stream processing framework (such as Apache Flink) and instantly detects minor abnormal deviations through a dynamically built normal operating condition model (such as using the isolated forest algorithm). The predictive analytics sub-engine runs complex machine learning models (such as LSTM long short-term memory networks). It calls on historical equipment operation data, maintenance records and current status for comprehensive analysis to predict the potential failure probability of specified equipment in a specific future time period and generate predictive maintenance alerts. The optimization suggestion sub-engine integrates the results of real-time monitoring and predictive analysis with the built-in process knowledge base and expert rules to automatically generate specific parameter adjustment suggestions or precise maintenance plans. In addition, the server also performs multi-source data correlation analysis tasks, such as aligning abnormal equipment data with operational anomalies identified by visual monitoring on the timeline to trace the root cause of the fault.

[0020] The control execution module communicates with the central server, and its output is also connected to the actuators on the production line. Based on the analysis results output by the artificial intelligence analysis engine, the control execution module generates control commands to automatically adjust the operating parameters of relevant production equipment or generate early warning information and push it to the management terminal.

[0021] The control execution module is used to receive primary control commands sent by edge computing nodes in emergency situations, emergency stop commands sent by the intelligent vision monitoring module when high-risk situations are detected, and optimization and adjustment commands based on deep analysis issued by the central server or new production schedules generated by the production planning optimization module. The control execution module has a priority arbitration logic inside, and safety-related emergency commands usually have the highest interrupt priority. After receiving the command, the control execution module parses it and converts it into standard control signals of specific industrial fieldbus protocols (such as PROFIBUS, PROFINET, EtherCAT) or industrial Internet of Things protocols (such as OPC UA, MQTT). These signals are sent to various actuators on the production line, such as: adjusting the frequency of the frequency converter to change the motor speed, driving the opening and closing of the solenoid valve to control the flow of fluid, or guiding the robot to complete specific actions. For information that is not directly controlled, such as predictive maintenance warnings, optimization suggestions, or production plan change notifications, the module is responsible for formatting them and accurately pushing them to the management terminals of relevant personnel through integrated messaging services or interfaces, ensuring timely information delivery and forming a complete closed loop of decision-making, execution, and feedback.

[0022] The central server's output is also connected to an intelligent visual monitoring module, which communicates with edge computing nodes. The intelligent visual monitoring module includes: an image acquisition unit, consisting of high-definition industrial cameras deployed at key workstations, continuously acquiring video streams of the production process; an image recognition unit, communicating with the image acquisition unit, running AI image recognition algorithms to analyze the video stream in real time to identify whether operator actions conform to preset specifications and detect the presence and location of production materials; and an alarm triggering unit, communicating with the image recognition unit, immediately triggering audiovisual alarms and generating alarm logs containing timestamps and workstation information when non-compliant operations or material anomalies are detected. The central server then correlates these alarm logs with real-time equipment data collected by the data acquisition module. The AI ​​image recognition algorithm is a deep learning-trained posture recognition model. The image recognition unit classifies detected non-compliant operations by severity level. For non-compliant operations classified as high-severity, the alarm triggering unit sends instructions to the control execution module to suspend the relevant production line segment. The edge computing node is configured to receive analysis results from the image recognition unit. When the analysis results indicate non-compliant operation and real-time device data from the data acquisition module indicates abnormal device operating parameters, the edge computing node is triggered to execute a predefined emergency intervention process.

[0023] The image acquisition unit consists of high-definition industrial cameras deployed in key areas such as assembly, inspection, and hazardous workstations. Each camera is equipped with a suitable lens and lighting system to continuously acquire high-quality video stream data at a specific frame rate. This video stream is transmitted in real-time to the image recognition unit, which incorporates a deep learning model trained on massive amounts of industrial scene image data (such as a YOLO or OpenPose-based object detection and pose recognition model). This model can analyze video frames in real-time, identify the skeletal key points of operators, compare their movements with preset standard operating procedures, determine operational compliance (e.g., whether safety equipment is worn, whether movements are standardized), and detect the presence, location, and basic appearance of production materials. The recognition results, along with their severity level classification, are sent to the alarm triggering unit, which immediately takes action according to a preset strategy. For low-level anomalies, trigger on-site audible and visual alarms and generate logs with timestamps and workstation information, which are then uploaded to the server. For high-severity anomalies (such as serious violations that may lead to accidents), an emergency stop command is sent directly to the control execution module to force the production line to stop, thereby realizing visual closed-loop management from perception to intervention and greatly improving the level of safety management.

[0024] The central server's output is also connected to a production planning optimization module, which includes: a data integration interface for acquiring historical sales data, market trend information, and raw material supply data from market data sources; a prediction model unit for running machine learning prediction algorithms, which predict product demand and optimal product mix for future specific periods based on the acquired data and current production efficiency assessments provided by the artificial intelligence analysis engine; and a plan generation unit for automatically generating detailed production schedules based on the prediction results and sending them to the control execution module to adjust the production line's pace and product types. The machine learning prediction algorithm is either a time series analysis model or a regression model. The prediction model unit is also configured to periodically receive real-time production data and equipment status data from the central server to dynamically update the prediction model's parameters, and re-predict demand and continuously adjust the production schedule based on the updated model parameters.

[0025] The production planning optimization module first automatically acquires key information such as historical sales data, market trend forecasts, and raw material prices and supply from external market data sources through a data integration interface. This external data, along with internal real-time production efficiency assessments (such as Overall Equipment Effectiveness (OEE) and equipment health status) provided by the central server, is input into the prediction model unit. This unit runs advanced machine learning algorithms, such as using time series analysis models to predict product demand for specific future periods (e.g., next week, next month) or using regression models to analyze the optimal product mix under the influence of multiple factors. Based on the fusion analysis of internal and external data, the model outputs demand forecasts and optimization suggestions. Subsequently, the planning generation unit is activated. It comprehensively considers factors such as forecast results, actual production line capacity, material inventory, and process constraints, and uses operations research optimization algorithms (such as linear programming) to automatically generate a detailed, executable production schedule, including production orders, sequence, start / end times, etc. This schedule is sent to the control execution module to guide the production line in adjusting its production rhythm and product types. The production planning optimization module also has rolling optimization capabilities, regularly receiving the latest production data, dynamically updating the parameters of the prediction model, and readjusting future plans to ensure that the production plan always closely reflects actual market changes and internal production status.

[0026] This application, firstly, utilizes edge computing nodes to perform localized real-time processing and feature extraction of massive amounts of raw data, significantly reducing the load on the network and central server, and achieving millisecond-level rapid response to emergency equipment anomalies. Secondly, the artificial intelligence analysis engine integrated into the central server performs in-depth mining and predictive analysis of the simplified data, achieving a leap from passive alarms to proactive prediction, accurately identifying potential equipment failures and production anomaly trends. Thirdly, the system further integrates intelligent visual monitoring, enabling real-time compliance detection of personnel operations and material status, forming a multi-dimensional collaborative safety protection network for people, machines, and materials. Finally, the production planning optimization module drives production planning from static experience-based decision-making to dynamic data-driven decision-making through internal and external data fusion and rolling optimization, enabling the production line to flexibly adapt to market fluctuations.

[0027] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A digital intelligent factory management and control platform system, characterized in that, include: The data acquisition module is used to collect real-time status data of production equipment and production environment parameters through multiple IoT sensors deployed on the production line. The status data includes temperature, pressure and vibration frequency. An edge computing node is communicatively connected to the data acquisition module. The edge computing node is used to perform preliminary processing on the acquired real-time data. The preliminary processing includes data filtering, outlier removal, and data compression. The preliminary processing is used to generate a simplified dataset. A central server, which is communicatively connected to the edge computing nodes, is used to receive the simplified dataset and run an artificial intelligence analysis engine to perform in-depth analysis on the simplified dataset in order to identify abnormal production trends. The control execution module is communicatively connected to the central server. The output of the control execution module is also communicatively connected to the actuators on the production line. The control execution module is used to generate control commands based on the analysis results output by the artificial intelligence analysis engine to automatically adjust the operating parameters of relevant production equipment or generate early warning information and push it to the management terminal.

2. The intelligent factory management platform system according to claim 1, characterized in that, The edge computing nodes include: The data preprocessing unit is used to clean and format real-time data by applying preset rules; A feature extraction unit is used to extract key feature indicators from the cleaned data. The local decision-making unit is used to compare the key feature indicators with preset thresholds. When an emergency anomaly is determined, it directly sends a primary control command to the control execution module and simultaneously marks the relevant data as high priority and sends it to the central server.

3. The intelligent factory management platform system according to claim 1, characterized in that, The artificial intelligence analysis engine includes: A real-time monitoring sub-engine continuously analyzes the simplified dataset based on streaming data processing technology to detect anomalies that deviate from normal operating conditions in real time. The predictive analysis sub-engine is based on a machine learning model and predicts the potential failure risk of a specified device within a preset time period based on historical data and the current operating status. An optimization suggestion sub-engine generates optimization information, including parameter adjustment suggestions or maintenance plans, based on the analysis results.

4. The intelligent factory management platform system according to claim 1, characterized in that, The output of the central server is also communicatively connected to an intelligent visual monitoring module, which is communicatively connected to the edge computing node. The intelligent visual monitoring module includes: An image acquisition unit, which is a high-definition industrial camera deployed at a key workstation, is used to continuously acquire video streams of the production operation process; An image recognition unit is communicatively connected to the image acquisition unit. The image recognition unit is used to run an AI image recognition algorithm to analyze the video stream in real time, so as to identify whether the operator's actions conform to the preset specifications and detect the existence and location of the production materials. An alarm triggering unit is communicatively connected to the image recognition unit. The alarm triggering unit is used to immediately trigger an audiovisual alarm and generate an alarm log containing timestamps and workstation information when non-compliant operations or material abnormalities are detected. The central server is used to perform correlation analysis between the alarm log and the real-time equipment data collected by the data acquisition module.

5. The intelligent factory management platform system according to claim 4, characterized in that, The AI ​​image recognition algorithm is a pose recognition model trained by deep learning, and the image recognition unit is used to classify the severity level of the identified non-compliant operations. For non-compliant operations classified as high severity, the alarm triggering unit sends an instruction to the control execution module to suspend the operation of the relevant production line segment.

6. The intelligent factory management platform system according to claim 4, characterized in that, The edge computing node is configured as follows: The edge computing node receives analysis results from the image recognition unit. When the analysis results indicate non-compliant operation and real-time device data from the data acquisition module indicates abnormal device operating parameters, the edge computing node is triggered to execute a predefined emergency intervention process.

7. The intelligent factory management platform system according to claim 1, characterized in that, The output of the central server is also communicatively connected to a production planning optimization module, which includes: A data integration interface, which is used to obtain historical sales data, market trend information and raw material supply data from market data sources; A prediction model unit is used to run a machine learning prediction algorithm. Based on the acquired data and the current production efficiency assessment provided by the artificial intelligence analysis engine, the prediction model unit predicts the product demand and optimal product mix for a specific future period. The planning generation unit is used to automatically generate detailed production plans and schedules based on the prediction results, and send them to the control execution module to adjust the production rhythm and product type of the production line.

8. The intelligent factory management platform system according to claim 7, characterized in that, The machine learning prediction algorithm is a time series analysis model or a regression model, and the prediction model unit is further configured as follows: The system periodically receives real-time production data and equipment status data from the central server to dynamically update the parameters of the prediction model. Based on the updated model parameters, it re-predicts demand and adjusts the production schedule on a rolling basis.

Images