Digitalized air suspension conveyor control system

Abstract

The application relates to the technical field of industrial control, and particularly discloses a digital and intelligent air suspension conveyor control system, and a method thereof, which comprises the following steps: a monitoring acquisition module: monitoring real-time equipment operation data, real-time air pressure data and real-time airflow data, and acquiring multiple historical conveying data and multiple edge control sets; a range module: acquiring equipment operation range data, determining air pressure range data and airflow range data; a determination module: determining multiple real-time edge abnormal data; an analysis module: determining multiple cause-phenomenon category abnormal data and a category decision tree model; a construction module: calculating multiple model evaluation values, and constructing multiple conveying decision tree models; and a control module: determining and executing multiple real-time abnormal control strategies. The application realizes accurate prediction and intelligent processing of abnormalities, improves the stability and reliability of air suspension conveyor operation, reduces equipment failure risk and maintenance cost, improves conveying efficiency, and better adapts to complex industrial production environments.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to a digital intelligent air suspension conveyor control system. Background Technology

[0002] Air suspension conveyors originated from the upgrade of traditional pneumatic conveying technology. In the early days, due to the limited level of industrial automation, their control systems were mainly based on manual intervention and simple mechanical control: they only monitored basic parameters such as fan speed and air box static pressure through pointer instruments. Anomaly judgment relied on fixed thresholds, which could not adapt to the conveying needs of different materials. The rigid thresholds often led to false shutdowns or missed faults.

[0003] With the popularization of PLC technology, air suspension conveyors have entered the stage of basic automation. Although PLC can realize the automatic adjustment of fan speed and airflow valves, there are still obvious limitations: First, there is a lack of historical data management capabilities, making it impossible to trace the root cause of anomalies; second, the centralized control architecture requires data from multiple monitoring points to be aggregated and processed in the central control room, resulting in a delay of more than 10 seconds in responding to anomalies in remote equipment; third, anomaly handling relies on the experience of maintenance personnel, there is no standardized strategy library, and the processing efficiency of the same fault varies significantly depending on the operator.

[0004] Therefore, this invention proposes a digital intelligent air suspension conveyor control system. Summary of the Invention

[0005] This invention provides a digital intelligent air-suspension conveyor control system. By acquiring historical conveying data, it determines normal conveying data, abnormal conveying data, air pressure range data, and airflow range data. Combining this with the edge control set of each edge controller, real-time equipment operation data, real-time air pressure data, and real-time airflow data, it determines the real-time edge anomaly data for each edge controller, the cause-phenomenon category anomaly data for each anomaly cause-phenomenon category, and the category decision tree model. The model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller is used to construct the conveying decision tree model for each edge controller, and to determine and execute the real-time anomaly control strategy for each edge controller. This system can monitor and classify anomalies of the air-suspension conveyor in real-time and accurately, achieving accurate prediction and intelligent processing of anomalies. This improves the stability and reliability of the air-suspension conveyor operation, reduces equipment failure risks and maintenance costs, and simultaneously increases conveying efficiency, better adapting to complex industrial production environments.

[0006] This invention provides a digital intelligent air-suspended conveyor control system, comprising:

[0007] Monitoring and Acquisition Module: Monitors real-time equipment operation data, real-time air pressure data, and real-time airflow data of the air suspension conveyor; acquires historical conveying data from multiple conveying operations; and acquires edge control sets from multiple edge controllers.

[0008] Range module: acquires equipment operating range data of air suspension conveyor, determines normal and abnormal conveying data based on historical conveying data, and determines air pressure range data and airflow range data based on normal conveying data;

[0009] Determination Module: Based on the edge control set, operating range data, air pressure range data, airflow range data, real-time equipment operation data, real-time air pressure data, and real-time airflow data of each edge controller, determine the real-time edge anomaly data of each edge controller;

[0010] Analysis module: Analyzes abnormal transmission data and the edge control set of each edge controller to determine the cause-phenomenon category of each abnormality for each edge controller and the category decision tree model;

[0011] Building Module: Based on abnormal delivery data, calculate the model evaluation value of the category decision tree model for each abnormal cause-phenomenon category of each edge controller, and construct the delivery decision tree model for each edge controller based on all category decision tree models of each edge controller;

[0012] Control module: Based on the real-time edge anomaly data of each edge controller and the conveying decision tree model, the module determines and executes the real-time anomaly control strategy of each edge controller to realize the intelligent control of the air suspension conveyor.

[0013] Preferably, a digital intelligent air suspension conveyor control system includes a monitoring and acquisition module, comprising:

[0014] First monitoring unit: Based on the first sensor group installed on each device of the air suspension conveyor, it monitors the real-time equipment operation sub-data of each device, wherein the equipment operation sub-data includes the device name, multiple operating parameters and the operating parameter value of each operating parameter;

[0015] Real-time equipment operation data unit: Based on the equipment operation sub-data of all equipment, determine the real-time equipment operation data of the air suspension conveyor;

[0016] Real-time air pressure data unit: Based on the distributed pressure sensor group installed on the air suspension conveyor, it monitors the real-time air pressure data of the air suspension conveyor. The real-time air pressure data includes real-time air pressure sub-data from multiple air pressure monitoring points. The real-time air pressure sub-data includes the location of the air pressure monitoring point, multiple air pressure parameters, and the air pressure parameter value of each air pressure parameter.

[0017] The third monitoring unit: Based on the second sensor group of each airflow monitoring point, it monitors the real-time airflow sub-data of each airflow monitoring point. The real-time airflow sub-data includes the location of the airflow monitoring point, multiple airflow parameters, and the airflow parameter value of each airflow parameter.

[0018] Real-time airflow data unit: Based on the real-time airflow sub-data from all airflow monitoring points, determine the real-time airflow data of the air suspension conveyor.

[0019] Preferably, in a digital intelligent air-suspended conveyor control system, the monitoring and acquisition module further includes:

[0020] Historical Conveying Data Unit: Acquires historical conveying data of the air suspension conveyor based on multiple conveyings within a specified historical time period. The historical conveying data includes historical material data, historical air pressure data, historical airflow data, and historical anomaly data. The historical anomaly data includes multiple historical anomaly sub-data and the anomaly cause, anomaly feature vector, anomaly phenomenon, anomaly start time, anomaly end time, anomaly tag, and anomaly control strategy for each historical anomaly sub-data. The anomaly tag includes the equipment name, air pressure monitoring point location, and airflow monitoring point location.

[0021] Edge control set unit: acquires the edge control set of multiple edge controllers installed on the air suspension conveyor, wherein the edge control set includes multiple device names, multiple air pressure monitoring point locations, and multiple airflow monitoring point locations.

[0022] Preferably, a digital intelligent air suspension conveyor control system includes a range module comprising:

[0023] Equipment operating range data unit: acquires the equipment operating range data of the air suspension conveyor, wherein the equipment operating range data includes equipment operating range sub-data of multiple devices, and the equipment operating range sub-data includes multiple operating parameters and the normal operating range of each operating parameter;

[0024] First division unit: Based on whether the historical transmission data contains historical abnormal data, the historical transmission data of all transmissions is divided to determine normal transmission data and abnormal transmission data;

[0025] Material-pressure adaptive model unit: Based on historical material data and historical pressure data of all historical transport data in the normal transport data, the material-pressure adaptive model is trained.

[0026] Material-airflow adaptive model unit: Based on historical material data and historical airflow data of all historical conveying data in normal conveying data, the material-airflow adaptive model is trained;

[0027] Material conveying data unit: acquires material conveying data, which includes multiple material parameters and the material parameter value of each material parameter;

[0028] Air pressure range data unit: Input the conveyed material data into the material-air pressure adaptive model, and determine the air pressure range data based on the output of the material-air pressure adaptive model. The air pressure range data includes air pressure range sub-data of multiple air pressure monitoring points, and the air pressure range sub-data includes the normal air pressure range of each air pressure parameter.

[0029] Airflow range data unit: The conveyed material data is input into the material-airflow adaptive model, and the airflow range data is determined based on the output of the material-airflow adaptive model. The airflow range data includes airflow range sub-data of multiple airflow monitoring points, and the airflow range sub-data includes the normal airflow range of each airflow parameter.

[0030] Preferably, a digital intelligent air suspension conveyor control system includes a defined module comprising:

[0031] Real-time equipment anomaly vector unit: Based on the equipment operating range sub-data of each equipment name in the edge control set of each edge controller of the air suspension conveyor and the real-time equipment operating sub-data, determine the operating label and real-time equipment anomaly vector of each equipment name in the edge control set of each edge controller. The operating label includes normal operation and abnormal operation.

[0032] Real-time edge device abnormal data unit: Based on the real-time device abnormal vector of all device names with abnormal operation tags in the edge control set of each edge controller, the real-time edge device abnormal data of each edge controller is determined;

[0033] Real-time air pressure anomaly vector unit: Based on the air pressure range sub-data and real-time air pressure sub-data of each air pressure monitoring point location in the edge control set of each edge controller of the air suspension conveyor, determine the air pressure label and real-time air pressure anomaly vector of each air pressure monitoring point location in the edge control set of each edge controller. The air pressure label includes normal air pressure and abnormal air pressure.

[0034] Real-time edge pressure anomaly data unit: Based on the real-time pressure anomaly vector of all pressure monitoring points with pressure labels of pressure anomaly in the edge control set of each edge controller, the real-time edge pressure anomaly data of each edge controller is determined;

[0035] Real-time airflow anomaly vector unit: Based on the airflow range sub-data and real-time airflow sub-data of each airflow monitoring point location in the edge control set of each edge controller of the air suspension conveyor, determine the airflow label and real-time airflow anomaly vector of each airflow monitoring point location in the edge control set of each edge controller. The airflow label includes normal airflow and abnormal airflow.

[0036] Real-time edge airflow anomaly data unit: Based on the real-time airflow anomaly vector of the airflow monitoring point corresponding to the location of all airflow monitoring points with airflow labels of airflow anomaly in the edge control set of each edge controller, the real-time edge airflow anomaly data of each edge controller is determined;

[0037] Real-time edge anomaly data unit: Based on the real-time edge device anomaly data, real-time edge air pressure anomaly data, and real-time edge airflow anomaly data of each edge controller, determine the real-time edge anomaly data of each edge controller.

[0038] Preferably, an intelligent air-suspended conveyor control system includes an analysis module comprising:

[0039] Historical edge anomaly data unit: Based on the edge control set of each edge controller and the anomaly tags in the historical anomaly data of each transmission in the anomaly transmission data, the historical transmission data of each transmission in the anomaly transmission data is extracted to determine the historical edge anomaly data of each transmission of each edge controller. The historical edge anomaly data includes the historical anomaly sub-data of multiple device names corresponding to multiple device names belonging to the edge control set, multiple air pressure monitoring points corresponding to multiple air pressure monitoring point locations, and multiple airflow monitoring points corresponding to multiple airflow monitoring point locations.

[0040] Cause Category Anomaly Data Unit: Based on the anomaly causes in all historical anomaly sub-data in all historical edge anomaly data transmitted by each edge controller, the historical edge anomaly data transmitted by each edge controller is classified, and the cause category anomaly data and the anomaly cause label of each cause category anomaly data are determined for each edge controller. The cause category anomaly data includes multiple historical anomaly sub-data.

[0041] Cause-Phenomenon Category Anomaly Data Unit: Based on the anomalies in all historical anomaly sub-data of cause-phenomenon category anomaly data for each edge controller based on each anomaly cause, the cause-phenomenon category anomaly data for each edge controller is classified, and the cause-phenomenon category anomaly data and the anomaly cause-phenomenon category label for each cause-phenomenon category anomaly data for each edge controller are determined. The cause-phenomenon category anomaly data includes multiple historical anomaly sub-data.

[0042] Preferably, in an intelligent air-suspended conveyor control system, the analysis module further includes:

[0043] The second partitioning unit: divides each abnormal cause-phenomenon category abnormal data of each edge controller into training category abnormal data and test category abnormal data;

[0044] Category decision tree model unit: Based on the training category anomaly data for each anomaly cause-phenomenon category of each edge controller, train a category decision tree model for each anomaly cause-phenomenon category of each edge controller;

[0045] Building Unit: The abnormal feature vectors and abnormal labels of all historical abnormal sub-data in the training category abnormal data of each abnormal cause-phenomenon category of each edge controller are used as inputs to the category decision tree model. The abnormal start time, abnormal end time and abnormal control strategy of all historical abnormal sub-data in the training category abnormal data of each abnormal cause-phenomenon category of each edge controller are used as inputs to the category decision tree model.

[0046] Preferably, a digital intelligent air suspension conveyor control system includes the following modules:

[0047] Prediction Unit: Input the abnormal feature vector and abnormal label of each historical abnormal sub-data in the test category abnormal data of each abnormal cause-phenomenon category of each edge controller into the category decision tree model of each abnormal cause-phenomenon category. Based on the output of the category decision tree model, determine the predictive control strategy and predict the duration of the abnormality in each historical abnormal sub-data in the test category abnormal data of each abnormal cause-phenomenon category of each edge controller.

[0048] Simulation data unit: Based on all historical anomaly sub-data in the training category anomaly data of all anomaly causes-phenomenon categories of all edge controllers, the simulation data is determined, wherein the simulation data includes multiple historical anomaly sub-data;

[0049] Simulation Unit: Input the abnormal feature vector, abnormal start time, abnormal label, and predictive control strategy of each historical abnormal sub-data in the simulation data into the transport simulation model. Based on the transport simulation model, determine the simulation abnormal end time of each historical abnormal sub-data.

[0050] Model evaluation value unit: Based on the predicted control strategy, predicted anomaly duration, and simulation anomaly end time of all historical anomaly sub-data in the test category anomaly data of each anomaly cause-phenomenon category of each edge controller, as well as the anomaly control strategy, anomaly start time, and anomaly end time in all historical anomaly sub-data, calculate the model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller.

[0051] Optimization Unit: Compare the model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller with the preset evaluation value. If the model evaluation value is less than the preset evaluation value, optimize the category decision tree model based on the model evaluation value of each anomaly cause-phenomenon category of each edge controller, all historical anomaly sub-data in the training category anomaly data, the predictive control strategy of each historical anomaly sub-data, and the predicted anomaly duration.

[0052] Delivery decision tree model unit: The delivery decision tree model for each edge controller is constructed based on all optimized category decision tree models whose model evaluation values ​​are less than the preset evaluation values ​​and all category decision tree models whose model evaluation values ​​are greater than or equal to the preset evaluation values.

[0053] Preferably, a digital intelligent air suspension conveyor control system includes a control module comprising:

[0054] Real-time anomaly control strategy unit: Each real-time device anomaly vector, each real-time air pressure anomaly vector, and each real-time airflow anomaly vector from the real-time edge anomaly data of each edge controller are input into the delivery decision tree model of each edge controller to determine the real-time anomaly control strategy of each edge controller. The real-time anomaly control strategy includes multiple sub-strategies.

[0055] Intelligent control unit: Executes all sub-strategies in the real-time abnormal control strategy of all edge controllers to realize intelligent control of the air suspension conveyor.

[0056] The beneficial effects of this invention compared to existing technologies are as follows: By acquiring historical conveying data, normal conveying data, abnormal conveying data, air pressure range data, and airflow range data are determined. Combined with the edge control set of each edge controller, real-time equipment operation data, real-time air pressure data, and real-time airflow data, the real-time edge anomaly data, cause-phenomenon category anomaly data, and category decision tree models for each edge controller are determined. The model evaluation value of the category decision tree model for each cause-phenomenon category of anomalies for each edge controller is also determined. This constructs the conveying decision tree model for each edge controller, and determines and executes the real-time anomaly control strategy for each edge controller. This allows for real-time and accurate monitoring and classification of anomalies in air-suspended conveyors, enabling accurate prediction and intelligent processing of anomalies. This improves the stability and reliability of air-suspended conveyor operation, reduces equipment failure risks and maintenance costs, and simultaneously increases conveying efficiency, better adapting to complex industrial production environments.

[0057] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0058] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0059] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0060] Figure 1 This is a schematic diagram of a digital intelligent air suspension conveyor control system according to an embodiment of the present invention. Detailed Implementation

[0061] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0062] Example 1: This invention provides a digital intelligent air suspension conveyor control system, referencing... Figure 1 ,include:

[0063] Monitoring and Acquisition Module: Monitors real-time equipment operation data, real-time air pressure data, and real-time airflow data of the air suspension conveyor; acquires historical conveying data from multiple conveying operations; and acquires edge control sets from multiple edge controllers.

[0064] Range module: acquires equipment operating range data of air suspension conveyor, determines normal and abnormal conveying data based on historical conveying data, and determines air pressure range data and airflow range data based on normal conveying data;

[0065] Determination Module: Based on the edge control set, operating range data, air pressure range data, airflow range data, real-time equipment operation data, real-time air pressure data, and real-time airflow data of each edge controller, determine the real-time edge anomaly data of each edge controller;

[0066] Analysis module: Analyzes abnormal transmission data and the edge control set of each edge controller to determine the cause-phenomenon category of each abnormality for each edge controller and the category decision tree model;

[0067] Building Module: Based on abnormal delivery data, calculate the model evaluation value of the category decision tree model for each abnormal cause-phenomenon category of each edge controller, and construct the delivery decision tree model for each edge controller based on all category decision tree models of each edge controller;

[0068] Control module: Based on the real-time edge anomaly data of each edge controller and the conveying decision tree model, the module determines and executes the real-time anomaly control strategy of each edge controller to realize the intelligent control of the air suspension conveyor.

[0069] In this embodiment, the monitoring and acquisition module collects real-time equipment operation data, real-time air pressure data, real-time airflow data, historical conveying data from multiple conveying operations, and edge control sets from multiple edge controllers. This data forms the basis for subsequent analysis and control.

[0070] In this embodiment, the range module first acquires the equipment operating range data, then distinguishes between normal and abnormal transport data based on historical transport data, and finally determines the air pressure range data and airflow range data based on the normal transport data, providing a standard for judging the equipment operating status.

[0071] In this embodiment, the determining module combines the edge control set, operating range data, air pressure range data, airflow range data, and real-time device operation data, real-time air pressure data, and real-time airflow data of each edge controller to determine the real-time edge anomaly data of each edge controller, that is, to determine whether the current device operation is abnormal.

[0072] In this embodiment, the analysis module analyzes the abnormal transmission data and the edge control set of each edge controller to determine the cause-phenomenon category of abnormal data for each edge controller and establishes a category decision tree model.

[0073] In this embodiment, the construction module uses abnormal transmission data to calculate the model evaluation value of the category decision tree model for each abnormal cause-phenomenon category of each edge controller, and then constructs the transmission decision tree model for each edge controller based on all category decision tree models, providing a decision basis for subsequent control strategies.

[0074] In this embodiment, the control module determines and executes the real-time anomaly control strategy for each edge controller based on the real-time edge anomaly data of each edge controller and the transport decision tree model, thereby realizing intelligent control of the air suspension conveyor and enabling the equipment to be handled in a timely manner when anomalies occur.

[0075] The beneficial effects of the above technology are as follows: By acquiring historical conveying data, normal conveying data, abnormal conveying data, air pressure range data, and airflow range data are determined. Combined with the edge control set of each edge controller, real-time equipment operation data, real-time air pressure data, and real-time airflow data, the real-time edge anomaly data, the cause-phenomenon category anomaly data, and the category decision tree model for each anomaly cause-phenomenon category of each edge controller are determined. The model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller is also determined, constructing the conveying decision tree model for each edge controller. This allows for the determination and execution of real-time anomaly control strategies for each edge controller. This enables real-time and accurate monitoring and classification of anomalies in air-suspended conveyors, achieving accurate prediction and intelligent processing of anomalies, improving the stability and reliability of air-suspended conveyor operation, reducing equipment failure risks and maintenance costs, while simultaneously increasing conveying efficiency and better adapting to complex industrial production environments.

[0076] Example 2: Based on Example 1, a digital intelligent air suspension conveyor control system, including a monitoring and acquisition module, comprises:

[0077] First monitoring unit: Based on the first sensor group installed on each device of the air suspension conveyor, it monitors the real-time equipment operation sub-data of each device, wherein the equipment operation sub-data includes the device name, multiple operating parameters and the operating parameter value of each operating parameter;

[0078] Real-time equipment operation data unit: Based on the equipment operation sub-data of all equipment, determine the real-time equipment operation data of the air suspension conveyor;

[0079] Real-time air pressure data unit: Based on the distributed pressure sensor group installed on the air suspension conveyor, it monitors the real-time air pressure data of the air suspension conveyor. The real-time air pressure data includes real-time air pressure sub-data from multiple air pressure monitoring points. The real-time air pressure sub-data includes the location of the air pressure monitoring point, multiple air pressure parameters, and the air pressure parameter value of each air pressure parameter.

[0080] The third monitoring unit: Based on the second sensor group of each airflow monitoring point, it monitors the real-time airflow sub-data of each airflow monitoring point. The real-time airflow sub-data includes the location of the airflow monitoring point, multiple airflow parameters, and the airflow parameter value of each airflow parameter.

[0081] Real-time airflow data unit: Based on the real-time airflow sub-data from all airflow monitoring points, determine the real-time airflow data of the air suspension conveyor.

[0082] In this embodiment, all devices cover all the core hardware of the air suspension conveyor, including but not limited to drive motors such as conveyor motors and fan motors, high-pressure fans such as main fans and redundant fans, material carrying trays, airflow regulating valves, guide wheels, etc. Each device is an independent monitoring object to avoid the difficulty of abnormal positioning caused by the mixing of data from multiple devices.

[0083] In this embodiment, each device is equipped with a dedicated first sensor group. The type and number of sensors are matched according to the functional characteristics of the device. For example, a current sensor, a temperature sensor, and a vibration sensor are configured for the fan motor; a speed sensor and a position sensor are configured for the conveyor tray; and an opening sensor is configured for the airflow regulating valve.

[0084] In this embodiment, the monitoring data of each device is ultimately integrated into device operation sub-data, which contains three types of key information: First, the device name: such as main fan motor A, conveyor tray No. 1, clearly indicating the data ownership; second, multiple operating parameters: such as the current, temperature, and speed of the fan motor, and the moving speed and suspension height of the tray, covering the core operating dimensions of the device; and third, the operating parameter value of each operating parameter: such as current = 15A, temperature = 45℃, speed = 2m / s.

[0085] In this embodiment, a distributed deployment method is used to configure pressure sensors. That is, according to the airflow system structure of the air suspension conveyor, such as the air box, fan inlet / outlet, and pipeline, pressure sensors are installed at key air pressure monitoring locations, such as the four corners and center of the air box, the fan inlet / outlet, and the start / middle / end point of the pipeline. All these sensors together form a distributed pressure sensor group to ensure that air pressure monitoring covers the entire link from airflow generation to action, and that data from each monitoring location is collected independently.

[0086] In this embodiment, the real-time air pressure data is not a single value, but a collection of multiple real-time air pressure sub-data. Each real-time air pressure sub-data corresponds to a specific air pressure monitoring point and contains three types of information: First, the location of the air pressure monitoring point: such as the left front corner of the air box P001, the fan outlet P002, and the midpoint of the pipeline P003, which clearly defines the spatial location of the air pressure data; second, multiple air pressure parameters: such as static pressure, dynamic pressure, pressure difference with adjacent monitoring points, and deviation rate from the standard value, which describe the air pressure status at this location from different dimensions; and third, the air pressure parameter value of each air pressure parameter: such as static pressure = 800Pa, pressure difference = 50Pa, and deviation rate = -5%, which provides specific quantitative air pressure data.

[0087] In this embodiment, the location of the airflow monitoring points is determined by selecting key nodes of the airflow system, such as the fan outlet, the airflow outlet of the air box, the air hole area on the table, the airflow turning point of the pipeline, and the area above the material suspension area. Each location is used as an independent airflow monitoring point to ensure coverage of the entire process of airflow from generation and transportation to action on the material.

[0088] In this embodiment, a dedicated second sensor group is configured for each airflow monitoring point. The sensor types are matched according to the airflow characteristics of that monitoring point: for example, the second sensor group at the fan outlet includes an airflow sensor, an air velocity sensor, and an airflow temperature sensor; the second sensor group in the air box platform vent area includes an air velocity sensor and a turbulence sensor; and the second sensor group at the pipe bend includes a wind direction sensor and an air velocity sensor. The second sensor group at each monitoring point only collects airflow data for that point, avoiding cross-point interference.

[0089] In this embodiment, the monitoring data of each airflow monitoring point forms real-time airflow sub-data, which includes three types of core information: First, the location of the airflow monitoring point: such as the fan outlet V001, the air box platform V002, and the pipeline turning point V003, which clearly defines the spatial ownership of the airflow data; Second, multiple airflow parameters: such as wind speed, air volume, airflow temperature, turbulence intensity, and flow direction, which completely describe the airflow state from dimensions such as speed, flow rate, temperature, stability, and direction; Third, the airflow parameter value of each airflow parameter: such as wind speed = 3m / s, air volume = 500m³ / h, temperature = 25℃, turbulence intensity = 8%, and flow direction = vertically upward, which realizes the quantitative monitoring of the airflow state.

[0090] The beneficial effects of the above technologies are: monitoring real-time equipment operation data, real-time air pressure data, and real-time airflow data of the air suspension conveyor can adapt to different monitoring needs of equipment, air pressure, and airflow, improve the targeting and accuracy of data collection, and provide data support for determining real-time edge anomaly data of each edge controller.

[0091] Example 3: Based on Example 1, a digital intelligent air suspension conveyor control system, including a monitoring and acquisition module, further includes:

[0092] Historical Conveying Data Unit: Acquires historical conveying data of the air suspension conveyor based on multiple conveyings within a specified historical time period. The historical conveying data includes historical material data, historical air pressure data, historical airflow data, and historical anomaly data. The historical anomaly data includes multiple historical anomaly sub-data and the anomaly cause, anomaly feature vector, anomaly phenomenon, anomaly start time, anomaly end time, anomaly tag, and anomaly control strategy for each historical anomaly sub-data. The anomaly tag includes the equipment name, air pressure monitoring point location, and airflow monitoring point location.

[0093] Edge control set unit: acquires the edge control set of multiple edge controllers installed on the air suspension conveyor, wherein the edge control set includes multiple device names, multiple air pressure monitoring point locations, and multiple airflow monitoring point locations.

[0094] In this embodiment, the specified historical time period can be set according to actual needs: such as the past 3 months, to ensure that the data is timely and representative.

[0095] In this embodiment, historical material data records the key attributes of the material in each historical conveying operation, such as material type (e.g., powder, granules, pallet carriers), material specific gravity (e.g., 1 kg / m³, 500 kg / m³), single conveying volume (e.g., 200 tons / hour, 500 kg / batch), and material temperature / humidity (e.g., if there are special requirements for conveying).

[0096] In this embodiment, the historical air pressure data corresponds to the historical record of the real-time air pressure data, including the air pressure parameters of each air pressure monitoring point during the historical transportation process: such as the static pressure at each point of the air box, the pressure difference between the inlet and outlet of the fan, and the air pressure in the pipeline before, during, and after transportation, and must be associated with the specific location of the air pressure monitoring point: such as the static pressure of the front left corner P001 of the air box at 10:00 is 850Pa.

[0097] In this embodiment, the historical airflow data corresponds to the historical record of the real-time airflow data, including the airflow parameters of each airflow monitoring point during the historical transportation process: such as the airflow of the fan outlet, the air velocity of the air box vent, the airflow temperature, the turbulence, etc. It also needs to be associated with the specific location of the airflow monitoring point: such as the airflow of the fan outlet V001 at 10:05 is 520m³ / h.

[0098] In this embodiment, the historical anomaly data is a complete archive of anomalies that occurred in the past, containing multiple historical anomaly sub-data. That is, each anomaly corresponds to an independent sub-data, and each sub-data needs to cover the key information of the entire life cycle of the anomaly, as follows: anomaly cause: such as fan filter blockage, pipeline leakage, sensor false alarm; anomaly phenomenon: such as material deviation, decrease in conveying speed, sudden drop in air pressure; anomaly start time; anomaly end time.

[0099] In this embodiment, the abnormal feature vector represents the feature set of key parameters when an abnormality occurs: such as the static pressure at point P001 dropping to 500Pa, the wind speed at point V001 dropping to 1m / s, and the motor current rising to 20A when an abnormality occurs. The abnormal state is described by quantitative parameters.

[0100] In this embodiment, the anomaly label marks the specific location and equipment associated with the anomaly, including equipment name: such as main fan A, conveyor motor B, air pressure monitoring point location: such as the left front corner of air box P001, airflow monitoring point location: such as the air outlet of fan V001, so as to clarify the source of the anomaly and avoid ambiguous anomaly location.

[0101] In this embodiment, the abnormal control strategy records the control measures executed when the abnormality occurs, such as switching redundant fans, adjusting the opening of the airflow regulating valve, and suspending delivery. The control effect is such that the air pressure returns to normal 10 minutes after execution.

[0102] In this embodiment, the intelligent control system of the air suspension conveyor is typically configured with multiple edge controllers: one edge controller is responsible for the air source system control, one for the material conveying control, and one for the safety protection control. The edge control set is that each edge controller has its own exclusive control range list, rather than all edge controllers sharing a unified set, to ensure the uniqueness and specificity of control responsibilities.

[0103] In this embodiment, each edge controller's edge control set includes three types of managed objects: devices, air pressure monitoring points, and airflow monitoring points. This clarifies the hardware devices operable by the edge controller and the monitoring / controllable points, specifically as follows: Multiple device names: Lists all devices controlled by the edge controller. For example, an edge controller responsible for the air source system includes device names such as: main fan A, redundant fan B, and airflow regulating valve C—clearly defining the hardware scope for which the edge controller can perform control actions. Multiple air pressure monitoring point locations: Lists the air pressure monitoring points monitored and associated with the edge controller. For example, an edge controller responsible for the air source system includes air pressure monitoring point locations such as: fan outlet P002 and air box center P003—clearly defining the air pressure data sources that the edge controller needs to monitor. Multiple airflow monitoring point locations: Lists the airflow monitoring points monitored and associated with the edge controller. For example, an edge controller responsible for the air source system includes airflow monitoring point locations such as: fan outlet V001 and pipeline midpoint V004—clearly defining the airflow data sources that the edge controller needs to monitor.

[0104] The beneficial effects of the above technologies are: acquiring historical transmission data from multiple transmissions, acquiring edge control sets from multiple edge controllers, clarifying the control scope of edge controllers, avoiding response delays caused by ambiguous control responsibilities, and providing data support for determining real-time edge anomaly data for each edge controller.

[0105] Example 4: Based on Example 1, a digital intelligent air suspension conveyor control system, including a range module, comprises:

[0106] Equipment operating range data unit: acquires the equipment operating range data of the air suspension conveyor, wherein the equipment operating range data includes equipment operating range sub-data of multiple devices, and the equipment operating range sub-data includes multiple operating parameters and the normal operating range of each operating parameter;

[0107] First division unit: Based on whether the historical transmission data contains historical abnormal data, the historical transmission data of all transmissions is divided to determine normal transmission data and abnormal transmission data;

[0108] Material-pressure adaptive model unit: Based on historical material data and historical pressure data of all historical transport data in the normal transport data, the material-pressure adaptive model is trained.

[0109] Material-airflow adaptive model unit: Based on historical material data and historical airflow data of all historical conveying data in normal conveying data, the material-airflow adaptive model is trained;

[0110] Material conveying data unit: acquires material conveying data, which includes multiple material parameters and the material parameter value of each material parameter;

[0111] Air pressure range data unit: Input the conveyed material data into the material-air pressure adaptive model, and determine the air pressure range data based on the output of the material-air pressure adaptive model. The air pressure range data includes air pressure range sub-data of multiple air pressure monitoring points, and the air pressure range sub-data includes the normal air pressure range of each air pressure parameter.

[0112] Airflow range data unit: The conveyed material data is input into the material-airflow adaptive model, and the airflow range data is determined based on the output of the material-airflow adaptive model. The airflow range data includes airflow range sub-data of multiple airflow monitoring points, and the airflow range sub-data includes the normal airflow range of each airflow parameter.

[0113] In this embodiment, the equipment operating range data of the air suspension conveyor is composed of sub-data of the equipment operating range of multiple devices. Each sub-data corresponds to an independent device and includes multiple operating parameters of that device and the normal operating range of each parameter. For example, for the main fan motor, its operating parameters may include motor current, motor temperature, and motor speed, with corresponding normal operating ranges of 10 to 18 amps, 30 to 50 degrees Celsius, and 1500 to 2000 revolutions per minute, respectively. For the conveyor tray, its operating parameters may include moving speed and suspension height, with corresponding normal operating ranges of 0.5 to 3 meters per second and 0.2 to 0.5 millimeters, respectively.

[0114] In this embodiment, two categories of data, normal and abnormal, are separated from historical transmission data. The classification is based on whether the historical transmission data contains historical abnormal data. If the historical transmission data of a certain transmission does not contain any historical abnormal sub-data, that is, no abnormal cause, abnormal phenomenon, abnormal feature vector, or other abnormal related information occurred during that transmission, then the historical transmission data of that transmission is classified as normal transmission data. If the historical transmission data of a certain transmission contains at least one historical abnormal sub-data, that is, an abnormality occurred during that transmission, regardless of the duration or impact of the abnormality, the historical transmission data of that transmission is classified as abnormal transmission data.

[0115] In this embodiment, the data source for training the material-pressure adaptive model is the historical material data and historical pressure data from all normal delivery data. The historical material data includes material parameters such as material type, specific gravity, and single delivery volume for each normal delivery; the historical pressure data includes pressure parameters such as static pressure, dynamic pressure, and differential pressure at each pressure monitoring point during each normal delivery, along with their corresponding values. The training logic is to uncover the inherent correlation between material parameters and pressure parameters. For example, when the specific gravity of the material increases, the static pressure at each monitoring point within the air box needs to be increased accordingly to maintain stable material suspension; when the single delivery volume increases, the pressure difference between the inlet and outlet of the blower needs to be maintained at a higher level to ensure sufficient air supply. The trained material-pressure adaptive model has the ability to: input new material parameters, and based on the normal pressure parameter range corresponding to similar material parameters in history, output a reasonable range of pressure parameters suitable for the current material, achieving dynamic matching between the pressure range and material characteristics.

[0116] In this embodiment, the data source for training the material-airflow adaptive model is historical material data and historical airflow data from all normal conveying data. Historical material data includes parameters such as material type, specific gravity, and single conveying volume; historical airflow data includes airflow parameters and corresponding values ​​such as wind speed, airflow volume, airflow temperature, and turbulence at each airflow monitoring point during each normal conveying. The training process focuses on analyzing the impact of changes in material parameters on airflow parameters. For example, when conveying lightweight and easily deflected materials, the wind speed at the airflow monitoring points needs to be controlled at a low range to prevent the material from being blown off course; when conveying granular materials, the turbulence of the airflow needs to be kept at a low level to prevent material splashing. The resulting material-airflow adaptive model can output the normal range of airflow parameters adapted to the input material parameters. For example, if the input material has a specific gravity of 0.8 kg / m³ and a single conveying volume of 100 kg per batch, the model can output specific ranges such as the wind speed at each airflow monitoring point should be between 2 and 3 m / s and the airflow volume should be between 400 and 500 m / h.

[0117] In this embodiment, detailed parameters of the materials in the upcoming conveying task are obtained, including multiple material parameters and the material parameter value for each. The types of material parameters are consistent with the parameter types in historical material data to ensure compatibility with the previously trained adaptive model. For example, if the material being conveyed is PCB board, the material parameters in the conveying material data may include material type as flat material, material specific gravity as 1.2 kg / m², single conveying quantity as 50 pieces per batch, and material thickness as 2 mm; if the material being conveyed is granular material, the material parameters may include material type as granular material, material specific gravity as 1500 kg / m³, single conveying quantity as 200 kg per batch, and material particle diameter as 0.5 to 1 mm.

[0118] In this embodiment, the material delivery data is input into a pre-trained material pressure adaptive model. Based on the input material parameters, the model retrieves historical normal pressure data for similar materials and calculates the appropriate pressure parameter range for the current material using internal correlation logic. The output is pressure range data, which consists of sub-data points representing the pressure ranges of multiple pressure monitoring points. Each sub-data point corresponds to a specific pressure monitoring point and includes the normal pressure range for each pressure parameter at that monitoring point. For example, for the monitoring point at the left front corner of the air box, its pressure range sub-data may include a normal static pressure range of 750 to 850 Pascals, a normal dynamic pressure range of 150 to 200 Pascals, and a normal pressure range of 0 to 50 Pascals for the pressure difference with adjacent monitoring points; for the fan outlet monitoring point, its pressure range sub-data may include a normal static pressure range of 1200 to 1300 Pascals and a normal pressure range of 800 to 900 Pascals for the pressure difference with the inlet.

[0119] In this embodiment, the material transport data is input into a trained material flow adaptive model. Based on the input material parameters and combined with historical normal flow data for the same or similar materials, the model calculates a suitable range of flow parameters for the current material transport. The final output flow range data consists of sub-data from multiple flow monitoring points. Each sub-data point corresponds to one flow monitoring point and includes the normal flow range for each flow parameter at that monitoring point. For example, for the fan outlet monitoring point, the flow range sub-data might include a normal flow range of 3 to 4 meters per second for wind speed, 450 to 550 cubic meters per hour for air volume, and 25 to 30 degrees Celsius for air temperature; for the air box platform monitoring point, the flow range sub-data might include a normal flow range of 2 to 3 meters per second for wind speed and 5% to 10% for turbulence intensity. These ranges will serve as the criteria for determining whether the flow parameters are normal in the current transport task.

[0120] The beneficial effects of the above technologies are as follows: obtaining equipment operating range data of the air suspension conveyor, determining normal and abnormal conveying data based on historical conveying data, and determining air pressure range data and airflow range data based on normal conveying data, which can make the air pressure range data and airflow range data fit the current material characteristics and avoid instability caused by general range.

[0121] Example 5: Based on Example 1, a digital intelligent air suspension conveyor control system, comprising a determination module, including:

[0122] Real-time equipment anomaly vector unit: Based on the equipment operating range sub-data of each equipment name in the edge control set of each edge controller of the air suspension conveyor and the real-time equipment operating sub-data, determine the operating label and real-time equipment anomaly vector of each equipment name in the edge control set of each edge controller. The operating label includes normal operation and abnormal operation.

[0123] Real-time edge device abnormal data unit: Based on the real-time device abnormal vector of all device names with abnormal operation tags in the edge control set of each edge controller, the real-time edge device abnormal data of each edge controller is determined;

[0124] Real-time air pressure anomaly vector unit: Based on the air pressure range sub-data and real-time air pressure sub-data of each air pressure monitoring point location in the edge control set of each edge controller of the air suspension conveyor, determine the air pressure label and real-time air pressure anomaly vector of each air pressure monitoring point location in the edge control set of each edge controller. The air pressure label includes normal air pressure and abnormal air pressure.

[0125] Real-time edge pressure anomaly data unit: Based on the real-time pressure anomaly vector of all pressure monitoring points with pressure labels of pressure anomaly in the edge control set of each edge controller, the real-time edge pressure anomaly data of each edge controller is determined;

[0126] Real-time airflow anomaly vector unit: Based on the airflow range sub-data and real-time airflow sub-data of each airflow monitoring point location in the edge control set of each edge controller of the air suspension conveyor, determine the airflow label and real-time airflow anomaly vector of each airflow monitoring point location in the edge control set of each edge controller. The airflow label includes normal airflow and abnormal airflow.

[0127] Real-time edge airflow anomaly data unit: Based on the real-time airflow anomaly vector of the airflow monitoring point corresponding to the location of all airflow monitoring points with airflow labels of airflow anomaly in the edge control set of each edge controller, the real-time edge airflow anomaly data of each edge controller is determined;

[0128] Real-time edge anomaly data unit: Based on the real-time edge device anomaly data, real-time edge air pressure anomaly data, and real-time edge airflow anomaly data of each edge controller, determine the real-time edge anomaly data of each edge controller.

[0129] In this embodiment, the device operation range sub-data and real-time device operation sub-data of each device name in the edge control set of each edge controller of the air suspension conveyor are used to make a judgment. If the operation parameter values ​​of all operation parameters in the real-time device operation sub-data of each device are within the normal operation range of the corresponding operation parameters in the device operation range sub-data, the operation tag of the device name is determined to be normal operation. If the operation parameter value of any one operation parameter in the real-time device operation sub-data of each device is not within the normal operation range of the corresponding operation parameters in the device operation range sub-data, the operation tag of the device name is determined to be abnormal operation. The real-time device abnormal vector of the device name is determined based on the operation parameter values ​​of all operation parameters that are not in the operation range sub-data. The real-time device abnormal vector includes the operation parameter values ​​of multiple operation parameters.

[0130] In this embodiment, the filtering object is all devices in the edge control set of each edge controller, and only the devices corresponding to the device names with the running label of abnormal operation are retained. The real-time device abnormal vectors of these abnormal devices are collected to determine the real-time edge device abnormal data of the edge controller.

[0131] In this embodiment, the air pressure range sub-data and real-time air pressure sub-data of each air pressure monitoring point corresponding to the edge control set of each edge controller of the air suspension conveyor are used to make a judgment. If the air pressure parameter values ​​of all air pressure parameters in the real-time air pressure sub-data of each air pressure monitoring point are within the normal air pressure range of the corresponding air pressure parameter in the air pressure range sub-data, the air pressure label of the air pressure monitoring point corresponding to the air pressure monitoring point is determined to be normal air pressure. If the air pressure parameter value of any air pressure parameter in the real-time air pressure sub-data of each air pressure monitoring point is not within the normal air pressure range of the corresponding air pressure parameter in the air pressure range sub-data, the air pressure label of the air pressure monitoring point corresponding to the air pressure monitoring point is determined to be abnormal air pressure. The real-time air pressure abnormality vector of the air pressure monitoring point corresponding to the air pressure monitoring point is determined based on the air pressure parameter values ​​of all air pressure parameters that are not in the air pressure range sub-data of the air pressure monitoring point. The real-time air pressure abnormality vector includes the air pressure parameter values ​​of multiple air pressure parameters.

[0132] In this embodiment, from all the air pressure monitoring points in the edge controller's edge control set, the monitoring points corresponding to the locations of the monitoring points with air pressure labels of air pressure anomalies are selected, and the real-time air pressure anomaly vectors of these abnormal monitoring points are collected to determine the real-time edge air pressure anomaly data of the edge controller.

[0133] In this embodiment, the airflow range sub-data and real-time airflow sub-data of each airflow monitoring point corresponding to the edge control set of each edge controller of the air suspension conveyor are used to make a judgment. If the airflow parameter values ​​of all airflow parameters in the real-time airflow sub-data of each airflow monitoring point are within the normal airflow range of the corresponding airflow parameters in the airflow range sub-data, the airflow label of the airflow monitoring point corresponding to the airflow monitoring point is determined to be normal airflow. If the airflow parameter value of any airflow parameter in the real-time airflow sub-data of each airflow monitoring point is not within the normal airflow range of the corresponding airflow sub-data, the airflow label of the airflow monitoring point corresponding to the airflow monitoring point is determined to be abnormal airflow. The real-time airflow anomaly vector of the airflow monitoring point corresponding to the airflow monitoring point is determined based on the airflow parameter values ​​of all airflow parameters that are not in the airflow range sub-data of the airflow monitoring point. The real-time airflow anomaly vector includes the airflow parameter values ​​of multiple airflow parameters.

[0134] In this embodiment, from all airflow monitoring points in the edge controller's edge control set, the monitoring points corresponding to the locations of the monitoring points with airflow labels indicating airflow anomalies are selected, and the real-time airflow anomaly vectors of these anomaly monitoring points are collected to determine the real-time edge airflow anomaly data of the edge controller.

[0135] The beneficial effects of the above technologies are as follows: Based on the edge control set, operating range data, air pressure range data, airflow range data, real-time equipment operation data, real-time air pressure data, and real-time airflow data of each edge controller, the real-time edge anomaly data of each edge controller can be determined, which can avoid cross-controller interference and improve the targeting of anomaly identification.

[0136] Example 6: Based on Example 5, a digital intelligent air suspension conveyor control system, including an analysis module, comprises:

[0137] Historical edge anomaly data unit: Based on the edge control set of each edge controller and the anomaly tags in the historical anomaly data of each transmission in the anomaly transmission data, the historical transmission data of each transmission in the anomaly transmission data is extracted to determine the historical edge anomaly data of each transmission of each edge controller. The historical edge anomaly data includes the historical anomaly sub-data of multiple device names corresponding to multiple device names belonging to the edge control set, multiple air pressure monitoring points corresponding to multiple air pressure monitoring point locations, and multiple airflow monitoring points corresponding to multiple airflow monitoring point locations.

[0138] Cause Category Anomaly Data Unit: Based on the anomaly causes in all historical anomaly sub-data in all historical edge anomaly data transmitted by each edge controller, the historical edge anomaly data transmitted by each edge controller is classified, and the cause category anomaly data and the anomaly cause label of each cause category anomaly data are determined for each edge controller. The cause category anomaly data includes multiple historical anomaly sub-data.

[0139] Cause-Phenomenon Category Anomaly Data Unit: Based on the anomalies in all historical anomaly sub-data of cause-phenomenon category anomaly data for each edge controller based on each anomaly cause, the cause-phenomenon category anomaly data for each edge controller is classified, and the cause-phenomenon category anomaly data and the anomaly cause-phenomenon category label for each cause-phenomenon category anomaly data for each edge controller are determined. The cause-phenomenon category anomaly data includes multiple historical anomaly sub-data.

[0140] In this embodiment, historical anomaly information related to the control range of each edge controller is filtered from the overall abnormal transmission data. Data extraction is based on two key dimensions: first, the edge control set of each edge controller, which specifies the name of the equipment, the location of the air pressure monitoring point, and the location of the airflow monitoring point under the controller's responsibility; second, the anomaly tags in the historical anomaly data of each transmission, which indicate the associated equipment name, air pressure monitoring point location, and airflow monitoring point location. The extraction process is carried out for each transmission in the abnormal transmission data, checking the anomaly tags of the historical anomaly data for that transmission one by one. If the equipment name, air pressure monitoring point location, or airflow monitoring point location in the anomaly tag belongs to the edge control set of a certain edge controller, the corresponding historical anomaly sub-data is extracted and classified as historical anomaly data for that edge controller. Ultimately, the historical edge anomaly data for each transmission by each edge controller contains only content related to the controller's control range.

[0141] In this embodiment, historical edge anomaly data for each edge controller is categorized using the core dimension of anomaly cause. The categorization is based on the anomaly causes contained in all historical anomaly sub-data within all historical edge anomaly data transmitted by each edge controller. These anomaly causes may include different types such as fan filter blockage, motor overload, pipeline leakage, and pressure sensor drift. The categorization process revolves around each edge controller. First, all historical anomaly sub-data from all historical edge anomaly data for that controller are collected. Then, these sub-data are grouped according to different anomaly causes. Historical anomaly sub-data with the same anomaly cause are grouped together, forming a cause category anomaly data. Simultaneously, each cause category anomaly data is assigned an anomaly cause label, with the label content consistent with the anomaly cause of that category of data. For example, all historical anomaly sub-data with the anomaly cause of fan filter blockage are grouped into one category, and the anomaly cause label for this category of data is "fan filter blockage." Each edge controller corresponds to multiple cause category anomaly data, and each dataset contains multiple historical anomaly sub-data with the same anomaly cause.

[0142] In this embodiment, the abnormal data of cause categories are further classified according to the abnormal phenomena. The classification is based on the abnormal phenomena contained in all historical abnormal sub-data of the cause category abnormal data of each edge controller for each abnormal cause. These abnormal phenomena may include different manifestations such as material deviation, decrease in conveying speed, sudden drop in static pressure of air box, and increased fan vibration. The classification process is carried out for each cause category abnormal data of each edge controller. First, a cause category abnormal data is selected. The historical abnormal sub-data in this data set have the same abnormal cause. Then, the abnormal phenomena of each of these sub-data are examined, and they are further grouped according to the different abnormal phenomena. The historical abnormal sub-data with the same abnormal phenomenon are grouped into one category to form a cause phenomenon category abnormal data. Each cause phenomenon category abnormal data is assigned an abnormal cause phenomenon category label. The label content contains both the abnormal cause and the abnormal phenomenon of the data in that category. For example, if the abnormal cause of a cause category abnormal data is fan filter blockage, and the abnormal phenomenon of some sub-data is sudden drop in static pressure of air box, then these sub-data are grouped into one category, and their abnormal cause phenomenon category label is "fan filter blockage and sudden drop in static pressure of air box". Each cause category of abnormal data can be broken down into multiple cause phenomenon categories of abnormal data. Each dataset contains multiple historical abnormal sub-data with the same abnormal cause and abnormal phenomenon.

[0143] The beneficial effects of the above technologies are as follows: by analyzing abnormal transmission data and the edge control set of each edge controller, the cause-phenomenon category of each abnormality data of each edge controller can be determined, which can avoid irrelevant abnormal interference, ensure that the data is accurately matched with the controller's responsibilities, and improve the efficiency of abnormal data management.

[0144] Example 7: Based on Example 6, a digital intelligent air suspension conveyor control system, including an analysis module, further includes:

[0145] The second partitioning unit: divides each abnormal cause-phenomenon category abnormal data of each edge controller into training category abnormal data and test category abnormal data;

[0146] Category decision tree model unit: Based on the training category anomaly data for each anomaly cause-phenomenon category of each edge controller, train a category decision tree model for each anomaly cause-phenomenon category of each edge controller;

[0147] Building Unit: The abnormal feature vectors and abnormal labels of all historical abnormal sub-data in the training category abnormal data of each abnormal cause-phenomenon category of each edge controller are used as inputs to the category decision tree model. The abnormal start time, abnormal end time and abnormal control strategy of all historical abnormal sub-data in the training category abnormal data of each abnormal cause-phenomenon category of each edge controller are used as inputs to the category decision tree model.

[0148] In this embodiment, for each abnormal cause phenomenon category data of each edge controller, independent samples are allocated for model training and performance verification. The partitioning logic is to separate the historical abnormal sub-data in the cause phenomenon category abnormal data according to a preset ratio or a random method. For example, it can be divided in a 7:3 ratio, with 70% of the historical abnormal sub-data as training category abnormal data for subsequent model parameter learning; and 30% of the historical abnormal sub-data as test category abnormal data to verify the recognition accuracy and reliability of the model after training.

[0149] In this embodiment, for each abnormal phenomenon category of each edge controller, a decision tree model capable of identifying that type of abnormality is constructed using training data. The training logic is based on the decision tree algorithm, extracting the association rules between abnormal features and abnormal categories from the training category abnormal data. For example, if a training set corresponds to the category "fan filter blockage and sudden drop in static pressure in the air box", the model will learn the correspondence between the abnormal feature vectors under this category, such as static pressure value below the threshold and fan current increase, and the abnormal category of fan filter blockage and sudden drop in static pressure in the air box, thus constructing multi-layer decision branches.

[0150] The beneficial effects of the above technologies are: determining the category decision tree model for each abnormal cause-phenomenon category of each edge controller can improve the model's relevance while ensuring the accuracy of single-category models.

[0151] Example 8: Based on Example 7, a digital intelligent air suspension conveyor control system, comprising the following modules:

[0152] Prediction Unit: Input the abnormal feature vector and abnormal label of each historical abnormal sub-data in the test category abnormal data of each abnormal cause-phenomenon category of each edge controller into the category decision tree model of each abnormal cause-phenomenon category. Based on the output of the category decision tree model, determine the predictive control strategy and predict the duration of the abnormality in each historical abnormal sub-data in the test category abnormal data of each abnormal cause-phenomenon category of each edge controller.

[0153] Simulation data unit: Based on all historical anomaly sub-data in the training category anomaly data of all anomaly causes-phenomenon categories of all edge controllers, the simulation data is determined, wherein the simulation data includes multiple historical anomaly sub-data;

[0154] Simulation Unit: Input the abnormal feature vector, abnormal start time, abnormal label, and predictive control strategy of each historical abnormal sub-data in the simulation data into the transport simulation model. Based on the transport simulation model, determine the simulation abnormal end time of each historical abnormal sub-data.

[0155] Model evaluation value unit: Based on the predicted control strategy, predicted anomaly duration, and simulation anomaly end time of all historical anomaly sub-data in the test category anomaly data of each anomaly cause-phenomenon category of each edge controller, as well as the anomaly control strategy, anomaly start time, and anomaly end time in all historical anomaly sub-data, calculate the model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller.

[0156] Optimization Unit: Compare the model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller with the preset evaluation value. If the model evaluation value is less than the preset evaluation value, optimize the category decision tree model based on the model evaluation value of each anomaly cause-phenomenon category of each edge controller, all historical anomaly sub-data in the training category anomaly data, the predictive control strategy of each historical anomaly sub-data, and the predicted anomaly duration.

[0157] Delivery decision tree model unit: The delivery decision tree model for each edge controller is constructed based on all optimized category decision tree models whose model evaluation values ​​are less than the preset evaluation values ​​and all category decision tree models whose model evaluation values ​​are greater than or equal to the preset evaluation values.

[0158] In this embodiment, a pre-trained category decision tree model is used to predict anomalous sub-data in the test data. Based on the learned anomalous features and processing rules, the model outputs two key results: first, a predicted control strategy, which is the control measures the model determines should be implemented when this type of anomalous event occurs. For example, for a fan filter blockage anomalous event, the model outputs a strategy of switching redundant fans and cleaning the filter; second, a predicted duration of the anomalous event, which is the model estimates the time required from the onset to the end of the anomalous event, for example, predicting that this type of anomalous event will last 12 minutes. Each historical anomalous sub-data in the test category of anomalous data corresponds to a predicted control strategy and a predicted duration of the anomalous event. These prediction results strictly correspond to the anomalous cause-phenomenon category to which the sub-data belongs, ensuring the relevance of the predictions.

[0159] In this embodiment, historical anomaly sub-data that can be used for simulation verification is extracted from the training data of all edge controllers to form a simulation data set. Simulation data is formed based on all historical anomaly sub-data in the training category anomaly data of all anomaly causes-phenomenon categories of all edge controllers.

[0160] In this embodiment, the simulation input consists of two parts: first, the information of each historical anomaly sub-data in the simulation data, namely, the anomaly feature vector, the anomaly start time, and the anomaly label; second, the predictive control strategy obtained by the predictive unit from the historical anomaly sub-data. These two parts of data are integrated into the model according to the format required by the simulation model. The simulation process involves the simulation model simulating the system state at the time of the anomaly based on the input anomaly feature vector and anomaly start time, then simulating the execution of control actions according to the predictive control strategy, calculating system parameter changes in real time, until the model determines that the anomaly state has been eliminated. The recorded time at this point is the simulation anomaly end time. Each historical anomaly sub-data input to the simulation model will output a corresponding simulation anomaly end time. This time reflects the time it takes for the anomaly to be resolved after the predictive control strategy is executed in the simulation environment, and can be used to compare with the actual anomaly end time to verify the effectiveness of the predictive control strategy.

[0161] In this embodiment, based on the predictive control strategy, predicted anomaly duration, and simulation anomaly end time of all historical anomaly sub-data in the test category anomaly data of each anomaly cause-phenomenon category for each edge controller, as well as the anomaly control strategy, anomaly start time, and anomaly end time in all historical anomaly sub-data, the model evaluation value of the category decision tree model for each anomaly cause-phenomenon category of each edge controller is calculated. The formula for calculating the model evaluation value can be expressed as:

[0162] ;

[0163] in, β1 represents the model evaluation value of the category decision tree model for the j-th anomaly cause-phenomenon category of the i-th edge controller, and β2 represents the policy weight. , These represent the first and second model sub-evaluation values, respectively, for the j-th anomaly cause-phenomenon category of the i-th edge controller. , These represent the first indicator function and the second exponential function, respectively. This represents the best prediction ratio for the j-th cause of an anomaly to the phenomenon category of the i-th edge controller. This represents the actual duration of the historical anomaly sub-data in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the start time of the anomaly in the k-th historical anomaly subdata within the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the end time of the anomaly in the k-th historical anomaly sub-data within the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the simulation anomaly end time of the k-th historical anomaly sub-data in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the duration of the simulated anomaly in the k-th historical anomaly subdata of the test category anomaly data for the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the predicted duration of the k-th historical anomaly subdata in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the predictive control strategy for the k-th historical anomaly subdata in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller. Let ijN1 represent the anomaly control strategy in the k-th historical anomaly sub-data in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller, and let ijN1 represent the number of historical anomaly sub-data in the test category anomaly data of the j-th anomaly cause-phenomenon category of the i-th edge controller.

[0164] In this embodiment, the first model sub-evaluation value This represents the error evaluation value of all historical anomaly subdata in the test category anomaly data where the predicted control strategy and the anomaly control strategy are the same as those in the test category anomaly data for the j-th anomaly cause-phenomenon category of the i-th edge controller.

[0165] In this embodiment, the second model sub-evaluation value This represents the error evaluation value of all historical anomaly subdata in the test category anomaly data where the predicted control strategy and the anomaly control strategy are different from those of the i-th edge controller for the j-th anomaly cause-phenomenon category.

[0166] In this embodiment, the best prediction ratio of the j-th anomaly cause-phenomenon category of the i-th edge controller. This represents the percentage of times the new strategy recommended by the model proves to be a better strategy in all cases where the strategy predicted by the model is inconsistent with the historical actual strategy.

[0167] In this embodiment, the error weight β1 can be 0.7 and the strategy weight β2 can be 0.3.

[0168] In this embodiment, the optimization trigger condition is to compare the model evaluation value of each category decision tree model with the preset evaluation value. If the model evaluation value is less than the preset evaluation value, it indicates that the model performance has not met the expected requirements, and the optimization process needs to be initiated; if the evaluation value is greater than or equal to the preset evaluation value, the model does not need optimization and is directly retained. The optimization basis includes multiple aspects of information: first, the model evaluation value corresponding to the model, which clarifies the dimensions in which the model has deficiencies: such as large control strategy prediction bias and low time prediction accuracy; second, all historical anomaly sub-data in the training category anomaly data of the model; and third, the predicted control strategy and predicted anomaly duration output by the model, which can be used to adjust the model's decision logic for these output results. The optimization process involves adjusting the structure and parameters of the category decision tree model based on the above basis. For example, if the model has a large deviation in predicting the time of a certain type of anomaly, the weight of the anomaly duration-related features in the training data can be increased; if the control strategy prediction is inaccurate, the association learning between the anomaly label and the actual control strategy can be strengthened. After optimization, the model evaluation value needs to be recalculated until the evaluation value reaches the preset standard.

[0169] In this embodiment, all compliant and optimized category decision tree models under each edge controller are integrated to form a comprehensive decision model covering all anomaly categories of that controller. Specifically, the model sources are two types of category decision tree models for each edge controller: one is the original compliant model with an evaluation value greater than or equal to a preset evaluation value; the other is a model with an evaluation value less than a preset evaluation value that has been optimized to meet the requirements. This ensures that all integrated models meet performance requirements. The integration logic involves classifying and sorting the models according to the anomaly cause-phenomenon category, constructing a multi-level delivery decision tree model. The top layer of this model is the anomaly category identification branch. Using the input anomaly feature vector and anomaly label, the corresponding anomaly cause-phenomenon category is first located, and then the category decision tree model corresponding to that category is called to output a specific predictive control strategy and the predicted anomaly duration. Each edge controller will have an independent delivery decision tree model that fully covers all anomaly cause-phenomenon categories under the controller's responsibility. This model can quickly identify anomaly types and output targeted processing solutions in practical applications, providing core decision support for real-time anomaly control of the edge controller.

[0170] The beneficial effects of the above technologies are as follows: Based on abnormal transmission data, the model evaluation value of the category decision tree model for each abnormal cause-phenomenon category of each edge controller is calculated. Based on all category decision tree models of each edge controller, the transmission decision tree model of each edge controller is constructed. This can verify the accuracy of control strategy prediction and the rationality of abnormal duration, improve the comprehensiveness and practicality of model evaluation, and provide more comprehensive support for actual operation and maintenance.

[0171] Example 9: Based on Example 1, a digital intelligent air suspension conveyor control system, comprising a control module including:

[0172] Real-time anomaly control strategy unit: Each real-time device anomaly vector, each real-time air pressure anomaly vector, and each real-time airflow anomaly vector from the real-time edge anomaly data of each edge controller are input into the delivery decision tree model of each edge controller to determine the real-time anomaly control strategy of each edge controller. The real-time anomaly control strategy includes multiple sub-strategies.

[0173] Intelligent control unit: Executes all sub-strategies in the real-time abnormal control strategy of all edge controllers to realize intelligent control of the air suspension conveyor.

[0174] In this embodiment, real-time anomaly data from the edge controller is input into a dedicated delivery decision tree model, which outputs an executable real-time anomaly control strategy. Specifically, the input data consists of three types of anomaly vectors derived from the real-time edge anomaly data of each edge controller. The first type is the anomaly vector for each real-time device in the real-time edge device anomaly data, containing parameter deviations of the abnormal device, such as excessive motor current or fan speed deviation. The second type is the anomaly vector for each real-time air pressure in the real-time edge air pressure anomaly data, containing parameter deviations of the abnormal air pressure monitoring point, such as static pressure below the threshold or excessive differential pressure. The third type is the anomaly vector for each real-time airflow in the real-time edge airflow anomaly data, containing parameter deviations of the abnormal airflow monitoring point, such as insufficient wind speed or excessive turbulence. These three types of anomaly vectors need to be formatted according to the requirements of the delivery decision tree model to ensure that the model can read and parse them. The strategy generation process involves inputting the formatted three types of anomaly vectors one by one into the delivery decision tree model of the edge controller. The model first identifies the corresponding anomaly cause / phenomenon category based on features in the anomaly vector, such as parameter deviation type and associated monitoring points / equipment. Then, it invokes the decision logic for that category to output handling measures for each type of anomaly. These measures are integrated to form the real-time anomaly control strategy for the edge controller. This strategy contains multiple sub-strategies, each corresponding to a specific type of anomaly. For example, for an anomaly vector indicating excessive motor current, a sub-strategy to reduce the motor load is output; for an anomaly vector indicating insufficient static pressure, a sub-strategy to adjust the opening of the airflow regulating valve is output. Each edge controller independently generates its own real-time anomaly control strategy, without affecting others.

[0175] In this embodiment, all sub-strategies within the real-time anomaly control policies of all edge controllers are uniformly executed to resolve system anomalies and maintain stable operation. Specifically, the execution targets are all sub-strategies within the real-time anomaly control policies output by all edge controllers. First, the real-time anomaly control policies of all edge controllers need to be collected, each sub-strategy extracted, and the execution target corresponding to each sub-strategy defined. For example, one sub-strategy corresponds to adjusting the speed of fan A, while another sub-strategy corresponds to closing the bypass valve of the air pressure monitoring point P001, along with execution parameters such as reducing the fan speed from 1800 rpm to 1500 rpm and adjusting the bypass valve opening from 50% to 0%.

[0176] Ultimately, by executing all sub-strategies, the abnormalities in equipment, air pressure, and airflow of the air suspension conveyor are eliminated, restoring the system operating parameters to the normal range, achieving intelligent control without human intervention, and ensuring the continuous and stable operation of the conveying operation.

[0177] The beneficial effects of the above technologies are as follows: Based on the real-time edge anomaly data of each edge controller and the transport decision tree model, the real-time anomaly control strategy of each edge controller is determined and executed, realizing intelligent control of the air suspension conveyor. This can improve the targeting of the control strategy, avoid cross-controller strategy interference, adapt to multiple anomaly concurrent scenarios, reduce manual intervention, shorten the link from anomaly identification to strategy output, and reduce control latency.

[0178] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A digitized air-suspended conveyor control system, characterized by, Comprising: Monitoring acquisition module: monitoring the real-time equipment operation data, real-time air pressure data and real-time airflow data of the air suspension conveyor, acquiring historical conveying data of multiple conveyances, and acquiring an edge control set of multiple edge controllers; Range module: acquiring equipment operation range data of the air suspension conveyor, determining normal conveying data and abnormal conveying data based on the historical conveying data, and determining air pressure range data and airflow range data based on the normal conveying data; Determination module: based on the edge control set of each edge controller, the operation range data, the air pressure range data, the airflow range data, and the real-time equipment operation data, the real-time air pressure data and the real-time airflow data, determining the real-time edge abnormal data of each edge controller; Analysis module: analyzing the abnormal conveying data and the edge control set of each edge controller, determining the reason-phenomenon class abnormal data of each abnormal reason-phenomenon class of each edge controller and the category decision tree model; Construction module: based on the abnormal conveying data, calculating the model evaluation value of the category decision tree model of each abnormal reason-phenomenon class of each edge controller, and constructing the conveying decision tree model of each edge controller based on all category decision tree models of each edge controller; Control module: based on the real-time edge abnormal data and the conveying decision tree model of each edge controller, determining and executing the real-time abnormal control strategy of each edge controller to realize intelligent control of the air suspension conveyor.

2. The digitized air levitation conveyor control system of claim 1, wherein, The monitoring acquisition module comprises: The first monitoring unit: based on the first sensor group installed on each device of the air suspension conveyor, monitoring the real-time device operation sub-data of each device, wherein the device operation sub-data includes device name, multiple operation parameters and operation parameter value of each operation parameter; Real-time equipment operation data unit: based on the device operation sub-data of all devices, determining the real-time equipment operation data of the air suspension conveyor; Real-time air pressure data unit: based on the distributed pressure sensor group installed on the air suspension conveyor, monitoring the real-time air pressure data of the air suspension conveyor, wherein the real-time air pressure data includes real-time air pressure sub-data of multiple air pressure monitoring points, and the real-time air pressure sub-data includes air pressure monitoring point position, multiple air pressure parameters and air pressure parameter value of each air pressure parameter; The third monitoring unit: based on the second sensor group of each airflow monitoring point, monitoring the real-time airflow sub-data of each airflow monitoring point, wherein the real-time airflow sub-data includes airflow monitoring point position, multiple airflow parameters and airflow parameter value of each airflow parameter; Real-time airflow data unit: based on the real-time airflow sub-data of all airflow monitoring points, determining the real-time airflow data of the air suspension conveyor.

3. The digitized air levitation conveyor control system of claim 1, wherein, The monitoring acquisition module further comprises: a historical conveying data unit configured to obtain historical conveying data of the air suspension conveyor based on a plurality of conveyings in a historical specified time period, wherein the historical conveying data comprises historical material data, historical air pressure data, historical air flow data, and historical abnormal data, the historical abnormal data comprises a plurality of historical abnormal sub-data, and each historical abnormal sub-data comprises an abnormal cause, an abnormal feature vector, an abnormal phenomenon, an abnormal start time, an abnormal end time, an abnormal label, and an abnormal control strategy, and the abnormal label comprises a device name, an air pressure monitoring point position, and an air flow monitoring point position; an edge control set unit configured to obtain an edge control set of a plurality of edge controllers installed on the air suspension conveyor, wherein the edge control set comprises a plurality of device names, a plurality of air pressure monitoring point positions, and a plurality of air flow monitoring point positions.

4. The digitized air levitation conveyor control system of claim 1, wherein, a range module, comprising: a device operation range data unit configured to obtain device operation range data of the air suspension conveyor, wherein the device operation range data comprises a plurality of device operation range sub-data, and each device operation range sub-data comprises a plurality of operation parameters and a normal operation range of each operation parameter; a first division unit configured to divide all the historical conveying data of the conveyings based on whether the historical conveying data comprises the historical abnormal data, to determine normal conveying data and abnormal conveying data; a material-air pressure adaptive model unit configured to train a material-air pressure adaptive model based on historical material data and historical air pressure data of all the historical conveying data of the conveyings in the normal conveying data; a material-air flow adaptive model unit configured to train a material-air flow adaptive model based on historical material data and historical air flow data of all the historical conveying data of the conveyings in the normal conveying data; a conveying material data unit configured to obtain conveying material data, wherein the conveying material data comprises a plurality of material parameters and a material parameter value of each material parameter; an air pressure range data unit configured to input the conveying material data into the material-air pressure adaptive model, to determine air pressure range data based on an output result of the material-air pressure adaptive model, wherein the air pressure range data comprises a plurality of air pressure range sub-data of air pressure monitoring points, and each air pressure range sub-data comprises a normal air pressure range of each air pressure parameter; an air flow range data unit configured to input the conveying material data into the material-air flow adaptive model, to determine air flow range data based on an output result of the material-air flow adaptive model, wherein the air flow range data comprises a plurality of air flow range sub-data of air flow monitoring points, and each air flow range sub-data comprises a normal air flow range of each air flow parameter.

5. The digitized air levitation conveyor control system of claim 1, wherein, a determination module, comprising: a real-time device abnormal vector unit configured to determine a running label of each device corresponding to each device name in the edge control set of each edge controller of the air suspension conveyor and a real-time device abnormal vector based on device operation range sub-data and real-time device operation sub-data of the device corresponding to each device name in the edge control set of each edge controller of the air suspension conveyor, wherein the running label comprises normal running and abnormal running. The real-time edge device exception data unit determines the real-time edge device exception data of each edge controller based on the real-time device exception vectors of all devices corresponding to all device names of which the running labels are abnormal running in the edge control set of each edge controller. The real-time air pressure exception vector unit determines the air pressure labels and the real-time air pressure exception vectors of the air pressure monitoring points corresponding to each air pressure monitoring point position in the edge control set of each edge controller of the air suspension conveyor based on the air pressure range sub-data and the real-time air pressure sub-data of the air pressure monitoring points corresponding to each air pressure monitoring point position in the edge control set of each edge controller of the air suspension conveyor, wherein the air pressure labels include air pressure normal and air pressure abnormal. The real-time edge air pressure exception data unit determines the real-time edge air pressure exception data of each edge controller based on the real-time air pressure exception vectors of all air pressure monitoring points corresponding to all air pressure monitoring point positions of which the air pressure labels are air pressure abnormal in the edge control set of each edge controller. The real-time airflow exception vector unit determines the airflow labels and the real-time airflow exception vectors of the airflow monitoring points corresponding to each airflow monitoring point position in the edge control set of each edge controller of the air suspension conveyor based on the airflow range sub-data and the real-time airflow sub-data of the airflow monitoring points corresponding to each airflow monitoring point position in the edge control set of each edge controller of the air suspension conveyor, wherein the airflow labels include airflow normal and airflow abnormal. The real-time edge airflow exception data unit determines the real-time edge airflow exception data of each edge controller based on the real-time airflow exception vectors of all airflow monitoring points corresponding to all airflow monitoring point positions of which the airflow labels are airflow abnormal in the edge control set of each edge controller. The real-time edge exception data unit determines the real-time edge exception data of each edge controller based on the real-time edge device exception data, the real-time edge air pressure exception data, and the real-time edge airflow exception data of each edge controller.

6. The cyber-physical air levitation conveyor control system of claim 5, wherein, The analysis module includes: The historical edge exception data unit extracts the historical transport data of each transport in the abnormal transport data based on the edge control set of each edge controller and the abnormal labels in the historical exception data of the historical transport data of each transport in the abnormal transport data, and determines the historical edge exception data of each transport of each edge controller, wherein the historical edge exception data includes the historical exception sub-data of the devices corresponding to the multiple device names, the air pressure monitoring points corresponding to the multiple air pressure monitoring point positions, and the airflow monitoring points corresponding to the multiple airflow monitoring point positions of which the abnormal labels belong to the edge control set. The cause category exception data unit classifies the historical edge exception data of all transports of each edge controller based on the abnormal causes in all historical exception sub-data in the historical edge exception data of all transports of each edge controller, and determines the cause category exception data of each edge controller based on each abnormal cause and the abnormal cause labels of each cause category exception data, wherein the cause category exception data includes multiple historical exception sub-data. Reason-phenomenon category abnormal data unit: based on the abnormal phenomena in all historical abnormal sub-data in the reason category abnormal data of each abnormal reason of each edge controller, classifying the reason category abnormal data of each abnormal reason of each edge controller, determining the reason-phenomenon category abnormal data of each abnormal reason-phenomenon category of each edge controller and the abnormal reason-phenomenon category label of each reason-phenomenon category abnormal data, wherein the reason-phenomenon category abnormal data comprises a plurality of historical abnormal sub-data.

7. The cyber-physical air levitation conveyor control system of claim 6, wherein, The analysis module further comprises: A second division unit: dividing the reason-phenomenon category abnormal data of each abnormal reason-phenomenon category of each edge controller into training category abnormal data and test category abnormal data; A category decision tree model unit: training the category decision tree model of each abnormal reason-phenomenon category of each edge controller based on the training category abnormal data of each abnormal reason-phenomenon category of each edge controller; A construction unit: taking the abnormal feature vector and the abnormal label of all historical abnormal sub-data in the training category abnormal data of each abnormal reason-phenomenon category of each edge controller as the input of the category decision tree model, and taking the abnormal start time, the abnormal end time and the abnormal control strategy of all historical abnormal sub-data in the training category abnormal data of each abnormal reason-phenomenon category of each edge controller as the input of the category decision tree model.

8. The cyber-physical gas levitation conveyor control system of claim 7, wherein, The construction module comprises: A prediction unit: inputting the abnormal feature vector and the abnormal label of each historical abnormal sub-data in the test category abnormal data of each abnormal reason-phenomenon category of each edge controller into the category decision tree model of each abnormal reason-phenomenon category, and determining the predicted control strategy and the predicted abnormal duration of each historical abnormal sub-data in the test category abnormal data of each abnormal reason-phenomenon category of each edge controller based on the output result of the category decision tree model; A simulation data unit: determining simulation data based on all historical abnormal sub-data in the training category abnormal data of all abnormal reason-phenomenon categories of all edge controllers, wherein the simulation data comprises a plurality of historical abnormal sub-data; A simulation unit: inputting the abnormal feature vector, the abnormal start time and the abnormal label in each historical abnormal sub-data in the simulation data, and the predicted control strategy of each historical abnormal sub-data into the conveying simulation model, and determining the simulation abnormal end time of each historical abnormal sub-data based on the conveying simulation model; A model evaluation value unit: calculating the model evaluation value of the category decision tree model of each abnormal reason-phenomenon category of each edge controller based on the predicted control strategy, the predicted abnormal duration and the simulation abnormal end time of all historical abnormal sub-data in the test category abnormal data of each abnormal reason-phenomenon category of each edge controller, and the abnormal control strategy, the abnormal start time and the abnormal end time in all historical abnormal sub-data. The optimization unit compares the model evaluation value of each abnormality cause-phenomenon category of each edge controller with a preset evaluation value, and if the model evaluation value is less than the preset evaluation value, optimizes the category decision tree model based on the model evaluation value of each abnormality cause-phenomenon category of each edge controller, all historical abnormality sub-data in the training category abnormality data, the predicted control strategy of each historical abnormality sub-data, and the predicted abnormality duration. The delivery decision tree model unit constructs a delivery decision tree model for each edge controller based on all optimized category decision tree models whose model evaluation values are less than the preset evaluation value and all category decision tree models whose model evaluation values are greater than or equal to the preset evaluation value.

9. The digitized air levitation conveyor control system of claim 1, wherein, The control module comprises: The real-time abnormality control strategy unit inputs each real-time device abnormality vector in the real-time edge device abnormality data, each real-time air pressure abnormality vector in the real-time edge air pressure abnormality data, and each real-time air flow abnormality vector in the real-time edge air flow abnormality data in the real-time edge abnormality data of each edge controller into the delivery decision tree model of each edge controller, respectively, to determine a real-time abnormality control strategy for each edge controller, wherein the real-time abnormality control strategy comprises a plurality of sub-strategies. The intelligent control unit executes all sub-strategies in the real-time abnormality control strategies of all edge controllers to achieve intelligent control of the air-suspended conveyor.

Images