A dust removal monitoring system

Abstract

The present application relates to a dust removal monitoring system, and to the technical field of dust removal monitoring, which includes a pulse bag dust collector; a sensor array; an edge device; the edge device communicates with the sensor array and the pulse bag dust collector; the edge device receives monitoring data sent by each sensor in the sensor array; a cloud device; the cloud device communicates with the edge device; the cloud device receives the monitoring data sent by the edge device, performs data analysis on the monitoring data through a configured strategy optimization model, and obtains a model output containing a control strategy optimization result; the cloud device transmits the model output to the edge device so that the edge device controls the pulse bag dust collector to adjust operating parameters based on the model output. The present application can reduce unnecessary energy consumption while ensuring dust removal efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of dust removal monitoring, and in particular to a dust removal monitoring system. Background Art

[0002] With the acceleration of industrialization, environmental protection has become one of the key issues of global concern. In the process of industrial production, a large amount of dust emissions not only pollutes the environment, but also poses a threat to human health. Therefore, an effective dust removal system is essential for protecting the environment and ensuring human health. Most traditional dust removal systems rely on manual monitoring and manual adjustment, which is not only inefficient, but also difficult to respond to changes in working conditions in real time, resulting in low dust removal efficiency and high energy consumption. As a high-efficiency dust removal equipment, pulse bag dust collectors have gradually replaced old dust removal methods such as electrostatic precipitators, wet dust removal, and cyclone dust removal in many industrial fields, and are widely used around the world. The main structure of the pulse bag dust collector includes the following parts: a filter unit composed of multiple bags, usually made of porous materials, to capture dust in the air; a pulse cleaning system is used to remove dust attached to the bags; and an ash hopper is used to collect dust dropped during the cleaning process.

[0003] In recent years, with the development of the Internet of Things (IOT), big data analysis and artificial intelligence technology, intelligent monitoring systems have begun to be widely used in various industrial fields. In particular, in dust removal systems, by deploying various sensors, the working status of the dust collector can be monitored in real time, and the data can be uploaded to the cloud for analysis and processing through edge computing devices. This intelligent monitoring system based on IoT technology can dynamically adjust the operating parameters of the dust collector, thereby improving dust removal efficiency and reducing energy consumption.

[0004] However, the current domestic dust removal monitoring system still has some shortcomings. Existing technical means can only make simple parameter adjustments according to preset rules and lack adaptive optimization capabilities. Secondly, anomaly detection usually relies on fixed threshold judgments and is difficult to adapt to changes under different working conditions.

[0005] These problems greatly increase the amount of repeated work on site and generate safety risks. How to solve this technical problem is a technical problem that technicians in this field need to overcome. Summary of the invention

[0006] In order to at least partially solve the above technical problems, the present application provides a dust removal monitoring system.

[0007] The present application provides a dust removal monitoring system, which adopts the following technical solution:

[0008] A dust removal monitoring system, comprising:

[0009] Pulse bag dust collector;

[0010] A sensor array; the sensor array includes a plurality of sensors of different types; at least one sensor of each type is provided; the sensors of the sensor array are used to obtain monitoring data of corresponding types;

[0011] Edge device; the edge device communicates with the sensor array and the pulse bag filter; the edge device receives monitoring data sent by each sensor in the sensor array;

[0012] A cloud device; the cloud device communicates with the edge device; the cloud device receives the monitoring data sent by the edge device, performs data analysis on the monitoring data through a configured strategy optimization model, and obtains a model output including a control strategy optimization result; the cloud device transmits the model output to the edge device so that the edge device controls the pulse bag dust collector to adjust operating parameters based on the model output.

[0013] Optionally, the configuration process of the strategy optimization model includes:

[0014] A state space S is defined; the state space includes all sensor data that affect the dust removal effect; the sensor data that affect the dust removal effect includes at least dust concentration, temperature and material level value;

[0015] An action space A is defined; the action space A includes all adjustable parameters; the adjustable parameters include at least a fan speed and a pulse frequency;

[0016] Define the reward function R: ;in For dust removal efficiency; is the energy efficiency; a is the weight coefficient, which is used to balance the relationship between dust removal efficiency and energy efficiency; the dust removal efficiency is obtained based on the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector; the energy efficiency is obtained based on the energy consumption of the pulse bag dust collector during operation;

[0017] S1, obtain some preprocessed training data to obtain the current state s1;

[0018] S2, outputting a first action through a neural network according to the current state s1;

[0019] S3, execute the first action, and observe the new state s2 and reward R after executing the first action;

[0020] S4, use reinforcement algorithm for model training;

[0021] Return to S2 and repeat S2, S3 and S4 until the reward R reaches the convergence requirement to obtain the policy optimization model.

[0022] Optionally, use a reinforcement algorithm for model training, including:

[0023] S801, initializing the parameters of the strategy network and the value network;

[0024] S802. Under the current strategy, collect a batch of trajectory data, each trajectory containing a series of states, actions, rewards, and next states;

[0025] S803, for each trajectory, calculate the reward at each time step t;

[0026] S804, using the value network to estimate the value of each state;

[0027] S805, calculating the advantage function of each time step;

[0028] S806. For each time step t, update the policy network according to the PPO algorithm, use the gradient descent method to minimize the loss function, update the policy network parameters, and update the value network parameters;

[0029] Steps S802 to S806 are repeated until a predetermined confidence level is reached.

[0030] Optionally, before the cloud device performs data analysis on the monitoring data through the configured strategy optimization model, the cloud device further includes: performing anomaly detection through the configured anomaly detection model; when the anomaly detection model outputs information indicating the occurrence of anomalies; the cloud device sends an alarm message to the edge device so that the edge device takes an abnormal feedback instruction; the abnormal feedback instruction includes at least one of triggering an alarm and stopping the pulse bag dust collector chamber;

[0031] The configuration process of the anomaly detection model includes:

[0032] Acquire some historical normal data sent by the sensor array; set the parameters of the isolation forest; the parameters include the number of trees and the size of sub-sampling;

[0033] Using the historical normal data to train an initial model to obtain an anomaly detection model;

[0034] The detection method of the anomaly detection model includes:

[0035] For each sample data, calculate its path length in the isolated forest; obtain an anomaly score based on the ratio of the expected value of the path length to the height of the tree; the path length is the number of splits required for the data point corresponding to the sample data to reach the leaf node from the root node;

[0036] The sample data with anomaly scores higher than the threshold are judged as abnormal data;

[0037] Output information representing the abnormality based on the type of abnormal data.

[0038] Optionally, the threshold is adjusted dynamically; the threshold adjustment process includes:

[0039] S401, setting a time period T, and continuously collecting sample data within the time period T;

[0040] S402, within the time period T, for each newly collected sample data, calculate its anomaly score in the isolation forest model;

[0041] S403: Count the abnormal scores of all sample data in the current time period T that are greater than the threshold of this time period, and calculate the average value of all abnormal scores that are greater than the threshold of this event period. and standard deviation ;The initial threshold is preset;

[0042] S404: Set a new threshold value to: , where k is a preset constant used to control the sensitivity of threshold adjustment;

[0043] S405, applying the new threshold to anomaly detection in the next time period T;

[0044] S401 to S405 are repeatedly executed.

[0045] Optionally, the sensor array includes:

[0046] Pressure sensor: A pressure sensor is provided at the air inlet and the air outlet of the pulse bag dust collector respectively; it is used to monitor the pressure difference between the air inlet and the air outlet of the pulse bag dust collector;

[0047] Temperature sensor; installed inside the pulse bag dust collector to monitor the temperature inside the pulse bag dust collector;

[0048] Dust concentration sensor: A dust concentration sensor is provided at the air outlet and air inlet of the pulse bag dust collector, respectively, to monitor the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector;

[0049] The material level meter is installed on the ash hopper or the surface of the filter bag; it is used to monitor the accumulation of dust in the ash hopper or on the surface of the filter bag;

[0050] Pulse valve state sensor; set on the pulse valve; used to monitor the action state of the pulse valve;

[0051] Flow sensor; installed at the air inlet and outlet; used to monitor the gas flow of the pulse bag dust collector;

[0052] Vibration sensor; installed at the motor installation location of the pulse bag dust collector; used to monitor the vibration of the motor;

[0053] The system also includes a communication status sensor; used to monitor the communication status between the edge device and the cloud device in real time; when the communication is interrupted, the alarm device is controlled to sound an alarm.

[0054] Optionally, the monitoring system further includes a plurality of environmental sensors for monitoring the environment in which the pulse bag filter is located; the environmental sensors include a temperature sensor and a humidity sensor; the environmental sensors communicate with the edge device; the defined state space of the strategy optimization model also includes environmental data;

[0055] Wherein, the cloud device receives the environmental data and updates the state space S based on the environmental data;

[0056] The strategy optimization model adjusts its model parameters according to the environmental data; the model parameters include weight coefficients and thresholds;

[0057] The strategy optimization model outputs the adjusted strategy optimization results after updating the model parameters.

[0058] Optionally, the system also includes: a local storage module, which is used to store monitoring data and operation records; when the communication between the edge device and the cloud device is interrupted, the local storage module stores the offline data and automatically uploads it to the cloud device after the communication is restored.

[0059] In summary, the technical solution of the present application realizes dynamic adjustment of the operating parameters of the pulse bag dust collector, thereby ensuring the dust removal efficiency and reducing unnecessary energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a system block diagram of a dust removal monitoring system according to an embodiment of the present application;

[0061] In the figure, 101 is a pulse bag dust collector; 102 is a sensor array; 103 is an edge device; 104 is a cloud device. DETAILED DESCRIPTION

[0062] The following is combined with Figure 1 The present application is further described with specific embodiments:

[0063] First of all, it should be noted that in the description of this application, if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside" and other directional words appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, which is only for the convenience of description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of this application; in addition, if the terms "first", "second", "third" and other numerical quantifiers appear, they are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in this application, unless otherwise clearly specified and limited, if the terms "installed", "connected", and "connected" appear, they should be understood in a broad sense, for example, it can be a fixed connection, or a detachable connection, a limited connection such as an interference fit, a transition fit, or an integral connection; it can be directly connected or indirectly connected through an intermediate medium; therefore, for ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0064] The present application embodiment discloses a dust removal monitoring system. Figure 1 As an implementation of a dust removal monitoring system, a dust removal monitoring system includes:

[0065] Pulse bag dust collector 101;

[0066] Sensor array 102; the sensor array includes several different types of sensors; at least one sensor of each type is provided; the sensors of the sensor array are used to obtain corresponding types of monitoring data;

[0067] Edge device 103; the edge device communicates with the sensor array and the pulse bag filter; the edge device receives monitoring data sent by each sensor in the sensor array;

[0068] A cloud device 104; the cloud device communicates with the edge device; the cloud device receives the monitoring data sent by the edge device, performs data analysis on the monitoring data through the configured strategy optimization model, and obtains a model output including a control strategy optimization result; the cloud device transmits the model output to the edge device so that the edge device controls the pulse bag dust collector to adjust the operating parameters based on the model output.

[0069] Specifically, various types of monitoring data are obtained through the sensor array, which can comprehensively and timely reflect the working status of the dust collector; the cloud device uses the configured strategy optimization model to conduct in-depth analysis and processing on the collected data, and the generated control strategy optimization results are fed back to the edge device, thereby realizing dynamic adjustment of the operating parameters of the pulse bag dust collector, ensuring dust removal efficiency while reducing unnecessary energy consumption.

[0070] As a specific implementation of a dust removal monitoring system, the configuration process of the strategy optimization model includes:

[0071] A state space S is defined; the state space includes all sensor data that affect the dust removal effect; the sensor data that affect the dust removal effect includes at least dust concentration, temperature and material level value;

[0072] An action space A is defined; the action space A includes all adjustable parameters; the adjustable parameters include at least a fan speed and a pulse frequency;

[0073] Define the reward function R: ;in For dust removal efficiency; is the energy efficiency; a is the weight coefficient, which is used to balance the relationship between dust removal efficiency and energy efficiency; the dust removal efficiency is obtained based on the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector; the energy efficiency is obtained based on the energy consumption of the pulse bag dust collector during operation;

[0074] S1, obtain some preprocessed training data to obtain the current state s1;

[0075] S2, outputting a first action through a neural network according to the current state s1;

[0076] S3, execute the first action, and observe the new state s2 and reward R after executing the first action;

[0077] S4, use reinforcement algorithm for model training;

[0078] Return to S2 and repeat S2, S3 and S4 until the reward R reaches the convergence requirement to obtain the policy optimization model.

[0079] Specifically, by defining detailed state space and action space, the system can accurately identify the key factors that affect dust removal efficiency, and adjust adjustable parameters such as fan speed and pulse frequency accordingly, thereby achieving refined control of the dust removal process. The defined reward function combines the dust removal efficiency (Ec) and energy efficiency (Ee), and introduces a weight coefficient a between the two for balance. This enables the system to minimize energy consumption while ensuring efficient dust removal, achieving dual optimization of dust removal performance and energy saving effects. Through the action selection mechanism output by the neural network, the system can automatically select the best operation plan based on the current state, and continuously adjust the strategy based on the feedback (new state and reward) after executing the action.

[0080] As a specific implementation of a dust removal monitoring system, a reinforcement algorithm is used for model training, including:

[0081] S801, initializing the parameters of the strategy network and the value network;

[0082] S802. Under the current strategy, collect a batch of trajectory data, each trajectory containing a series of states, actions, rewards, and next states;

[0083] S803, for each trajectory, calculate the reward at each time step t;

[0084] S804, using the value network to estimate the value of each state;

[0085] S805, calculating the advantage function of each time step;

[0086] S806. For each time step t, update the policy network according to the PPO algorithm, use the gradient descent method to minimize the loss function, update the policy network parameters, and update the value network parameters;

[0087] Steps S802 to S806 are repeated until a predetermined confidence level is reached.

[0088] Specifically, by initializing the parameters of the policy network and the value network and collecting trajectory data under the current policy, the system can quickly enter the learning state; by calculating the reward of each time step and using the value network to estimate the value of each state, the system can accurately evaluate the quality of each decision, thereby dynamically adjusting the strategy to achieve optimal control of the dust removal process. The PPO algorithm is used to update the policy network, minimize the loss function through the gradient descent method, and continuously iterate and optimize the parameters of the policy network and the value network, so that the system can continuously improve its behavior strategy and gradually approach the optimal solution. The entire training process aims to achieve a predetermined confidence level to ensure the quality of model training, so that the final strategy can be stably applied to the actual dust removal leakage monitoring system, improving the reliability and effectiveness of the system.

[0089] As a specific implementation of a dust removal monitoring system, before the cloud device performs data analysis on the monitoring data through the configured strategy optimization model, it also includes: performing anomaly detection through the configured anomaly detection model; when the anomaly detection model outputs information representing the occurrence of anomalies; the cloud device sends an alarm message to the edge device so that the edge device takes an abnormal feedback instruction; the abnormal feedback instruction includes triggering an alarm and stopping at least one of the pulse bag dust collector chambers;

[0090] The configuration process of the anomaly detection model includes:

[0091] Acquire some historical normal data sent by the sensor array; set the parameters of the isolation forest; the parameters include the number of trees and the size of sub-sampling;

[0092] Using the historical normal data to train an initial model to obtain an anomaly detection model;

[0093] The detection method of the anomaly detection model includes:

[0094] For each sample data, calculate its path length in the isolated forest; obtain an anomaly score based on the ratio of the expected value of the path length to the height of the tree; the path length is the number of splits required for the data point corresponding to the sample data to reach the leaf node from the root node;

[0095] The sample data with anomaly scores higher than the threshold are judged as abnormal data;

[0096] Output information representing the abnormality based on the type of abnormal data.

[0097] Specifically, by configuring the anomaly detection model, possible anomalies can be discovered and marked in time before the strategy optimization model conducts in-depth analysis of the monitoring data, thereby providing early warning. When the anomaly detection model detects an anomaly, the cloud device immediately sends an alarm message to the edge device, prompting the edge device to take corresponding emergency measures, such as triggering an alarm or stopping the dust collector, avoiding production interruptions or environmental pollution caused by failures not being discovered in time. The anomaly detection model trained with historical normal data using the isolation forest algorithm can effectively identify abnormal patterns by setting appropriate model parameters (such as the number of trees and subsampling size). The anomaly score calculation method based on path length can adapt to different types of data distributions, maintain high detection performance even when data features change slightly, and enhance the adaptability of the model. The anomaly detection model is based on the isolation forest algorithm, and determines the anomaly score by statistically analyzing the number of splits required for a data point to reach a leaf node. This method helps reduce false alarms and ensures that only truly abnormal situations trigger alarms or actions.

[0098] As one implementation of a dust removal monitoring system, the threshold is dynamically adjusted; the threshold adjustment process includes:

[0099] S401, setting a time period T, and continuously collecting sample data within the time period T;

[0100] S402, within the time period T, for each newly collected sample data, calculate its anomaly score in the isolation forest model;

[0101] S403: Count the abnormal scores of all sample data in the current time period T that are greater than the threshold of this time period, and calculate the average value of all abnormal scores that are greater than the threshold of this event period. and standard deviation ;The initial threshold is preset;

[0102] S404: Set a new threshold value to: , where k is a preset constant used to control the sensitivity of threshold adjustment;

[0103] S405, applying the new threshold to anomaly detection in the next time period T;

[0104] S401 to S405 are repeatedly executed.

[0105] Specifically, by repeatedly collecting data, calculating anomaly scores, and adjusting thresholds, the anomaly detection model can be continuously optimized over time, thereby better reflecting the anomaly characteristics under actual operating conditions. At the end of each time period, the system adjusts the threshold based on the latest statistical data and immediately applies the new threshold to anomaly detection in the next time period, ensuring the system's real-time response to current conditions.

[0106] As one implementation of a dust removal monitoring system, the sensor array includes:

[0107] Pressure sensor: A pressure sensor is provided at the air inlet and the air outlet of the pulse bag dust collector respectively; it is used to monitor the pressure difference between the air inlet and the air outlet of the pulse bag dust collector;

[0108] Temperature sensor; installed inside the pulse bag dust collector to monitor the temperature inside the pulse bag dust collector;

[0109] Dust concentration sensor: A dust concentration sensor is provided at the air outlet and air inlet of the pulse bag dust collector, respectively, to monitor the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector;

[0110] The material level meter is installed on the ash hopper or the surface of the filter bag; it is used to monitor the accumulation of dust in the ash hopper or on the surface of the filter bag;

[0111] Pulse valve state sensor; set on the pulse valve; used to monitor the action state of the pulse valve;

[0112] Flow sensor; installed at the air inlet and outlet; used to monitor the gas flow of the pulse bag dust collector;

[0113] Vibration sensor; installed at the motor installation location of the pulse bag dust collector; used to monitor the vibration of the motor;

[0114] The system also includes a communication status sensor; used to monitor the communication status between the edge device and the cloud device in real time; when the communication is interrupted, the alarm device is controlled to sound an alarm.

[0115] As one implementation of a dust removal monitoring system, the monitoring system further includes a plurality of environmental sensors for monitoring the environment of the pulse bag dust collector; the environmental sensors include a temperature sensor and a humidity sensor; the environmental sensors communicate with the edge device; the defined state space of the strategy optimization model also includes environmental data;

[0116] Wherein, the cloud device receives the environmental data and updates the state space S based on the environmental data;

[0117] The strategy optimization model adjusts its model parameters according to the environmental data; the model parameters include weight coefficients and thresholds;

[0118] The strategy optimization model outputs the adjusted strategy optimization results after updating the model parameters.

[0119] Specifically, by introducing environmental sensors (such as temperature sensors and humidity sensors), the system can monitor the conditions of the environment in which the dust collector is located in real time and incorporate these environmental data into the state space of the strategy optimization model. This enables the system to dynamically adjust the dust removal strategy according to environmental changes, enhancing its ability to adapt to environmental changes. By updating the policy optimization model after the model parameters are updated, the system can better match the actual working environment.

[0120] As one of the implementation methods of a dust removal monitoring system, the system also includes: a local storage module, which is used to store monitoring data and operation records; when the communication between the edge device and the cloud device is interrupted, the local storage module stores the offline data and automatically uploads it to the cloud device after the communication is restored.

[0121] It should be noted that the above embodiments are only used to illustrate the present application and are not intended to limit the technical solutions described in the present application. Although the present application has been described in detail in this specification with reference to the above embodiments, a person of ordinary skill in the art should understand that a person of ordinary skill in the art can still modify or make equivalent substitutions to the present application, and all technical solutions and improvements thereof that do not depart from the spirit and scope of the present application should be included in the scope of the claims of the present application.

Claims

1. A dust removal monitoring system, characterized in that: include: Pulse bag dust collector; Sensor arrays; The sensor array includes a number of different types of sensors; At least one sensor of each type is provided; the sensors of the sensor array are used to obtain monitoring data of the corresponding type; Edge device; the edge device communicates with the sensor array and the pulse bag filter; the edge device receives monitoring data sent by each sensor in the sensor array; Cloud devices; The cloud device communicates with the edge device; the cloud device receives the monitoring data sent by the edge device, performs data analysis on the monitoring data through the configured strategy optimization model, and obtains a model output including a control strategy optimization result; the cloud device transmits the model output to the edge device so that the edge device controls the pulse bag filter to adjust the operating parameters based on the model output; The configuration process of the strategy optimization model includes: A state space S is defined; the state space includes all sensor data that affect the dust removal effect; the sensor data that affect the dust removal effect includes at least dust concentration, temperature and material level value; An action space A is defined; the action space A includes all adjustable parameters; the adjustable parameters include at least a fan speed and a pulse frequency; Define the reward function R: ;in For dust removal efficiency; is the energy efficiency; a is the weight coefficient, which is used to balance the relationship between dust removal efficiency and energy efficiency; the dust removal efficiency is obtained based on the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector; the energy efficiency is obtained based on the energy consumption of the pulse bag dust collector during operation; S1, obtain some preprocessed training data to obtain the current state s1; S2, outputting a first action through a neural network according to the current state s1; S3, execute the first action, and observe the new state s2 and reward R after executing the first action; S4, use reinforcement algorithm for model training; Return to S2 and repeat S2, S3 and S4 until the reward R reaches the convergence requirement to obtain the strategy optimization model; The sensor array comprises: Pressure sensor: A pressure sensor is provided at the air inlet and the air outlet of the pulse bag dust collector respectively; it is used to monitor the pressure difference between the air inlet and the air outlet of the pulse bag dust collector; Temperature sensor; installed inside the pulse bag dust collector to monitor the temperature inside the pulse bag dust collector; Dust concentration sensor: A dust concentration sensor is provided at the air outlet and air inlet of the pulse bag dust collector, respectively, to monitor the dust concentration at the air outlet and the dust concentration at the air inlet of the pulse bag dust collector; The material level meter is installed on the ash hopper or the surface of the filter bag; it is used to monitor the accumulation of dust in the ash hopper or on the surface of the filter bag; Pulse valve state sensor; set on the pulse valve; used to monitor the action state of the pulse valve; Flow sensor; installed at the air inlet and outlet; used to monitor the gas flow of the pulse bag dust collector; Vibration sensor; installed at the motor installation location of the pulse bag dust collector; used to monitor the vibration of the motor; The system also includes a communication status sensor; used to monitor the communication status between the edge device and the cloud device in real time; when the communication is interrupted, the alarm device is controlled to sound an alarm.

2. A dust removal monitoring system according to claim 1, characterized in that: Model training using enhanced algorithms, including: S801, initializing the parameters of the strategy network and the value network; S802. Under the current strategy, collect a batch of trajectory data, each trajectory containing a series of states, actions, rewards, and next states; S803, for each trajectory, calculate the reward at each time step t; S804, using the value network to estimate the value of each state; S805, calculating the advantage function of each time step; S806. For each time step t, update the policy network according to the PPO algorithm, use the gradient descent method to minimize the loss function, update the policy network parameters, and update the value network parameters; Steps S802 to S806 are repeated until a predetermined confidence level is reached.

3. A dust removal monitoring system according to claim 2, characterized in that: Before the cloud device performs data analysis on the monitoring data through the configured strategy optimization model, the cloud device also includes: performing anomaly detection through the configured anomaly detection model; when the anomaly detection model outputs information indicating anomaly; the cloud device sends an alarm message to the edge device so that the edge device takes an abnormal feedback instruction; the abnormal feedback instruction includes at least one of triggering an alarm and stopping the pulse bag dust collector chamber; The configuration process of the anomaly detection model includes: Acquire some historical normal data sent by the sensor array; set the parameters of the isolation forest; the parameters include the number of trees and the size of sub-sampling; Using the historical normal data to train an initial model to obtain an anomaly detection model; The detection method of the anomaly detection model includes: For each sample data, calculate its path length in the isolated forest; obtain an anomaly score based on the ratio of the expected value of the path length to the height of the tree; the path length is the number of splits required for the data point corresponding to the sample data to reach the leaf node from the root node; The sample data with anomaly scores higher than the threshold are judged as abnormal data; Output information representing the abnormality based on the type of abnormal data.

4. A dust removal monitoring system according to claim 3, characterized in that: The threshold is adjusted dynamically; the threshold adjustment process includes: S401, setting a time period T, and continuously collecting sample data within the time period T; S402, within the time period T, for each newly collected sample data, calculate its anomaly score in the isolation forest model; S403: Count the abnormal scores of all sample data in the current time period T that are greater than the threshold of this time period, and calculate the average value of all abnormal scores that are greater than the threshold of this event period. and standard deviation ;The initial threshold is preset; S404: Set a new threshold value to: , where k is a preset constant used to control the sensitivity of threshold adjustment; S405, applying the new threshold to anomaly detection in the next time period T; S401 to S405 are repeatedly executed.

5. A dust removal monitoring system according to claim 4, characterized in that: The monitoring system also includes a number of environmental sensors for monitoring the environment of the pulse bag filter; the environmental sensors include a temperature sensor and a humidity sensor; the environmental sensors communicate with the edge device; the defined state space of the strategy optimization model also includes environmental data; Wherein, the cloud device receives the environmental data and updates the state space S based on the environmental data; The strategy optimization model adjusts its model parameters according to the environmental data; the model parameters include weight coefficients and thresholds; The strategy optimization model outputs the adjusted strategy optimization results after updating the model parameters.

6. A dust removal monitoring system according to claim 5, characterized in that: The system also includes: a local storage module, which is used to store monitoring data and operation records; when the communication between the edge device and the cloud device is interrupted, the local storage module stores the offline data and automatically uploads it to the cloud device after the communication is restored.

Images