Intelligent inspection system and method for coal conveying belt of thermal power generating unit based on artificial intelligence

Abstract

The invention discloses a thermal power generating unit coal conveying belt intelligent inspection system and method based on artificial intelligence, and relates to the field of artificial intelligence and machine learning, and the method comprises the steps: collecting multi-source data, dividing a coal conveying belt into a grid coordinate system, distributing an initial risk score for each grid through an artificial intelligence model, and obtaining a preliminary inspection path; analyzing the time change trend of the coal flow speed through an artificial intelligence attention mechanism, processing the preliminary inspection path, marking a high-risk area, generating an optimized inspection path, and fusing multi-source data to obtain a comprehensive feature set; according to the method, the time change trend of the coal flow speed is analyzed through the attention mechanism, the routing inspection path is optimized according to the real-time coal flow trend and space correlation, blindness of artificial experience and hysteresis of static design are avoided, and efficient distribution of routing inspection resources is ensured.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and machine learning, and in particular to an intelligent inspection system and method for coal conveyor belts of thermal power units based on artificial intelligence. Background Art

[0002] The coal conveyor belt of thermal power units is the core component of material transportation in coal-fired power plants. The monitoring and maintenance of its operating status is directly related to production efficiency and safety and stability. In recent years, with the advancement of industrial automation and intelligent technology, the coal conveyor belt inspection technology has gradually evolved from traditional manual visual inspection to an automated monitoring method based on sensors and data analysis.

[0003] Existing technologies have made good progress in the field of coal conveyor belt inspection, but there are still many problems, such as dynamic adaptability and defect recognition accuracy. First, traditional inspection methods mostly use fixed thresholds or manually preset rules, which cannot accurately capture transition states or complex defects, resulting in high missed detection and false detection rates; in addition, traditional inspection path planning is basically static design or manual experience, lacking the ability to optimize in real time, making resource utilization efficiency very low. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an artificial intelligence-based intelligent inspection method for the coal conveyor belt of a thermal power unit to solve the problem that traditional inspection methods cannot accurately capture transition states or complex defects and lack the ability of real-time optimization.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides an intelligent inspection method for a coal conveyor belt of a thermal power unit based on artificial intelligence, which comprises: Collect multi-source data, divide the coal conveyor belt into a grid coordinate system, and assign an initial risk score to each grid through an artificial intelligence model to obtain a preliminary inspection path; The time variation trend of coal flow speed is analyzed through the artificial intelligence attention mechanism, the preliminary inspection path is processed, high-risk areas are marked, and an optimized inspection path is generated. At the same time, multi-source data is fused to obtain a comprehensive feature set; The optimized inspection route is sent to a drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, classifies the defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

[0007] As a preferred solution of the artificial intelligence-based intelligent inspection method for coal conveyor belts of thermal power units described in the present invention, the artificial intelligence model is built with recurrent neural networks and graph neural networks as the basic architecture, and outputs the initial risk score of each grid point after training is completed.

[0008] As a preferred solution of the artificial intelligence-based intelligent inspection method for coal conveyor belts of thermal power units of the present invention, wherein: according to the initial risk score of each grid point, the priority is divided in descending order; Get a preliminary inspection path based on priority.

[0009] As a preferred solution of the artificial intelligence-based intelligent inspection method for coal conveyor belts of thermal power units described in the present invention, it utilizes an artificial intelligence attention mechanism based on a recurrent neural network to receive real-time data from a coal flow velocity sensor, analyze the time series change trend of each grid point, adjust the initial risk score of each grid point, divide the high-risk areas, and generate an optimized inspection path.

[0010] As a preferred solution of the artificial intelligence-based intelligent inspection method for coal conveyor belts of thermal power units described in the present invention, the weights of multi-source data are adjusted, and the multi-source data are fused through weighted average fusion to obtain a comprehensive feature set.

[0011] As a preferred solution of the artificial intelligence-based intelligent inspection method for coal conveyor belts of thermal power units described in the present invention, the adjustment of the weights of multi-source data refers to adjusting the weights of multi-source data through preset weight adjustment rules and dynamic correction based on the stability of multi-source data.

[0012] As a preferred solution of the intelligent inspection method for coal conveyor belt of thermal power unit based on artificial intelligence described in the present invention, the optimized inspection path is sent to a drone equipped with a multi-spectral polarization imaging system, the drone collects three-band polarization light data, and classifies defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant, specifically including the following steps: The drone equipped with a multispectral polarization imaging system executes the optimized inspection path, transmits the collected three-band polarization light data to the polarization light model, and obtains the defect feature vector; The defect feature vector and the comprehensive feature set are fused by weighted average to obtain a fused feature set; The fused feature set is classified using a classification model, and a defect report is finally generated and transmitted to the thermal power plant control center.

[0013] In a second aspect, the present invention provides an intelligent inspection system for coal conveyor belts of thermal power units based on artificial intelligence, comprising: The path generation module collects multi-source data, divides the coal conveyor belt into a grid coordinate system, and assigns an initial risk score to each grid through an artificial intelligence model to obtain a preliminary inspection path; The path optimization module analyzes the temporal variation trend of coal flow speed through the artificial intelligence attention mechanism, processes the preliminary inspection path, marks high-risk areas, generates an optimized inspection path, and fuses multi-source data to obtain a comprehensive feature set; The inspection module sends the optimized inspection path to a drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, classifies the defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence as described in the first aspect of the present invention.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence as described in the first aspect of the present invention.

[0016] The beneficial effects of the present invention are as follows: by combining the artificial intelligence model of recurrent neural network and graph neural network, analyzing time trends and spatial associations, avoiding the limitation of a single threshold, being able to capture transition states and complex risks, and improving the accuracy of initial defect screening, and dynamically analyzing transition states and complex trends through the attention mechanism, improving the sensitivity to potential defects, making up for the static deficiency of artificial rules, and fusing multidimensional data to capture complex defects, avoiding the defect of ignoring superposition states due to a single threshold, and improving recognition accuracy. In addition, the attention mechanism analyzes the time variation trend of coal flow speed, achieving the optimization of the inspection path according to the real-time coal flow trend and spatial association, avoiding the blindness of artificial experience and the lag of static design, and ensuring the efficient allocation of inspection resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1This is a diagram of the intelligent inspection system for the coal conveyor belt of a thermal power unit based on artificial intelligence in this embodiment.

[0019] Figure 2 The figure is a flow chart of the intelligent inspection method of the coal conveyor belt of a thermal power unit based on artificial intelligence in this embodiment.

[0020] Figure 3 Schematic diagram of data fusion of the intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence in this embodiment.

[0021] Figure 4 This is a defect optimization diagram of the intelligent inspection method for the coal conveyor belt of a thermal power unit based on artificial intelligence in this embodiment. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0025] In this embodiment, refer to Figure 1~Figure 4 This embodiment provides an intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence, comprising the following steps: S1. Collect multi-source data and divide the coal conveyor belt into a grid coordinate system.

[0026] The specific steps include: The specific parameters of the coal conveyor belt are determined in the thermal power plant. For example, the length is 100 meters, the width is 1.5 meters, and the operating environment is an open-air or semi-enclosed coal bunker. A sensor deployment plan is formulated based on the specific parameter information of the coal conveyor belt. A group of sensors is installed every 5 meters, for a total of 20 groups, covering the entire length of the belt.

[0027] After determining that a group of sensors would be installed every 5 meters, the installation of sensors began, including high-definition visible light cameras, infrared cameras, integrated multi-band cameras (also known as multi-spectral polarization imaging systems, which complete a three-band polarized light scan every 2 seconds to generate multi-layer images for detecting the reflection and scattering characteristics of the belt surface and interior, especially sub-surface defects, installed on inspection drones), coal flow velocity sensors, voiceprint sensors and environmental monitors.

[0028] Use the installed sensors to start collecting data. Use the coal flow velocity sensor to collect real-time coal flow velocity values ​​once per second; use a high-definition visible light camera to collect high-definition images of the coal conveyor belt surface (continuous video stream) at a frequency of 30 frames per second; use an infrared camera to collect thermal images of the belt surface at a frequency of once per second; use a voiceprint sensor to collect vibration waveforms and noise decibels at a frequency of once every 10 milliseconds; use an integrated multi-band camera to collect three-band polarized light images (multi-layer image data) at a frequency of once every 2 seconds; use an environmental monitor to collect dust concentration, humidity, and vibration intensity at a frequency of once every minute. The edge computing node receives real-time data from all sensors through the industrial Ethernet to form a continuous multi-source data stream, such as recording a coal flow velocity of 2.3 meters per second and a dust concentration of 500 micrograms per cubic meter at a certain moment.

[0029] S1.1. The edge computing nodes are divided into a grid coordinate system according to the length and width of the coal conveyor belt. Taking a 100-meter-long and 1.5-meter-wide coal conveyor belt as an example, it is divided into a 100×5 grid coordinate system, with each grid on the horizontal axis being 1 meter and each grid on the vertical axis being 0.3 meters, for a total of 500 grid points. Each grid point is bound to the nearest sensor group data. For example, (50,3) corresponds to the 50th meter and the third width segment.

[0030] S2. Assign an initial risk score to each grid through the artificial intelligence model to obtain a preliminary inspection path.

[0031] The specific steps include: First, prepare to train the artificial intelligence model. The preparation process before training is as follows: Collect historical coal flow speed values, such as the speed values ​​recorded per second in the past 6 months, covering normal operation, sudden increase in coal volume, no-load and other states. Collect historical environmental data such as dust concentration, humidity, and vibration intensity to simulate different working conditions. Collect coal conveyor belt status labels, and the actual belt status recorded by manual inspection or equipment logs, such as coal powder accumulation, belt deviation or wear at a certain time period and location, as supervision signals for the artificial intelligence model. Sort the historical coal flow speed values, historical environmental data, and coal conveyor belt status labels by timestamp and location (such as grid coordinates (50,3)) to form a sample set of time series and spatial distribution.

[0032] The artificial intelligence model of this solution is a hybrid model that combines recurrent neural networks (RNN) and graph neural networks (GNN). Because recurrent neural networks are good at processing time series data, they can capture the changing trend of coal flow speed over time, such as sudden increase or continuous low speed. Graph neural networks are suitable for analyzing spatial relationships and can identify the association between grid points, such as whether the high flow of (50,3) and (51,4) forms a local high-load area.

[0033] The sample set is divided into a training set (80%) and a validation set (20%). For example, the training set contains 5 months of data and the validation set contains 1 month of data. Set the training goal: The AI ​​model predicts the risk score of each grid point and makes the predicted risk score of each grid point as close as possible to the actual risk level of the manual mark, for example, (50,3) is predicted to be 0.8 when piled up. Adjust the AI ​​model parameters through repeated iterations, for example, input a 5-minute window of coal flow velocity sequence and corresponding grid state in each round of training, and optimize the sensitivity of the AI ​​model to time changes and spatial distribution. Finally, test the performance of the AI ​​model by using the validation set, for example, input a certain period of historical data to check whether the AI ​​model correctly identifies high-risk areas (such as (50,3) with a sudden increase in coal flow) and low-risk areas (such as (80,1) with no load). If a misjudgment is found, such as rating a normal area as high risk, adjust the AI ​​model weight until the AI ​​model accuracy reaches more than 90%. Finally, save the trained model parameters and deploy them to the edge computing node.

[0034] Preferably, the artificial intelligence model of the present invention combines recurrent neural networks and graph neural networks, and learns to identify the temporal changes and spatial distribution of coal flow by training historical data, wherein the recurrent neural network captures temporal trends and indicates potential accumulation risks; the graph neural network analyzes spatial associations to increase the accuracy of risk scoring. Moreover, the artificial intelligence model of the present invention not only relies on a single threshold (such as speed>2 meters / second), but also integrates temporal and spatial characteristics to avoid misjudgment, such as distinguishing between stable operation and unstable fluctuations.

[0035] S2.1. The trained AI model receives real-time coal flow velocity values ​​from the coal flow velocity sensor, such as a measurement value once per second, covering all 500 grid points, such as (50,3) is 2.3 m / s, (80,1) is 0.4 m / s. The real-time coal flow velocity value data is organized in time order and spatial position to form an input sequence. The AI ​​model processes the coal flow velocity of each grid point in a time series, such as extracting the velocity record of (50,3) in the past 10 minutes, which increases from 1.5 m / s to 3 m / s. The RNN part of the AI ​​model analyzes this sequence and identifies the trend of change. For example, if the speed is continuously higher than 2 m / s or increases rapidly, it indicates that the coal flow load has increased, which may lead to accumulation and is marked as a high-risk mode. If the speed is continuously lower than 0.5 m / s, for example, (80,1) maintains 0.4 m / s for 10 minutes, it indicates an unloaded state, which may cause wear and tear, and is marked as a low-risk mode but needs attention. This time window can be adjusted dynamically, for example, set to 5 minutes or 15 minutes according to the frequency of coal flow fluctuations. Customize according to specific needs.

[0036] At the same time, the GNN part of the AI ​​model processes the spatial relationship of the grid points, such as checking whether the 2.3 m / s of (50,3) is associated with the 2.8 m / s of the adjacent grid point (51,4). For example, if multiple adjacent grid points (such as (50,3), (51,4), and (52,3)) show high speeds at the same time, it indicates that the coal flow in the local area is concentrated and the risk of accumulation is high. The AI ​​model increases the risk score of the adjacent grid points. However, if a certain area (such as (80,1) to (85,1)) has a continuous low speed and is isolated, it indicates that the unloaded is evenly distributed and the risk is low. The AI ​​model will lower the risk score. The distance and connectivity between grids are considered in spatial analysis. For example, when the speed difference between adjacent points is less than 0.5 m / s, it is considered a consistent pattern.

[0037] The artificial intelligence model integrates the results of temporal changes and spatial distribution, and outputs a risk score for each grid point, which is divided into three types: high-risk mode, low-risk mode, and intermediate mode.

[0038] A high-risk mode refers to a risk score greater than or equal to 0.8. For example, if the speed at the grid point (50,3) increases suddenly and the neighboring points have synchronous high flow, the risk score is set to 0.8, indicating accumulation or congestion risk.

[0039] Low risk mode means that the risk score is less than or equal to 0.2. For example, if the grid point (80,1) has a long period of low speed and no adjacent anomalies, the risk score is set to 0.2, indicating no load or normal operation.

[0040] The risk score between 0.3-0.6 is an intermediate mode, for example, the speed fluctuates between 0.5-2 m / s and there is no obvious clustering in the spatial distribution. At this time, the risk score will be adjusted according to the specific trend. The so-called adjustment according to the specific trend means that the risk score is not fixed at a certain value between 0.3-0.6, but will be dynamically adjusted according to the specific change trend of the coal flow speed. The specific trend refers to the pattern in the time series, such as whether the speed is gradually increasing, gradually decreasing, or fluctuating irregularly, as well as the duration and amplitude of the pattern changes in the time series. The purpose of the adjustment is to make the score more accurately reflect the potential risks, rather than simply applying a static range, because when the risk score is between 0.3-0.6, the coal conveyor belt state is very likely to be in a transition stage, which may be normal operation or a precursor to problems. Fixed scores cannot accurately reflect this uncertainty, so they need to be adjusted dynamically according to trends.

[0041] The logic of the adjustment is: Trend 1: The speed gradually increases but does not exceed the limit. If the speed of a grid point (50,3) slowly increases from 0.5 m / s to 1.8 m / s within 5 minutes, and the speed of the adjacent grid points also increases slightly (for example, 1.3 m / s), the artificial intelligence model will think that the coal flow load may be increasing. Although it has not reached a high risk (>2 m / s), the risk is slightly higher, so the score may be adjusted to 0.5 or 0.6, indicating that a little attention is needed. Trend 2: The speed is stable at the median. If the speed of (50,3) is stable at around 1.2 m / s for a long time, the fluctuation range is only ±0.2 m / s, and the adjacent grid points are similar, in this case the artificial intelligence model will think that the operation is smooth and there are no obvious problems. The risk score may be adjusted to 0.3 or 0.4, indicating a lower risk. Trend 3: The speed fluctuates irregularly. If the speed of (50,3) frequently jumps between 0.8m / s and 1.6m / s, and the spatial distribution is irregular (the speed difference between adjacent points is large), the AI ​​model may consider that there is instability. Although it is not very serious, it needs to be vigilant, and the score may be set to 0.5, which is in the middle. Trend 4: Approaching the boundary but not breaking through. If the speed of (50,3) occasionally approaches 2m / s (for example, 1.9m / s) but quickly falls back, and the spatial distribution is uniform, the AI ​​model will adjust the risk score according to the fallback speed and duration. For example, if it falls back quickly, it will be set to 0.4, and if it continues to approach, it may rise to 0.6, indicating that the potential risk is slightly higher.

[0042] Preferably, for the intermediate mode, the artificial intelligence model of the present invention will dynamically adjust the score according to the specific trend, for example, when the speed of (50,3) increases from 0.5 m / s to 1.8 m / s, it is adjusted to 0.6, and when it stabilizes at 1.2 m / s, it is adjusted to 0.3. This flexible adjustment reflects the transitional state of the coal conveyor belt, ensuring that the risk score is closer to the actual risk. It can automatically adapt to different working conditions, and improve the precision and practicality of the risk score.

[0043] Based on the output risk score results, an inspection priority list (also known as the preliminary inspection path) is automatically generated to guide the inspection frequency and order of inspection equipment (such as drones) along the belt. In layman's terms, according to the risk score of each grid point, the grid points are divided into different priorities and the corresponding inspection intervals are assigned. For example, high-risk grid points need to be inspected more frequently, while low-risk grid points can be inspected less frequently.

[0044] It is further explained that compared with fixed thresholds, manual rules or single-dimensional methods, this solution has comprehensive improvements in dynamics, intelligence, adaptability and real-time performance, significantly reducing the missed detection rate (such as potential risks of intermediate modes) and false detection rate (such as misjudgment of normal fluctuations), providing more efficient and intelligent guarantees for the safe operation of coal conveyors in thermal power plants.

[0045] S3. Analyze the temporal variation trend of coal flow speed through the artificial intelligence attention mechanism, process the preliminary inspection path, mark high-risk areas, and generate an optimized inspection path.

[0046] The specific steps include: An artificial intelligence attention mechanism module is trained based on the recurrent neural network of the artificial intelligence model, which is specifically responsible for dynamically analyzing the time variation trend of the coal flow velocity and optimizing it based on the preliminary inspection path.

[0047] An attention layer is added on top of the recurrent neural network to highlight key moments in the time series. For example, when analyzing the 10-minute sequence of (50,3), the AI ​​attention mechanism will assign higher weights to the moment when the speed suddenly increases to 3 meters per second, rather than treating the entire sequence equally. The attention mechanism simulates the human ability to focus on changes. The input is the coal flow speed time series at each grid point, such as the 10-minute record of (50,3); the output is a trend score (range 0-1), which reflects the risk level of dynamic changes, such as a sudden increase trend is rated 0.8 and a slow trend is rated 0.2.

[0048] The AI ​​attention mechanism module is trained through supervised learning. The training goal is to predict the trend score of each grid point so that the predicted trend score of each grid point is as close to the real trend label as possible. For example, the sudden increase trend target of (50,3) is 0.8, and the slow trend target of (80,1) is 0.2.

[0049] The trained AI attention mechanism module receives real-time data from the coal flow velocity sensor, such as the measurement value once per second, such as 2.3 m / s for (50,3) and 0.4 m / s for (80,1). The AI ​​attention mechanism module organizes the real-time data of the coal flow velocity sensor in time series, for example, extracting the velocity record of (50,3) in the past 10 minutes, and finding that it has increased from 1.5 m / s to 3 m / s. Start analyzing the change trend and adjusting the initial risk score: If the speed is continuously higher than 2 m / s or increases rapidly (such as (50,3)), it indicates that the coal flow load has increased, which may lead to accumulation, and the initial risk score is adjusted to 0.8 points. If the speed is continuously lower than 0.5 m / s (such as 0.4 m / s for (80,1) for 10 consecutive minutes), it indicates an unloaded state, and the initial risk score is adjusted to 0.2 points. If it is an intermediate trend (such as 0.5-2 m / s fluctuation), it is adjusted according to the specific mode, such as slowly rising to 1.8 m / s.

[0050] Adjustment based on specific patterns means that when the coal flow speed is in the middle range of 0.5-2 m / s, the AI ​​attention mechanism module will not simply assign a value, but analyze the specific pattern of speed change (such as rising, stable, and fluctuating), and dynamically adjust the trend score or inspection frequency according to the trend, amplitude, and duration of the change pattern of the coal flow speed time series. For example, the speed of the grid point (60,2) slowly rises to 1.8 m / s, and the score is adjusted from the initial value to 0.6. This method ensures accurate evaluation of the transition state and improves the pertinence and efficiency of inspection path optimization.

[0051] While analyzing the time trend, the AI ​​attention mechanism module combines spatial associations to check the speed changes of adjacent grid points. For example, if the speed of (51,4) increases synchronously to 2.8 m / s, which is consistent with the sudden increase trend of (50,3), it is judged as a local high-load area, and the initial risk score of (51,4) is further adjusted to 0.75 to strengthen the high-risk assessment.

[0052] Preferably, the AI ​​attention mechanism module accurately identifies and marks high-risk areas by analyzing the temporal trend and spatial correlation of coal flow speed. This targeted adjustment enables the inspection path to prioritize potential problem points, such as accumulation or blockage risk areas.

[0053] Based on the analysis of temporal changes and spatial correlation, the artificial intelligence attention mechanism module dynamically adjusts the preliminary inspection path. Re-evaluate the risk priority of each grid point: For (50,3), because the initial risk score is adjusted to 0.8 and the load is continuously high, the high-risk status of grid point (50,3) (i.e., high-risk area) is confirmed, and the 15-minute inspection frequency is maintained or strengthened. For (80,1), the initial risk score is adjusted to 0.2 and there are no abnormalities. The priority of grid point (80,1) is reduced, and the inspection frequency is adjusted from 20 minutes to 30 minutes. For intermediate mode grid points (such as (60,2, with an initial risk score of 0.6), adjust the frequency of drone inspections according to the trend of the temporal change pattern of the coal flow velocity (such as rising or stable), for example, fine-tuning from 20 minutes to 18 minutes or 22 minutes.

[0054] After adjusting the priority, an optimized inspection path is planned for the high-risk area. For example, the flight time of the inspection equipment drone from (1,1) to (50,3) is calculated. It takes 24.5 seconds to fly 49 meters at a speed of 2 meters per second. Adding the hovering and collection time, it is planned to complete a high-risk area cycle every 14 minutes. The optimized inspection path covers (50,3) and (51,4) from (1,1) in sequence. This adjustment process is set to dynamic refresh mode, and re-analyze every 5 minutes based on the latest coal flow speed data. For example, if (80,1) remains at a low speed, it will continue to be skipped. If (50,3) and (51,4) continue to have high flow, they will maintain priority coverage. The optimized inspection path is sent to the drone via the wireless network to ensure real-time execution.

[0055] Preferably, by dynamically adjusting the inspection frequency, resources are concentrated in high-risk areas, while redundant inspections in low-risk areas are reduced. For example, high-risk areas are inspected about 4 times per hour (60 minutes ÷ 14 minutes ≈ 4.3 times), and low-risk areas are reduced to 2 times (60 minutes ÷ 30 minutes = 2 times), which significantly improves resource utilization. In addition, the inspection path is refreshed every 5 minutes based on the latest coal flow speed data to ensure that the inspection is always synchronized with the current status. This saves resources while ensuring coverage of high-risk points.

[0056] S4. Fuse multi-source data to obtain a comprehensive feature set.

[0057] The specific steps include: Extract high-definition images of the coal conveyor belt surface collected by a high-definition visible light camera, thermal images of the belt surface collected by an infrared camera, and voiceprint data collected by a voiceprint sensor.

[0058] Read the environmental data collected by the environmental monitor. The environmental data reflects the environmental conditions. For example, if the dust concentration reaches 600 micrograms / cubic meter, it is already higher than the normal range of 500 micrograms / cubic meter; and if the vibration intensity reaches 6 meters / second², it indicates that the machine is running violently. Environmental conditions will affect the reliability of sensor data. For example, high dust may blur the visible light image (referring to the high-definition image of the coal conveyor belt surface collected by the high-definition visible light camera), and high vibration may amplify the voiceprint signal. Reading environmental data provides a basis for adjusting the weight in the next step to ensure that the fusion result adapts to the actual working conditions.

[0059] After reading the environmental data, the weights of each sensor are adjusted according to the adjustment rules.

[0060] For example, the adjustment rules are as follows: the default weight is 33% for each sensor (1 / 3 for visible light, infrared, and voiceprint). If it is a high dust scene, when the dust concentration is 600 micrograms / cubic meter (>500 micrograms / cubic meter), the visible light image may be distorted due to occlusion, and the weight is reduced to 20%; the infrared thermal image is less affected by dust, and the weight is increased to 40%; the voiceprint weight is increased to 35%. If it is a high vibration scene, when the vibration intensity is 6 meters / second², the voiceprint signal can better reflect mechanical abnormalities, and the weight is further increased to 40%, the infrared is reduced to 35%, and the visible light is maintained at 20%.

[0061] For example, in the current environment (dust 600 μg / m3, vibration 6 m / s²), the weights are adjusted to 20% for visible light, 35% for infrared, and 40% for voiceprint.

[0062] Preferably, the weights are adjusted dynamically to make the fusion results more dependent on reliable data, such as giving priority to infrared and voiceprints under high dust conditions, and highlighting voiceprints under high vibrations, to avoid unreliable data interfering with the evaluation.

[0063] Check the data stability of each sensor and analyze the fluctuation of sensor data within a short time window (1 minute). For example, if the frequency of the voiceprint of (50,3) jumps irregularly (which may be noise or sensor failure), the data is considered unstable. If an anomaly is detected, reduce the weight of unreliable data (such as voiceprint from 40% to 10%), and increase the weight of other stable data (such as infrared from 35% to 45%, visible light maintained at 20%). By checking the stability, the influence of abnormal data is eliminated, ensuring that the fusion result is based on reliable data and improving the accuracy of the evaluation.

[0064] S4.1. Normalize the weighted data (normalize the weighted data to the range of 0-1), and then fuse them by weighted average. For example, integrate the multimodal data of (50,3). The weighted data of (50,3) is 20% visible light (surface texture features, such as wear marks), 45% infrared (high temperature point, 48°C, high weight, large contribution), and 10% voiceprint (vibration mode, abnormal increase, low weight, small impact). After fusion, a multidimensional feature vector (also known as a comprehensive feature set) is generated, describing the status of (50,3) as "abnormal high temperature and increased vibration", indicating potential deviation or roller problems. This comprehensive feature set is updated once a second to ensure that it can reflect the latest status.

[0065] Preferably, the fusion of multi-source data forms a comprehensive description of the coal conveyor belt status, which can reveal more complex problems than a single data, and achieve comprehensive and accurate monitoring of the coal conveyor belt status, high reliability evaluation, real-time dynamic update, efficient inspection and maintenance coordination and resource utilization optimization, significantly improving the intelligence level of inspection.

[0066] S5. Send the optimized inspection path to the drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, and classifies the defects of each grid point through the polarization light model combined with the comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

[0067] The specific steps include: S5.1. The optimized inspection path is transmitted to two inspection drones. The coal conveyor belt is 100 meters long and divided into 500 grid points (100×5 grid system). To improve efficiency, the inspection task is divided into two sections. The first drone is responsible for the first 50 meters (grid (1,1) to (50,5), focusing on covering the high-risk point (50,3)), and the second is responsible for the last 50 meters (grid (51,1) to (100,5), focusing on covering (51,4)). This partitioning strategy ensures that the two drones work in parallel and cover the entire belt. After receiving the optimized inspection path, the two drones are started, and the built-in navigation adjusts the direction according to the inspection path coordinates to ensure movement along the optimized path.

[0068] After the inspection drone reaches the high-risk grid point, it hovers and activates the multispectral polarization imaging system to start scanning the coal conveyor belt. The multispectral polarization imaging system integrates three cameras: a visible light lens (used to capture the reflected light on the surface of the coal conveyor belt), a near-infrared lens (detecting shallow scattered signals), and a short-wave infrared lens (penetrating into the interior of the coal conveyor belt). A polarization filter is also added to measure the polarization angle and intensity of light to enhance the detection of surface characteristics (such as texture, cracks) and internal defects. The three cameras output images separately to form a multi-layer data structure (that is, three-band polarization light data). For example, the visible light image of (50,3) shows wear, the near-infrared image highlights scattering anomalies, and the short-wave infrared image reveals deep problems.

[0069] S5.2 To analyze the three-band polarized light data, a polarized light model is established. First, extract the three-band polarized light data from the historical inspection records; the three-band polarized light data in the historical inspection records are collected by the multispectral polarization imaging system carried by the drone, covering normal operation and abnormal conditions. Collect at least 6 months to 1 year of data, including thousands of sets of images, such as near-infrared polarized light images of (50,3) grid points, visible light surface reflection images, etc. At the same time, collect defect annotations for each set of images, and record the environmental conditions at the time of collection, such as dust concentration (for example, 600 micrograms / cubic meter) and vibration intensity (for example, 6 meters / second²) as auxiliary information to help the polarized light model understand the impact of the environment on the polarized light data. Standardize the three-band polarized light data to the range of 0-1 to avoid the magnitude difference of different bands affecting the training. Then divide it into a training set (80%, such as 8 months of data) and a validation set (20%) to ensure the independence of training and evaluation.

[0070] CNN (convolutional neural network) is used as the basic architecture because CNN is good at extracting spatial features of images and is suitable for processing multi-spectral polarized light data; the input layer is set to receive three-band polarized light images, such as three layers of data (visible light, near infrared, short-wave infrared) of (50,3), each layer is a two-dimensional matrix (for example, 256×256 pixels), and the input dimension is 3 channels. The convolution layer extracts reflection characteristics (such as surface texture) and scattering characteristics (such as near-infrared intensity enhancement) through multi-layer convolution and pooling operations. For example, the convolution kernel captures the abnormal area of ​​the (50,3) near-infrared image and generates a feature map. In addition, prior knowledge, such as the photon scattering equation (based on Monte Carlo simulation), is embedded into the polarized light model to enhance the polarized light model's ability to interpret the scattering signal and help distinguish the scattering patterns of internal cracks and surface foreign objects. Finally, the output layer outputs a defect feature vector, such as the internal stress or crack signal of (50,3), which contains the defect type and quantitative parameters (such as scattering intensity).

[0071] The training set is input into the polarization model in batches, for example, 32 sets of images per batch. The polarization model processes the (50,3) three-band data, and the convolution layer extracts features, such as the scattering enhancement area in the near infrared, and outputs the predicted crack signal.

[0072] The cross entropy loss is used to measure the difference between the predicted features (such as the crack depth of 4 mm) and the actual annotation. For example, if the predicted depth is 3.8 mm and the actual depth is 4 mm, the error is calculated and back-propagated. The Adam optimizer is used to adjust the polarized light model parameters. The learning rate is initially set to 0.001 and decays dynamically with iterations. The weights are updated at each iteration to make the prediction closer to the actual defect. In addition, scattering physical constraints are added to the loss function, such as the positive correlation between the near-infrared scattering intensity and the crack depth, to ensure that the output of the polarized light model conforms to the optical law. Until the loss converges. Finally, input the validation set data, such as the polarized light image of (50,3), to check whether the polarized light model correctly predicts the internal crack with a depth of 4 mm. Calculate the accuracy (such as the correctness of the defect type) and the mean square error (such as the depth prediction error). Until the accuracy of the polarized light model reaches 90%, for example, 90% of the grid points have the correct defect type and the depth error is less than 0.5 mm, then save the optimal parameters.

[0073] Preferably, the comprehensiveness and accuracy of defect detection are significantly improved by collecting three-band polarized light data through a multi-spectral polarization imaging system and combining it with precise analysis of the polarized light model. This multi-dimensional data analysis is more comprehensive than the traditional single sensor method and provides a reliable basis for the diagnosis of complex defects (such as deviation or roller problems).

[0074] S5.3. After the trained polarized light model outputs the defect feature vector (for example, the internal stress or crack signal of (50,3)), it is fused with the comprehensive feature set through weighted averaging. The defect feature vector provides internal defect clues (such as near-infrared scattering enhancement), and the comprehensive feature set provides external state information (high temperature and vibration). The two are combined to generate a more comprehensive feature vector (that is, the fused feature set). For example, the fusion result of (50,3) may indicate that high temperature and vibration lead to increased internal stress, and the polarized light anomaly is further confirmed as a crack risk. This integration of multi-source information improves the accuracy of defect judgment and avoids the misjudgment that may occur from a single data (such as polarized light only).

[0075] The classification model uses a support vector machine, and the fused feature set is input into the support vector machine to determine the defect type and severity of each grid point. Before this, a defect library needs to be preset. This defect library is a classification set that contains common coal conveyor belt defect types, such as "internal cracks", "surface wear", "deviation", "fiber breakage", "accumulation blockage", etc. Each defect type is associated with a set of typical features for comparison and classification matching between the support vector machine and the fused feature set. For example, for internal cracks, the polarized light scattering intensity is enhanced, accompanied by high temperature and vibration; for surface wear, the visible light reflectivity is reduced, and there is no obvious internal scattering change; for deviation, the vibration frequency is abnormal, and the polarized light shows local stress concentration.

[0076] It is further explained that the weighted average method of integrating the comprehensive feature set and polarized light data improves the reliability of defect assessment. The comprehensive feature set integrates external conditions (such as high temperature 48°C and enhanced vibration of (50,3)), and the polarized light data provides internal clues (such as scattering intensity 0.8). The fusion feature set generated by the fusion of the two can comprehensively reflect the synergistic effects of high temperature, vibration and cracks, avoiding the judgment bias that may be caused by single modal data, and is closer to actual working conditions.

[0077] The fused feature set (e.g., abnormal high temperature with enhanced vibration and polarized light scattering of (50,3)) is compared with the feature templates in the defect library one by one using Euclidean distance to find the most matching defect feature type. The type with the highest matching degree is selected as the preliminary classification result, for example, (50,3) is determined to be an internal crack.

[0078] After feature matching, the support vector machine evaluates the severity based on the preset defect threshold, further confirms the classification and quantifies the defects. For example: for internal cracks, the scattering intensity > 0.7 and the depth > 1 mm are the basis for confirmation, and the depth is estimated by scattering value mapping. (Mapping estimation is based on historical data to establish a mapping relationship between scattering intensity and depth. The scattering intensity is measured on crack samples with known depths. For example, a scattering intensity of 0.5 refers to a depth of about 1 mm, a scattering intensity of 0.7 refers to a depth of about 2 mm, and a scattering intensity of 0.8 refers to a depth of about 4 mm); for surface wear, a reflectivity drop of > 10% without high temperature is mild, and a drop of > 30% is moderate. (The preset defect threshold can be customized according to personal needs and actual business needs).

[0079] Example, (50,3), scattering intensity 0.8 (>0.7), depth estimated 4 mm (>1 mm), internal cracks confirmed, severity is moderate to severe. (80,1), reflectivity decreased by 15%, no scattering anomaly, surface wear confirmed, severity is mild.

[0080] The severity of each grid point is checked according to the preset severity threshold (e.g. crack depth 3 mm). For example, for (50,3), the crack depth of 4 mm exceeds the severity threshold, and the edge computing node automatically triggers the early warning mechanism. However, for the mild wear of (80,1), because it does not reach the verification threshold, it is only recorded without triggering an early warning, which reflects the flexibility of priority management.

[0081] It is further explained that the support vector machine and the preset defect library are used for classification, which enables the rapid quantification and priority management of defects. For (50,3), the scattering intensity of 0.8 and the depth of 4 mm exceed the 3 mm threshold, triggering an early warning, prompting the risk of medium to severe cracks in a timely manner; while the mild wear (reflectivity decreases by 15%) of (80,1) is only recorded without an alarm, reflecting flexibility and resource optimization.

[0082] S5.4. Generate a defect report, including the defect location, such as grid (50,3), type of internal crack, severity of 4 mm depth, recommended repair time of replacing the belt within 72 hours, and format of text plus coordinate annotation. The detailed report provides maintenance personnel with a clear action guide to ensure that defect handling is based on evidence.

[0083] The analysis results and warning status of all grid points are integrated into a complete data packet. For example, the internal crack at (50,3), 4 mm in depth, warning trigger and the surface wear at (80,1), light, no warning are packaged with timestamp and priority tag. Then it is transmitted to the thermal power plant control center via industrial Ethernet.

[0084] This embodiment also provides an intelligent inspection system for coal conveyor belts of thermal power units based on artificial intelligence, including: The path generation module collects multi-source data, divides the coal conveyor belt into a grid coordinate system, and assigns an initial risk score to each grid through an artificial intelligence model to obtain a preliminary inspection path; The path optimization module analyzes the temporal variation trend of coal flow speed through the artificial intelligence attention mechanism, processes the preliminary inspection path, marks high-risk areas, generates an optimized inspection path, and fuses multi-source data to obtain a comprehensive feature set; The inspection module sends the optimized inspection path to a drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, classifies the defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

[0085] This embodiment also provides a computer device, which is suitable for the case of an intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence as proposed in the above embodiment.

[0086] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0087] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence is implemented as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0088] In summary, the present invention combines the artificial intelligence models of recurrent neural networks and graph neural networks to analyze time trends and spatial associations, avoiding the limitations of a single threshold, capturing transition states and complex risks, and improving the accuracy of initial defect screening. The attention mechanism dynamically analyzes transition states and complex trends, improving the sensitivity to potential defects, making up for the static deficiencies of artificial rules, and integrating multidimensional data to capture complex defects, avoiding the defects of ignoring superposition states due to a single threshold, and improving recognition accuracy. In addition, the attention mechanism analyzes the time variation trend of the coal flow speed, achieving the optimization of the inspection path according to the real-time coal flow trend and spatial association, avoiding the blindness of artificial experience and the lag of static design, and ensuring efficient allocation of inspection resources.

[0089] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent inspection method for coal conveyor belts of thermal power units based on artificial intelligence, characterized in that: include, Collect multi-source data, divide the coal conveyor belt into a grid coordinate system, and assign an initial risk score to each grid through an artificial intelligence model to obtain a preliminary inspection path; The time variation trend of coal flow speed is analyzed through the artificial intelligence attention mechanism, the preliminary inspection path is processed, high-risk areas are marked, and an optimized inspection path is generated. At the same time, multi-source data is fused to obtain a comprehensive feature set; The optimized inspection route is sent to a drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, classifies the defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

2. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 1, characterized in that: The artificial intelligence model is built on a recurrent neural network and graph neural network infrastructure, and outputs an initial risk score for each grid point after training is completed.

3. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 2, characterized in that: According to the initial risk score of each grid point, priority is assigned in descending order; Get a preliminary inspection path based on priority.

4. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 3, characterized in that: Utilizing the artificial intelligence attention mechanism based on recurrent neural networks, the real-time data of the coal flow velocity sensor is received, the time series change trend of each grid point is analyzed, the initial risk score of each grid point is adjusted, and high-risk areas are divided to generate an optimized inspection path.

5. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 4, characterized in that: The weights of multi-source data are adjusted, and the multi-source data are fused through weighted average fusion to obtain a comprehensive feature set.

6. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 5, characterized in that: The adjusting of the weights of the multi-source data refers to adjusting the weights of the multi-source data through a preset weight adjustment rule and a dynamic correction based on the stability of the multi-source data.

7. The method for intelligent inspection of coal conveyor belts of thermal power plants based on artificial intelligence according to claim 6, characterized in that: The optimized inspection path is sent to a drone equipped with a multispectral polarization imaging system. The drone collects three-band polarization data, classifies defects at each grid point through a polarization model combined with a comprehensive feature set, and generates a defect report that is pushed to the thermal power plant control center. The specific steps include the following: The drone equipped with a multispectral polarization imaging system executes the optimized inspection path, transmits the collected three-band polarization light data to the polarization light model, and obtains the defect feature vector; The defect feature vector and the comprehensive feature set are fused by weighted average to obtain a fused feature set; The fused feature set is classified using a classification model, and a defect report is finally generated and transmitted to the thermal power plant control center.

8. An intelligent inspection system for coal conveyor belts of thermal power plants based on artificial intelligence, based on an intelligent inspection method for coal conveyor belts of thermal power plants based on artificial intelligence according to any one of claims 1 to 7, characterized in that: include, The path generation module collects multi-source data, divides the coal conveyor belt into a grid coordinate system, and assigns an initial risk score to each grid through an artificial intelligence model to obtain a preliminary inspection path; The path optimization module analyzes the temporal variation trend of coal flow speed through the artificial intelligence attention mechanism, processes the preliminary inspection path, marks high-risk areas, generates an optimized inspection path, and fuses multi-source data to obtain a comprehensive feature set; The inspection module sends the optimized inspection path to a drone equipped with a multi-spectral polarization imaging system. The drone collects three-band polarization light data, classifies the defects of each grid point through a polarization light model combined with a comprehensive feature set, and generates a defect report, which is pushed to the control center of the thermal power plant.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent inspection method for the coal conveyor belt of a thermal power unit based on artificial intelligence according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the intelligent inspection method for the coal conveyor belt of a thermal power unit based on artificial intelligence according to any one of claims 1 to 7 are implemented.

Images