Method, system and medium for recognizing abnormity of coarse grid equipment based on AI model combination

Abstract

The invention relates to the technical field of sewage treatment equipment monitoring, and discloses a method, a system and a medium for recognizing abnormity of coarse screen equipment based on AI model combination. The method comprises the following steps: inputting an optimized coarse grid image of coarse grid equipment into an optimized target detection model for processing to obtain target position data corresponding to a detection target, segmenting the optimized coarse grid image according to the target position data to obtain a target image, extracting to obtain image feature data, and processing according to the image feature data to obtain the target image. The method comprises the following steps: acquiring an operation initial evaluation index of coarse grid equipment, acquiring an initial evaluation state of the coarse grid equipment through threshold comparison, inputting image feature data into a preset anomaly detection model for evaluation processing if the coarse grid equipment is suspected to be abnormal, acquiring an operation comprehensive measurement index of the coarse grid equipment, and finally determining a final evaluation state of the coarse grid equipment through threshold comparison. Therefore, through combined application of the large and small models, accurate judgment of the abnormality of the coarse grid equipment is realized, and the misjudgment rate is reduced while the judgment efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment equipment monitoring, and more specifically, to a method, system and medium for identifying abnormal conditions of coarse grid equipment based on an AI model combination. Background Art

[0002] The coarse grid equipment undertakes the important task of intercepting large suspended solids and floating objects in sewage. The stability of its operating state directly affects the effect and efficiency of the subsequent treatment process. The traditional monitoring method of coarse grid equipment mainly relies on manual regular inspections; however, this method has relatively low inspection efficiency, a relatively long inspection cycle, and it is difficult to achieve real-time monitoring, which may lead to the failure to detect equipment abnormalities in a timely manner. Manual judgment is prone to errors, affecting the accuracy of judgment. Moreover, the inspection personnel need to be in close contact with sewage and equipment, posing safety risks such as slipping, poisoning, and electric shock, and the operation is highly dangerous. The intelligent monitoring technology based on computer vision and deep learning can achieve real-time and accurate monitoring of equipment, improving the intelligent management level of sewage treatment plants; related technologies have been explored in some fields. For example, a sewage treatment coarse grid equipment operation monitoring and control system involved in Patent CN202411283204.0, but the accuracy of abnormal identification and determination of coarse grid equipment needs to be further optimized. Summary of the Invention

[0003] In view of the above problems, the purpose of the present invention is to provide a method, system and medium for identifying abnormal conditions of coarse grid equipment based on an AI model combination.

[0004] To solve the above technical problems, the technical solution of the present invention is as follows:

[0005] The first aspect of the present invention provides a method for identifying abnormal conditions of coarse grid equipment based on an AI model combination, including the following steps:

[0006] Obtain the initial coarse grid image of the coarse grid equipment, and perform preprocessing to obtain an optimized coarse grid image;

[0007] Obtain the initial object detection model, and perform optimization processing to obtain an optimized object detection model;

[0008] Input the optimized coarse grid image into the optimized object detection model for processing to obtain the target position data corresponding to the detected target;

[0009] Segment the optimized coarse grid image according to the target position data to obtain a target image, and perform extraction processing to obtain the image feature data of the target image;

[0010] Process according to the image feature data to obtain the initial operation evaluation index of the coarse grid equipment, and compare it with a preset initial operation evaluation threshold of the equipment to obtain the initial evaluation state of the coarse grid equipment;

[0011] If the preliminary evaluation status is suspected to be abnormal, input the image feature data into a preset abnormal detection model for evaluation processing to obtain the operation comprehensive evaluation index of the coarse grille device;

[0012] Compare the operation comprehensive evaluation index with a preset device calibration operation evaluation threshold, and determine the final evaluation status of the coarse grille device according to the comparison result.

[0013] Optionally, in the method for identifying abnormal conditions of a coarse grille device based on an AI model combination in this application, the obtaining of the initial target detection model and performing optimization processing to obtain an optimized target detection model includes:

[0014] Obtain the initial target detection model;

[0015] Optimize the FPN module in the initial target detection model through an adaptive weighted fusion method to obtain a weighted optimized target detection model;

[0016] Optimize the network structure in the weighted optimized target detection model by using a channel attention module and a spatial attention module to obtain an optimized target detection model.

[0017] Optionally, in the method for identifying abnormal conditions of a coarse grille device based on an AI model combination in this application, the inputting of the optimized coarse grille image into the optimized target detection model for processing to obtain the target position data corresponding to the detection target includes:

[0018] Input the optimized coarse grille image into the optimized target detection model for processing to obtain the target position data corresponding to the detection target;

[0019] The target position data includes the coarse grille device position data, the rake bucket position data, and the garbage position data.

[0020] Optionally, in the method for identifying abnormal conditions of a coarse grille device based on an AI model combination in this application, the segmenting of the optimized coarse grille image according to the target position data to obtain a target image and performing extraction processing to obtain the image feature data of the target image includes:

[0021] Segment the optimized coarse grille image according to the coarse grille device position data, the rake bucket position data, and the garbage position data to obtain a target image;

[0022] The target image includes a coarse grille device image, a rake bucket image, and a garbage image;

[0023] Extract image feature data from the coarse grille equipment image to obtain coarse grille equipment feature data, including coarse grille shape feature data, coarse grille color feature data, and coarse grille texture feature data;

[0024] Extract image feature data from the scraper bucket image to obtain scraper bucket image feature data, including scraper bucket shape feature data, scraper bucket color feature data, scraper bucket texture feature data, and scraper bucket inclination angle.

[0025] Optionally, in the method for identifying abnormal conditions of a coarse grille equipment based on an AI model in this application, it further includes:

[0026] Perform pixel statistics on the garbage image to obtain the number of pixels in the garbage area;

[0027] Perform pixel statistics on the optimized coarse grille image to obtain the number of pixels in the monitoring area;

[0028] Compare the number of pixels in the garbage area with the number of pixels in the monitoring area to obtain the garbage occupancy ratio.

[0029] Optionally, in the method for identifying abnormal conditions of a coarse grille equipment based on an AI model in this application, the process of processing the image feature data to obtain the initial evaluation index of the operation of the coarse grille equipment and comparing it with a preset initial evaluation threshold of equipment operation to obtain the initial evaluation status of the coarse grille equipment includes:

[0030] Compare the inclination angle of the scraper bucket with the preset designed inclination angle of the scraper bucket to obtain the deviation rate of the scraper bucket inclination angle;

[0031] Compare the garbage occupancy ratio with the preset allowable value of the garbage occupancy ratio to obtain the garbage overcapacity rate;

[0032] Perform weighted summation processing based on the deviation rate of the scraper bucket inclination angle and the garbage overcapacity rate to obtain the initial evaluation index of the operation of the coarse grille equipment;

[0033] Compare the initial evaluation index of the operation with the preset initial evaluation threshold of equipment operation to obtain the initial evaluation status of the coarse grille equipment;

[0034] If the initial evaluation index of the operation is less than or equal to the preset initial evaluation threshold of equipment operation, it is determined that the initial evaluation status is normal;

[0035] If the initial evaluation index of the operation is greater than the preset initial evaluation threshold of equipment operation, it is determined that the initial evaluation status is suspected of being abnormal.

[0036] Optionally, in the method for identifying abnormal conditions of a coarse grille equipment based on an AI model in this application, if the initial evaluation status is suspected of being abnormal, then input the image feature data into a preset abnormal detection model for evaluation processing to obtain the comprehensive evaluation index of the operation of the coarse grille equipment, including:

[0037] If the preliminary evaluation status is suspected to be abnormal, input the shape feature data, color feature data, and texture feature data of the coarse grille into a preset coarse grille anomaly detection model for evaluation processing to obtain a quality evaluation index of the coarse grille;

[0038] Input the shape feature data, color feature data, and texture feature data of the rake bucket into a preset rake bucket anomaly detection model for evaluation processing to obtain a quality evaluation index of the rake bucket;

[0039] Process according to the quality evaluation index of the coarse grille and the quality evaluation index of the rake bucket in combination with the rake bucket inclination deviation rate and the garbage overcapacity rate to obtain an operation comprehensive evaluation index of the coarse grille equipment.

[0040] Optionally, in the method for identifying anomalies in a coarse grille device based on an AI model combination in this application, the comparing the operation comprehensive evaluation index with a preset device calibration operation evaluation threshold and determining the final evaluation status of the coarse grille device according to the comparison result includes:

[0041] Compare the operation comprehensive evaluation index with a preset operation parameter index to obtain a relative value of the operation index;

[0042] Compare the relative value of the operation index with a preset device calibration operation evaluation threshold to obtain the final evaluation status of the coarse grille device;

[0043] If the relative value of the operation index is less than or equal to the preset device calibration operation evaluation threshold, determine that the final evaluation status is an abnormal status and output a warning response;

[0044] If the relative value of the operation index is greater than the preset device calibration operation evaluation threshold, determine that the final evaluation status is a normal status.

[0045] The second aspect of the present invention provides a system for identifying anomalies in a coarse grille device based on an AI model combination. The system includes: a memory and a processor. The memory includes a program for the method for identifying anomalies in a coarse grille device based on an AI model combination. When the program for the method for identifying anomalies in a coarse grille device based on an AI model combination is executed by the processor, the following steps are implemented:

[0046] Obtain an initial coarse grille image of the coarse grille device and perform preprocessing to obtain an optimized coarse grille image;

[0047] Obtain an initial object detection model and perform optimization processing to obtain an optimized object detection model;

[0048] Input the optimized coarse grille image into the optimized object detection model for processing to obtain target position data corresponding to the detection target;

[0049] Segment the optimized coarse grid image according to the target position data to obtain a target image, and perform extraction processing to obtain the image feature data of the target image;

[0050] Process according to the image feature data to obtain the initial evaluation index of the operation of the coarse grid device, and compare it with the preset initial evaluation threshold of device operation to obtain the initial evaluation status of the coarse grid device;

[0051] If the initial evaluation status is suspected of being abnormal, input the image feature data into a preset abnormal detection model for evaluation processing to obtain the comprehensive evaluation index of the operation of the coarse grid device;

[0052] Compare the comprehensive evaluation index of the operation with the preset calibration operation evaluation threshold of the device, and determine the final evaluation status of the coarse grid device according to the comparison result.

[0053] The third aspect of the present invention provides a computer-readable storage medium, in which a program for identifying abnormal methods of a coarse grid device based on an AI model is stored. When the program for identifying abnormal methods of a coarse grid device based on an AI model is executed by a processor, the steps of the method for identifying abnormal methods of a coarse grid device based on an AI model as described in any one of the above are implemented.

[0054] By using the optimized coarse grid image and the optimized target detection model of the coarse grid device, calculate the initial evaluation index of the operation of the coarse grid device. Through threshold comparison, obtain the initial evaluation status of the coarse grid device. If it is suspected of being abnormal, input the image feature data into a preset abnormal detection model for evaluation processing to obtain the comprehensive evaluation index of the operation of the coarse grid device. Finally, through threshold comparison, determine the final evaluation status of the coarse grid device. Through the combined application of large and small models, accurate judgment of the abnormality of the coarse grid device is realized, improving the judgment efficiency and reducing the false alarm rate at the same time.

[0055] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0056] Improve the monitoring efficiency: Compared with manual inspection, this method can realize real-time monitoring of the coarse grid device, greatly improving the monitoring efficiency.

[0057] Reduce the false alarm rate: Through the combination of large and small models, the accuracy of abnormal judgment is improved, and the false alarm rate is reduced.

[0058] Reduce safety risks: Reduce the number of times of personnel entering the pool for inspection, and reduce the safety risks of operators. Description of the Drawings

[0059] Figure 1 It is a flowchart of the method for identifying abnormal methods of a coarse grid device based on an AI model provided by an embodiment of the present invention;

[0060] Figure 2 Flow chart for obtaining an optimized object detection model for the method of identifying abnormal coarse screen equipment based on an AI model provided by an embodiment of the present invention;

[0061] Figure 3 Flow chart for obtaining the preliminary evaluation status of a coarse screen equipment for the method of identifying abnormal coarse screen equipment based on an AI model provided by an embodiment of the present invention;

[0062] Figure 4 Flow chart for obtaining the operation comprehensive measurement index of a coarse screen equipment for the method of identifying abnormal coarse screen equipment based on an AI model provided by an embodiment of the present invention. Detailed implementation manners

[0063] In order to be able to more clearly understand the above objects, features, and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0064] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0065] Embodiment 1

[0066] As Figure 1 shown, this embodiment discloses a method for identifying abnormal coarse screen equipment in combination with an AI model, which is characterized by including the following steps:

[0067] S11. Obtain the initial coarse screen image of the coarse screen equipment, and perform preprocessing to obtain an optimized coarse screen image;

[0068] S12. Obtain the initial object detection model, and perform optimization processing to obtain an optimized object detection model;

[0069] S13. Input the optimized coarse screen image into the optimized object detection model for processing to obtain the target position data corresponding to the detection target;

[0070] S14. Segment the optimized coarse screen image according to the target position data to obtain a target image, and perform extraction processing to obtain the image feature data of the target image;

[0071] S15. Process according to the image feature data to obtain the initial evaluation index of the operation of the coarse screen equipment, and compare the threshold with a preset initial evaluation threshold of equipment operation to obtain the initial evaluation status of the coarse screen equipment;

[0072] S16. If the preliminary evaluation status is suspected to be abnormal, input the image feature data into a preset abnormal detection model for evaluation processing to obtain the operation comprehensive evaluation index of the coarse grid device;

[0073] S17. Compare the operation comprehensive evaluation index with a preset device calibration operation evaluation threshold, and determine the final evaluation status of the coarse grid device according to the comparison result.

[0074] It should be noted that in order to accurately judge the abnormality of the coarse grid device, first, the initial coarse grid images are taken at regular intervals by the shooting devices arranged in the area of the coarse grid device, and preprocessing such as denoising, sharpening, and grayscale conversion is performed. At the same time, in order to improve the accuracy of target detection, the initial target detection model is optimized. The preprocessed optimized coarse grid images are input into the optimized target detection model for processing to obtain the target position data corresponding to the detection target. Then, based on the target position data, the optimized coarse grid images are segmented to identify the regional images of the coarse grid device, the rake bucket, and the garbage, and the image feature data of the target images are extracted for subsequent abnormality judgment. In order to improve the judgment efficiency and reduce unnecessary computing resources, first, a preliminary judgment is made on the possible abnormalities according to the rake bucket inclination angle and garbage ratio in the image feature data to obtain the preliminary evaluation status of the coarse grid device. If it is suspected to be abnormal, then a final judgment is made through multiple large models to further improve the judgment accuracy and reduce the false judgment rate.

[0075] Embodiment 2

[0076] As Figure 2 shown, this embodiment discloses a flowchart for obtaining an optimized target detection model based on an AI model combined with a method for identifying abnormalities in a coarse grid device. According to an embodiment of the present invention, the obtaining of the initial target detection model and performing optimization processing to obtain the optimized target detection model includes:

[0077] S21. Obtain an initial target detection model;

[0078] S22. Optimize the FPN module in the initial target detection model through an adaptive weighted fusion method to obtain a weighted optimized target detection model;

[0079] S23. Optimize the network structure in the weighted optimized target detection model by using a channel attention module and a spatial attention module to obtain an optimized target detection model.

[0080] It should be noted that in the target detection stage, YOLOv5 is adopted as the basic model in this embodiment. This model features strong real-time performance and high accuracy, and is suitable for real-time monitoring scenarios. Since the coarse grille device has multi-scale targets of different sizes, in terms of feature fusion for the basic model, an adaptive weighted fusion method is used to optimize the simple addition operation of the traditional FPN module. The FPN module is the Feature Pyramid Network module. The optimized FPN assigns a learnable weight to the feature map of each scale, and automatically adjusts these weights by training the network, enabling the network to more reasonably fuse feature information according to the contribution degree of feature maps of different scales to target detection. Then, a channel attention module is introduced to enable the model to automatically learn the importance of each channel and assign different weights to each channel according to the importance. In the detection of the coarse grille device, different channels may contain different types of information. For example, some channels may be sensitive to the edge features of the coarse grille, while others may be sensitive to the color features of the rake bucket. The channel attention module can enable the model to automatically assign weights according to the importance of these channels, making the model pay more attention to the channels containing key information, suppressing the interference of irrelevant channels, and improving the model's ability to extract key features of the coarse grille device, thereby enhancing the detection accuracy. At the same time, a spatial attention module is introduced to enable the model to focus on the key spatial regions in the image. In the image of the coarse grille device, different spatial regions may have different importance. For example, the working area of the coarse grille, the movement trajectory area of the rake bucket, etc. are the areas that need to be focused on. The spatial attention module can enable the model to automatically identify these key spatial regions, enhance the features of these regions, and suppress the features of non-key regions, improving the model's attention to the key regions of the coarse grille device, and thus more accurately detecting the target and further improving the detection accuracy.

[0081] According to the example of the present invention, processing the optimized coarse grille image in the optimized target detection model to obtain target position data corresponding to the detection target includes:

[0082] Processing the optimized coarse grille image in the optimized target detection model to obtain target position data corresponding to the detection target;

[0083] The target position data includes the position data of the coarse grille device, the position data of the rake bucket, and the position data of the garbage.

[0084] It should be noted that the optimized coarse grille image obtained after preprocessing is input into the optimized and improved optimized target detection model for processing to obtain target position data including the position data of the coarse grille device, the position data of the rake bucket, and the position data of the garbage, which serves as the basis for subsequent image segmentation processing.

[0085] According to an embodiment of the present invention, segmenting the optimized coarse grid image according to the target position data to obtain a target image, and performing extraction processing to obtain image feature data of the target image, including:

[0086] Segmenting the optimized coarse grid image according to the coarse grid device position data, the rake bucket position data, and the garbage position data to obtain a target image;

[0087] The target image includes a coarse grid device image, a rake bucket image, and a garbage image;

[0088] Performing image feature data extraction on the coarse grid device image to obtain coarse grid device feature data, including coarse grid shape feature data, coarse grid color feature data, and coarse grid texture feature data;

[0089] Performing image feature data extraction on the rake bucket image to obtain rake bucket image feature data, including rake bucket shape feature data, rake bucket color feature data, rake bucket texture feature data, and rake bucket inclination angle.

[0090] It should be noted that, for facilitating the extraction of image feature data of different regions, the optimized coarse grid image is segmented into a target image including a coarse grid device image, a rake bucket image, and a garbage image according to the determined coarse grid device position data, rake bucket position data, and garbage position data, and then image feature data extraction is performed on the coarse grid device image and the rake bucket image. Among them, the rake bucket inclination angle is the angle between the rake bucket and the horizontal line, providing data support for anomaly determination.

[0091] According to an example of the present invention, it further includes:

[0092] Performing pixel statistics on the garbage image to obtain the number of pixels in the garbage area;

[0093] Performing pixel statistics on the optimized coarse grid image to obtain the number of pixels in the monitoring area;

[0094] Comparing the number of pixels in the garbage area with the number of pixels in the monitoring area to obtain the garbage proportion.

[0095] It should be noted that, in order to determine the garbage situation on the coarse grid device, the garbage proportion is determined through pixel statistics. For example, if the number of pixels in the garbage area obtained by statistics is 30, and the number of pixels in the monitoring area obtained by statistics is 100, then 30 / 100 = 0.3 is the garbage proportion.

[0096] Embodiment 3

[0097] Such as Figure 3As shown in the figure, this embodiment discloses a flowchart for obtaining the preliminary evaluation status of a coarse grille device based on an AI model combined with a method for identifying abnormal conditions of the coarse grille device. According to an example of the present invention, processing the image feature data to obtain a preliminary operation evaluation index of the coarse grille device, and comparing the index with a preset preliminary operation evaluation threshold of the device to obtain the preliminary evaluation status of the coarse grille device, including:

[0098] S31. Compare the rake bucket inclination angle with a preset designed rake bucket inclination angle to obtain a rake bucket inclination deviation rate;

[0099] S32. Compare the garbage occupancy ratio with a preset allowable garbage occupancy ratio to obtain a garbage overcapacity rate;

[0100] S33. Perform weighted summation processing based on the rake bucket inclination deviation rate and the garbage overcapacity rate to obtain a preliminary operation evaluation index of the coarse grille device;

[0101] S34. Compare the preliminary operation evaluation index with a preset preliminary operation evaluation threshold of the device to obtain the preliminary evaluation status of the coarse grille device;

[0102] S35. If the preliminary operation evaluation index is less than or equal to the preset preliminary operation evaluation threshold of the device, determine that the preliminary evaluation status is normal;

[0103] S36. If the preliminary operation evaluation index is greater than the preset preliminary operation evaluation threshold of the device, determine that the preliminary evaluation status is suspected of being abnormal.

[0104] It should be noted that the rake bucket inclination deviation rate refers to the ratio of the absolute value of the difference between the rake bucket inclination angle and the preset designed rake bucket inclination angle to the preset designed rake bucket inclination angle. The garbage overcapacity rate refers to the ratio of the difference between the garbage occupancy ratio and the preset allowable garbage occupancy ratio to the preset allowable garbage occupancy ratio. If the garbage overcapacity rate is negative, it means that the garbage occupancy ratio is relatively small, and this garbage overcapacity rate is excluded. Perform weighted summation processing based on the rake bucket inclination deviation rate and the garbage overcapacity rate to obtain a preliminary operation evaluation index of the coarse grille device, where the corresponding weight values are obtained by querying a third-party preset sewage treatment operation detection platform. Finally, compare with the preset preliminary operation evaluation threshold of the device to obtain the preliminary evaluation status of the coarse grille device.

[0105] Embodiment 4

[0106] As Figure 4 shown in the figure, this embodiment discloses a flowchart for obtaining the comprehensive operation evaluation index of a coarse grille device based on an AI model combined with a method for identifying abnormal conditions of the coarse grille device. According to an example of the present invention, if the preliminary evaluation status is suspected of being abnormal, input the image feature data into a preset abnormal detection model for evaluation processing to obtain the comprehensive operation evaluation index of the coarse grille device, including:

[0107] S41. If the preliminary evaluation status is suspected to be abnormal, input the rough grid shape feature data, rough grid color feature data, and rough grid texture feature data into a preset rough grid anomaly detection model for evaluation processing to obtain a rough grid quality evaluation index;

[0108] S42. Input the scraper bucket shape feature data, scraper bucket color feature data, and scraper bucket texture feature data into a preset scraper bucket anomaly detection model for evaluation processing to obtain a scraper bucket quality evaluation index;

[0109] S43. Process according to the rough grid quality evaluation index and the scraper bucket quality evaluation index in combination with the scraper bucket inclination deviation rate and the garbage overcapacity rate to obtain an operation comprehensive evaluation index of the rough grid equipment.

[0110] It should be noted that in order to improve the accuracy of anomaly recognition and reduce the false alarm rate, after the preliminary evaluation status is suspected to be abnormal, the obtained rough grid equipment feature data and scraper bucket image feature data are respectively input into a preset large model for further determination, such as Swin Transformer and CLIP, to obtain a rough grid quality evaluation index and a scraper bucket quality evaluation index. Among them, the preset rough grid anomaly detection model is trained by obtaining the rough grid shape feature data, rough grid color feature data, and rough grid texture feature data of a large number of historical samples and the corresponding rough grid quality evaluation index. The preset scraper bucket anomaly detection model is trained by obtaining the scraper bucket shape feature data, scraper bucket color feature data, and scraper bucket texture feature data of a large number of historical samples and the corresponding scraper bucket quality evaluation index; then process in combination with the scraper bucket inclination deviation rate and the garbage overcapacity rate to obtain an operation comprehensive evaluation index of the rough grid equipment;

[0111] The calculation formula of the operation comprehensive evaluation index is:

[0112]

[0113] Among them, c r is the operation comprehensive evaluation index, e c , e s , t a , g a are the rough grid quality evaluation index, the scraper bucket quality evaluation index, the scraper bucket inclination deviation rate, and the garbage overcapacity rate respectively, and ε 1 , ε 2 , χ are preset feature coefficients (the feature coefficients are obtained by querying through a preset sewage treatment operation detection platform).

[0114] According to an example of the present invention, comparing the operation comprehensive evaluation index with a preset equipment calibration operation evaluation threshold, and determining the final evaluation status of the rough grid equipment according to the comparison result includes:

[0115] Compare the operation comprehensive evaluation index with a preset operation parameter index to obtain a relative value of the operation index;

[0116] Compare the relative value of the operation index with a preset device calibration operation evaluation threshold to obtain the final evaluation status of the coarse screen device;

[0117] If the relative value of the operation index is less than or equal to the preset device calibration operation evaluation threshold, determine that the final evaluation status is an abnormal status and output a warning response;

[0118] If the relative value of the operation index is greater than the preset device calibration operation evaluation threshold, determine that the final evaluation status is a normal status.

[0119] It should be noted that by comparing the obtained comprehensive operation evaluation index with the preset operation parameter index, the relative value of the operation index is obtained. The preset operation parameter index refers to the comprehensive operation evaluation benchmark index set according to actual needs. For example, if the obtained comprehensive operation evaluation index is 7 and the preset operation parameter index is 10, then 7 / 10 = 0.7 is the relative value of the operation index. Then, compare the relative value of the operation index with the preset device calibration operation evaluation threshold to obtain the final evaluation status of the coarse screen device. In this embodiment, the preset device calibration operation evaluation threshold is set to (0, 0.8] and (0.8, 1], corresponding to the abnormal status and the normal status respectively. For example, if the obtained relative value of the operation index is 0.7, which is less than the preset device calibration operation evaluation threshold, it is determined that the final evaluation status is an abnormal status and a warning response is output. If the obtained relative value of the operation index is 0.85, which is greater than the preset device calibration operation evaluation threshold, it is determined that the final evaluation status is a normal status, indicating that the initial evaluation of suspected abnormality is a false alarm.

[0120] It is worth mentioning that according to the example of the present invention, it further includes:

[0121] Obtain the pollution interception effectiveness data and the water passing capacity monitoring data of the coarse screen device;

[0122] The pollution interception effectiveness data includes the interception efficiency and the moisture content of the screen residue;

[0123] The water passing capacity monitoring data includes the measured value of the water passing volume and the head loss;

[0124] Compare the measured value of the water passing volume with the preset designed water passing volume to obtain the water passing rate, and compare the head loss with the preset reference head amount to obtain the head loss rate;

[0125] Process according to the interception efficiency and the moisture content of the screen residue in combination with the water passing rate and the head loss rate to obtain the performance evaluation index of the coarse screen device.

[0126] It should be noted that in addition to evaluating the operation of the coarse grid equipment through monitoring images, attention should also be paid to the operation effectiveness of the coarse grid equipment. The interception efficiency is the ratio of the amount of solid pollutants intercepted by the coarse grid per unit time to the total amount of pollutants entering the grid. The water passing rate is the ratio of the difference between the preset designed water passing volume and the measured water passing volume to the preset designed water passing volume. The head loss rate is the ratio of the difference between the preset head reference quantity and the head loss quantity to the preset head reference quantity;

[0127] The calculation formula for the performance evaluation index of the coarse grid equipment is as follows:

[0128]

[0129] Among them, p e is the performance evaluation index, n e , m c , w p , w s are the interception efficiency, the moisture content of the grid residue, the water passing rate, and the head loss rate respectively, and κ 1 , κ 2 , λ 1 , λ 2 are preset characteristic coefficients (the characteristic coefficients are obtained by querying through a preset sewage treatment operation detection platform).

[0130] It is worth mentioning that according to the embodiments of the present invention, it further includes:

[0131] Obtain the running speed of the scraper bucket, and compare it with the preset design parameter speed to obtain the running speed change rate;

[0132] Obtain the tension data of the chain of the coarse grid equipment, and compare it with the preset design parameter tension to obtain the tension change rate;

[0133] Input the running speed change rate and the tension change rate into a preset operation evaluation model of the coarse grid equipment for processing to obtain the operation stability and effectiveness index of the coarse grid equipment.

[0134] It should be noted that when the scraper bucket runs with problems such as jamming, tooth skipping or deviation, the running speed will be quite different from the designed value. Whether the chain tension is too loose or too tight will affect the operation of the equipment. The running speed change rate refers to the ratio of the absolute value of the difference between the running speed and the preset design parameter speed to the preset design parameter speed. The tension change rate refers to the ratio of the absolute value of the difference between the tension data and the preset design parameter tension to the preset design parameter tension. Those skilled in the art can evaluate the tension by measuring the sag of the chain or checking the meshing condition of the chain and the sprocket. Input the obtained running speed change rate and tension change rate into the preset operation evaluation model of the coarse screen equipment for processing to obtain the operation stability and efficiency index of the coarse screen equipment. Among them, the preset operation evaluation model of the coarse screen equipment is obtained by training with the running speed change rate, tension change rate and corresponding operation stability and efficiency index of a large number of historical samples.

[0135] It is worth mentioning that according to the embodiments of the present invention, it further includes:

[0136] Process according to the performance evaluation index and the operation stability and efficiency index to obtain the abnormal identification correction coefficient of the coarse screen equipment;

[0137] Correct the operation comprehensive measurement index according to the abnormal identification correction coefficient to obtain the operation comprehensive measurement correction index.

[0138] It should be noted that in order to further improve the accuracy of abnormal identification and determination by combining the operation effectiveness and the operation condition of the power equipment, process according to the performance evaluation index and the operation stability and efficiency index to obtain the abnormal identification correction coefficient of the coarse screen equipment, and correct the operation comprehensive measurement index of the coarse screen equipment;

[0139] The calculation formula of the abnormal identification correction coefficient is:

[0140]

[0141] Wherein, c f is the abnormal identification correction coefficient, p e , s t are the performance evaluation index and the operation stability and efficiency index respectively, is the preset characteristic coefficient (the characteristic coefficient is obtained by querying through the preset sewage treatment operation detection platform);

[0142] The calculation formula of the operation comprehensive measurement correction index is:

[0143] c mx =(1 + μ×c f )×c m ;

[0144] Wherein, c mx is the operation comprehensive measurement correction index, c f , cm They are the anomaly recognition correction coefficient and the operation comprehensive measurement index respectively, and μ is a preset characteristic coefficient (the characteristic coefficient is obtained by querying through a preset sewage treatment operation detection platform).

[0145] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed with each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0146] The units described above as separate components may or may not be physically separated. The components shown as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0147] In addition, in each embodiment of the present invention, the various functional units can all be integrated in a processing unit, or each unit can be separately used as a unit, or two or more units can be integrated in one unit; the above-mentioned integrated units can be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0148] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: various media that can store program codes such as mobile storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs.

[0149] Alternatively, if the above integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as removable storage devices, ROM, RAM, magnetic disks, or optical discs.

Claims

1. A method for identifying abnormalities of coarse grid equipment based on an AI model, characterized in that: The following steps are involved: Acquire an initial coarse grid image of a coarse grid device, and perform preprocessing to obtain an optimized coarse grid image; Obtain an initial target detection model, and perform optimization processing to obtain an optimized target detection model; Inputting the optimized coarse grid image into the optimized target detection model for processing to obtain target position data corresponding to the detection target; Segmenting the optimized coarse grid image according to the target position data to obtain a target image, and extracting the image to obtain image feature data of the target image; Processing is performed according to the image feature data to obtain a preliminary evaluation index of the operation of the coarse grid equipment, and a threshold value comparison is performed with a preset preliminary evaluation threshold value of the equipment operation to obtain a preliminary evaluation status of the coarse grid equipment; If the initial evaluation status is suspected abnormal, the image feature data is input into a preset abnormality detection model for evaluation and processing to obtain a comprehensive operation index of the coarse grid equipment; The operation comprehensive measurement index is compared with the preset equipment calibration operation evaluation threshold, and the final evaluation status of the coarse grid equipment is determined according to the comparison result.

2. According to claim 1, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: The obtaining of the initial target detection model and performing optimization processing to obtain the optimized target detection model includes: Get the initial target detection model; The FPN module in the initial target detection model is optimized by an adaptive weighted fusion method to obtain a weighted optimized target detection model. A channel attention module and a spatial attention module are used to optimize the network structure in the weighted optimization target detection model to obtain an optimized target detection model.

3. According to claim 2, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: The step of inputting the optimized coarse grid image into the optimized target detection model for processing to obtain target position data corresponding to the detection target includes: Inputting the optimized coarse grid image into the optimized target detection model for processing to obtain target position data corresponding to the detection target; The target location data includes coarse screen equipment location data, rake bucket location data and garbage location data.

4. According to claim 3, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: The step of segmenting the optimized coarse grid image according to the target position data to obtain the target image, and extracting the image feature data of the target image comprises: Segmenting the optimized coarse grid image according to the coarse grid equipment position data, the rake bucket position data and the garbage position data to obtain a target image; The target images include a coarse grid equipment image, a rake bucket image, and a garbage image; Extracting image feature data from the coarse grid device image to obtain coarse grid device feature data, including coarse grid shape feature data, coarse grid color feature data, and coarse grid texture feature data; Image feature data of the bucket image is extracted to obtain bucket image feature data, including bucket shape feature data, bucket color feature data, bucket texture feature data and bucket inclination angle.

5. According to claim 4, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: Also includes: Perform pixel counting according to the garbage image to obtain the number of pixels in the garbage area; Perform pixel statistics according to the optimized coarse grid image to obtain the number of pixels in the monitoring area; The number of pixels in the garbage area is compared with the number of pixels in the monitoring area to obtain the garbage ratio.

6. According to claim 5, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: The processing according to the image feature data to obtain the preliminary evaluation index of the operation of the coarse grid equipment, and comparing the threshold with the preset preliminary evaluation threshold of the equipment operation to obtain the preliminary evaluation status of the coarse grid equipment, includes: Comparing the inclination angle of the rake bucket with the preset design inclination angle of the rake bucket to obtain the deviation rate of the inclination angle of the rake bucket; Compare the garbage percentage with a preset garbage percentage allowable value to obtain a garbage excess rate; A weighted summation process is performed based on the bucket inclination deviation rate and the garbage over-capacity rate to obtain a preliminary evaluation index of the operation of the coarse screen equipment; Compare the operation preliminary evaluation index with the preset equipment operation preliminary evaluation threshold to obtain the preliminary evaluation status of the coarse grid equipment; If the operation preliminary evaluation index is less than or equal to the preset equipment operation preliminary evaluation threshold, the preliminary evaluation state is determined to be normal; If the operation preliminary evaluation index is greater than the preset equipment operation preliminary evaluation threshold, the preliminary evaluation status is determined to be suspected abnormal.

7. According to claim 6, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: If the initial evaluation status is suspected abnormal, the image feature data is input into a preset abnormality detection model for evaluation and processing to obtain a comprehensive operation index of the coarse grid equipment, including: If the initial evaluation status is suspected abnormality, the coarse grid shape feature data, the coarse grid color feature data and the coarse grid texture feature data are input into a preset coarse grid abnormality detection model for evaluation processing to obtain a coarse grid quality evaluation index; The bucket shape feature data, bucket color feature data and bucket texture feature data are input into a preset bucket anomaly detection model for evaluation and processing to obtain a bucket quality evaluation index; The operation comprehensive index of the coarse screen equipment is obtained by processing the coarse screen quality evaluation index and the scraper bucket quality evaluation index in combination with the scraper bucket inclination deviation rate and the garbage over-capacity rate.

8. According to claim 7, the method for identifying abnormalities of coarse grid equipment based on AI model is characterized in that: The operation comprehensive measurement index is compared with a preset equipment calibration operation evaluation threshold, and the final evaluation status of the coarse grid equipment is determined according to the comparison result, including: Comparing the operation comprehensive measurement index with a preset operation parameter index to obtain a relative value of the operation index; Compare the relative value of the operation index with the preset equipment calibration operation evaluation threshold to obtain the final evaluation status of the coarse grid equipment; If the relative value of the operation index is less than or equal to the preset equipment calibration operation evaluation threshold, the final evaluation state is determined to be an abnormal state, and an early warning response is output; If the relative value of the operation index is greater than the preset equipment calibration operation evaluation threshold, the final evaluation status is determined to be a normal status.

9. Based on AI model combined with the system for identifying abnormalities of coarse grid equipment, it is characterized by: The invention comprises a memory and a processor, wherein the memory comprises a method program for identifying abnormalities of a coarse grid device based on an AI model, and the method program for identifying abnormalities of a coarse grid device based on an AI model is executed by the processor to implement the following steps: Acquire an initial coarse grid image of a coarse grid device, and perform preprocessing to obtain an optimized coarse grid image; Obtain an initial target detection model, and perform optimization processing to obtain an optimized target detection model; Inputting the optimized coarse grid image into the optimized target detection model for processing to obtain target position data corresponding to the detection target; Segmenting the optimized coarse grid image according to the target position data to obtain a target image, and extracting the image to obtain image feature data of the target image; Processing is performed according to the image feature data to obtain a preliminary evaluation index of the operation of the coarse grid equipment, and a threshold value comparison is performed with a preset preliminary evaluation threshold value of the equipment operation to obtain a preliminary evaluation status of the coarse grid equipment; If the initial evaluation status is suspected abnormal, the image feature data is input into a preset abnormality detection model for evaluation and processing to obtain a comprehensive operation index of the coarse grid equipment; The operation comprehensive measurement index is compared with the preset equipment calibration operation evaluation threshold, and the final evaluation status of the coarse grid equipment is determined according to the comparison result.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a method program for identifying abnormalities of coarse grid equipment based on an AI model. When the method program for identifying abnormalities of coarse grid equipment based on an AI model is executed by a processor, the steps of the method for identifying abnormalities of coarse grid equipment based on an AI model as described in any one of claims 1 to 8 are implemented.

Images