Intelligent detection and identification method for remotely monitoring overhead line system insulator defects

Abstract

The invention relates to the technical field of insulator defect detection, in particular to an intelligent detection and identification method for remotely monitoring overhead line system insulator defects, which comprises the following steps: acquiring a first-stage cascade model for intercepting and correcting insulator image information and a second-stage cascade model for selecting and calibrating an insulator defect area, and periodically determining whether the operation of the cascade model is qualified or not according to the false detection rate of the two-stage cascade model, adjusting the operation parameters of the cascade model when the operation of the cascade model is judged to be abnormal, monitoring the operation of the cascade model in the use process of the cascade model, and timely calibrating the model, so that the detection precision of the defective insulator is improved, and the detection efficiency is improved. And the detection efficiency of the defective insulator is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of insulator defect detection, and particularly to an intelligent detection and identification method for remotely monitoring the defects of catenary insulators. Background Art

[0002] The catenary cantilever insulator plays an important role in the electrical insulation and power support of electrified railways. Its insulation performance is crucial for the safe and stable operation of the railway power supply system. The service environment of catenary insulators is complex, and their surfaces are inevitably eroded by the outside world, resulting in defects of varying degrees. Typical ones include bird pecking damage, umbrella skirt brittle fracture, foreign object attachment, cracks, etc. The expansion of defects may cause insulation failure and trigger short-circuit faults, resulting in large-scale train suspension accidents. Therefore, it is necessary to monitor the status of cantilever insulators for a long time to ensure the safe operation of the catenary system.

[0003] Chinese Patent Publication No.: CN118839245A discloses an insulator defect detection method based on improved YOLOv8, including designing a new C2f module based on ScConv and introducing GhostConv. In addition, an Adown module is also introduced. It can be seen that the above technical solutions have the following problems: during the use of the model, the operation of the model cannot be monitored, and the model cannot be calibrated in time, which affects the detection accuracy of defective insulators. Summary of the Invention

[0004] Therefore, the present invention provides an intelligent detection and identification method for remotely monitoring the defects of catenary insulators to overcome the problems in the prior art that during the use of the model, the operation of the model cannot be monitored, the model cannot be calibrated in time, and the detection accuracy of defective insulators is affected.

[0005] To achieve the above object, the present invention provides an intelligent detection and identification method for remotely monitoring the defects of catenary insulators, including: S1, using the insulator region image information obtained by the insulator detector as the experimental data set to train a one-stage cascade model for outputting the intercepted and corrected insulator image information; S2, using the insulator image information as the training set to train a two-stage cascade model for framing and calibrating the insulator defect region; S3, sequentially inputting the insulator image information obtained by each insulator detector into the one-stage cascade model and the two-stage cascade model to obtain the framed and calibrated insulator region image information; S4, taking the preset monitoring duration as a cycle, periodically determining whether the operation of the cascade model is qualified according to the false detection rate of the two-stage cascade model, including, When determining the abnormal operation of the cascaded model, adjust the operation parameters of the cascaded model based on the recognition difference amount, including using the insulator region image information obtained by the insulator detector as an experimental data set to calibrate the cascaded model, or adjusting the layout density of the insulator detector to the corresponding value; Or, when determining that the operation of the cascaded model is qualified, continuously use the two-stage cascaded model to perform box selection and calibration on each insulator region image information.

[0006] Further, the process of periodically determining whether the operation of the cascaded model is qualified based on the false detection rate of the two-stage cascaded model includes: Compare the obtained false detection rate with the first preset false detection rate and the second preset false detection rate respectively; If the false detection rate is less than or equal to the first preset false detection rate, it is determined that the operation of the cascaded model is qualified, and continuously use the two-stage cascaded model to perform box selection and calibration on each insulator region image information; If the false detection rate is less than or equal to the second preset false detection rate and greater than the first preset false detection rate, mark the false detection rate of the current cycle as an abnormal false detection rate, and determine whether the operation of the cascaded model is qualified in combination with the obtained historical false detection rates; If the false detection rate is greater than the second preset false detection rate, it is determined that the operation of the cascaded model is abnormal, and adjust the operation parameters of the cascaded model based on the recognition difference amount; Determine the ratio of the number of insulator image information obtained by the two-stage cascaded model within the preset monitoring duration to the preset number of insulators as the recognition difference amount. Further, the process of determining whether the operation of the cascaded model is qualified in combination with the obtained historical false detection rates includes: Calculate the ratio of the number of abnormal false detection rates in the historical false detection rates to the total number of historical false detection rates to obtain the abnormal frequency; Calculate the ratio of the abnormal false detection rate of the current cycle to the average value of each historical false detection rate to obtain the abnormal deviation amount; If the abnormal frequency is less than or equal to the preset abnormal frequency and the abnormal deviation amount is greater than the preset deviation amount, it is determined that the operation of the cascaded model is qualified, continuously use the two-stage cascaded model to perform box selection and calibration on each insulator region image information, and adjust the preset monitoring duration to the corresponding value based on the abnormal deviation amount; If the abnormal frequency is greater than the preset abnormal frequency and / or the abnormal deviation amount is less than or equal to the preset deviation amount, it is determined that the operation of the cascaded model is abnormal, and adjust the operation parameters of the cascaded model based on the recognition difference amount.

[0007] Further, the process of adjusting the operation parameters of the cascaded model based on the recognition difference amount includes: If the recognized difference amount is less than or equal to the preset recognized difference amount, obtain the time nodes when the insulator detector acquires the image information of each mis-calibrated insulator area, and determine the concentrated time interval of each time node; Use the image information of the preset calibration quantity of insulator areas acquired by the insulator detector within the concentrated time interval as the experimental data set to calibrate the cascade model; Adjust the preset calibration quantity to the corresponding value based on the recognized difference amount.

[0008] Furthermore, the process of adjusting the operating parameters of the cascade model based on the recognized difference amount also includes: If the recognized difference amount is greater than the preset recognized difference amount, adjust the cascade model by combining the recognition accuracy evaluation value determined from the image information of each insulator area; Calculate the average value of the areas of the framed regions of the image information of each insulator area to obtain the recognition accuracy evaluation value; If the recognition accuracy evaluation value is less than or equal to the preset recognition accuracy evaluation value, adjust the layout density of the insulator detectors to the corresponding value based on the average value of the distances between the insulator detectors corresponding to the image information of each mis-calibrated insulator area; If the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, select the image information of the insulator areas acquired by the preset calibration quantity of insulator detectors as the experimental data set to calibrate the cascade model, and adjust the preset calibration quantity to the corresponding value based on the recognition accuracy evaluation value.

[0009] Furthermore, the specific process of obtaining the false detection rate includes, Input the image information of the preset verification quantity of insulator images acquired by the insulator detector into the cascade model to obtain the image information of the framed and calibrated insulator areas; Determine the quantity of the image information of the insulator areas with mis-framed calibration, and calculate the ratio of the quantity of mis-calibrations to the preset verification quantity to obtain the false detection rate.

[0010] Furthermore, adjust the preset monitoring duration to the corresponding value based on the abnormal deviation amount, where: The reduction amplitude of the preset monitoring duration is proportional to the abnormal deviation amount.

[0011] Furthermore, adjust the preset calibration quantity to the corresponding value based on the recognized difference amount, where, The increase amplitude of the preset calibration quantity is inversely proportional to the recognized difference amount.

[0012] Furthermore, adjust the layout density of the insulator detectors to the corresponding value based on the average value of the distances between the insulator detectors corresponding to the image information of each mis-calibrated insulator area, where, The increase amplitude of the layout density is inversely proportional to the average value of the distances.

[0013] Further, adjust the preset correction quantity to a corresponding value based on the recognition accuracy evaluation value, where the increase amplitude of the preset correction quantity is directly proportional to the recognition accuracy evaluation value.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: obtain a one-stage cascade model for intercepting and correcting insulator image information, and a two-stage cascade model for framing and calibrating the defective area of the insulator. Periodically determine whether the operation of the cascade model is qualified according to the false detection rate of the two-stage cascade model, and adjust the operation parameters of the cascade model when it is determined that the operation of the cascade model is abnormal. During the use of the cascade model, monitor the operation of the cascade model and calibrate the model in a timely manner, improving the detection accuracy of defective insulators and the detection efficiency of defective insulators.

[0015] Further, determine whether the operation of the cascade model is qualified according to the false detection rate of the cascade model. The false detection rate characterizes the accurate situation of the detection of the cascade model. When the false detection rate is less than or equal to the first preset false detection rate, the detection accuracy of the cascade model is relatively high, and it can frame and calibrate the insulator area image information more accurately. In this case, it is determined that the operation of the cascade model is qualified. When the false detection rate is greater than the second preset false detection rate, the false detection rate is relatively high, and the false detection situation of the cascade model is relatively serious. Further adjust the operation parameters of the cascade model according to the recognition difference amount. Timely discover the abnormal operation situation of the model, ensure the stability and reliability of the cascade model, and further improve the detection accuracy of insulator defects.

[0016] Further, when the false detection rate is less than or equal to the second preset false detection rate and greater than the first preset false detection rate, judge the operation situation of the model in combination with the historical false detection rate. The abnormal frequency characterizes the frequency of the abnormal false detection rate in the historical false detection rate. When the abnormal frequency is less than or equal to the preset abnormal frequency, the occurrence of the abnormal false detection rate is accidental. The abnormal deviation amount characterizes the deviation degree of the current cycle's abnormal false detection rate from the average value of the historical false detection rate. When the abnormal deviation amount is greater than the preset deviation amount, the current abnormal false detection rate deviates greatly from the historical average level; in the case of a low abnormal frequency and the cascade model suddenly deviating greatly from the historical average level, the abnormal false detection rate is too high due to extreme environments or emergencies, and the model can still operate normally. In this case, adjust the preset monitoring duration to further observe. Comprehensively evaluate the stability of the model, avoid misjudgment caused by accidental factors, improve the accuracy of the model evaluation, and further improve the detection accuracy of defective insulators.

[0017] Furthermore, the operating parameters of the cascaded model are adjusted based on the recognition difference amount, which reflects the recognition situation of the insulator by the one-stage cascaded model. When the recognition difference amount is less than or equal to the preset recognition difference amount, there is a large difference between the number of insulators recognized by the one-stage cascaded model and the actual number of insulators to be detected in the area. At this time, the model has problems during certain time periods and cannot accurately recognize insulators due to the influence of the external environment. In this case, the concentrated time interval is determined, and the image information within this interval is obtained to calibrate the model, specifically calibrating the model. When the recognition difference amount is greater than the preset recognition difference amount, the recognition difference amount is large. In this case, the model is further adjusted in combination with the recognition accuracy evaluation value. This improves the adaptability and accuracy of the model. The recognition accuracy evaluation value characterizes the general situation of the area selected by the model. When the recognition accuracy evaluation value is less than or equal to the preset recognition accuracy evaluation value, the area selected by the model is generally small, and the model can still accurately recognize the defects and abnormalities in small areas. In this case, due to the insufficient layout density of the insulator detectors, the obtained image information is incomplete. In this case, the layout density of the insulator detectors is adjusted. When the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, the areas selected by the model are all large. In this case, due to the abnormal recognition accuracy of the model, the abnormalities in small areas cannot be accurately recognized, and there are false detection situations. In this case, the insulator area image information is used as the experimental data set to calibrate the cascaded model. The layout quantity of the insulator detectors is optimized, the accuracy of insulator defect detection is improved, and the detection efficiency of defective insulators is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 FIG. is a flowchart of the steps of the intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to an embodiment of the present invention; Figure 2 FIG. is a logical decision diagram for determining whether the operation of the cascaded model is qualified based on the false detection rate of the cascaded model according to an embodiment of the present invention; Figure 3 FIG. is a logical decision diagram for determining whether the operation of the cascaded model is qualified by combining the historical false detection rate according to an embodiment of the present invention; Figure 4 FIG. is a logical decision diagram for adjusting the operating parameters of the cascaded model based on the recognition difference amount according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0021] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of the present invention.

[0022] In addition, it should also be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0023] Please refer to Figure 1 、 Figure 2 、 Figure 3 and Figure 4 as shown, which are respectively the step flow chart of the intelligent detection and recognition method for remotely monitoring the defects of catenary insulators in the embodiments of the present invention, the logical decision diagram for determining whether the cascade model is qualified based on the false detection rate of the cascade model, the logical decision diagram for determining whether the cascade model is qualified by combining the historical false detection rate, and the logical decision diagram for adjusting the operating parameters of the cascade model based on the recognition difference amount; An intelligent detection and recognition method for remotely monitoring the defects of catenary insulators in an embodiment of the present invention includes: S1, using the insulator region image information obtained by the insulator detector as an experimental data set to train a one-stage cascade model for outputting the intercepted and corrected insulator image information; S2, using the insulator image information as a training set to train a two-stage cascade model for framing and calibrating the insulator defect region; S3, sequentially inputting the insulator image information obtained by each insulator detector into the one-stage cascade model and the two-stage cascade model to obtain the framed and calibrated insulator region image information; S4, taking the preset monitoring duration as a period, periodically determining whether the cascade model is qualified based on the false detection rate of the two-stage cascade model, including, When determining the abnormal operation of the cascaded model, adjust the operation parameters of the cascaded model based on the recognition difference amount, including calibrating the cascaded model with the insulator region image information obtained by the insulator detector as the experimental data set, or adjusting the layout density of the insulator detector to the corresponding value; Or, when determining that the operation of the cascaded model is qualified, continuously use the two-stage cascaded model to perform box selection and calibration on each insulator region image information.

[0024] Specifically, input the insulator region image information obtained by the insulator detector into the one-stage cascaded model to obtain the cropped and corrected insulator image information; input the insulator image information into the two-stage cascaded model, and the two-stage cascaded model identifies the defects in the insulator image information, calibrates and boxes the defect regions, maps the coordinates of the defect regions in the insulator image information to the insulator region image information, and obtains the insulator region image information with the defect regions calibrated and boxed, thereby obtaining the detection result of the cascaded model.

[0025] Specifically, in the step S1, the insulator region image information used as the experimental data set is the image information actually obtained by the insulator detector in the railway department. Use the Labelme software to annotate all the insulator region image information obtained by the insulator detector, add labels of defect types and defect positions to the insulators in each image information, and randomly divide the annotated image information into a training set and a validation set according to a ratio of 7:3 to train a one-stage cascaded model for identifying, positioning, cropping, and correcting target insulators in the global image.

[0026] Specifically, the specific structure of the insulator detector is not limited, and it can be a catenary suspension state detection and monitoring device, which is the prior art and will not be elaborated here.

[0027] Specifically, the one-stage cascaded model adopts the YOLOv8 object detection algorithm to identify the insulator and locate its position in the insulator region image information, intercept the insulator in the insulator region image information, and use the Canny algorithm and the Hough transform algorithm to correct the intercepted insulator. This is the prior art and will not be elaborated here.

[0028] Specifically, in the step S2, use the insulator image information output by the one-stage cascaded model as the training set to train the YOLO v8 instance segmentation algorithm to obtain the two-stage cascaded model; the two-stage cascaded model identifies, boxes, and calibrates five types of surface defects of the insulator, and maps the defect coordinates to the insulator region image information using the homography matrix as the final output of the insulator surface defect detection result. The five types of surface defects include bird peck damage, skirt brittle fracture, crack, foreign object attachment, and bird droppings.

[0029] Specifically, a one-stage cascade model for obtaining and correcting insulator image information and a two-stage cascade model for bounding and calibrating the defective areas of insulators are acquired. Periodically, it is determined whether the operation of the cascade model is qualified based on the false detection rate of the two-stage cascade model. When it is determined that the operation of the cascade model is abnormal, the operation parameters of the cascade model are adjusted. During the use of the cascade model, the operation of the cascade model is monitored, and the model is calibrated in a timely manner, improving the detection accuracy and efficiency of defective insulators.

[0030] Specifically, the process of periodically determining whether the operation of the cascade model is qualified based on the false detection rate of the two-stage cascade model includes: Comparing the obtained false detection rate with a first preset false detection rate and a second preset false detection rate respectively; If the false detection rate is less than or equal to the first preset false detection rate, it is determined that the operation of the cascade model is qualified, and the two-stage cascade model is continuously used to bound and calibrate the image information of each insulator area; If the false detection rate is less than or equal to the second preset false detection rate and greater than the first preset false detection rate, the false detection rate of the current cycle is marked as an abnormal false detection rate, and it is determined whether the operation of the cascade model is qualified in combination with the obtained historical false detection rates; If the false detection rate is greater than the second preset false detection rate, it is determined that the operation of the cascade model is abnormal, and the operation parameters of the cascade model are adjusted based on the recognition difference amount; The ratio of the number of insulator image information obtained by the two-stage cascade model within a preset monitoring duration calculated to the preset number of insulators is determined as the recognition difference amount.

[0031] Specifically, the preset number of insulators is the number of insulators actually set within a preset monitoring area.

[0032] Specifically, it can be understood that during the process of determining whether the operation of the cascade model is qualified based on the false detection rate of the two-stage cascade model, the false detection rate represents the false detection situation of the output result of the two-stage cascade model.

[0033] It should be noted that the data in this embodiment are all obtained through comprehensive analysis and evaluation of the historical detection data and corresponding historical detection results in the three months before this detection by the method of the present invention. Those skilled in the art can understand that the determination method of the present invention for a single parameter can be to select the value with the highest proportion according to the data distribution as the preset standard parameter, use weighted summation to take the obtained value as the preset standard parameter, or other selection methods, as long as it satisfies that the system of the present invention can clearly define different specific situations in the single-item determination process through the obtained values.

[0034] Specifically, the first preset false detection rate is selected within the range of [0.04, 0.07], and the second preset false detection rate is selected within the range of [0.1, 0.13].

[0035] Specifically, the specific process of obtaining the false detection rate includes Sequentially inputting the preset verification number of insulator image information obtained by the insulator detector into the one-stage cascade model and the two-stage cascade model to obtain the image information of the insulator region after frame selection and calibration; Determine the number of image information of the insulator region where the frame selection and calibration are incorrect, calculate the ratio of the number of incorrect calibrations to the preset verification number to obtain the false detection rate.

[0036] Specifically, the Labelme software can be used to label the preset verification number of insulator region image information obtained by the insulator detector, and compare the image information of the insulator region after labeling by the Labelme software with the image information of the insulator region after frame selection and calibration output by the cascade model. The ratio of the area of the intersection of the two frame selection regions to the area of the union of the two frame selection regions is used as the comparison coincidence degree; the image information of the insulator region where the comparison coincidence degree is lower than the preset coincidence degree is determined as the image information of the insulator region where the frame selection and calibration are incorrect.

[0037] Specifically, based on the false detection rate of the cascade model, it is determined whether the operation of the cascade model is qualified. The false detection rate characterizes the accuracy of the cascade model's detection. When the false detection rate is less than or equal to the first preset false detection rate, the detection accuracy of the cascade model is relatively high, and it can accurately frame and calibrate the image information of the insulator region. In this case, it is determined that the cascade model operates qualified. When the false detection rate is greater than the second preset false detection rate, the false detection rate is relatively high, and the false detection situation of the cascade model is relatively serious. Further, the operating parameters of the cascade model are adjusted according to the recognition difference amount. Timely detecting the abnormal operation situation of the model ensures the stability and reliability of the cascade model and further improves the accuracy of insulator defect detection.

[0038] Specifically, the process of determining whether the operation of the cascade model is qualified by combining the obtained historical false detection rate includes: Calculate the ratio of the number of abnormal false detection rates in the historical false detection rate to the total number of historical false detection rates to obtain the abnormal frequency; Calculate the ratio of the abnormal false detection rate in the current period to the average value of each historical false detection rate to obtain the abnormal deviation amount; If the abnormal frequency is less than or equal to the preset abnormal frequency and the abnormal deviation amount is greater than the preset deviation amount, it is determined that the operation of the cascade model is qualified, and the two-stage cascade model is continuously used to frame and calibrate the image information of each insulator region, and the preset monitoring duration is adjusted to the corresponding value based on the abnormal deviation amount; If the abnormal frequency is greater than the preset abnormal frequency and / or the abnormal deviation amount is less than or equal to the preset deviation amount, it is determined that the cascade model runs abnormally, and the operating parameters of the cascade model are adjusted based on the recognition difference amount.

[0039] Specifically, the preset abnormal frequency is selected within the interval [0.15, 0.21], and the preset deviation amount P0 is selected within the interval [1.18, 1.23].

[0040] Specifically, there is no limit on the number of selected historical false detection rates for determining the abnormal frequency and abnormal deviation amount. It can be understood that in order to obtain the occurrence law of the abnormal false detection rate to determine whether the accidental abnormal false detection rate is caused by extreme environments or emergencies, the number of false detection rates selected from the historical data should be no less than 30, which will not be elaborated here.

[0041] Specifically, when the false detection rate is less than or equal to the second preset false detection rate and greater than the first preset false detection rate, the operating condition of the model is judged in combination with the historical false detection rate. The abnormal frequency characterizes the frequency of the abnormal false detection rate in the historical false detection rate. When the abnormal frequency is less than or equal to the preset abnormal frequency, the occurrence of the abnormal false detection rate is accidental. The abnormal deviation amount characterizes the deviation degree of the current cycle's abnormal false detection rate from the average value of the historical false detection rates. When the abnormal deviation amount is greater than the preset deviation amount, the current abnormal false detection rate deviates greatly from the historical average level. In the case of a low abnormal frequency and a large sudden deviation of the cascade model from the historical average level, the abnormal false detection rate is high due to extreme environments or emergencies, and the model can still operate normally. In this case, the preset monitoring duration is adjusted for further observation. The stability of the model is comprehensively evaluated, avoiding misjudgment caused by accidental factors, improving the accuracy of the model evaluation, and further improving the detection accuracy of defective insulators.

[0042] Specifically, the process of adjusting the operating parameters of the cascade model based on the recognition difference amount includes: If the recognition difference amount is less than or equal to the preset recognition difference amount, obtain the time nodes when the insulator detector acquires the image information of each calibrated faulty insulator area, and determine the concentrated time interval of each time node; Take the image information of the preset correction number of insulator area acquired by the insulator detector within the concentrated time interval as the experimental data set to calibrate the cascade model; Adjust the preset correction number to the corresponding value based on the recognition difference amount.

[0043] Specifically, there is no limitation on the specific method for determining the concentrated time interval. Each time node can be sorted in chronological order, and a sliding window can be used. The window is determined to be 1 hour. The window is slid on the time axis, and the number of time nodes within the window is counted. The time period corresponding to the window with the number exceeding the preset critical number is determined as the concentrated time interval. It can also be that each time node is sorted in chronological order, the number of time nodes within 1 hour near each time node is counted, and the time period with a larger number of time nodes is determined as the concentrated time interval.

[0044] Specifically, the preset recognition difference amount B0 is selected within the interval [0.94, 0.97].

[0045] Specifically, the process of adjusting the operating parameters of the cascade model based on the recognition difference amount further includes: If the recognition difference amount is greater than the preset recognition difference amount, the cascade model is adjusted in combination with the recognition accuracy evaluation value determined by the image information of each insulator region; The average value of the areas of the framed regions of the image information of each insulator region is calculated to obtain the recognition accuracy evaluation value; If the recognition accuracy evaluation value is less than or equal to the preset recognition accuracy evaluation value, the layout density of the insulator detectors is adjusted to the corresponding value based on the average value of the distances between the insulator detectors corresponding to the image information of each calibrated faulty insulator region; If the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, the image information of the insulator regions obtained by selecting a preset number of corrected insulator detectors is used as the experimental data set to calibrate the cascade model, and the preset number of corrections is adjusted to the corresponding value based on the recognition accuracy evaluation value.

[0046] Specifically, the preset recognition accuracy evaluation value J0 is selected within the interval [420, 480], and the unit of the preset recognition accuracy evaluation value is pixels.

[0047] Specifically, when the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, there is no limitation on the acquisition time of the insulator detectors for the image information of the insulator regions selected for calibrating the cascade model. It can be obtained at random time nodes, which will not be elaborated here.

[0048] Specifically, the layout density of the insulator detectors is the ratio of the number of insulator detectors arranged to the area of the area to be detected.

[0049] Specifically, the operating parameters of the cascaded model are adjusted based on the recognition difference amount, which reflects the recognition situation of the insulator by the one-stage cascaded model. When the recognition difference amount is less than or equal to the preset recognition difference amount, there is a large difference between the number of insulators recognized by the one-stage cascaded model and the actual number of insulators to be detected in the area. At this time, the model has problems during certain time periods and cannot accurately recognize insulators due to the influence of the external environment. In this case, a concentrated time interval is determined, and the image information within this interval is obtained to calibrate the model, specifically calibrating the model. When the recognition difference amount is greater than the preset recognition difference amount, the recognition difference amount is large. In this case, the recognition accuracy evaluation value is combined to further adjust the model. This improves the adaptability and accuracy of the model. The recognition accuracy evaluation value characterizes the general situation of the area enclosed by the model. When the recognition accuracy evaluation value is less than or equal to the preset recognition accuracy evaluation value, the area enclosed by the model is generally small. The model can still accurately recognize defects in small areas. In this case, due to the insufficient layout density of the insulator detectors, the obtained image information is incomplete. In this case, the layout density of the insulator detectors is adjusted. When the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, the areas enclosed by the model are all large. In this case, due to the abnormal recognition accuracy of the model, the abnormalities in small areas cannot be accurately recognized, and there are false detection situations. In this case, the insulator area image information is used as an experimental data set to calibrate the cascaded model. The layout quantity of the insulator detectors is optimized, the accuracy of insulator defect detection is improved, and the detection efficiency of defective insulators is further improved.

[0050] Specifically, the specific process of obtaining the false detection rate includes: Input the insulator image information of the preset verification quantity obtained by the insulator detector into the two-stage cascaded model to obtain the insulator area image information after frame selection and calibration; Determine the quantity of the insulator area image information with incorrect frame selection and calibration, and calculate the ratio of the quantity of incorrect calibration to the preset verification quantity to obtain the false detection rate.

[0051] Specifically, the preset monitoring duration is adjusted to the corresponding value based on the abnormal deviation amount, where: The reduction amplitude of the preset monitoring duration is proportional to the abnormal deviation amount.

[0052] In this embodiment, optionally, Compare the abnormal deviation amount with the first preset deviation amount comparison threshold and the second preset deviation amount comparison threshold; If the abnormal deviation amount is less than or equal to the first preset deviation amount comparison threshold, adjust the preset monitoring duration to 0.92 times the initial preset monitoring duration; If the abnormal deviation amount is less than or equal to the second preset deviation amount comparison threshold and greater than the first preset deviation amount comparison threshold, the preset monitoring duration is adjusted to 0.82 times the initial preset monitoring duration; If the abnormal deviation amount is greater than the second preset deviation amount comparison threshold, the preset monitoring duration is adjusted to 0.72 times the initial preset monitoring duration; The first preset deviation amount comparison threshold is taken as 1.2P0, and the second preset deviation amount comparison threshold is taken as 1.35P0.

[0053] Specifically, based on the recognition difference amount, the preset correction quantity is adjusted to the corresponding value, where The increase amplitude of the preset correction quantity is inversely proportional to the recognition difference amount.

[0054] In this embodiment, optionally, The recognition difference amount is compared with the first preset recognition comparison threshold and the second preset recognition comparison threshold; If the recognition difference amount is less than or equal to the first preset recognition comparison threshold, the preset correction quantity is adjusted to 1.23 times the initial preset correction quantity; If the recognition difference amount is less than or equal to the second preset recognition comparison threshold and greater than the first preset recognition comparison threshold, the preset correction quantity is adjusted to 1.17 times the initial preset correction quantity; If the recognition difference amount is greater than the second preset recognition comparison threshold, the preset correction quantity is adjusted to 1.11 times the initial preset correction quantity; The first preset recognition comparison threshold is taken as 0.72B0, and the second preset recognition comparison threshold is taken as 0.85B0.

[0055] Specifically, based on the average value of the distances between the insulator detectors corresponding to the image information of each calibrated faulty insulator region, the layout density of the insulator detectors is adjusted to the corresponding value, where The increase amplitude of the layout density is inversely proportional to the average value of the distances.

[0056] In this embodiment, optionally, The average value of the distances between the insulator detectors corresponding to the image information of each calibrated faulty insulator region is denoted as the average distance value; The average distance value is compared with the first preset average distance value and the second preset average distance value; If the average distance value is less than or equal to the first preset average distance value, the layout density of the insulator detectors is adjusted to 1.23 times the initial layout density; If the average distance value is less than or equal to the second preset average distance value and greater than the first preset average distance value, the layout density of the insulator detectors is adjusted to 1.18 times the initial layout density; If the distance average value is greater than the second preset distance average value, adjust the layout density of the insulator detector to 1.12 times the initial layout density; The first preset distance average value is taken as 10 m, and the second preset distance average value is taken as 20 m.

[0057] Specifically, adjust the preset correction quantity to the corresponding value based on the recognition accuracy evaluation value, where The increase amplitude of the preset correction quantity is proportional to the recognition accuracy evaluation value.

[0058] In this embodiment, optionally, Compare the recognition accuracy evaluation value with the first preset evaluation value comparison threshold and the second preset evaluation value comparison threshold; If the recognition accuracy evaluation value is less than or equal to the first preset evaluation value comparison threshold, adjust the preset correction quantity to 1.2 times the initial preset correction quantity; If the recognition accuracy evaluation value is less than or equal to the second preset evaluation value comparison threshold and greater than the first preset evaluation value comparison threshold, adjust the preset correction quantity to 1.3 times the initial preset correction quantity; If the recognition accuracy evaluation value is greater than the second preset evaluation value comparison threshold, adjust the preset correction quantity to 1.35 times the initial preset correction quantity; The first preset evaluation value comparison threshold is taken as 1.2J0, and the second preset evaluation value comparison threshold is taken as 1.4J0.

[0059] So far, the technical solution of the present invention has been described in combination with the preferred embodiments shown in the drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or replacements to the relevant technical features, and the technical solutions after these changes or replacements will all fall within the protection scope of the present invention.

[0060] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent detection and recognition method for remotely monitoring the defects of catenary insulators, characterized in that, Including: S1. Using the insulator region image information obtained by the insulator detector as an experimental data set to train a one-stage cascade model for outputting the intercepted and corrected insulator image information; S2. Using the insulator image information as a training set to train a two-stage cascade model for bounding and calibrating the insulator defect region; S3. Sequentially inputting the insulator image information obtained by each insulator detector into the one-stage cascade model and the two-stage cascade model to obtain the bounded and calibrated insulator region image information; S4. Taking the preset monitoring duration as a cycle, periodically determining whether the cascade model operates qualified according to the false detection rate of the two-stage cascade model, including when it is determined that the operation of the cascade model is abnormal, adjusting the operation parameters of the cascade model based on the recognition difference amount, including using the insulator region image information obtained by the insulator detector as an experimental data set to calibrate the cascade model, or adjusting the layout density of the insulator detector to the corresponding value; or, when it is determined that the operation of the cascade model is qualified, continuously using the two-stage cascade model to bound and calibrate each insulator region image information.

2. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 1, characterized in that The process of periodically determining whether the cascade model operates qualified according to the false detection rate of the two-stage cascade model includes: Comparing the obtained false detection rate with a first preset false detection rate and a second preset false detection rate respectively; if the false detection rate is less than or equal to the first preset false detection rate, it is determined that the operation of the cascade model is qualified, and continuously using the two-stage cascade model to bound and calibrate each insulator region image information; if the false detection rate is less than or equal to the second preset false detection rate and greater than the first preset false detection rate, marking the false detection rate of the current cycle as an abnormal false detection rate, and determining whether the cascade model operates qualified in combination with the obtained historical false detection rates; if the false detection rate is greater than the second preset false detection rate, it is determined that the operation of the cascade model is abnormal, and adjusting the operation parameters of the cascade model based on the recognition difference amount; Determining the ratio of the number of insulator image information obtained by the two-stage cascade model within the preset monitoring duration to the preset number of insulators as the recognition difference amount.

3. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 2, characterized in that, The process of determining whether the cascade model operates qualified in combination with the obtained historical false detection rates includes: Calculating the ratio of the number of abnormal false detection rates in the historical false detection rates to the total number of historical false detection rates to obtain the abnormal frequency; Calculating the ratio of the abnormal false detection rate of the current cycle to the average value of each historical false detection rate to obtain the abnormal deviation amount; if the abnormal frequency is less than or equal to the preset abnormal frequency and the abnormal deviation amount is greater than the preset deviation amount, it is determined that the operation of the cascade model is qualified, continuously using the two-stage cascade model to bound and calibrate each insulator region image information, and adjusting the preset monitoring duration to the corresponding value based on the abnormal deviation amount; if the abnormal frequency is greater than the preset abnormal frequency and / or the abnormal deviation amount is less than or equal to the preset deviation amount, it is determined that the operation of the cascade model is abnormal, and adjusting the operation parameters of the cascade model based on the recognition difference amount.

4. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 3, characterized in that, The process of adjusting the operation parameters of the cascade model based on the recognition difference amount includes: If the identified difference amount is less than or equal to the preset identified difference amount, obtain the time nodes when the insulator detector acquires the image information of each calibrated faulty insulator area, and determine the concentrated time intervals of each time node; Use the image information of the preset calibration number of insulator areas acquired by the insulator detector within the concentrated time interval as the experimental data set to calibrate the cascade model; Adjust the preset calibration number to the corresponding value based on the identified difference amount.

5. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 4, characterized in that The process of adjusting the operating parameters of the cascade model based on the identified difference amount further includes: If the identified difference amount is greater than the preset identified difference amount, adjust the cascade model by combining the recognition accuracy evaluation values determined from the image information of each insulator area; Calculate the average value of the areas of the framed regions of the image information of each insulator area to obtain the recognition accuracy evaluation value; If the recognition accuracy evaluation value is less than or equal to the preset recognition accuracy evaluation value, adjust the layout density of the insulator detectors to the corresponding value based on the average value of the distances between the insulator detectors corresponding to the image information of each calibrated faulty insulator area; If the recognition accuracy evaluation value is greater than the preset recognition accuracy evaluation value, select the image information of the insulator areas acquired by the preset calibration number of insulator detectors as the experimental data set to calibrate the cascade model, and adjust the preset calibration number to the corresponding value based on the recognition accuracy evaluation value.

6. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 5, characterized in that, The specific process of obtaining the false detection rate includes, Input the image information of the preset verification number of insulator images acquired by the insulator detector into the cascade model to obtain the image information of the framed and calibrated insulator areas; Determine the number of the image information of the insulator areas where the framing and calibration are faulty, calculate the ratio of the number of faulty calibrations to the preset verification number, and obtain the false detection rate.

7. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 6, characterized in that, Adjust the preset monitoring duration to the corresponding value based on the abnormal deviation amount, where the reduction amplitude of the preset monitoring duration is proportional to the abnormal deviation amount.

8. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 7, characterized in that, Adjust the preset calibration number to the corresponding value based on the identified difference amount, where the increase amplitude of the preset calibration number is inversely proportional to the identified difference amount.

9. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 8, characterized in that, Adjust the layout density of the insulator detectors to the corresponding value based on the average value of the distances between the insulator detectors corresponding to the image information of each calibrated faulty insulator area, where the increase amplitude of the layout density is inversely proportional to the average value of the distances.

10. The intelligent detection and recognition method for remotely monitoring the defects of catenary insulators according to claim 9, wherein, Adjust the preset calibration number to the corresponding value based on the recognition accuracy evaluation value, where the increase amplitude of the preset calibration number is proportional to the recognition accuracy evaluation value.

Images