Multi-model transmission line forest fire point identification and positioning method

Abstract

The embodiment of the invention discloses a multi-model power transmission line forest fire point identification and positioning method, which comprises the steps of obtaining a to-be-identified and positioned power transmission line image, inputting the to-be-identified and positioned power transmission line image into a preset CNN model, and obtaining a first target power transmission line image, and obtaining target intermediate layer features output by an intermediate layer of the preset CNN model, inputting the target intermediate layer features and the to-be-identified and positioned power transmission line image into a preset SVM model to obtain a second target power transmission line image, and then inputting the to-be-identified and positioned power transmission line image into a preset Teager energy operator model to obtain a second target power transmission line image. And finally, determining a forest fire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image and the third target power transmission line image, so that the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power systems, and in particular to a method for identifying and locating mountain fire points of a multi-model power transmission line. Background Art

[0002] With the continuous development of the power system and the increase in electricity demand, the safe monitoring and maintenance of transmission lines have become particularly important. Among them, the timely and accurate identification and positioning of wildfire points along the transmission lines has become a technical problem that needs to be solved urgently.

[0003] Traditional methods for identifying and locating wildfire points on transmission lines mainly rely on manual inspections or algorithms based on simple image processing. However, due to limitations in real-time response speed and identification and positioning accuracy, they are unable to meet the needs of modern power grid management.

[0004] In recent years, with the rapid development of network model technology, especially the continuous emergence of advanced technologies such as deep learning and machine learning, new hope has been brought to the identification and positioning of wildfire points on power transmission lines. However, although the wildfire point identification and positioning method of power transmission lines based on network models has improved the efficiency and accuracy of identification and positioning to a certain extent, there are still many shortcomings, such as low computational efficiency caused by excessive model complexity, or poor adaptability to complex environmental changes, which limits its effect in practical applications, that is, there are still problems of low identification and positioning efficiency and inaccurate identification and positioning. Summary of the invention

[0005] Based on this, it is necessary to propose a multi-model method for identifying and locating wildfire points on transmission lines to address the above problems. This method can effectively solve the problems of low efficiency and inaccurate identification and positioning in the existing technology, and realize efficient and accurate identification and positioning of wildfire points on transmission lines, providing strong technical guarantee for the safe operation of the power grid.

[0006] To achieve the above objectives, the present invention provides, in a first aspect, a multi-model method for identifying and locating wildfire points on power transmission lines, the method comprising:

[0007] Acquire an image of a power transmission line to be identified and located;

[0008] Inputting the power transmission line image to be identified and located into a preset CNN model to obtain a first target power transmission line image, and obtaining target intermediate layer features output by an intermediate layer of the preset CNN model;

[0009] Inputting the target intermediate layer features and the power transmission line image to be identified and located into a preset SVM model to obtain a second target power transmission line image;

[0010] Inputting the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image;

[0011] A wildfire point identification and positioning result is determined based on the first target power transmission line image, the second target power transmission line image, and the third target power transmission line image.

[0012] Optionally, the method further comprises:

[0013] Inputting the to-be-identified and located power transmission line image, the first target power transmission line image, and the second target power transmission line image into a preset random forest model to obtain a fourth target power transmission line image;

[0014] Then, determining the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image, and the third target power transmission line image includes:

[0015] The wildfire point identification and positioning result is determined according to the first target power transmission line image, the second target power transmission line image, the third target power transmission line image, and the fourth target power transmission line image.

[0016] Optionally, determining the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image, the third target power transmission line image, and the fourth target power transmission line image includes:

[0017] Determine a first initial wildfire point identification and positioning result according to the first target power transmission line image;

[0018] Determine a second initial wildfire point identification and positioning result according to the second target power transmission line image;

[0019] Determine a third initial wildfire point identification and positioning result according to the third target power transmission line image;

[0020] Determining a fourth initial wildfire point identification and positioning result according to the fourth target power transmission line image;

[0021] The mountain fire point identification and positioning result is determined according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result.

[0022] Optionally, determining the mountain fire point identification and positioning result according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result includes:

[0023] Determine the error between any two of the initial mountain fire point identification and positioning results according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result;

[0024] The wildfire point identification and positioning results are determined according to the error results between the two initial wildfire point identification and positioning results.

[0025] Optionally, determining the fire point identification and positioning result according to the error result between the two initial fire point identification and positioning results includes:

[0026] Among the error results between the initial wildfire point identification and positioning results, the two error results with the smallest error results are taken as the target error results;

[0027] The average value of the three initial wildfire point identification and positioning results corresponding to the target error result is used as the wildfire point identification and positioning result.

[0028] Optionally, inputting the to-be-identified and located power transmission line image into a preset Teager energy operator model to obtain a third target power transmission line image includes:

[0029] The transmission line image to be identified and located is input into the preset Teager energy operator model: the preset Teager energy operator model extracts the feature vector of the transmission line image to be identified and located; using the Teager energy algorithm, multiple energy amplitudes are determined according to multiple feature points in the feature vector; according to the multiple energy amplitudes and the characteristic energy amplitude of the wildfire point obtained by the preset Teager energy operator model during training, the third target transmission line image is obtained.

[0030] Optionally, the using the Teager energy algorithm to determine a plurality of energy amplitudes according to a plurality of feature points in the feature vector includes:

[0031] When the multiple feature points in the feature vector are continuous feature points, the formula determining a plurality of energy amplitudes;

[0032] In the case where the plurality of feature points in the feature vector are discrete feature points, a plurality of energy amplitudes are determined using the formula ψ[f(i)]=[f(i)]-f(i+1)f(i-1);

[0033] Among them, the conversion relationship between ψ[f(t)] and ψ[f(i)] is:

[0034] Wherein, ψ[f(t)] is the energy amplitude obtained by two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, is the first-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, f(t) is the feature point in the feature vector at the extraction sampling time t of the feature vector, is the second-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, ψ[f(i)] is the energy amplitude obtained by three adjacent feature points in the feature vector, f(i) is the i-th feature point in the feature vector, f(i+1) is the i+1-th feature point in the feature vector, f(i-1) is the i-1-th feature point in the feature vector, and T is the extraction sampling period of the feature vector.

[0035] Optionally, the method further comprises:

[0036] Acquire multiple historical transmission line images to be identified and located;

[0037] For each historical transmission line image to be identified and located, wildfire point identification and location marking are performed to obtain multiple historical target transmission line images;

[0038] Inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training in sequence to obtain the preset Teager energy operator model;

[0039] The step of sequentially inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training to obtain the preset Teager energy operator model includes:

[0040] The nth historical transmission line image to be identified and located and the nth historical target transmission line image are input into the initial Teager energy operator model for the nth training: the initial Teager energy operator model extracts the first feature vector of the nth historical transmission line image to be identified and located, and extracts the second feature vector of the nth historical target transmission line image; using the Teager energy algorithm, multiple first energy amplitudes are determined according to multiple first feature points in the first feature vector, and multiple second energy amplitudes are determined according to multiple second feature points in the second feature vector; the initial wildfire point characteristic energy amplitude is determined according to the multiple first energy amplitudes and the multiple second energy amplitudes; wherein n successively takes integers greater than 0 until n is equal to the total number of historical transmission line images to be identified and located, and the initial wildfire point characteristic energy amplitude obtained in each training is obtained;

[0041] Determine the target fire point characteristic energy amplitude according to all the initial fire point characteristic energy amplitudes obtained through training;

[0042] Using the initial Teager energy operator model having the characteristic energy amplitude of the target wildfire point as the predicted Teager energy operator model;

[0043] The target wildfire point characteristic energy amplitude is used as the wildfire point characteristic energy amplitude obtained during the training of the preset Teager energy operator model.

[0044] Optionally, the transmission line image to be identified and located is input into a preset CNN model to obtain target intermediate layer features output by an intermediate layer of the preset CNN model, including:

[0045] The image of the power transmission line to be identified and located is input into the preset CNN model: the middle layer of the preset CNN model performs convolution operation on the image of the power transmission line to be identified and located by means of a sliding window to obtain the target middle layer features.

[0046] Optionally, the acquiring the image of the power transmission line to be identified and located includes:

[0047] Obtaining original transmission line images;

[0048] The original power transmission line image is processed in sequence by median filtering, Gaussian filtering, bilateral filtering, histogram equalization, adaptive histogram equalization, contrast enhancement, threshold segmentation, edge segmentation and region segmentation to obtain the power transmission line image to be identified and located.

[0049] To achieve the above object, the present invention provides, in a second aspect, a multi-model transmission line fire point identification and positioning device, the device comprising:

[0050] An acquisition module, used to acquire an image of a power transmission line to be identified and located;

[0051] A first model prediction module is used to input the transmission line image to be identified and located into a preset CNN model to obtain a first target transmission line image and obtain target intermediate layer features of the intermediate layer output of the preset CNN model;

[0052] A second model prediction module is used to input the target intermediate layer features and the power transmission line image to be identified and located into a preset SVM model to obtain a second target power transmission line image;

[0053] A third model prediction module is used to input the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image;

[0054] A determination module is used to determine a wildfire point identification and positioning result based on the first target power transmission line image, the second target power transmission line image and the third target power transmission line image.

[0055] To achieve the above-mentioned object, the present invention provides in a third aspect a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the method as described in any one of the first aspects.

[0056] To achieve the above-mentioned purpose, the present invention provides a computer device in a fourth aspect, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the method as described in any one of the first aspects.

[0057] The embodiment of the present invention has the following beneficial effects: the above method obtains the transmission line image to be identified and located, and then inputs the transmission line image to be identified and located into a preset CNN model to obtain a first target transmission line image, and obtains the target intermediate layer features of the intermediate layer output of the preset CNN model, and then inputs the target intermediate layer features and the transmission line image to be identified and located into a preset SVM model to obtain a second target transmission line image, and then inputs the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image, and finally determines the wildfire point identification and positioning result according to the first target transmission line image, the second target transmission line image and the third target transmission line image; that is, the wildfire point identification and positioning result is determined by integrating the image feature extraction capability of the preset CNN model, the nonlinear classification advantage of the preset SVM model and the signal feature detection advantage of the preset Teager energy operator model, which can effectively solve the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art, and realizes efficient and accurate identification and positioning of wildfire points of transmission lines, so as to provide a strong technical guarantee for the safe operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] in:

[0060] Figure 1 A schematic diagram of a multi-model method for identifying and locating wildfire points on power transmission lines in an embodiment of the present application;

[0061] Figure 2 A schematic diagram of a multi-model transmission line fire point identification and positioning device in an embodiment of the present application;

[0062] Figure 3 1 is a diagram of the internal structure of a computer device in some embodiments. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] With the continuous development of the power system and the increase in electricity demand, the safe monitoring and maintenance of transmission lines have become particularly important. Among them, the timely and accurate identification and positioning of wildfire points along the transmission lines has become a technical problem that needs to be solved urgently.

[0065] Traditional methods for identifying and locating wildfire points on transmission lines mainly rely on manual inspections or algorithms based on simple image processing. However, due to limitations in real-time response speed and identification and positioning accuracy, they are unable to meet the needs of modern power grid management.

[0066] In recent years, with the rapid development of network model technology, especially the continuous emergence of advanced technologies such as deep learning and machine learning, new hope has been brought to the identification and positioning of wildfire points on power transmission lines. However, although the wildfire point identification and positioning method of power transmission lines based on network models has improved the efficiency and accuracy of identification and positioning to a certain extent, there are still many shortcomings, such as low computational efficiency caused by excessive model complexity, or poor adaptability to complex environmental changes, which limits its effect in practical applications, that is, there are still problems of low identification and positioning efficiency and inaccurate identification and positioning.

[0067] In response to the above problems, the present application proposes a multi-model method for identifying and locating wildfire points on transmission lines, which can effectively solve the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art, and realize efficient and accurate identification and positioning of wildfire points on transmission lines, so as to provide strong technical guarantee for the safe operation of the power grid. The specific implementation principle will be described in detail in the following embodiments.

[0068] In a first aspect, the present application provides a multi-model method for identifying and locating wildfire points on power transmission lines.

[0069] See also Figure 1, is a schematic diagram of a multi-model transmission line fire point identification and positioning method in an embodiment of the present application, the method comprising:

[0070] Step 110: Acquire an image of the power transmission line to be identified and located.

[0071] The transmission line image to be identified and located refers to the transmission line image that needs to be used for wildfire point identification and location.

[0072] In some embodiments, before obtaining the image of the power transmission line to be identified and located, the original power transmission line image may be obtained first, and then the original power transmission line image may be preprocessed to obtain a standard power transmission line image, and finally the standard power transmission line image may be used as the power transmission line image to be identified and located; wherein the preprocessing includes but is not limited to denoising processing, enhancement processing, and segmentation processing.

[0073] It can be understood that by preprocessing the original transmission line image, the image quality can be improved. When there are wildfire points in the image, the characteristics of the wildfire points can be highlighted to provide more accurate input data for subsequent wildfire point identification and positioning tasks.

[0074] Step 120: Input the power transmission line image to be identified and located into a preset CNN model to obtain a first target power transmission line image and target intermediate layer features output by the intermediate layer of the preset CNN model.

[0075] Among them, the preset CNN model here is a pre-trained CNN model that is used to predict and output the first target transmission line image based on the input transmission line image to be identified and located; the first target transmission line image refers to the image in which the wildfire point has been identified and located after prediction using the preset CNN model; the middle layer of the preset CNN model refers to any layer located between the input layer and the output layer.

[0076] In some embodiments, the intermediate layer of the preset CNN model is any one of the convolutional layer, activation layer, pooling layer, normalization layer and fully connected layer of the preset CNN model.

[0077] As for the method of obtaining the preset CNN model, in some embodiments, a large number of different historical transmission lines to be located (i.e., different historical transmission line images to be identified and located, such as transmission lines, time points, and wildfire points) and historical target transmission line images corresponding to each historical transmission line image to be identified and located are collected, and then each historical transmission line image to be identified and located and its corresponding historical target transmission line image are sequentially input into the initial CNN model for training. After training to a certain extent, a trained preset CNN model is obtained; wherein, each historical target transmission line image can be used as the true value of the initial CNN model, that is, by comparing the true value with each first historical target transmission line image output during the training process, it can be determined whether the initial CNN model is well trained and has met the expected requirements.

[0078] Regarding the method of acquiring historical target transmission lines, in some embodiments, after collecting a large number of different historical transmission lines to be located, wildfire point identification and positioning can be performed on each historical transmission line image to be identified and located to obtain multiple historical target transmission line images.

[0079] Of course, the above embodiment is only one of the embodiments provided by the present application. In other embodiments, any CNN model that can predict and output the first target transmission line image based on the input transmission line image to be identified and located can be used as the preset CNN model of the present application, and the present application does not limit this.

[0080] It should be noted that in the process of inputting the transmission line image to be identified and located into the preset CNN model for prediction, identification and positioning are included, that is, the first target transmission line image predicted and output is an image in which the wildfire point is identified and located; and for the image in which the wildfire point cannot be identified, since there is no wildfire point, there is no wildfire point located. At this time, it is assumed that the transmission line image to be identified and located corresponding to the first target transmission line image that is not output after the prediction is an image without a wildfire point.

[0081] That is, in other embodiments, the image of the power transmission line to be identified and located is input into the preset CNN model. If the first target power transmission line image is not obtained, no feedback is given, or the operator is prompted that the image of the power transmission line to be identified and located is an image without a wildfire point.

[0082] Step 130: Input the target intermediate layer features and the power transmission line image to be identified and located into a preset SVM model to obtain a second target power transmission line image.

[0083] Among them, the preset SVM model here is a pre-trained SVM model for predicting and outputting a second target transmission line image based on the input target intermediate layer features and the transmission line image to be identified and located; the second target transmission line image refers to an image in which the wildfire point has been identified and located after prediction using the preset SVM model.

[0084] As for the method of obtaining the preset SVM model, in some embodiments, a large number of different historical transmission lines to be located (i.e., different historical transmission line images to be identified and located, such as transmission lines, time points, and wildfire points), as well as historical target transmission line images and historical target intermediate layer features corresponding to each historical transmission line image to be identified and located are collected, and then each historical transmission line image to be identified and located and its corresponding historical target transmission line image and corresponding historical target intermediate layer features are sequentially input into the initial SVM model for training. After training to a certain extent, a trained preset SVM model is obtained; wherein, each historical target transmission line image can be used as the true value of the initial SVM model, that is, by comparing the true value with each second historical target transmission line image output during the training process, it can be determined whether the initial SVM model is well trained and has met the expected requirements.

[0085] The method for acquiring the historical target transmission line is the same as that in the above embodiment. The relevant contents of step 120 in the above embodiment may be referred to and will not be described in detail here.

[0086] Regarding the method of acquiring the intermediate layer features of historical targets, in some embodiments, each historical transmission line image to be identified and located can be sequentially input into a preset CNN model to obtain multiple intermediate layer features of historical targets sequentially output by the intermediate layer of the preset CNN model.

[0087] Of course, the above embodiment is only one of the embodiments provided by the present application. In other embodiments, as long as the SVM model can predict and output the second target transmission line image based on the input target intermediate layer features and the transmission line image to be identified and located, it can be used as the preset SVM model of the present application, and the present application does not limit this.

[0088] It should be noted that in the process of inputting the target intermediate layer features and the power transmission line image to be identified and located into the preset SVM model for prediction, identification and positioning are included, that is, the second target transmission line image output by the prediction is an image in which the wildfire point is identified and located; and for the image in which the wildfire point cannot be identified, since there is no wildfire point, there is no wildfire point located. At this time, it is assumed that the power transmission line image to be identified and located corresponding to the second target transmission line image that is not output after the prediction is an image without a wildfire point.

[0089] That is, in other embodiments, the target intermediate layer features and the transmission line image to be identified and located are input into a preset SVM model. If the second target transmission line image is not obtained, no feedback is given, or the operator is prompted that the transmission line image to be identified and located is an image without a wildfire point.

[0090] Step 140: input the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image.

[0091] Among them, the preset Teager energy operator model here is a pre-trained Teager energy operator model for predicting and outputting a third target transmission line image based on an input transmission line image to be identified and located; the third target transmission line image refers to an image in which a wildfire point has been identified and located after prediction using the preset Teager energy operator model.

[0092] As for the method of obtaining the preset Teager energy operator model, in some embodiments, a large number of different historical transmission lines to be located (i.e., different historical transmission line images to be identified and located, such as transmission lines, time points, and wildfire points) and historical target transmission line images corresponding to each historical transmission line image to be identified and located are collected, and then each historical transmission line image to be identified and located and its corresponding historical target transmission line image are sequentially input into the initial Teager energy operator model for training. After training to a certain extent, a trained preset Teager energy operator model is obtained; wherein, each historical target transmission line image can be used as the true value of the initial Teager energy operator model, that is, by comparing the true value with each third historical target transmission line image output during the training process, it can be determined whether the initial Teager energy operator model is well trained and has met the expected requirements.

[0093] The method for acquiring the historical target transmission line is the same as that in the above embodiment. The relevant contents of step 120 in the above embodiment may be referred to and will not be described in detail here.

[0094] Of course, the above embodiment is only one of the embodiments provided by the present application. In other embodiments, any Teager energy operator model that can predict and output a third target transmission line image based on the input transmission line image to be identified and located can be used as the preset Teager energy operator model of the present application, and the present application does not limit this.

[0095] It should be noted that in the process of inputting the transmission line image to be identified and located into the preset Teager energy operator model for prediction, identification and positioning are included, that is, the predicted output third target transmission line image is an image that identifies and locates the wildfire point; and for the image in which the wildfire point cannot be identified, since there is no wildfire point, there is no located wildfire point. At this time, it is assumed that the transmission line image to be identified and located corresponding to the third target transmission line image that is not output after the prediction is an image without a wildfire point.

[0096] That is, in other embodiments, the image of the transmission line to be identified and located is input into the preset Teager energy operator model. If the third target transmission line image is not obtained, no feedback is given, or the operator is prompted that the image of the transmission line to be identified and located is an image without a wildfire point.

[0097] Step 150: Determine the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image, and the third target power transmission line image.

[0098] It should be noted that since the first target transmission line image, the second target transmission line image and the third target transmission line image are all images that have identified and located the wildfire point, in some embodiments, the identification and positioning results of the first target transmission line image, the second target transmission line image and the third target transmission line image can be combined to determine the wildfire point identification and positioning result; for example, the method of combining the identification and positioning results of the three can be a voting mechanism, averaging or weighted averaging, etc., which is not limited here.

[0099] In the embodiment of the present application, the wildfire point identification and positioning results are determined by integrating the image feature extraction capability of the preset CNN model, the nonlinear classification advantage of the preset SVM model, and the signal feature detection advantage of the preset Teager energy operator model. This can effectively solve the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art, and achieve efficient and accurate identification and positioning of wildfire points on transmission lines, providing strong technical guarantee for the safe operation of the power grid.

[0100] It should also be noted that in actual applications, the present application can monitor wildfire points in real time, that is, it can obtain images of power transmission lines to be identified and located in real time, and then identify and locate wildfire points. If there is no wildfire point, no feedback is required. If a wildfire point is identified and located, an early warning prompt is required, and the determined wildfire point identification and location results are converted into corresponding geographic location information and output to the operator, so that the operator can perform corresponding operations based on the geographic location information.

[0101] In a feasible implementation, the method in the above embodiment further includes: inputting the transmission line image to be identified and located, the first target transmission line image and the second target transmission line image into a preset random forest model to obtain a fourth target transmission line image.

[0102] Then step 150 in the above embodiment, determining the wildfire point identification and positioning result based on the first target transmission line image, the second target transmission line image and the third target transmission line image, includes: determining the wildfire point identification and positioning result based on the first target transmission line image, the second target transmission line image, the third target transmission line image and the fourth target transmission line image.

[0103] Among them, the preset random forest model here is a pre-trained random forest model for predicting and outputting a fourth target transmission line image based on the input transmission line image to be identified and located, the first target transmission line image and the second target transmission line image; the fourth target transmission line image refers to an image in which a wildfire point has been identified and located after prediction using the preset random forest model.

[0104] As for the method of obtaining the preset random forest model, in some embodiments, a large number of different historical transmission lines to be located (i.e., different historical transmission line images to be identified and located, such as transmission lines, time points, and wildfire points) can be collected, as well as the historical target transmission line image, the first historical target transmission line image, and the second historical target transmission line image corresponding to each historical transmission line image to be identified and located, and then each historical transmission line image to be identified and located and its corresponding historical target transmission line image, the corresponding first historical target transmission line image, and the corresponding second historical target transmission line image are sequentially input into the initial random forest model for training. After training to a certain extent, a trained preset random forest model is obtained; wherein, each historical target transmission line image can be used as the true value of the initial random forest model, that is, by comparing the true value with each fourth historical target transmission line image output during the training process, it can be determined whether the initial random forest model is well trained and has met the expected requirements.

[0105] The method for acquiring the historical target transmission line is the same as that in the above embodiment. The relevant contents of step 120 in the above embodiment may be referred to and will not be described in detail here.

[0106] Regarding the method of acquiring the first historical target transmission line image and the second historical target transmission line image, in some embodiments, each historical transmission line image to be identified and located can be sequentially input into a preset CNN model to obtain multiple first historical target transmission line images, and multiple historical target intermediate layer features sequentially output by the intermediate layer of the preset CNN model; each first historical target transmission line image and its corresponding historical target intermediate layer feature are input into a preset SVM model to obtain multiple second historical target transmission line images.

[0107] Of course, the above embodiment is only one of the embodiments provided by the present application. In other embodiments, as long as the random forest model can predict and output the fourth target transmission line image based on the input transmission line image to be identified and located, the first target transmission line image and the second target transmission line image, it can be used as the preset random forest model of the present application, and the present application does not limit this.

[0108] It should be noted that in the process of inputting the transmission line image to be identified and located, the first target transmission line image and the second target transmission line image into the preset random forest model for prediction, identification and positioning are included, that is, the fourth target transmission line image output by the prediction is an image in which the wildfire point is identified and the wildfire point is located; and for the image in which the wildfire point cannot be identified, since there is no wildfire point, there is no wildfire point located. At this time, it is assumed that the transmission line image to be identified and located corresponding to the fourth target transmission line image that is not output after the prediction is an image without a wildfire point.

[0109] That is, in other embodiments, the image of the transmission line to be identified and located, the first target transmission line image, and the second target transmission line image are input into a preset random forest model. If the fourth target transmission line image is not obtained, no feedback is given, or the operator is prompted that the image of the transmission line to be identified and located is an image without a wildfire point.

[0110] It should be further explained that since the first target transmission line image, the second target transmission line image, the third target transmission line image and the fourth target transmission line image are all images that have identified and located the wildfire point, in some embodiments, the identification and positioning results of the first target transmission line image, the second target transmission line image, the third target transmission line image and the fourth target transmission line image can be combined to determine the wildfire point identification and positioning result; for example, the method of combining the identification and positioning results of these four can be a voting mechanism, averaging or weighted averaging, etc., which is not limited here.

[0111] In the embodiment of the present application, the wildfire point identification and positioning results are determined by integrating the image feature extraction capability of the preset CNN model, the nonlinear classification advantage of the preset SVM model, the signal feature detection advantage of the preset Teager energy operator model, and the efficient feature processing characteristics of the preset random forest model. This can effectively solve the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art, and further realize efficient and accurate identification and positioning of wildfire points on transmission lines, providing strong technical guarantee for the safe operation of the power grid.

[0112] In a feasible implementation method, the method of determining the mountain fire point identification and positioning result based on the first target transmission line image, the second target transmission line image, the third target transmission line image and the fourth target transmission line image in the above embodiment includes: determining the first initial mountain fire point identification and positioning result based on the first target transmission line image; determining the second initial mountain fire point identification and positioning result based on the second target transmission line image; determining the third initial mountain fire point identification and positioning result based on the third target transmission line image; determining the fourth initial mountain fire point identification and positioning result based on the fourth target transmission line image; determining the mountain fire point identification and positioning result based on the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result and the fourth initial mountain fire point identification and positioning result.

[0113] It should be noted that since the first target transmission line image, the second target transmission line image, the third target transmission line image and the fourth target transmission line image are images that have identified and located the wildfire points, in some embodiments, the identification and positioning results of their wildfire points can be determined separately.

[0114] Further, in some embodiments, the first initial wildfire point identification and positioning result, the second initial wildfire point identification and positioning result, the third initial wildfire point identification and positioning result, the fourth initial wildfire point identification and positioning result and the wildfire point identification and positioning result can all be wildfire point identification and positioning coordinates; that is, the coordinates corresponding to the wildfire point identification and positioning result are determined according to the first coordinates corresponding to the first initial wildfire point identification and positioning result, the second coordinates corresponding to the second initial wildfire point identification and positioning result, the coordinates corresponding to the third initial wildfire point identification and positioning result and the coordinates corresponding to the fourth initial wildfire point identification and positioning result.

[0115] In an embodiment of the present application, by adopting a comprehensive analysis strategy of multi-source image data of multiple models (i.e., the first, second, third, and fourth target transmission line images), the recognition and positioning accuracy can be improved, that is, each model has its advantages, and through cross-validation and complementarity of multi-source image data of multiple models (i.e., the first, second, third, and fourth target transmission line images), the location of the wildfire point can be determined more accurately.

[0116] In a feasible implementation method, the method of determining the wildfire point identification and positioning result based on the first initial wildfire point identification and positioning result, the second initial wildfire point identification and positioning result, the third initial wildfire point identification and positioning result, and the fourth initial wildfire point identification and positioning result in the above embodiment includes: determining the error result between the pairwise initial wildfire point identification and positioning results based on the first initial wildfire point identification and positioning result, the second initial wildfire point identification and positioning result, the third initial wildfire point identification and positioning result, and the fourth initial wildfire point identification and positioning result; determining the wildfire point identification and positioning result based on the error result between the pairwise initial wildfire point identification and positioning results.

[0117] It should be noted that since the first initial wildfire point identification and positioning result, the second initial wildfire point identification and positioning result, the third initial wildfire point identification and positioning result, the fourth initial wildfire point identification and positioning result and the wildfire point identification and positioning result can all be wildfire point identification and positioning coordinates, in some embodiments, the Euclidean distance formula can be used to determine the Euclidean distance between the pairwise initial wildfire point identification and positioning results based on the first initial wildfire point identification and positioning result, the second initial wildfire point identification and positioning result, the third initial wildfire point identification and positioning result and the fourth initial wildfire point identification and positioning result, and the Euclidean distance between the pairwise initial wildfire point identification and positioning results is used as the error result between the pairwise initial wildfire point identification and positioning results.

[0118] In the embodiment of the present application, the final wildfire point identification and positioning result is determined by integrating multiple initial wildfire point identification and positioning results (i.e., the first to fourth initial wildfire point identification and positioning results), and paying special attention to the error between the two initial results, which can effectively improve the identification and positioning accuracy, enhance robustness, etc.

[0119] It can be understood that the accuracy of identification and positioning can be improved: by calculating the errors between the initial wildfire point identification and positioning results, it is possible to identify which results are closer or have significant deviations. This error analysis helps to eliminate or adjust those results with large deviations, thereby obtaining more accurate wildfire point identification and positioning; enhance robustness: in actual applications, due to the influence of various external factors (such as weather, light, equipment accuracy, model accuracy, etc.), a single wildfire point identification and positioning result may be uncertain. By comprehensively considering multiple initial wildfire point identification and positioning results and calculating their errors, the ability to resist adverse factors can be significantly enhanced, thereby improving the robustness of identification and positioning.

[0120] In a feasible implementation method, the above-mentioned embodiment determines the wildfire point identification and positioning result based on the error result between the pairwise initial wildfire point identification and positioning results, including: among the error results between the pairwise initial wildfire point identification and positioning results, taking the two error results with the smallest error results as the target error results; and taking the average value of the three initial wildfire point identification and positioning results corresponding to the target error results as the wildfire point identification and positioning result.

[0121] It should be noted that the present application uses the two error results with the smallest error results as the target error results instead of using the error result with the smallest error result as the target error result in order to avoid randomness. This method can further improve its recognition and positioning accuracy.

[0122] Of course, in other embodiments, the error result with the smallest error result between the initial wildfire point identification and positioning results may be used as the target error result, and then the average value of the two initial wildfire point identification and positioning results corresponding to the target error result may be used as the wildfire point identification and positioning result.

[0123] In an embodiment of the present application, by selecting the two error results with the smallest error results as the target error results, and taking the average of the three initial wildfire point identification and positioning results corresponding to these two error results as the final wildfire point identification and positioning result, the identification and positioning accuracy can be further improved, the data reliability can be enhanced, and the calculation efficiency can be optimized.

[0124] It can be understood that improving the accuracy of identification and positioning: by selecting the two error results with the smallest error, it can be ensured that the final identification and positioning result is based on the most reliable and accurate initial data, which helps to reduce the problem of inaccurate identification and positioning caused by individual data deviations or large errors, thereby improving the overall identification and positioning accuracy; enhancing data reliability: by comparing the errors of different initial wildfire point identification and positioning results, it can be identified which data is more reliable and which data may have large deviations. This screening mechanism helps to enhance the data reliability of the final identification and positioning results; optimizing calculation efficiency: selecting the two error results with the smallest error from multiple initial wildfire point identification and positioning results for calculation can significantly reduce the amount of calculation and improve calculation efficiency compared to averaging or weighting all results. At the same time, since only the results with the smallest error are focused on, it can converge to accurate identification and positioning results more quickly.

[0125] In a feasible implementation, step 140 in the above embodiment, inputting the image of the transmission line to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image, includes: inputting the image of the transmission line to be identified and located into the preset Teager energy operator model: the preset Teager energy operator model extracts the feature vector of the transmission line image to be identified and located; using the Teager energy algorithm to determine multiple energy amplitudes based on multiple feature points in the feature vector; and obtaining the third target transmission line image based on the multiple energy amplitudes and the characteristic energy amplitude of the wildfire point obtained by the preset Teager energy operator model during training.

[0126] In an embodiment of the present application, the transmission line image to be identified and located is processed by a preset Teager energy operator model, and the energy amplitude of the feature points in the feature vector of the transmission line image to be identified and located is used to identify and locate the wildfire point, so as to obtain a third target transmission line image, which can effectively improve the recognition and positioning accuracy.

[0127] It can be understood that the accuracy of recognition and positioning is improved: the Teager energy algorithm is used to determine multiple energy amplitudes based on multiple feature points in the feature vector. The Teager energy operator is very sensitive to the sudden changes in the signal and can effectively reflect local changes in the image, such as the flickering of flames, the spread of smoke, etc. This is particularly important for the detection of abnormal situations such as wildfires; further, by comparing the calculated multiple energy amplitudes with the wildfire point characteristic energy amplitudes obtained by the preset Teager energy operator model during training, the wildfire points in the transmission line image to be identified and positioned can be accurately identified and positioned.

[0128] In addition, this method overcomes the sensitivity of traditional image processing methods to interference factors such as lighting changes and complex backgrounds, and improves the accuracy and robustness of recognition and positioning.

[0129] In a feasible implementation, the Teager energy algorithm in the above embodiment is used to determine multiple energy amplitudes according to multiple feature points in the feature vector, including:

[0130] When multiple feature points in the feature vector are continuous feature points, the formula is used determining a plurality of energy amplitudes;

[0131] When the multiple feature points in the feature vector are discrete feature points, multiple energy amplitudes are determined using the formula ψ[f(i)]=[f(i)]-f(i+1)f(i-1);

[0132] Among them, the conversion relationship between ψ[f(t)] and ψ[f(i)] is:

[0133] Among them, ψ[f(t)] is the energy amplitude obtained by two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, is the first-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, f(t) is the feature point in the feature vector at the extraction sampling time t of the feature vector, is the second-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, ψ[f(i)] is the energy amplitude obtained by three adjacent feature points in the feature vector, f(i) is the i-th feature point in the feature vector, f(i+1) is the i+1-th feature point in the feature vector, f(i-1) is the i-1-th feature point in the feature vector, and T is the extraction sampling period of the feature vector.

[0134] In the embodiments of the present application, a rigorous energy amplitude calculation formula is provided from a mathematical perspective. The accuracy of the calculated energy amplitude can be ensured from the rigor of mathematical logic, and the above calculation formula is preferably shown to provide reference, understanding and calculation for technical personnel; in addition, by adopting the above calculation formula to calculate the energy amplitude, a more accurate energy amplitude can be obtained to improve the recognition and positioning accuracy.

[0135] In addition, this method can flexibly process continuous feature points and discrete feature points in the feature vector and calculate the energy amplitude through different formulas. This flexibility enables the algorithm to be applied to a wider range of scenarios, improving the adaptability and versatility of the algorithm. The calculations in the above calculation formula are relatively simple and mainly involve basic mathematical operations (such as addition, multiplication and differentiation), which makes the algorithm computationally efficient. This high efficiency is particularly important when processing transmission line images in real time.

[0136] In a feasible implementation, the method in the above embodiment also includes: obtaining multiple historical transmission line images to be identified and located; performing wildfire point identification and positioning annotation on each historical transmission line image to be identified and located, and obtaining multiple historical target transmission line images; inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training in turn, and obtaining a preset Teager energy operator model; inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training in turn, and obtaining a preset Teager energy operator model, including: inputting the nth historical transmission line image to be identified and located and the nth historical target transmission line image into the initial Teager energy operator model for the nth training process: the initial Teager energy operator model extracts the nth historical transmission line image to be identified and located The invention relates to a method for extracting a first feature vector of a power transmission line image and a second feature vector of an nth historical target power transmission line image; using a Teager energy algorithm, determining a plurality of first energy amplitudes according to a plurality of first feature points in the first feature vector, and determining a plurality of second energy amplitudes according to a plurality of second feature points in the second feature vector; determining an initial wildfire point characteristic energy amplitude according to the plurality of first energy amplitudes and the plurality of second energy amplitudes; wherein n successively takes integers greater than 0 until n is equal to the total number of historical power transmission line images to be identified and located, and obtaining an initial wildfire point characteristic energy amplitude obtained in each training; determining a target wildfire point characteristic energy amplitude according to the initial wildfire point characteristic energy amplitudes obtained in all training; using an initial Teager energy operator model having a target wildfire point characteristic energy amplitude as a predicted Teager energy operator model; and using the target wildfire point characteristic energy amplitude as the wildfire point characteristic energy amplitude obtained by the preset Teager energy operator model during training.

[0137] Among them, multiple historical transmission line images to be identified and located refer to the large number of different multiple historical transmission lines to be located mentioned in the above embodiments (that is, multiple historical transmission line images to be identified and located with different transmission lines, time points and wildfire points, etc.).

[0138] The manner of determining the first energy amplitude and the second energy amplitude is similar to that in the above embodiment, and reference may be made to the relevant contents in the above embodiment, which will not be described again here.

[0139] It should be noted that in the present application, during the training process, the initial wildfire point characteristic energy amplitude obtained in each training is obtained, and then the initial wildfire point characteristic energy amplitudes obtained in all trainings can be compared to determine the target wildfire point characteristic energy amplitude.

[0140] In an embodiment of the present application, multiple historical images of transmission lines to be identified and located are obtained, annotated and trained as an initial Teager energy operator model. During the training process, the characteristic energy amplitudes of the initial wildfire points obtained from all trainings are compared to determine the characteristic energy amplitudes of the target wildfire points, and a preset Teager energy operator model is obtained, which can improve the accuracy of the model and the depth of feature learning.

[0141] It can be understood that improving model accuracy: by using multiple historical transmission line images to be identified and their corresponding historical target transmission line images for training, the model can learn more diverse wildfire point features, which helps to improve the model's accuracy and generalization ability when identifying new transmission line images to be identified; the depth of feature learning: during the training process, the Teager energy algorithm is used to calculate the energy amplitude corresponding to the feature points in the feature vectors of different historical transmission lines to be identified. This deep feature learning enables the model to more accurately capture the unique characteristics of wildfire points.

[0142] In a feasible implementation, step 120 in the above embodiment, inputting the image of the power transmission line to be identified and located into a preset CNN model to obtain target intermediate layer features output by the intermediate layer of the preset CNN model, includes: inputting the image of the power transmission line to be identified and located into the preset CNN model: the intermediate layer of the preset CNN model performs a convolution operation on the image of the power transmission line to be identified and located by means of a sliding window to obtain target intermediate layer features.

[0143] In some embodiments, the intermediate layer of the preset CNN model of the present application may be composed of multiple convolutional layers.

[0144] In an embodiment of the present application, by inputting the image of the power transmission line to be identified and located into a preset CNN model, a convolution operation is performed on the image using a sliding window through the intermediate layer of the preset CNN model to obtain the target intermediate layer features, which can improve the feature extraction efficiency, enhance feature representation, and achieve accurate identification and positioning.

[0145] It can be understood that the efficiency of feature extraction is improved: by sliding the window, the model traverses the entire image of the power line to be identified and located with a certain step size when processing the image, and performs convolution operations on each local area. This method can significantly reduce the computational complexity and improve the efficiency of feature extraction; enhance feature representation: by applying sliding windows for convolution in multiple convolutional layers in the middle layer, the feature information of the power line image to be identified and located at different scales and positions can be captured. These multi-scale feature representations help improve the model's ability to understand complex scenes; achieve accurate identification and positioning: the convolution operation of the sliding window not only extracts features, but also retains the position information of the features in the image of the power line to be identified and located, which can achieve accurate identification and positioning.

[0146] In a feasible implementation, the acquisition of the image of the power transmission line to be identified and located in the above embodiment includes: acquiring the original power transmission line image; performing median filtering, Gaussian filtering, bilateral filtering, histogram equalization, adaptive histogram equalization, contrast enhancement, threshold segmentation, edge segmentation and region segmentation on the original power transmission line image in sequence to obtain the image of the power transmission line to be identified and located.

[0147] Among them, median filtering, Gaussian filtering and bilateral filtering are denoising processing; histogram equalization, adaptive histogram equalization and contrast enhancement are enhancement processing; threshold segmentation, edge segmentation and region segmentation are segmentation processing.

[0148] Regarding the method of acquiring the original transmission line image, in some embodiments, a drone, a surveillance camera, or other image acquisition equipment may be used to shoot to obtain the original transmission line image.

[0149] In an embodiment of the present application, the original transmission line image is sequentially subjected to median filtering, Gaussian filtering, bilateral filtering, histogram equalization, adaptive histogram equalization, contrast enhancement, threshold segmentation, edge segmentation and region segmentation to obtain a transmission line image to be identified and located, namely: through median filtering, a filter can be used to replace the value of each pixel in the original transmission line image with the median of its surrounding pixels to remove discrete noise such as salt and pepper noise; through Gaussian filtering, a Gaussian kernel can be used to smooth the original transmission line image to reduce continuous noise such as Gaussian noise; through bilateral filtering, the information of the original transmission line image in the spatial domain and the grayscale value domain can be combined to remove noise while maintaining edge information; through histogram equalization, the original transmission line image can be adjusted Pixel value distribution can enhance the contrast and details of the original transmission line image, making the wildfire points more prominent; the original transmission line image can be divided into small blocks through adaptive histogram equalization processing, and each block can be histogram equalized to avoid excessive enhancement caused by global equalization; the contrast and brightness of the original transmission line image can be adjusted through contrast enhancement processing to highlight the details of the wildfire points; the threshold segmentation processing can compare the grayscale value of the pixels in the original transmission line with the preset threshold, and divide the pixels into foreground and background, so that the grayscale difference between the wildfire points and the background is large; the edge segmentation processing can detect the edge information in the original transmission line image, so that the wildfire points are separated from the background; the region segmentation processing can divide the original transmission line image into regions with similar features, and the wildfire points are segmented out by the connectivity and similarity of the regions.

[0150] In a second aspect, the present application provides a multi-model wildfire point identification and positioning device for power transmission lines.

[0151] See also Figure 2 , is a schematic diagram of a multi-model transmission line fire point identification and positioning device in an embodiment of the present application, the device 210 includes:

[0152] An acquisition module 211 is used to acquire an image of a power transmission line to be identified and located;

[0153] A first model prediction module 212 is used to input the transmission line image to be identified and located into a preset CNN model to obtain a first target transmission line image and a target intermediate layer feature output by an intermediate layer of the preset CNN model;

[0154] A second model prediction module 213 is used to input the target intermediate layer features and the power transmission line image to be identified and located into a preset SVM model to obtain a second target power transmission line image;

[0155] The third model prediction module 214 is used to input the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image;

[0156] The determination module 215 is used to determine the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image and the third target power transmission line image.

[0157] In the embodiment of the present application, the relevant contents of the acquisition module 211, the first model prediction module 212, the second model prediction module 213, the third model prediction module 214 and the determination module 215 can be referred to. Figure 1 The contents in the illustrated embodiment will not be described in detail here.

[0158] It should be noted that the device 210 of the present application also includes some other modules. It can be understood that the method of the present application and the device 210 have a one-to-one correspondence. Therefore, some other modules of the device 210 of the present application are the contents corresponding to the method of the present application in the above-mentioned embodiment.

[0159] In the embodiment of the present application, the wildfire point identification and positioning results are determined by integrating the image feature extraction capability of the preset CNN model, the nonlinear classification advantage of the preset SVM model, and the signal feature detection advantage of the preset Teager energy operator model. This can effectively solve the problems of low identification and positioning efficiency and inaccurate identification and positioning in the prior art, and achieve efficient and accurate identification and positioning of wildfire points on transmission lines, providing strong technical guarantee for the safe operation of the power grid.

[0160] In a third aspect, the present application further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes a multi-model method for identifying and locating wildfire points on power transmission lines in the above method embodiment.

[0161] In a fourth aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes a multi-model method for identifying and locating wildfire points on power transmission lines in the above method embodiment.

[0162] Figure 3 The internal structure diagram of a computer device in some embodiments is shown. The computer device may be a terminal, a server, or a gateway. Figure 3 As shown, the computer device includes a processor, a memory and a network interface connected through a system bus.

[0163] The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement each step in the above method embodiment. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can implement each step in the above method embodiment. Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0165] Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0166] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0167] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A multi-model transmission line fire point identification and positioning method, characterized in that: The method comprises: Acquire an image of a power transmission line to be identified and located; Inputting the power transmission line image to be identified and located into a preset CNN model to obtain a first target power transmission line image, and obtaining target intermediate layer features output by an intermediate layer of the preset CNN model; Inputting the target intermediate layer features and the power transmission line image to be identified and located into a preset SVM model to obtain a second target power transmission line image; Inputting the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image; A wildfire point identification and positioning result is determined based on the first target power transmission line image, the second target power transmission line image, and the third target power transmission line image.

2. The method according to claim 1, characterized in that The method further comprises: Inputting the to-be-identified and located power transmission line image, the first target power transmission line image, and the second target power transmission line image into a preset random forest model to obtain a fourth target power transmission line image; Then, determining the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image, and the third target power transmission line image includes: The wildfire point identification and positioning result is determined according to the first target power transmission line image, the second target power transmission line image, the third target power transmission line image, and the fourth target power transmission line image.

3. The method according to claim 2, characterized in that The determining the wildfire point identification and positioning result according to the first target power transmission line image, the second target power transmission line image, the third target power transmission line image, and the fourth target power transmission line image includes: Determine a first initial wildfire point identification and positioning result according to the first target power transmission line image; Determine a second initial wildfire point identification and positioning result according to the second target power transmission line image; Determine a third initial wildfire point identification and positioning result according to the third target power transmission line image; Determining a fourth initial wildfire point identification and positioning result according to the fourth target power transmission line image; The mountain fire point identification and positioning result is determined according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result.

4. The method according to claim 3, characterized in that The determining the mountain fire point identification and positioning result according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result includes: Determine the error result between the initial mountain fire point identification and positioning results according to the first initial mountain fire point identification and positioning result, the second initial mountain fire point identification and positioning result, the third initial mountain fire point identification and positioning result, and the fourth initial mountain fire point identification and positioning result; The wildfire point identification and positioning results are determined according to the error results between the two initial wildfire point identification and positioning results.

5. The method according to claim 4, characterized in that The step of determining the fire point identification and positioning result according to the error result between the two initial fire point identification and positioning results comprises: Among the error results between the initial wildfire point identification and positioning results, the two error results with the smallest error results are taken as the target error results; The average value of the three initial wildfire point identification and positioning results corresponding to the target error result is used as the wildfire point identification and positioning result.

6. The method according to claim 1, characterized in that The step of inputting the transmission line image to be identified and located into a preset Teager energy operator model to obtain a third target transmission line image includes: The transmission line image to be identified and located is input into the preset Teager energy operator model: the preset Teager energy operator model extracts the feature vector of the transmission line image to be identified and located; using the Teager energy algorithm, multiple energy amplitudes are determined according to multiple feature points in the feature vector; according to the multiple energy amplitudes and the characteristic energy amplitude of the wildfire point obtained by the preset Teager energy operator model during training, the third target transmission line image is obtained.

7. The method according to claim 6, characterized in that The method of using the Teager energy algorithm to determine a plurality of energy amplitudes according to a plurality of feature points in the feature vector includes: When the multiple feature points in the feature vector are continuous feature points, the formula determining a plurality of energy amplitudes; In the case where the plurality of feature points in the feature vector are discrete feature points, a plurality of energy amplitudes are determined using the formula ψ[f(i)]=[f(i)]-f(i+1)f(i-1); Among them, the conversion relationship between ψ[f(t)] and ψ[f(i)] is: Wherein, ψ[f(t)] is the energy amplitude obtained by two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, is the first-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, f(t) is the feature point in the feature vector at the extraction sampling time t of the feature vector, is the second-order derivative of two adjacent feature points in the feature vector at the extraction sampling time t of the feature vector, ψ[f(i)] is the energy amplitude obtained by three adjacent feature points in the feature vector, f(i) is the i-th feature point in the feature vector, f(i+1) is the i+1-th feature point in the feature vector, f(i-1) is the i-1-th feature point in the feature vector, and T is the extraction sampling period of the feature vector.

8. The method according to claim 6, characterized in that The method further comprises: Acquire multiple historical transmission line images to be identified and located; For each historical transmission line image to be identified and located, wildfire point identification and location marking are performed to obtain multiple historical target transmission line images; Inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training in sequence to obtain the preset Teager energy operator model; The step of sequentially inputting each historical transmission line image to be identified and located and its corresponding historical target transmission line image into the initial Teager energy operator model for training to obtain the preset Teager energy operator model includes: The nth historical transmission line image to be identified and located and the nth historical target transmission line image are input into the initial Teager energy operator model for the nth training: the initial Teager energy operator model extracts the first feature vector of the nth historical transmission line image to be identified and located, and extracts the second feature vector of the nth historical target transmission line image; using the Teager energy algorithm, multiple first energy amplitudes are determined according to multiple first feature points in the first feature vector, and multiple second energy amplitudes are determined according to multiple second feature points in the second feature vector; the initial wildfire point characteristic energy amplitude is determined according to the multiple first energy amplitudes and the multiple second energy amplitudes; wherein n successively takes integers greater than 0 until n is equal to the total number of historical transmission line images to be identified and located, and the initial wildfire point characteristic energy amplitude obtained in each training is obtained; Determine the target fire point characteristic energy amplitude according to all the initial fire point characteristic energy amplitudes obtained through training; Using the initial Teager energy operator model having the characteristic energy amplitude of the target wildfire point as the predicted Teager energy operator model; The target wildfire point characteristic energy amplitude is used as the wildfire point characteristic energy amplitude obtained during the training of the preset Teager energy operator model.

9. The method according to claim 1, characterized in that: Inputting the transmission line image to be identified and located into a preset CNN model, and obtaining target intermediate layer features of the intermediate layer output of the preset CNN model, including: The image of the power transmission line to be identified and located is input into the preset CNN model: the middle layer of the preset CNN model performs convolution operation on the image of the power transmission line to be identified and located by means of a sliding window to obtain the target middle layer features.

10. The method according to claim 1, characterized in that The step of obtaining the image of the power transmission line to be identified and located comprises: Obtaining original transmission line images; The original power transmission line image is processed in sequence by median filtering, Gaussian filtering, bilateral filtering, histogram equalization, adaptive histogram equalization, contrast enhancement, threshold segmentation, edge segmentation and region segmentation to obtain the power transmission line image to be identified and located.

Images