Power transmission line bird recognition algorithm based on big data analysis

Data is collected through high-resolution cameras and infrared sensors, and bird recognition is combined with machine learning models, which solves the problems of identification accuracy and model generalization capabilities, realizes real-time monitoring and identification of birds, and improves the safety of transmission lines and processing efficiency.

CN118981685AActive Publication Date: 2024-11-19国网四川省电力公司阿坝供电公司 +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410997856.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-19
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

The prior art has the recognition accuracy in bird recognition due to noise, fuzzy and occlusion, the model generalization ability is limited, the processing efficiency is low, and the complex environment has a great impact on the performance of the recognition algorithm.

Method used

Using high-resolution cameras, infrared sensors and sound recognition equipment to collect data, preprocess and feature extraction through machine learning models, build a support vector machine algorithm model, conduct bird recognition and behavior monitoring, and identify bird species and behavior in real time to improve Data quality and model generalization capabilities.

Benefits of technology

Through data preprocessing and feature extraction, identification accuracy and model adaptability are improved, real-time monitoring and identification of complex environments are realized, manual monitoring needs are reduced, and the safety and efficiency of the power system are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118981685A_ABST
    Figure CN118981685A_ABST
Patent Text Reader

Abstract

The invention discloses a transmission line bird recognition algorithm based on big data analysis, and relates to the technical field of big data analysis. The method specifically comprises the steps of collecting data, collecting a high-resolution image of a power transmission line in real time, and obtaining the collected data; preprocessing the collected data to obtain preprocessed data so as to improve the accuracy of the data; according to the obtained preprocessing data, bird feature extraction is carried out; based on the extracted bird features, adopting a machine learning algorithm to construct a bird recognition model, and training the extracted bird features; and inputting real-time acquired data into the constructed bird identification model to identify bird types and behaviors so as to realize real-time monitoring of the power transmission line. Through the data preprocessing step, the machine learning model is used for training, the generalization ability of the model is enhanced, and the safety and reliability of the power transmission line are improved based on the algorithm design of big data analysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and in particular to a transmission line bird recognition algorithm based on big data analysis. Background Art

[0002] With the rapid development of big data technology, deep learning, the Internet of Things, and cloud computing, the transmission line bird recognition algorithm based on big data analysis provides data processing capabilities and algorithm support for bird recognition. Big data technology makes it possible to collect and process massive image and video data. Deep learning, especially the breakthrough of convolutional neural network (CNN) in image recognition, lays the foundation for accurate identification of bird species. The Internet of Things technology ensures the real-time collection and transmission of data, while cloud computing provides powerful computing resources to support the training and deployment of complex models.

[0003] However, despite the great potential that technological advances have brought to bird identification, there are still some problems and challenges. The collected images and videos may affect the recognition accuracy due to noise, blur or occlusion, making the data quality uncertain; due to the limited generalization ability of the model, the trained model may not perform well when facing new environments or new species of birds; processing large amounts of data requires efficient algorithms and sufficient computing resources, and the data cannot be updated in real time; in addition, the complex environment in which the transmission lines are located, such as weather changes and lighting conditions, will also affect the performance of the recognition algorithm.

[0004] To overcome these problems, it is necessary to continuously optimize algorithms, improve data quality, enhance the generalization ability of models, and explore more efficient and economical solutions. Summary of the invention

[0005] The purpose of the present invention is to provide a transmission line bird recognition algorithm based on big data analysis, which collects and analyzes bird activity data around the transmission line to achieve real-time monitoring and recognition of bird behavior, so as to prevent the impact of bird activities on the safety of the transmission line. The algorithm uses high-resolution cameras, infrared sensors and sound recognition equipment to collect data, processes and analyzes the data through machine learning models, identifies different bird species, behavior patterns and their potential threats to transmission lines, so as to solve the problems raised in the above background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a transmission line bird recognition algorithm based on big data analysis, specifically including:

[0007] Step S1, collecting data through a high-resolution camera, an infrared sensor and a sound recognition device, collecting high-resolution images of the transmission line in real time, and obtaining collected data;

[0008] Step S2, preprocessing the data collected in step S1 to obtain preprocessed data to improve the accuracy of the data;

[0009] Step S3, extracting bird features based on the preprocessed data obtained in step S2;

[0010] Step S4: Based on the bird features extracted in step S3, a machine learning algorithm is used to construct a bird recognition model, and the extracted bird features are trained, specifically:

[0011] S41, extracting n features from the bird features in step S3, represented by x=[x1, x2, ... x n ], where x i is the i-th eigenvalue;

[0012] S42, normalize the extracted data and divide the data into a training set, a validation set and a test set. The calculation formula is:

[0013]

[0014] Among them, u i and σ i are the mean and standard deviation of the i-th feature respectively;

[0015] S43, select the support vector machine algorithm to build a bird recognition model, the model formula is:

[0016]

[0017] The constraints are:

[0018] y i (w·x j +b)≥1―ξ j ,ξj ≥ 0;

[0019] Among them, w is the weight vector, b is the bias term, ξ j is the slack variable, C is the regularization parameter;

[0020] S44. The SVM model is trained using the training set through the gradient descent optimization algorithm to minimize the loss function. The calculation formula is:

[0021]

[0022] Where η is the learning rate and L(w,b) is the loss function;

[0023] S45, using the validation set and the test set to evaluate the trained SVM model to determine the reliability and stability of the SVM model;

[0024] S46, adjusting the hyperparameters and feature selection of the SVM model according to the result of the SVM model evaluation to ensure the accuracy of the SVM model evaluation;

[0025] Step S5: input the real-time collected data into the constructed bird recognition model to identify the bird species and behaviors, so as to realize real-time monitoring of the transmission line.

[0026] Furthermore, the model evaluation in step S45 includes the evaluation of accuracy, precision, recall, and F1 score, specifically:

[0027] S451, the accuracy calculation formula is:

[0028]

[0029] in, is the label predicted by the model, and ∏ is the indicator function;

[0030] S452, the accuracy calculation formula is:

[0031]

[0032] Among them, TP is a true positive example and FP is a false positive example;

[0033] S453, the recall rate calculation formula is:

[0034]

[0035] Among them, FN is a false negative example;

[0036] S454, F1 score calculation formula is:

[0037]

[0038] Furthermore, step S2 is further:

[0039] S21, using Gaussian filtering to perform denoising, reduce or eliminate the noise in the image to improve the image quality. The specific calculation formula is:

[0040] I filtered (x,y)=∑ (i,j)∈Ω I(i,j)·G(i―x,j―y);

[0041] Among them, G(i―x,j―y) is a Gaussian kernel function, which represents the influence weight of pixel (i,j) on the central pixel (x,y);

[0042] S22. Enhance the high-frequency components of the captured image by sharpening to make the edges and details of the image clearer. The calculation formula is:

[0043] I enhanced (x,y)=I(x,y)+λ(I(x,y)-I blurred (x,y));

[0044] Among them, I blurred( x,y) is the blurred image, and λ is the sharpening strength parameter;

[0045] S23, image segmentation is performed by edge detection, and the calculation formula is:

[0046]

[0047] Where E(x,y) is the edge strength, and are the gradients of the image in the x and y directions.

[0048] Furthermore, step S3 is further:

[0049] S31. Extract the color distribution in the image through the color histogram to distinguish different types of birds. The specific calculation formula is:

[0050] H(i)=∑ x,y δ(I(x,y)―i);

[0051] Where H(i) is the value of the i-th bin of the color histogram, (I(x,y) is the color value of the image at position (x,y), and δ is the Kronecker delta function;

[0052] S32, obtaining the shape of the bird through Hu moment calculation to identify the morphological characteristics of the bird;

[0053] S33, extracting the statistical characteristics of bird texture by gray-level co-occurrence matrix calculation to identify the texture characteristics of bird feathers;

[0054] S34. Use the Canny edge detection algorithm to extract edge information to identify the outline of the bird.

[0055] Furthermore, step S5 is further as follows:

[0056] Integrate the results of bird identification and behavior analysis into the detection system of power transmission lines to achieve real-time identification and monitoring of power transmission lines;

[0057] Detect the presence of birds and the degree of threat posed to power transmission lines by abnormal activities;

[0058] The detection results are fed back to the system in real time so that measures can be taken to protect the transmission lines.

[0059] The beneficial effects of the present invention are as follows: through the data preprocessing step, the collected images and videos are denoised, sharpened and segmented, which can effectively reduce the impact of noise, blur and occlusion on recognition accuracy, thereby improving data quality; using machine learning models for training can improve the adaptability of the model to different environments and bird species, enhance the generalization ability of the model, and enable it to perform better when facing new environments or new types of birds; algorithm design based on big data analysis can efficiently process large amounts of data, ensure real-time updating and processing of data, improve data processing efficiency, and through continuous optimization of algorithms and models, enable it to adapt to complex environments, improve the stability and accuracy of the recognition algorithm, and improve overall performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the overall process of the present invention.

[0061] Figure 2 It is a schematic flow chart of step S2 of the present invention.

[0062] Figure 3 It is a schematic flow chart of step S4 of the present invention. DETAILED DESCRIPTION

[0063] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features well known in the art are not described.

[0064] This solution provides a transmission line bird recognition algorithm based on big data analysis, referring to Figure 1 , the specific steps are:

[0065] Step S1, collecting data through a high-resolution camera, an infrared sensor and a sound recognition device, collecting high-resolution images of the transmission line in real time, and obtaining collected data;

[0066] Step S2, preprocessing the data collected in step S1 to obtain preprocessed data to improve the accuracy of the data;

[0067] Step S3, extracting bird features according to the preprocessed data obtained in step S2;

[0068] Step S4: Based on the bird features extracted in step S3, a machine learning algorithm is used to construct a bird recognition model, and the extracted bird features are trained;

[0069] Step S5: input the real-time collected data into the constructed bird recognition model to identify the bird species and behaviors, so as to realize real-time monitoring of the transmission line.

[0070] In this embodiment, through automated data collection and processing, the need for manual monitoring is reduced, and the efficiency and coverage of monitoring are greatly improved; the preprocessing data and feature extraction steps ensure the quality of data input into the machine learning model, thereby improving the accuracy of bird identification, and through real-time collection and analysis of data, timely identification of bird species and behaviors helps to quickly respond to potential security threats or environmental changes, help prevent contact between birds and transmission lines, and reduce the resulting power outages or equipment damage to the transmission lines, thereby improving the safety of the entire power system.

[0071] In a further embodiment, the collected data is preprocessed to obtain preprocessed data, mainly by denoising, sharpening and segmenting the data to improve the quality of the data, specifically:

[0072] S21. When using high-resolution cameras, infrared sensors, and sound recognition equipment to collect data, you will encounter various types of noise, such as Gaussian noise and salt and pepper noise. Therefore, you need to select a denoising method based on the noise type to remove random noise in the image while retaining the edge and detail information of the image as much as possible. In this solution, Gaussian filtering is used for denoising to reduce or eliminate noise in the image to improve image quality. The specific calculation formula is:

[0073] I filtered (x,y)=∑ (i,j)∈Ω I(i,j)·G(i―x,j―y);

[0074] Among them, G(i―x,j―y) is a Gaussian kernel function, which represents the influence weight of pixel (i,j) on the central pixel (x,y);

[0075] S22, sharpening processing can enhance the edges and details of the image, making the image look clearer and more vivid. This solution enhances the high-frequency components of the collected image by sharpening, making the edges and details of the image clearer. The calculation formula is:

[0076] I enhanced (x,y)=I(x,y)+λ(I(x,y)-I blurred (x,y));

[0077] Among them, I blurred (x, y) is the blurred image, λ is the sharpening strength parameter;

[0078] S23, the process of dividing the image into multiple regions or objects, which is crucial for subsequent feature extraction and recognition. In this solution, the edge detection segmentation method is used to perform image segmentation, identify and separate different objects or regions in the image. The calculation formula is:

[0079]

[0080] Where E(x,y) is the edge strength, and are the gradients of the image in the x and y directions.

[0081] According to the obtained pre-processed data, bird feature extraction is performed as follows:

[0082] S31. Extract the color distribution in the image through the color histogram to distinguish different types of birds. The specific calculation formula is:

[0083] H(i)=∑ x,y δ(I(x,y)―i);

[0084] Where H(i) is the value of the i-th bin of the color histogram, (I(x,y) is the color value of the image at position (x,y), and δ is the Kronecker delta function;

[0085] S32, obtaining the shape of the bird through Hu moment calculation to identify the morphological characteristics of the bird, specifically:

[0086] The Hu moment calculation formula is:

[0087]

[0088] Among them, η pq is the normalized central moment, μ pq is the central moment, μ 00 is the zeroth-order moment;

[0089] S33, extracting the statistical characteristics of bird texture through gray-level co-occurrence matrix calculation to identify the texture characteristics of bird feathers, the calculation formula is:

[0090] P(i,j│Δx,Δy)=∑ x,y δ(I(x,y)―i)δ(I(x+Δx,y+Δy)―j);

[0091] Among them, P(i,j│Δx,Δu) is the element of the gray-level co-occurrence matrix, which represents the frequency of gray values ​​i and j appearing simultaneously at the distance (Δx, Δy);

[0092] S34, using the Canny edge detection algorithm to extract edge information to identify the outline of the bird, specifically:

[0093] S341, calculate the weight matrix of Gaussian filtering, the formula is:

[0094]

[0095] Where σ is the standard deviation of the Gaussian distribution, x and y are the pixel positions;

[0096] S342, use the Sobel operator to calculate the gradient of the image. The Sobel operator is in the horizontal direction G x and vertical direction G y The convolution kernels are:

[0097]

[0098] The calculation formula for the gradient strength M(x,y) and direction θ(x,y) is:

[0099]

[0100]

[0101] S343, performing non-maximum suppression in the gradient direction to refine the edge;

[0102] S344, set two thresholds, high threshold T H and low threshold T L , the gradient strength is greater than T H Pixels with a gradient intensity of T are considered strong edges. L and T H The pixels between them are considered weak edges, strong edges are retained directly, and weak edges are retained only when they are connected to strong edges.

[0103] In this embodiment, denoising, sharpening, segmentation, etc. can significantly improve the quality of collected data, remove noise and interference, and enhance image features, thereby providing a clearer and more accurate data basis for subsequent analysis and identification; through feature extraction, key features of birds such as shape, texture, and color can be effectively extracted from the preprocessed data, and different bird species and behaviors can be distinguished, which helps to build a more accurate recognition model; through preprocessing and feature extraction, false alarms and missed alarms can be reduced, and the reliability of the monitoring system can be improved, which is of great significance for ensuring the safe operation of transmission lines and reducing unnecessary human resource investment.

[0104] In a further embodiment, a machine learning algorithm is used to construct a bird recognition model and train the extracted bird features, specifically:

[0105] S41. Extract n features from the bird features, expressed as x = [x1, x2, ... xn ], where x i is the i-th eigenvalue;

[0106] S42, normalize the extracted data and divide the data into a training set, a validation set and a test set. The calculation formula is:

[0107]

[0108] Among them, u i and σ i are the mean and standard deviation of the i-th feature respectively;

[0109] S43, select the support vector machine algorithm to build a bird recognition model, the model formula is:

[0110]

[0111] The constraints are:

[0112] y i (w·x j +b)≥1―ξ j ,ξ j ≥0;

[0113] Among them, w is the weight vector, b is the bias term, ξ j is the slack variable, C is the regularization parameter;

[0114] S44. The SVM model is trained using the training set through the gradient descent optimization algorithm to minimize the loss function. The calculation formula is:

[0115]

[0116] Where η is the learning rate and L(w,b) is the loss function;

[0117] S45, using the validation set and the test set to evaluate the trained SVM model to determine the reliability and stability of the SVM model;

[0118] The model evaluation in step S45 includes the evaluation of accuracy, precision, recall, and F1 score, specifically:

[0119] S451, the accuracy calculation formula is:

[0120]

[0121] in, is the label predicted by the model, and ∏ is the indicator function;

[0122] S452, the accuracy calculation formula is:

[0123]

[0124] Among them, TP is a true positive example and FP is a false positive example;

[0125] S453, the recall rate calculation formula is:

[0126]

[0127] Among them, FN is a false negative example;

[0128] S454, F1 score calculation formula is:

[0129]

[0130] S46. According to the result of the SVM model evaluation, the hyperparameters and feature selection of the SVM model are adjusted to ensure the accuracy of the SVM model evaluation, specifically:

[0131] S461, select linear function

[0132] Regularization parameter C, C controls the degree of penalty for misclassification. The larger C is, the greater the penalty for misclassification, which may lead to overfitting; the smaller C is, the smaller the penalty for misclassification, which may lead to underfitting kernel function parameters to determine hyperparameters;

[0133] S462, select Bayesian optimization to guide parameter search;

[0134] S463. Through correlation analysis, calculate the Pearson correlation coefficient between the eigenvalue and the target variable. The formula is:

[0135]

[0136] S464. Evaluate the accuracy, recall and F1 score of the model, perform model evaluation, and perform K-fold cross validation.

[0137] The real-time collected data is input into the constructed bird recognition model to identify the bird species and behaviors, so as to realize the real-time monitoring of the transmission line, specifically:

[0138] The shape, texture, color and behavioral characteristics of birds obtained through bird identification and behavior analysis are integrated into the detection system of the transmission line to perform behavior analysis. The formula is:

[0139] B pred =behaviorAnalysis(y pred ,y);

[0140] Among them, Bpred is the predicted bird behavior, which enables real-time identification and monitoring of transmission lines;

[0141] According to the bird species and behavior, the threat level to the transmission line is assessed using the formula:

[0142] T = ThratAssessment(y pred ,B pred );

[0143] Among them, T is the threat level score, which detects the threat level posed by the presence or abnormal activity of birds to the transmission line;

[0144] The test results are fed back to the system in real time, and measures are taken to protect the transmission lines, such as adjusting the operating parameters of the transmission lines, dispatching personnel for inspection, and installing protective devices.

[0145] In this embodiment, by constructing a machine learning model based on the extracted bird features, a highly accurate bird recognition model can be trained, which can effectively distinguish different bird species and behaviors according to the features, thereby improving the accuracy of recognition; the real-time collected data is input into the trained bird recognition model to realize real-time monitoring of the transmission line, and can timely detect and respond to the potential threats of bird activities to the transmission line; through the bird recognition model constructed by the machine learning algorithm, the system can automatically process and analyze a large amount of real-time data, reducing the need for manual intervention, improving the efficiency and intelligence level of monitoring, and the automated real-time monitoring system can reduce the demand for human resources and reduce operating costs. At the same time, by timely identifying and responding to bird activities, it can reduce transmission line failures caused by birds, further reduce maintenance costs, improve the safety and reliability of transmission lines, and reduce power outages and equipment damage caused by bird activities.

[0146] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes may be made to it in form and detail without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. Bird identification algorithm on power transmission lines based on big data analysis, It is characterized in that Specifically include: Step S1, collecting data through a high-resolution camera, an infrared sensor and a sound recognition device, collecting high-resolution images of the transmission line in real time, and obtaining collected data; Step S2, preprocessing the data collected in step S1 to obtain preprocessed data to improve the accuracy of the data; Step S3, extracting bird features based on the preprocessed data obtained in step S2; Step S4: Based on the bird features extracted in step S3, a machine learning algorithm is used to construct a bird recognition model, and the extracted bird features are trained, specifically: S41, extracting n features from the bird features in step S3, expressed as x=[x 1 ,x 2 ,…x n ], where x i is the i-th eigenvalue; S42, normalize the extracted data and divide the data into a training set, a validation set and a test set. The calculation formula is: Among them, u i and σ i are the mean and standard deviation of the i-th feature respectively; S43, select the support vector machine algorithm to build a bird recognition model, the model formula is: The constraints are: y i (w x j +b)≥1-ξ j ,x j ≥0; Among them, w is the weight vector, b is the bias term, ξ j is the slack variable, C is the regularization parameter; S44. The SVM model is trained using the training set through the gradient descent optimization algorithm to minimize the loss function. The calculation formula is: Where η is the learning rate and L(w,b) is the loss function; S45, using the validation set and the test set to evaluate the trained SVM model to determine the reliability and stability of the SVM model; S46, adjusting the hyperparameters and feature selection of the SVM model according to the result of the SVM model evaluation to ensure the accuracy of the SVM model evaluation; Step S5: input the real-time collected data into the constructed bird recognition model to identify the bird species and behaviors, so as to realize real-time monitoring of the transmission line.

2. According to the transmission line bird identification algorithm based on big data analysis according to claim 1, It is characterized in that The model evaluation in step S45 includes the evaluation of accuracy, precision, recall, and F1 score, specifically: S451, the accuracy calculation formula is: in, is the label predicted by the model, and ∏ is the indicator function; S452, the accuracy calculation formula is: Among them, TP is a true positive example and FP is a false positive example; S453. The recall rate calculation formula is: Among them, FN is a false negative example; S454, F1 score calculation formula is:

3. According to the transmission line bird identification algorithm based on big data analysis according to claim 1, It is characterized in that Step S2 is further as follows: S21, using Gaussian filtering to perform denoising, reduce or eliminate the noise in the image to improve the image quality. The specific calculation formula is: I filtered (x,y)=∑ (ij)∈Ω I(i,j)·G(i-x,j-y); Among them, G(ix, jy) is a Gaussian kernel function, which represents the influence weight of pixel (i, j) on the central pixel (x, y); S22. Enhance the high-frequency components of the captured image by sharpening to make the edges and details of the image clearer. The calculation formula is: I enhanced (x,y)=I(x,y)+λ(I(x,y)-I blurred (x,y)); Among them, I blurred (x, y) is the blurred image, λ is the sharpening strength parameter; S23, image segmentation is performed by edge detection, and the calculation formula is: Where E(x,y) is the edge strength, and are the gradients of the image in the x and y directions.

4. The transmission line bird identification algorithm based on big data analysis according to claim 1, It is characterized in that Step S3 is further as follows: S31. Extract the color distribution in the image through the color histogram to distinguish different types of birds. The specific calculation formula is: H(i)=∑ x,y δ(I(x,y)-i); Where H(i) is the value of the i-th bin of the color histogram, (I(x,y) is the color value of the image at position (x,y), and δ is the Kronecker delta function; S32, obtaining the shape of the bird through Hu moment calculation to identify the morphological characteristics of the bird; S33, extracting the statistical characteristics of bird texture by gray-level co-occurrence matrix calculation to identify the texture characteristics of bird feathers; S34. Use the Canny edge detection algorithm to extract edge information to identify the outline of the bird.

5. The transmission line bird identification algorithm based on big data analysis according to claim 1, It is characterized in that Step S5 is further as follows: Integrate the results of bird identification and behavior analysis into the detection system of power transmission lines to achieve real-time identification and monitoring of power transmission lines; Detect the presence of birds and the degree of threat posed to power transmission lines by abnormal activities; The detection results are fed back to the system in real time so that measures can be taken to protect the transmission lines.

Citation Information

Patent Citations

  • Power transmission line bird nest detection method based on improved SSD algorithm

    CN111241905A

  • Method for classifying and identifying bird species harmful to power grid based on deep convolution features

    CN114067368A

  • Bird identification method based on fine-grained feature fusion

    CN115272763A

  • Bird monitoring system with combination of dual-light camera carried by unmanned aerial vehicle and deep learning

    CN118196660A

  • Curable composition, cured layer using the same and display device including the cured layer

    KR102927846B1