CN-121998437-A - Big data-based dynamic adjustment system and method for bird prevention measure parameters of power transformation facilities

CN121998437ACN 121998437 ACN121998437 ACN 121998437ACN-121998437-A

Abstract

The invention discloses a system and a method for dynamically adjusting bird prevention measure parameters of a power transformation facility based on big data, and relates to the technical field of bird prevention big data processing of the power transformation facility, wherein the system comprises a data acquisition module, and a plurality of types of equipment are integrated to acquire bird, environment and bird prevention device data; the big data processing module is used for preprocessing data and constructing a standardized data set, the bird species recognition submodule recognizes bird species and marks the infringement level, the bird condition analysis module outputs the risk level, the environment adaptation submodule corrects the risk level, the parameter adjustment decision module solves the optimal parameter, the control module is used for driving equipment adjustment, and the feedback monitoring module is used for collecting data to form a closed loop. The bird damage risk prediction method improves data quality and bird species recognition accuracy, accurately predicts bird damage risk, optimizes bird-preventing parameter adaptation scenes, reduces energy consumption and operation and maintenance cost of the device, and meets the intelligent operation and maintenance requirements of the power system.

Inventors

RAO BINBIN
HU JING
LU YUXIN
HUANG KUN
LIU YINKANG

Assignees

国网江西省电力有限公司电力科学研究院

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. Big data-based power transformation facility bird prevention measure parameter dynamic adjustment system is characterized by comprising: The data acquisition module is used for integrating video acquisition equipment and multiple sensors and synchronously acquiring the picture of the power transformation facility area, the surrounding environment parameters and the running state parameters of the bird prevention device; The big data processing module is used for preprocessing the data acquired by the data acquisition module and constructing a standardized substation facility bird condition data set based on the preprocessed data; the bird species identification submodule is used for presetting a sample library covering various invasive birds and non-invasive bird samples, matching birds in a power transformation facility area picture shot in real time with the bird samples in the sample library, determining the types of the birds and marking the invasive grades; The bird situation analysis module is used for constructing a deep learning model, carrying out association of space time, bird types and environmental dimensions on the real-time bird situation recognition results output by the standardized power transformation facility bird situation data set and the bird type recognition sub-module to form a complete power transformation facility bird situation data set; the environment adaptation sub-module is used for constructing a correlation model based on real-time environment parameters and outputting bird activity liveness correction coefficients; The comprehensive damage risk correction module corrects the initial risk level of bird damage of the power transformation facility area at different time intervals in the future based on a preset multidimensional weight coefficient and a bird activity correction coefficient; the parameter adjustment decision module is used for constructing a multi-objective optimization function by taking the risk level corrected by the attack risk comprehensive correction module as an optimization target, wherein the risk level is reduced to a preset threshold value, the energy consumption of the bird prevention device is reduced, and the operation interference of a power transformation facility is minimum, and solving the optimal bird prevention measure parameter combination by adopting an intelligent algorithm; And the execution control module comprises a driving circuit and an execution mechanism, and adjusts the bird preventing device based on the solved optimal bird preventing measure parameter combination.
2. The system for dynamically adjusting bird prevention measure parameters of power transformation facilities based on big data as set forth in claim 1, wherein the standardized power transformation facility bird condition data set comprises: Historical bird condition data, which is the type of historical bird and the degree of invasion; Corresponding to the historical environmental parameters; Historical bird space-time activity data, namely time dimension, observation period and season, and space dimension, namely activity area coordinates, high-risk point stay records and nesting positions.
3. The big data-based dynamic adjustment system for the bird prevention measure parameters of the power transformation facility, as set forth in claim 2, is characterized in that the bird species identification submodule specifically includes: Receiving a real-time monitoring picture of the power transformation facility area, which is acquired by a data acquisition module, carrying out component decomposition on each pixel point in the real-time monitoring picture of the power transformation facility area, and separating out a reflection component reflecting the inherent outline and texture of birds and an illumination component reflecting the ambient illumination interference; Performing self-adaptive brightness equalization processing on the separated illumination components, performing brightness improvement on a low illumination area in a real-time monitoring picture of a power transformation facility area, and performing brightness inhibition on an overexposure high illumination area; based on the illumination component and the reflection component after the self-adaptive brightness equalization processing, performing self-adaptive illumination enhancement calculation, and outputting a standardized picture for eliminating illumination interference; Constructing a double-branch parallel multi-scale feature fusion convolutional neural network, wherein the multi-scale feature fusion convolutional neural network is divided into three core units, namely a shared shallow feature layer, a main classification branch and a small target bird detection branch, after a standardized picture is input into the network, the shared shallow feature layer is firstly entered, the shared shallow feature layer extracts general basic features of all birds in the standardized picture, including bird edges, outlines and basic textures, through the convolutional layer, the output general basic features are synchronously transmitted to the main classification branch and the small target bird detection branch, the main classification branch further performs semantic extraction on the general basic features through the deep convolutional layer to obtain deep semantic features of large and medium-sized invasive birds, the small target bird detection branch retains the general basic features output by the shared shallow feature layer, and simultaneously sets a special anchor point frame which is adaptive to the small invasive birds to extract high-resolution detail features of the small invasive birds; constructing a bird sample library covering the whole scene of the power transformation facility, wherein the sample library comprises a plurality of invasive birds and noninvasive bird samples, and covers the scene of backlight, night infrared, overcast and rainy blur and strong wind shake which occur at different illumination, shooting angles, shooting distances and high frequency of the transformer substation; completing multidimensional basic labeling of each type of bird sample in the bird sample library, wherein the multidimensional basic labeling comprises body type parameters, typical behavior characteristics and different perch point risk coefficients of birds; matching the standardized bird species matching characteristics with the bird sample characteristics in the bird sample library in a cosine similarity mode, and taking the bird type corresponding to the bird sample with the highest similarity as a final recognition result; calculating the scores of the body type parameters, typical behavior characteristics and risk coefficients of different habitat points of the identified bird species after the bird species identification is completed, and weighting and calculating the final basic score of the damage of the bird species based on the scores of the body type parameters, typical behavior characteristics and risk coefficients of different habitat points of the identified bird species Based on the final bird species invasion basal score Marking the corresponding infringement level.
4. The big data-based power transformation facility bird prevention measure parameter dynamic adjustment system of claim 3, wherein in the bird situation analysis module, the model training data set constructed based on the standardized power transformation facility bird situation data set comprises historical bird situation data, corresponding historical environment parameters, historical bird space-time activity data and actual infringement result labels; the process of training the deep learning model is that two types of feature extraction are synchronously executed on the data in the model training data set: Extracting time dimension characteristics in a model training data set by a sliding window method; extracting spatial distribution characteristics in a training data set of a model through spatial convolution; fusing the time dimension characteristics and the space distribution characteristics to obtain space-time characteristics; after the space-time characteristics are fused with the types, the invasiveness level and the environmental parameters of the historical birds, the space-time characteristics are input into a deep learning model for training, and an Adam optimizer is adopted for training; The risk prediction data set extracted based on the complete power transformation facility bird condition data set comprises real-time bird types and attack levels output by the bird species identification submodule, real-time environment parameters preprocessed by the big data processing module, and historical data extracted from the standardized power transformation facility bird condition data set and corresponding to current bird condition data, environment parameters and bird space-time activity data; The method comprises the steps of respectively extracting real-time dimension characteristics from data in the risk prediction data set through a sliding window method, extracting real-time space distribution characteristics through space convolution, fusing the real-time dimension characteristics with the real-time space distribution characteristics to obtain real-time space-time characteristics, inputting the real-time space-time characteristics, the real-time bird types, the invasion levels and the environmental parameters into a trained deep learning model after fusing the real-time space-time characteristics with the real-time bird types, the invasion levels and the environmental parameters, and outputting the bird invasion initial risk levels of the future transformer facility areas in different time periods.
5. The system for dynamically adjusting bird-preventing measure parameters of power transformation facilities based on big data as set forth in claim 4, wherein the big data processing module further comprises a data standardization sub-module for performing normalization processing on the established standardized power transformation facility bird condition data set by adopting a standardization method, mapping all data to a 0-1 interval, simultaneously establishing a data quality evaluation index, scoring the normalized standardized power transformation facility bird condition data set by three dimensions of data integrity, accuracy and consistency, wherein the score satisfies a preset score, judging that the standardized power transformation facility bird condition data set is qualified, and the standardized power transformation facility bird condition data set is used for a bird condition analysis module, wherein the score does not satisfy the preset score, judging that the standardized power transformation facility bird condition data set is unqualified, and returning to a data preprocessing link for reprocessing.
6. The system for dynamically adjusting the bird prevention measure parameters of the power transformation facility based on big data as set forth in claim 5, wherein the comprehensive risk of attack correction module comprises: Setting a time adaptation score based on bird activity periods and non-activity periods Setting a bird group density correction score based on the number of invasive birds in the region Calculating a bird history invasion feedback score based on the bird history invasion frequency coefficient and the latest invasion interval coefficient ; Constructing an index system of a correction dimension, wherein the correction dimension comprises a bird species invasion dimension, an environment adaptation dimension, a time dimension, a bird group density dimension and a historical invasion feedback dimension; The method comprises the steps of calculating subjective weight of each correction dimension by adopting an analytic hierarchy process, calculating objective weight of each correction dimension by adopting an entropy weight process, fusing the subjective weight and the objective weight of each correction dimension based on a minimum identification information principle to obtain a dynamic self-adaptive weight coefficient of each correction dimension, and constructing a comprehensive risk level correction formula based on the dynamic self-adaptive weight coefficient of each correction dimension: ; in the formula, Representing a corrected bird infestation risk level; A dynamic adaptive weight coefficient representing a dimension of invasion of the bird species; a dynamic adaptive weight coefficient representing an environmental adaptation dimension; A dynamic adaptive weight coefficient representing a time dimension; a dynamic adaptive weight coefficient representing a bird group density dimension; a dynamic adaptive weight coefficient representing a historical violation feedback dimension; and the environmental adaptation correction score is represented and is obtained by converting the bird activity liveness correction coefficient output by the environmental adaptation sub-module.
7. The system for dynamically adjusting the bird-preventing measure parameters of the power transformation facility based on big data as set forth in claim 6, wherein the optimal bird-preventing measure parameter combination obtained by solving in the parameter adjustment decision module comprises frequency and volume of the sound wave bird repeller, scanning angle and frequency of the laser bird repeller, and expanding distance and height of bird thorns.
8. A big data-based dynamic adjustment method for anti-bird measure parameters of a power transformation facility, which is realized based on the big data-based dynamic adjustment system for anti-bird measure parameters of the power transformation facility according to any one of claims 1 to 7, and is characterized by comprising the following steps: Starting various devices of the data acquisition module to acquire a power transformation facility area picture, surrounding environment parameters and running state parameters of the bird prevention device; preprocessing the data acquired by the data acquisition module through the big data processing module, and constructing a standardized substation facility bird condition data set based on the preprocessed data; matching birds in a power transformation facility area picture shot in real time with bird samples in a preset sample library through a bird species identification submodule, determining the types of the birds and marking the infringement level; the bird situation analysis module is used for generating initial risk levels of bird invasion of the power transformation facility areas at different time periods in the future based on the output of the bird species identification sub-module and the big data processing module; through the environment adaptation sub-module, an association model is constructed based on real-time environment parameters, and a bird activity liveness correction coefficient is output; The parameter adjustment decision-making module is used for constructing a multi-objective optimization function by taking the risk level corrected by the attack risk comprehensive correction module as an optimization target, wherein the risk level is reduced to a preset threshold value, the energy consumption of the bird prevention device is reduced, and the operation interference of a power transformation facility is minimum, and an intelligent algorithm is adopted to solve the optimal bird prevention measure parameter combination; And adjusting the bird preventing device based on the solved optimal bird preventing measure parameter combination by executing the control module.
9. An electronic device, comprising a processor, a memory and a bus, wherein the processor and the memory are connected through the bus, the memory is used for storing a set of program codes, and the processor is used for calling the program codes stored in the memory and executing a big data-based dynamic adjustment method for the bird prevention measure parameters of the power transformation facilities according to claim 8.
10. A non-volatile computer storage medium having stored thereon computer executable instructions for performing a big data based dynamic adjustment method of anti-bird measure parameters of a power transformation facility as claimed in claim 8.

Description

Big data-based dynamic adjustment system and method for bird prevention measure parameters of power transformation facilities Technical Field The invention relates to the technical field of big data processing of anti-bird measures of a power transformation facility, in particular to a system and a method for dynamically adjusting parameters of anti-bird measures of the power transformation facility based on big data. Background The power transformation facilities are used as core nodes for power transmission and distribution of the power system, the operation safety of the power transformation facilities directly determines the stability of power supply, and bird activities are important causes for causing faults of the power transformation facilities. Foreign matters such as branches, feathers and the like falling in the nesting and perching processes of birds easily cause short circuits of power transformation equipment, the direct contact of birds with electrified parts also causes problems such as insulation flashover, grounding faults and the like, the local power failure is caused by light faults, the regional power grid fluctuation is caused by heavy faults, and obvious economic loss is caused. The current transformer facility bird prevention measures mainly comprise traditional fixed schemes, including bird thorn prevention installation with preset intervals, operation of a sound wave bird repellent device with fixed frequency, scanning of a laser bird repellent device with fixed angle and the like, and parameter setting of the measures completely depends on field experience of operation and maintenance personnel, so that dynamic adaptation capability to bird activity rules is lacked, and complex and changeable bird situation scenes are difficult to deal with. The sensitivity difference of different birds to bird-expelling means is obvious, the frequency response of large-scale infringement birds such as magpie, crow and the like to conventional sound waves is weak, the sound wave bird-expelling device with fixed parameters is difficult to effectively expel the large-scale infringement birds such as sparrow, pigeon and the like, adaptability is easy to be generated in the fixed-parameter bird-expelling environment, the bird-expelling effect is greatly attenuated after long-term operation, and continuous protection cannot be formed. The existing partial bird prevention system tries to introduce sensors to collect bird condition data, but the data processing link has obvious short plates, so that simple screening of single-dimension data can be realized, multi-dimension data such as historical bird conditions, environmental factors and the like are not integrated for deep analysis, a correlation model of bird activities and environmental conditions cannot be built, the prejudging precision of bird invasion risks is low, the bird prevention strategy is difficult to adjust in advance, and the bird prevention system is still in a passive state of 'post-fault coping'. The traditional image recognition algorithm adopted in the bird species recognition link is limited, the coverage scene of a sample library is limited, the influence of different illumination, shooting angles and distances on the recognition result is not fully considered, invasive and non-invasive birds cannot be accurately distinguished, and the problems of excessive bird repelling or bird missing judgment of invasive birds are easy to occur. Excessive bird repelling can increase the energy consumption and operation and maintenance cost of equipment, such as bird's nest, woodpecker and other non-invasive birds, not only waste energy, but also possibly destroy ecological balance, and if the bird is missed, the risk of facility failure can be directly increased, such as the phenomenon that the bird is not identified as severe invasive bird like magpie, and the like, so that the bird is nest-built to cause short circuit of the equipment. In addition, the existing bird prevention system lacks a complete closed-loop regulation mechanism, after bird prevention measures parameters are adjusted, key indexes such as bird activity change, facility running state and the like cannot be quantified in real time only by evaluating the effect through a manual inspection mode, and the evaluation result is not fed back to a parameter decision-making link, so that the parameter adjustment is not subjected to iterative optimization basis, and remains on a passive corresponding level all the time, and the requirements of intelligent and lean operation and maintenance of the electric power system on the accuracy, the dynamics and the economy of the bird prevention measures cannot be met. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a system and a method for dynamically adjusting the parameters of anti-bird measures of a power transformation facility based on big data, and aims to solve the problems in the background ar