Bird control method and system for power systems

Abstract

The application provides a bird repelling control method and system for a power system, relates to the field of adaptive bird repelling control, and comprises the following steps: setting vibration sensing devices at multiple key positions of an overhead line; determining a first bird repelling probability according to sensing data collected by the multiple vibration sensing devices at multiple continuous time points; adjusting the first bird repelling probability according to key meteorological information of the overhead line to generate a second bird repelling probability; judging whether to perform bird identification according to the second bird repelling probability; collecting an overhead line image when it is determined to perform bird identification; identifying birds according to the overhead line image; determining ultrasonic bird repelling parameters according to the identified birds; determining bird repelling pose parameters according to the overhead line image; and controlling ultrasonic bird repelling equipment to repel birds according to the ultrasonic bird repelling parameters and the bird repelling pose parameters, so that the intelligent level and the bird repelling effect of the power system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bird repelling adaptive control, and particularly relates to a bird repelling control method and system for a power system. BACKGROUND

[0002] With the development of social economy, the demand for electricity in industrial and agricultural production and residential life gradually increases, and the overhead lines used in the power transmission and distribution network also increase. Due to the improvement of the ecological environment, the number of bird species increases rapidly, and birds nesting and inhabiting on the overhead lines can easily cause line short circuit, tripping and other accidents.

[0003] In the prior art, a device for expelling birds by high-intensity sound waves is generally used. Under the control of the existing control system, the bird repelling device periodically emits sound waves and scans according to a fixed path. The control system used for bird repelling generally includes a central processor and a memory connected to the central processor. Meanwhile, the bird repelling device is also controlled by the central processor. In the memory, a fixed path and a fixed time are set in advance, so that the central processor controls the bird repelling device to periodically emit sound waves to the upper space of the specified area according to the path and time, so as to expel the birds in the area.

[0004] Although such a bird repelling device and its control system can achieve the purpose of expelling birds, the bird repelling device controlled by the bird repelling control system blindly emits sound waves during the entire bird repelling process. The bird repelling device is always in an open state and covers the entire upper space of the specified area with sound waves without distinction. This not only causes great waste of energy, but also has poor bird repelling effect due to the blindness.

[0005] Therefore, it is necessary to provide a bird repelling control method and system for a power system to improve the intelligent level of bird repelling of the power system, thereby improving the bird repelling effect. SUMMARY

[0006] The present application provides a bird repelling control method for a power system, comprising: setting vibration sensing devices at a plurality of key positions of an overhead line; determining a first bird repelling probability according to sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points; adjusting the first bird repelling probability according to key meteorological information of the overhead line to generate a second bird repelling probability; determining whether to perform bird identification according to the second bird repelling probability; collecting an overhead line image when it is determined to perform bird identification; identifying birds according to the overhead line image; determining ultrasonic bird repelling parameters according to the identified birds; determining bird repelling pose parameters according to the overhead line image; and controlling an ultrasonic bird repelling device to perform bird repelling according to the ultrasonic bird repelling parameters and the bird repelling pose parameters.

[0007] Further, the plurality of key positions of the overhead line are determined, including: determining a plurality of to-be-screened positions on the simulated overhead line at uniform intervals according to a preset interval, and setting a vibration sensing device at each to-be-screened position; performing a plurality of bird dropping experiments on the simulated overhead line, wherein the distance between the bird dropping positions of any two bird dropping experiments is greater than a preset distance threshold; for each to-be-screened position, determining an initial sensing feature vector of the to-be-screened position according to the sensing data collected by the vibration sensing device at the to-be-screened position in each bird dropping experiment; determining the weight of each to-be-screened position according to the historical bird dropping data of the overhead line; for each to-be-screened position, generating a weighted sensing feature vector according to the weight and the initial sensing feature vector of each to-be-screened position; and determining the plurality of key positions from the plurality of to-be-screened positions according to the weighted sensing feature vector of each to-be-screened position.

[0008] Further, the plurality of key positions of the overhead line are determined, including: determining a plurality of to-be-screened positions on the simulated overhead line at uniform intervals according to a preset interval, and setting a vibration sensing device at each to-be-screened position; performing a plurality of bird dropping experiments on the simulated overhead line, wherein the distance between the bird dropping positions of any two bird dropping experiments is greater than a preset distance threshold; for each to-be-screened position, determining an initial sensing feature vector of the to-be-screened position according to the sensing data collected by the vibration sensing device at the to-be-screened position in each bird dropping experiment; determining the weight of each to-be-screened position according to the historical bird dropping data of the overhead line; for each to-be-screened position, generating a weighted sensing feature vector according to the weight and the initial sensing feature vector of each to-be-screened position; and determining the plurality of key positions from the plurality of to-be-screened positions according to the weighted sensing feature vector of each to-be-screened position.

[0009] Further, the first bird repelling probability is determined according to the sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points, including: calculating the correlation coefficient of any two key positions according to the weighted sensing feature vectors of the any two key positions; screening a plurality of key vibration sensing devices from the plurality of vibration sensing devices according to the sensing data collected by the plurality of vibration sensing devices at the plurality of continuous time points and the correlation coefficient of any two key positions; for each key vibration sensing device, performing variational mode decomposition on the sensing data collected by the key vibration sensing device at the plurality of continuous time points to extract a key vibration feature of the key position where the key vibration sensing device is located; and determining the first bird repelling probability according to the key vibration feature of the key position where each key vibration sensing device is located and the correlation coefficient of any two key positions through a probability prediction model.

[0010] Further, the key meteorological information of the overhead line is obtained, including: determining a plurality of meteorological factors; obtaining meteorological data and bird repelling times of a plurality of historical time periods; for each meteorological factor, calculating the correlation coefficient of the meteorological factor and the bird repelling times according to the meteorological data and the bird repelling times of the plurality of historical time periods; screening a plurality of key meteorological factors from the plurality of meteorological factors according to the correlation coefficient of each meteorological factor and the bird repelling times; and obtaining the key meteorological information of the overhead line according to the plurality of key meteorological factors.

[0011] Further, the first bird repelling probability is adjusted according to the key meteorological information of the overhead line to generate a second bird repelling probability, including: predicting an adjustment coefficient according to the key meteorological information of the overhead line through a coefficient prediction model; and adjusting the first bird repelling probability according to the adjustment coefficient to generate the second bird repelling probability.

[0012] Further, the bird species is identified according to the overhead line image, including: obtaining images of a plurality of sample birds, wherein the plurality of sample birds are different in species and / or growth stage; extracting biological features of each sample bird; dividing the plurality of sample birds into a plurality of sample bird clusters according to the biological features of each sample bird through a multi-level clustering algorithm; for each sample bird cluster, determining a comprehensive biological feature corresponding to the sample bird cluster according to the biological features of each sample bird included in the sample bird cluster; segmenting a region of interest image from the overhead line image; extracting biological features of the bird to be repelled from the region of interest image; and identifying the bird species based on the biological features of the bird to be repelled and the comprehensive biological feature corresponding to each sample bird cluster.

[0013] Further, the ultrasonic bird repelling parameters are determined according to the identified bird species, including: determining optimal ultrasonic bird repelling parameters for each sample bird; and determining the ultrasonic bird repelling parameters according to the identified bird species and the optimal ultrasonic bird repelling parameters for each sample bird.

[0014] Further, the bird repelling pose parameters are determined according to the overhead line image, including: determining a spatial position of the bird to be repelled according to the overhead line image; and determining the bird repelling pose parameters according to the spatial position of the bird to be repelled and the ultrasonic bird repelling parameters, wherein the bird repelling pose parameters at least include an ultrasonic emission spatial position, an ultrasonic emission horizontal angle and an ultrasonic emission pitch angle.

[0015] The application provides a bird repelling control system for a power system for executing the bird repelling control method for a power system described above, including: a bird species sensing module for setting vibration sensing devices at a plurality of key positions of an overhead line, determining a first bird repelling probability according to sensing data collected by a plurality of vibration sensing devices at a plurality of consecutive time points, adjusting the first bird repelling probability according to key meteorological information of the overhead line to generate a second bird repelling probability, and determining whether to perform bird species identification according to the second bird repelling probability; a bird species identification module for determining to perform bird species identification, collecting an overhead line image, and identifying a bird species according to the overhead line image; and a bird repelling control module for determining ultrasonic bird repelling parameters according to the identified bird species, determining bird repelling pose parameters according to the overhead line image, and controlling an ultrasonic bird repelling device to repel birds according to the ultrasonic bird repelling parameters and the bird repelling pose parameters.

[0016] Compared with the prior art, the bird repelling control method and system for a power system provided by the application at least have the following beneficial effects:

[0017] Vibration sensing devices are installed at key locations on the overhead line, and the first bird repelling probability is determined based on the sensing data collected at multiple consecutive time points. Birds' activities on the overhead line, such as staying, flying, nesting, etc., will generate vibrations, and the vibration sensing devices can capture these vibration signals. By analyzing the vibration data, the likelihood of birds appearing in the area can be preliminarily judged, providing a basis for subsequent bird repelling decisions and avoiding blind bird repelling.

[0018] The first bird repelling probability is adjusted according to the key meteorological information of the overhead line to generate the second bird repelling probability. Meteorological conditions, such as temperature, humidity, wind speed, weather conditions, etc., will affect the activity habits of birds. For example, in sunny and warm weather, birds may be more active and have a wider range of activities; in adverse weather, birds may seek shelter from the wind and rain. Adjusting the bird repelling probability in combination with meteorological information can more accurately reflect the likelihood of birds appearing, making the bird repelling decision more scientific and reasonable.

[0019] Bird identification is determined based on the second bird repelling probability, and images of the overhead line are collected only when identification is needed. This approach avoids unnecessary image collection and identification operations when the probability of birds appearing is low, saving system resources and computing costs and improving system efficiency.

[0020] Bird identification is performed after collecting images of the overhead line. Image recognition technology can accurately determine whether birds are present in the image and the species of the birds. Different birds may cause different damage to the overhead line, and accurate identification of birds helps to take more targeted bird repelling measures.

[0021] Ultrasonic bird repelling parameters are determined based on the identified birds. Different species of birds have different sensitivities to ultrasonic frequency, intensity, and other parameters. For example, some small birds may be more sensitive to high-frequency ultrasonic waves, while large birds may require lower-frequency, higher-intensity ultrasonic waves to achieve the repelling effect. By setting appropriate ultrasonic parameters for different birds, the effectiveness of bird repelling can be improved.

[0022] Bird repelling pose parameters, i.e., the emission direction and angle of the ultrasonic bird repelling device, are determined based on the images of the overhead line. This allows the ultrasonic waves to more accurately cover the location of the birds, improving the accuracy of bird repelling and avoiding waste and ineffective diffusion of ultrasonic waves.

[0023] Through the above series of steps, birds that may cause harm to the overhead line can be found and driven away in time. The activities of birds on the overhead line, such as nesting and pecking insulators, may cause line faults, such as short circuit and tripping, which seriously affect the safe and stable operation of the power system. The bird driving control method can effectively reduce the harm of birds to the overhead line and ensure the safe power supply of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0024] The present specification will be further illustrated in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, the same numbers represent the same structures, wherein:

[0025] Figure 1 is a flowchart of a bird driving control method for a power system according to some embodiments of the present specification;

[0026] Figure 2 is a flowchart of determining a plurality of key positions of an overhead line according to some embodiments of the present specification;

[0027] Figure 3 is a flowchart of identifying birds according to some embodiments of the present specification;

[0028] Figure 4 is a module diagram of a bird driving control system for a power system according to some embodiments of the present specification. DETAILED DESCRIPTION

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings needed to be used in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some examples or embodiments of the present specification, and for those skilled in the art, the present specification can also be applied to other similar scenarios without creative labor. Unless it is obvious from the language environment or otherwise stated, the same reference numbers in the drawings represent the same structure or operation.

[0030] Figure 1 is a flowchart of a bird driving control method for a power system according to some embodiments of the present specification, as shown in Figure 1 The bird driving control method for a power system can include the following steps.

[0031] Step 110, at a plurality of key positions of the overhead line, a vibration sensing device is arranged.

[0032] Figure 2 is a flowchart of determining a plurality of key positions of an overhead line according to some embodiments of the present specification, as shown in Figure 2As shown, in some embodiments, a plurality of key positions of the overhead line are determined, including:

[0033] At preset intervals, a plurality of to-be-screened positions are determined on the simulated overhead line, and a vibration sensing device is arranged at each to-be-screened position;

[0034] A plurality of bird dropping experiments are performed on the simulated overhead line, wherein the distance between the bird dropping positions of any two bird dropping experiments is greater than a preset distance threshold;

[0035] For each to-be-screened position, an initial sensing feature vector of the to-be-screened position is determined according to the sensing data collected by the vibration sensing device at the to-be-screened position in each bird dropping experiment;

[0036] According to the historical bird dropping data of the overhead line, the weight of each to-be-screened position is determined;

[0037] For each to-be-screened position, a weighted sensing feature vector is generated according to the weight and the initial sensing feature vector of each to-be-screened position;

[0038] According to the weighted sensing feature vector of each to-be-screened position, a plurality of key positions are determined from the plurality of to-be-screened positions.

[0039] Specifically, the structure of the simulated overhead line is consistent with the structure of the actual overhead line. Specifically, the basic interval is determined according to the line voltage level. For example, 220kV lines are arranged at an interval of 50 meters, and 500kV lines are arranged at an interval of 30 meters, to ensure intensive monitoring of high-voltage lines. Ensure that the distance between any two experiments is > 5 meters (preset threshold), to avoid vibration signal superposition interference. When the distance is < 3 meters, the cross-sensitivity of the vibration sensing device can reach more than 40%. Use a mechanical bird model equipped with an accelerometer to control the landing accuracy ± 0.2 meters, simulate the mass (0.3-1.5kg) and falling speed (4-7m / s) of real birds. Collect 10 seconds of vibration data before and after each experiment, covering the pre-vibration stage before the bird touches the wire and the continuous vibration stage after the bird touches the wire.

[0040] A vibration amplitude threshold can be set to determine whether the vibration sensing device at the to-be-screened position senses the bird in each bird dropping experiment, for example, if the maximum vibration amplitude extracted from the sensing data collected by the vibration sensing device at the to-be-screened position in the bird dropping experiment is greater than the vibration amplitude threshold, it is determined that the vibration sensing device at the to-be-screened position senses the bird in the bird dropping experiment.

[0041] The initial sensing feature vector of the position to be screened can include a sensing mark corresponding to each bird dropping experiment. If the vibration sensing device at the position to be screened senses a bird dropping in a bird dropping experiment, the sensing mark corresponding to the bird dropping experiment is 1, otherwise it is 0. For example, the initial sensing feature vector of the position to be screened can consist of 5 elements, one element corresponding to one bird dropping experiment. When the vibration sensing device at the position to be screened senses a bird dropping in the first three bird dropping experiments, the initial sensing feature vector of the position to be screened can be (1, 1, 1, 0, 0).

[0042] Extract the bird dropping event records of the past years (for example, 3 years) from the line operation database to construct a spatio-temporal distribution heat map. The weight of each position to be screened can be calculated by kernel density, which can reflect the frequency of bird dropping at the position to be screened.

[0043] A weight vector can be generated according to the position to be screened in each bird dropping experiment. For example, the weight of the position to be screened 1 is 0.3, the weight of the position to be screened 2 is 0.2, the weight of the position to be screened 3 is 0.1, the weight of the position to be screened 4 is 0.2, and the weight of the position to be screened 5 is 0.2. The corresponding positions to be screened of 5 bird dropping experiments are position to be screened 2, position to be screened 3, position to be screened 1, position to be screened 4 and position to be screened 5, and the weight vector is (0.2, 0.1, 0.3, 0.2, 0.2).

[0044] The weight vector and the weight and initial sensing feature vector of the position to be screened are multiplied element by element to obtain the weighted sensing feature vector of the position to be screened.

[0045] It can be understood that through the four-layer architecture of simulation experiment-data fusion-space optimization-dynamic adaptation, the transition from "experience-driven" to "data-model dual-driven" is realized. Under the premise of ensuring line safety, the equipment performance is maximized, the deployment cost is minimized, and a replicable and scalable intelligent bird repelling system deployment scheme is formed.

[0046] In some embodiments, a plurality of key positions are determined from a plurality of positions to be screened according to the weighted sensing feature vector of each position to be screened, comprising:

[0047] For each position to be screened, the effective value of the position to be screened is calculated according to the weighted sensing feature vector of the position to be screened;

[0048] The position to be screened with an effective value greater than an effective threshold is taken as a candidate position;

[0049] According to the weighted sensing feature vectors of any two candidate positions, the vector similarity of any two candidate positions is calculated;

[0050] The plurality of candidate positions are divided into a plurality of position clusters according to the vector similarity of any two candidate positions by a clustering algorithm.

[0051] For each position cluster, a candidate position is selected from the position cluster as a key position.

[0052] Specifically, for each position to be screened, the weighted induced feature vector of the position to be screened can be summed to obtain an effective value of the position to be screened.

[0053] The vector similarity of two candidate positions can be calculated according to the following formula:

[0054]

[0055] Wherein, S (i,j) is the vector similarity of the i th candidate position and the j th candidate position, a (i,n) is the value of the n th element of the weighted induced feature vector of the i th candidate position, a (j,n) is the value of the n th element of the weighted induced feature vector of the j th candidate position, p is a parameter greater than 0 and less than 1, and N is the total number of elements included in the weighted induced feature vector.

[0056] The hierarchical clustering (Hierarchical Clustering) algorithm can be used to merge similar positions layer by layer based on the vector similarity of any two candidate positions, form a tree structure, and finally obtain a plurality of position clusters by pruning according to requirements, which specifically includes the following steps:

[0057] S11, constructing a similarity matrix:

[0058] The vector similarity of any two candidate positions forms an MxM matrix S, where M is the number of candidate positions.

[0059] Example: for 10 candidate positions, a 10x10 similarity matrix is generated, the diagonal line is 1 (self-similarity), and the value range of the remaining elements is [0, 1].

[0060] S12, initializing clusters:

[0061] Each candidate position is initialized as an independent cluster, and there are M clusters in total.

[0062] S13, iteratively merging clusters:

[0063] Each time, the two clusters with the highest similarity are selected for merging, and the similarity matrix is updated.

[0064] Merging strategy: Ward method is adopted to minimize the increment of intra-cluster variance after merging to ensure that the positions within the cluster are as similar as possible.

[0065] S14, prune to generate final clusters:

[0066] Prune the hierarchical tree according to the preset maximum cluster diameter (e.g., 50 meters) or target cluster number to form final location clusters.

[0067] S15, key parameter design:

[0068] Spatial constraint: set the maximum allowed distance D between any two locations within a cluster to ensure spatial continuity of the cluster. max

[0069] Dynamic adjustment: adjust according to the line voltage level, for example:

[0070] 220kV line: D max = 40 meters;

[0071] 500kV line: D max = 50 meters.

[0072] Stopping condition: stop merging when the average of cluster inter-vector similarity is below the similarity threshold S min (e.g., 0.75), avoiding over-aggregation and ensuring high consistency of location features within a cluster.

[0073] The key location can be any candidate location from the location cluster, or the candidate location corresponding to the cluster center of the location cluster.

[0074] It can be understood that by calculating the effective value of each to-be-screened location and taking the to-be-screened location with an effective value greater than the effective threshold as a candidate location, those relatively unimportant locations in feature performance can be effectively filtered out. In this way, the subsequent calculation and clustering process only needs to be performed on the candidate locations, greatly reducing the amount of data to be processed, thereby improving the running efficiency of the entire algorithm.

[0075] Focus on key information: the effective value can be regarded as a quantitative measure of the importance of the to-be-screened location. Only keeping candidate locations with high effective values is equivalent to focusing on those locations with key features in a specific scenario, which helps to more accurately extract key locations and avoid being disturbed by a large number of irrelevant or low-value location information.

[0076] Calculating the vector similarity of the weighted induced feature vector of any two candidate locations can quantitatively measure the similarity between the candidate locations. In actual situations, many locations may have similarities in function, environment, etc. Vector similarity can accurately identify these similar locations and provide a basis for subsequent clustering operations.

[0077] ​The plurality of candidate positions are divided into a plurality of position clusters according to the vector similarity by using a clustering algorithm, and the candidate positions with high similarity can be grouped together. The candidate positions in each position cluster have high similarity, and if all candidate positions are used as key positions, a large amount of redundant information will exist between the key positions. Selecting one candidate position from each position cluster as a key position can effectively reduce the number of key positions while ensuring the representativeness of the key positions, and avoid information redundancy.

[0078] Through a series of steps, a representative key position can be efficiently and accurately determined from a plurality of positions to be screened, which has obvious advantages in data processing efficiency, information focusing degree and key position quality.

[0079] In step 120, the first bird repelling probability is determined according to the sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points.

[0080] Specifically, it includes:

[0081] According to the weighted sensing feature vectors of any two key positions, the correlation coefficient of any two key positions is calculated.

[0082] According to the sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points and the correlation coefficient of any two key positions, a plurality of key vibration sensing devices are selected from the plurality of vibration sensing devices.

[0083] For each key vibration sensing device, the sensing data collected by the key vibration sensing device at a plurality of continuous time points is subjected to variational modal decomposition, and the key vibration features of the key position where the key vibration sensing device is located are extracted.

[0084] According to the key vibration features of the key position where each key vibration sensing device is located and the correlation coefficient of any two key positions, the first bird repelling probability is determined by a probability prediction model.

[0085] Specifically, the elements of the weighted sensing feature vectors of any two key positions can be substituted into the calculation formula of the correlation coefficient (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) to calculate the correlation coefficient of the two key positions.

[0086] As a preferred embodiment, according to the sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points and the correlation coefficient of any two key positions, a plurality of key vibration sensing devices are selected from the plurality of vibration sensing devices, which can include the following steps:

[0087] S21, for each key position, the key position whose correlation coefficient is greater than the correlation coefficient threshold can be regarded as the relevant key position of the key position.

[0088] S22, for each vibration sensing device, a maximum vibration amplitude can be extracted from the sensing data collected by the vibration sensing device, and the vibration sensing device with the maximum vibration amplitude greater than a vibration amplitude threshold is taken as a central vibration sensing device;

[0089] S23, for each central vibration sensing device, a vibration sensing device set at a relevant key position of a key position where the central vibration sensing device is located is taken as a relevant vibration sensing device of the central vibration sensing device;

[0090] S24, the relevant vibration sensing devices of all the central vibration sensing devices are de-duplicated, and the vibration sensing devices remaining after de-duplication are taken as key vibration sensing devices.

[0091] It can be understood that the variational modal decomposition is a self-adaptive and completely non-recursive signal processing method, which decomposes a signal into a plurality of modal functions (Intrinsic Mode Functions, IMFs) with a center frequency and a limited bandwidth. The core idea is to construct and solve a variational problem, so that the error between the sum of the modal functions obtained by decomposition and the original signal is minimized, while each modal function meets certain constraint conditions, such as limited bandwidth, center frequency, etc.

[0092] The key vibration features of the key position where the key vibration sensing device is located can include modal energy features, center frequency features, instantaneous frequency features, and modal correlation features, wherein the modal energy features include the energy of each modal function obtained by decomposition, the center frequency features can include the center frequency of each modal function obtained by decomposition, the instantaneous frequency features can include the instantaneous frequency of each modal function obtained by decomposition, and the modal correlation features can include the correlation coefficient between different modal functions obtained by decomposition.

[0093] The probability prediction model can be a convolutional neural network model, the input of the probability prediction model can include the key vibration features of the key position where each key vibration sensing device is located and the correlation coefficient of the key position where any two key vibration sensing devices are located, and the output of the probability prediction model can include a first bird repelling probability, wherein the first bird repelling probability is used to represent the probability of a bird falling event occurring on the overhead line.

[0094] It can be understood that the screened key vibration sensing devices are located in the areas more closely related to the current occurring bird falling events. This enables the subsequent analysis and prediction to be more focused on these key areas, improves the pertinence of bird falling event prediction, reduces the interference of irrelevant information, and helps to improve the accuracy of prediction. The collected sensing data of the key vibration sensing devices are subjected to variational modal decomposition, which can decompose the complex vibration signal into a series of modal components with different frequency characteristics, and then extract key vibration features. These features can more accurately reflect the internal laws and essential characteristics of the vibration signal, and compared with the original data, they can better reflect the vibration patterns related to bird falling events. For example, when birds fall on overhead lines, they will produce vibrations with specific frequencies and amplitudes. Through variational modal decomposition, these features can be extracted, providing a more reliable data basis for subsequent probability prediction.

[0095] In step 130, the first bird repelling probability is adjusted according to the key meteorological information of the overhead line to generate a second bird repelling probability.

[0096] In some embodiments, the key meteorological information of the overhead line is obtained, including:

[0097] determining a plurality of meteorological factors;

[0098] obtaining meteorological data and bird repelling times of a plurality of historical time periods;

[0099] For each meteorological factor, the correlation coefficient of the meteorological factor and the bird repelling times is calculated according to the meteorological data and the bird repelling times of the plurality of historical time periods;

[0100] According to the correlation coefficient of each meteorological factor and the bird repelling times, a plurality of key meteorological factors are screened from the plurality of meteorological factors;

[0101] According to the plurality of key meteorological factors, the key meteorological information of the overhead line is obtained.

[0102] Specifically, meteorological conditions can affect the activities of birds. For example, in sunny weather, birds are more likely to see the location of overhead lines and thus choose to perch or stay in these places. In addition, sunny weather usually means good flying conditions, and birds may be more willing to fly and may choose overhead lines as temporary resting places. In calm or light wind weather, birds are more likely to maintain flight balance and are more willing to stay near overhead lines. Strong wind weather can make it difficult for birds to maintain flight balance, thereby reducing their activities near overhead lines.

[0103] The meteorological data can be obtained from professional meteorological observation stations, meteorological satellites, meteorological radars and the like. The data provided by these channels has high accuracy and reliability, and can truly reflect the meteorological conditions of the area where the overhead line is located. The bird driving frequency data can be obtained through the operation and maintenance records of the overhead line, the monitoring system of the bird driving equipment and the like. The operation and maintenance records will record the time, location, driving method and driving reason of each bird driving in detail. Through the arrangement and analysis of these records, the bird driving frequency of different historical time periods can be obtained.

[0104] The meteorological data and bird driving frequency of multiple historical time periods can be used to calculate the correlation coefficient of meteorological factors and bird driving frequency, which is substituted into the calculation formula of the correlation coefficient to obtain the correlation coefficient of meteorological factors and bird driving frequency. The meteorological factors with a correlation coefficient greater than a correlation coefficient threshold can be regarded as key meteorological factors. The key meteorological information of the overhead line can include the values corresponding to each key meteorological factor.

[0105] In some embodiments, step 130 specifically includes:

[0106] The adjustment coefficient is predicted according to the key meteorological information of the overhead line through a coefficient prediction model, wherein the coefficient prediction model can be a long short-term memory network model.

[0107] The first bird driving probability is adjusted according to the adjustment coefficient to generate a second bird driving probability. Specifically, the product of the adjustment coefficient and the first bird driving probability can be taken as the second bird driving probability.

[0108] It can be understood that the key meteorological information is an important factor affecting the activities of birds on the overhead line. Different meteorological conditions will change the activity patterns of birds, thereby affecting the possibility of bird falling events. The coefficient prediction model can process and remember the long-term dependence relationship in the time series data. By predicting the adjustment coefficient according to the key meteorological information through the model, the dynamic changes of meteorological factors can be taken into account. For example, in the case of continuous high temperature, birds may change their activity areas to find water sources and cool environments. The coefficient prediction model can capture this meteorological trend and predict the corresponding adjustment coefficient to make more accurate adjustments to the first bird driving probability, so that the second bird driving probability is more in line with the actual situation.

[0109] Step 140, determining whether to perform bird identification according to the second bird driving probability.

[0110] Specifically, if the second bird driving probability is greater than a second bird driving probability threshold, it is determined to perform bird identification.

[0111] Step 150, collecting an overhead line image when it is determined to perform bird identification.

[0112] Step 160, identifying birds according to the overhead line image.

[0113] Figure 3 is a flowchart illustrating a process of identifying a bird according to some embodiments of the present specification, as shown in Figure 3 In some embodiments, step 160 specifically comprises:

[0114] obtaining images of a plurality of sample birds, wherein the plurality of sample birds are of different species and / or growth stages;

[0115] extracting biological features of each sample bird;

[0116] dividing the plurality of sample birds into a plurality of sample bird clusters according to the biological features of each sample bird by using a multi-level clustering algorithm;

[0117] for each sample bird cluster, determining a comprehensive biological feature corresponding to the sample bird cluster according to the biological features of each sample bird included in the sample bird cluster;

[0118] segmenting a region of interest image from the overhead line image;

[0119] extracting biological features of the bird to be driven from the region of interest image;

[0120] identifying the bird based on the biological features of the bird to be driven and the comprehensive biological feature corresponding to each sample bird cluster.

[0121] Specifically, a plurality of representative birds are selected as sample birds. These sample birds should cover different families, different habits of birds commonly living near overhead lines, which may pose a threat to the safety of the line, to ensure that the research results have wide applicability.

[0122] The biological features of the sample birds can include morphological features (e.g., body length, wing span, tail length, head-body ratio, wing-body ratio, etc.), feather features (e.g., feather color, feather pattern, etc.), head features (e.g., beak shape, eye size, color, etc.).

[0123] The plurality of sample birds are divided into a plurality of sample bird clusters according to the biological features of each sample bird by using a multi-level clustering algorithm, which can include the following steps:

[0124] S31, for each biological characteristic, the difference value between any two sample birds in the biological characteristic can be calculated, for example, the difference value between two sample birds in body length is calculated using absolute difference value or relative difference value, for another example, the beak can be geometrically modeled, the length, width, curvature and other geometric parameters of the beak are extracted, and then the difference between the geometric parameters of the beaks of the two birds is calculated. For example, the absolute difference value of the beak length, the relative difference value of the width, and the like, and then the difference values of multiple parameters are weighted and summed to obtain the comprehensive difference value of the beak shape, for another example, the color can be converted into color coordinates (such as RGB, HSV, etc.), and the Euclidean distance between the feather colors of the two birds in the color space is calculated as the difference value. The standard deviation of the difference value between any two sample birds in the biological characteristic can be calculated as the comprehensive difference fluctuation coefficient of the biological characteristic;

[0125] S32, according to the comprehensive difference fluctuation coefficient of each biological characteristic, the key biological characteristics for the first clustering are calculated, for example, the top three biological characteristics can be selected as the key biological characteristics for the first clustering according to the comprehensive difference fluctuation coefficient of each biological characteristic;

[0126] S33, according to the key biological characteristics for the first clustering, the first key biological characteristic similarity between any two sample birds is calculated, and the multiple sample bird clusters are determined by clustering algorithm according to the first key biological characteristic similarity between any two sample birds.

[0127] S34, for each first sample bird cluster, whether to perform the next clustering is determined according to the comprehensive difference fluctuation coefficient of the sample birds included in the first sample bird cluster in the remaining biological characteristics not used for clustering, for example, if there is a biological characteristic with a comprehensive difference fluctuation coefficient greater than a comprehensive difference fluctuation coefficient threshold, the next clustering is performed, and the biological characteristic with a comprehensive difference fluctuation coefficient greater than a comprehensive difference fluctuation coefficient threshold is determined as the biological characteristic for the next clustering to determine the second sample bird cluster included in the first sample bird cluster.

[0128] S35, for each second sample bird cluster included in the first sample bird cluster, whether to perform the next clustering is determined in the same manner as S34 until the comprehensive difference fluctuation coefficient of each sample bird cluster in the remaining biological characteristics not used for clustering is less than or equal to the comprehensive difference fluctuation coefficient threshold.

[0129] The first sample bird cluster to which the driven bird belongs can be determined according to the feature value of the driven bird in the key biological characteristics used for the first clustering, and then the second sample bird cluster to which the driven bird belongs can be determined according to the feature value of the driven bird in the key biological characteristics used for the second clustering in the first sample bird cluster to which the driven bird belongs. The bird species is identified by analogy.

[0130] It can be understood that various biological characteristics are selected, such as morphological characteristics (body length, wing span, etc.), feather characteristics (color, pattern, etc.), head characteristics (beak shape, eye size, etc.). Different biological characteristics reflect the characteristics of birds from different angles. By combining these characteristics for identification, misjudgment caused by relying on a single characteristic can be avoided. For example, it can be difficult to distinguish some birds of different species with similar colors based on feather color alone, but combining body length, beak shape, and other characteristics can more accurately identify the birds. Through a multi-level clustering algorithm, the sample birds are gradually subdivided into multiple sample bird clusters. In the first clustering, the sample birds are divided into multiple first sample bird clusters according to the key biological characteristics, and then further subdivided according to the remaining biological characteristics. This hierarchical clustering method can more accurately capture the differences between birds and improve the accuracy of identification. For example, after the first clustering, some birds with similar body size and beak shape can be classified into one category, but through feather patterns and other characteristics, these birds can be further distinguished in subsequent clustering.

[0131] Step 170, determining the ultrasonic bird repelling parameters according to the identified bird species.

[0132] In some embodiments, step 170 specifically includes:

[0133] determining the optimal ultrasonic bird repelling parameters for each sample bird;

[0134] determining the ultrasonic bird repelling parameters according to the identified bird species and the optimal ultrasonic bird repelling parameters for each sample bird.

[0135] Specifically, for each sample bird, a series of ultrasonic bird repelling experiments with different parameter combinations are designed. These parameters can include the frequency range, sound pressure level, emission mode (such as continuous emission, intermittent emission, pulse emission, etc.), emission duration, etc. For example, for the frequency range, multiple different intervals from low frequency to high frequency can be set for testing; for the sound pressure level, different decibel values are set for comparison.

[0136] It can be understood that when identifying the bird to be repelled, the key biological characteristics used in different levels of clustering are used to gradually determine the sample bird cluster to which the bird to be repelled belongs. This step-by-step positioning method avoids large-scale similarity calculations in all sample birds, reducing the amount of calculation and improving the identification efficiency. For example, the first sample bird cluster to which the bird to be repelled belongs is determined according to the key biological characteristics of the first clustering, and then the second sample bird cluster to which it belongs is determined according to the key biological characteristics of the second clustering within the cluster. In this way, the bird species is accurately identified.

[0137] During the experiment, the behavior of the sample birds in response to different combinations of ultrasonic parameters is closely observed. The various behaviors of the birds after hearing the ultrasonic waves, such as whether they appear to be startled, move away from the sound source, change flight trajectory, emit alarm calls, and other behaviors, as well as the frequency and duration of these behaviors, are recorded. At the same time, a series of evaluation indicators are set to quantify the bird repelling effect, such as the percentage of reduction in the birds' stay time in a specific area, the degree of reduction in the number of times entering the danger zone, and so on.

[0138] In-depth analysis of experimental data is conducted, and the best combination of ultrasonic parameters in various evaluation indicators is found through statistical methods, which is determined as the optimal ultrasonic bird repelling parameters for the sample bird. For example, through data analysis, it is found that when the ultrasonic frequency is 20-25 kHz, the sound pressure level is 90 dB, and the intermittent emission mode is used (1 second emission every 5 seconds), the stay time of a certain bird near the overhead line is significantly reduced, and the number of times entering the danger zone is almost zero, so this parameter combination is determined as the optimal ultrasonic bird repelling parameters for this bird.

[0139] It can be understood that after identifying the bird species near the overhead line through image recognition, sound recognition and other technical means, the identified bird is matched with the pre-determined sample bird. If the identified bird is exactly one of the sample birds, the optimal ultrasonic bird repelling parameters corresponding to the sample bird can be directly called.

[0140] If the identified bird does not have a complete corresponding species in the sample birds, but belongs to a certain family or has similar habits, the optimal ultrasonic bird repelling parameters of the sample bird closest to the bird can be referred to, and appropriate adjustments can be made in combination with the specific characteristics of the bird. For example, a certain bird may be similar to a certain sample bird in terms of body size, flight habits, etc., but may have slight differences in sound sensitivity, so the sound pressure level of the ultrasonic wave can be fine-tuned based on the optimal parameters of the sample bird to adapt to the characteristics of this bird.

[0141] It can be understood that by determining specific ultrasonic bird repelling parameters for different birds, the auditory system of the birds can be more accurately stimulated to trigger their discomfort or alertness response, thereby improving the bird repelling effect and reducing the birds' stay and nesting behavior near the overhead line. Using the optimal ultrasonic bird repelling parameters can ensure the bird repelling effect while avoiding excessive emission of ultrasonic waves, reducing energy consumption of the bird repelling device, prolonging the service life of the device, and reducing unnecessary interference to the surrounding environment. It can adapt to the changes in bird species in different regions and different seasons, and through flexible adjustment of the ultrasonic bird repelling parameters, it can ensure that the bird repelling system always operates efficiently, providing reliable protection for the safe operation of the overhead line.

[0142] Step 180, determine the bird repelling pose parameters according to the overhead line image.

[0143] In some embodiments, step 180 specifically comprises:

[0144] According to the overhead line image, the spatial position of the bird to be driven is determined.

[0145] According to the spatial position of the bird to be driven and the ultrasonic bird driving parameters, the bird driving pose parameters are determined, wherein the bird driving pose parameters at least include the ultrasonic wave emission spatial position, the ultrasonic wave emission horizontal angle and the ultrasonic wave emission pitch angle.

[0146] Specifically, in combination with the installation position, focal length, resolution and other parameters of the image acquisition device, and the pixel coordinate information of the birds in the image, the pixel coordinates of the birds are converted into actual three-dimensional space coordinates through geometric transformation and spatial positioning algorithm (such as monocular vision positioning algorithm), so as to determine the specific spatial position of the bird to be driven target in the environment around the overhead line. For example, by knowing the installation height and angle of the camera, and the horizontal and vertical pixel positions of the birds in the image, the distance and height of the birds relative to the camera are calculated using trigonometric functions, and then the spatial position is determined.

[0147] According to the spatial position of the bird to be driven target, the best spatial position of the ultrasonic wave emitter is determined. The ultrasonic wave emitter should be as close as possible to the bird to be driven target, so as to reduce the attenuation of ultrasonic waves in the propagation process and improve the bird driving effect.

[0148] The ultrasonic wave emission horizontal angle refers to the pointing angle of the ultrasonic wave emitter in the horizontal direction. According to the position of the bird to be driven target in the horizontal plane, the horizontal angle that the ultrasonic wave emitter needs to rotate is calculated to ensure that the ultrasonic wave can accurately point to the target. For example, by calculating the included angle between the line connecting the bird to be driven target and the installation position of the bird driving device in the horizontal plane and a certain reference direction (such as the north direction), the horizontal emission angle of the ultrasonic wave emitter is determined.

[0149] The ultrasonic wave emission pitch angle refers to the inclination angle of the ultrasonic wave emitter in the vertical direction. Considering the height of the bird to be driven target and the propagation characteristics of the ultrasonic wave, a suitable pitch angle needs to be determined to make the ultrasonic wave cover the target area at the best angle. For example, if the bird to be driven target is located at a certain height above the overhead line, the ultrasonic wave emitter needs to be inclined upward at a certain angle to ensure that the ultrasonic wave can effectively reach the target position. When calculating the pitch angle, the horizontal distance between the ultrasonic wave emission spatial position and the target, and other factors are calculated using trigonometric functions.

[0150] The bird driving pose parameters can be determined according to the spatial position of the bird to be driven and the ultrasonic bird driving parameters through a parameter determination model, wherein the parameter determination model can be a convolutional neural network model.

[0151] In step 190, the ultrasonic bird repelling device is controlled to repel birds according to the ultrasonic bird repelling parameters and the bird repelling pose parameters.

[0152] Specifically, the pose of the ultrasonic bird repelling device is adjusted according to the bird repelling pose parameters, and after the pose adjustment is completed, the ultrasonic bird repelling device emits ultrasonic waves to repel birds according to the ultrasonic bird repelling parameters.

[0153] Figure 4 Fig. 1 is a schematic diagram of a bird repelling control system for a power system according to some embodiments of the present specification. Figure 4 As shown in Fig. 1, the bird repelling control system for a power system can include a bird sensing module, a bird identification module, and a bird repelling control module.

[0154] The bird sensing module is configured to set vibration sensing devices at a plurality of key positions of the overhead line, determine a first bird repelling probability according to sensing data collected by the plurality of vibration sensing devices at a plurality of continuous time points, adjust the first bird repelling probability according to key meteorological information of the overhead line to generate a second bird repelling probability, and determine whether to perform bird identification according to the second bird repelling probability.

[0155] The bird identification module is configured to, when it is determined to perform bird identification, collect an image of the overhead line, and identify birds according to the image of the overhead line.

[0156] The bird repelling control module is configured to determine ultrasonic bird repelling parameters according to the identified birds, determine bird repelling pose parameters according to the image of the overhead line, and control the ultrasonic bird repelling device to repel birds according to the ultrasonic bird repelling parameters and the bird repelling pose parameters.

[0157] The bird repelling control system for a power system can perform a bird repelling control method for a power system, which will not be described herein.

[0158] Finally, it should be understood that the embodiments described in the specification are only used to illustrate the principles of the embodiments of the specification. Other variations can also be within the scope of the specification. Therefore, as an example but not limitation, alternative configurations of the embodiments of the specification can be considered consistent with the teachings of the specification. Accordingly, the embodiments of the specification are not limited to the embodiments explicitly introduced and described in the specification.

Claims

1. A bird-repelling control method for a power system, characterized in that: include: Vibration sensing devices are installed at several key locations on overhead lines; determining a first bird-repelling probability based on sensing data collected by the plurality of vibration sensing devices at a plurality of consecutive time points; The first bird-repelling probability is adjusted according to key meteorological information of the overhead line to generate a second bird-repelling probability; According to the second bird-repelling probability, determine whether to perform bird identification; When determining bird identification, collect images of overhead lines; Identify birds based on overhead power line images; Determine ultrasonic bird repellent parameters based on the identified birds; Determine bird-repelling posture parameters based on overhead line images; According to the ultrasonic bird repellent parameters and bird repellent posture parameters, the ultrasonic bird repellent equipment is controlled to repel birds; Among them, several key locations of overhead lines were determined, including: According to preset intervals, multiple locations to be screened are evenly determined on the simulated overhead line, and a vibration sensing device is set at each location to be screened; Conduct multiple bird landing experiments on the simulated overhead line, wherein the distance between the landing positions of any two bird landing experiments is greater than a preset distance threshold; For each position to be screened, an initial sensing feature vector of the position to be screened is determined based on sensing data collected by a vibration sensing device at the position to be screened in each bird landing experiment; Determine the weight of each location to be screened based on historical bird landing data on overhead lines; For each position to be screened, a weighted sensing feature vector is generated according to the weight of each position to be screened and the initial sensing feature vector; Determining a plurality of key positions from a plurality of positions to be screened according to a weighted sensing feature vector of each position to be screened; According to the weighted sensing feature vector of each position to be screened, multiple key positions are determined from the multiple positions to be screened, including: For each position to be screened, the effective value of the position to be screened is calculated according to the weighted sensing feature vector of the position to be screened; The locations to be screened whose effective values ​​are greater than the effective threshold are taken as candidate locations; Calculate the vector similarity between any two candidate positions based on their weighted sensing feature vectors; Through the clustering algorithm, multiple candidate locations are divided into multiple location clusters according to the vector similarity between any two candidate locations; For each location cluster, a candidate location is screened from the location cluster as the key location.

2. The bird-repelling control method for a power system according to claim 1, characterized in that: Determining a first bird-repelling probability based on sensing data collected by multiple vibration sensing devices at multiple consecutive time points includes: Calculate the correlation coefficient of any two key positions based on the weighted induction feature vectors of any two key positions; screening a plurality of key vibration sensing devices from the plurality of vibration sensing devices according to sensing data collected by the plurality of vibration sensing devices at a plurality of consecutive time points and correlation coefficients of any two key positions; For each key vibration sensing device, performing variational modal decomposition on sensing data collected by the key vibration sensing device at multiple consecutive time points to extract key vibration features of the key position where the key vibration sensing device is located; The first bird-repelling probability is determined through a probability prediction model according to the key vibration characteristics of the key position where each key vibration sensing device is located and the correlation coefficient between any two key positions.

3. The bird-repelling control method for a power system according to claim 1 or 2, characterized in that: Obtain key meteorological information for overhead lines, including: Identify multiple meteorological factors; Obtain meteorological data and bird-scaring times for multiple historical time periods; For each meteorological factor, the correlation coefficient between the meteorological factor and the number of bird-scaring times was calculated based on the meteorological data and the number of bird-scaring times in multiple historical time periods; According to the correlation coefficient between each meteorological factor and the number of bird-scare events, multiple key meteorological factors are selected from multiple meteorological factors; Obtain key meteorological information of overhead lines based on multiple key meteorological factors.

4. The bird-repelling control method for a power system according to claim 1 or 2, characterized in that: The first bird-repelling probability is adjusted based on key meteorological information of overhead lines to generate a second bird-repelling probability, including: Through the coefficient prediction model, the adjustment coefficient is predicted based on the key meteorological information of the overhead line; The first bird-repelling probability is adjusted according to the adjustment coefficient to generate a second bird-repelling probability.

5. The bird-repelling control method for a power system according to claim 1 or 2, characterized in that: Identify birds based on overhead power line images, including: Acquiring images of a plurality of sample birds, wherein any plurality of the sample birds are of different species and / or growth stages; Extract biological characteristics of each sample bird; Through the multi-level clustering algorithm, the various sample birds are divided into multiple sample bird clusters according to the biological characteristics of each sample bird; For each sample bird cluster, determining the comprehensive biological characteristics corresponding to the sample bird cluster according to the biological characteristics of each sample bird included in the sample bird cluster; Segmenting the region of interest image from the overhead line image; Extracting biological features of the birds to be driven away from the image of the region of interest; Birds are identified based on the biological characteristics of the birds to be driven away and the comprehensive biological characteristics corresponding to each sample bird cluster.

6. The bird-repelling control method for a power system according to claim 1 or 2, characterized in that: According to the identified birds, determine the ultrasonic bird repellent parameters, including: Determine the optimal ultrasonic bird repellent parameters for each sample bird; The ultrasonic bird repellent parameters are determined based on the identified birds and the optimal ultrasonic bird repellent parameters for each sample bird.

7. The bird-repelling control method for a power system according to claim 1 or 2, characterized in that: Determine bird-repelling posture parameters based on overhead line images, including: Determine the spatial location of birds to be driven away based on the overhead line image; The bird-repelling posture parameters are determined according to the spatial position of the bird to be repelled and the ultrasonic bird-repelling parameters, wherein the bird-repelling posture parameters at least include the ultrasonic emission spatial position, the ultrasonic emission horizontal angle, and the ultrasonic emission pitch angle.

8. A bird-repelling control system for a power system, characterized in that: A method for executing a bird-repelling control method for a power system according to any one of claims 1 to 7, comprising: A bird sensing module is configured to install vibration sensing devices at multiple key locations of the overhead line, determine a first bird-repelling probability based on sensing data collected by the multiple vibration sensing devices at multiple consecutive time points, adjust the first bird-repelling probability based on key meteorological information of the overhead line, generate a second bird-repelling probability, and determine whether to perform bird identification based on the second bird-repelling probability; A bird identification module is used to collect overhead line images when determining bird identification and identify birds based on the overhead line images; The bird repellent control module is used to determine the ultrasonic bird repellent parameters according to the identified birds, determine the bird repellent posture parameters according to the overhead line image, and control the ultrasonic bird repellent equipment to repel birds according to the ultrasonic bird repellent parameters and the bird repellent posture parameters.

Images