Research on Bird Pest Management in Substation Based on Big Data

Analysis

Ping Qian

, Donglei Weng

, Yong Zhang

, Guoyi Wang

, Jiang Lin

3,* e

and Guofu Gao

State Grid Zhejiang Electric Power Co., Ltd., Hangzhou, 310007, China

State Grid Zhejiang Electric Power Co., Ltd. Ningbo Power Supply Company, Ningbo, 315000, China

Nanjing Desoft Information Technology Development Co., Ltd., Nanjing, 210000, China

College of Life Sciences, Nanjing Agricultural University, Nanjing, 210095, China

805657092@qq.com

Keywords: Big Data, Substation, Bird Pest Management, Prevention and Control Technology.

Abstract: For better and more exact management of bird pest in substation based on birds' habits, this paper analyses

various bird's data collected from representative substations in Zhejiang Province through data clean,

features extraction and data analysis, establishes hazard level for bird species based on estimation of

damage caused by birds in substation in Zhejiang Province and proposes measures and suggestions for bird

pest management accordingly. The results from sites adopted the strategy demonstrate that, from Jan. 2022

to Sep. 2022, the trip rate of 220kV line was reduced by 80% and the trip rate of 110kV line was reduced by

65%, which obtained good application effect.

1 INTRODUCTION

Most of the substations are outdoor open type,

covering a large area, and are located in plains, hills

or mountains with good ecological environment.

Bird activities in these areas are relatively frequent,

which brings great hidden danger to the safe

operation of the substation. At the same time,

unattended mode has been widely adopted in

substations, which adds some difficulties to bird pest

control. In the actual operation of the substation,

many equipment failures caused by bird damage

have also occurred, and bird damage accidents are

showing a growing trend (Chen 2019). However, at

present, there is no effective method to effectively

prevent birds from entering the substation, which

requires us to master the specific situation of bird

damage in the substation, so as to formulate practical

bird damage prevention measures.

https://orcid.org/0000-0002-4937-3479

https://orcid.org/0000-0003-0829-7833

https://orcid.org/0000-0001-6100-170X

https://orcid.org/0000-0002-1309-6625

https://orcid.org/0000-0001-7099-2605

https://orcid.org/0000-0001-9179-7331

At present, the management and prevention of

bird damage in substations are mainly concentrated

on the equipment level, such as bird drive devices

and bird prevention devices. The literature

(Carbonell I 2016) summarized that bird repelling

devices include sound, laser, chemical, wind and

other types. Through sound, smell, light and other

media, they stimulate birds' hearing, vision and taste

systems, cause birds to have fear and tension, and

worsen birds' living environment, so as to achieve

the effect of bird repelling. The literature (Egwumah

2022) points out that bird prevention devices mainly

include bird thorn and bird baffle, which are mainly

used to destroy the nesting space of birds by

installing baffles or bird thorns in the hollows of

door frame, gate knife base and other equipment, so

as to achieve the goal of bird prevention. Although

bird prevention equipment has played a certain role

in the application of bird pest prevention, the effect

is difficult to meet the expected requirements,

mainly because birds have strong adaptability to

various bird repellents, and the effect is difficult to

continue to be effective (Mori 2021).

In the aspect of bird research involved in the

problem of bird pest in substations, the current focus

is on the impact of bird behavior on the equipment

120

Qian, P., Weng, D., Zhang, Y., Wang, G., Lin, J. and Gao, G.

Research on Bird Pest Management in Substation Based on Big Data Analysis.

DOI: 10.5220/0012070900003624

In Proceedings of the 2nd International Conference on Public Management and Big Data Analysis (PMBDA 2022), pages 120-126

ISBN: 978-989-758-658-3

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

in the substation, while the research on bird species

and distribution in substations in different seasons

and regions is less, which makes the research and

prevention of bird pest in substations less systematic

and targeted (Sun 2017). Under the above

background, based on big data analysis technology,

this paper conducts data mining analysis on the bird

species and distribution in the substation, so as to

propose targeted prevention and control measures to

better strengthen the management of bird pest in the

substation.

2 ACQUISITION OF BIRD DATA

IN SUBSTATION

Zhejiang Province is located in the south coast of

China. Within a year, it will experience weather

changes such as fog, plum rain, humidity, rainstorm,

typhoon, etc. This climate also makes the substation

more prone to bird damage accidents.

2.1 Selection of Substation

In mid June 2021, early July 2021 and late August

2021, a field survey of birds was carried out in 43

substations in Zhejiang Province. The stations cover

all cities in the province and are well representative

of substations in all regions. The types of substations

are shown in Table 1 below:

Table 1: Substation type.

Voltag

e level

110

220

500

1000

Numbe

4 8 29 1 1

The specific data studied include the

environmental data inside and outside the station,

the type and quantity statistics of bird species inside

and outside the station, and the number and type of

bird nests inside the station.

2.2 Collection of Bird Population

The survey time is usually 6:30-10:30 in the

morning and 2:00-5:30 in the afternoon. The survey

of birds outside the substation is mainly based on the

sample line method. Walk around the fence of the

substation for 1-1.5km/hr for a week, and record the

bird species and number within a distance of about

70m around the substation. The sample point

method is used to survey some inconvenient places.

The bird survey in the station adopts the

combination of sampling line and sampling point to

comprehensively record the bird species and

quantity on the ground and equipment within the

enclosure, and the birds flying over the substation

are also included in the station. In order to avoid

repeated statistics, individual birds flying in or out of

the station from outside will be included in the

number of birds in the station and not counted

outside the station (Wu 2020). Observation and

photographing equipment adopts 8×Binoculars and

Canon 7D body, equipped with Shima 150-600mm

or Canon 100-400mm lens.

2.3 Collection of Bird Species

During the collection of bird species, 43 substations

distributed in various regions were selected to

conduct bird nest survey in the station, count the

number of bird nests, observe and take photos to

record the size, shape, nesting materials and other

characteristics of the bird nests, nesting sites, eggs,

nestlings and parent birds in and out activities. For

the data of transmission line tripping caused by bird

damage, the statistics of companies in various cities

and regions in Zhejiang Province are used.

3 BIRD PEST MANAGEMENT IN

SUBSTATION BASED ON BIG

DATA ANALYSIS

3.1 Data Cleaning and Feature

Extraction

Data cleaning is a process of finding and eliminating

errors or invalid data. When species identification is

carried out according to various relevant

characteristics of bird nests, many bird nests are not

typical due to their attachment to various equipment.

The identification of such data should try to combine

the characteristics of bird eggs, nestlings and parent

birds' activities to improve the accuracy of bird

group species data collection. At the same time,

image preprocessing and data dimensionality

reduction techniques are also needed to extract the

key features of bird nests and flocks.

3.1.1 Image Preprocessing

For image preprocessing, the bird nest image is

grayed first, and the image in RGB space is

converted into a grayscale image. The calculation

formula is as follows (1):

Research on Bird Pest Management in Substation Based on Big Data Analysis

121

(, ) 0.299 (, ) 0.587 (, ) 0.144 (, )

rayxy Rxy Gxy Bxy=++ggg

(1)

Where:

𝑔𝑟𝑎𝑦𝑥, 𝑦 is the gray value,

(, )Rxy ,𝐺𝑥,𝑦 and𝐵𝑥,𝑦 are the red, green and

blue pixel values respectively. The image after gray

processing is shown in the following figure.

Figure 1: Image after grayscale processing.

Then gamma correction is carried out. Different

correction values

will make the transformation of

gray areas different. If 𝛾1, the gray value of the

image can be increased. If 𝛾1, the gray value of

the image can be reduced, and the gamma correction

processing formula is as follows (2):

(, ) (, )Yxy Ixy

(2)

Where: 𝑌𝑥, 𝑦 is the image after gamma

correction processing,

𝐼𝑥, 𝑦 is the input image, and

the correction value

can be taken as 0.5.

Finally, the radiation transformation is carried

out, and the calculation formula is as follows (3):

uabm

vcdn



  



  



  





(3)

Where: x,y are the coordinates in the source two-

dimensional coordinate system, and u,v are the

coordinates in the transformed coordinate system.

At the same time, edge detection is also required

to retain important feature data in the image and

reduce the amount of data to be processed, as shown

in the following figure:

Figure2: Flow chart of edge detection.

After the threshold processing, the image can be

clearly displayed, but the image needs to be

binarized to reduce the dimension of the pixel matrix

and enhance the contour of the bird nest and birds.

3.1.2 Feature Extraction

For the feature extraction of bird flock and bird nest

images, this paper mainly adopts the Histogram of

Oriented Gradient (Hog). After the image is grayed

and gamma corrected, the gradient size and direction

of each pixel can be calculated. Sobel horizontal

operator [- 1,0,1] and vertical operator [- 1,0,1] T

can be used to calculate the gradient and direction of

the pixel (x, y) in the x, y direction, as shown in the

following equations (4) and (5):

( , ) ( 1, ) ( 1, )

Gxy Ix y Ix y=+ −−

(4)

(, ) (, 1) (, 1)

Gxy Ixy Ixy=+−−

(5)

Where:

(, )

Gxy

(, )

Gxy

are the gradient of

the pixel in the horizontal and vertical directions

respectively, and

(, )Ixy

are the pixel values of the

pixel. The gradient amplitude

(, )Gxy

and direction

(, )xy

of pixels are shown in the following

equations (6) and (7):

(, ) (, ) (, )

Gxy G xy G xy=+

(6)

(, )

(, ) tan ( )

(, )

Gxy

−

(7)

Then the histogram channel and overlapping

block normalization processing are constructed to

obtain the required image features.

3.2 Clustering of Data

After the feature data is extracted, clustering

analysis can be carried out according to the feature

data. This paper mainly introduces K-means

clustering. In this algorithm, set the eigenvector set

X of samples as shown in formula (8) below:

{, ,, }

XXX X= 

(8)

In equation (8),

are the first to nth

eigenvectors. For the algorithm step, the first step is

the initial classification, so that k=0, and each

sample can be regarded as a category, namely:

(0)

{}( 1,2,,)

GXi N==

(9)

The second step is to calculate the distance

between various types, on this basis, further generate

a symmetric distance matrix:

()

ij m m

(10)

PMBDA 2022 - International Conference on Public Management and Big Data Analysis

122

In the formula, m is the number of categories

(m=N at the beginning of calculation). The third step

is to find the smallest element in the matrix D

(k)

obtained in the previous step. Let it be the distance

between and. Combine the two categories and into

one, and then generate a new clustering

(1) (1)

,,,



.Order k=k+1, m=m-1. After

the above steps have been completed, it is also

necessary to check whether the number of sample

categories generated meets the requirements. When

the number of categories meets the requirements, the

above calculation process can be ended (Yang 2017).

At the same time, according to the given data set, K

samples were initially selected as the initial center

by random method, and then the iterative calculation

was carried out step by step according to the

principle of the shortest distance. The following

figure 3 shows the application of K-means clustering

algorithm in the analysis of bird data.

Figure 3: Application of K-means Clustering Algorithm in

bird data analysis.

It can be seen from the above figure 3 that in the

application process of K-means clustering algorithm,

the first step is to select the initial condensation

point and carry out the initial classification of bird

data, but the abnormal outliers in the sample data

should be removed to ensure the accuracy of the

sample data. At the same time, the wrong data in the

sample should be corrected and verified. Then, if the

algorithm program judges that the initial

classification is reasonable, it will be the final

classification result (Zhang 2019). If the

classification of bird data is unreasonable, modify

the classification of bird data according to K-means

clustering algorithm until the requirements are met,

and complete the analysis of bird data characteristics.

European distance can be used for evaluation, and

the objective function is shown in equation (11)

below:

111

(, , , )

cci cn i ijij

iij

JU X X X J u d

===



，

(11)

Where, U is the membership matrix,

(0,1)

u ∈

representing the membership of the jth sample to the

ith category;

represents the cluster center of

category i, as shown in

(, , , )

cci cn

JU X X X，

;

ij ci j

dXX=−

is the Euclidean distance from

and

representing the relative distance between the

jth sample feature vector

and the cluster center

of category i; m is the weighted index (Zhang

2020). By synthesizing the above formula and using

the Lagrangian transformation, we can obtain the

necessary conditions to minimize the above formula:

ij i



(12)

()

−



(13)

3.3 Data Analysis

3.3.1 Analysis Index Selection

When using big data for bird pest management, it is

necessary to reasonably select analysis indicators to

improve the pertinence of bird pest management, as

shown in Figure 4 below:

Whether it will enter the

substation

Whether it will enter the

substation

Size, quantity, material and

location of the nest

Size, quantity, material and

location of the nest

Size and shape of bird

Individual number of bird

species

Individual number of bird

species

bird

harm

Tube

reason

branch

Analys

finger

mark

bird

harm

Tube

reason

branch

Analys

finger

mark

Birds group flight

Figure 4: Analysis indicators of bird pest management.

In case of bird damage in the substation, birds

need to enter the substation first, and the larger the

Research on Bird Pest Management in Substation Based on Big Data Analysis

123

bird body, the more the bird number, the more the

bird nest, and the frequent group flight will increase

the probability of bird damage accidents (Zhang

2021).

3.3.2 Species Composition

According to big data analysis, 72 species of birds

belonging to 12 orders and 31 families were

recorded inside and outside 43 substations in

Zhejiang Province. Passeriformes has the largest

number of birds, with 42 species belonging to 19

families, while non Passeriformes has 30 species

belonging to 12 families. The results by residence

type are shown in the following table:

Table 2: Classification results by residence type.

birds

Number of

ecies

percentage

resident bird 48 66.67

Summer bird 23 31.94

Winter Migratory

Birds

1 1.39

In terms of fauna, there are 31 species (43.05%)

in the Oriental realm, 28 species (38.89%) in the

widespread realm, and at least 13 species (18.06%)

in the Palaearctic realm.

3.3.3 Comparison of Bird Species and

Individual Numbers Outside and

Inside the Station

A total of 72 species of birds were counted outside

the 43 substations and 28 species were counted

inside. The maximum number of bird species

outside the station is 22, and the minimum number is

5, with an average of 12.58 species per station.

There are 11 species with the largest number in the

station, and 3 species with the smallest number. The

average number of species in each station is 6. The

average number of species in the station is 46% of

that outside the station.

Through analysis, the number of bird species in

different substations is mainly related to the

environment, and the substation close to or near the

mountain has the most bird species. In addition,

there are ponds or ditches and other waters around

the substation, and there are many bird species.

However, the substation far away from mountains or

in plain areas, and the surrounding areas are mainly

farmland, has relatively few bird species.

In terms of the number of birds, 5422 birds were

counted inside and outside the substation. There are

35 species of birds with less than 5 individuals,

accounting for nearly 50% of all species. There are

31 species with more than 10 individuals, and the

dominant species with more individuals are Passer

montanus, Hirundo rustica, Streptopelia chinensis,

Pycnonotus sinensis, Acrothers cristatellus, Turdus

merula, Passer rutilans, Egretta garzetta, Motacilla

alba, Lonchura punctulata, Sinouthora webbiana,

Cyprus daurica, etc. There are 1297 birds recorded

in the station, about 1/4 of the total 5422 birds inside

and outside the station, and 1/3 of the total 4135

birds outside the station.

The data analysis results show that more than

half of the bird species distributed in the substation

environment will not enter the station or the

probability of entering is very low, which is closely

related to the living habits of birds. These species

include Gallinula chlopropus, Tachybaptus ruficollis,

waders, etc., which have close activities with the

water environment, and birds that highly depend on

shrubs, arbor forests and other special vegetation

activities, such as Sinouthora webbiana, Prinia

inornata, Horornis fortipes, Zosterops japonicus,

long tailed tits, cuckoos, Babblers, etc. Although

some species, such as the brown headed Brucea

javanica and dark green hydrangea, have a large

number of individuals outside the station, they will

not pose a security threat to the equipment in the

station.

3.3.4 Analysis of the Hidden Danger of the

Bird's Nest

In this study, 359 bird nests were observed on

electrical equipment, 329 bird nests were identified,

with a recognition rate of 92%. 30 bird nests could

not be identified due to incomplete or atypical shape.

Table 3 shows the types, numbers and stations of

bird nests.

Table 3: Type and quantity of bird nests in substation.

Scientific name

The number of

bird nests

The number of

ubstations

site of nests

Streptopelia chinensis 46 13 Knife base

Acridotheres cristatellus 30 12 Knife base

Pica pica 10 4 Longmen frame

PMBDA 2022 - International Conference on Public Management and Big Data Analysis

124

Turdus merula 63 16 Knife base

Pycnonotus sinensis 1 1

Area above insulating

orcelain bottle

Passer montanus 103 15

Heat shrinkable sleeve, main

transformer radiator

Lonchura punctulata 76 5

Main transformer radiator,

conservator, heat shrink

sleeve, support structure

In this study, 7 species of nests were identified,

all of which were resident birds. Among them,

sparrows have the most nests and bulbuls have the

least. The bird species for nest breeding in the

substation are ranked as: Passer montanus, Lonchura

punctulata, Turdus merula, Streptopelia chinensis,

Acridotheres cristatellus, Pica pica and Pycnonotus

sinensis according to the number dominance of nests.

The occurrence frequency of each bird species is

shown in the following figure:

Figure 5: Occurrence frequency of each bird species.

The above nest identification and analysis results

show that the hidden danger of bird nests in

substations in Zhejiang Province mainly comes from

six bird species: Turdus merula, Passer montanus,

Streptopelia chinensis, Acridotheres cristatellus,

Pica pica and Lonchura punctulata. Nesting birds are

common birds in the station, and there is no nesting

phenomenon found in many birds such as white

wagtail, golden waist swallow, mountain sparrow,

brown backed shrike, which is related to the nesting

environment and habits of these birds.

3.3.5 Hazard Analysis of Bird Species

Bird hazards include scattering of nest materials,

defecation, short circuit of transmission lines caused

by flashover during flight, etc. According to the

degree of damage, each bird species is divided into 5

hazard levels. Among them, there are 38 types of

non hazard level, 16, 9 and 3 types of hazard level 1,

2 and 3, and 2 and 4 types of hazard level 4 and 5, as

shown in Figure 6 below:

Figure 6: Hazard level of birds.

Among them, Level 4 and Level 5 are the most

harmful, including the following 6 bird species:

Streptopelia chinensis, Turdus merula, Acridotheres

cristatellus, Passer montanus, Pica pica and

Lonchura punctulata, all of which are resident birds.

Among them, the streptopelia chinensis has a long

body and wide wings, and the nest material structure

is loose and easy to scatter, so the probability of

causing an accident is greater. Most species have

low potential hazards, and more than half of birds

have no potential hazards to the substation.

4 MEASURES AND

SUGGESTIONS FOR BIRD

PEST MANAGEMENT

Measures and suggestions for bird pest management

mainly includes the following:

1) Placing artificial bird’s nests

Build some artificial nests of appropriate size

and shape for birds in appropriate areas near the

edge of substations or on iron towers outside the

substation to induce birds to nest on them reducing

and avoiding birds from nesting on other parts of the

station. It also helps to reduce birds' activity

frequency in the station. This method is mainly

aimed at Pica pica.

2) Environmental treatment in the station

Try to reduce the number of plants in the station,

timely trim the lawn, weeds and trees. That will

reduce birds' activity frequency and the probability

78%

65%

25%

20%

40%

60%

80%

100%

Turdus

merula and

Passer

montanus

Streptopelia

chinensis and

Acridotheres

cristatellus

Lonchura

punctulata

and Pica pica

hazard

Level 1Level 2Level 3Level 4Level 5

Research on Bird Pest Management in Substation Based on Big Data Analysis

125

of nesting and breeding in the station. In autumn and

winter, Passer montanus, Acridotheres cristatellus

and other birds are easy to group. Reducing the

vegetation in the station will help to reduce the

probability of accidents caused by birds' group

foraging and crossing.

3) Environmental treatment outside the station

The first is to clean up the iron wire and other

objects outside the station. The iron wire is often

used as nesting material by birds, and it is easy to

cause short circuit accidents when falling off on

related equipment during flight. Secondly, garbage

should be avoided near the substation, and there

should be no water source near the substation as far

as possible, increasing the difficulty for birds to

drink. Finally, trees against the wall should be

pruned in time.

4) Design anti-bird-nesting tools

Each station shall reasonably use bird repeller

according to the specific situation of bird damage,

and the substation far away from the residential area

can use tweeter to improve the bird repelling effect.

At the same time, the substation management

department shall uniformly design and customize

various bird stingers and other blocking tools

according to the equipment structure and bird nest

type of each station to effectively prevent birds from

nesting.

A power supply company in Zhejiang Province

which has adopted the above comprehensive bird

pest prevention management strategy has reduced

the 220kV line tripping rate by 80% and the 110kV

line tripping rate by 65% during the period from

January 2022 to September 2022 resulting good

application effect.

5 CONCLUSIONS

This paper combines big data analysis technology

with substation bird pest management, aiming to

explore the relationship between bird living habits,

distribution and bird pest accidents, so that power

supply enterprises can take targeted substation bird

pest management measures according to the actual

situation of local substations. Adopting the method

described in this paper in the substation bird pest

management can make the bird pest management

and prevention more accurate, achieve more

effective substation bird pest prevention effect, and

ensure the safe and stable operation of substations

and transmission lines.

FUNDS

2021 Science and Technology Project of State Grid

Zhejiang Electric Power Co., Ltd. (5211NB20013A)

REFERENCES

Chen Weisheng (2019). Improved the small animal

prevention measures and improved the anti short

circuit capability of the substation. J. Power System

Equipment .18, 86-87.

Carbonell I (2016). The ethics of big data in big

agriculture. J. Internet Policy Review. 5(1), 76-79.

Egwumah, F. A., Edet, D. I., & Okechukwu, O. C. (2022).

Effects of Land use on Composition of Birds Species.

Paramount Ecological Resources, 2(1), 1-8.

Mori, E., & Menchetti, M. (2021). The Ecological Impacts

of introduced parrots. Naturalized Parrots of the World:

Distribution, Ecology, and Impacts of the World’s

Most Colorful Colonizers, 87-101.

Sun Haihong, Ma Junlin (2017). Laser three-dimensional

bird pest control system for 110 kV substation. J.

Rural Electrification.12, 43-44.

Wu Yunhao, Liu Wei, Tai Deyun, et al (2020). Bird

community structure and bird strike prevention at

Changzhou Benniu Airport. J. Journal of Wildlife. 41

(3), 616 – 625.

Yang C, Huang Q, Li Z, et al (2017). Big Data and cloud

computing: innovation opportunities and challenges. J.

International Journal of Digital Earth.10(1), 13-53.

Zhang Xuri, Li Hepeng, Yue Chunlei, et al (2019). Bird

Diversity Directory of Zhejiang Province. J. Journal of

Wildlife. 40 (3), 685 – 699.

Zhang Hao, Wang Shuxin, Xu Ligang, et al (2020).

Development of bird proof grid with gate structure in

substation. J. Power Security Technology .22(2), 47-

50.

Zhang, F., Bales, C., & Fleyeh, H. (2021). From time

series to image analysis: A transfer learning approach

for night setback identification of district heating

substations. Journal of Building Engineering, 43,

102537.

PMBDA 2022 - International Conference on Public Management and Big Data Analysis

126