Correlation Analysis of Wind Farm Wind Speed Output Typical

Fluctuation Process Based on Deep Learning Algorithm

Jing Wang

1,2

, Ronaldo Juanatas

, Jing Bi

and Jianye Wei

Technological University of the Philippines, Manila, The Philippines

ChaohuUniversity,238024, China

Technological University of the Philippines Manila, The Philippines

China Anhui Huaxia Optoelectronics Co., LTD, Wuhu City, Anhui Province, 241002, China

Chaohu University, Hefei City, Anhui Province, 238024, China

Keywords: Deep Learning Theory, Deep Learning Algorithm, Wind Speed in the Wind Farm, Typical Fluctuations.

Abstract: With the increasing improvement of people's quality of life and the increase in household appliances, the

capacity of wind turbines increases, the diameter of wind turbines increases, and a large amount of rotational

inertia is generated during the rotation process of wind turbines. However, the output road cannot be changed

in real-time based on wind speed changes, leading to the problem of wind speed wavelet fluctuations. This

small fluctuation can cause large power fluctuations and even reduce power quality, affecting the safety and

stability of power system operation. To this end, a probability distribution model is established for wind farm

wind speed and wind power. A deep learning algorithm is used to analyze the correlation analysis of typical

fluctuation processes of wind farm speed output. The model is used to simulate wind farm power fluctuations,

and the correlation and smoothness analysis is used to analyze the spatiotemporal fluctuation of wind power

output, forming a correlation analysis evaluation plan. The results of the correlation analysis of typical

fluctuation processes of wind farm wind speed output are comprehensively evaluated. The results show that

compared with statistical methods, the deep learning algorithm has a smaller difference in the correlation

coefficient between wind speed fluctuations and wind power.

1 INTRODUCTION

The correlation between wind speed fluctuations and

wind power is one of the important contents of wind

power generation systems, which is of great

significance for the development of wind power

generation systems (Supraja and Salameh, et al.

2022). However, in the process of correlation analysis

evaluation, there is a problem of poor correlation in

the correlation analysis evaluation scheme, which

leads to serious power losses in wind power

generation systems (M, and Priyadi, , et al. 2022).

Some scholars believe that applying deep learning

algorithms to the correlation analysis of typical wind

farm wind speed output fluctuations can effectively

study the correlation analysis evaluation scheme and

provide corresponding support for the correlation

analysis evaluation (Yadav, and Sharma. et al. 2023).

On this basis, this article proposes a deep learning

algorithm to optimize the association analysis

evaluation scheme and verify the effectiveness of the

model (Bhatt. and Shikka, et al. 2022).

2 RELATED CONCEPTS

2.1 Mathematical Description of Deep

Learning Algorithms

The deep learning algorithm utilizes deep learning

theory to optimize the correlation analysis evaluation

scheme, and based on various indicators in the

correlation analysis evaluation, finds unqualified

values in the correlation analysis of wind farm wind

speed output typical fluctuation process (Bernardo

and Xing, et al. 2022), and integrates the correlation

analysis evaluation scheme to ultimately determine

the feasibility of the correlation analysis of wind farm

wind speed output typical fluctuation process. The

deep learning algorithm combines the advantages of

Wang, J., Juanatas, R., Bi, J. and Wei, J.

Correlation Analysis of Wind Farm Wind Speed Output Typical Fluctuation Process Based on Deep Learning Algorithm.

DOI: 10.5220/0013540200004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futur istic Technology (INCOFT 2025) - Volume 1, pages 295-299

ISBN: 978-989-758-763-4

295

deep learning theory and utilizes the correlation

analysis of typical wind farm wind speed output

fluctuations for quantification, which can improve the

correlation analysis evaluation of wind speed

fluctuations and wind power.

Assuming that the evaluation requirement for

correlation analysis is

max−t

v , the evaluation plan for

correlation analysis is

min−t

v , the satisfaction of the

evaluation plan for correlation analysis is

, and the

judgment function for the evaluation plan for

correlation analysis is

A as shown in formula (1).

−

−−

minmax

)

2.2 Selection of Correlation Schemes

Between Wind Speed Fluctuations

and Wind Power

Assumption II The correlation analysis function of

wind farm wind speed output typical fluctuation

process is, and the weight coefficient is. Therefore,

the correlation analysis evaluation requires the

correlation analysis of unqualified wind farm wind

speed output typical fluctuation process as shown in

formula (2).

−

−−

minmax

)

2.3 Analysis of Typical Fluctuation

Process of Correlation Analysis

Evaluation Scheme

Before conducting deep learning algorithms,

multidimensional analysis should be conducted on

the correlation analysis evaluation scheme, and the

requirements for correlation analysis evaluation

should be mapped to the correlation analysis library

of wind farm wind speed output typical fluctuation

processes, and unqualified correlation analysis

evaluation schemes should be eliminated. Firstly,

conduct a comprehensive analysis of the typical

fluctuation process of wind farm wind speed output

through correlation analysis, and set thresholds and

indicator weights for the correlation analysis

evaluation scheme to ensure the accuracy of deep

learning algorithms. The correlation analysis of wind

farm wind speed output typical fluctuation process is

a system testing correlation analysis evaluation

scheme that requires typical fluctuation process

analysis. If the correlation analysis of the typical wind

speed output fluctuation process of the wind farm is

in a non normal distribution, its correlation analysis

and evaluation scheme will be affected, reducing the

accuracy of the overall correlation analysis and

evaluation. In order to improve the accuracy of deep

learning algorithms and improve the level of

association analysis evaluation, it is necessary to

select the evaluation scheme for association analysis.

The specific scheme selection is shown in Figure 1.

Definition of fluctuation

process

Quantitative description

of fluctuation process

Fluctuation process based on

deep learning algorithm

Wind speed

fluctuation

Wind power

wavefront

correlation

Typical Fluctuation

Process of Wind Speed

Output in Wind Farm

Correlation

analysis results

Figure 1: Selection results of correlation schemes between

wind speed fluctuations and wind power

The investigation correlation analysis evaluation

plan shows that the correlation between wind speed

fluctuations and wind power shows a

multidimensional distribution, which is in line with

objective facts. The correlation analysis of wind farm

wind speed output typical fluctuation process has no

directionality, indicating that the correlation scheme

between wind speed fluctuation and wind power has

strong randomness, so it is considered as a higher

analysis research. The correlation analysis of wind

farm wind speed output typical fluctuation process

meets the normal requirements, mainly through deep

INCOFT 2025 - International Conference on Futuristic Technology

296

learning theory to adjust the correlation analysis of

wind farm wind speed output typical fluctuation

process, remove duplicate and irrelevant schemes,

and supplement the default schemes, making the

dynamic correlation of the entire correlation analysis

evaluation scheme strong.

3 OPTIMIZATION STRATEGY

FOR CORRELATION

ANALYSIS OF WIND FARM

WIND SPEED OUTPUT

TYPICAL FLUCTUATION

PROCESS

The deep learning algorithm adopts a stochastic

optimization strategy for the correlation analysis of

wind farm wind speed output typical fluctuation

process, and adjusts the fluctuation process

parameters to achieve the optimization of the scheme

for the correlation analysis of wind farm wind speed

output typical fluctuation process. The deep learning

algorithm divides the correlation analysis of wind

farm wind speed output typical fluctuation process

into different correlation analysis evaluation levels,

and randomly selects different schemes. During the

iteration process, the correlation analysis evaluation

schemes with different levels of correlation analysis

evaluation are optimized for typical fluctuation

processes. After optimizing the typical fluctuation

process, compare the correlation analysis evaluation

levels of different schemes and record the optimal

wind farm wind speed output typical fluctuation

process correlation analysis.

4 A PRACTICAL CASE STUDY

ON THE CORRELATION

ANALYSIS OF WIND SPEED

OUTPUT TYPICAL

FLUCTUATION PROCESS IN

WIND FARM

4.1 Introduction to Correlation

Analysis and Evaluation

For the convenience of correlation analysis and

evaluation, this article takes the correlation analysis

of wind farm wind speed output typical fluctuation

process under complex conditions as the research

object, with 12 paths and a testing time of 12 hours.

The specific correlation analysis and evaluation plan

of wind farm wind speed output typical fluctuation

process correlation analysis is shown in Table 1.

Table 1: Requirements for wind power system correlation

analysis and evaluation

Fluctuation

type

Volatility Wind

speed

fluctuation

Wind

power

wave

uantit

Upward

fluctuation

31.4457 24.3128 49.5477

Decline

fluctuation

31.6993 25.3570 50.2546

Stable

fluctuations

32.5422 26.4628 48.3050

The correlation analysis evaluation process in

Table 1 is shown in Figure 2.

Correlation Analysis of Fluctuation

Proces ses

Fluctuation process

First descending and then

ascending

Rise first and then fallDecline fluctuationSt a ble fluctu atio nsUp and down fluctuations

Figure 2: The Fluctuation Process of Wind Farm Wind

Speed Output Typical Fluctuation Process Correlation

Analysis

Compared with statistical methods, the

correlation analysis evaluation scheme of deep

learning algorithms is closer to the actual

requirements of correlation analysis evaluation. The

deep learning algorithm outperforms statistical

methods in terms of the rationality and fluctuation

amplitude of the correlation analysis of typical wind

speed output fluctuations in wind farms. According to

the correlation analysis evaluation scheme changes in

Figure 2, it can be seen that the deep learning

algorithm has better stability and faster judgment

speed. Therefore, the correlation analysis evaluation

scheme of deep learning algorithm has better

fluctuation speed, wind speed fluctuation amount,

and wind power stability.

Correlation Analysis of Wind Farm Wind Speed Output Typical Fluctuation Process Based on Deep Learning Algorithm

297

4.2 Correlation Analysis of Wind Farm

Wind Speed Output Typical

Fluctuation Process

The correlation analysis evaluation scheme for wind

farm wind speed output typical fluctuation process

includes unstructured information, semi structured

information, and structural information. After pre

selection of deep learning algorithms, a preliminary

correlation analysis evaluation scheme for wind farm

wind speed output typical fluctuation process

correlation analysis was obtained, and the feasibility

of the correlation analysis evaluation scheme for wind

farm wind speed output typical fluctuation process

correlation analysis was analyzed. In order to more

accurately verify the innovative effect of wind farm

wind speed output typical fluctuation process

correlation analysis, different correlation analysis

evaluation levels of wind farm wind speed output

typical fluctuation process correlation analysis were

selected, and the correlation analysis evaluation

scheme is shown in Table 2.

Table 2: Overall situation of wind speed fluctuations and

wind power correlation schemes

Fluctuation type Cluster center Number of data

units

Upward

fluctuation

49.0213 61.2641

Decline

fluctuation

47.3967 61.1110

Stable

fluctuations

50.0778 59.1769

4.3 The Correlation and Stability

Between Wind Speed Fluctuations

and Wind Power Fluctuations

Evaluated by Correlation Analysis

To verify the accuracy of the deep learning algorithm,

a correlation analysis evaluation scheme was

compared with statistical methods. The correlation

analysis evaluation scheme is shown in Figure 3.

As shown in Figure 3, the correlation between

wind speed fluctuations and wind power in the deep

learning algorithm is higher than that of statistical

methods, but the error rate is lower, indicating that the

correlation analysis evaluation of the deep learning

algorithm is relatively stable, while the correlation

analysis evaluation of statistical methods is uneven.

The average correlation analysis evaluation schemes

of the above three algorithms are shown in Table 3.

Figure 3: Correlation between wind speed fluctuations and

wind power using different algorithms

Table 3: Comparison of the accuracy of correlation

analysis evaluation using different methods

Algorithm Correlation

between

wind speed

fluctuations

and wind

owe

Change

amplitude

Erroneousness

Deep

learning

orith

91.0224 90.9412 89.7616

Statistical

methods

61.9487 62.9441 59.1769

Physical

methods

32.3613 31.0165 32.6413

From Table 3., it can be seen that statistical

methods have shortcomings in the correlation

analysis of wind farm wind speed output typical

fluctuation process, such as the correlation and

stability between wind speed fluctuation and wind

power. The correlation analysis of wind farm wind

speed output typical fluctuation process shows

significant changes, with a high error rate. The

general results of deep learning algorithms show a

high correlation between wind speed fluctuations and

wind power, which is superior to statistical methods.

Meanwhile, the correlation between wind speed

fluctuations and wind power of the deep learning

algorithm is greater than 90%, and there is no

significant change in accuracy. To further validate the

superiority of deep learning algorithms. In order to

further validate the effectiveness of the proposed

method in this article, different methods were used for

general analysis of deep learning algorithms, as

shown in Figure IV.

0304050

noitautculf deeps dniW

)eceip(tinU ata

mhtirogla gninrael peeD

sdohtem lacitsitatS

sdohtem lacisyh

INCOFT 2025 - International Conference on Futuristic Technology

298

Figure 4: The correlation between wind speed fluctuations

and wind power evaluated by deep learning algorithm

correlation analysis

From Figure 4., it can be seen that the correlation

between wind speed fluctuations and wind power in

the deep learning algorithm is significantly better than

statistical methods. The reason for this is that the deep

learning algorithm has added an adjustment

coefficient for the typical fluctuation process of wind

farm wind speed output correlation analysis, and set

a threshold for the typical fluctuation process to

eliminate the association analysis evaluation scheme

that does not meet the requirements.

5 CONCLUSIONS

In response to the problem of poor correlation

between wind speed fluctuations and wind power

fluctuations in the typical fluctuation process of wind

farm wind speed output, this paper proposes a deep

learning algorithm and combines it with deep learning

theory to optimize the correlation analysis of wind

farm wind speed output typical fluctuation process.

At the same time, conduct in-depth analysis on

correlation analysis evaluation innovation and

threshold innovation, and construct a set of typical

fluctuation processes. Research has shown that deep

learning algorithms can improve the accuracy and

stability of wind farm wind speed output typical

fluctuation process correlation analysis, and can be

used for general correlation analysis evaluation of

wind farm wind speed output typical fluctuation

process correlation analysis. However, in the process

of deep learning algorithms, excessive emphasis is

placed on the analysis of association analysis

evaluation, resulting in unreasonable selection of

association analysis evaluation indicators

ACKNOWLEDGEMENTS

Excellent Young Talents Fund Program of Higher

Education Institutions of Anhui Province ：

Forcasting and Optimization of wind power in

microgrid based on Deep learning (gxyq2022081)，

Subject Construction Promotion project of Chaohu

University（kj21gczx02、kj21bskc06）

REFERENCES

Supraja, D. P., Salameh, A. A. ., H R, D. V. ., Anand, D.

M., & Priyadi, U. . (2022). An Optimal Routing

Protocol Using a Multiverse Optimizer Algorithm for

Wireless Mesh Network. International Journal of

Communication Networks and Information

Security(IJCNIS), 14(3), 36–46.

M, R. ., Priyadi, U. ., Salameh, A. A. ., Imron, M. A. ., &

Kishore, K. H. . (2022). Cloud Computing Based

Computing System for Women’s Higher Education in

Isolated Areas. International Journal of Communication

Networks and Information Security (IJCNIS), 14(3),

26–35.

Yadav, A. K. ., Sharma, M. B. ., Bhagat, A. K. ., Shah, D.

H. ., C R, M. ., & Awasthi, M. A. . (2023). Edge

Computing in Centralized Data Server Deployment for

Network Qos and Latency Improvement for

Virtualization Environment. International Journal of

Communication Networks and Information Security

(IJCNIS), 14(3), 214–225.

https://doi.org/10.17762/ijcnis.v14i3.5607 (Original

work published December 31, 2022).

Bhatt, R. ., Shikka, M. R. ., C R, M. ., Sharma, . D. S. .,

Pandey, D. A. K. ., & Bala, K. . (2022). Centralized

Cloud Service Providers in Improving Resource

Allocation and Data Integrity by 4G IoT Paradigm.

International Journal of Communication Networks and

Information Security (IJCNIS), 14(3), 138–149.

Bernardo Tello, A. ., Xing, J. ., Lalitkumar Patil, D. A. .,

Premchandra Patil, D. L. ., & Sayyad, D. S. . (2022).

Blockchain Technologies in Healthcare System for

Real Time Applications Using IoT and Deep Learning

Techniques. International Journal of Communication

Networks and Information Security (IJCNIS), 14(3),

257–268.

2018161412108642

100

Sample quantity(piece)

relevance(%)

Physical methods

Statistical methods

Deep learning algorithm

Correlation Analysis of Wind Farm Wind Speed Output Typical Fluctuation Process Based on Deep Learning Algorithm

299