Multi-Unit Wind Power Prediction Based on LSTM and PSO

Multi-Unit Wind Power Prediction Based on LSTM and Particle Swarm Algorithm

Wu Zhenlong1, Mo Yipeng1, Wang Ronghua2, Fan Xinyu1, Liu Yanhong1, Guo Xiaolian3

1. Zhengzhou University, School of Electrical and Information Engineering; 2. Shandong Labor Vocational Technical College; 3. Zhejiang Special Equipment Science Research Institute

Click “Read the Original” at the end of the article to view the literature!

Contents

Multi-Unit Wind Power Prediction Based on LSTM and PSO

01

Abstract

Multi-Unit Wind Power Prediction Based on LSTM and PSO

02

Chart Appreciation

Multi-Unit Wind Power Prediction Based on LSTM and PSO

03

Cite This Article

Multi-Unit Wind Power Prediction Based on LSTM and PSO

04

Author Introduction

Abstract

Wind energy is favored globally due to its widespread distribution, pollution-free, and renewable characteristics. However, the scheduling of wind energy is affected by various factors, including wind speed, atmospheric pressure, terrain, and altitude, which can cause significant disturbances to wind power output, leading to intermittency and instability. Therefore, predicting wind power output and minimizing uncertainty is necessary. Based on this, this paper proposes a multi-unit wind power prediction method based on Long Short-Term Memory (LSTM) networks. First, the Spearman correlation coefficient method is used for quantitative analysis of the data, followed by principal component analysis to reduce dimensionality and extract key information. The LSTM is used to predict wind power output in a single-step rolling manner, predicting only the next time step’s power value and adjusting inputs through a sliding window. To address the difficulty in tuning LSTM parameters, Particle Swarm Optimization (PSO) is used to optimize LSTM hyperparameters with the root mean square error as the optimization objective. For the multi-unit wind power prediction problem, this paper starts with the single unit, identifies the best-performing model in the single unit, and then extends the single unit wind power prediction to multi-units. Experimental results show that compared to other models, the proposed method reduces the root mean square error by 11.8% and the mean absolute error by 5.03%. This indicates that the LSTM optimized by PSO can achieve higher accuracy and better generalization ability.

Chart Appreciation

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 1: LSTM Internal Structure Diagram

Figure 1: LSTM internal structure diagram

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 2: Single-Step Rolling Prediction Method

Figure 2: Single-step rolling prediction method

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 3: Particle Swarm Algorithm Flow Chart

Figure 3: Particle swarm algorithm flow chart

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 4: PSO Iterative Process

Figure 4: PSO iterative process

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 5: PSO-LSTM Prediction Results of Unit 1

Figure 5: PSO-LSTM prediction results of unit 1

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 6: Principal Component Cumulative Variance Contribution Rate

Figure 6: Principal component cumulative variance contribution rate

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 7: Multi-Unit Wind Power Prediction Flow Chart

Figure 7: Multi-unit wind power prediction flow chart

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Figure 8: Multi-Unit PSO-LSTM Prediction Results

Figure 8: Multi-unit PSO-LSTM prediction results

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Cite This Article

Wu Zhenlong, Mo Yipeng, Wang Ronghua, et al. Multi-Unit Wind Power Prediction Based on Long Short-Term Memory and Particle Swarm Optimization[J]. Journal of Zhengzhou University (Engineering Science), 2024, 45(6):114-121.

WU Z L, MO Y P, WANG R H, et al. Multi-unit wind power prediction based on long short-term memory and particle swarm optimization[J]. Journal of Zhengzhou University (Engineering Science), 2024, 45(6):114-121.

Author Introduction

1 Wu Zhenlong

Associate Professor, School of Electrical and Information Engineering, Zhengzhou University, PhD, Master’s Supervisor.

Main Research and Teaching Experience: Obtained PhD from Tsinghua University in 2020, visiting scholar at the University of California, Merced, and Southern University of Science and Technology. Engaged in research and teaching in control theory and its application in energy systems, new energy power prediction, and consumption. Member of the Self-Resilient Control Professional Committee of the Chinese Society of Command and Control, member of the Chinese Association of Automation and the Chinese Society of Command and Control.

Research Achievements: Published over 40 academic papers in international journals, conferences, and domestic core journals, representative papers include: “Performance Analysis of Improved ADRCs for a Class of High-Order Processes with Verification on Main Steam Pressure Control”, “Modified Active Disturbance Rejection Control Design Based on Gain Scheduling for Selective Catalytic Reduction Denitrification Processes”, “Economic Operation of Islanded Micro-Grids via Modified Active Disturbance Rejection Control”, etc.; published 2 monographs; applied for over 3090 patents; received the second prize for scientific and technological progress in Henan Province, and multiple excellent paper awards at conferences. Led and completed projects such as the National Natural Science Foundation project, Henan Province Science and Technology Key Project, National Key Laboratory Open Fund of Power Systems, Zhengzhou University Professor Team Support for Enterprise Innovation-Driven Development Project, and Zhengzhou University Youth Talent Enterprise Cooperation Innovation Team Support Program, with multiple achievements applied in various types of units including 1000MW, 600MW, and 300MW.

Recruitment Direction and Application Requirements: Recruiting master’s students in Control Engineering.

Contact Information:[email protected]

About Us
We
Are
Here
“Journal of Zhengzhou University (Engineering Science)” was founded in 1980, originally named “Journal of Zhengzhou Institute of Technology” and “Journal of Zhengzhou University of Industrial Technology”, is a comprehensive academic journal published by Zhengzhou University, issued both domestically and internationally, bimonthly. This journal has always been a core journal in the Chinese Academy of Sciences, a core journal in Chinese science and technology, and an excellent journal in the national university journal category. It mainly publishes academic papers in engineering disciplines such as computer science, mechanical engineering, power engineering, control science, hydraulics, electrical engineering, civil engineering, transportation, environmental science, communication, chemical engineering, materials science, as well as interdisciplinary fields. It has won multiple national awards for excellent university journals, first prize for excellent scientific and technological journals in national universities, execution excellence journal award from the Chinese Academic Journal (CD), demonstration case of excellent team in the construction of scientific and technological journals in Chinese universities, first prize for excellent scientific and technological journals in Henan Province, excellent natural science journal in Henan Province, and first-level journal in Henan Province.
This journal has been included by 8 well-known foreign abstracting and indexing services, including American Chemical Abstracts (CA), Russian Abstracts Journal (VINITIRAN), British Science Abstracts (INSPEC), American Cambridge Scientific Abstracts (CSA), American Mathematical Reviews (MathScinNeT), Polish Copernicus Index (IC), American Ulrich’s Periodicals Directory (UPD), and American EBSCO, as well as domestic core journals in the Chinese Core Journals Overview, Chinese Science and Technology Core Journals, and various domestic abstracting and indexing services.

Contact Information for Journal of Zhengzhou University (Engineering Science):

Submission Website:http://gxb.zzu.edu.cn

WeChat Official Account: zdxbgxb

Contact Email: [email protected]

Contact Method: 0371-67781276

Contact Method: 0371-67781277

For more exciting content, please follow us

Multi-Unit Wind Power Prediction Based on LSTM and PSO

Copyright Statement:

This article is original content from the editorial department of the “Journal of Zhengzhou University (Engineering Science)”, welcome to reprint!

Click here to “Read the Original” for journal content

Leave a Comment