Construction of a Distribution Network Line Loss Rate Prediction Model Integrating LSTM and Particle Swarm Optimization
DOI:
https://doi.org/10.54097/4e1swq04Keywords:
Distribution Network Line Loss Rate; Long Short-term Memory Network; Particle Swarm Optimization; Prediction Model; Parameter Optimization.Abstract
Distribution network line loss rate prediction is a core component of power system economic operation and safety management. However, traditional prediction methods are prone to lag effects when dealing with non-stationary line loss series. Furthermore, the hyperparameters of a single long short-term memory (LSTM) network rely on empirical selection, resulting in insufficient prediction accuracy and stability, making it difficult to meet the requirements of refined modern power grid management. This study aims to construct an LSTM distribution network line loss rate prediction model optimized by a particle swarm optimization (PSO). Leveraging PSO's global optimization capabilities, the LSTM hyperparameter configuration challenge is addressed, improving the model's prediction performance. Research demonstrates that the PSO-optimized LSTM model significantly improves prediction accuracy and training efficiency compared to traditional methods. Experimental results show that the PSO-LSTM improves the mean absolute percentage error (MAPE) and root mean square error (RMSE) by 32.5% and approximately 56%, respectively, compared to the LSTM model, demonstrating the effectiveness of the particle swarm optimization algorithm for neural network parameter optimization. This model effectively reduces the uncertainty of manual parameter adjustment and can accurately adapt to the characteristics of complex distribution network line loss data. It provides reliable technical support for power supply companies to reduce costs and increase efficiency, as well as for grid operation status monitoring and loss control. It is of great significance to the development of intelligent distribution network line loss prediction technology.
References
[1] Ma Qian, Guo Jiangtao, Yang Xia, Dilinia Dilixiati, Zhao Gang, Chen Tianyu. Line loss prediction method based on multidimensional information matrix and DAM-LSTNet [J]. Power System Technology, 2024.
[2] Luo Wei, Chen Yuan. Vibration Fault Diagnosis Method for Hydropower Units Based on PSO-LSTM Network [J]. Mechanical Design and Manufacturing Engineering, 2024.
[3] Li Shuitian, Huang Xueyou, Jiang Jing, Liang Qianyi, Jiao Xianan. Distribution network equipment failure rate prediction method based on improved LSTM algorithm [J]. Hebei Electric Power Technology, 2024.
[4] Wang Zhengguo, Guan Jiewen, Zhuang Xiaoning. Power load forecasting model based on long short-term memory neural network based on particle swarm optimization [J]. Electronic Products World, 2024.
[5] Wu Zhenlong, Mo Yipeng, Wang Ronghua, Fan Xinyu, Liu Yanhong, Guo Xiaolian. Multi-unit wind power forecasting based on LSTM and particle swarm algorithm [J]. Journal of Zhengzhou University (Engineering Edition), 2024.
[6] Chen Zaixin, Qi Guangye, Lu Jianxun. Calculation Method of Distribution Network Line Loss Rate Based on Artificial Intelligence Algorithm[J]. Automation and Instrumentation, 2024.
[7] Wang Shouxiang, Zhang Bingjie, Zhao Qianyu, Guo Luyang, Zhang Sheng. Two-layer Estimation Model of Feeder Statistical Line Loss Rate Based on Ensemble Tree and MoE[J]. Transactions of the Chinese Society of Electrotechnical Engineering, 2024.
[8] Wang Qi, Chai Yuhuan, Wang Pengcheng, Liu Baichuan, Liu Xiang. Prediction of SO2 Emission Mass Concentration Based on Improved INFO-Bi-LSTM Model[J]. Journal of Power Engineering, 2024.
[9] Qin Yinping, Wang Jinpeng. Analysis of Distribution Network Line Loss Calculation Method Based on Machine Learning[J]. Electronic Technology, 2024.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Xinkai Hu

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.







