Distributed Photovoltaic Power Output Anomaly Detection Method Based on FCM-BOA-TCN-GRU
Peng Yu1, Fu Chen1, Guo Xin2, Huang Shoudao3, Su Sheng1
1. School of Electrical and Information Engineering Changsha University of Science and Technology Changsha 410114 China; 2. School of Intelligent Manufacturing Hunan First Normal University Changsha 410205 China; 3. College of Electrical and Information Engineering Hunan University Changsha 410082 China
Abstract：To address the challenges that the anomaly detection technology for centralized photovoltaic systems is difficult to apply in distributed photovoltaic systems due to the widespread and diverse nature of distributed photovoltaic points, this paper proposes a distributed photovoltaic power generation anomaly detection method based on the fuzzy C-means (FCM) clustering algorithm and Bayesian optimization algorithm (BOA) optimized TCN-GRU network. The method involves feature selection, the FCM-Frechet model for clustering similar days, and the BOA-TCN-GRU model for predicting normal photovoltaic power output, combined with a dual dynamic threshold method to determine anomalies in distributed photovoltaic power generation systems. This approach reduces the need for annotated datasets and improves the accuracy of photovoltaic system anomaly detection. Firstly, the original data is treated for outliers and Pearson correlation analysis is conducted to select current, horizontal total radiation, temperature, humidity, horizontal scattered radiation, and wind speed as inputs to construct a similar day clustering model. Secondly, to mitigate the impact of weather volatility on photovoltaic power prediction results, a weighted FCM-Frechet algorithm is proposed for two-stage similar day clustering, categorizing weather into clear, cloudy, and rainy similar days. Then, using mutual information to screen redundant features, a BOA-optimized TCN-GRU network model is proposed. TCN processes local and global spatial features of time series, while GRU handles temporal features, achieving spatiotemporal joint modeling of features and improving the prediction accuracy of normal photovoltaic power output. Finally, the dual dynamic threshold method with an adaptive factor is used to determine photovoltaic system anomalies. The effectiveness of the proposed method is verified using the Alice Springs dataset from the Australian desert area and a building photovoltaic power generation dataset from Changsha City. The results show that the proposed method achieves higher detection accuracy. This paper implements two simulations using Pytorch. The first simulation aims to predict normal photovoltaic power output and detect anomalies under different similar days using the proposed BOA-TCN-GRU model. The results indicate that under cloudy conditions, the proposed method reduces RMSE and MAE by 0.223 9 kW and 0.170 6 kW, respectively, compared to FCM-CNN-LSTM-Attention; under rainy conditions, RMSE and MAE are reduced by 0.385 7 kW and 0.398 9 kW, respectively. The proposed method demonstrates better weather generalizability in normal photovoltaic power prediction tasks. On clear days, the FPR is reduced by 40.38% and 33.71% compared to CNN-LSTM and Transformer-BiLSTM, respectively, with an accuracy increase of 11.24 and 3.92 percentage points. The second simulation aims to verify the generalization ability of the proposed anomaly detection model. The results show that the accuracy rates of CNN-LSTM and Transformer-BiLSTM are 81.32% and 85.43%, respectively, and the proposed method can still detect photovoltaic anomalies on other datasets, with an accuracy rate reaching 90%. From the simulation analysis, the following conclusions can be drawn: (1) The weighted FCM-Frechet clustering model effectively reduces the impact of weather volatility on photovoltaic prediction results by clustering similar days. (2) Considering environmental conditions, the BOA-optimized TCN-GRU network model, combined with the dual dynamic threshold method with an adaptive factor, improves the accuracy and versatility of the model in photovoltaic anomaly detection. (3) Experimental results show that the proposed model improves accuracy by 11.24 and 3.92 percentage points compared to CNN-LSTM and Transformer-BiLSTM. This paper identifies an issue with the inaccuracy of the model between moving clouds and photovoltaic output under cloudy weather conditions. The next step will involve researching high-precision spatiotemporal modeling methods for the relationship between cloud movement and photovoltaic output in cloudy weather and proposing detection methods for identifying photovoltaic anomalies under this model.
[1] 王小宇, 刘波, 孙凯, 等. 光伏阵列故障诊断技术综述[J]. 电工技术学报, 2024, 39(20): 6526-6543. Wang Xiaoyu, Liu Bo, Sun Kai, et al.A review of photovoltaic array fault diagnosis technology[J]. Transactions of China Electrotechnical Society, 2024, 39(20): 6526-6543. [2] 王玉庆, 徐飞, 刘志坚, 等. 基于动态关联表征与图网络建模的分布式光伏超短期功率预测[J]. 电力系统自动化, 2023, 47(20): 72-82. Wang Yuqing, Xu Fei, Liu Zhijian, et al.Ultra-short-term power forecasting of distributed photo-voltaic based on dynamic correlation characterization and graph network modeling[J]. Automation of Electric Power Systems, 2023, 47(20): 72-82. [3] 赵耀, 高少炜, 李东东, 等. 基于天气相似聚类与QRNN的短期光伏功率区间概率预测[J]. 电力系统自动化, 2023, 47(23): 152-161. Zhao Yao, Gao Shaowei, Li Dongdong, et al.Short-term interval probability prediction of photovoltaic power based on weather similarity clustering and quantile regression neural network[J]. Automation of Electric Power Systems, 2023, 47(23): 152-161. [4] 时珉, 许可, 王珏, 等. 基于灰色关联分析和GeoMAN模型的光伏发电功率短期预测[J]. 电工技术学报, 2021, 36(11): 2298-2305. Shi Min, Xu Ke, Wang Jue, et al.Short-term photovoltaic power forecast based on grey relational analysis and GeoMAN model[J]. Transactions of China Electrotechnical Society, 2021, 36(11): 2298-2305. [5] Balachandran G B, Devisridhivyadharshini M, Ramachandran M E, et al.Comparative investigation of imaging techniques, pre-processing and visual fault diagnosis using artificial intelligence models for solar photovoltaic system—a comprehensive review[J]. Measurement, 2024, 232: 114683. [6] 赖昌伟, 黎静华, 陈博, 等. 光伏发电出力预测技术研究综述[J]. 电工技术学报, 2019, 34(6): 1201-1217. Lai Changwei, Li Jinghua, Chen Bo, et al.Review of photovoltaic power output prediction technology[J]. Transactions of China Electrotechnical Society, 2019, 34(6): 1201-1217. [7] Qu Jiaqi, Qian Zheng, Pei Yan, et al.An unsupervised hourly weather status pattern recognition and blending fitting model for PV system fault dete-ction[J]. Applied Energy, 2022, 319: 119271. [8] Lu Xiaoyang, Lin Peijie, Cheng Shuying, et al.Fault diagnosis for photovoltaic array based on con-volutional neural network and electrical time series graph[J]. Energy Conversion and Management, 2019, 196: 950-965. [9] Alippi C, Ntalampiras S, Roveri M.Model-free fault detection and isolation in large-scale cyber-physical systems[J]. IEEE Transactions on Emerging Topics in Computational Intelligence, 2017, 1(1): 61-71. [10] Wang Kejun, Qi Xiaoxia, Liu Hongda.Photovoltaic power forecasting based LSTM-convolutional net-work[J]. Energy, 2019, 189: 116225. [11] Harrou F, Dairi A, Taghezouit B, et al.An unsupervised monitoring procedure for detecting anomalies in photovoltaic systems using a one-class support vector machine[J]. Solar Energy, 2019, 179: 48-58. [12] 乔楠, 蒋波涛, 郑雨, 等. 基于深度模糊神经网络的太阳总辐射预测研究[J]. 太阳能学报, 2024, 45(2): 59-64. Qiao Nan, Jiang Botao, Zheng Yu, et al.Research on global solar radiation forecast based on deep fuzzy neural network[J]. Acta Energiae Solaris Sinica, 2024, 45(2): 59-64. [13] Wang Chaofan, Shuai Jing, Ding Liping, et al.Comprehensive benefit evaluation of solar PV projects based on multi-criteria decision grey relation projection method: evidence from 5 counties in China[J]. Energy, 2022, 238: 121654. [14] 丛伟伦, 张博, 夏亚东, 等. 基于马尔可夫链的光伏电站遮挡实时诊断算法[J]. 太阳能学报, 2020, 41(4): 67-72. Cong Weilun, Zhang Bo, Xia Yadong, et al.Diagnosis algorithm for real-time shaded analysis of photo-voltaic power station based on Markov chain[J]. Acta Energiae Solaris Sinica, 2020, 41(4): 67-72. [15] Ramírez A F, Valencia C F, Cabrales S, et al.Simulation of photo-voltaic power generation using copula autoregressive models for solar irradiance and air temperature time series[J]. Renewable Energy, 2021, 175: 44-67. [16] Sahin G, Isik G, van Sark W G J H M. Predictive modeling of PV solar power plant efficiency con-sidering weather conditions: a comparative analysis of artificial neural networks and multiple linear regression[J]. Energy Reports, 2023, 10: 2837-2849. [17] Lindig S, Kaaya I, Weiss K A, et al.Review of statistical and analytical degradation models for photovoltaic modules and systems as well as related improvements[J]. IEEE Journal of Photovoltaics, 2018, 8(6): 1773-1786. [18] Dey M, Rana S P, Simmons C V, et al.Solar farm voltage anomaly detection using high-resolution μ PMU data-driven unsupervised machine learning[J]. Applied Energy, 2021, 303: 117656. [19] Xiang Ling, Yang Xin, Hu Aijun, et al.Condition monitoring and anomaly detection of wind turbine based on cascaded and bidirectional deep learning networks[J]. Applied Energy, 2022, 305: 117925. [20] 蔡雨思, 李泽文, 刘萍, 等. 基于间接健康特征优化与多模型融合的锂电池SOH-RUL联合预测[J]. 电工技术学报, 2024, 39(18): 5883-5898. Cai Yusi, Li Zewen, Liu Ping, et al.Joint prediction of lithium battery state of health and remaining useful life based on indirect health features optimization and multi-model fusion[J]. Transactions of China Electro-technical Society, 2024, 39(18): 5883-5898. [21] 杨童亮, 胡东, 唐超, 等. 基于SMA-VMD-GRU模型的变压器油中溶解气体含量预测[J]. 电工技术学报, 2023, 38(1): 117-130. Yang Tongliang, Hu Dong, Tang Chao, et al.Prediction of dissolved gas content in transformer oil based on SMA-VMD-GRU model[J]. Transactions of China Electrotechnical Society, 2023, 38(1): 117-130. [22] Yu Wennian, Kim I Y, Mechefske C.Analysis of different RNN autoencoder variants for time series classification and machine prognostics[J]. Mechanical Systems and Signal Processing, 2021, 149: 107322. [23] Ibrahim M, Alsheikh A, Awaysheh F, et al.Machine learning schemes for anomaly detection in solar power plants[J]. Energies, 2022, 15(3): 1082. [24] Chang Zhonghao, Han Te.Prognostics and health management of photovoltaic systems based on deep learning: a state-of-the-art review and future perspectives[J]. Renewable and Sustainable Energy Reviews, 2024, 205: 114861. [25] de Souza Silva J L, Mahmoudi E, Carvalho R R M, et al. Classification of anomalies in photovoltaic systems using supervised machine learning techniques and real data[J]. Energy Reports, 2024, 11: 4642-4656. [26] Liu Shujun, Xu Tong, Du Xiaoze, et al.A hybrid deep learning model based on parallel architecture TCN-LSTM with Savitzky-Golay filter for wind power prediction[J]. Energy Conversion and Management, 2024, 302: 118122. [27] Limouni T, Yaagoubi R, Bouziane K, et al.Accurate one step and multistep forecasting of very short-term PV power using LSTM-TCN model[J]. Renewable Energy, 2023, 205: 1010-1024. [28] Yang Shun, Du Yajun, Huang Jiaming, et al.Few-shot intent detection with mutual information and contrastive learning[J]. Applied Soft Computing, 2024, 167: 112338. [29] Fantini D G, Silva R N, Siqueira M B B, et al. Wind speed short-term prediction using recurrent neural network GRU model and stationary wavelet transform GRU hybrid model[J]. Energy Conversion and Management, 2024, 308: 118333.
2025, Vol. 40 
