Research on Photovoltaic Inverter Temperature Prediction Method Based on CNN-LSTM

• Published in: 2024 9th International Conference on Intelligent Computing and Signal Processing (ICSP) pp. 1647 - 1651
• Main Authors: Dong, Haifeng, Liu, Hao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19.04.2024
• Subjects: Accuracy; CNN; Convolutional neural networks; Data models; Deep learning; Inverters; Long short term memory; LSTM; Neural networks; Photovoltaic inverter; Photovoltaic systems; Predictive models; Temperature distribution; Temperature prediction; Time series analysis
• DOI: 10.1109/ICSP62122.2024.10743359

This paper comes from the "Photovoltaic Power Generation Intelligent Management Cloud Platform" project in a domestic oil field that the author participated in. After analyzing various factors that affect photovoltaic inverter temperature prediction, this paper uses a method combining convolutional neural network and long short-term memory network to predict photovoltaic inverter temperature. This model can obtain the spatial relationship between each feature value of the data, thereby overcoming the inability of the LSTM (Long Short-Term Memory Network) layer to capture the spatial components of the data, and the features extracted are still sequential. The research results show that the hybrid neural network combining CNN and LSTM exhibits high accuracy and stability in predicting the temperature of photovoltaic inverters. By calculating the absolute error (MAE) of the two models, the performance of the CNN-LSTM hybrid neural network is improved by about 18.57% compared to the LSTM neural network model.