High-Precision Prediction and Analysis Method of Power Line Noise Characteristics Based on Aggregated Frequency-Domain Neural Network

With the continuous expansion of the power grid, massive power electronic devices are deployed in the low-voltage station area, and the noise interference is particularly complex. The power line noise characteristic analysis is essential to improve the communication performance. This paper firstly proposes a power line noise-device matching method based on the database prior information to realize the matching of noise data and power electronic devices. On this basis, a power line noise analysis method based on the aggregated frequency domain neural network is proposed. The high-precision prediction and analysis of noise characteristics is achieved to updated the device matching degree weights based on the prediction error and update the database in real time based on the noise analysis results. Finally, the feasibility and effectiveness of the proposed method are verified by simulations.