Hydroelectric generating set temperature prediction method based on time domain convolution and recurrent neural network

Abstract

The invention discloses a hydroelectric generating set temperature prediction method based on a time domain convolution and recurrent neural network. The method comprises the following steps: carrying out the preprocessing of the data of a sample, dividing a data set into a training set and a test set, constructing a time domain convolution neural network comprising a time domain convolution layer and a maximum pooling layer, and inputting a training data set into the network, firstly, carrying out effective information extraction on large-scale time sequence data of the hydroelectric generating set and reducing the data scale by introducing a sliding window mechanism and a maximum pooling structure, and then carrying out time sequence prediction by utilizing an RNN neural network. According to the invention, the training speed and precision are improved, and the prediction accuracy of large-scale time series data is improved.

Description

technical field [0001] The invention relates to the fields of safety detection and temperature prediction of hydropower stations, in particular to a method for predicting the temperature of hydropower units based on time-domain convolution and cyclic neural networks. Background technique [0002] The structure of the water wheel motor and the generator of the hydropower station is complex, and it is a nonlinear device. The bearing pad and stator coil of the hydroelectric unit are the key equipment of the whole unit. If the temperature of the bearing pad and stator coil of the hydroelectric unit fluctuates abnormally due to some reasons, the temperature rises rapidly, resulting in the burning of the bearing pad surface and the coil, which will affect the entire unit. The normal operation and power generation safety, and even forced shutdown. Therefore, it is an important task for the safe and stable operation of hydropower stations to carry out safety detection and temperatu...

AI Technical Summary

Problems solved by technology

[0006] At present, the more common deep learning models are the cyclic neural network in the deep network, etc., but due to the complexity of the data of the hydropower unit, it is still an unsolved problem.

Images