Long Short Term Memory Based Status Prediction for Diesel Generator Set

The diesel generator set is a kind of equipment which uses the diesel engine as the prime mover to drive the synchronous generator to generate electricity. Diesel generator sets play an important role in production and emergency power supply. To detect and supervise the operation status and running states of diesel generating sets, the prediction of the operation status of diesel generating sets is becoming more and more important. Long short term memory (LSTM) neural network can control the memory and forgetting degree of the previous data and the current data in the time series data by forgetting gates, input gates, output gates and so on, so that the neural network has the function of long-term memory. LSTM model and prediction method for diesel generator set is proposed. Oil inlet temperature data are collected with sensor network from diesel generator set, and the data are used for status prediction. Compared with support vector regression, RNN and gated recurrent unit, proposed LSTM model higher prediction accuracy, and it can be used for status prediction for diesel generator set.