Wearing condition trend prediction of turbine machine based on multi-parameter joint monitoring of turbine oil

It is critical for the well running of electricity factory to monitor and predict the statues of turbine oil used by turbine generator. However, there are a lot of monitoring parameters for turbine oil and the statues trend is hard to predict. So the principle component analysis method was used to research 7 groups and monitor data for 10 months. The two principle components obtained from analysis are able to reflect the trend for the synthetic statues variation of the turbine oil within this period. The fact is that the actual data is less. In order to use more data to predict, the three times linear interpolation method is used to increase the data volume so that the data obtained from the curve is smoother. Then we forecast data analysis with dynamic neural network. The analysis and discussion to the experiment result shows that the PCA method can get a useful synthetic information from a couple of groups of monitoring data for the turbine oil, so that the trend of oil getting bad and the interval time or period of oil changing can be predicted according to the information.