Fault Diagnosis of Gas Turbine Actuator Based on Improved Convolutional Neural Network

As a kind of internal combustion engine, gas turbine has become the core of industrial technology in the 21st century because of its excellent technical characteristics such as low pollution, low noise, high efficiency and high reliability. Aiming at the common faults of the actuator of gas turbine control system, a modified One dimensional convolution neural network (IDCNN) was used to construct the fault diagnosis model of the actuator. Considering that Gabor filter has good resolution in both time domain and frequency domain, a one-dimensional convolutional neural network (G-IDCNN) fault diagnosis model based on Gabor filter is proposed. Gabor filter is used to enhance the feature extraction ability of IDCNN and improve the model operation speed without affecting the accuracy of fault diagnosis. The experimental results show that the feature extraction ability of the network is greatly enhanced when the software and hardware conditions and training parameters are kept the same, and the high fault diagnosis accuracy (99.26% and 99.52% under the valve opening condition of 50% and 100%) can be achieved while the network depth is reduced, which has a good fault diagnosis effect.