Research of an EPB shield pressure and depth prediction model based on deep neural network and its control device

Based on the construction data of Fuzhou Metro Line 4 in Fujian Province, China, this paper proposes a soil pressure prediction model that combines Long Short-Term Memory (LSTM) and Particle Swarm Optimization (PSO). The values of Mean Absolute Error, Mean Squared Error, and Coefficient of Determination are 0.007MPa, 0.007%, and 0.93, respectively, indicating an improvement in accuracy.Wang-Mendel algorithm is used to establish fuzzy rules. The Mean Absolute Error and Mean Squared Error of the rotating speed of the screw machine are 0.065rpm and 1.528%, and the Coefficient of Determination is 0.82. The calculation accuracy of this algorithm is high.A set of knob intelligent control device is developed.The Mean Absolute Error and Mean Squared Error of 0.015rpm and 0.392%, respectively, and the Coefficient of Determination of 0.95, indicating a small execution error of the device. This paper provides a new and effective method for the control of EPB shield pressure.