Embedded Mechanical Engineering Controller Design Using LSTM Technology

Abstract

Embedded controllers are very competitive in the market and can be seen worldwide in a wide range of applications from office automation equipment, consumer electronics, telecommunication, smart instrumentation to automotive electronics, financial electronics, industrial control and other different fields. The embedded construction machinery controller designed in this paper is mainly used in automotive electronics and construction machinery, and the design process mainly defines the functions of the controller according to the structural system of construction machinery and the functional module requirements of the control. In this paper, a fault time series prediction method based on long short-term memory (LSTM) recurrent neural network is proposed from the data, including network structure design, network training and prediction process implementation algorithm, etc. Further, with the goal of minimizing the prediction error, a multi-layer grid search-based LSTM prediction model parameter preference algorithm is proposed, and through experiments with a variety of typical time series prediction models The proposed LSTM prediction model and its parameter selection algorithm are verified to be highly applicable and accurate in the implementation of the embedded controller.