Modified Discrete Neural Network PID Controller for Controlling the Helicopters Turboshaft Engines Free Turbine Speed

The focus of this study is the enhancement of the neuro-PID controller for regulating the free turbine speed in helicopters turboshaft engines. This is achieved through the integration of a nonlinear PID controller, implemented using a dynamic neural network with direct data transmission. The neural network comprises neurons with a radial-basis activation function in the first layer and adalines – neurons with a linear function activation – in the second layer. The optimization of PID controller coefficients is accomplished by transitioning to a discrete representation of the PID controller. This reduced the transient time from 3.0 seconds to 0.75 seconds, and reduced the overshoot from 5.0 % to 0.5 %.

I. Introduction

The predominant type of industrial controller in use today is the PID controller, with approximately 90% of controllers in commercial operation employing the PID algorithm. This widespread adoption can be attributed to the controller's straightforward construction, suitability for industrial applications, operational transparency, versatility in addressing practical issues, and cost-effectiveness [1], [2]. However, existing methods for computing PID controller parameters are tailored to linear systems, as the controller itself is considered a linear dynamic component in such cases. In instances where the control object is fundamentally nonlinear, achieving satisfactory control quality becomes challenging. When tasked with regulating complex nonlinear systems, such as helicopters turboshaft engines (TE), PID controllers may fall short in delivering the requisite control precision for the parameters of helicopters TE, system stability, and robustness under varying operational conditions and malfunctions. Thus, the modification of the PID controller by using neural network technologies to control the helicopters TE is an urgent scientific and practical task.