一种基于单神经元的模糊自整定PID控制器

A Single Neuron Based Fuzzy Self-Tuning PID Controller

为解决普通的PID控制器设计对系统模型有一定依赖性问题,本文将单神经元网络和模糊逻辑相结合,提出了一种基于单神经元的模糊自整定PID控制器。单神经元控制部分本身还是一个PID控制器,但由于权值可以在线调整,具有较强的自学习和自适应能力,而通过模糊控制器对输出整定,可加速控制过程。将该控制器应用于船舶运动控制仿真中,仿真结果表明,单神经元模糊自整定PID控制器能使船舶更快地达到设定航向,且航向曲线和打舵更加平稳,可明显改善对大纯滞后、大惯性系统的控制效果,在工业控制中有广泛的应用前景。

In order to solve the problem that parameter design of PID controller commonly depends on the model of the system, a single neuron based fuzzy adaptive PID controller （SNFSPID） is proposed in this paper by the combination of single neuron network and fuzzy logic. The single neuron part of the controller is, in fact, a PID controller with links adjustable online, which has the ability of self-learning and self-adaptability, and the control process can be accelerated by tuning the output using fuzzy logic. The controller is applied to ship motion control system. Simulations showed that the proposed controller has the ability of making the ship reache its course more quickly, together with smooth course curve and helm operation, which can obviously improve the control effect for large pure time-delay and large inertia system, so has extensive application in industry.