عنوان مقاله
شبکه عصبی آداپتیو تناسبی-انتگرالی-مشتقی برای کنترل سیستم غیرخطی پیچیده
فهرست مطالب
مقدمه
شبکه های عصبی آداپتیو تناسبی-انتگرالی-مشتقی
تجزیه و تحلیل پایداری
شبیه سازی
نتیجه گیری
بخشی از مقاله
ساختار سیستم کنترلی
سیستم کنترلی یک کنترل حلقه بسته اتخاذ می کند، و عمدتا شامل 2 قسمت می باشد: کنترلر و سیستم کنترل شده، کنترلر بر اساس شبکه عصبی آداپتیو تناسبی - انتگرالی - مشتقی ساخته شده است. در تمام سیستم کنترلی، X بردار هدف، E بردار خطا، Y مقدار خروجی سیستم کنترلی وU قانون کنترلی سیستم کنترل می باشد.
کلمات کلیدی:
An adaptive PID neural network for complex nonlinear system control Jun Kang a,b , Wenjun Meng a , Ajith Abraham c,d,e , Hongbo Liu c,e,n a School of Mechanical Engineering, Taiyuan University of Science and Technology, Taiyuan 030024, China b Software School, North University of China, Taiyuan 030051, China c Machine Intelligence Research Labs (MIR Labs), Scientific Network for Innovation and Research Excellence, Auburn 98071, WA, USA d IT4Innovations - Center of excellence, VSB - Technical University of Ostrava, Ostrava 70833, Czech Republic e School of Information Science and Technology, Dalian Maritime University, Dalian 116026, China article info Article history: Received 18 October 2012 Received in revised form 20 February 2013 Accepted 8 March 2013 Available online 4 January 2014 Keywords: Complex nonlinear system Adaptive PID neural network PSO Gradient descent abstract Usually it is difficult to solve the control problem of a complex nonlinear system. In this paper, we present an effective control method based on adaptive PID neural network and particle swarm optimization (PSO) algorithm. PSO algorithm is introduced to initialize the neural network for improving the convergent speed and preventing weights trapping into local optima. To adapt the initially uncertain and varying parameters in the control system, we introduce an improved gradient descent method to adjust the network parameters. The stability of our controller is analyzed according to the Lyapunov method. The simulation of complex nonlinear multiple-input and multiple-output (MIMO) system is presented with strong coupling. Empirical results illustrate that the proposed controller can obtain good precision with shorter time compared with the other considered methods. & 2014 Elsevier B.V. All rights reserved. 1. Introduction In the industrial field, the controlled system usually has great nonlinearity, including spacecraft system, vehicle system, robot system, power system, chemical reaction system, and so on. It is hard to get a precise control performance even by the intelligent control methods, including adaptive control [1,2], fuzzy control [3–5], neural network control [6–8] and decoupling control [9–11]. So many mixed control methods are presented, such as PID neural network. Due to the characteristics of self-learning, self-organizing and self-adaptation, PID neural network would automatically identify the parameters of controlled system and adjust them according to system changes. In this paper, we present a controller model based on adaptive PID neural networks. To prevent the weights of neural networks falling into local optima, PSO algorithm is adopted to select initial weights. The parameters of PID neural network are self-regulating without intervention. The improved gradient descent method is used to optimize the weights of networks. 2. Related works Since it is difficult to control a complex nonlinear system [12–14], neural network was introduced to solve the problems. In [15], Jafarnejadsani et al. proposed an adaptive control based on radialbasis-function neural network (NN) for different operation modes of variable-speed variable-pitch wind turbines. In [16], Lin et al. presented an interactively recurrent self-evolving fuzzy neural network to predict and identify the dynamic systems. They derived the consequent update parameters by a variable-dimensional Kalman filter algorithm. In [17], Chemachema introduced a direct adaptive control algorithm based on neural networks for a class of single input single output nonlinear systems. These signals involved in the closed loop were proven to be exponentially bounded and hence the system stability, without any additional control term to the NN adaptive controller. However, researches still confront some difficulties. For example, network parameter training is time-consuming and easily falls into local minimum. Particle swarm optimization (PSO) algorithm is a new globe optimization algorithm, which has the advantage of fast convergence speed [18,19]. In [20], Selvakumaran et al. proposed a new design of decentralized load-frequency controller for interconnected power systems with ac–dc parallel using PSO algorithm. The experiment result illustrated that their method have rapid dynamic response ability. In [21], Hasni et al. used PSO algorithm to parameters selection, and used genetic algorithm to optimize the choice of parameters by minimizing a cost function. The study was applied to a greenhouse environment with Continuous Roof Vents, and obtained satisfactory effect. Nevertheless, it is difficult to apply directly these methods to complex nonlinear system with strong coupling. Adaptive controller has the ability to adjust of control parameters without the help of human intelligence. It can tune complex systems better by combining nonlinear controlling methods and intelligent control technology [22,23]. The results show that adaptive control has the advantage to solve effectively the
