The paper presents a novel artificial neural network (A") speed PID controller for the permanent magnet synchronous motor CpWsn;r> drive system. The ANN speed PID is a MIMO controller and operates the mal-torqudcurrent control strategy. A general computer is used as hardware platform. The simulation and experiment results verify the success of drive system with ANN speed PID controller.Keyword-Permanent Magnet Synchronous Motor, Artificial Neural Network I .INTRODUCTIONArtificial neural networks are composed of simple elements operating in parallel. As in nature, the ANN can be trained to operate a particular function by adjusting the weight values between elements. They have the abilities of self-studying and adaptation. Recently ANN is applied successfully in various fields, including pattern recognition, speech and image identification, etc.One of the typical methods in development for control PWSM speed is by using the double feedback control loops, that is, speed-loop and current-loop. The conventional speed control is often adopted PID control method. But in some cases, when the known constructs and parameters of motor motion system are changed as the outside conditions variety, for example, the air temperature, the performance of the controller will be spoiled.The paper makes a research in application ANN in motor vector control for resolving this kind of problem in the motor drive system. The general microcomputer is used as control unit, which reduces the computing time of neural network and makes it possible to realize the A " speed PID algorithm in the real time control of PMSM driver. In the paper, the neural network PID model is described in second section. In third section the controller platform is described. At last, simulation and experiment results are given, and conclusions verified the PMSM driver possesses favorable performance. II. MODEL of ANN PIDThe learning algorithm of multi-layer feedforward ANN is the error backpropagation algorithm that was created by generalizing the Widrow-Hoff learning rule to multiple-layer networks and nonlinear differentiable transfer functions. Backpropagation refers to the manner in which the gradient is computed for nonlinear multi-layer networks. This feedforward A " is called BP network. Basing on the .digital PID controller, the common ANN PJD model is derived as follow in Fig 1 . ul (k-110, / u2 (k-1) 0 Fig. 1. Conventional'neural network PID modelBecause the two-layer ANN PID controller has no hidden layer, which is poor in the generalization, and is ditliculty to approximate the whole speed transient procedure, a novel multi input multi output W O ) neural network PID controller is proposed in the paper, showed in Fig.2. The 3-laylers neural network with input node number n=4,hidden node number 1=3,0utput node number m=2.Networks as shown in Fig2 with biases, a sigmoid layer, and a linear output layer are capable of approximating any function with a finite number of Fig.2. MtMO neural network PID model -679 -
Six-step commutation control widely used in brushless DC (BLDC) motor can be applied to consequent pole permanent magnet (CPPM) belt starter generator (BSG) with trapezoidal back electromotive force (EMF) in the starter state. However, rotor position detection with three Hall sensors in BLDC motor can hardly be employed in CPPM BSG due to asymmetric flux distribution in each pole side of CPPM BSG. This paper presents a low-cost rotor position detection method for CPPM BSG in which six Hall sensors are proposed to be used based on the analysis of flux distribution by 3D FEA. In the method, the six Hall sensors are divided into three groups and two signals in each group are combined through performing logic operations. In addition, offset angle between back EMF and the related Hall signal can be compensated by moving the Hall sensors. Experiments of a 2 kW CPPM BSG prototype have also been performed to verify the proposed method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
334 Leonard St
Brooklyn, NY 11211
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.