Font Size: a A A

The Research On New Internal Model Control Methods For Complex System

Posted on:2016-11-27Degree:DoctorType:Dissertation
Country:ChinaCandidate:B Y JiangFull Text:PDF
GTID:1108330473462557Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
As a new type of control strategy in the advanced control, Internal model control (IMC) has been widely used in complex multivariable control of industrial process for its advantage of simple design principle, easy to adjust, and strong robustness, etc. However, the current research of internal model control for multivariable system mainly targets to full rank square system or rectangular system. When the controlled object is rank-defect system or generalized system, how to adopt IMC method to study control problems of these unconventional system are the main research contents of this paper, include:1. An internal model control method based on frequency domain analysis identification is proposed in this paper by combining them together. Firstly, frequency domain analysis identification is used to obtain an accurate mathematical model, and then design internal model controller based on the identified model to achieve expected performance of the system. At the same time, through analysis of relations between attenuation factor and the parameters of the system, the scope of attenuation factor is given. In order to improve the accuracy of model parameters, the NPSO choose optimization algorithm is used to search the optimal attenuation factor.2. By studying the internal model control problem for a kind of large delay rectangular system, a directly decoupling internal model control method based on the constraint of equivalent transfer functions is proposed. According to the relationship of the equivalent transfer function and the system inverse model, an internal model controller is designed to achieve the complete decoupling of system and tracking the setpoint. Through the reasonable design of feedback loop filter parameters, the robustness of the system is improved.3. For the proplem of singularity plants cannot obtain the inverse model directly, a new decoupling internal model control method is first proposed based on pseudo-feed-forward decoupling in this paper. In order to overcome this problem, a pseudo-feed-forward decoupling is first designed to decouple the system. Then the internal model controller is designed by the decoupling generalized system. And the stability condition for the system are given by analyzing the the robust stability. Adopted integral of absolute error as evaluation index, new Luus-Jaakola(NLJ) optimization algorithm is used to to optimal the filter of controller for improvoing the system robustness.4. Two internal model control methods are proposed for rectangle rank deficient systems. One is based on full rank decomposition; the other is based on damped pseudo-inverse. The former uses full rank decomposition to design feedforward compensator and feedback compensator. Furthermore, the non-square relative gain subsystem selection criterion is used to choose the square subsystem and to realize satisfactory control of controlled subsystem under the premise of guarantee the stability of the aux output, which is easy to realize. For the latter, a damped pseudo-inverse is used to solve the problem. By introducing a damping factor, an internal model control is designed based on the damped pseudo-inverse. The principle of the method is simple and intuitively.5. Broken the limitation that internal model control is only viable for conventional system, an internal model control method is proposed for multivariable descriptor system. Through transforming the given system into fast and slow subsystem, two kinds of internal model control structure based on time scale transform are presented in this chapter. One is internal model control structure with fast and slow subsystem in parallel. The other is master-slave internal model control method. Both of the methods are use a feedforward compensator to eliminate the pulse of fast subsystem. Take the output of the fast subsystem as the external disturbance or uncertainties of slow subsystem. Then an internal model controller is designed for slow subsystem to realize system performance. Compared to the former, the latter has stronger robustness and anti-jamming ability by adding a PI controller in the forward channel.
Keywords/Search Tags:internal model control, frequency domain identification, rectangular system, structured rank deficient, robust analysis, decoupling internal model controller, Moore-Penrose generalized inverse, descriptor system
PDF Full Text Request
Related items