Qiang Bi

A technique for frequency response identification from relay feedback

In this paper, a new method for process frequency response identification is proposed, which can identify multiple points on a process frequency response from a single relay feedback test. The process output and input responses to the test are decomposed into the transient parts and the stationary cycle parts. These parts are then transformed to their frequency responses using the DFT and digital integration, respectively, to obtain the process frequency response. Real-time tests of the method on various processes give quite accurate process frequency responses, especially in the important frequency range from zero to the critical frequency. The testing conditions for our identification are the same as those for the step testing. The multiple points on process frequency response identified are very useful for process modeling and controller tuning. A real-time PID controller auto-tuning experiment which employs the proposed identification method is presented to demonstrate the applicability of our method.

Partial internal model control

Internal model control (IMC) is a well-known and effective control scheme. However, when unstable processes are concerned, the original IMC structure cannot be directly used for control system implementation. In this paper, a new scheme called partial internal model control (PIMC) is proposed, which is capable of controlling both stable and unstable processes. In PIMC, a process model is expressed as the sum of the stable and antistable parts and only the stable part of the process model is used as the internal model. The process stable part is canceled by the internal model and the remaining antistable part is stabilized and controlled with a primary controller, which is usually a PID-type regulator when the antistable part is of a low order. Various properties of a PIMC system such as internal stability and robust stability are analyzed. The design of PIMC is discussed in detail. Various simulation examples are included for illustration and a real-time implementation on a motor system is presented.

Relay-based estimation of multiple points on process frequency response

In this paper, a new technique for process frequency-response identification is proposed, which can identify multiple points on a process frequency response from a single relay feedback test. A modified relay is proposed and it can effectively excite a process not only at the process critical frequency gwc but also at 0.5ωc and 1.5ωc. As a result, the process frequency responses at multiple frequencies around ωc can be accurately estimated from one relay test using the FFT algorithm. The technique has been tested on various processes in real-time. The results show that the method is insensitive to noise and step-like load disturbance. The technique can also be easily extended to identify other frequency-response points of interest.

A Frequency Domain Controller Design Method

With the advent of the relay feedback auto-tuning technique, a number of PI/PID auto-tuning rules based on the results of the relay test are reported. Many of them employ only one or two points of the process frequency response. The designed controllers are limited to certain simple structured processes. In this paper, a multiple-point of frequency response fitting controller design method is proposed. First, multiple points on the process frequency response are estimated from a relay feedback test. The objective closed-loop response is then properly chosen with respect to the process frequency response. The controller is determined by minimizing the error between the actual and desired closed-loop response in a certain frequency region. Both over-damped and oscillatory processes are covered in the discussion. Simulation examples are included to demonstrate the proposed method.

Re-design of Smith predictor systems for performance enhancement

In this paper, a new design for Smith Predictor systems is presented. It employs a deliberately mismatched model to enhance performance over a perfectly matched system while using a simple primary controller. The design methodology is formulated in the frequency domain as an optimization problem and it turns out that an approximate solution can be obtained using the linear least squares method. To improve the performance further, a modified Smith Predictor system structure is also proposed, and it reduces the system into one involving second-order dynamics for the primary controller design. Illustrative simulation of several typical processes are included.

Multivariable controller auto-tuning with its application in HVAC systems

In this paper, an advanced PID auto-tuner for both single- and multivariable processes is developed and its application to HVAC systems presented. The auto-tuner exploits relay plus step test identification method to suit a wide range of processes including not only linear but also some nonlinear processes. It incorporates two latest PID design rules to achieve high control performance for processes with different dynamic characteristics. Decoupling control is used for multivariable processes. The auto-tuner was implemented on a distributed control system and practically used in industry. Experimental results exhibit the effectiveness and superior performance of the implemented auto-tuner over the manually tuned PID controller and the standard relay auto-tuner.

Implementation and testing of an advanced relay auto-tuner

Abstract In this paper, an advanced relay feedback auto-tuner for single-loop controllers is presented. The tuner combines an improved relay feedback process frequency response identification method and a multiple-point frequency response fitting controller design method. In process identification, multiple accurate points on the frequency response curve of a process are obtained in one single relay feedback test. A controller is then designed based on these estimated points using frequency response fitting methodology. The auto-tuner was implemented on a personal computer with AD/DA card. The software was developed on a commercial real-time control platform. The algorithms were written in C++ and compiled as a Dynamic Link Library. The auto-tuner has been tested on a dual-process simulator which can be easily configured as various typical processes, and on pilot coupled-tank and heat exchanger. The testing results show that the tuner gives significant performance improvement over the standard relay auto-tuner.

Co-operative control of multi-input and single-output processes: Problem formulation and solution

Abstract In control engineering, a process variable is normally regulated by a process manipulated variable. Due to physical limitations, any manipulated variable will have constraints on its amplitude. A practical problem frequently encountered in industry but less addressed by researchers is the output unreachability under input saturation: for a large disturbance or set-point change, the process variable may never reach the set-point even when the manipulated variable has been driven to saturation. The process variable can be brought back to the set-point only by activating auxiliary manipulated variables. However, timing of activation remains an open problem, A common practice adopted in industry is to wait for the process variable to reach a new steady state and then take action accordingly. This is time-consuming and may cause serious loss of productivity. In this paper, a new strategy based on disturbance estimation and output prediction is presented- It can detect unreachability early and activate the auxiliary variables timely so that the process variable will return to the set-point much faster. Simulation and real-time testing on a pilot mini-HVAC system show the effectiveness of the proposed strategy.

Continual Self-Tuning from Load Disturbance Responses

In this paper, a new continual self-tuning control scheme based on an unknown process response under load disturbance is proposed. It can re-estimate the possibly changed process frequency response using the benign process transients under an unknown step or impulse, like load disturbance. According to the disturbance response pattern, the unknown step/impulse like load disturbance is assumed to be generated by passing a step or an impulse through unknown rational function dynamics. With process input and output frequency response computed by FFT, the best estimation of the process frequency response is obtained using the least squares method. Based on the re-estimated process frequency response, a PID controller is re-tuned to adapt to the possible changes in the process. The effectiveness of the method is demonstrated through simulation examples and real-time tests.