G. Bialic

On the Impact of Plant Model and Controller Sophistication on Performance of Disturbance Attenuation and System Robustness

In the paper the problem of disturbance attenuation performance in single-loop control system is investigated. Modified MV control strategy with bounded control variance is used as benchmark for both LQG and classically and optimally tuned PID-type controllers. Certain time and frequency domain functions are then examined to further assess the control performance in terms of robustness. It has been shown that as long as the system to be controlled is delay-free, and there is no extreme demand on performance, simple lag-delay system model along with optimally tuned PID control algorithm provides control quality similar to that of LQG controlled original system assuming the same bound on control signal variance. Although the classical performance measure based on substitute delay indicates then almost optimal performance, there is still room for additional improvement that can be attained using both PID and LQG controllers and more sophisticated original delay-free model. The price paid is much larger control variance. It is interesting to note that while LQG systems remain robust this is not longer valid for PID controllers

On control performance assessment based on lag-delay models

In the paper, suitability of control quality assessment based on delay approximation for delay-free plants is investigated. To this end, the LQG control benchmark, which can be seen as a MV benchmark with bounded control variance, is compared for both linear delay-free continuous-time plants with outputs corrupted by a stochastic disturbance and their lag-delay models. Using approximated plant models, the area of achievable accuracy is then defined for control performance assessment. Control systems with discrete-time PID controllers tuned both classically and optimally in such way that disturbance characteristics are taken into account are considered and compared with the benchmark.

Design of the scales for the power boiler fuel feeding system based on the process identification

Abstract In the paper the mathematical modeling methods were employed to solve the problem of fuel mass estimation supplied to the power generating boiler grid. It was shown that popular polynomial and neural network based identification algorithms can be used to measure amount of fuel supplied to the boiler combustion chamber. Furthermore the original conception and exemplary realization of the mass stream measuring system was presented.

Developments in the Control Loops Benchmarking

The problem of control performance assessment and monitoring is getting more and more significant because control systems have much bigger influence on accomplishing aims determined by companies. These aims mean achieving goals connected with quality, safety and profits. This justify academic and commercial interest in development of methods for analyzing the quality of such systems which allow to avoid unreliable human factor. The control system performance cannot be depicted by means of only one simple statistics. The whole procedure called control loops benchmarking (Harris & Seppala 2002) require much more complicated multi-stage process consisting of: data acquisition, analysis and diagnostics (making the tool based on mathematical model of process, the lower bound estimation, the existing control loop performance assessment, testing of the performance improvement using existing controller structure), retuning or control algorithm replacement.