Parameters Estimation of PID Controller
|Course||Circuits and Systems|
|Keywords||Controller parameters Series form Analog controller Standard form Settings Differential link Digitized Parameter estimation Input signal System Identification|
The predominant means of achieving control throughout a wide range of industries is by means of the PID controller. The primary aim of the control engineer is to select these gains in such a way as to optimise the behaviour of the controlled system. Much research has been carried out in this area leading to a variety of tuning methods being put forward. Setting the required gains can be hampered by a number of practical difficulties:? PID control can be implemented in a variety of ways and although these are equivalent the values of the gains will be affected by the implementation of the PID algorithm.? There may be no direct means of ascertaining the values of the parameters set. In the case of analogue controllers the relevant dials may be out of calibration or even without a scale.? The manual for the controller may have been lost.? Other difficulties include the existence of saturation limits, in the case of digital controllers, of an anti-alias filter to prevent the sampling procedure from distorting the (low frequency band) of the input signal.These problems provided the motivation for this paper whose aim was to determine the gains set on an analogue controller using system identification.Algorithms were developed which were capable of identifying the parameters of an analogue PID controller modelled in SIMULINK. The algorithms were tailored to the standard and series implementations. They used both ARX and ARMAX parameter estimation methods and could achieve accuracies better than 99%.The real test, however, of the method would be to apply it to a variety of analogue controllers of both the standard and series structures. If these prove successful it would be relatively easy to extend the method to the parallel implementation.Currently, the identification is performed "off-line", however, it would be much more convenient if the estimation process were conducted in tandem with the adjustment of the controller’s gains. To achieve this, a recursive approach should be investigated.