The classical approaches to process monitoring and supervision are based on mathematical methods for numerical signal processing. They constitute extensions and complement with respect to classical control theory methods. The most important issues performed by classical supervision systems include the following tasks:
reading the numerical data at every instant determined by samplingperiod; this include also analog to digital conversion, scaling, shifting,etc. (if necessary), filtering operations performed on the numerical input signals, calculation of common indices, such as mean value, amplitude, variance, square or absolute error, etc. within predefined time windows,. on-line checking if the signals are within required ranges and alarm generation in case of limit crossing, calculation of global indices, e.g. performance evaluation, energy consumption, material consumption, etc., model evaluation and adaptation (in case of adaptive control), adaptation of control (if applicable). The classical process monitoring and supervision tasks are performed with numerical software and basic logical software such as Programmable Logic Controllers (PLC).
A systematic view on process supervision
Systems science provides a general methodological view and approach on how to develop control, supervisory and decision making systems in a rational way. It highlights the necessity of analysing and defining, and taking into account the following factors:
goal of the analysis; usually understood as optimisation of certain performance indices, may also take into account satisfaction of certain criteria, and improvement of certain characteristics. In the considered case the main goals include improvement of the process reliability and safety, extension of state representation by knowledge-level situation assessment, providing the possibility of qualitative analysis of process behaviour, detection of faults, diagnosis, and improving man-machine communication. Satisfaction of goal(s) should be evaluated, even if in a qualitative way.
model of the system; it is necessary to have some model simply to know, what the required behaviour of the system should be, to have the possibility of predicting, simulating and analysing the behaviour, to control the system in a goal-oriented mode, and finally, to be able to perform diagnosis in case of failure,
input (controls, noise, environment) and output; what in fact constitute input, what can be controlled and what just identified (measured), what is disturbing the work, and what should be the output,
constraints imposed on the system; it is necessary to take into account all the constraints defining what conditions must be satisfied by the system during operation. Constraints define necessary conditions, i.e. ones to be satisfied by any solution. Constraints may also be related to theoretical properties and characteristics of the system and its operation.
human, environment, and other systems impact; what we usually forget about are the problems of man-machine communication and the influence of the developed solutions on environment. Further, the interconnection to other systems and possible influence (in both directions) cannot be overlooked.
The above five fundamental elements of any analysis and design in context
of supervisory systems lead to rational detailed analysis and design of
approach, tools, methods, etc. And seems to appear mandatory. Applied in
particular solution. Depending on the type of problem under analysis some
further elements of analysis – including for example economical analysis
(cost), problems of data and knowledge acquisition, specific resources
or even political problems – should be incorporated in the study.
