Project: Probabilistic distributed industrial system monitor

Nowadays industrial control systems consist of number of components of several types: sensors and actuators, controllers, man-machine interfaces, database servers, computer networks, etc. Importance of particular units to the overall operation of the system may differ according to their role in the system. Total malfunction of most of the components can be quickly recognized in most cases. However, there exist cases, when for example a hidden fault of a sensor may impact negatively on proper operation of the control system or on the product quality but the fault is revealed with a significant delay._x000D__x000D_Even more tricky may be a partial degradation of function of a system component: worsening of the overall operation of the control system can be consequently recognized only statistically and it is difficult to reveal its cause. Examples of such situations can be a worn sensor or actuator, overloaded computer resources, unevenly delayed network communication, etc._x000D__x000D_Of course, there exist number of means for monitoring and fault detection of system components. The project aims at development of a consistent approach to this problem resulting in the advanced probabilistic monitor of a distributed control system. CO features of the solution will be:_x000D__x000D_a) Consistent treatment of pervasive uncertainty._x000D_Reliability and preciseness of each piece of information which is available in the control system can be considered as uncertain. Calculus of probability can be utilized as measure of the uncertainty. Consistent probabilistic approach will thus enable to evaluate measure of malfunction of a single system component and propagation of its influence within the control system._x000D__x000D_b) Hierarchical evaluation of control system "health"._x000D_Using the chain rule and utilizing models of relations among particular system components, the approach will enable to evaluate the overall control system health. Such "pyramidal" structure enables to express proper operation of the whole system by a single variable and, on the other hand, to trace a trouble down to its source._x000D__x000D_c) Modularity._x000D_Specific type of evaluation of proper operation will be assigned to particular types of system components. Applied to sensors, it will take into account physical quantity being measured, electrical properties, type of communication etc. Another method will be used for evaluation of proper operation of a network or a computer, to mention a few. Once particular evaluation modules are developed they can be used repeatedly according to the structure of the control system._x000D__x000D_d) Modeling of inner relations within the control system._x000D_Relations among particular components will be modeled to evaluate impact of potential malfunction of a component on another one(s). As an example, reliability of output of a single controller which uses inputs from several sensors will depend on reliability of particular sensors the weights of which can be considered as parameters of the related model. On the other hand controller reliability also depends on proper function of its hardware, timing, etc., which should be taken into account as well._x000D__x000D_Due to nowadays ubiquitous effort for decreasing of COtenance costs and improving quality of production, the monitor may be attractive for operators of virtually any industrial control system._x000D_To make the accomplishment of the project aims realistic, the project will consider an industrial control system of a reasonable extent, for example in the scale of a production line or an array of annealing furnaces. Data from the real control system will be utilized for off-line tests and experiments._x000D__x000D_Four Ps with complementary expertise intend to cooperate to accomplish the project aim:_x000D__x000D_1. COMPUREG Plzen, s.r.o., CZ - innovative SME producing industrial control systems with advanced features. COMPUREG will be responsible for data collection, design and programming of specific modules and their integration into the monitoring system, experiments and dissemination of practical results._x000D__x000D_2. UTIA AVCR, Prague, CZ - research institute renowned for its results on the field of Bayesian statistics and probability theory. UTIA will be responsible for research in the field of Bayesian information treatment and decision making._x000D__x000D_3. JSI, Ljubljana, SLO - research institute oriented towards applied research in systems, control and transfer of advanced methodologies into industrial practice. JSI will be responsible for research and development of algorithms for specific modules of the the monitoring system. _x000D__x000D_4. INEA d.o.o, Ljubljana, SLO - SME oriented to control and management of industrial processes. INEA will be responsible for implementation of specific modules within the PLC environment._x000D__x000D_The project Ps have been already successfully cooperating within the Eurostars project ProBaSensor._x000D__x000D_ProDiSMon will be prepared for the market within two years after completion of the project as an optional part of the industrial control system._x000D_

Acronym ProDiSMon (Reference Number: 7262)
Duration 01/01/2013 - 31/12/2015
Project Topic The project is aimed to bring a novel type of monitors of the overall control system condition based on hierarchical assessment of its components. The idea combines mathematical models of system's inner relations and priors on components reliability by using a consistent probabilistic approach.
Network Eurostars
Call Eurostars Cut-Off 8

Project partner

Number Name Role Country
4 INEA - informatizacija, energetika, avtomatizacija d.o.o. Partner Slovenia
4 Jozef Stefan Institute Partner Slovenia
4 COMPUREG Plzen, s.r.o. Coordinator Czech Republic
4 Institute of Information Theory and Automation, Academy of Sciences of the Czech Republic Partner Czech Republic