A new method has been developed that is capable of nonlinear dynamical systems modelling by means of the so called interval fuzzy model. The proposed approach is useful in the case of robust control design as well as in the case of fault detection in systems with uncertain parameters. When using nonlinear model predictive control, where the output prediction is based on the system model, the prediction has to be precise enough while in nonlinear fault detection methods an unsuitable system model results in false alarms or in too slow responses to faults.
COBISS.SI-ID: 4571988
The modified method of model-based predictive control algorithm which enables the implementation in simple and low cost hardware was developed. Predictive control was implemented in its modified form for rocket control, in cooperation with Japan Space Agency, in the case of hybrid semi-batch reactors, being the most commonly used in chemical, pharmaceutical and biochemical industry and on the simulator of the semi-batch reactor where different approaches of adaptive predictive control were tested exhibiting good practical properties.
COBISS.SI-ID: 5592404
New feedback control of mobile robots was developed and verified enabling safe and efficient operation of mobile systems in the certain environment. Interaction with the latter was achieved by the methods for sensing the environment. In control law for mobile robots reference trajectory tracking, predictive capabilities were introduced. For motion tracking of a group of mobile agents a robust and efficient approach using computer vision was suggested and developed. The mentioned approaches were validated on the different types of real mobile robots and on the developed digital simulators.
COBISS.SI-ID: 5894996
An algorithm for predictive control of hybrid systems with discrete inputs has been developed which was first designed for the optimal control of a multipurpose batch plant and then generalized to an arbitrary hybrid systems with discrete inputs. The new technique of transfer between different hybrid models representations was developed and the nonlinear modelling inside sub-spaces, using fuzzy approach, was introduced. So more efficient predictive control is enabled, which is computationally less demanding. The approach has been successfully applied to temperature control in a batch reactor.
COBISS.SI-ID: 4217428
An algorithm for automated model building for scheduling purposes was developed where the integration of the scheduling procedures into the production information system is important.. A comparison between the rule based scheduling and scheduling with heuristic search based exploration of a Petri net reachability tree was performed. Both methods give better results than the commercial scheduler. The developed algorithm was also included in the previously developed Petri net software tool, which was extended by the timed simulation support and scheduling algorithms for the obtained models.
COBISS.SI-ID: 20756263