Unfalsified Adaptive CONTROL (UAC) is a recently proposed robust adaptive CONTROL strategy. In this paper, the UAC principles and algorithms are reviewed and Multi-Model UAC is followed as an intermediate between UAC and multiple model CONTROL. Different approaches in UAC and MMUAC are studied. Also, Multi-Model Unfalsified Generalized Predictive CONTROL (MMUGPC) is proposed, which is a new CONTROL design strategy in the UAC framework. For an uncertain system, by utilizing several generalized predictive CONTROLlers and discrete switching between them with unfalsified CONTROL, a new structure is proposed and appropriate equations are derived. Simulation results show the effectiveness of proposed Multi-Model Unfalsified Generalized Predictive CONTROL.

Unfalsified adaptive CONTROL strategy is a data-driven approach in robust adaptive CONTROL that selects the stabilizing CONTROLler from an the available CONTROL bank based on the input-output system’s data. Selection is done without activating the CONTROLlers by using the virtual reference signal and a cost function. Stability of the closed loop system is guaranteed. In this paper, inspired by the unfalsified CONTROL approach, the goal is to select a CONTROLler from the available pre-designed CONTROLler bank that has the highest level of disturbance attenuation. The selection is based on the system’s data by using a cost function without disturbance measurement. By introducing a new concept called virtual disturbance, the performance of CONTROLlers is evaluated without activation. The convergence of the algorithm and the disturbance attenuation level of the proposed method have been proven in a theorem. Simulation results show performance of the proposed method on a wind turbine for various wind speed disturbances.

In this paper, using hierarchical CONTROL system, we have introduced a new method for CONTROL of double inverted pendulum. In this method first we separate the system into three sub-system and for each sub-systems we design a local CONTROLler which is independent from the others. Then, in the higher level considering the interaction between the sub-systems we design a SUPERVISORY CONTROL. (this is done by producing a number of weight function corresponding to our sub-systems). The results show that fuzzy CONTROLlers stabilize the system in a wide range of initial condition in about 20 to 24 seconds.

The design and implementation of a novel fuzzy SUPERVISORY CONTROL approach for the motion CONTROL of a seismic shake table is addressed in this work. For this purpose, a single degree of freedom laboratory-scale electric shake table was developed. The CONTROL scheme comprises two loops: a PI inner loop and a fuzzy outer loop as the supervisor. Three separate SUPERVISORY CONTROLlers are proposed and implemented in the shake table and their performance in tracking two real earthquakes is assessed via extensive shake table testing. The test results reveal the effectiveness of the fuzzy SUPERVISORY CONTROLler in reducing displacement and acceleration tracking errors.

In the SUPERVISORY CONTROL of systems, an observer monitors discrete system responses to the events of the system environment and will report an unsafe or critical situation if the response is undesired. The undesired response means that it does not adhere to user’s requirements. Several methods have already been proposed to model and simulate SUPERVISORY CONTROL of discrete systems; however, there is a lack of a systematic method relying on problem domain data. To present a domain-based method, we use a three-step method. In the first step, we commence with discrete data. Discrete data are primary ingredients of the discrete systems environment and used by system users as a gauge to express their requirements playing a vital role in safety-critical systems, such as medical and avionic ones. We use the discrete data for definition of events and conditions constituting requirements definition. Having extracted events, conditions, and user’s requirements from problem domain data, a Petri-Net automaton is constructed for identifying violation of user’s requirements in the second step. The net constitutes the core of the observer and it is used to identify undesired responses of the system. In the third step, run-time simulation of the observer is suggested using multithreading mechanism and Task Parallel Library (TPL) technology of Microsoft. Finally, by proposing the Train Protection System as a case study of a discrete concurrent system, we tangibly show how one can apply the steps of our method for modeling and simulation of the observer. The simulation results were analyzed based on the system implementation on a multi-core computer.

CONTROL of seismic shake table in order to track the predefined earthquake profile is a key concern in design of seismic shake tables. This paper proposes a vision-based real time displacement measurement system using image processing techniques to CONTROL a laboratory-scale seismic shake table. The shake table is CONTROLled via a fuzzy-SUPERVISORY CONTROLler, an inner PID loop and a Fuzzy outer one, whose feedback is provided by the vision-based measurement system. To minimize tracking errors, the fuzzy CONTROLler uses displacement and acceleration responses as its feedbacks. For this purpose, a camera and an image processing application are utilized to measure the motions directly in real time. Results are sent to a host PC through a network as the CONTROLler feedback. Proposed system performance is compared with an alternative system which utilizes a linear encoder as displacement sensor and CONTROLler feedback. Test results prove effectiveness of the proposed fuzzy system in cutting back the tracking errors. In addition, the vision-based system uses a very low cost camera to measure the displacement directly. It has appropriate accuracy, works in real time, and doesn't need any contact with the table, comparing to the encoder version.

Distributed SUPERVISORY CONTROL is a method to synthesize local CONTROLlers in discrete-event systems with a systematic observation of the plant. Some works were reported on extending this method by which local CONTROLlers are constructed so that observation properties are preserved from monolithic to distributed SUPERVISORY CONTROL, in an up-down approach. In this paper, we find circumstances in which observation properties are preserved from monolithic to distributed SUPERVISORY CONTROL. Local observation properties, i.e. local normality and local relative observability are employed for investigating observation properties of each local CONTROLler, which are constructed by any localization algorithm that preserves CONTROL equivalency to the monolithic supervisor with respect to the plant. These properties enable us to investigate the observation properties from monolithic to distributed SUPERVISORY CONTROL. Moreover, observation equivalence property is defined according to the CONTROL equivalence in a distributed SUPERVISORY CONTROL with partial observation. It is proved that with preserving observation equivalence of the local CONTROLlers to the monolithic supervisor, the CONTROL equivalence is satisfied, if and only if the intersection of local event sets is a subset of or equal to the global observable event set.

When the process is highly uncertain, even linear minimum phase systems must sacrifice desirable feedback CONTROL benefits to avoid an excessive ‘cost of feedback’, while preserving the robust stability. In this paper, the CONTROL structure of SUPERVISORY based switching Quantitative Feedback Theory (QFT) CONTROL is proposed to CONTROL highly uncertain plants. According to this strategy, the uncertainty region is suitably divided into smaller regions. It is assumed that a QFT CONTROLler-prefilter exits for robust stability and robust performance of the individual uncertain sets. The proposed CONTROL architecture is made up by these local CONTROLlers, which commute among themselves in accordance with the decision of a high level decision maker called the supervisor. The supervisor makes the decision by comparing the candidate local model behavior with the one of the plant and selects the CONTROLler corresponding to the best fitted model. A hysteresis switching logic is used to slow down switching for stability reasons. Besides, each CONTROLler is designed to be stable in the whole uncertainty domain, and as accurate in command tracking as desired in its uncertainty subset to preserve the robust stability from any failure in the switching.

In this paper, a SUPERVISORY based switching strategy is employed to solve the H2/H¥ multi objective CONTROLler design. Using linear matrix inequalities, two distinict CONTROLlers are designed to meet the H2 and H¥ performance specifications. A state realization for each CONTROLler transfer matrix is found such that the asymptotical stability of the closed-loop switching system is guaranteed for any switching sequence. A supervisor is used to diagnose the exogenous input changes and switch to the relevant CONTROLler. Simulation results show that switching strategy improves the performance of the CONTROLler and reduces the conservation in comparison with the common H2/H¥ CONTROLler.