Search

This paper proposes a novel parallelization of ABC Algorithm to solve Short-term Hydro-Thermal Scheduling (SHTS) problem. Parallelization is a novel way of improving the convergence of the algorithm. In case of normal ABC algorithms accelerating operation or modification can be used for better convergence and diversity. But with large number of variables the non-linearity of the search space increases and these modifications are not sufficient to give enough diversity in the crowds of swarms for finding better solutions. However, parallelization of the algorithm increases the diversity in solutions and the search capability. The performance of the proposed algorithm is experimented on two short term multi reservoir cascaded hydrothermal systems. First test case is moderately large while the second test case is a considerably large one. Experimental results demonstrate the superior performance of the proposed algorithm as compared to other recently reported ones in terms of convergence and better quality of solutions. [ABSTRACT FROM AUTHOR]

Ductile damage mechanics is essential to predict failure during cold metal forming applications. Several damage models can be found in the literature. These damage models are coupled with the mechanical behavior so as to model the progressive softening of the material due to damage growth. However, the identification of damage parameters remains an issue. In this paper, an inverse analysis approach is set-up to identify ductile damage parameters, based on different kind of mechanical tests and observables. The Lemaitre damage model is used and damage is coupled with the material behavior. The Efficient Global Optimization (EGO) method is used in a parallel environment. This global algorithm based on kriging meta-model enables the identification of a set of damage parameters based on experimental observables. Global and local observables are used to identify these parameters and a special attention is paid to the computation of the cost function. Finally, an identification procedure based on displacement field measurements is presented and applied for damage parameters identification. [ABSTRACT FROM AUTHOR]

Methods which provide good conditioning of model identification task in immune inspired, steady-state controller SILO (Stochastic Immune Layer Optimizer) are presented in this paper. These methods are implemented in a model based optimization algorithm. The first method uses a safe model to assure that gains of the process’s model can be estimated. The second method is responsible for elimination of potential linear dependences between columns of observation matrix. Moreover new results from one of SILO implementation in polish power plant are presented. They confirm high efficiency of the presented solution in solving technical problems. [ABSTRACT FROM AUTHOR]