Search results

This paper aims to propose a reliable local search algorithm having steepest descent pivot rule for computationally expensive optimization problems. In particular, an application to the design of Permanent Magnet Synchronous Motor (PMSM) drives is shown.

A surrogate assisted Hooke‐Jeeves algorithm (SAHJA) is proposed. The SAHJA is a local search algorithm with the structure of the Hooke‐Jeeves algorithm, which employs a local surrogate model dynamically constructed during the exploratory move at each step of the optimization process.

Several numerical experiments have been designed. These experiments are carried out both on the simulation model (off‐line) and at the actual plant (on‐line). Moreover, the off‐line experiments have been considered in non‐noisy and noisy cases. The numerical results show that use of the SAHJA leads to a saving in terms of computational cost without requiring any extra hardware components.

The surrogate approach in the design of electric drives is novel. In addition, implementation of the proposed surrogate model allows the algorithm not only to reduce computational cost but also to filter noise caused by the sensors and measurement devices.

This paper aims to design an algorithm able to locate all the possible dangerous areas generated by the leaking of a fault current in a grounding system (i.e. the areas where the limits of the technical standards are not respected) and thus locate, inside each area, the point which takes locally the maximum value of touch voltage.

A fast evolutionary‐deterministic algorithm to solve constrained multimodal optimization problems is proposed. The algorithm is composed by three algorithmic blocks: a Quasi Genetic Algorithm to find a population of feasible solutions, a Fitness Sharing Selection to choose a subpopulation of feasible and fitter solutions having high diversity, a Hooke‐Jeeves Algorithm to find all the global and local feasible maxima.

The proposed algorithm has been successfully applied to various current field (i.e. to many shapes of grounding grids) problems to find the dangerous values of touch voltages generated by various grounding systems having any shape and it has turned out to be fast and reliable.

For this kind of problems, in fact, there is a lack, in literature, of multimodal optimization methods under safety constraints and the application of classical methods (e.g. genetic algorithms or deterministic methods) would be often inadequate since these methods are made so as to converge towards a single maximum point and so they unavoidably lose the information related to all the other possible maxima. On the contrary, a good application of the proposed allows the overcoming of these limits.

Sets out a method for determining the dangerous areas on the soil surface. The touch voltages are calculated by a Maxwell's subareas program. The search for the areas in which the touch voltages are dangerous is performed by a suitably modified genetic algorithm. The fitness is redefined so that the genetic algorithm does not lead directly to the only optimum solution, but to a certain number of solutions having pre‐arranged “goodness” characteristics. The algorithm has been called “quasi‐genetic” algorithm and has been successfully applied to various grounding systems.