Particle Swarm Optimization of a Hybrid Energy Storage System

Abstract

The paper presents the energy loss minimization of a hybrid energy storage system used in an electric vehicle, composed by a battery and a supercapacitor. The optimization is carried out by searching the optimal power sharing between the energy storage devices. The power sharing factor is defined as a discrete time variable, with constant values during each subdivision of the driving cycle. The elements of the optimal solution vector are the power sharing factors and the time instants that define the subdivisions. The particle swarm optimization algorithms have been validated using the Rastrigin test function, and three versions of the boundary behaviour have been compared in case of the constrained optimization. The algorithms have been tested for the energy loss minimization in case of a simple driving cycle, and their performance has been assessed by statistical analysis for different swarm sizes.

Optimal Power Sharing Control of the Hybrid Energy Storage System of an Electric Vehicle Along a Standard Driving Cycle

“…In the following, either a single partition (the entire driving cycle) or multiple partitions are being used, with subdivision of each partition into two segments. The optimization vector in each partition is formed by the power shares of the supercapacitor in the two subintervals, defined by (1), extended with the length of the first subinterval normalized to the length of the partition [16].…”

“…In this way the dimension of the solution space is only 3, and the complexity of the problem is moderate [16].…”

“…Both the power sharing ratios and the time instant of the subdivision are elements of the optimization variable. The corresponding particle swarm optimization (PSO) algorithm is described in detail in [16]. The number of individuals in the swarm is 25, and a number of 8 constraints are applied during the optimization, including the minimum required energy storage, as a time function.…”

“…In [16], [17] we introduced a constrained particle swarm optimization (PSO) algorithm [15], [18], [19] to minimize the energy losses of a HESS from an electric vehicle, for a simple driving cycle. In this paper an extension of the PSO is presented for the standard New European Driving Cycle (NEDC) [20].…”

Electrical Power Generator Faults Analysis Using Fault Tree and Bayesian Network

“…Other studies proposed artificial intelligence methods for industrial systems faults analysis, such as: artificial neural networks for fault diagnosis [5], fuzzy logic [6], and support vector machine [7]. Other authors have used artificial techniques for optimizing EPG operation and improve its availability such as: particle swarm optimization [8] and genetic algorithms [9]. These artificial intelligence techniques are given a strong contribution in the mastering of the EPG function, and they have improved the fault diagnosis and prediction when combined with traditional methods.…”