Optimization of mechanical structures using serial and parallel artificial immune systems

Abstract

A shape optimization problem of structures can be solved using methods based on sensitivity analysis information or non gradient methods based on genetic algorithms or on artificial immune systems. This paper is devoted to the method based on the serial and parallel artificial immune system. Artificial immune systems are developed on the basis of mechanism discovered in biological immune systems [9]. An immune system is a complicated, distributed group of specialized cells and organs. The main purpose of the immune system is to recognize and destroy pathogens - funguses, viruses, bacteria and improper functioning cells. The artificial immune systems (AIS) [1] take only few elements from the biological immune systems. The most frequently used are mutation of the B cells, proliferation, memory cells, and recognition using the B and T cells. The artificial immune systems are used to optimization, classification and also computer viruses recognition. A parallel artificial immune system (PAIS) was introduced in [2] for classification problems. The applications of an artificial immune system in optimization need only information about values of an objective function. The objective function is calculated for each B cell in each iteration by solving the boundary - value problem of elasticity by means of the finite element method (FEM). The main drawback of this approach is the long time of calculations. The applications of the parallel artificial immune system can shorten the time of calculations but additional requirements are needed: a multiprocessor computer or a cluster of computers are necessary. The message passing paradigm of parallel computations is used in presented approach. An artificial immune system is implemented as one master process, other processes - workers evaluate objective functions for B cells.