The foundation for the Genetic Algorithm is based on the principle of survival of the fittest. It uses the analogy of the evolution of a population of gene sequences over several generations to yield the optimal accelerometer set. In this analogy each gene represents a potential accelerometer location, each gene sequence represents a potential accelerometer set (set of genes), and the population is composed of several genetic sequences. At every generation an objective function is used to evaluate each sequence’s fitness, and those accelerometer sets with the highest fitness (or lowest error) have the highest probability of reproducing to form the next generation. Three forms of reproduction are implemented: direct reproduction, crossover, and mutation. These methods of reproduction are described in detail in the Generation Control Table section.
One of the key features of this implementation is that the error always decreases from one generation to the next, assuming the reproduction variable is non-zero. By default the best gene sequences (accelerometer sets) are directly reproduced to each subsequent generation, guaranteeing that the minimum error never increases. In addition, because of the randomness of this nondeterministic process, different accelerometer sets will most likely result if the GA is run more than once helping to avoid local minimums of the objective function.