Improved population initialization method and its application in bridge optimization

Abstract

When solving bridge optimization problems by traditional genetic algorithms, a random method is often used to generate the initial population, which can’t ensure that the initial population is evenly distributed in the solution space and reasonably represents the solution space. In order to solve the shortcomings of random method, an improved population initialization method is proposed by combining random method with good point set. Based on it, aimed at the characteristic of implicit non-linear objective function or constraint function in bridge structure optimization, a BP neural network is used to simulate the relations between the objective function or constraint function and design variables. Then this paper takes a steel truss bridge as an example of optimization. Taken the allowable deflection of bridges as limit and aiming at minimizing the amount of steel, optimization model of bridge is established. Finally, the improved population initialization method is applied to solve the optimization model. The results show that the improved population initialization method can not only improve the uniformity of the population, but also be suitable for optimization design of bridges.