The Study On Design Optimization Of Sound Baffle With Genetic Alogrithms

Nowadays in sonar design, the structure of transducer array becomes more and more complex with the development of the transducer technique, which leads to enchance the performance of sound baffles. Sound baffle is one of important components in sonar system which can diminish the environmental diffusing sound at neighborhood of transducers due to effects of reflection, absorption and decoupling, and improved the receiving character, changed the directivity of transducer array. In this paper, genetic algorithm had been used to optimize the sound baffle in order to decrease the reflection of sound for different incident directions and frequencies, as well as to reduct the environmental diffusive sound for some 3-D transducer array.Taking layered composite structure as an example, the parameters of material required for optimum baffle design are determined theorically. The relationship between different parameters of material and the variation of reflection coefficient for different sound frequencies and incident angles is analysed by simulation. The absorption ratio of each layer, and its contribution, when located at different layer, for different parameters of material and different frequencies had been calculated and analysed by genetic algorithm.Also taking the periodic structure with variable sectional cylindrical cavities as another example, the mathmatical model for optimization of cavity structure is established. The influence about the change in form of cavity on reflection coefficient for different sound frequencies is analysed. For the optimization problem, minimum of mean values and variance of reflection coefficient at different incident angles had been required. And suitable weight function had been used in establishing of the object function, after analysis about the variation of mean value and variance of reflection coefficient at different incident angles.As respect to the early maturing effect, which usually appeared in genetic evolution, after a thorough analysis of standing operators for recombination, a new recombination operator based on fitness measure had been introduced in design of the genetic algorithm for searching space control, which had been examined by different kinds of test function, and be proved that it can keep the required diversity of species during the procedure of evolution.A searching program named valley-searching is used to identify the different local optimal solutions in order to dodge all of the local optimal solutions during evolution. The validity of applied algorithm had been verified by multi-peak test functions of different dimension.Inversion of material parameters had been used to verify the realiablity of this optimization algorithm. Measurements had been done about the reflection coefficients of sound baffle made from known material, and the proposed algorithm is used to determine the parameters of material. A new fitness function for the error between theoretical and measured values of reflection coefficient is designed and be calculated, thus the realiability of this algorithm is testified.