多目标优化设计分频网络方法研究

Research of Designing Crossover Networks Using Multiple Objective Optimization

论述了运用多目标优化设计扬声器系统分频网络的方法 .使用实测的扬声器阻抗、频响数据 ,模拟扬声器接入分频网络后扬声器系统的响应 ,并运用多目标优化方法对系统响应进行优化设计 ,使系统的多个响应达到预定的指标 .给出的实例表明 。

The design of an electric crossover network for a multiway loudspeaker system is often a difficult and tedious task. The reason is that there are many complex and sometimes conflicting parameters to be accounted for during the design process. In this paper, a method of designing loudspeaker crossover networks using multiple objective optimization is described. Designing loudspeaker crossover networks is regarded as a constrained multiple objective optimization problem. Some sound-pressure-frequency response of the loudspeaker system is the objective to be optimized. And each sound-pressure-frequency response has a desired response. Some error functions are defined to indicate the difference between the calculated response and the disired response. The limit of the impedance-frequency response of the system is the constrained condition. Using linear weighting method and punitive function method, the constrained multiple objective optimization problem is transformed to a unconstrained single objective optimization problem. To ensure an accurate prediction of the frequency response of the system, the measured sound-pressure-frequency response and the impedance-frequency-response of the loudspeakers are used to simulate the frequency responses of the system when the loudspeakers are connected to the crossover. The component values of the crossover are selected by multiple objective optimization so that the difference between some desired sound-pressure-frequency responses and the calculated sound-pressure-frequency responses of the system can be minimized and the constrained condition can be met. Using this design method, the traditional requirement for trial-and-error adjustment of the values of the actual crossover components is unnecessary. A design example shows that the calculated and measured frequency responss are accordant.