基于遗传算法的计算机水冷循环泵叶轮优化

Design Optimization of a Centrifugal Pump Impeller for CPU Refrigerating Using Genetic Algorithm

在通过实算的方法设计计算机水冷循环泵叶轮的时候,需要手动多次调整叶轮的各种参数直到大体上水泵的各项性能能接近对应的设计要求的性能,所以需要大量重复的计算,并且得到的结果不一定是最好的。为了达到简化设计过程并优化设计参数的目的,引入了水泵优化设计模型。首先,考虑到简化多目标优化的复杂性,建立优化模型来使原本多目标优化问题转化为单目标优化问题,使得仅用单一参数的大小便能衡量设计出的水泵满足要求的程度。另外,考虑到设计参数的多样性,采用了遗传算法对该模型进行求解,得到在遗传算法的迭代范围内最优的设计参数。最后,通过算例来验证这种方法,比较了实算法设计出来的结果和优化模型的方法设计出来的结果,发现了后者的水泵性能更能满足设计要求,因此,达到了叶轮优化的目的。

While designing a centrifugal pump impeller for CPU refrigerating, the parameters of the impeller should be dealt with manually to achieve the required performances of pump. Therefore, considerable repeated calculations are conducted to receive the satisfying results, which may not be the best ones. To streamline the designing process and optimize the parameters of the impeller, an optimization model of the pump is introduced. Firstly, considering of the complexity of multi-objective optimization （MOO） , an optimization model is developed aiming at converting the problem of MOO to the problem of single-objective optimization （SOO） , which can evaluate the de- gree of how well the pump will fit the requirements. Moreover, genetic algorithm （GA） is adopted to solve the model and obtains the best parameters by taking the diversity of the parameters into consideration . Finally, a study case is taken to verify the proposed method, and by comparing the calculated results with the ones that has already been studied, a series of better parameters are found by the proposed method. Thus, the goal of optimizationis achieved.