基于遗传算法的片上螺旋电感模型优化方法

给出了一种基于遗传算法的片上螺旋电感宽带电路模型优化方法。通过对螺旋电感Y参数变化规律的分析,首先给出一种能够工作在高频段的电路模型结构。对每个频段内电路模型的参数优化被认为是一个误差最小化的多目标优化问题,将具有全局搜索能力和鲁棒性能的遗传算法引入到该优化过程中。介绍了遗传算法的优化过程以及其中采用的子频带划分方法。优化结果表明,针对不同几何参数的螺旋电感,该优化方法都能得到误差在5%以内的宽频带电感模型。

Genetic Algorithms for Model Optimization of On-Chip Spiral Inductors

A wide-band modeling approach for on-chip spiral inductors based on Genetic Algorithms(GA) is presented.Based on the analysis of the variation of spiral inductor's Y parameters,an equivalent circuit model is proposed,which can match the performance of spiral inductors in the high frequency range.The modeling approach is then converted into a multi-objective optimization problem,and GA is introduced to optimize the model parameters because of its robustness and its global convergence.The implementation of GA and a frequency range division technique are also illustrated in details.Optimization result of inductors with a variety of geometry sizes shows that our approach is effective to achieve accurate models with RMSE below 5% in a wide frequency range.