关键词: 延时宏模型, 无源性, 等式约束, 二次优化

Abstract:

Passivity violated delayed macromodels will lead to exponential divergence of the waveform in  time domain simulation. The causality of the delayed macromodels was ensured by conventional modeling strategies. But the passivity was not guaranteed. Based on this consideration, an efficient method for the passivity enforcement of delayed macromodels was presented, which is a generalization of the passivity enforcement method for lumped models. The method was based on quadratic programming with equality constraint, which was solved by using conjugate gradient. This method could enforce the passivity of the delayed macromodels while being optimal in the least squares sense. This procedure can greatly improve the efficiency. Numerical examples showed the efficiency and the accuracy of the method.

Key words:  delayed macromodels, passivity, equality constraint, quadratic programming