基于核估计的成品率分析算法及其实现

A Method of Yield Analysis Based on Kernel Density Estimate

本文改进了NBA(Non-parametric Boundary Analysis)算法,给出了一种基于核估计法的成品率分析算法。利用该算法不必事先假定参数服从何种分布,仅通过较少次数的仿真,便可获得成品率分析结果。而且可有效提高NBA算法的准确性,降低了算法复杂度。文中详细介绍该算法的实现方法。算例表明,该方法对电路设计进行快速成品率分析及电路的稳定性设计具有较好的应用前景。

A yield analysis method based on kernel density estimate by improving NBA (Non_parametric Boundary Analysis) is proposed. Using this improved method, result of analysis on the data of any distribution by fewer simulations can be attained. Accuracy and efficiency of NBA can also be improved. Implementation procedure is presented in detail. Experience example demonstrates that the proposed method is very useful in yield analysis of electronic circuit design.