To design approximately linear-phase complex coefficient finite impulse response （FIR） digital filters with arbitrary magnitude and group delay responses, a novel neural network approach is studied. The approach is ...To design approximately linear-phase complex coefficient finite impulse response （FIR） digital filters with arbitrary magnitude and group delay responses, a novel neural network approach is studied. The approach is based on a batch back-propagation neural network algorithm by directly minimizing the real magnitude error and phase error from the linear-phase to obtain the filter＇s coefficients. The approach can deal with both the real and complex coefficient FIR digital filters design problems. The main advantage of the proposed design method is the significant reduction in the group delay error. The effectiveness of the proposed method is illustrated with two optimal design examples.展开更多
随着功率模块集成化程度的提高,其散热结构优化已成为研发中的关键。拓扑优化可通过变换散热器形貌、结构来最大化地提升散热效果,因此受到了广泛关注。但在拓扑优化过程中,每步迭代均需要计算模块与散热器温度分布,占用较庞大的计算资...随着功率模块集成化程度的提高,其散热结构优化已成为研发中的关键。拓扑优化可通过变换散热器形貌、结构来最大化地提升散热效果,因此受到了广泛关注。但在拓扑优化过程中,每步迭代均需要计算模块与散热器温度分布,占用较庞大的计算资源和计算时间。为加速传统散热器拓扑优化进程,在基于传统固体各向同性材料惩罚SIMP(solid isotropic material with penalization)散热器拓扑优化方法的基础上,提出一种嵌套神经网络NN(neural network)同步学习的快速迭代方法。首先,构建散热器基于编码器-解码器结构的NN预测模型,即基于散热器形貌迭代进化过程实现优化结构的快速预测;其次,将NN模型与散热器SIMP拓扑优化流程相嵌套,利用迭代过程中的中间形貌同步训练NN;最后,针对单芯片、两芯片模块结构,对比所提方法与传统迭代方法的拓扑优化结果,验证了所提NN同步学习方法的准确性和快速性。展开更多
基金supported by the National Natural Science Foundation of China(6087602250677014)+2 种基金the High-Tech Research and Development Program of China(2006AA04A104)the Hunan Provincial Natural Science Foundation of China (06JJ202407JJ5076).
文摘To design approximately linear-phase complex coefficient finite impulse response （FIR） digital filters with arbitrary magnitude and group delay responses, a novel neural network approach is studied. The approach is based on a batch back-propagation neural network algorithm by directly minimizing the real magnitude error and phase error from the linear-phase to obtain the filter＇s coefficients. The approach can deal with both the real and complex coefficient FIR digital filters design problems. The main advantage of the proposed design method is the significant reduction in the group delay error. The effectiveness of the proposed method is illustrated with two optimal design examples.
文摘随着功率模块集成化程度的提高,其散热结构优化已成为研发中的关键。拓扑优化可通过变换散热器形貌、结构来最大化地提升散热效果,因此受到了广泛关注。但在拓扑优化过程中,每步迭代均需要计算模块与散热器温度分布,占用较庞大的计算资源和计算时间。为加速传统散热器拓扑优化进程,在基于传统固体各向同性材料惩罚SIMP(solid isotropic material with penalization)散热器拓扑优化方法的基础上,提出一种嵌套神经网络NN(neural network)同步学习的快速迭代方法。首先,构建散热器基于编码器-解码器结构的NN预测模型,即基于散热器形貌迭代进化过程实现优化结构的快速预测;其次,将NN模型与散热器SIMP拓扑优化流程相嵌套,利用迭代过程中的中间形貌同步训练NN;最后,针对单芯片、两芯片模块结构,对比所提方法与传统迭代方法的拓扑优化结果,验证了所提NN同步学习方法的准确性和快速性。
