支持向量回归音频均衡器性能评价策略

An evaluation strategy for audio frequency equalizer performance based on support vector regression

针对因缺少大量分析数据样本而制约评价与分析音频均衡器性能的问题,提出了基于支持向量回归的音频均衡器性能评价方法。该方法以音频均衡器所应用的带通滤波器为实验研究对象。首先通过幅频特性测试仪采样,得到带通滤波器幅频特性数据集,然后进行支持向量回归,得到带通滤波器幅频特性曲线的逼近函数,用该函数对四项评价指标进行测定。实验表明,该方法能在获得模型参数全局最优点的同时,保证训练误差为0.0002,适于用示波法测量电子产品性能的评价。

Focusing on experimental samples lack,the new evaluation strategy of audio frequency equalizer performance based on support vector machine was presented.Considering the BPF that applied to audio frequency equalizer as experiment object,three things are done： obtaining the BPF datasets of the amplitude frequency by amplitude frequency characteristics detecting equipment to get samples,then using support vector regression to get the appoximating function of the amplifier amplitude frequency characteristics of the BPF,and using the function to detect the four performance parameters.The simulation results show that the perfect nonlinear mapping ability ensures zero training error and leads to the global optimal model parameters,and it is suitable for evaluation of the electronic production performance that tested by oscillograph.