改进的LMS算法在光全散射法粒度反演中的应用

Application of improved LMS algorithm in the inversion of particle size using total light scattering method

在光全散射法颗粒粒度测量中,正确地选择以及使用反演算法是非常关键的。首先介绍了光全散射法的基本原理,然后提出了一种基于非负PT直接算法的改进的LMS迭代算法,最后将改进的LMS迭代算法应用到光全散射颗粒粒度分布函数的求解中。对理想状态和加噪状态分别采用传统的LMS迭代算法和改进的LMS迭代算法进行求解,通过对比单峰分布颗粒粒度和双峰分布颗粒粒度的实验结果,改进后的LMS迭代算法使粒度相对误差均减少了4%左右,整个运行时间小于3s,不仅提高了LMS算法的运算时间、收敛速度和稳定性,而且还解决了LMS算法中初始值对结果的影响。

In the total light scattering particle sizing technology,the correct selection and use of inversion algorithm is very important.The basic principle of total light scattering method is introduced firstly,and then an improved LMS algorithm is proposed based on the nonnegative PT algorithm.At last this improved LMS method is used in solving the grain size distribution.Both the traditional and improved method are respectively used to deal with the ideal signal and the signal with noise.Comparing the error result between the single peak distribution and the bimodal distribution inversion,the improved method can reduce a 4% error than the traditional method.And the entire running time of this improved method is less than 3s,which not only saves the operation time,improve the convergence speed,and enhance the stability,but also solves the influence on the initial value of the LMS algorithm.