基于RBF神经网络的PM2.5检测方法

PM2.5 Measuring Method Based on RBF Neural Network

针对现有PM2.5检测装置存在的自动化程度低,检测重复性差、有介质材料消耗等问题,设计了一套基于夫琅禾费衍射理论的激光衍射PM2.5检测系统。通过采用RBF(Radial Basis Function)神经网络对多路激光微粒子衍射信号进行建模计算,有效提高了系统的检测精度和自学习能力;对于系统中光电探测器输出的多路微弱信号,采用信号独立放大、分组采集和通道校准新方法,实现了多路微弱信号的精确调理和实时采集。仿真和实验结果表明,所设计的PM2.5检测系统有效克服了常规检测系统存在的问题,检测精度高,达到了设计要求。

Considering the disadvantage of existing PM2.5 detection devices, such as low degree of automation, poor test repeatability, dielectric material loss and so on, this paper presents a new PM2.5 detection system using laser diffraction technique based on Fraunhofer diffraction theory. The detection precision and self-learning ability are improved efficiently by using RBF neural network model with the inputs of multiple laser diffraction signals from the micro particles. The multi-channeI output signals of the photoelectric detectors have the properties of small and weak in the system, then some new technologies such as independent amplifying, group signal sampling, channel calibrating, and so on, are employed to realize the precise control and sampling in real time. The simulation and experiment results show that the new PM2.5 detection system satisfies the detection requirements of high calculating precision, and effectively overcomes the problems existing in the conventional detection system.