Research for fluid impurity detection based on ANN and infrared spectrum analysis technology

Abstract

A series of economic losses is caused by the biofilm of water pipe in industrial real water systems. Combined optical fiber self-relative technology with infrared spectrum analysis technology, real time detection technique for forming thickness and ingredient is put forward in the paper, which provides technical support and reliable data for analyzing biofouling influencing factors, contaminant separation and warning. Schematic diagram of biofouling detection is presented. Compensation technology based on radial basis function (RBF) neural network and learning algorithm are studied in order to solve the problem of measurement precision and range. Biofouling forming and optical characteristics in industrial real water systems are researched and standard specimen collection is set up. Correcting model explaining quantitatively relation between substance ingredient content and infrared spectrum based on partial least squares (PLS) method. A new method is provided for the research on biofouling in real water system, which can be used in other fields such as mining, environment protection, medical treatment and transportation of oil, gas and water.