More attention is paid to on-line monitoring of biofouling in industrial water process systems. Combined optical fiber self-referencing technology with artificial neural networks (ANN) technology, real time detection technique for forming thickness and ingredient is put forward, which provides technical support and reliable data for analyzing biofouling influencing factors, contaminant separation and warning. Unclad fiber sensing mechanism and self-referencing fiber optic sensor are 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. 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.