Rapid Particle Size Analysis of Suspensions Based on Video Technology and Artificial Neural Network with Additional Training During Operation

摘要

The original system for rapid determination of suspensions particle size distribution is proposed. The system is based on video technology and artificial neural network (ANN) that has training procedure not only in time of initial calibration but during operation too. The system structure and functioning algorithm has been considered. Results of some experiments are shown. The proposed system has an on-line laser nephelometric analyzer based on a simple measurement chamber with transparent window. The image of scattered light is registered by the camera. Further treatment converts some special parameters of images into input variables of the ANN which continuously generates parameters of particle size distribution on its outputs. The ANN learning is implemented on the basis of the reliable (but not so fast) microscopic method which is switched in use only in some moments when the registered image is utterly different from the known ones. Such approach allows to perform rapid analysis and to maintainits high metrological reliability.