Study On Electronic Clearing Using Nerual Networks

In textile industry, it is desirable to produce yarn as high quality as possible. It is benefit for exporting and further processing. One of the most important steps is using electronic clearing.The first electronic clearing appeared since 1951, half century has pasted; it has been remarkable in the efficient for clearing yarn faults. But there are some gaps between human intelligence and electronic clearing to inspect the yarn faults. The main reasons are the clearing efficient not enough and clearing curved line cannot be changed.To find a new way for the foundation principle of electronic clearing, I have analyzed the principle of electronic clearing and known it is a device of classify, it should substitute for Neural Networks. After the study on Neural Networks and compare the algorism, I have realized its optimized scheme is BP (back propagation learning algorithm) Neural Networks for clearing yarn faults.At the Lab of the Artificial Intelligence, University of Zurich, I have did the experiment by "Virtuelles Labor fur Neuronale Netze" which is a platform of Neural Networks. Selected the parameter for the "training pattern set" and "work test pattern set" . I studied and checked the parameter that are used non-linear classify for clearing. It has been sure that Neural Networks can be classifying to the yarn faults. It will set up a new clearing foundation principle. The paper also discusses the further technique problem.