Automatic Diagnosis for Profibus Networks

Abstract

This research proposes the use of intelligent systems to diagnose industrial network communication via Profibus DP Protocol. These diagnostics are based on the information provided by the physical, data link and user layers from the Profibus DP. In order to analyze the physical layer, an artificial neural network first analyzes signal samples transmitted through industrial network. Moreover, some diagnostics can be acquired by using an expert system that analyzes frames which are transmitted by the data link layer. Finally, this research proposes a fuzzy system, which indicates to the user an ideal value to the target rotation time variable. The project was validated by data obtained from concrete Profibus protocol and by some synthetic data. The results were satisfactory, proving the great strength and versatility that intelligent computer systems have when applied to the outlined purposes in this work.