A Study on the Real-Time Modeling Capabilities of the IDS Method

This paper deals with a fuzzy-based modeling technique called the ink drop spread (IDS) method. The IDS method has good convergence, and its modeling process, which is based on computing that uses intuitive pattern information instead of complex formulas, is simple and efficient. The computing structure of the IDS modeling system is characterized by heavy parallel processing on distributed processing units. This is similar to a case in which neural networks and fuzzy inference systems gain the real-time performance on their parallelized hardware architecture. The real-time performance of modeling methods depends on their algorithmic architecture and implementation. In this paper, the convergence of the IDS model is studied from the perspective of the algorithm, and it is compared against that of a neural network. In addition, some ideas for achieving fast convergence are presented