The Parallel Reasoning Algorithm Based On Fuzzy Petri Nets

As a knowledge representation method, Fuzzy Petri Nets can not only describle the expert system reasonablely which is uncertainty, fuzzy, asynchronous and concurrent, but also use more rigorous mathematical theories to reason. It is using the language variables to describe the knowledge of expert system, which is easy to be understood and mastered. The reasoning algorithm based on Fuzzy Petri Net which broaden the application fields of Petri Nets is also more and more important. Since 1965, the scholars have done a lot of researches on the fuzzy reasoning algorithm. This article proposes an improved algorithm on the basis of the existing fuzzy reasoning algorithm, and programmes it with C/C++.The forward reasoning algorithm makes full use of the parallel power of expression of Fuzzy Petri Nets, which starts from the initial confidence set out to launch new confidence value of all places. The reverse search method has more real-time performance, which starts from the target place to solve the confidence value of all the associated paces, but Its applicable scope is small. The time complexity of the algorithm will increase when the scale of the Fuzzy Petri Nets become large. So combining with the matrix command that proposes a matrix operation mechanism of Bi-direction reasoning. And analyzes the feasibility and the complexity of the algorithm. Experimental analysis shows that the reasoning algorithm can effectively improve the parallelism of the whole process of reasoning, and reduce the time complexity of algorithm and improve the efficiency of reasoning, compared with a general fuzzy reasoning algorithm and a reverse search algorithm. Lastly programmes the reasoning process with C/C++.