A two-level reasoning method based on SVM_RETE algorithm in industrial environments

Abstract

In the industrial environments, the efficient automatic control of terminal devices depends on the changing of reception data and customized rules. As the development of Industrial Internet of Things (IIoT), more and more industrial data can be achieved to generate the big data of IIoT. Therefore, efficient matching and processing of dynamic IIoT big data and customized rules becomes increasingly important. This paper presents a two-level reasoning method in improving performance of rule engine. The first level uses a decision function trained by Support Vector Machine (SVM) to classify reported data from sensors based on the semantic data interface. In this stage, the useless data is filtered in order to reduce subsequent process. In the second level an improved RETE, in this paper is called SVM_RETE algorithm for matching rules and performing actions is presented to increase efficiency of reasoning processing. The proposed scheme is performed in a practical industrial environment. The experiment results show that the method can be performed efficiently and flexibly when massive data is involved.