Research And Development Of The SDG Extensible Inference Engine Based On Rules

Safety assessment has become one of the important research subjects, because of the losses in economy and human, which caused by the safety accidents frequently happened in petrochemical industry in the past years. In all the kinds of safety analysis methods, SDG-HAZOP method has stood out owing to its ability in effective and complete analysis. It combines the SDG inference and HAZOP analysis mechanism, and then finishes the computer-aided HAZOP study of the SDG model, so that it overcomes the shortcomings of the conventional HAZOP, such as time-consuming, tedious and so on. It is said that the combination of SDG and HAZOP makes the results much more accurate and complete.However, conventional SDG model only considers the one-to-one influential relationship between the process variables, and neglects the other restricted relationships, so that its ability of knowledge representation is deficient. At the same time, the conventional inference engine only relies on the single intrinsic rules, and can not be added new ones according to the process. Because of those, the inference engine is not flexible enough, and lack of exchange with the user.In this paper, an SDG extensible inference engine based on rules is presented. The inference engine is integrated with VBA, so that the exchange with users can be enhanced. Meanwhile, conventional SDG model is improved. By introducing exchangeable object into the model, users can define the rules and operations for the process in VBA, then the new rules can be added to the inference engine. During the process of reasoning, the new inference engine adopts inference algorithm based on rules. It chooses the general rules or user-defined rules according to the attribute value of the VBA object model of the node and branch, and then reasons only after the rule is satisfied. The case study showed that the results of safety assessment could be filtered, and be much more accurate by reasoning the improved SDG model with the new inference engine based on rules.