An Inference And Decision Logic Approach For Process Planning Based On Mathematical Logic

Process preparation serves as a vital link between design and manufacturing functions, which plays a key role in determining the agility of manufacturing, competitiveness, response to market. Modern manufacturing system requires digitization and rapid response of process preparation while Process planning is one of the primary concerns of Digitalized rapid process preparation. Process reasoning and decision logic is the kernel of intelligent process planning, which is on demand with the manufacturing resources on the shop floor and requires a variety of knowledge of design, process, and manufacturing. However, a successful intelligent machining process planning system relies greatly on a good process reasoning and decision logic which is still the basic crux of process planning. To date, the basic crux of process planning involves the following problems: (1) lack of systematic methods of modeling the manufacturing resources for process planning; (2) need of process knowledge representation approaches especially from the mathematical viewpoint; (3) lack of microcosmic process reasoning mechanism from the standpoint of informationalization of manufacturing bottom; (4) need of systematic decision logic approach of process planning. To solve these problems, this research is to address the following issues.In this research, a global scheme strategy for process planning, which consists of knowledge customization strategy and decision logic strategy with reasoning mechanism, is proposed based on analyzing process knowledge. A decision logic model for process planning is presented using multi-agent technology while a structural model for agent is proposed. Since each agent involves many atomic-level tasks, three relationship models for atomic-level tasks are built. It lays the foundation for in-depth decision logic process of intelligent process planning.The clustering-based method of the manufacturing resources for process planning is presented by combining clustering analysis method with averaging method. A clustering analysis strategy is proposed, which involves three parts: classifying by machining methods, determining clustering attributes of each class, and clustering analyzing. The average value of similarity is applied to clustering granularity so as to reasonably partition and classify the manufacturing resources on a shop floor. Since one object may simultaneously belong to two or many clusters, the averaging method along with the clustering analysis method is utilized to determine the uniqueness of belongingness of clustering samples. Thus, it makes clustering results more reasonable and improves the application and feasibility of clustering results.This research addresses representation approach for process knowledge from the mathematical viewpoint, which is based on knowledge customization strategy above. The meta-modeling paradigm of the manufacturing resources for process planning is proposed using the first- and second-order logic and mapping theory, which is based on the idea of encapsulating process knowledge into the meta machine. Infrastructure model and information model for the meta machine are built, which is represented by first-order logic. Mathematical model of mapping relationship between logic sets is presented by combining second-order logic with mapping theory as well as approach for determining logic sets is described. Customized process knowledge with dynamic information, which integrates the manufacturing resources with process knowledge, is basis for process reasoning and decision logic and thus improves feasibility of a process plan.The decision logic schema of process planning is proposed by combining backward chaining reasoning in mathematical logic with the meta-modeling paradigm to process reasoning in process planning. Machining features, as goal-oriented constraints, are analyzed, studied, and modeled along with strategy. Based on the proposed decision logic flow, systematic decision logic model is presented particularly with regards to coping with selective decision while atomic inference engine model is built from the microcosmic viewpoint. Finally, an illustrative example is applied to analyze and verify the decision logic approach.