Research And Practice On Design Of Solid Rocket Motor Based On Knowledge Engineering

This dissertation, focusing on the Solid Rocket Motor (SRM) conceptual demonstration, explores the design method based on knowledge engineering and their applications. In detail, the system and the framework of the SRM design based on knowledge engineering are established. SRM Structure-behavior-function (SBF) representation model is established, characteristic and requirement of SRM design are analyzed. The methods of knowledge representation, knowledge acquisition and knowledge reasoing are thus established to suggest some efficient appraoches to inheritance of SRM design knowledge.In order to obtain the quantitative design demand used for the design reasoning, function modeling of design is produced. SBF mapping modeling of design is established, the task of system and subsystems design, and main points of acquisition and representation of model knowledge and experiential knowledge are resulted.Object-Oriented hybird knowledge representation is studied. Some single knowledge representations, such as semantic web of quantitative design demand, frame structure of material, qualitative rules of heuristic knowledge, neural rules of experiential knowledge and decision table of system and subsystems design, are developed. Encapsulation of single knowledge representations are finished by Object-Oriented, and the case of SRM design is established.Design knowledge acquisition under domain expert direction is studied. According to quantitative parameters mapping relation, heuristic knowledge acquisition strategy based on qualitative reasoning and data mining is proposed. According to its source, experiential knowledge acquisition methods from product design cases and domain experts are proposed. For the sake of completeness and granularity, using information entropy, experiential knowledge acquisition based on combination of design cases and domain experts experience is established.Case-based reasoning (CBR) of SRM design is studied. To improve the diversity of similar cases, similarity retrieval method based on partitioned clustering and General Fuzzy Min-Max neural network is developed. According to fuzzy, coupling and uncertain parameters mapping relation, case adaptation using neural rules and qualitative rules is developed. CBR processes of SRM system and subsystems design are established.The design method based on knowledge engineering was used for design examples, including single-chamber single-thrust and single-chamber dual-thrusts SRM design. Conceptual design of two examples is realized. The design task includes system design, propellant grain design, chamber design, nozzle design and igniter design.The research in this dissertation provides great impetus to SRM design for conceptual demonstration. It can realize knowledge inheritance, improve the diversity of feasible design results, and enhance the design speed. The concepts and methods described in the dissertation are quite general and will be applicable to other engineering systems.