Research On Knowledge-Based Scheme Design Technologies Of Circulating Fluidized Bed Boiler

Circulating Fluidized Bed (CFB) is a new generation of clean combustion technology with high-efficiency and low-pollution in recent twenty years. CFB can not only realize 90% desulfurization efficiency, low emission of NOx and similar combustion efficiency compared with pulverized coal fired boiler, but also has many advantages, such as wild fuel adaptability and excellent performance of load adjustment. Therefore, it gets rapid and worldwide dissemination for business.The Scheme design of CFB boiler is a complex, highly empirical, long-term, highly reliability and economical efficiency required project, which needs researchers rich in the relative knowledge and experience. It is one of the frontier topics in the technique of computer aided designing boilers that stores the knowledge of CFB boiler in the way of digitalization to enhance the intelligent process of CFB boiler design, shorten design period and reduce design cost. The paper was in the subsidization of "eleventh-five national science and technology support plan projects" (2006BAF01A46) and Zhejiang Technology Plan Program (2006C21SA160008) and researched the computer aided method of CFB boiler scheme design. In addition, it combined the method of knowledge engineering to explore an intelligent solving technology supporting the whole research cycle of scheme design. The main research contents on this subject were explained as follows:Based on the cognition of the research process, the scheme design of CFB boiler was divided into four courses: scheme design, scheme analysisg, scheme evaluation and scheme optimization. Moreover, an integrated method of intelligent design was applied to the research: the scheme was established by the method of case study based reasoning; the model of scheme was structured and solved based on the theory of process system; the scheme was evaluated by the method of rule based reasoning. In the end, the scheme would be optimized according to the evaluated results.The scheme design of CFB boiler need to do a lot of calculations, such as thermal calculation, smoke resistance calculation and (hydrodynamic calculation. Design a model which include all calculations by adopting the thinking of integration, realize the function of data transferring, code reusing and tools sharing in calculations. Cognate the CFB boiler as network model by the made of part element and fluid dram line in the view of process system, using sequential modular method to solve the model automatically.Propose a CFB boiler heat transfer calculation model based on experience. The model solved the problem that ash concentration and heating surface were inhomogeneous distribution by the method of divide the furnace into serial sections in the height direction. Picked out empirical parameters, such as heat transfer coefficients, ash concentration and combustion fraction of fuel, derivate general formulas in heat balance calculations of section, calculate the change of heat, ash, smoke components of furnace in the condition of desulfurization.Propose case based reasoning method to structure the model of design and modify scheme system. Firstly, disassembled past design scheme, proposed a model of level-division and multi-library to storage cases. The library model was beneficial to heighten an effort and improve efficiency. Secondly, adopted the method of dynamic configuration to customize the index. In the end, proposed a kind of nearest neighbor algorithm based on mathematical statistics to retrieval cases. The nearest neighbor algorithm could find out the most similar cases automatically without setting feature weights.Propose rule based reasoning method to evaluative scheme CFB boiler. Cognized and classified the knowledge in process of evaluation, decomposed the process of evaluating scheme into three sub processes: correctness checking of inputted model, feasibility judgment of calculation results and performance analysis of scheme. Inference engine used control strategy of goal-driven, which only search the related rules, to improve efficiency and accuracy.