建筑热工系统模糊-神经网络控制理论与技术开发

Nowadays, there are many problems in building thermal systems and equipments such as weak self-adjust of energy and low efficiency. In addition, traditional control systems are of no effect to systems with characteristic of multi-variables, non-linear, long-time delayed, multi-input and multi-output and quick time variant performance. In order to solve all these problems above, this item, consulting studying status in overseas and our country, takes fuzzy-neural network theory for real-time control system as the key, and it consists of study of intelligent control theory，development of single-chip machine and computer, experiment of building thermal systems and equipments and the application of intelligent control theory in related field. Details are as followed.(1) Study on fuzzy-neural network theoryFunctioning-fuzzy-subset inference (FFSI) and the method of the true-value -assigned fuzzifying have been proposed, firstly. The estimated method of the fuzzy control rules coherence has been erfected, secondly. Then, the self-optimized method of fuzzy control rules has been put orward based on fuzzy subsets coherence; Further more, the fuzzy control method of self-tuning in domain has been perfected, and the rules self-tuning fuzzy controller on the basis of the trajectory performance on the phase plane has been presented; The multi-step forecast method based on BP and Elman neural network and the fuzzy neural network multi-step forecast method have been proposed.(2) Development of fuzzy controller based on single-chip machine and computer Single-chip fuzzy controller and multi-channel data sampling system have been developed, and the self-switch of their multi-channel input data has been realized; The method of realization of fuzzy control based on FFSI by hardware has been studied, and the design theory and method of digital/analog mixed inference have been presented, the simulation, subsequently, of the HSPICE electrocircuit characteristic, Interface circuit from inference machine to user circuit has been designed next; Fuzzy neural network control system based on industry computer has been developed. .(3) Study on design theory of heat-exchanger for fuzzy neural network control and its and xperiment Heat-exchanger with an incremental adjustable structure and the design of its minimum heat-exchanging unit have been proposed, and heat-exchanger with an incremental adjustable structure has been manufactured; The method of the AHU design with an adjustable structure in heat-exchanger, exchange in working order and control and adjust have been presented, and the AHU and the single-chip machine control system have been completed; The computer simulation model of air -conditioning system and test room has been developed, and their thermal characteristic has been analysed; The test system of indoor environment intelligent control has been established, and the study on intelligent control of indoor temperature by test has been completed..(4) Application of intelligent control theory in related fields (grain drying)..Grain drying and building thermal system are with the same characteristic of multi-variables, long-time delayed and non-linear, so our group, according to what have studied above, complete the sub-item of "Grain dryer special item" from National Grain Bureau -"The intelligent control and measurement system of 10 thousand million kilogram grain dryer" and 6 example systems. Till the end of 2002, this system have been run for 2 drying periods steadily in Boli and Lanxi of Heilongjiang province, Xiersu, Xizi and Tuquan of Neimengu province and Kaiyuan of Liaoning province, and gained a lot of benefites.After researches listed above, a postdoctoral report has been finished, and 10 papers in core journal and above, 7 papers in international symposium and 1 paper in national symposium have been published. At the same time, a single-chip fuzzy controller, 1 multi-data collecting apparatus, 1 heat-exchanger with djustable structure, 1 AHU with an adjustable structure and 1 test bed for intelligent control

研究建筑热工系统模糊-神经网络在线控制理论和方法;以热湿交换设备为对象,试验研究该刂评砺鄣氖导视τ眉际?研制开发单片机模糊-神经网络控制器及新型热湿调控装置.将理论研究,装置开发和试验研究紧密结合,为解决非线性,大滞后和多输入多输出系统的控制提供砺刍?对推动该控制理论的发展和实现机电一体化具有重要学术和应用价值.