Study On Optimization Control Techniques Of Heating Ventilating Air Conditioning

With gradual popularization of heating ventilating air conditioning (HVAC), energy consumption of HVAC commonly occupies more than 50 percent of energy consumption of the whole building. In fact, a majority of HVAC systems inefficiently runs, and wastes a mass of energy. The reason of the phenomenon is that loads of HVAC are calculated according to the most loads, and it adopts invariable working point and parameters of controller to devise the whole HVAC system including some characters of nonlinearity, time-variety, time-delay, big inertia and so on.According to HVAC characters of nonlinearity> time-variety and so on, the thesis adopts method basing on radial basic function (RBF) fuzzy neural network to build dynamic model of HVAC. The dynamic model of HVAC can adapt to variety of run environment and disturbances action, and satisfy demands of HVAC control algorithm and optimization of HVAC system. Taking into account the PMV value, using BP neural network designs optimal room temperature, the optimal temperature offers desired trajectory of temperature to generalized predictive control of HVAC. According to characters of nonlinearity, time-variety, long time-delay, big inertia and so on, a generalized predictive controller basing on radial basic function (RBF)fuzzy neural network is designed. A RBF fuzzy neural network is continuously on-line revised in the process of control to offer information that on-line roll optimization needs. The controller finishing roll optimization is realized by RBF fuzzy neural network to reduce calculating. The controller of RBF fuzzy neural network can obtain optimal rule of control by on-line optimization using some information that the predictive network offers and beforehand designing goal function.Simulation experiment proves that modeling basing on RBF fuzzy neural network can accurately obtain the dynamic model of HVAC. Using BP neural network can design optimal room temperature satisfying requirements of human body. At the same time, the thesis designs the generalized predictive controller basing on RBF fuzzy neural network that can regulate comfortable room temperature and reduce energy consumption.