Fault Detection and Diagnosis for Thermal Energy Storage HVAC System

本研究计画之目的在于探讨具有蓄热功能之空调系统之最佳运作方式。为了实现故障检测与诊断，研究了存储操作的性能指标、故障检测与诊断工具以及计算机仿真优化操作。以下是所得结果的总结。1.提出了用于评价实际系统运行状况的调峰运行指标。并通过实际系统的实测数据验证了这些指标的有效性.将故障检测与诊断的模糊溯因算法与故障原因和征兆知识库相结合，构成了一个专家系统。研究了隶属度函数对诊断结果的影响.根据实际系统的实测数据，对蓄能系统中冷水机组和空气处理机组的物理模型参数进行了辨识。并将这些模型与蓄热水箱的混合模型相结合，编制了一个模拟运行工况的计算机程序，作为故障改进的成本/效益分析工具.利用水箱混合模型和实测数据，预测了水箱内的温度分布。并通过收敛计算得到预测温度分布的最小误差，估算了通过水箱、冷水机组和空气处理机组的流量，以及实际系统中冷水机组和空气处理机组的特性。这个程序被发现允许准确地模拟运行条件，没有任何物理模型的冷水机组和空气处理单元。

The objective of this research project is the optimum operation of HVAC system having thermal energy storage without any faults. In order to realize the fault detection and diagnosis, the performance indices of storage operation, the tool for fault detection and diagnosis and the optimum operation with computer simulation were studied. The following is the summary of the obtained results.1. The indices for the peak shift and peak cut operations were presented to evaluate the operation condition in actual systems. And, the usefulness of these indices were validated by using the measured data in actual systems.2. An expert system was constructed by combining the fuzzy abduction algorithm for fault detection and diagnosis with knowledge database for reasons and symptoms of faults. And, the effect of the membership function on the obtained diagnosis result was studied.3. The model parameters of physical models for chiller and air-handling unit in thermal energy storage system were identified from the measured data in actual systems. And, these models were combined with the mixing model of water tank for thermal energy storage, and a computer program allowing to simulate the operation condition was built up as a tool for cost/benefit analysis of fault improvement.4. The temperature profile in water tank was predicted by using the mixing model of water tank and the measured data. And, by convergence calculation to obtain the minimum error of the predicted temperature profile, the flow rate through the water tank, chillers and air-handling units were estimated as well as the characteristics of chillers and air-handling units in actual systems. This procedure was found to allow to simulate the operation condition accurately without any physical models for chillers and air-handling units.