A Fuzzy Compositional Inference Rule for Real-Time Dynamic Operating Reservoir Normal Elevation Based on Forecast 24h Rainfall

To deal with the fuzziness in hydrology, fuzzy set theory and method have been widely used for decision making in last decades. For the purpose of more water conservancy without adding flood risk, the method of dynamic operating reservoir normal elevation (RNE) has taken an important role in recent years. While dynamic operating RNE, fuzzy compositional inference (FCI) can be used. Taking Biliuhe Reservoir as an example, we form a FCI rule to real time operate RNE. Firstly, we analyze the uncertainty of forecasted 24 h rainfall. Secondly, we determine the upper boundary of dynamic operating range of RNE with pre-discharge model. Thirdly, we form the macro and miner premise of FCI. Fourthly, we build an inference structure. And finally, we develop the FCI rule. By an example, we describe in detail how the FCI rule works.