论文部分内容阅读
为了实现计算机实时联机预报,需要将手编预报方案的“相应水位法”形成数学模型。由于上下游水位的非线性现象,用多元非线性回归分析建立这种模型往往比较困难。如果在上下游水位站各自建立虚拟的H~Q关系,先将水位转换成流量作多元线性回归计算,再将计算的虚流量反推水位。这样就可以基本克服这种非线性现象。另外,还可以将上游洪峰水位的前一时段、前二时段以及下游同时水位的前一时段、前二时段水位均作为预报因子转换为虚流量计算。经过这种处理后的预报精度比原手工编制方案的预报精度有了显著的提高。
In order to realize real-time on-line forecasting of computers, it is necessary to form a mathematical model for “corresponding water level method” of hand-knit forecasting schemes. Due to the non-linear phenomenon of upstream and downstream water level, it is often difficult to establish such a model by multivariate nonlinear regression analysis. If each of the upstream and downstream water level stations to establish their own virtual H ~ Q relationship, the first water level into multiple linear regression calculation, and then calculate the virtual backflow of water level. This can basically overcome this non-linear phenomenon. In addition, the upstream flood peak water level before the first period, the first two periods and the lower reaches of the same time before the water level, the first two periods of water level as a forecast factor into a virtual flow calculation. After this treatment, the accuracy of forecasting is significantly improved compared with that of the original hand-made plan.