| Time series is always affected by astronomy, geographic, artificial activity, etc. when it isapplied in the domain of hydrology and water resource. As a result, the data of hydrology and waterresource have some characteristics as non-linear, versatile, random attribute, and so on. At the sametime, it has periodic nature and tendency nature. These problems are resolved hardly through thenormal linear model. In order to solve these intricate problems, several kind of wavelet neuralnetwork (WNN) has been brought forward. Using MATLAB, the computer language, we haveestablished the mathematical models. Then the models are applied and the good effect has beentaken. The innovations in the article:1. The tight wavelet neural network based on morlet wavelet function has been improved atthe following aspects: data invert, parameters set up, parameters adjustment. The model is appliedfor the first time to predict the amount of paddy rice needing water. The good effect has beenobtained.2. The pulse coupled neural network (PCNN) is called that the third generation artificial nervenet. Based on the original abbreviate model, the model is modified profoundly. The improve PCNNevaluation model for the water resource using condition and the PCNN annual rainfall forecastingmodel based on a trous are proposed at the first time. There are great innovations theoretically.3. The ant colony optimization is the new produce in the domain of modem bionics. Themethod has global optimization ability. The conjuncture point of the ant colony optimization andthe Marr WNN has been found. The model is applied to predict the groundwater level. The goodeffect has been obtained.Based on the above established model, the theory and the practice have been associatedorganically. The research result offers new method for the complex non-linear problem in thedomain of hydrology and water resource. At the same time, the applied range of WNN model isamplified. Moreover, the theory is improved accordingly.
|
PDF Full Text Request