Neural Network Based Predication and Optimization for Groundwater / Surface Water System

Abstract

The present invention relates to a method and apparatus, based on the use of a neural network (NN), for (a) predicting important groundwater/surface water output/state variables, (b) optimizing groundwater/surface water control variables, and/or (c) sensitivity analysis, to identify physical relationships between input and output/state variables used to model the groundwater/surface water system or to analyze the performance parameters of the neural network.

1 . A method of predicting output/state variables in a groundwater/surface water system, comprising the steps of (a) providing and training a neural network with hydrologic and control data for a groundwater/surface water system, and (b) operating the trained neural network to predict output/state variables in the groundwater/surface water system, the output variables comprising one or more of the following: surface water elevations, surface water flow rates, groundwater heads and elevations, groundwater gradients, groundwater velocities, and chemical concentrations.  
   
   
       2 . A method as defined in  claim 1 , wherein the step of providing and training the neural network further comprises providing and training the neural network with meteorological data.  
   
   
       3 . A method as defined in  claim 2 , wherein the step of providing and training the neural network further comprises providing and training the neural network with water quality data.  
   
   
       4 . A method as defined in  claim 1 , wherein the step of providing and training the neural network further comprises providing and training the neural network with water quality data.  
   
   
       5 . A method as defined in  claim 1 , wherein the groundwater/surface water system is a groundwater system, and wherein the neural network is operated to predict output/state variables, at least one of which comprises groundwater elevation/head.  
   
   
       6 . A method as defined in  claim 1 , wherein the groundwater/surface water system is a surface water system, and wherein the neural network is operated to predict output/state variables, at least one of which comprises surface water elevations.  
   
   
       7 . Apparatus for predicting output/state variables in a groundwater/surface water system, comprising (a) a neural network that has been provided and trained with hydrologic and control data for a groundwater/surface water system, and (b) the trained neural network being configured to predict output/state variables in the groundwater/surface water system, the output variables comprising one or more of the following: surface water elevations, surface water flow rates, groundwater head and elevations, groundwater gradients, groundwater velocities, and chemical concentrations.  
   
   
       8 . A method of providing sensitivity analysis for a neural network for a groundwater/surface water system, comprising the steps of (a) providing and training a neural network with input data for a groundwater/surface water system, the trained neural network configured to produce output comprising output/state variables of the groundwater/surface water system, and (b) operating the neural network to define relationships between selected input data and selected output variables.  
   
   
       9 . A method as set forth in  claim 8 , wherein the input data comprises a plurality of input variables of the groundwater/surface water system, and the step of operating the neural network comprises operating the neural network to define physical relationships between a least one of the real input variables and at least one of the output variables of the groundwater/surface water system.  
   
   
       10 . A method of providing transition equations for an optimization procedure for a groundwater/surface water system, comprising the steps of (a) providing a trained neural network that has been trained with hydrologic and pumping data for the groundwater/surface water system and (b) configuring the neural network to produce transition equations that can be used either as function routines or constraints in the optimization procedure.  
   
   
       11 . An apparatus for use in an optimization procedure for a groundwater/surface water system, comprising a trained neural network that has been trained with hydrologic and control data for the groundwater/surface water system and is configured to produce transition equations that can be used either as function routines or constraints in the optimization procedure.