Neural network based predication and optimization for groundwater/surface water system
OwnerEmery J. Coppola, Jr., Mary M. Poulton +1inventors
InventorsEmery J. Coppola, Jr., Mary M. Poulton, Ferenc Szidarovszky
Life of this patent
Estimating the term…
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.
Claims9
A method of providing transition equations for an optimization procedure for a groundwater or surface water system, comprising the steps of (a) providing a trained neural network that has been trained with real world data for the groundwater or surface water system, which real world data includes time dependent variables that may change over stress periods and comprises state and control variables for the groundwater or surface water system, the trained neural network being interfaced to an automated on-line data system that provides real time input to the trained neural network, (b) configuring the trained neural network to produce transition equations that can be used as any or all of function routines and constraints in the optimization procedure, (c) initializing time dependent variables that are input to the transition equations of the trained neural network from the automated on-line data system to real time conditions for an initial stress period and running the optimization procedure to compute the optimal values of the control variables and their predicted effect on the system over the initial stress period, and (d) using the optimal values to control the system in the most effective or appropriate manner, given the objectives and constraints of the system.
A method of providing transition equations for an optimization procedure for a groundwater or surface water system, comprising the steps of:
- (a) providing a trained neural network that has been trained with real-world data for the groundwater or surface water system, which real world data includes time dependent variables that may change over stress periods and comprises state and control variables for the groundwater or surface water system, the trained neural network being interfaced to an automated on-line data system that provides real time input to the trained neural network,
- (b) configuring the trained neural network to produce transition equations that can be coupled or combined with interpolation equations in the optimization procedure,
- (c) initializing time dependent variables that are input to the transition equations of the trained neural network from the automated on-line data system to real time conditions for an initial stress period and running the optimization procedure to compute the optimal values of the control variables and their predicted effect on the system over the initial stress period, and
- (d) using the optimal values to control the system in the most effective or appropriate manner, given the objectives and constraints of the system.
A method of providing transition equations for an optimization procedure for a groundwater or surface water system, comprising the steps of:
- (a) providing a trained neural network that has been trained with real-world data for the groundwater or surface water system, which real world data includes time dependent variables that may change over stress periods and comprises state and control variables for the groundwater or surface water system, the trained neural network being interfaced to an automated on-line data system that provides real time input to the trained neural network,
- (b) configuring the trained neural network to produce transition equations that can be coupled or combined with physical equations in the optimization procedure,
- (c) initializing time dependent variables that are input to the transition equations of the trained neural network from the automated on-line data system to real time conditions for an initial stress period and running the optimization procedure to compute the optimal values of the control variables and their predicted effect on the system over the initial stress period, and
- (d) using the optimal values to control the system in the most effective or appropriate manner, given the objectives and constraints of the system.
Dependencies are inferred from the claim text.