A method for optimizing groundwater depth of BP neural network based on genetic algorithm

Abstract

A method for optimizing groundwater depth of BP neural network based on genetic algorithm includes collecting data, analyzing and screening independent variable by set pair, and taking samples corresponding to said target independent variable; the output result of BP neural network is obtained by setting BP neural network; setting initial parameters of genetic algorithm; individual selection operation: In the old population, the preset probability is used to select the individual to produce the next generation. The selection principle is as follows: selecting according to the fitness value ofthe individual from large to small; crossover: the generation of new individuals through the crossover of chromosomes; mutation: An individual selected from a population to mutate a segment of a chromosome to produce a new individual; the fitness is calculated and compared with the original population when the number of evolution reached the upper limit. The chromosome corresponding to the fitnessoptimal solution is the threshold value and weight value corresponding to BP neural network. The application of the embodiment of the invention can improve the accuracy of the groundwater burial depth.

Description

technical field [0001] The invention relates to the technical field of groundwater depth prediction, in particular to a method for predicting groundwater depth based on genetic algorithm optimization of BP neural network. Background technique [0002] The groundwater system is a complex system involving multiple factors. It is affected by many physical factors such as atmospheric circulation, solar activity, previous hydrometeorological elements, and regional cover changes. The short-term prediction of groundwater depth is an important content of regional water security management. It is also an important subject of long-term research in the water science community. In recent years, BP neural network and set pair analysis have been widely used and promoted in water resources. Wu Hongbin established a multiple linear regression model to dynamically predict the groundwater quality of Hailongba in Zunyi City. The results show that the prediction accuracy is high. The establishe...

AI Technical Summary

Problems solved by technology

However, most of the current studies on groundwater depth prediction focus on the introduction of new model methods, and seldom start from the screening of groundwater independent variables and improving the representativeness of input factors. The selection of predictive model independent variables is an important basis for improving the prediction level of groundwater depth. The sample selection of the input layer is less representative and may have a greater impact on the predicted results. How to retain the more representative independent variables is the key

At the same time, the ordinary neural network is very sensitive to the initial weight, it is easy to converge to the local minimum, and the global search ability is poor.

After self-adaptive improvement, although the BP neural network algorithm can solve the problems of slow convergence speed and local minimum of the BP network, it still cannot completely overcome the inherent defects of the BP neural network algorithm in practical applications.

Images