KPCA and thinking optimization BP neural network-fused total water consumption prediction method

Abstract

The invention discloses a KPCA and thinking optimization BP neural network-fused total water consumption prediction method. The method comprises the following steps of determining a prediction factorby using a correlation coefficient method; using the kernel principal component analysis (KPCA) for carrying out dimension reduction processing on prediction factors, and solving nonlinear characteristics among data; establishing a total water consumption prediction model by using the BP neural network; and predicting the total water consumption by using the model. The first two steps of the method are data preprocessing, and the method aims to extract the useful information in the water data and eliminate the interference of the redundant information on prediction, the third step is to put the BP neural network into the prediction of the total water consumption, and optimize the weight and the threshold value of the BP neural network by adopting a thinking evolution learning algorithm atthe same time; and the last step is used for checking the model effect. According to the method, experiments are carried out in the annual open statistical water consumption data of the national statistical bureau, and results show that the water consumption total amount prediction model based on the KPCA and thinking optimization BP neural network can well predict the future water consumption total amount.

Description

technical field [0001] The invention relates to a technology for predicting total water consumption, in particular to a method for predicting total water consumption that integrates KPCA and thinking-optimized BP neural network. Background technique [0002] At present, in some areas, the lack of water resources can no longer meet the needs of population growth and industrial and agricultural development, which seriously hinders the social and economic benefits of the area. Therefore, it is extremely urgent to plan and dispatch water resources rationally. The forecasting of total water consumption is an important measure for the rational allocation, regulation, analysis and estimation of water resources under the severe situation of water resources today, and plays an important role in the sustainable development of social economy. [0003] Total water use forecast involves many aspects: industrial water use forecast, agricultural water use forecast, and short-term and long-...

AI Technical Summary

Problems solved by technology

Literature [Shan Jinlin, Dai Xiongqi, Li Jiangtao. Using BP network to establish a model for predicting urban water consumption [J]. China Water Supply and Drainage, 2001,17(8):61-63.] Using BP neural network to predict urban water consumption , and the water consumption data are arranged in time series order, but this prediction method is only suitable for short-term water consumption prediction and ignores the possible uncertain factors of water consumption, so the results are not so accurate

Literature [Sang Huiru, Wang Lixue, Chen Shaoming, et al. Application of RBF neural network based on principal component analysis in water demand forecasting [J]. Hydropower Energy Science, 2017(7): 58-61.] using principal components The analysis reduces the dimensionality of each factor that affects the total water consumption. However, the principal component analysis is a linear transformation. The data of the total water consumption factors are nonlinear, complex, and uncertain. Although the linear dimensionality reduction can remove the redundancy information, but some useful information will be lost

Images