GRU neural network-based residential water consumption prediction system and prediction method

Abstract

The invention discloses a GRU neural network-based residential water consumption prediction system and prediction method, and the system comprises: a prediction model construction module which is used for constructing a two-layer GRU residential daily water consumption prediction model; a prediction model training module which is used for collecting influence factors such as daily water consumption of residents, weather types, highest and lowest temperatures and workday types, and inputting feature vectors corresponding to the influence factors into the two-layer GRU network to achieve training of a daily water consumption prediction model of the residents; a test prediction model module which is used for taking the feature vectors corresponding to the influence factors as test data and inputting the test data into a trained resident daily water consumption prediction model to predict the resident daily water consumption; and a water supply management module for managing the daily water supply amount of the residents according to the prediction. According to the invention, daily water supply management of residents can be timely and accurately carried out according to prediction, early warning is realized, countermeasures are taken, and normal water consumption of residents is guaranteed.

Description

technical field [0001] The invention belongs to the technical field of residential water use prediction methods, and in particular relates to a residential water use prediction system based on a GRU neural network, and also relates to a residential water use prediction method based on a GRU neural network. Background technique [0002] With the rapid development of society, the water demand of residents is constantly increasing, and the forecast of residential water use is the basis for the water supply department to ensure water supply and scientific dispatch. In recent decades, many countries and cities have actively carried out research on methods and systems for forecasting residential water use due to water shortages. [0003] The traditional residential water consumption forecast is based on the year and month, which cannot provide a basis for daily water supply scheduling; however, the daily water consumption of residents is nonlinear and non-stationary, and the daily...

AI Technical Summary

Problems solved by technology

However, when processing longer data sequences, RNN is easily troubled by gradient disappearance or gradient explosion, so that the training cannot propagate the gradient down.

The LSTM model is an improved RNN model. It adds a cell control mechanism on the basis of the RNN model to solve the long-term dependence of RNN and the gradient explosion problem caused by too long time series. It can remember long-term intervals Historical data information, but due to the long training time of the model, it cannot be used in time to predict residents' daily water consumption

Images