WATER DISTRIBUTION QUANTITY PREDICTING METHOD AND DEVICE

PURPOSE: To apply a water distribution quantity predicting method even if the number of data is small and quickly respond to the changes of the environment by determining the predicted value of the periodic fluctuation component of the distribution quantity, determining the remainder component except the fluctuation component with a self-regression model, and using the sum of two predicted values as the distribution quantity predicted value. CONSTITUTION: Water is distributed from a distributing reservoir 1 via natural flow-down. The distribution flow is detected by a distribution detector 2, and the actual result, weather, predicted weather, and predicted temperature are inputted to an input device. The input data collected by a measured value storage device 5 are accumulated, and average fluctuation patterns for the day of the week, time, and weather are generated by an average fluctuation pattern generating device 6. The correction coefficient for the rainy day is determined by a correction coefficient generating device 7 based on the ratio between the average values of the distribution quantities on rainy days and on days other than rainy days in the past. The stored measured value of the measured value storage device 5, the average fluctuation pattern of the average fluctuation pattern generating device 6, and the correction coefficient are inputted to a system variable data generating device 8, and the data excluded with the effect of weather and day of the week and the system variable data of temperature are generated.