Abstract
The optimal hydrothermal scheduling is the distribution of load among the generating stations. The objective function for the problem is to minimize total generating cost considering the electrical and hydrological constraints. This problem is usually solved in two stages. The first is to find the hydropower generation share, then to find the thermal generation share. This research concerns the first stage. This work uses the Artificial Neural Network (ANN) to find the optimal scheduling of the monthly water discharge over the year. The main power station of Mosul dam is considered as an application example for this study. Six input variables are chosen to be the input to the ANN. They are monthly inflow water, monthly demand water, number of the month in a year, expected next year water inflow, available stored water (water from the past year). The ANN is trained and tested by the available water flow data over the past 65 years (1931- 1995). It is found that this technique enables the utilization of whole inflow water for most of the years (within considered constraints) in spite of the great fluctuation of inflow water for these years. Besides, this technique takes into account the status of the water for last year and next year in addition to the year under study. This means that the water distribution improves the utilization of available water over three years
