Use Of Data Mining Techniques For Short-term Generation Scheduling Of Cascaded Hydropower Plants

The daily task of electrical industry is scheduling of power system.The purpose of scheduling is to find the feasible generation schedules that minimize the total cost and increase the power generation of hydro power plant.The electric power industry is facing other problem called unit commitment.The objective of unit commitment is to find the feasible schedules of each unit in each hour while satisfying the demand.In this study various techniques mathematical optimization like dynamic programming,nonlinear approach,fuzzy logic,genetic algorithm,mixed integer programing,artificial neural network and classification algorithms like decision tree are discussed.The dynamic programming is good in dealing with different constraints but its drawback is curse of dimensionality.The lagrange relaxation method handles the pumped storage hydropower plants constraints and give feasible results The head sensitive cascaded hydropower is better deal by mixed integer linear programming,mixed integer non-linear programming.For obtaining the results in a short time genetic algorithm is used.Moreover,in comparison with mathematical and classification techniques,the classification techniques like decision tree algorithms are used for finding the quick schedules of generation and optimal unit dispatch.The challenging task for hydro-based power systems is to solve the short-term scheduling of cascaded hydropower plants,until now the decision tree algorithms are not used for finding the generation schedules and generation of each unit of hydropower plant.In cascaded hydropower plants,the small change in water level of upstream hydropower plant results in large fluctuation of the downstream hydropower plant due to its poor regulation-ability.The objective of current study is to find a fast and practical method for predicting and classifying the future schedules of hydropower plants by using decision tree algorithms.The proposed method consists of data mining techniques and approaches.First,the energy production and generation of units is determined for upstream and downstream hydropower plant by using multiple linear regression.Then,the cluster analysis is used to find the generation curve with the help of historical data.The decision tree algorithms C4.5,ID3-IV,improved C4.5 and CHAID are adopted to predict the quick generation schedules and comparison between different algorithms are made.The unit generation schedules are also determined by C4.5,results showed the decision rules from which speedy unit schedules are determined for each hour,the optimal unit dispatch is helpful in increasing the unit generation.The decision tree algorithms are solved by SIPINA software.Results show that the C4.5 algorithms is more feasible for rapidly generating the schedules of cascaded hydropower plants.The proposed method is helpful for the researchers to make fast decisions in order to schedule cascaded hydropower plants every day.