QUASI REAL-TIME ECONOMICAL OPERATION OF HYDROPOWER PLANT

摘要

Optimum regulation can improve the economic benefits for hydropower plant. Economical Operation of Hydropower Plant (EOHP) is one type of optimum regulation in a day, which can improve the economic results at least 1%. It aims at dispatch the load among the units of hydropower plant. As the time-interval being 1 hour in load diagram supplied by power system, EOHP always selects 1 hour as calculating time-interval in the past. That means EOHP calculates at the beginning of every hour to dispatch load for this hour. In this way, the real-time system of EOHP cannot be obtained. With the development of power system and monitoring and controlling technique of hydropower plant, the load can be forecasted in a shorter period and real-time EOPH can be realized step by step. So a quasi real-time EOHP is proposed, which is close to real-time EOHP. Two improvements are made to realize quasi real-time EOHP, one is to improve accuracy of EOHP, and the other is to increase the speed of calculation. EOHP distributes load according to the load requirement of the electrical network and state of water level as well as situation of each hydroelectric power unit. The accuracy of EOHP is affected by unit state and calculation speed greatly. In order to gain the correct state of each water turbine generator units, in this paper a radial basis function (RBF) network is proposed to forecast the flow consumption. Efficiency test data for typical water heads is used to train the network. Then groups of water heads and generator output data are used as input vector of the network, with the flow consumption as outputs vector, to train the RBF network. After the network being trained, the proper weights and biases are obtained, with which the flow consumption of water turbine generator unit can be forecasted by the network. In practice, the weights and biases of the RBF network can be modified by the last period water head and generator unit output. According to the change of water heads and electrical loads, optimal calculation of EOHP is realized by dynamic programming in this paper. If the changes reach the computational condition, optimal calculation of EOHP will begin to dispatch the load for each unit; else the operating states of units in last period will be continued. Improvements in these two respects can increase accuracy and calculation speed of EOHP. In a case study, the speed of calculation exceeds the speed of necessary changes of the controlled hardware. Thus quasi real-time economic hydropower plant operation is achieved.