Research And Application Of Spares Demand Forecasting For Water Diversion Run-of-River Hydropower Plant

Hydropower, as the major regenerated power we have now, has been energetically supported by our government for its ability to provide a green and cheap electricity supply. Water diversion run-of-river hydropower plant plays a critical role in the hydropower industry thanks to its comparatively low construction cost and environmentally friendly features. However, it is quite dependent on its equipment reliability since it generates power based on its influent flow while without reservoirs to control it. A proper amount of spare storage helps improve the equipment reliability by facilitating the timely equipment maintenance and reducing the equipment downtime. Nevertheless, too much spare storage can cause unnecessary floating capital and extra work for warehouse keepers. Consequently, a proper amount of spare storage is of significance. Furthermore, a precise spares demand forecast ensures a proper amount of spare storage to a large extent. Currently, water diversion run-of-river hydropower plant adopts the method of moving averages to forecast its spare demand, which provides an undesirable accuracy and leads to an excess amount of spare storage. To ensure a high accuracy of the spares demand forecasting and thus get a proper amount of spare storage, this paper focuses on the spares demand forecast research for water diversion run-of-river hydropower plant. The main research mainly includes the following sections:First, illustrate the spares demand forecasting theories including the definition, the function and the regular methods of spares demand forecasting; investigate the current spares demand forecasting situation for water diversion run-of-river hydropower plant; analyze reasons leading to spares demand for water diversion run-of-river hydropower plant, and classify the features of its spares demand and spares demand forecasting; based on this, put forward the technical route for spares demand forecasting and explain its key questions..Second, study the spares demand forecasting models for the water diversion run-of-river hydropower plant, which has four major parts: firstly determine the object of the spares demand forecasting, and construct the influencing factors set for the spares demand in water diversion run-of-river hydropower plant, then quantify these factors based on its spares demand features; secondly, build the spares demand forecasting model by using Rough Set and Least Squares Support Vector Machine Algorithm(LS- SVM); thirdly, describe how to solve the model in a detailed way and introduce into the adaptive genetic algorithm during the parameter setting period to seek for an optimized solution; fourthly, conduct the model evaluations in terms of its demand time accuracy and demand data accuracy.Finally, apply this spares demand forecasting model into practice for some hydropower enterprise, and verify its effectiveness and accuracy for the clip-on butterfly valve.