MEDIUM RANGE LOAD FORECASTING USING DIGITAL IMAGE PROCESSING TECHNIQUES

Medium range forecasts of hourly load demand for a span of 24 hours to 168 hours (one week) are required for the preparation of short term maintenance schedules of unit auxiliaries and peaking stations apart from maintaining security constraints and minimising operational costs. The varying nature of power system data having multiple periodicity of 24 hours/168 hours makes it suitable for the application of Digital Image Processing technique whereby hourly forecasts for a span of one or two weeks can be made with equal variance of forecast error. An attempt has been made to represent one year data in the form of image replacing the time variables by space variables. Thus the inter pixel gap of the image represents the sampling time of 1 hour along the horizontal axis and 24 hours along the vertical axis. 2-Dimensional image data representing one-year load is suitably augmented to represent 26 blocks each of two weeks load data consisting of 16 days load. First 25 blocks (50 weeks) data is used for modelling by time series analysis using Discrete Cosine Transform (DCT). The predicted DCT coefficients for 26th block are inverse transformed for the forecast of the hourly load data for 51st and 52nd week of the year.

کلیدواژه ها:

Medium Range Forecasting ، Time Series Analysis ، Discrete Cosine Transform