Filter‐Based Short‐Term Electric Load Prediction Considering Characteristics of Load Curve

Abstract

This paper deals with  ∞ filter-based short-term electric load prediction taking into consideration the characteristics of the load curve. We propose a predictive method to forecast the future electric load demand for 36 h from 12:00 PM, and evaluate the peak and bottom of the load curves on the next day. We propose a load model, estimate the unknown parameters of the model by means of an  ∞ filter using the data separated for nonworking days and weekdays, with the same pattern of the previous data chosen and assigned to the model parameters. The simulation results show the effectiveness of the proposed prediction methodology. C⃝ 2014 Wiley Periodicals, Inc. Electron Comm Jpn, 97(12): 1-10, 2014; Published online in Wiley Online Library (wileyonlinelibrary.com).

Short‐Term Wind Power Prediction for Wind Turbine via Kalman Filter Based on JIT Modeling

“…By establishing a model using this data, in this paper, we proposed a prediction method guaranteeing the convergence of unknown coefficient, which reduces the amount of data used and the computational burden as compared to a conventional method [9] and decreases the prediction error. By processing the data to be used for prediction by JIT modeling, the data to be used for the predictions were turned into data closer to the weather forecast at that prediction location.…”

“…By processing the data to be used for prediction by JIT modeling, the data to be used for the predictions were turned into data closer to the weather forecast at that prediction location. By establishing a model using this data, in this paper, we proposed a prediction method guaranteeing the convergence of unknown coefficient, which reduces the amount of data used and the computational burden as compared to a conventional method [9] and decreases the prediction error. Through reduction of the error covariance in the estimation of unknown parameters, we were able to demonstrate convergence of the unknown coefficients and to achieve more reliable predictions.…”

“…In Ref. [9], the unknown coefficients are estimated using 2-month portion of data. However, there is a problem, in which the unknown coefficients do not converge, which occurs as a result of estimations using past data distant from the nearest data.…”

“…[9] is that it involves a method capable of predictions based on 2-month portion of data, and is useful for wind power generation predictions, in which power generation output fluctuates significantly and available data are insufficient. The feature of Ref.…”

“…The feature of Ref. [9] is that it involves a method capable of predictions based on 2-month portion of data, and is useful for wind power generation predictions, in which power generation output fluctuates significantly and available data are insufficient. Specifically, the proposed method represents a wind speed model and wind power generation model using regression equations, and estimates the unknown coefficients using a constrained Kalman filter based on past wind speed and wind power generation levels.…”