Research Of Steady-state Power Quality Forecasting And Early Warning Based On Data Mining Technology

With the development of power enterprise,and power electronic devices and shock load widely used in electric power system, power quality is becoming more and more serious, and the predicting of the power quality is also becoming more and more important.In response to that demand, this paper presents a mathod that using data mining technology to predict power quality and to produce early warning for power quality.The main work ot this program is to forecast the five power quality steady state indicators in the next period by using the historical data of active power and the five conventional indicators of power quality.Then the program will generate early warnings according to the comparison of the forecast data and the limit value.According to the early warning, people will more easily find the problems in the power system, and the aidministrator can put forward more practical measures to ensure security, stability, economic operation of the power system.In this paper, through analyzing the variation regularity about historical data of active power and the five conventional indicators, first carries on the forecast about the active power, and then using the correlation between the active power and the five indexes to predict the five conventional indicators, finally the forecasting data of five conventional indicators are compared with the limit data to judge whether the prediction data exceed the standard. If one indicator’s forecast data exceeds the standard,then the early warning will be produced. All these study contents of this dissertation is in my paper.During the program, The article uses ARIMA algorithm of time series to set up the active power predicting model to predict the active power. By using the correlation of the active power and the five indicators, The article respectively uses the ARIMA algorithm and neural network algorithm to predict the five conventional indicators. And the results were analyzed according to the prediction of the two algorithms by comparing the error of the two kinds of prediction algorithms.In the last this article chooses the appropriate algorithm to predict for each indicator, this paper stores the predictive results in tables, and the predictive results and warning information will be displayed on web page.Then users can view the predictive results and the warning information through the interface.