Electrical power quality has become an important issue in power systems in the last two decades. Some of the reasons are increasing use of non-linear loads like arc and power electronic loads, fast increase of critical loads such as computers and microprocessors, development of inter connected electrical networks, restructuring in electricity industry and creation of competitive electricity market. In order to understand the electrical power quality, having adequate information about its phenomena and their standards is essential. Due to stochastic nature of power quality phenomena, it is necessary to determine standard levels that is usually expressed in the statistical form. Considering technological progresses in the field of measurement equipments, standards for the power quality are ever changing ; So research and revision of the existing standards as well as new standards is required in a permanent form. Therefore, the nead for a fairly accurate method for evaluating an estimate of power quality level for the network and also obtaining a unique index to include power quality in cost the electricity tariff of the customers is required. Some standard indices have been defined for the power quality ( for example power factor for the reactive power, flicker for estimating low frequency changes of voltage)however, there is no a unique index that can assess the level of system power quality with all above phenomena. The purpose of this research is to obtain a quantitative index, in a manner that this index covers all power quality phenomena. Since the problem of power quality is a function of time varying stochastic phenomena, and in general it is a stochastic problem which causes make difficult solving this problem by analytical methods and a combination of statistical, mathematical, and intelligent methods are used here to overcome the problem. Data mining methods present proper algorithms for solution of this problem and Fast _ICA algorithm will be introduced and used for our problem. In this thesis, at first we use the collected data from various points in country and divide them on the basis of types of loads. Some levels are defined for each phenomenon of power quality that according to it each phenomenon is divided into 7 justify; LINE-HEIGHT: 90%; MARGIN: 0cm 0cm 0pt; unicode-bidi: embed; DIRECTION: ltr" key words: power quality, general and unique index, data mining , Fast-ICA
