Application Study In The Thermal Electric Power Plant Equipment Fault Detection Of Changeable Capacity Statistical Process Control

It is more and more higher as the power production enterprise to the requirement of production management efficiency to the constantly increase of power sources of energy requirements to produce the life along with modern times, and generating electricity, the enterprise put forward the higher requirement to the power production brain. To the power production course is in progress online monitoring the fault diagnosis oneself through a very important research direction that becomes at present the power trade process control domain, makes that the research of this one side possesses the very important real meaning.This text sets out from the descriptive statistical interpretation of data, the systematic introduction the application in the Thermal electric power plant equipment fault detection course of changeable capacity statistical process control engineering. One kind of active fault detection method - linear model law of the smallest error rate was put forward, checkout sensitivity and the checkout precision that many grades of checkout tactics pattern and trends come to achieve better to calculate the speed was built that to the static state after by way of adopts linear dividing into groups and very good, and passes through that the simulation experiment research verified this kind of effectiveness and feasibility of failure detection method.Thesis is chiefly innovate a little as follows:1 Should principal component analysis falls that the dimension thought arouses to put forward linear to divide into groups that the method tieed up is fallen in the projection, had already reduced to build the dimension according to the space of modulus greatly, and makes to find the solution that the most excellent linear recurrence equation poor lifts searching for becoming possibility, and like PCA's method not to fall the loss that the dimension can be known according to information.2 The being mutually related knowledge to use the number reason to add up to poor lifts searching for the most excellent linear computer that builds the pattern course algorithm separated to be in progress at 4 optimizing, and to raise the speed of procedure operation greatly, and makes this method apply in the large-scale data to calculate into for probably, and active solves the data analysis theory and the contradiction of practice application.3 Uses many grades of dynamic checkout tactics to raise sensitivity and the precision of model checkout, and passes through the corresponding checkout tactics of formulation to every model making up the shortcoming of model checkout information.4 Uses many models realizes monitoring to a checkout variable at the same time, and rather than goes to monitor many checkout variables with a model, and active solves the mistake newspaper in the failure detection course and fails to report the appearance.5 Compile the complete computer processor, from data analysis, data pretreatments and the complete computer processor was all compiled to linear a series of courses laying down to the trouble simulation checkout building pattern and checkout tactics, and makes the application of this method become simply workably; At the same time, the procedure possesses stronger transplanting, and the suitable adjustment is tested after data and the parameter, and available analyses dealing with in the automation of this kind of question.