Power grid basic construction project cost prediction method based on multi-core support vector regression

Abstract

The invention discloses a power grid basic construction project cost prediction method based on a multi-core support vector regression. The method comprises the following steps: carrying out attribute pretreatment on descriptive data of a transmission and distribution project, and using a basic kernel function to calculate a multi-core matrix, wherein the attribute pretreatment comprises the steps of standardization of continuous variables and type conversion of enumerative variables; constructing a prediction model of the multi-core support vector regression, and optimizing a regression parameter of the prediction model and the weighing of the multi-core matrix; and utilizing the prediction model to carry out cost prediction by taking data obtained from the attribute pretreatment as a sample to be tested. According to the application of the method, the power grid basic construction project cost can be reasonably predicted so as to facilitate construction management personnel to improve the budget estimate making precision; and therefore, the basic construction investment is efficiently controlled.

Description

technical field [0001] The invention relates to data mining technology, in particular to a method for predicting the cost of power grid infrastructure projects based on a multi-core support vector regression machine. Background technique [0002] In recent years, the construction pace of the power grid industry has been rapid. In 2010, the investment in power grid construction was 45.4 billion yuan, and in 2011, the investment in power grid construction was 29.7 billion yuan. Affected by the internal and external environment, there is an urgent need to reasonably control the project cost of the power grid and improve the lean level of power grid infrastructure project cost management. [0003] External environment, in recent years, the CPI (Consumer Price Index, consumer price index) index has continued to rise, and the prices of major equipment and materials have risen rapidly. Submit the relevant cost analysis materials of typical projects; internal environment, the balan...

AI Technical Summary

Problems solved by technology

[0004] The traditional support vector regression machine can effectively predict the data on the data set of a single stable data source with given parameters, but in the face of more complex heterogeneous data, the traditional support vector regression machine is difficult because it only uses a single kernel function. In this kernel function space, a reasonable mapping relationship between input and output is established, so a more accurate regression function cannot be effectively obtained, or an over-generalized regression function will be obtained, resulting in inaccurate predictions

Images