The Application Of A Functional Partial Linear Regression Model In The Coal Spectral Data

Currently coal is used in industrial production more widely, about seventy-five percent or more of electricity is generated by the thermal power, the fuel of thermal power plant is mainly coal. But in the coal combustion process, if the quality of coal is not good, it will cause direct damage to the equipment we use, affect the interests of the plant, but it also will lead to air pollution, and affect our lives, so determining the merits of coal quality is particularly important.Whenever some coal is shipped into the plant, if by random sampling, then physical methods we measure the content of the various elements in coal, it losts timeconsuming and laborious, and the accuracy is very low, so the plant wants to develop a scanning instrument who can scan the coal spectrum, then we can predict the precise content of the element using the spectrum and content of the coal by roughly measuring main elements in coal. Lastly we compare the precise content with the element content standards given in the plant again, and ultimately determine the quality of the coal quality.Traditional coal spectrum data analysis treats coal spectrum and the main elements in coal as a random vector, and analyse them by using the methods of time series and artificial neural network, but it will lose some important information, and can not meet the expectations of more accurate coal spectrum analysis.In out paper,spectrum is treated as a function curve,a functional partial linear model is setted up in coal spectrum and elements such as carbon, hydrogen, oxygen, nitrogen,sulphur and so on,the model assumes that the main variable is a curve. Through simulation and analysis of coal spectrum data, we found that the accuracy of using the functional partial linear model is the highest.