【正文】 of the object to be detected can be targeted to work. First to conduct a detailed analysis of the physical characteristics of the furnace pulverized coal flame, and in particular relates to the analysis of the infrared radiation characteristics of the pulverized coal flame, and through analysis that the pulverized coal flame signal is a quasistationary signal. Determine the infrared flame detection device detection field of view of the principle of selecting. Flame detection device to be used by the system sampling frequency is determined according to the characteristics of the pulverized coal flame, and to determine the range of the antialiasing filter parameters. 成都理工大學碩士學位論文 The Fourier transform is one of the most monly used approach in modern signal processing, a brief introduction of the theoretical basis of the Fourier transform, fast Fourier transform algorithm mathematical process is described in more detail. Pulverized coal flame using the fast Fourier transform method of information extraction, and extract the information. Since the standard Fourier transform in the frequency domain with local analytical capacity, and do not have the local analysis capabilities in the time domain, the Dennis Gabor in 1946 introduced the concept of shorttime Fourier transform. This paper introduces the theoretical basis of the short time Fourier transform, a more detailed description of the short time Fourier transform algorithm mathematical process, and the pulverized coal flame data using the short time Fourier transform method of information extraction. After the analysis that this method can accurately identify the existence of a pulverized coal flame at the burner signal, but more accurately determine the bustion stability information. The wavelet transform is a new transformation method can not meet the requirements of the analysis of nonstationary signals generated in the traditional Fourier transform. Essentially, it is a timefrequency analysis method, with a multiresolution analysis of excellent quality, successfully overe the defects of the shorttime Fourier transform of a single resolution. The more important is the wavelet transform with the characterization of signal characteristics in the time domain. This paper introduces the theoretical basis of wavelet transform, the mother wavelet analysis on how to choose a more detailed description of the Mallat algorithm specific mathematical process, pulverized coal flame data and information extraction using wavelet transform method. After analysis, this method can accurately identify the existence of a pulverized coal flame at the burner signal, and can accurately judge bustion stability information, and the ability to accurately provide the critical timefrequency domain information. Flame detection device hardware system taking into account the current development and use of the short time Fourier transform is the better overall performance of a pulverized coal flame signal information extraction method. The wavelet analysis method has very significant information extraction, in the case of system hardware allows, should focus on information extraction using wavelet analysis of pulverized coal flame signal, wavelet analysis method in the future will increasingly Abstract V wide range of specific applied to the analysis of flame engineering practice. Prone to saturation distortion introduced in the probe circuit of the logarithmic amplification circuit, the dynamic range is pressed to the degree of ease of handling within the probe output signal for conventional flame detecting apparatus. The traditional flame detector device signal filtering methods improved by the passive filter active filter more efficiently filter out the useful band interference signals, in favor of subsequent digital signal processing. The flame detection device core processor is AT91SAM9G45 embedded processor with powerful data processing capability and business management capabilities. Successfully in the flame detection device to achieve shorttime Fourier transformbased information retrieval and information extraction based on wavelet transform. The Model flame detection device is used in a number of different furnace large boilers. They are able to correctly identify the presence of a pulverized coal flame and give more accurate information of the bustion stability, and achieved good social and economic benefits. Keywords: Flame detection device Infrared Fourier transform Short time Fourier transform Wavelet transform Logarithmic amplification Embedded processor 成都理工大學碩士學位論文 VI 目 錄 第 1 章 引言 ....................................................................................................... 1 課題研究的背景及意義 ................................................................................ 1 光能式火焰檢測裝置的發(fā)展簡史 ................................................................ 2 紅外型火焰檢測裝置的檢測方法 ................................................................ 2 紅外型火焰檢測裝置的常規(guī)檢測方法 ............................................ 2 對紅外型火焰檢測裝置的改進 ........................................................ 3 本文主要內容及成果 .................................................................................... 3 第 2 章 爐膛煤粉火焰的物理特征 ................................................................... 5 爐膛煤粉火焰物理特征 ................................................................................ 5 煤粉火焰的物理特征 ........................................................................ 5 鍋爐燃燒器處主火焰特征和背景特性 ............................................ 5 第 3 章 煤粉火焰的信號處理與分析 ............................................................... 8 煤粉火焰燃燒信號的特性 ............................................................................ 8 火焰信號的采樣率和抗混疊濾波 ...................................................