Analysis on Automatic Discrimination of High Temperature Based on Fowler-Milne Method
Jiang Fan,Li Yuan-feng,Chen Shu-jun,Li Cheng
DOI: https://doi.org/10.3964/j.issn.1000-0593(2019)02-0370-07
2019-01-01
Spectroscopy and spectral analysis
Abstract:In the research field of arc plasma spectrum diagnosis, combined with the advanced image sensing technology, the Fowler-Milne method uses spectral image to obtain temperature information of arc plasma. Because of its high time and space resolution, the Fowler-Milne method is widely used in arc plasma temperature measurement. However, the relationship between line emission coefficient and temperature is not monotonous, and traditional Fowler-Milne method selects one Ar I spectral line to complete the measurement, which leads to the decrease of line intensity, and the process of measurement needs researchers to determine the temperature range of different locations to complete the calculation of temperature. The whole process can' t be automatically completed by software. In view of this problem, based on PLTE conditions of arc plasma, applying the partial LTE, model of arc plasma, modified Fowler-Milne method based on two spectrum line, which combines the Ar I spectral line emitted by Ar atoms in an outer low temperature region of arc, and the Ar II spectral line emitted by Ar ionization ion in hightemperature area of arc, to determine the different temperature range of the arc plasma, and then the whole temperature is calculated by the temperature corresponding to the intensity of the Ar I spectral line in low-temperature area and high-temperature area, eliminating the adverse effects of single Ar I spectral line emission coefficient field. A light splitting system was also designed and built, dividing the light of arc into two beams by a spectroscope. Two sets of reflectors and narrow-band filters were used to collect the image of two sets of arc spectral images though one exposure, of which parameters such as the focal length aperture are exactly the same, achieving good time and spatial consistency and reducing the error of emission coefficient fusion. In order to verify the feasibility of measurement system and arc image extraction, black and white chessboard was used as a target, and the extraction of corners extracting of two image proved the system satisfies the demand of collecting two groups of arc spectral images, and was also used to normalize two images for the extraction of arc image in the later stage. Based on the assumption that the plasma arc has axisymmetric properties, with the brightness information of spectral image CMOS collected as the integration of emission coefficient under different angle projection, after the median filter noise reduction processing, ML-EM method was used to reconstruct the 3D emission coefficient distribution from the 2D luminance distribution. In the experiment, Ar 696, 5 nm spectral line and Ar II 480. 6 nm spectral line with little self-absorption effect were selected. The ODO. 4 ND filter was added in the pathway of 696. 5 nm spectral line, to make the maximum brightness value of the two spectral images consistent. 150 A welding plasma arc was measured in the experiment. After three - dimensional reduction of ML-EM method, the two spectral line emission coefficient fields were fused. At the pixel point location where the Ar I spectral line reaches the maximum value, Ar II spectral line reaches epsilon(rp), which were used to determine the high-temperature zone or low-temperature zone. The measurement of the plasma arc of 150 A showed that 696. 5 and 480. 6 nm spectral line can automatically identify the high temperature zone in welding arc plasma, making it more possible for the arc temperature real-time monitoring to be realized.