IE64270B1

IE64270B1 - A multiwavelengths pyrometer

Info

Publication number: IE64270B1
Application number: IE343690A
Authority: IE
Inventors: Claudio Ronchi; Jean-Pol Hiernaur; Rutger Beukers; Wilheim Heinz; Raoul Francois Selfslag
Original assignee: Euratom
Priority date: 1989-09-25
Filing date: 1990-09-24
Publication date: 1995-07-26
Also published as: CA2067248A1; IE903436A1; PT95405A; ES2056328T3; ATE107021T1; CA2067248C; EP0420108A1; LU87595A1; AU6418590A; DE59006014D1; DK0420108T3; RU2083961C1; JPH06500387A; EP0420108B1; AU639029B2; WO1991004472A1; PT95405B

Abstract

The invention concerns a multi-wavelength pyrometer for the measurement of the temperature and emissivity of a surface above 900 K. The pyrometer has several radiation detectors sensitive in different wavebands lambda 1... lambda i... lambda n as well as a data processor to which the detector outputs are fed after digitalization and which calculates the temperature of the surface from these detector outputs using Wien-Planck's law, assuming the surface is a perfect black body. The emissivity is then calculated, as a function of the temperature and the wavelength, from these calculated temperature values using an approximation method, and from this, the required temperature. The invention calls for the differences between the actual pyrometer signals and the signals which would be expected from the assumed emissivity and the temperature calculated from this emissivity to be determined for various approximations and various wavelengths. The approximation which gives, for all the wavelengths, the least sum of the squares of these differences is then selected.

Description

The invention relates to a method for measuring the ¢temperature and emission rate of a surface above 900 K, comprising a multiwavelengths pyrometer which is sensitive to ‘ different wavelengths λ1...λΐ...λη and a data processor which receives the output signals after A-D conversion and deduces therefrom, by means of the Wien-Planck law, the temperature, assuming that the surface is an ideal black body, the emission rate being then computed from these temperature values according to an approximation law as a function of temperature and the wavelength, the desired temperature being deduced therefrom.

From the journal Temperature, vol. 5, 1982, pages 439 to 446, a rapid pyrometer of the type mentioned above is known. An optical system is directed onto the surface to be measured, which system splits up into six channels by means of a glass fiber bundle, and is led to the photodiodes via narrow band filters. The detector signals are then converted into digital signals and evaluated in a processor.

The evaluation is based on the Wien-Planck equation for black bodies L = Cl. λ-5[exp(C2/XT)-1)]_1 (1) where L is the beam density at the wavelength λ, Cl and C2 are constant terms and T is the temperature of the black body.

Since the surface to be examined is normally no ideal black body, the emission rate E must be taken into account, which represents the ratio between the beam density of the black body and the real body.

This emission rate is a function of temperature and wavelength and can be expressed by a Taylor series of the following kind: In E = βθ + a^X + a2^·2 + ··· (2) According to experience, the dependence on wavelength in limited wavelength ranges is a steady function, so that the series (2) can be cut off after a few terms.

In the cited article, it is thus proposed to choose a linear approximation of the function (2) and to evaluate respectively pairs of wavelengths out of the six measured values of the beam density according to the six wavelengths of the pyrometer and then to find out the temperature by the analysis of the squares of the deviations of the different results.

It has been found out that in difficult cases this method leads to results which do not permit a reliable state10 ment as to their precision.

Thus, pyrometrical measurements of highly reflective surfaces, where the emission rate is very low and very unstable due to possible surface reactions (for example aluminum during metallurgical treatments), are reputed to be difficult.

Document DE-A-3 611 634 refers to a pyrometer measuring method in which the differences between the measured spectral signal voltages and the assumed voltages are calculated. The calculation of the assumed voltages makes use of stored data for different materials.

Another method for exactly calculating the temperature establishes differences between measured and assumed signals and is described in document JP-A-60 152 924 (see English abstract).

It is thus the aim of the invention to improve a method for a multiwavelengths pyrometer of the kind cited above in such a way that the computation complexity and the residual error are diminished and that usable results can be obtained even under very unfavourable measurement conditions.

According to the invention, this aim is achieved by the method as defined in claim 1. Regarding a feature of a preferred implementation of this method, reference is made to the dependent claim.

The invention will now be described more in detail with reference to two figures.

Figure 1 shows a flow diagram for the operations to be carried out by the processor.

Figure 2 shows schematically a pyrometer according to v the invention.

A six wavelengths pyrometer 1, as it is described in 4 the above-mentioned essay in Temperature, delivers simultaneously six radiation intensity values of a body, or its surface respectively, observed by the pyrometer, the wavelengths used in practice lying between 400 and 2000 nm. The bandwidth of a measurement channel lies under 100 nm.

The measurement values which are proportional to intensity are obtained in a known way in photodiodes and then applied to a processor 2 in digitalized form. The latter firstly finds out whether the signals are sufficiently stable, i.e. whether the noise level is sufficiently low. Only if this is the case, the temperature can be computed with a sufficiently small error (signal standard deviation: SQ). Then the law for the determination of the emission rate E is chosen according to equation 2. It must be differentiated between a model of zero order, in which ln E is a constant a, independent of the wavelength, a model of first order, in which ln E linearly depends on the wavelength (the law is defined by the determination of βθ and a^) and models of higher order, in which further members of the Taylor series must be evaluated.

First, a model of first order is taken as basis and βθ and a^ and thus the emission rate for the six wavelengths are determined, βθ and a^ having the same value in all six determination equations. Basically, the evaluation consists in a sub-routine which minimizes the sum of the squares of the deviations between the measured signals and the beam density value computed by means of the emission value defined by βθ and alz and which evaluates the resulting standard deviation SK of the fitting procedure.

The expected temperature and emission rate errors are computed as the differentials which are obtained by successive incrementation of the signals at the respective error and by repeated computation of temperature and emission rate. It is recommended to control whether a model of zero order would also be applicable, since it offers a smaller absolute error in the temperature detection. This is the case when the constant from equation 2 lies below a given value, i.e. when the emission rate practically does not depend on the wavelength. In this case, six independent temperature measurements are obtained for the different wavelengths.

The choice of models of higher order leads to a reduction of the standard deviation SK, but not forcibly of the temperature error. On the contrary, when S„ reaches the K value of Sq, any following increase of the model order (overfitting) results mostly not in a lesser, but in a higher imprecision of the temperature. When during error evaluation it is found out that the error increases, the optimal model has been found and the constants a., a-,...a. have been deter1 2 j mined.

If the error analysis has shown that the error is especially small, then it is recommended to memorize for later utilization the group of curves connecting the temperature, the wavelength and the emission rate. Thus, a data bank organized according to the kind of materials of the surface to be examined is established which can be made use of lateron. This is especially valuable when during a later measurement there are very unfavourable measuring conditions, for example colour differentiated vapour development in the optical path of the pyrometer or instabilities in the electronics due to high environmental temperature. In this case, the pyrometer measurement values are simply compared to groups of curves evaluated at earlier times and the temperature can be directly computed therefrom. Such a data bank fed by the least disturbed signals is shown in Figure 2 with the reference 3. Also other unicolored pyrometers can be operated with the data of this data bank in parallel.

With the pyrometer according to the invention, the desired evaluations can be carried out during a millisecond even under unfavourable conditions, so that on a screen 4 tem porary evolution of the temperature or the emission rate can be shown practically in real time also for rapidly developing processes, such as for example the pulse heating by means of laser. This opens new possibilities for the analysis of rapid ly developing processes in the temperature range above 700 K and up to 10.000 K.

Claims

1. A method for measuring the temperature and emission rate of a surface above 900 K comprising a multiwavelengths pyrometer which is sensitive to different wavelengths λϊ... λϊ... λη, and a data processor which receives the output signals of the pyrometer after A-D conversion, and deduces therefrom, by means of the Wien-Planck law, the temperature, assuming that the surface is an ideal black body, the emission rate being then computed from these temperature values according to an approximation law as a function of temperature and the wavelength and from this the wanted temperature is computed, characterized in that the differences between the pyrometer signals and the pyrometer signals to be expected due to the assumed emission rate and the desired temperature deduced therefrom are computed for several of the approximation laws and the different wavelengths, and that then that approximation law is selected which represents for all the wavelengths the lowest sum of the squares of these differences.

2. A method according to claim 1, characterized in that a data bank is established for the emission rate of certain materials as a function of temperature and wavelength, and that the processor also uses this data bank for computing the temperature when the same materials are subjected to a pyrometer measurement.

3. A method according to claim 1 substantially as hereinbefore described with reference to and as illustrated in the accompanying drawing.