Optical manufacturing process sensing and status indication system

Abstract

An optical manufacturing process sensing and status indication system is taught that is able to utilize optical emissions from a manufacturing process to infer the state of the process. In one case, it is able to use these optical emissions to distinguish thermal phenomena on two timescales and to perform feature extraction and classification so that nominal process conditions may be uniquely distinguished from off-nominal process conditions at a given instant in time or over a sequential series of instants in time occurring over the duration of the manufacturing process. In other case, it is able to utilize these optical emissions to derive corresponding spectra and identify features within those spectra so that nominal process conditions may be uniquely distinguished from off-nominal process conditions at a given instant in time or over a sequential series of instants in time occurring over the duration of the manufacturing process.

Claims

1. An additive manufacturing system comprising: a powder bed arranged to hold a workpiece; an energy beam arranged to generate a molten region at the workpiece; a sensor arranged to collect data related to the molten region; and a processor adapted to determine a coupling efficiency of the energy beam to the workpiece based on the data.

2. The additive manufacturing system of claim 1 wherein the energy beam is a laser beam.

3. The additive manufacturing system of claim 1 wherein the powder bed comprises a layer of powder that is selectively fused to the workpiece.

4. The additive manufacturing system of claim 1 wherein the sensor is an optical pyrometer.

5. The additive manufacturing system of claim 1 wherein the energy beam moves relative to the workpiece and the sensor has a field of view that moves with the energy beam.

6. The additive manufacturing system of claim 1 wherein the energy beam moves relative to the workpiece and the sensor has a field of view that remains stationary.

7. The additive manufacturing system of claim 1 wherein the sensor has a field of view at the powder bed that is larger than a size of the molten region.

8. The additive manufacturing system of claim 1 wherein the sensor is arranged to detect optical radiation emitted from the molten region.

9. The additive manufacturing system of claim 8 wherein the processor analyzes the data using a fast fourier transfer (FFT) function to generate transformed data.

10. The additive manufacturing system of claim 9 wherein the processor determines the coupling efficiency from the transformed data.

11. A method comprising: generating an energy beam; directing the energy beam at a workpiece to create a molten region at the workpiece; acquiring data from a sensor arranged to collect input related to the molten region; and calculating a coupling efficiency of the energy beam to the workpiece based on the data.

12. The method of claim 11 wherein the energy beam is a laser beam.

13. The method of claim 11 further comprising a powder bed that includes a layer of powder that is selectively fused to the workpiece.

14. The method of claim 11 wherein the sensor is an optical pyrometer.

15. The method of claim 11 wherein the energy beam moves relative to the workpiece and the sensor has a field of view that moves with the energy beam.

16. The method of claim 11 wherein the energy beam moves relative to the workpiece and the sensor has a field of view that remains stationary.

17. The method of claim 11 wherein the sensor has a field of view at the workpiece that is larger than a size of the molten region.

18. The method of claim 11 wherein the sensor is arranged to detect optical radiation emitted from the molten region.

19. The method of claim 18 wherein the calculating comprises analyzing the data using a fast fourier transfer (FFT) function to generate transformed data.

20. The method of claim 19 wherein the coupling efficiency is determined from the transformed data.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a schematic illustration of a heat source impinging upon a workpiece.

(2) FIG. 2 is a schematic illustration of a workpiece giving off optical radiation due to heating by a heat source and the optical sensor is a non-contact pyrometer.

(3) FIG. 3 is a schematic illustration of raw pyrometer signals, which are then transformed by extracting features therefrom.

(4) FIG. 4 is a schematic illustration of an optical sensor that is capturing the optical emissions from the radiation process, comprising a spectrometer where the raw spectrum shows integrated spectral intensity as a function of wavelength.

MODES OF CARRYING OUT THE INVENTION, AND INDUSTRIAL APPLICABILITY

(5) In FIG. 1, a heat source 100 is shown impinging upon a workpiece 101. There is a molten region 102 on the workpiece directly below the region of the workpiece 101 affected by the energy source 100. There also could be a region of ionized or vaporized material 103 which is above the molten region 102. Both the molten region 102 and the vaporized or ionized region 103 will emit optical radiation 104. This radiation is detected by an optical sensor 105. The sensor can be stationary with respect to the energy source, also known as an Eulerian reference frame, or it can be in the same reference frame as the moving energy source, also known as a Lagrangian reference frame.

(6) In FIG. 2, the workpiece 200 gives off optical radiation 201 due to heating by the heat source 202 and the optical sensor is a non-contact pyrometer 203. This pyrometer collects a thermal signal 204. It is assume that the size of the heat source 202 is smaller than the size of the field of view of the pyrometer. The thermal signal comprises two type of features: a slower moving signal 205 that is associated with the heat source 202 gradually coming into the field of the view of the pyrometer as well as representing the background temperature of the workpiece 200; and a faster moving signal 206 that represents individual high temperature excursions caused by the heat source 202 as it moves into and out of the field of view of the pyrometer.

(7) In FIG. 3, the raw pyrometer signals 300 are then transformed by extracting features from this data. There are several types features that can be extracted. First, from the slower varying data 301 the heating rate 302 and cooling rate 303 as well as the peak background temperature 304 can be extracted. Secondly, from the faster varying data 305, the heating rate 306, the cooling rate 307 and the peak temperature 308 can be derived. The slower varying data features indicated by 302, 303, and 304 correspond to the material response, as it is largely dictated by the local thermal boundary conditions such as the thermal conductivity, heat sinking properties, etc. Additionally, the slower varying thermal data indicates the background temperature of the workpiece in between scan events from the moving heat source. This again is related to the material response as opposed to the thermal input from the process. The more rapidly varying data features indicated by 306, 307, and 307 are more representative of the process inputs and the energy inputs to the manufacturing process on account of the scanning energy source.

(8) In FIG. 4, the optical sensor that is capturing the optical emissions from the radiation process is a spectrometer 400 which results in the raw spectrum 401 showing integrated spectral intensity as a function of wavelength. In general any spectrum will have both background blackbody radiation features 402 as well as characteristic radiation features 403. It is possible to extract features from raw spectrum by taking the FFT and examining the peak heights in the FFT at some intermediate values of the inverse wavelength. For example the region of more rapid change in the wavelength domain peak 404 can correspond to a local maximum in the FFT 405 in the inverse wavelength domain. The slower the rise of the wavelength domain peak 404 would then correspond to a lower value of the corresponding FFT peak 405. This particular example could happen for example when the wavelength domain peak is caused by evaporation of given chemical species. The lower this evaporation, the lower the spectral intensity and the slower the rise towards the spectral peak because of the nature of the Gaussian fit of the spectral peak. This for example could happen when the energy coupling to the workpiece is optimal in the sense that energy is being absorbed, there is a stable liquid melt pool, and there is sufficient superheat to overcome the latent heat of melting for new powders being sintered but not so much superheat so as to cause excessive vaporization.

(9) Irrespective of how the features are derived, whether they are from the thermal sensor or the spectrometer, the classification scheme can be the same. First, the features associated with a baseline condition are identified as one set of data. Then the features from any given test case can be compared to the baseline condition as follows. First the features from the baseline case are averaged and a vector of the mean of these features M is created. he test vector X has the same dimensionality as the vector of feature means because it has the same number of features, which will be also called the degrees of freedom. Then a classification scheme as taught in this present invention involves the use of the Mahalanobis distance, which is simply given by:
MD.sup.2=[X−M].sup.T.Math.COV.sub.X.Math.[X−M]  (5)

(10) Where COV.sub.X is the covariance matrix of X. It can be shown that when the features are normally distributed, then the square of the MD distance will be Chi-Square distributed. The Chi Squared probability density distribution is given by:

(11) $\begin{array}{cc}f\left(x;k\right)=\{\begin{array}{cc}\frac{x^{\left(k\text{/}2\right)-1}{e}^{-x\text{/}2}}{2^{k\text{/}2}\Gamma \left(\frac{k}{2}\right)},& x\ge 0;\\ 0,& \mathrm{otherwise}.\end{array}& \left(6\right)\end{array}$

(12) Where is the Gamma Function and k is the number of degrees of freedom, which in this case is identical to the number of features. The critical value of the Chi-Squared distribution at a given confidence level and a given number of degrees of freedom can be calculated. This is a threshold value of the distribution above which a point could be considered as an outlier within the context of fitting the MD Distance t a Chi-Squared distribution. For example, at a 95% confidence level, or a critical p-value of 0.05, the corresponding table of critical values of the Chi-Squared distribution and therefore the MD distance squared as well are given by the following table:

(13) TABLE-US-00001 Degrees of Freedom Critical Value of the (also the number Chi-Squared Distribution - also of Features in critical value of the square of the the Feature Vector) MD distance 1 3.84 2 5.99 3 7.82 4 9.49 5 11.07 6 12.59 7 14.07 8 15.51 9 16.92 10 18.31

(14) The present invention provides a method of utilizing optical data through a variety of sensors as well as a variety of feature extraction techniques to enable the classification of nominal vs. off-nominal conditions found in a variety of manufacturing processes that involve the application of heat by a high energy or high temperature transient heat source.

(15) The present invention has been described in the context of various example embodiments. It will be understood that the above description is merely illustrative of the applications of the principles of the present invention, the scope of which is to be determined by the claims viewed in light of the specification. Other variants and modifications of the invention will be apparent to those of skill in the art.