Method for detecting defects in addition-processed powder layers deposited on a processing area.

Abstract

The present invention relates to a method for detecting defects in an additively manufactured powder bed deposited on a processing area, comprising the steps of: i. acquiring an image of the additively manufactured powder bed; ii. determining a discrete spectral representation of the acquired image; iii. frequency filtering the discrete spectral representation of the acquired image; iv. determining a filtered image from the filtered discrete spectral representation of the acquired image; and v. analysing the filtered image to detect defects.

Description

Technical Field

[0001] The present invention relates to the field of additive manufacturing, more specifically, to the field of selectively additive manufacturing three-dimensional objects from layers of powder. In particular, the present invention proposes a method that enables the detection of defects in a layer of powder before the layer of powder is selectively solidified.

Background Art

[0002] Selective additive manufacturing consists of creating a three-dimensional object by solidifying selected zones within successive layers of a powder material (metal powder, ceramic powder). The solidified zones correspond to successive cross-sections of the three-dimensional object. The solidification is performed layer by layer by overall or partial selective melting using a focused radiation source such as a light source (e.g., a high-power laser) or alternatively a particle beam source (e.g., an electron beam, i.e., a technique known as EBM or "electron beam melting" according to the terminology commonly used in this field).

[0003] The method of selectively additive manufacturing a three-dimensional object from a layer of powder makes it possible to manufacture parts that combine accuracy and surface quality with a level of detail of about 10 micrometers. However, this accuracy and this surface quality can be significantly deteriorated if the layer of powder has deposition defects (premature deposition, lack of powder (e.g., "cat's tongue", or ripples resulting from insufficient spreading of the powder by a roller or a scraper responsible for spreading the powder)).

[0004] In order to avoid these problems that affect the quality of the parts produced, it is necessary to be able to inspect the quality of the deposited layer of powder so as to identify the presence of these defects before they are selectively solidified.

[0005] Many methods have been developed for this purpose. For example, reference can be made to European Patent Application Publication No. 3378039 and German Patent Application Publication No. 102018207405.

[0006] These methods typically involve the use of specialized equipment such as industrial cameras, infrared imaging devices, and specific lighting.

[0007] They are therefore expensive.

[0008] These methods are further generally based on complex algorithms, such as 3D image reconstruction methods for comparing different heights of powder within the same layer, for example, for post-hoc analysis and statistical modeling of various height and illumination defects.

[0009] These methods are essentially related to laboratory methodologies and are difficult to integrate into manufacturing processes and selective additive manufacturing machinery intended for industrial applications. [Prior art documents] [Patent Documents]

[0010] [Patent Document 1] European Patent Application Publication No. 3378039 [Patent Document 2] German Patent Application Publication No. 102018207405 Specification [Overview of the project] [Problems that the invention aims to solve]

[0011] One objective of the present invention is to propose a method for detecting defects within a layer of additively manufactured powder that can be carried out using simple equipment and without major modifications to a powder bed fusion printing machine.

[0012] Another objective of the present invention is to enable the use of such a method by making it economical in terms of computer resources (execution time, required memory, etc.) during the printing process between each coating and melting step. [Means for solving the problem]

[0013] In a first aspect, the present invention is i. Step of obtaining an image of the additive manufacturing powder layer, ii. A step to determine the discrete spectral representation of the acquired image. iii. A filter having at least one cutoff frequency for filtering the discrete spectral representation of the acquired image with respect to frequency, iv. A step of determining a filtered image from the filtered discrete spectral representation of the acquired image; v. A step of analyzing the filtered image to detect defects in the layer of deposited powder. The present invention relates to a method for detecting defects in a layer of additive manufacturing powder deposited on a processing zone equipped with a processing zone.

[0014] Step v, which analyzes the filtered image, Step va) to enhance the contrast of the filtered image, Step vb) Detecting edges and / or shapes in a contrast-enhanced filtered image Steps include: processing the detected edges and / or shapes to classify the type of defect associated with the detected edges and / or shapes (vc), and Step vd) calculate a score associated with a layer of deposited additive manufacturing powder, which characterizes the number and / or size of defects present in the layer. It can be further equipped with...

[0015] The frequency filter applied in step iii. may be a Gaussian bandpass filter.

[0016] In a second aspect, the present invention relates to a method for detecting defects in a layer of additively manufactured powder, wherein training the model comprises the step of performing the method according to the first aspect.

[0017] According to a third aspect, the present invention relates to a method for selectively additive manufacturing a three-dimensional object from a layer of powder, the method comprising a device for selectively additive manufacturing a three-dimensional object from a layer of powder that performs the following steps repeated over a determined number of iterations: A. adding a layer of additive manufacturing powder on a base or on a previously solidified layer; B. detecting defects in the layer of additive manufacturing powder deposited according to the first aspect before it is solidified; C. emitting a laser beam over a first point of the layer of additive manufacturing powder to solidify a first zone of the layer of powder comprising the first point; The step of detecting defects in the layer of additive manufacturing powder deposited over the processing zone further comprises processing means (21, 31) that perform the following sub-steps: B.i. acquiring an image of the layer of additive manufacturing powder; B.ii. determining a discrete spectral representation of the acquired image; B.iii. filtering the discrete spectral representation of the acquired image with respect to frequency; B.iv. determining a filtered image from the filtered discrete spectral representation of the acquired image; B.v. analyzing the filtered image to detect defects in the deposited layer of powder.

[0018] Other features and advantages of the present invention will become apparent upon reading the following description of the preferred embodiments. This description is given below with reference to the accompanying drawings.

Brief Description of the Drawings

[0019] [Figure 1] It is a diagram of an architecture for implementing the method according to the present invention. [Figure 2] It is a diagram of an alternative architecture for implementing the method according to the present invention. [Figure 3] It is a diagram showing the steps of a method for selectively additive manufacturing a three-dimensional object from a layer of powder according to the present invention. [Figure 4] This figure shows the steps of a method for detecting defects within a layer of additively manufactured powder according to the present invention. [Figure 5] This figure shows the steps of the filtering image analysis method according to the present invention. [Modes for carrying out the invention]

[0020] architecture The selective additive manufacturing device 1 in Figure 1 comprises the following: -Additional manufacturing unit 3, - A processor-type data processing means 31 configured to implement the defect detection method according to the present invention, and - A data storage means 32 such as a hard disk computer memory in which code instructions for executing the defect detection method according to the present invention are stored.

[0021] The additive manufacturing unit 3 is a machine for selective additive manufacturing using layers of powder of the type currently sold by AddUp®. Such a machine conventionally has a manufacturing chamber 80 comprising an enclosure 30 and a work surface 40. The enclosure 30 has side walls and a top cover and covers the work surface 40. The enclosure may be made of, for example, metal or ceramic.

[0022] The manufacturing chamber 80 also has a manufacturing plate 50 intended to receive additive manufacturing powder 90 in the form of a continuous layer and to support the parts while they are being manufactured. For this purpose, the plate 50 preferably has mechanical strength properties that enable it to support parts weighing tens of kilograms or even hundreds of kilograms. That is, the platform 50 can be made of metal, for example.

[0023] In the embodiments presented herein, the plate 50 slides through an opening in the machining surface 40 into a sleeve 41 positioned below the machining surface 40. The machining surface 40 surrounds the upper edge of the manufacturing sleeve, allowing the manufacturing sleeve to keep the manufactured part and the surrounding non-solidified powder on the plate 50 within a substantially closed volume. However, in another embodiment, the plate 50 can be positioned alongside the machining surface 40, for example.

[0024] The plate 50 is positioned on the plane of the machining surface 40 at the start of the manufacturing cycle, and then on a plane that is substantially parallel to the plane of the machining surface 40 as the plate descends into the sleeve.

[0025] The manufacturing plate 50 may have a circular, rectangular, square, or triangular shape. Furthermore, the manufacturing chamber 80 includes a carriage 60 that slides above the processing surface 40 and the plate 50.

[0026] The carriage 60 allows for the distribution of powder onto the plate 50 or onto a preceding layer of powder from the standpoint of manufacturing parts, and the powder can be distributed to either side of the processing zone by a weighing device 62 that delivers the powder to a slide 66 that moves parallel below the weighing device to obtain a line of powder in front of the processing zone. The carriage 60 may be equipped with, for example, a scraper and / or rollers.

[0027] In addition, the manufacturing chamber 80 includes a power supply component 70 that enables the melting of metal powder. The power supply component 70 may be, for example, a laser beam source, an electron beam source, etc.

[0028] Furthermore, the manufacturing chamber 80 includes an illumination means 34 such as a lamp or photographic flash, and an imaging means 33 such as a photographic sensor. The illumination means 34 and the imaging means 33 can also be integrated outside the manufacturing chamber behind an opening closed by a windowpane.

[0029] As a variation, it should be noted that these processing and storage means may be transferred to a remote server 2. This is illustrated in Figure 2, where unit 3 is connected to server 2 by a data exchange network 10, and server 2 comprises: - A processor-type data processing means 21 configured to implement the defect detection method according to the present invention, and - A data storage means 22, such as a computer memory or hard disk, on which code instructions for executing the defect detection method according to the present invention are stored.

[0030] The data processing means 21 or 31 is configured to carry out the manufacturing method described below.

[0031] Additive manufacturing method The manufacturing method described in relation to Figure 3 comprises a first step A called "coating," during which a layer of additive manufacturing powder is deposited onto the plate 50 by a carriage 60.

[0032] Following this first step, the quality of the additive manufacturing powder layer is evaluated in step B, which involves detecting defects within the deposited additive manufacturing powder layer. It is during this step that defects such as mistimed powder deposition or absence of powder can be detected.

[0033] Depending on the quality of the additive manufacturing powder layer evaluated during the defect detection step (step B), it is possible to decide to repeat the coating step (step A) to eliminate any defects that appear. Furthermore, in cases where defects are repeatedly detected or when these defects are excessively numerous within the layer, it is also possible to warn the operator of the device for selectively additive manufacturing three-dimensional objects from the powder layer, enabling them to assess the severity of the defects, diagnose the cause of the defects, and decide whether to continue the manufacturing process (step F, described later).

[0034] Finally, with the quality of the additive manufacturing powder layer assessed as sufficient, the third melting step (step C) involves emitting a laser beam over the zone of the powder layer that will be solidified to form a part.

[0035] Defect detection (Step B) The coating defect detection step described in Figure 4 is performed between coating step A and melting step C, and ensures that melting step C is not performed on a layer of powder containing defects. The presence of defects in the powder layer can lead to the manufacture of defective parts. This defect detection is particularly important in additive manufacturing from powder layers, as manufacturing times can be particularly long, and therefore it is crucial to detect them as quickly as possible to correct potential defects or to avoid wasting manufacturing time by ending the manufacturing process as soon as possible if these defects cannot be corrected.

[0036] To perform detection step B, the additive manufacturing device 1 simply needs to be equipped with a conventional imaging means and a directional illumination means. Therefore, it is not necessary to use an ultra-high frequency industrial camera or a camera adapted to electromagnetic radiation outside the visible spectrum. Furthermore, the proposed detection step B advantageously includes a substep that suppresses the optical halo effect generated by a unidirectional illumination device, i.e., avoids the need to equip the additive manufacturing device 1 with an omnidirectional illumination device, which is far more complex to integrate.

[0037] To this end, defect detection step B includes substep i) an imaging means for acquiring an image of the layer of deposited additive manufacturing powder.

[0038] Next, the discrete spectral representation of the acquired image is computed in substep ii) by the Discrete Fourier Transform (DFT) method. A frequency filter is then applied to this discrete spectral representation of the acquired image in substep iii). More specifically, the applied frequency filter may be a Gaussian bandpass filter, which, in contrast to a simple high-pass filter that is thought to eliminate only the halo, allows for both the suppression of the halo effect and the suppression of noise present in the acquired image. After this, the filtered image is reconstructed from the filtered spectrum in substep iv) by the Inverse Discrete Fourier Transform (Inverse DFT) method.

[0039] Finally, the filtered image is analyzed in substep v) to detect and evaluate potential coating defects. The analysis substep includes operation a) of contrast optimization to generate a contrast image using a local contrast enhancement method, for example, by the contrast-restricted adaptive histogram equalization (CLAHE) technique. Subsequently, an edge detection or area detection (blob detection) method is performed in operation b), such as a difference in Gaussian distributions to extract variations in the texture of the additive manufacturing powder layer corresponding to the coating defect. A defect classification operation c) is then performed to classify defects, for example, having categories such as "delayed deposition," "powder deficiency," and "roller defects." Next, a score may be calculated for the overall image according to the defects detected during operation d). This score is calculated according to the geometric properties of all edges detected in the image, as well as their distribution. More specifically, the score can be calculated according to, for example, the total area of ​​the detected edges, the area of ​​the largest edge among the detected edges, the number of detected edges with an area above a threshold, the number of detected edges, the percentage of the image occupied by defects (the ratio of the total area of ​​detected edges to the total area of ​​the image), their distance from the zone in which the parts were manufactured, and their connectivity (in a topological sense). In one preferred embodiment, all of these parameters are combined according to a linear combination to obtain a score depending on several parameters. It is also possible to use an artificial learning method that directly calculates the score associated with the powder layer.

[0040] Finally, the score associated with the layer is compared to various thresholds. In the first actuation e), the score is compared to a first threshold to evaluate whether the additive manufacturing device needs to re-add the layer of additive manufacturing powder to a base or a previously solidified layer before proceeding to the next step (step B) of the manufacturing method. It is also possible to use actuation f) a counter that is updated each time the score exceeds a second threshold (which may be different from or the same as the first threshold) to determine whether it is necessary to notify the operator of manufacturing device 1 that there are too many defects in the layer of additive manufacturing powder, in other words, whether it is necessary to notify the operator when the number of counted defects exceeds a predetermined number of defects, for example, to enable him / her to terminate the manufacturing process and / or diagnose the possible causes of the defects.

[0041] Machine learning models It is also possible to use machine learning models such as support vector machines (SVMs), random forest classifiers, or neural networks to detect defects within powder layers. Next, a fault detection step can be used to generate a training dataset for training the machine learning model. That is, a score, and / or a decision to re-add a layer of additively manufactured powder before proceeding to the next step in the manufacturing process, is stored in a training database based on the images from which this score was calculated and / or the reasoning behind the decision. This database may be used, in connection with training a machine learning model, for binary classification tasks (determining whether a layer was actually added or not), or other multi-category or multi-label classification tasks (e.g., giving an assessment of coating quality), or other ordering tasks that use numerical evaluations similar to scores that allow classifying a set of coating images from best to worst (in terms of defects). [Explanation of symbols]

[0042] i. Imaging means for capturing an image of the added additive manufacturing powder layer. ii. Step of calculating the filtered discrete spectral representation of the acquired image. iv. Step of calculating the filtered image from the filtered discrete spectral representation of the acquired image. Steps to analyze filtered images B Detection step

Claims

1. A method for detecting defects in a layer of additive manufacturing powder deposited on a processing zone, Before selectively solidifying the additive manufacturing powder layer, the following steps are taken: i. A step of obtaining an image of the deposited layer of additively manufactured powder, ii. A step of determining the discrete spectral representation of the acquired image, iii. A filter having at least one cutoff frequency for filtering the discrete spectral representation of the acquired image with respect to frequency, iv. A step of determining a filtered image from the filtered discrete spectral representation of the acquired image, v. A step of analyzing the filtered image to detect defects in the layer of deposited powder, The system includes processing means (21, 31) for carrying out the following: Step v., which analyzes the filtered image, consists of the following steps: v. a) A step of enhancing the contrast of the filtered image, v. b) Step of detecting edges and / or shapes in the contrast-enhanced filtered image, v. c) A step of processing the detected edge and / or shape in order to classify the type of defect associated with the detected edge and / or shape, Equipped with, The method further comprises step v. d) calculating a score associated with a layer of deposited additive manufacturing powder, which characterizes the number and / or size of defects present in the layer. A method for detecting defects within a layer of additively manufactured powder, characterized by the following:

2. The method for detecting defects in a layer of additively manufactured powder according to claim 1, characterized in that the frequency filter applied in step iii is a band-pass filter.

3. A method for detecting defects in a layer of additively manufactured powder, comprising processing means (21, 31) that implement a machine learning model configured to detect defects in the additively manufactured layer, The step of training the model comprises the step of carrying out the method according to claim 1 or 2. A method characterized by the following:

4. A method for selectively adding a three-dimensional object from a powder layer (11), comprising a device for selectively adding a three-dimensional object from a powder layer (11), It is repeated over a predetermined number of repetitions. A. Step of adding a layer of additively manufactured powder to a base or a previously solidified layer. B. A step of detecting defects in the deposited additive manufacturing powder layer before it solidifies, C. A step of radiating a laser beam onto a first point in the layer of additively manufactured powder and solidifying a first zone of the powder layer having the first point. We will implement the following: The step of detecting defects in the layer of additive manufacturing powder deposited on the processing zone is: B. i. A substep to acquire an image of the layer of the additive manufacturing powder, B. ii. A substep to determine the discrete spectral representation of the acquired image, B. iii. A substep of filtering the discrete spectral representation of the acquired image with respect to frequency. B. iv. A substep in which a filtered image is determined from the filtered discrete spectral representation of the acquired image. B. v. A substep of analyzing the filtered image to detect defects in the layer of deposited powder, The system further comprises processing means (21, 31) for carrying out the process, Substep B.v., which analyzes the filtered image, B. v. a) Operation to enhance the contrast of the filtered image, B. v. b) Operation to detect edges and / or shapes in the contrast-enhanced filtered image, B. v. c) Operations to process the detected edge and / or shape in order to classify the type of defect associated with the detected edge and / or shape, Equipped with, The system further comprises operation B. v. d) for calculating a score associated with the layer of deposited additive manufacturing powder, A method for selectively adding and manufacturing a three-dimensional object from a powder layer, characterized by the following features.

5. The device further comprises operation B. v. e) for selectively additively manufacturing a three-dimensional object from the powder layer (1) by performing step B, when the score calculated during operation B. v. d) is above a first threshold, by re-adding the additive manufacturing powder layer to a base or a previously solidified layer, and then performing step B. A method for selectively adding and manufacturing a three-dimensional object from a powder layer as described in feature 4.

6. When the score calculated during operation B. v. d) is above the second threshold, the counter for the number of defects is updated. The method is Operation B. v. f) notifies an abnormal score when the number of counted defects exceeds a predetermined number of defects. It also has, A method for selectively adding and manufacturing a three-dimensional object from a powder layer as described in feature 4.

Images