Apparatus and system for providing optical real-time information about a process using an optical neural network and method for providing an apparatus

Abstract

An embodiment in accordance with the present invention includes an optical device for providing optical real-time information regarding a process, the optical device configured to detect an optical input signal emitted and / or reflected by the process, the optical device comprising an optical neural network configured to provide optical real-time information regarding the process based on the optical input signal. Additionally, a method for providing a system and apparatus for providing optical real-time information is disclosed.

Description

[Technical Field]

[0001] SUMMARY OF THE INVENTION Embodiments in accordance with the present invention relate to an apparatus and system for providing optical real-time information about a process using an optical neural network, and a method for providing such an apparatus.

[0002] Further, embodiments include optical process monitoring using diffractive neural networks, optical process monitoring using diffractive deep neural networks, and / or optical image processing using diffractive deep neural networks. [Background technology]

[0003] In laser material processing and laser measurement technology, systems are used to monitor the process state during processing. This process monitoring is often based on a system consisting of illumination, monitoring optics, and a detector (CCD chip). An image of the process zone is generated on the detector, which is evaluated by image processing methods in a connected computer. Based on this evaluation, the process quality can be inferred and the treatment process can be adjusted. In modern approaches to process monitoring, neural networks are used to process image data (inline) and adapt the process parameters. So far, this has been carried out exclusively in the electronic computing units that control the treatment plant.

[0004] Conventional technology These and very similar approaches can also be found in patents or utility models or respective applications (e.g., U.S. Patent No. 65974491, Chinese Utility Model No. 216680796, U.S. Patent No. 5517420, Canadian Patent Application Publication No. 2467221, Chinese Utility Model No. 201052570). The drawbacks of the prior art are, among others, latency (the duration between data acquisition and output of control parameters) and the hardware required for image processing (often a high-performance computer with multiple GPUs). The hardware is expensive, large (server racks), and consumes high energy during operation. Energy consumption can be an obstacle for some mobile applications. The resources required for a high-performance computer scale depending on the number of sensors used. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] U.S. Patent No. 65974491 [Patent Document 2] China Utility Model No. 216680796 [Patent Document 3] U.S. Patent No. 5,517,420 [Patent Document 4] Canadian Patent Application Publication No. 2467221 [Patent Document 5] China Utility Model No. 201052570 Summary of the Invention [Problem to be solved by the invention]

[0006] There is therefore a need for improved concepts for process monitoring.It is therefore an object of the present invention to provide a concept for process monitoring that allows an improved trade-off between the complexity of the required devices, the energy consumption and the quality of the process monitoring.

[0007] This object is solved by the subject matter of the independent claims. Developments of the invention are defined in the dependent claims. [Means for solving the problem]

[0008] An embodiment in accordance with the present invention includes an optical device for providing optical real-time information regarding a process, the optical device configured to detect an optical input signal emitted and / or reflected by the process, the optical device comprising an optical neural network configured to provide optical real-time information regarding the process based on the optical input signal.

[0009] The embodiments are based on the idea of ​​performing process evaluation by an optical neural network. In general, the optical neural network can be, for example, a diffractive neural network. The diffractive neural network can, for example, comprise a sequence of multiple diffractive elements configured to modulate radiation, i.e., for example, an optical input signal or an optical signal derived therefrom, in phase and amplitude. The inventors have found that the optical neural network can perform process evaluation in real time, such that the device can provide real-time information about the process.

[0010] Here, real-time information refers to information that is generated at the speed of light, for example, from detection to provision, i.e., the delay in the travel path from the location of emission of the optical input signal in the process to the location of emission of the real-time information in the device occurs based on the speed of light. Thus, real-time information may be information whose provision or processing is limited solely by the speed of light.

[0011] Furthermore, the inventors have found that this can result in significant energy savings. Each process evaluation can be performed entirely optically by an optical neural network, so that no energy needs to be provided for electronic evaluation. Alternatively, for example, only particularly computationally intensive parts of the evaluation can be performed optically. Optionally, any part of the evaluation can be performed optically. In either case, energy savings can be achieved compared to electronic evaluation, thereby enabling, for example, moving process monitoring.

[0012] It is noted here that, for example, with regard to production machines, high mobility may in some cases, or even usually, be less relevant, but embodiments may still have the advantage that it may be designed in a very compact manner, e.g., with a small volume, and / or may enable a compact, e.g., spatially small configuration of such production machines. Furthermore, according to embodiments, no (large, e.g., powerful) high-performance computer in the machine or access to cloud computing with the respective infrastructure is required.

[0013] Furthermore, costs for process monitoring can be saved, since due to the possible optical pre-processing less powerful computing units can be used for the electronic evaluation.

[0014] The optical input signal can be, for example, a signal emitted and / or reflected by the process. Thus, the evaluation can be based, for example, only on light generated by the process itself. Alternatively, light generated independently of the process, for example, ambient light or light from a specially provided light source, can be reflected by the process, for example, by a workpiece to be measured or processed, and serve as the input signal.

[0015] The process may be, for example, processing and / or measurement of a workpiece (e.g., a material processing process of the workpiece and / or a measurement process on the workpiece by using a laser), where a signal emitted or reflected by the workpiece is used as the optical input signal.

[0016] In particular, the optical device may be configured to use such radiation immediately, eg, as emitted or reflected by the workpiece, ie, without further electronic capture or rendering, for example.

[0017] In other words, the input signal can be provided using, for example, process light (eg, thermal radiation), ambient light, and / or specific external illumination (eg, coherent illumination and / or structured illumination).

[0018] Here, embodiments can be used, for example, for process monitoring or measurement techniques in general (e.g., water jet cutting, tape placement with IR radiation, etc.), and in particular for process monitoring or measurement techniques in laser-based methods (e.g., laser beam welding, drilling, additive manufacturing, LIBS, laser triangulation, etc.) As mentioned above, the use of optical neural networks for process monitoring, for example in the above-mentioned application fields, is an inventive idea.

[0019] In particular, certain embodiments include novel components and methods for detecting and evaluating process images. Here, a diffractive neural network (DNN, or generally, an optical neural network) can be used to directly evaluate light (i.e., emitted light and / or reflected light) coming from the process zone in the processing optics, outputting an intensity pattern (e.g., instead of or in addition to 2D image data). The results of the evaluation or new control and / or adjustment signals can be encoded in the intensity pattern. Compared to conventional methods (e.g., evaluation in an electronic computing unit), embodiments have the advantage that data processing occurs at the speed of light. This is based on the inventors' discovery that the duration of image processing, derivation of quality features, and new control signals is crucial for the speed of adjustment. Thus, image analysis can be practically instantaneously available to the adjustment unit. Thus, embodiments can avoid or reduce longer computation times in computers (e.g., in the case of further electronic processing of real-time information).

[0020] According to an embodiment, the optical device is further configured to capture the optical input signal immediately prior to the process to detect the optical input signal emitted and / or reflected by the process.

[0021] For example, the device can be directly integrated into elements close to the process, as opposed to an "external" evaluation of the image, for example, electronically detected by a computer remote from the process. For example, optical devices or components of optical devices, such as optical neural networks, can also be integrated into the processing optics of the respective process.

[0022] Thus, the input signal can be directly acquired by the process without being transferred, for example, using a waveguide. For example, an optical device can be configured to be placed above the same fluid in which the process is carried out in order to capture the optical input signal directly from the fluid.

[0023] The inventors have found that the optical device of the present invention requires only a small installation space and can therefore be integrated into the elements required for the process (i.e., simply put, elements that already exist), thus saving space on the one hand and preventing or reducing the effects of possible error sources resulting from signal transmission (e.g., signal loss in electrical lines, electromagnetic interference on lines) on the other hand.

[0024] Further, according to an embodiment, the optical device comprises optical means configured to obtain an optical input signal and to provide, based on the input signal, an optical signal to the optical neural network for determining real-time information. Optionally, the optical means may include at least one of a lens, a beam splitter, a mirror, a wavelength filter, and / or an aperture.

[0025] The inventors have found that optical input signals can be rendered for optical neural networks by optical means, for example, to enable improved classification results. Furthermore, the use of optical elements allows for flexibility when integrating optical devices, since the optical input signals can be guided (e.g., by mirrors) to advantageous installation positions for the optical device.

[0026] According to an embodiment, the process includes beamforming of the process beam by the processing optics, and the optical device is configured to obtain the optical input signal via the processing optics and / or individual optical subelements of the processing optics.

[0027] Here, the process of the present invention is not limited to a specific process and each process beam. The process beam can be, for example, a laser beam or an electron beam (E-beam). Here, the processing optical system can include any combination of optical elements, such as one or more optical mirrors, lenses, and / or prisms.

[0028] Furthermore, it should be noted that the processing optics, or parts thereof, may also act as or be used as optical means.

[0029] The inventors have found that the processing optics, or even individual optical sub-elements of such optics, can be shared by both the process and the process monitor, which can save device and installation space.

[0030] According to an embodiment, the optical device includes a beam splitter configured to receive the optical input signal via the processing optics and / or via individual optical sub-elements of the processing optics. Furthermore, the optical device is configured to separate the optical input signal from the process beam by the beam splitter and provide the optical input signal to the optical neural network by the beam splitter. Splitting the beam can include, for example, redirecting it in different directions.

[0031] The inventors have found that this enables, for example, coaxial process monitoring: each process optic can form a process beam and direct it to the workpiece, and radiation emitted and / or reflected by the workpiece can be directed through the same process optic (or at least the same part thereof) as the optical input signal, and the optical path can be adapted by a beam splitter so that the optical input signal is provided to the optical neural network and is not directed, for example, to the region where the process beam is generated.

[0032] Specifically, for example, in so-called coaxial process monitoring, sensor technology, such as the device of the present invention, can be integrated directly into the process head and simultaneously influence the process laser. The trick here is that both beams (process beam, e.g., process laser beam, and optical input signal, e.g., measurement beam from the process zone) partially pass through the same optical element and are separated via one (or more) beam splitters away from the process zone. Thus, according to the present invention, both parts (sensor technology and application) can work together. This allows for efficient component integration.

[0033] According to an embodiment, the optical neural network comprises at least one of a diffractive optical element, a spatial light modulator, and / or a meta-optics system. For example, the optical device and / or optical neural network (ONN) configuration of the present invention can include any combination of at least two diffractive optical elements (DOEs) and / or spatial light modulators (SLMs) (or one each, or, for example, just one or the other, or just multiple DOEs or just multiple SLMs), classical optical components (e.g., lenses, mirrors, apertures, wavelength filters, etc.), and / or classical (computer-based) AI methods (the ONN, for example, replaces several computationally intensive levels).

[0034] According to an embodiment, the process is a controlled and / or regulated process, and the optical neural network is configured to determine control and / or regulation signals based on the optical input signals and to encode the control and / or regulation signals into optical real-time information.

[0035] In this way, control and / or regulation information can be provided in real time, particularly for very fast processes, i.e., those with high recovery time requirements. The respective regulation can also be performed in a completely similar manner, for example, based on the detection of the coded information without AD conversion. Here, for example, the amplitude or phase of the detected signal can be used directly for regulation and / or control.

[0036] According to an embodiment, the optical neural network is configured to determine a process parameter based on the optical input signal and encode the process parameter into optical real-time information.

[0037] Here, the process parameter is, for example, a quantity related to the process. In the context of processing a workpiece, the process parameter can, for example, describe the quality of the processed workpiece. Furthermore, measurement results from a process beam can also form the process parameter. Thus, for example, in an optical cutting method, a laser can provide a process beam that is reflected by a surface such that the reflected signal forms an optical input signal. Here, the process parameter can, for example, include height measurement information. The inventors have found that, according to the present invention, process evaluation can be provided quickly and with low energy consumption.

[0038] It should be noted here that the process according to the embodiments does not necessarily have to be controlled or regulated. The embodiments include and / or address specific applications in which, for example, a quality feature (as an example of a process parameter) is "only" monitored. This may be, for example, the formation of splatter or errors in a weld seam. Thus, the respective information can be encoded into the optical real-time information.

[0039] According to embodiments, the optical device is configured to modify the phase and / or amplitude of an optical input signal to provide optical real-time information in the form of a light spot, a line beam, and / or a two-dimensional light array.

[0040] Providing real-time information as light points allows for representation of information that is easy to evaluate. A line beam, for example, by the centroid of the line, can indicate a scalar value as a process parameter or regulation and / or control information (e.g., in the interval [0,1], e.g., the centroid on one side of the line corresponds to zero, the centroid on the other side corresponds to one, and each intermediate value). Thus, a two-dimensional light array can indicate two-dimensional information, e.g., encoding by pattern or centroid.

[0041] It should be noted that an optical input signal, for example in the form of an input optical field, may contain exactly one phase and amplitude, and as a result, the device may be configured to modify the phase and / or amplitude of the optical input signal. Furthermore, embodiments may address or process input signals with partial coherence by multiple overlapping fields with different phases.

[0042] According to an embodiment, the optical neural network comprises at least one static optical element and at least one adaptive optical element, the adaptive optical element being configured to modify the processing of an optical input signal in the optical neural network.

[0043] Thus, for example, transfer learning methods can be used, where static elements are generated, for example, according to a general pre-training, and dynamic elements are adapted in an application-specific manner with respect to a second training. Furthermore, in this way, readjustments can be made during the lifetime of the device.

[0044] According to an embodiment, the optical device further comprises an optical filter configured to wavelength-selectively filter the optical input signal to provide a filtered optical input signal to the optical neural network.

[0045] In this way, for example, wavelengths can be selected that allow particularly important process analysis.

[0046] According to an embodiment, the optical device further comprises a light source configured to illuminate the process and generate an optical input signal that is emitted and / or reflected by the process. Such a light source may, for example, be coaxially integrated into the process element.

[0047] According to an embodiment, the optical neural network is configured to process light having particular optical properties, and the light source is configured to provide light having the particular optical properties.

[0048] In other words, for example, the light source and the optical neural network can be tuned to each other. The optical characteristics can be wavelength specific. In this way, the accuracy and / or efficiency of the neural network can be improved.

[0049] According to an embodiment, the optical device comprises a beam splitter, a beam splitter or another beam splitter, and the optical device is configured to separate light of the light source from the optical input signal by the beam splitter or another beam splitter, and to provide the optical input signal to the optical neural network by the beam splitter or another beam splitter.

[0050] As mentioned above in relation to the processing optics, the integration of the light source and the optical neural network, eg coaxially, together with the process beam source can also be performed, eg in the processing head or processing optics.

[0051] It should be noted that, according to embodiments, for illumination (or illumination range), wavelengths between 10 μm (thermal radiation) and 400 nm (visible light) can be used. As mentioned above, external illumination (e.g., by a light source) can be performed coaxially.

[0052] An advantage of embodiments in wavelength ranges beyond UV (ultraviolet) light (i.e., wavelengths above 400 nm, for example) may lie in simple or, for example, simpler fabrication of optical elements (e.g., DOEs and / or SLMs) for, for example, optical neural networks.

[0053] According to an embodiment, the optical device is configured to be powered for providing optical real-time information solely by the optical input signal, thus addressing mobile applications and applications with high requirements on energy consumption.

[0054] Further, embodiments of the present invention include an optical system including an optical device according to any of the embodiments disclosed herein and a detector (e.g., a CCD chip) configured to detect optical real-time information and provide an electrical signal based on the optical real-time information, wherein the detector includes at least one of a photodiode, a line detector, and / or an area detector.

[0055] In this way, processed or pre-processed electrical signals can be provided at high speed for optical processing in an optical neural network.

[0056] Further, according to an embodiment, the optical system comprises processing means configured to control and / or regulate the process based on the electrical signal of the detector and / or to provide information regarding the process based on the electrical signal of the detector.

[0057] Here, the processing means can operate, for example, digitally or, in the case of very fast regulation circuits, analogously. The control and / or regulation of the process can in particular include the control and / or regulation of a process plant. For example, a plant including a laser can be regulated and / or controlled. The control and / or regulation can include, for example, the feed rate, the adaptation of the material supply, and / or the adaptation of or for the process gas.

[0058] According to an embodiment, the process is a material treatment process and / or a measurement process by using a laser (e.g. a laser measurement process, e.g. Laser Induced Plasma Spectroscopy, LIBS), and the processing means is configured to control and / or adjust the laser based on the electrical signal of the detector. Furthermore, for example in a machine of the process (e.g. a machine including a laser), other parameters can be controlled and / or adjusted, e.g. axes, scanner, protective gas supply.

[0059] Additionally, embodiments in accordance with the present invention include a method for providing an optical device (e.g., one of the optical devices described above), the optical device comprising an optical neural network configured to provide optical real-time information about a process based on an optical input signal emitted and / or reflected by the process, where the optical neural network comprises at least one first optical element and at least one adaptable optical element.

[0060] Here, the method includes simulated pre-training of a virtual model of the optical neural network using a first set of training data, and at least one first optical element and at least one adaptive optical element are mapped to the virtual model.

[0061] The method further includes generating an optical neural network based on the pre-trained model; adapting at least one adaptable optical element in the virtual pre-trained model of the optical neural network based on simulated training of the virtual pre-trained model with a second set of training data; and adapting the at least one adaptable optical element of the optical neural network according to the adapted virtual model of the optical neural network to provide an optical device.

[0062] The present inventors have found that transfer learning methods can be applied to optical neural networks by using adaptive optical elements. For example, general pre-training can be considered using at least one first optical element, and then application-specific "fine training" can be mapped using adaptive elements. Here, training can be efficiently performed in a computer-assisted manner.

[0063] It should be noted here that, according to an embodiment, training can be performed simply by using images in a computer, without using adaptive or dynamic elements. In this way, traditional training methods (from applications in image recognition) can also be used for training.

[0064] Furthermore, it should be noted that, in general, according to embodiments, the at least one first optical element and the at least one adaptive optical element can be arranged in any order within the optical neural network. Along the optical path, starting from the process, the at least one first optical element can be arranged, for example, before or after the at least one adaptive optical element. The adaptive optical element can, for example, form the front or rear optical element of the optical neural network. Thus, for example, the first (e.g., in the sense of being the front with respect to the optical path starting from the process) optical element can be adaptive, and all optical elements behind it can be static. Conversely, for example, the front optical element can be static, and only the rear optical element can be adaptive or dynamic.

[0065] According to an embodiment, the at least one first optical element is a static or adaptive (e.g. dynamic) optical element, so that configurations with only adaptive optical elements are possible.

[0066] According to an embodiment, the method further includes generating the first and / or second set of training data by using a camera to capture optical input signals via optical means and using the optical means to provide optical signals to the optical neural network for determining real-time information based on the optical input signals.

[0067] The inventors have found that simply using two optical means can improve training efficiency, since both the training data and the optical input signal can be provided by the same optical means. Thus, errors caused by different optical characteristics (e.g., component tolerances) when using two different optical means can be offset. Note that the optical means can also be formed by a subelement of the processing optics.

[0068] Examples according to the present disclosure are discussed in more detail below with reference to the accompanying drawings: With regard to the illustrated schematic diagrams, it should be noted that the illustrated functional blocks can be considered both as elements or features of the apparatus of the present disclosure as well as the respective method steps of the method according to the present disclosure, from which the respective method steps of the method according to the present disclosure can be derived. [Brief explanation of the drawings]

[0069] [Figure 1] FIG. 1 is a schematic diagram of an optical device for providing optical real-time information about a process according to an embodiment of the present invention. [Figure 2] 1 is a schematic diagram of an optical device including additional optional features according to an embodiment of the present invention. [Figure 3A] 1 is a schematic diagram of an embodiment according to the invention, in which the process includes beam shaping of the process beam by processing optics; [Figure 3B] 1 is a schematic diagram of an embodiment according to the invention, in which the process includes beam shaping of the process beam by processing optics; [Figure 3C] 1 is a schematic diagram of an embodiment according to the invention, in which the process includes beam shaping of the process beam by processing optics; [Figure 4] 1 is a schematic diagram of a system according to an embodiment of the present invention. [Figure 5] 1 is a schematic diagram of an optical device according to an embodiment having an adaptive optical element. [Figure 6] 1 is a schematic block diagram of a method of the present invention for providing an optical device; [Figure 7] FIG. 10 is a schematic diagram of a system according to an embodiment having further optional features. DETAILED DESCRIPTION OF THE INVENTION

[0070] Before the embodiments of the present invention are discussed in more detail below based on the drawings, it should be noted that identical, functionally equivalent or equivalent elements, objects and / or structures in different drawings are provided with the same or similar reference numerals so that the descriptions of these elements shown in different embodiments are interchangeable or inter-applicable.

[0071] 1 shows a schematic diagram of an optical apparatus for providing optical real-time information about a process according to an embodiment of the present invention. The optical apparatus 100 includes an optical neural network 110.

[0072] The optical input signal 101 may, for example, be emitted by the process (e.g., in the case of thermal emissions) and / or reflected by the process (e.g., when processing or measuring a workpiece using a laser where laser radiation is reflected from the workpiece).

[0073] Thus, the optical device 100 is configured to detect an optical input signal. The optical neural network 110 is configured to provide optical real-time information 112 about the process based on the optical input signal.

[0074] Figure 2 shows a schematic diagram of an optical apparatus having additional optional features according to an embodiment of the present invention. Figure 2 shows apparatus 200 including optical neural network 210, optional filter 220, optional optical means 230, and optional light source 240. As described with reference to Figure 1, optical neural network 210 is configured to provide optical real-time information 212 regarding process P.

[0075] As shown in FIG. 2, optical device 200 may optionally be configured to capture an optical input signal directly from process P to detect an optical input signal 210 emitted and / or reflected by the process.

[0076] 2 shows an example of an embodiment in which the filter 220 is arranged before the optical means 230 (in the signal path starting from the process P). According to an embodiment, the optical means 230 can also be arranged after the filter 220. The filter 220 is configured to, for example, wavelength-selectively filter the optical input signal 201, for example, emitted and / or reflected by the process P. In this way, wavelengths that make little or no contribution to the process classification can be removed.

[0077] For example, in the case of a process that emits no radiation or insufficient radiation (to determine real-time information), the apparatus 200 of the present invention can include a light source 240. The light source can illuminate or irradiate the process such that light 241 can cause a reflected signal 201 that can form an input signal. However, it should be noted here that each input signal 201 can also include a superposition of process radiation, such as thermal radiation, with the reflected radiation of the illumination 240 or the reflected radiation of a process beam, such as a processing laser. Obviously, a light source can also be used for a process that itself emits radiation, for example, to add information in a different wavelength range to the input signal.

[0078] Optionally, the optical diffraction neural network 210 can be configured to process light having, for example, specific optical properties, such as specific wavelengths, and thus the light source can be configured to provide light having precisely those determined optical properties (e.g., also wavelength ranges), and thus the filter can be configured to transmit only light having the specific optical properties.

[0079] Here again, an advantage of using a filter 220 may be that a simple, e.g., inexpensive light source 240 with a broad spectrum can be used, and the filter can filter out wavelengths that are irrelevant or carry little information for process classification.

[0080] The filtered input signal 221 is thus provided to optical means 230. Again, the optical means is configured to take the input signal 201, e.g., a further processed form of the filtered input signal 221, and provide it as optical signal 231 to the optical neural network 210.

[0081] The optical means may include, for example, at least one of a lens, a beam splitter, a mirror, and / or an aperture. The optical neural network may further include, for example, at least one of a diffractive optical element, a spatial light modulator, and / or a meta-optics system.

[0082] Here, the optical real-time information may be provided in the form of, for example, a light spot, a line beam, and / or a two-dimensional light array, or may include such representations. To this end, the optical neural network 210 may be configured to modify the phase and / or amplitude of the optical input signal, i.e., in particular the optionally filtered optical signal 231 and / or the filtered input signal 221.

[0083] For example, in the absence of light source 240 , optical device 200 can optionally be powered solely by optical input signal 201 to provide optical real-time information 212 .

[0084] As a further optional feature, control and / or adjustment signals determined by the optical neural network 210 can be provided in the real-time information 212, which can, for example, control and / or adjust the process P. Alternatively or additionally, process parameters determined by the optical neural network 210 can be encoded in the optical real-time information 212.

[0085] 3A-3C show schematic diagrams of an embodiment according to the invention, in which a process includes beamforming of a process beam by processing optics. Figures 3A-3C show an optical neural network ONN, a process beam source L, an optical subelement O, a beam splitter T, a light source B, and a workpiece W. The process beam source can be, for example, a laser.

[0086] 3A, for example, shows a schematic diagram of a typical coaxial configuration according to an embodiment. A process beam is formed and directed to a workpiece W by processing optics 320, which includes a first optical subelement 322 and a second optical subelement 324.

[0087] Here, an optical apparatus 330 including an optical neural network ONN 332 is configured to obtain an optical input signal 340 via an optical subelement 324 of processing optics 320. To this end, apparatus 330 comprises, as an additional optional feature, a beam splitter T, 332. Here, beam splitter 350 is configured to separate process beam 310 from reflected input signal 340 and provide the optical input signal to the optical neural network.

[0088] Such a configuration allows the optical device to be coaxially integrated into the processing optics of the process. It should be noted that the processing optics can also include only one optical subelement. The optical subelement can be, for example, a classic optical device such as a lens, an aperture, or protective glasses. It should be noted again that the processing optics or a subelement thereof can function as the optical means of the device.

[0089] 3B shows a schematic diagram of a further structure according to an embodiment, where the apparatus 300 comprises an optional light source 334 and a further beam splitter 336. As explained above, the light source 334 is configured to illuminate the process to generate an optical input signal 340 that is emitted and / or reflected by the process.

[0090] It should be noted that, in general, according to an embodiment, the input signal can include both light emitted by the process itself and light generated and reflected by the light source. The light source 334 then generates the input signal 340 such that a portion of the input signal is provided that is not emitted by the process itself.

[0091] 3B, beam splitter 332 can separate or split process beam 310 from the optical input signal. Beam splitter 336 can again separate light 335 from light source 334 from the optical input signal, thereby illuminating workpiece W and providing optical input signal 340 to optical neural network 332.

[0092] FIG. 3B shows a schematic diagram of a structure according to an embodiment in which additional illumination, optionally present apart from the process beam, i.e., for example the process beam or the measurement beam, can also be coaxially integrated.

[0093] It should be noted that, depending on the embodiment, only one of the beam splitters may be provided. In other words, illumination using a beam splitter can be used not only in combination with a beam splitter for the process beam. Furthermore, a beam splitter with, for example, multiple optical sub-elements can separate both the process beam and the light source light from the optical input signal.

[0094] Furthermore, it should be noted that the processing optics, e.g., the illumination optics capable of shaping the process beam, do not necessarily have to consist of only one element. According to embodiments, multiple lenses, mirrors, and additional individual optics can be combined to obtain the desired beam shaping. Therefore, it may be the case that only some of the elements of the processing optics, e.g., 340, are relevant to the ONN measurement beam.

[0095] FIG. 3C shows a schematic diagram of a further structure according to an embodiment, in which the process includes beamforming of the process beam by processing optics 320 (which may also include, for example, multiple subelements), and the optical device 320 is configured to acquire an optical input signal 320 via the processing optics.

[0096] One advantage of the configuration according to Figures 3A and 3B compared to configuration 3C may be, for example, that smaller optical elements can be used in the processing optics.

[0097] However, the configuration according to Fig. 3C can accommodate special cases, for example, where both beams are guided through a protective glass, which is arranged after the last beam-forming element in the optical path, for example to separate the process zone from the environment and prevent contamination. Each protective glass can also form an optical element or optical subelement of the process optics or optical means.

[0098] The configuration according to Figures 3A-3C allows for detection of feedback from the process and / or component without the optical elements of a neural network, which may destroy, for example, high power spatial light modulators as typically required in material processing processes.

[0099] 4 shows a schematic diagram of a system according to an embodiment of the present invention. The system 400 includes an optical device 410 and a detector 420 configured to detect optical real-time information 412 and provide an electrical signal 421 based on the optical real-time information. An optical input signal of a process P is indicated by reference numeral 411.

[0100] The device 410 may include, among other things, the optional features described above, both individually and in combination.

[0101] As a further optional feature, the system 400 includes a processing means 430 configured to control and / or regulate the process P based on the electrical signal 421 of the detector 420. To this end, an actuator 431 is shown as an optional feature in the process P. Alternatively or additionally, the processing means 430 may be configured to provide information about the process based on the electrical signal of the detector. Here, for example, the actuator 431 can be omitted and instead information about the process can be provided by the processing means 430.

[0102] In particular, the process may be, for example, a material processing process and / or a measurement process using a laser, and the processing means 430 may therefore be configured to control and / or adjust the laser based on the electrical signal 421 of the detector.

[0103] 5 shows a schematic diagram of an optical device according to an embodiment having an adaptive optical element. The device 500 includes an optical neural network 510 and, as a further optional feature, an optical means 520. The optical means is configured to provide an optical signal 510 to the optical neural network 510 based on an optical input signal 510, and the optical neural network 510 is again configured to provide real-time information 503.

[0104] As an optional feature, the optical diffractive neural network includes at least one static optical element 512 and at least one adaptive optical element 514, where the adaptive optical element is configured to modify the processing of the optical input signal 501 in the optical diffractive neural network. In other words, the adaptive optical element may be a dynamic optical element that is adjustable or, for example, can be changed.

[0105] In the following, a method according to the invention for providing an optical device, for example device 500 (see FIG. 5, optionally without optical means 520), is described.

[0106] FIG. 6 shows a schematic block diagram of such a method of the present invention for providing an optical apparatus, e.g., 500, comprising an optical neural network, e.g., 510, configured to provide optical real-time information, e.g., 503, about a process based on an optical input signal, e.g., 501, emitted and / or reflected by the process, the optical neural network comprising at least one first optical element, e.g., 512, and at least one adaptive optical element, e.g., 514.

[0107] Here, method 600 includes simulated pre-training 610 of a virtual model of an optical neural network with a first set of training data, wherein at least one first optical element, e.g., 512, and at least one adaptive optical element, e.g., 514, are mapped to the virtual model. Method 600 further includes generating an optical neural network, e.g., 510, based on the pre-trained model, adapting at least one adaptive optical element in the virtual pre-trained model of the optical neural network based on simulated training of the virtual pre-trained model with a second set of training data, and adapting the at least one adaptive optical element, e.g., 514, of the optical neural network according to the adapted virtual model of the optical neural network to provide an optical device, e.g., 500.

[0108] Here, one or more first optical elements, e.g., 512 in diagram 500, can optionally be static or adaptive, i.e., can be, for example, dynamic optical elements, where a combination of static and adaptive elements is also possible.

[0109] According to embodiments, some of the optical elements of the optical neural network can be adjusted, e.g., by fixed or e.g., static optical elements, based on a first training, and a second part of the optical elements of the adaptive and / or tunable network can be retuned based on the second training. In this way, for example, methods of transfer learning can be applied.

[0110] It should again be noted that training according to embodiments can be performed solely on a computer using images, and the inventive arrangements may not include or require any dynamic elements. Thus, known training methods (e.g., for applications in image recognition) can also be used to provide the inventive apparatus.

[0111] Furthermore, establishing the first and / or second sets of training data can optionally be performed using the same optical means. The inventors have found that training quality can be improved when the training data is generated using the same optical means used to provide optical signals from optical input signals for the neural network.

[0112] In this way, mapping errors can be prevented or reduced, for example, when using different optics to generate training data compared to "field use" of the optical device.

[0113] An exemplary process for the present invention's training of an ONN can include, for example: 1. Optional setup of external lighting where necessary / advantageous 2. For example, take classical camera photos for training data. Depending on the process requirements, the camera can be used only as a sensor, but also with an objective lens or in a coaxial optical path with the processing optics. This objective lens can then optionally remain in the structure (e.g., as an optical means of an optical device), or the integration of the ONN can optionally be performed coaxially.

[0114] 3. Labeling, e.g., classification of (e.g., electronic) training data: for simple features, e.g., the size of the molten bath, this can optionally be done automatically, e.g., by established image processing methods; for more complex data, this can be done manually. The training data can optionally be artificially scaled, e.g., by mirroring / rotating, if necessary.

[0115] 4. Training an ONN with labeled training data in a computer: The complexity of the ONN can be determined based on the complexity of the task, e.g., a pre-trained ONN on an image net dataset can be used with transfer learning to simply map the last, e.g., 1-2 levels, by an adaptive optical element, e.g., an SLM (which can or even needs to be trained anew). ->The benefits of rapid adaptation in serial production 5. Manufacture the required first optical element, e.g. a DOE (e.g. externally, e.g. using methods known from the prior art) and optionally fully integrate the system (e.g. together with additional optical components, e.g. depending on further application characteristics) within the process structure, e.g. coaxially.

[0116] In the following, stated differently, for example, embodiments with further optional features will again be described, and further embodiments will be described.

[0117] Embodiments are not limited to process monitoring in laser material processing. In general, image data from production processes or other fields (e.g., facial recognition, environmental analysis, autonomous driving, etc.) can be evaluated very quickly with compact sensor technology.

[0118] To this end, the embodiments include novel components and methods for detecting and evaluating process images. Here, an optical neural network, for example in the form of a diffractive neural network (DNN), is used to directly evaluate light coming from the process zone, for example in the processing optics, and thus (instead of 2D image data, for example) output an intensity pattern that can encode the results of the evaluation or new control signals. Compared to conventional methods (e.g., evaluation in an electronic computing unit or exclusively), the embodiments offer the advantage that data processing, or at least part of the data processing, can be performed at the speed of light. Image analysis is, for example, available to the control unit virtually instantly. Thus, the embodiments can avoid longer computation times in the computer.

[0119] As explained above, embodiments according to the present invention address the shortcomings of the prior art by, among other things, performing process image analysis in an optical neural network (e.g., 110, 210, 310, 332). The information carrier for the (optical) calculation is, for example, light collected from the process zone itself (e.g., in the form of an optical input signal 111, 201, 340, 411, 501). The light thus passes through the optical system and can undergo modifications (e.g., in phase and / or amplitude) that can be performed similarly to processing information in a (digital) neural network. The difference is, for example, that information processing occurs at the speed of light, and no additional (electronic) hardware is required for the calculation, apart from optical elements (e.g., 512, 514) used for phase and amplitude manipulation. Depending on the type and number of outputs of the network, detectors with individual pixels, line arrays, or area detectors (e.g., 420) can be used.

[0120] According to embodiments, image recognition and image processing are combined in an optical system that can be composed of, for example, a series of conventional optical systems (e.g., lenses, beam splitters, mirrors, ... e.g., optical means 230, 520 and / or processing optics 320), diffractive optical elements, DOEs (e.g., 512), and spatial light modulators (SLMs) (e.g., 514) such as liquid crystal-based technology or micromirror arrays. For example, an optical neural network in the form of a diffractive optical neural network (DNN) can be composed of, for example, a sequence of phase or amplitude masks and can be implemented in reflection or transmission, both statically and dynamically, using diffractive optical elements (DOEs), SLMs, or meta-optics. Here, the embodiments are not limited to a specific diffractive neural network. According to embodiments, different concepts and different types of implementations of DNNs can be used. According to embodiments, DNNs can be, for example, directly integrated into monitoring optics (see, for example, Figures 3A-3C). Evaluation can be performed, for example, using a CCD chip or, depending on the application, by line detectors or individual photodiodes. The choice of detector may depend, for example, on the number and type (scalar or binary) of the output parameters of the DNN. The optical system or optical device may further optionally include one or more wavelength-selective filters (e.g., 220), for example, to analyze only the light of a specific illumination (e.g., 240) of the process. Optionally, the illumination of the process may be performed using a coherent light source. In other words, the light source may be, for example, a coherent light source, for example, in the form of a laser.

[0121] If the optical neural network, e.g., a DNN, includes one or more dynamic elements, e.g., an SLM, the functionality of the sensor can be modified, e.g., over time. Adaptation can be used, for example, to train the DNN to adapt to environmental conditions of the process, or to change the evaluation algorithm with modified quality requirements or modifications of the process (e.g., material, product type, etc.).

[0122] In the following, inter alia, further inventive advantages of the embodiments compared to conventional solutions are explained.

[0123] Embodiments can enable very fast image evaluation in process monitoring at the speed of light, making electronic computing units otherwise used for image processing (e.g., including neural network evaluation) obsolete. Electronic computing units typically also include specific hardware, such as a graphics processing unit (GPU), for high-speed image evaluation. Each sensor for process monitoring, e.g., an optical device or system according to embodiments, can be constructed in a very small, compact, and cost-effective manner. The sensor unit can, for example, include at least one or more DOEs and a detector, e.g., a photodiode (e.g., for detecting whether the process is running within defined quality standards). These sensors may not require a connected or integrated electronic computing unit. The sensor can, for example, be directly connected to a process controller. In this way, process parameters, such as process speed, material supply, or laser power, can be directly modified without detouring through a computing unit.

[0124] In an optical neural network, the calculations for the image evaluation can be performed by the process light itself, so no additional energy is required. In other words, for example, no additional energy is consumed or needs to be consumed. The energy consumption of the entire sensor, for example, a system according to an embodiment, is, for example, related only or to a large extent to the evaluation of the detector, and optionally also to the lighting used.

[0125] Thus, embodiments may include or address applications such as laser processing and OEMs (original equipment manufacturers), particularly in the field of sensor technology.

[0126] Embodiments include, enable, or address further applications, such as analysis of production processes, environments, objects, e.g., that are based on evaluation of image data by neural networks, or that are too slow or inefficient for such evaluation.

[0127] 7 shows a schematic diagram of a system according to an embodiment with further optional features. System 700 includes an optical device 710, e.g., a sensor, having a diffractive optical element 712, optional optical means 714, e.g., including an optical element, and an optional light source in the form of an illumination source 716. Furthermore, system 700 includes a detector 720. As a further optional feature, system 700 includes a plant control 730 as well as processing means in the form of a signal line 731.

[0128] Additionally, FIG. 7 shows a processing head 740 , for example a laser processing head, as well as a workpiece 750 .

[0129] Thus, workpiece 750 can be processed by processing head 740, for example by laser beam 741. Starting from the workpiece, optical input signal 701, or a signal that may form at least a part of optical input signal 701, can be reflected by the laser beam in the direction of device 710. Furthermore, input signal 701 can include a radiation portion emitted by the workpiece itself, for example due to thermal radiation. As a further option, light source 716 can illuminate workpiece 750 and thus provide a further portion of input signal 701 based on further reflections. Such illumination can also be provided, for example, by ambient light (e.g., daylight or room lighting).

[0130] It should again be noted that any combination for generating the optical input signal can be used. For example, radiation reflected by a laser can form the input signal alone (e.g., without using an additional light source 716 in the form of a laser). For example, in processes such as waterjet cutting, only a light source can be used to generate the input signal (since there is no reflection of the laser radiation of the processing laser). In all cases, reflected room light, e.g., daylight, can form the input signal or part of it. In other processing operations, the input signal 701 can again be formed solely by the thermal radiation of the workpiece. However, any combination of these input signal generation options can also be used according to the present invention.

[0131] In other words, according to an embodiment, a scene can be made visible with both active illumination and ambient light, or the heat emissions of a process are analyzed.

[0132] Based on an input signal 701, which may be, for example, additionally filtered, for example, wavelength selectively, an optical signal 702 providing real-time information 703 is fed to an optical neural network including, for example, element 712. This may be detected by a detector 720 and passed in electronic form to a plant control 730.

[0133] It should be noted here that the detector 720 can be integrated, for example, into the device 710 or simply into a common housing. An external configuration is also possible. Based on the detected real-time information, further evaluations can be made in regulating or controlling the plant control section 730 and / or the operating head 740 (e.g., in terms of speed, power).

[0134] In other fields of application, the processing head may for example be a cutting head. All enumerations of materials, environmental effects, electrical properties, and optical properties herein are to be considered merely exemplary and not exhaustive.

[0135] Although some aspects have been described in the context of an apparatus, it will be apparent that these aspects also represent a description of the corresponding method, such that an apparatus block or device corresponds to each method step or feature of a method step. Similarly, aspects described in the context of a method step also represent a description of the corresponding block or detail or feature of the corresponding apparatus. Some or all of the method steps may be performed by (or using) a hardware apparatus, such as a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, some or more of the most important method steps may be performed by such an apparatus.

[0136] Depending on specific implementation requirements, embodiments of the present invention can be implemented in hardware or software. Implementations can be performed using digital storage media, such as floppy disks, DVDs, Blu-ray disks, CDs, ROMs, PROMs, EPROMs, EEPROMs or flash memories, hard drives, or other magnetic or optical memories on which electronically readable control signals are stored, which cooperate or can cooperate with a programmable computer system to perform the respective methods. Thus, the digital storage media can be computer-readable.

[0137] Some embodiments according to the present invention include a data carrier containing electronically readable control signals that can cooperate with a programmable computer system to cause one of the methods described herein to be performed.

[0138] In general, embodiments of the present invention may be implemented as a computer program product having program code operable to perform one of the methods when the computer program product is run on a computer.

[0139] The program code may for example be stored on a machine readable carrier.

[0140] Other embodiments comprise the computer program for performing one of the methods described herein, wherein the computer program is stored on a machine readable carrier.

[0141] In other words, an embodiment of the inventive methods is, therefore, a computer program comprising a program code for performing one of the methods described herein, when the computer program runs on a computer.

[0142] A further embodiment of the inventive methods is, therefore, a data carrier (or digital storage medium or computer-readable medium) having recorded thereon a computer program for performing one of the methods described herein. The data carrier, digital storage medium, or computer-readable medium is typically tangible or non-volatile.

[0143] A further embodiment of the inventive method is, therefore, a data stream or a sequence of signals representing the computer program for performing one of the methods described herein, The data stream or sequence of signals may for example be adapted to be transferred via a data communication connection, for example via the Internet.

[0144] A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured to or adapted to perform one of the methods described herein.

[0145] A further embodiment comprises a computer having installed thereon the computer program for performing one of the methods described herein.

[0146] Further embodiments according to the present invention include an apparatus or system configured to transmit a computer program for performing at least one of the methods described herein to a receiver. The transmission may be, for example, electronic or optical. The receiver may be, for example, a computer, a mobile device, a memory device, or a similar device. The apparatus or system may, for example, include a file server for transmitting the computer program to the receiver.

[0147] In some embodiments, a programmable logic device (e.g., a field programmable gate array, FPGA) may be used to perform some or all of the functionality of the methods described herein. In some embodiments, the field programmable gate array may cooperate with a microprocessor to perform one of the methods described herein. In general, the methods are preferably performed by any hardware apparatus. This may be universally applicable hardware, such as a computer processor (CPU), or method-specific hardware, such as an ASIC.

[0148] The apparatus described herein may be implemented, for example, using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.

[0149] The devices described herein or any components of the devices described herein may be implemented at least in part in hardware and / or software (computer programs).

[0150] The methods described herein may be implemented, for example, by using a hardware apparatus, or by using a computer, or by using a combination of a hardware apparatus and a computer.

[0151] The methods described herein or any components of the methods described herein may be performed at least in part by hardware and / or software.

[0152] The above-described embodiments are merely illustrative of the principles of the present invention. It is understood that modifications and variations of the arrangements and details described herein will be apparent to others skilled in the art. It is therefore intended that the present invention be limited only by the scope of the appended claims, and not by the specific details presented by the description and explanation of the embodiments herein.

[0153] References US6597449B1 Real time process control of optical components using linearly swept tunable laser US5517420A Method and apparatus for real-time control of laser processing of materials CA2467221A1 Method and system for real-time monitoring and controlling height of deposit by using image photographing and image processing technology in laser cladding and laser-aided direct metal manufacturing process CN000216680796U Laser power real-time online monitoring device and laser processing system CN000214161805U Laser processing quality monitoring system CN000210967526U System for monitoring laser processing performance in real time CN000206775660U Three cascade optical network monitoring devices CN000203981562U Real-time monitoring device for interaction process between optical material and laser CN000201052570Y Real time monitoring device for three-dimensional laser welding and cutting process CN000114549479A Process monitoring system and method for laser additive remanufacturing equipment CN000114486687A Multi-scale continuous observation feedback method and device for femtosecond laser processing of cells CN000114450120A Laser processing monitoring method and laser processing monitoring device CN000203981562U Real-time monitoring device for interaction process between optical material and laser CN000114549479A Process monitoring system and method for laser additive remanufacturing equipment Knaak et Al.,"A Spatio-Temporal Ensemble Deep Learning Architecture for Real-Time Defect Detection during Laser Welding on Low Power Embedded Computing Boards",Sensors, 2021 Schindhelm,"In-process quality assurance for laser beam cutting of metals," Herbert Utz, 2014 Gehrke,"Monitoring of structural properties during surface structuring with ultrashort laser pulses", Herbert Utz, 2013 Lin et al.,"All-optical machine learning using diffractive deep neural networks", Science, 2018 Chen et al.,"Diffractive Deep Neural Networks at Visible Wavelengths", Engineering, 2021

Claims

1. 1. An optical device (100, 200, 330, 410, 500, 710) for providing optical real-time information (112, 212, 412, 503, 703) regarding a process, comprising: the optical device is configured to detect an optical input signal (111, 201, 340, 411, 501, 701) emitted and / or reflected by the process; The optical device comprises an optical neural network (110, 210, 310, 332, 510) configured to provide the optical real-time information about the process based on the optical input signal.

2. The optical device (100, 200, 330, 410, 500, 710) of claim 1, configured to capture the optical input signal (111, 201, 340, 411, 501, 701) immediately prior to the process to detect the optical input signal emitted and / or reflected by the process.

3. 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, further comprising optical means (230, 520, 714) configured to acquire the optical input signal (111, 201, 340, 411, 501, 701) and provide an optical signal (231, 502, 702) to the optical neural network (110, 210, 310, 332, 510) for determining the real-time information (112, 212, 412, 503, 703) based on the input signal.

4. The optical device (100, 200, 330, 410, 500, 710) of claim 3, wherein the optical means (230, 520, 714) comprises at least one of a lens, a beam splitter, a mirror, a wavelength filter, and / or an aperture.

5. 10. The optical device of claim 1, wherein the process includes beamforming of a process beam (310, 741) by a processing optical system (320), and the optical device is configured to acquire the optical input signal (111, 201, 340, 411, 501, 701) via the processing optical system (320) and / or via individual optical subelements (322, 324, 326) of the processing optical system.

6. the optical device comprises a beam splitter (332); the beam splitter is configured to receive the optical input signal (111, 201, 340, 411, 501, 701) via the processing optics (320) and / or via individual optical subelements (332, 324, 326) of the processing optics; the optical device is configured to separate the optical input signal from the process beam (310, 741) by the beam splitter; The optical device (100, 200, 330, 410, 500, 710) of claim 5, wherein the optical device is configured to provide the optical input signal to the optical neural network (110, 210, 310, 332, 510) by the beam splitter.

7. 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, wherein the optical neural network (110, 210, 310, 332, 510) comprises at least one of a diffractive optical element, a spatial light modulator, and / or a meta-optics system.

8. the process is a controlled and / or regulated process; 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, wherein the optical neural network (110, 210, 310, 332, 510) is configured to determine control and / or adjustment signals based on the optical input signal (111, 201, 340, 411, 501, 701) and encode the control and / or adjustment signals into the optical real-time information (112, 212, 412, 503, 703).

9. 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, wherein the optical neural network (110, 210, 310, 332, 510) is configured to determine process parameters based on the optical input signals and encode the process parameters into the optical real-time information (112, 212, 412, 503, 703).

10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, configured to modify the phase and / or amplitude of the optical input signal (111, 201, 340, 411, 501, 701) to provide the optical real-time information (112, 212, 412, 503, 703) in the form of a light spot, a line beam, and / or a two-dimensional light array.

11. the optical neural network (110, 210, 310, 332, 510) comprises at least one static optical element (512, 712) and at least one adaptive optical element (514); 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, wherein the adaptable optical element is configured to modify the processing of the optical input signal (111, 201, 340, 411, 501, 701) in the optical neural network.

12. 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, further comprising an optical filter (220) configured to wavelength-selectively filter the optical input signal (111, 201, 340, 411, 501, 701) to provide a filtered optical input signal (221) to the optical neural network (110, 210, 310, 332, 510).

13. 10. The optical device (100, 200, 330, 410, 500, 710) of any one of the preceding claims, further comprising a light source (240, 334, 716), the light source configured to illuminate the process and generate the optical input signal (111, 201, 340, 411, 501, 701) that is emitted and / or reflected by the process.

14. the optical neural network (110, 210, 310, 332, 510) is configured to process light having specific optical properties; 14. The optical device (100, 200, 330, 410, 500, 710) of claim 13, wherein the light source (240, 334, 716) is configured to provide the light having the particular optical characteristics.

15. the optical device comprises a beam splitter (336), the beam splitter (332), or a further beam splitter (336); the optical device is configured to separate the light of the light source (240, 334, 716) from the optical input signal (111, 201, 340, 411, 501, 701) by the beam splitter or the further beam splitter; 15. The optical device (100, 200, 330, 410, 500, 710) of claim 13 or 14, wherein the optical device is configured to provide the optical input signal to the optical neural network (110, 210, 310, 332, 510) by means of the beam splitter or the further beam splitter.

16. The optical device (100, 200, 330, 410, 500, 710) of any one of claims 1 to 13, wherein the optical device is configured to be energized to provide the optical real-time information (112, 212, 412, 503, 703) solely by the optical input signal (111, 201, 340, 411, 501, 701).

17. An optical system (400, 700) comprising: An optical device (100, 200, 330, 410, 500, 710) according to any one of the preceding claims, a detector (320, 720) configured to detect the optical real-time information (112, 212, 412, 503, 703) and provide an electrical signal (421) based on the optical real-time information; The optical system, wherein the detector comprises at least one of a photodiode, a line detector, and / or an area detector.

18. 18. The optical system (400, 700) of claim 17, further comprising processing means (430, 730) configured to control and / or regulate the process based on the electrical signal (421) of the detector (320, 720) and / or processing means configured to provide information about the process based on the electrical signal of the detector.

19. 19. The optical system (400, 700) of claim 18, wherein the process is a material processing process and / or a measurement process by using a laser, and the processing means (430, 730) is configured to control and / or adjust the laser based on the electrical signal (421) of the detector (320, 720).

20. A method (600) for providing an optical device, comprising: the optical device (100, 200, 330, 410, 500, 710) comprises an optical neural network (110, 210, 310, 332, 510) configured to provide optical real-time information (112, 212, 412, 503, 703) about the process based on optical input signals (111, 201, 340, 411, 501, 701) emitted and / or reflected by the process; the optical neural network comprises at least one first optical element (512, 712) and at least one adaptive optical element (514); The method comprises: Simulating pre-training (610) a virtual model of the optical neural network using a first set of training data, wherein the at least one first optical element and the at least one adaptive optical element are mapped into the virtual model; generating (620) the optical neural network based on the pre-trained model; adapting (630) the at least one adaptable optical element in the virtual pre-trained model of the optical neural network based on simulated training of the virtual pre-trained model with a second set of training data; and adapting (640) the at least one adaptable optical element of the optical neural network according to the adapted virtual model of the optical neural network to provide the optical device.

21. 21. The method (600) of claim 20, wherein the at least one first optical element is a static or adaptive optical element.

22. generating said first and / or second training data sets by using a camera that detects said optical input signal (111, 201, 340, 411, 501, 701) via optical means (230, 320, 520, 714); 22. The method (600) of claim 20 or 21, further comprising: using the optical means to provide an optical signal to the optical neural network (110, 210, 310, 332, 510) for determining the real-time information (112, 212, 412, 503, 703) based on the optical input signal.

23. generating said first and / or second training data sets by using a camera that detects said optical input signal (111, 201, 340, 411, 501, 701) via optical means (230, 320, 520, 714); 22. The method (600) of claim 20 or 21, further comprising: using the same optical means to provide an optical signal to the optical neural network (110, 210, 310, 332, 510) for determining the real-time information (112, 212, 412, 503, 703) based on the optical input signal.

24. 1. An optical device (100, 200, 330, 410, 500, 710) for providing optical real-time information (112, 212, 412, 503, 703) regarding a process, comprising: the process is a material processing process and / or a measurement process by using a laser in which a workpiece is measured or processed, the optical device is configured to detect an optical input signal (111, 201, 340, 411, 501, 701) emitted and / or reflected by the process; The optical device comprises an optical neural network (110, 210, 310, 332, 510) configured to provide the optical real-time information about the process based on the optical input signal.

25. 1. An optical device (100, 200, 330, 410, 500, 710) for providing optical real-time information (112, 212, 412, 503, 703) regarding a process, comprising: the process is a laser-based material processing and / or measurement process; the optical device is configured to detect an optical input signal (111, 201, 340, 411, 501, 701) emitted and / or reflected by the process; the optical device comprises an optical neural network (110, 210, 310, 332, 510) configured to provide the optical real-time information regarding the process based on the optical input signal; The optical device is configured to capture the optical input signal (111, 201, 340, 411, 501, 701) immediately prior to the process to detect the optical input signal emitted and / or reflected by the process.

26. the process is a controlled and / or regulated process; 26. The optical device (100, 200, 330, 410, 500, 710) of claim 25, wherein the optical neural network (110, 210, 310, 332, 510) is configured to determine control and / or adjustment signals for the material treatment process and / or the measurement process based on the optical input signal (111, 201, 340, 411, 501, 701) and encode the control and / or adjustment signals into the optical real-time information (112, 212, 412, 503, 703).

27. 1. An optical device (100, 200, 330, 410, 500, 710) for providing optical real-time information (112, 212, 412, 503, 703) regarding a process, comprising: the process is a laser-based material processing and / or measurement process; the optical device is configured to detect an optical input signal (111, 201, 340, 411, 501, 701) emitted and / or reflected by the process; the optical device comprises an optical neural network (110, 210, 310, 332, 510) configured to provide the optical real-time information regarding the process based on the optical input signal; The process includes beamforming of a process beam (310, 741) by processing optics (320); the processing optics are configured to shape and direct the process beam to a workpiece; the optical device comprises a beam splitter (332); the beam splitter is configured to obtain the optical input signal (111, 201, 340, 411, 501, 701) in the form of radiation emitted and / or reflected by the workpiece via the processing optics (320) and / or individual optical subelements (322, 324, 326) of the processing optics; the optical device is configured to separate the optical input signal from the process beam (310, 741) by the beam splitter; The optical device is configured to provide the optical input signal to the optical neural network (110, 210, 310, 332, 510) by the beam splitter.

Images