WO2025021776A1 - Quality assurance in additive manufacturing - Google Patents

Abstract

The invention relates to a method for additive manufacturing of a three-dimensional object comprising a device having a preparation unit for at least one plasticisable material, at least one printing head having at least one discharge nozzle, a construction platform that can be moved by motors along a plurality of axes, sensors for recording physical values of the motors, an open-loop and closed-loop control unit and a human-machine operator interface. The method comprises a sensor data evaluation and regulated control of the motors, wherein plasticised material from the discharge nozzle is deposited on the construction platform and the construction platform is moved in order to manufacture the object layer-by-layer. Previously determined statistical data (10, 20, 30) of the motors (200, 201, 202) are compared with process data (40) determined during the manufacture of the object, in order to determine a deviation, wherein the statistical data (10, 20, 30) and the process data (40) comprise the recorded physical values of the sensors. A deviation of the statistical data and the process data is evaluated in order to determine anomalies of the object to be manufactured, wherein an assessment-dependent communication takes place with an artificial intelligence and/or an operator for process control.

Description

Quality Assurance in Additive Manufacturing

Description

reference to related applications

The present application relates to and claims the priority of the German patent application 10 2023 119 616.2, filed on July 25, 2023, the disclosure content of which is hereby expressly made the subject matter of the present application in its entirety.

field of the invention

The present invention relates to a quality assurance method with the features of claim 1 for the additive production of a three-dimensional object by means of a device. Furthermore, the invention relates to a device operating according to the method with the features of claim 12.

State of the art

There are already known methods for the additive manufacture of a three-dimensional object in which the manufacturing process is automatically monitored using camera surveillance. The monitoring is carried out using complex models and image processing algorithms. If anomalies occur, the manufacturing process is interrupted so that an operator can check the previous result and, if necessary, intervene in the manufacturing process to make corrections.

WO 2021/021469 A1 describes a method and a device for quality assurance when printing at least one three-dimensional object, in which two process parameters that directly relate to the object to be produced are analyzed by means of two sensors. The method comprises (a) analyzing data collected by a first sensor to identify a first deviation from a first expected value, wherein the first sensor is configured to detect a first aspect of the printing of the at least one three-dimensional object, and (b) analyzing data collected by a second sensor to identify every second deviation from a second expected value, wherein the second sensor is configured to detect a second aspect of the printing of the at least one three-dimensional object. and (c) assessing the quality of the printing of the at least one three-dimensional object taking into account the first deviation and the second deviation. Two sensor data are thus used that are directly related to the printing of the three-dimensional object, e.g. temperature (object and environment), ambient pressure, contour of the object (camera), material density, material weight, flow properties, etc., in order to detect anomalies of the object in connection with correspondingly complex models and to change process data if necessary.

EP 3 921 166 A1 describes a 3D printing method with a controller for a liquid dispensing system, in particular a print head, in which the print head current is detected by means of a sensor and then it is determined whether there is a problem in the device in response to whether the measurement meets an expected property based on the control data. Thus, it is determined whether there is an anomaly within the print head, whereupon the printing process can be stopped or process data can be changed.

EP 4 104 955 A1 and EP 4 094 867 A1 disclose a method for additively manufacturing a component and an additive manufacturing device for carrying out a method. Specifically, this is an arc wire deposition welding method or an arc wire deposition welding device in which sensors are used to measure current and voltage on a print head electrode in connection with a welding wire. Furthermore, the component to be manufactured is monitored by a camera in order to detect its contour, structure and quality defects. Additional sensors are used to detect further process data, such as the temperature of the liquid phase, the feed rate of the print head and the feed rate of the welding wire. A digital twin (model) of the object to be manufactured is generated from this data and compared with a stored model in order to identify anomalies, whereupon additive manufacturing can be stopped or process data can be changed.

EP 2 996 006 A1 discloses a method for monitoring a manufacturing and/or assembly process, which is monitored in at least three steps with at least one sensor per step. A result of the monitoring is derived from this, whereby the sensor can recognize an expression of a feature detected by it. The method is preferably based on optical sensors and particularly preferably cameras that generate image signals as sensor signals. However, the sensors can also be other devices that generate image signals as sensor signals, such as X-ray cameras, infrared cameras, thermography sensors, laser scanners and the like. Sensors can, however, but also light barriers or tactile sensors that do not generate image signals. An observed feature can be the shape and/or the size of an object processed in the corresponding step, whereby the corresponding shape or the corresponding size represents the expression of this feature. Accordingly, the nature of at least one surface of a processed object can also be a feature, whereby the concrete surface nature can be the expression of this feature. Advantageously, a limited number of surface properties can be present or specified. In addition, properties of the interior of an object can also be features, particularly if the sensor can provide information about the interior of the workpiece, such as an X-ray device or a fluoroscopy device for transparent objects. The expression of this feature is then the concrete realization of the corresponding property inside the object. The results determined over a specified period of time or over a large number of observation times can be categorized in order to derive information about the overall process. This can be the sum of identical characteristic characteristics or identical results deduced from them over an analysis period. The statistical evaluation of, for example, mean value, dispersion and trends of the results over time can provide further information about continuous changes in the system. The sensor signals of the observed steps are each compared with corresponding target sensor signals, which may have been defined in a learning mode, for example. The result of this learning mode is concrete characteristic characteristics that have been linked to target sensor signals. The target sensor signals are each assigned, at least in some areas, to a characteristic of at least one characteristic of the step that is observed by the sensor generating the corresponding sensor signal. The characteristic of at least one characteristic for new sensor sequences can be derived from this comparison. These independently recognized characteristics are stored in a characteristic memory together with their properties such as sensor location or sensor number, time and duration of the persistent characteristic and the associated generic characteristic for the characteristic. From this feature memory, the characteristics of several steps together with their properties at at least one of the observation times are compared together with other characteristics of the corresponding characteristics with certain characteristics and a monitoring result is derived from the joint comparison of individual characteristics or the combination of several synchronous or specifically asynchronous characteristics. This result can be cumulated over a monitoring period and thus statistically evaluated or fed back into the production and/or assembly system in near real time in order to assume a target state. US 2021/0308950 A1 discloses a method for the quality-assured additive production of a three-dimensional object by means of a device. The device comprises a processing unit for at least one plasticizable or plasticized material, at least one print head with at least one discharge nozzle that can be moved along two axes by means of at least two motors assigned to the axes, a construction platform that can be moved along an axis by means of at least one motor assigned to the axis, sensors for recording physical values of the motor and, if necessary, at least one print head and/or at least one processing unit, a first control and regulating unit for controlling the print head, the motor and for evaluating the sensors, and a human-machine interface. The method comprises an evaluation of sensor data, a regulated control of the motors and, if necessary, the print head and/or the processing unit, by the control and regulating unit. At least one plasticized material is deposited from the discharge nozzle onto the build platform, with the build platform and the print head being moved to produce the object layer by layer. The first control and regulating unit receives stored control codes and setting data with time information, including limit values for the setting data, for controlling the motors, the print head and the processing unit from a second control and regulating unit. The first control and regulating unit checks the data supplied by the sensors to see if the limit values have been exceeded or not met, generates so-called quality data from this, including setting data with time information, status data such as sensor data and, if applicable, error data, and supplies this to the second control and regulating unit. The stored setting data can be changed according to an error that has occurred that causes a deviation between the expected and actual process data. The second control and regulating unit outputs the quality data via the human-machine interface. The second control unit also includes further analysis of the quality data using information acquired over a longer period of time than the information used in the first control unit. Alternatively, using machine learning or artificial intelligence, patterns observed in the sensor data for quality defects are learned in advance by a learning model, and the model is used for determination after learning. The input to a learning model is not limited to the sensor data, but may also include an intended control code, the defect data, the setting and state information of the three-dimensional manufacturing, image information, and 3D measurement data.

The solutions disclosed in the prior art therefore either check components of manufacturing devices using sensors for their process-compliant function or for deviations from it, or use complex models taking into account several other process parameters, which are detected by sensors, to determine anomalies of objects to be manufactured in the form of deviations from the process or object model.

Summary of the Invention

The invention is therefore based on the object of specifying a method for the additive production of a three-dimensional object or a device for carrying out the method, which enables the detection of an anomaly in an object to be manufactured without using several different process parameters and complex models.

representation of the invention

This object is firstly achieved by a method according to the features of patent claim 1. This is a method for the quality-assured additive production of a three-dimensional object by means of a device comprising a preparation unit for at least one plasticizable or plasticized material, at least one print head with at least one discharge nozzle, a construction platform that can be moved along several axes by means of at least one motor assigned to each axis, sensors for detecting physical values of the motors and, if necessary, at least one of the elements from the group comprising the at least one print head and the preparation unit, a control and regulating unit and a human-machine interface for an operator of the device. The method comprises an evaluation of sensor data from the sensors and a controlled control of the motors and, if necessary, of the at least one element from the group comprising the at least one print head and the processing unit, by the control and regulating unit, wherein at least one plasticized material is deposited under pressure from the at least one discharge nozzle on the construction platform and wherein the construction platform is moved in order to produce the object layer by layer, comprising the following steps: a comparison of previously determined statistical data of the motors with process data determined during the manufacture of the object in the control and regulating unit to determine deviations, wherein the statistical data and the process data comprise the recorded physical values of the sensors and wherein the statistical data, consisting of process parameters and a process prediction, which form a process model that depicts the manufacturing process of the object, are obtained from process data of repeated real and/or simulated manufacture of identical objects, an evaluation of the deviations of the statistical data and the process data in the control and regulation unit to detect anomalies in the object to be manufactured and an evaluation-dependent communication of the control and regulation unit with an artificial intelligence and/or via the human-machine interface with the operator to control the process, whereby the artificial intelligence automatically adapts the control of the process in conjunction with the control and regulation unit and/or issues recommendations for action to the operator if the process data deviate from the statistical data.

The operator is advantageously supported by artificial intelligence in avoiding faults or incorrect settings or in quickly recognizing faults and correcting them based on the stored and learned information in order to create high-quality products. Countermeasures can be taken during the respective process. The artificial intelligence draws on the training data, but also learns continuously, possibly in interactive contact with an operator. If the artificial intelligence cannot or should not correct a fault independently, for example because manual intervention by an operator is still required, it can guide the operator step by step.

The previously determined statistical data can be obtained from process data of repeated real and/or simulated production of identical objects in order to advantageously have information that improves the accuracy of the anomaly determination.

Secondly, the object is achieved by a device having the features of claim 12, which is adapted to carry out the method.

Advantageous further developments are the subject of the dependent patent claims.

In a preferred embodiment of the method which advantageously facilitates the acquisition of relevant process data, the physical values comprise a motor current and/or a motor voltage and/or a dynamic electrical resistance determined from a motor current and a motor voltage and/or a mechanical resistance, i.e. physical values which can be advantageously derived or determined in particular on the basis of the performance of the motors.

In order to advantageously improve the accuracy of the method, the process data for determining deviations preferably include, in addition to the motor data of the axes the construction platform a comparison of previously determined statistical data of the print head and/or the preparation unit with process data during the manufacture of the object in the control and regulation unit, wherein the statistical data and the process data include the recorded physical values of the sensors.

Preferably, the statistical data include construction times, shift times, material consumption, speeds, energy consumption, current strengths, lengths of contours and volumes of fillings, and the process data include shift times, material consumption, temperatures, process pressures and electrical signals that are recorded with other sensors of the device. This also contributes advantageously to the accuracy of the anomaly determination.

Preferably, in order to evaluate deviations, additional process data are considered in relation to one another in order to advantageously obtain a data field that favors an assessment of anomalies.

The accuracy of the anomaly determination is advantageously promoted by the fact that the device preferably comprises additional internal, e.g. optical measuring devices, such as a triangulation sensor for distance measurement, in particular a laser-based triangulation sensor, and/or a camera and/or an interferometer, in particular a white light interferometer, for generating further process data and the method preferably comprises a determination of the surface topography or the surface quality using the further process data of the triangulation sensor and/or the camera and/or the interferometer. Tactile measuring devices are also conceivable.

In order to enable the operator to more easily determine the anomaly, a visualization of the process data is preferably provided for direct quality control by the operator.

In order to advantageously increase the application possibilities and improve the accuracy of the anomaly determination, the visualization is preferably carried out on the human-machine interface and/or on an external computer and/or a production analysis and management tool, e.g. by means of colored markings and/or diagrams and/or camera images, whereby the object is shown in a 2D or 3D image.

In order to facilitate the control and regulation of the construction platform, the construction platform is preferably moved by means of linear axes or a multi-unit robot. Preferably, the human-machine interface and/or the control and regulation unit learns from the operator's inputs by means of artificial intelligence and/or issues recommendations for action to the operator in order to advantageously improve the result of interventions.

Likewise, in the control and regulation unit, the statistical data can be preferably improved by using the process data using artificial intelligence in order to advantageously further increase the accuracy of the anomaly determination.

Productivity and quality can be advantageously improved by allowing the operator to use the human-machine interface to set a deviation limit, which, if reached or exceeded, will interrupt the production of the object.

In an embodiment which advantageously facilitates the operability of the device and the executability of the method, the human-machine interface and/or the control and regulation unit preferably contains an artificial intelligence.

The features listed individually in the patent claims can be combined with one another in a technologically meaningful manner and can be supplemented by explanatory facts from the description and by details from the figures, whereby further embodiments of the invention are shown.

Short description of the characters

The invention will now be explained in more detail using an embodiment. Shown are:

Fig. 1 is a flow chart of the method,

Fig. 2 a possible visual representation of the manufacturing process on a human-machine interface,

Fig.3 is a schematic representation of a device for additive manufacturing.

Detailed description of preferred embodiments

The invention will now be explained in more detail by way of example with reference to the accompanying drawings. However, the embodiments are only examples that are not intended to restrict the inventive concept to a specific arrangement. Before the invention is described in detail, it should be pointed out that it is not restricted to the respective components of the device and the respective method steps, since these components parts and methods may vary. The terms used herein are intended to describe particular embodiments only and are not to be used in a limiting sense. In addition, when the description or claims use the singular or indefinite articles, this also refers to the plural of these elements, unless the overall context clearly indicates otherwise.

Fig. 1 shows the principle of the method. A process model 30 is formed from process parameters 10 and a process prediction 20, which depicts the manufacturing process of an object using a method for quality-assured additive manufacturing of the three-dimensional object 290. The process model 30 can be formed based on the data from a one-time simulation and/or manufacturing of an object. The accuracy of the method can be improved by repeated real and/or simulated manufacturing of identical objects.

These - hereinafter - statistical data 10, 20, 30 form the basis for a later determination and evaluation of anomalies of the object 290 to be produced. The object is produced by means of a device comprising a processing unit 230 for at least one plasticizable or plasticized material, at least one print head 210 with at least one discharge nozzle 220, a construction platform 250 that can be moved along several axes, such as x, y and z coordinate axes, by means of at least one motor 200, 201, 202 assigned to each axis, sensors 260 for detecting physical values of the motors 200, 201, 202 and possibly of the one print head 210 and/or the processing unit or also other elements of the device. However, the construction platform can also be located on a robot arm, for example, in which case the sensors can be assigned to the parts/axes of the robot arm. Also provided is a control and regulation unit 270, which is internal and/or external depending on the embodiment, and a human-machine interface 280, which is internal and/or external depending on the embodiment, for an operator of the device.

The method comprises an evaluation of the sensor data and a controlled control of the motors 200, 201, 202 and, if necessary, at least one of the elements from the group comprising the at least one print head 210 and the processing unit 230 by the control and regulation unit 270, wherein at least one plasticized material is deposited, e.g. under pressure, from the discharge nozzle 220 onto the construction platform 250 and wherein the construction platform 250 and/or the discharge nozzle are moved relative to one another in order to produce the object 290 layer by layer. In the further course of the method, a comparison A of previously determined statistical data 10, 20, 30 of the motors 200, 201, 202 with process data 40 determined during the manufacture of the object is carried out in the control and regulation unit 270 in order to determine deviations, wherein the statistical data 10, 20, 30 and the process data 40 comprise the recorded physical values of the sensors 260 and wherein the statistical data 10, 20, 30, consisting of process parameters 10 and a process prediction 20, which form a process model 30 which depicts the manufacturing process of the object 290, are obtained from process data 40 of repeated real and/or simulated manufacture of identical objects 290.

In a further step, an evaluation of the deviations 50 of the statistical data 10, 20, 30 and the process data 40 is carried out in the control and regulation unit 270 to determine anomalies of the object 290 to be manufactured and finally an evaluation-dependent communication takes place between the control and regulation unit 270 and an artificial intelligence K1 and/or via the human-machine interface 280 with the operator to control the process, whereby the artificial intelligence automatically adapts the control of the process in conjunction with the control and regulation unit 270 if the process data 40 deviates from the statistical data 10, 20, 30 and/or issues recommendations for action to the operator. The blocks 70, 80 and 90 with dashed lines are optional blocks, the meaning of which will be discussed later. The corresponding structure of the device can be seen in Fig. 3.

The artificial intelligence derives a solution to a problem and/or a recommendation for action from a knowledge base in which, for example, a large amount of expert knowledge has been collected, which includes the knowledge and experience of a long-standing machine operator, as well as process data, process parameters, procedures, quality parameters and faults that have been learned or recorded, and this from past and current processes or cycles. A semantic database can contain additional or supplementary information from manuals, expert information, data sheets, etc.

Training data for the artificial intelligence Kl are, for example, statistical values, process information, fault classes, context information and process information (e.g. material). Quality parameters can also be learned, such as surface quality, dimensional accuracy, gloss, weight, underfilling, overfilling, local defects, surface topology (e.g. roughness), foreign material, foreign particles, material carryover. The training data also includes other parameters that depend on the process sequence, such as material consumption, production time or the like. For the production time, in particular the Shift construction time and the total construction time or production time are considered and/or learned. In principle, other parameters/criteria can be considered and/or learned

The knowledge base enables artificial intelligence to draw its own conclusions and recommendations for action, generate new knowledge and explain to the operator how the problem solution and recommendation for action came about. The most important tasks of artificial intelligence and the expert system include: interpreting data by comparing statistical data 10, 20, 30 and process data 40, classifying events, identifying causes of errors and reducing errors, eliminating critical conditions by, for example, changing control data,

Stopping the process and/or interacting with the operator, providing dialog-oriented advice to the operator, predicting events based on deviations in the process data 40 and the statistical data 10, 20, 30.

When implementing artificial intelligence and expert systems, various models can be used, such as case-based, rule-based or classifying systems. In case-based systems, similar cases to the current problem are searched for in the knowledge base and the solution is transferred to the current problem. Rule-based systems work with predefined if-then rules and solve the problem by finding and applying the rules that fit the problem. In a classifying system, artificial intelligence uses decision trees to generate independent learning processes, whereby theses for new problems are derived from given facts.

A typical process involving artificial intelligence might look like this:

A fault is detected

The disturbance is classified by a Kl model based on statistical values, process information, etc. (e.g. overfilling).

Disturbance class, context information and process information (e.g. material) are searched in a semantic database.

The database returns the most relevant sections. The return of the database, fault class, context information, process information and statistical values are passed to the chatbot/large language model (LLM) in a prompt.

Based on this information, the LLM generates instructions for correcting the fault. If possible, the control system carries out these steps independently (e.g. if only parameters need to be changed).

If this is not possible, e.g. because the operator has to intervene manually, the operator is guided through step by step.

The LLM and the database can run either on the machine's control system or in the cloud.

The coordinates of each material discharge are known from the statistical data 10, 20, 30 of the motors 200, 201, 202. At each point on the axes 240, 241, 242, e.g. at each x, y, z coordinate point, mechanical resistance data from the motors 200, 201, 202 can also be stored in the statistical data 10, 20, 30, which allow conclusions to be drawn about the correct amount of material deposited. Depending on the position and weight of the object 290 and the mechanical equipment of the machine, the axes 240, 241, 242 and thus their process data 40, e.g. in the form of mechanical resistance, experience a change in mechanical resistance during the process. If, for example, the resistance increases, the resistance increases. For example, if the mechanical resistance during movement of the axes (locally) is lower than the statistical data, the component cross-section may be overfilled at this point. This can be noticeable across layers. The same applies if the mechanical resistance is reduced compared to the statistical data 10, 20, 30. This can indicate underfilling.

Various error cases 60 - 63, e.g. error case 1, error case 2, ... error case n, ... error case n+i can thus be output by means of the human-machine interface 280. In addition to the error cases 60 - 63, statistical data 10, 20, 30 and process data 40 can also be output in a visualized 90 manner at the human-machine interface 280. The components belonging to the human-machine interface are shown in dash-dotted lines in Fig. 1.

Fig. 2 shows an example of a human-machine interface 280 with output and input areas. In one area, the path 100 is shown in a specific layer. However, a camera image 110 can also be shown in another area, which allows the operator to directly inspect the discharged material. A third area shows a 3D representation 120 of the object, as well as the display of a specific layer in the object. In a fourth area, a deviation curve 130 of process data values P, e.g. the specific current of the axes, from expected values is shown. The operator can make inputs via a final operating area 140. The output areas can also be equipped with a touchscreen function instead of or in addition to this.

The acquisition of relevant process data 40 can be made easier if the physical values include a motor current and/or a motor voltage and/or a dynamic electrical resistance determined from a motor current and a motor voltage and/or a mechanical resistance. The acquisition of currents and voltages can be carried out easily using measuring resistors and corresponding analog/digital converters, for example. An increase in mechanical resistance correlates, for example, with an increase in motor current, so that the more complex determination of mechanical resistance can be avoided.

In order to improve the accuracy of the method, the determination of deviations can, in addition to the motor data of the axes 240, 241, 242 of the construction platform 250, include a comparison of previously determined statistical data 10, 20, 30 of the print head 210 and/or the processing unit 230 with further process data 70 during the manufacture of the object 290 in the control and regulation unit 270, wherein the statistical data 10, 20, 30 and the further process data 70 include the recorded physical values of the sensors 260. If, for example, the mechanical resistance or the current increases (locally) during a movement of the axes while the process pressure increases at the same time, the component cross-section is overfilled at this point. However, if there is no increase in the mechanical resistance or the current of the axes 240, 241, 242, but an increase in the process pressure due to a blocked nozzle outlet, the component may be underfilled because not enough material can be discharged.

For the data preparation of the statistical data 10, 20, 30, the 3D data of the object 290 to be manufactured can be broken down into layers and, depending on the selected construction parameters and/or construction strategy, machine instructions can be generated that specify how much material is deposited layer by layer along which paths and per layer. Even before the component is manufactured, data can be taken from these machine instructions for process prediction. The data can be evaluated per layer and/or construction time and can include the following expected information: construction times, shift times, material consumption, speeds, energy consumption, current strengths, lengths of contours and volumes of fillings. During the manufacturing process, additional process data 70 can be continuously recorded and analyzed. This data can include the following expected information: shift times, material consumption, temperatures (e.g. in the construction space or the preparation unit 230, in particular in the form of a plasticizing unit), process pressures and electrical signals.

Taking into account the above-mentioned statistical data 10, 20, 30 and further process data 70 can further improve the accuracy of the anomaly determination. This can also be the case if process data 40, 70 in particular are considered in relation to one another.

Actual layer construction times can deviate from calculated times, for example, due to dosing processes taking too long, which can indicate insufficient material supply (e.g. trickle problems).

The material consumption is calculated based on the number of drops. If this deviates from the expected value, this could indicate a leak in the system (leakage flow from the locking ring, leakage from the nozzle/cylinder, etc.) caused by faulty components, material fatigue or faulty assembly. If this leak is located at the discharge nozzle 220 within the build space, this could lead to a reduction in component quality and even to process termination.

A drop in process pressure can indicate underfilling caused by insufficient material discharge. An increase in process pressure can be caused by local overfilling of the component layer, but also due to a partially blocked nozzle outlet caused by foreign bodies in the material or insufficiently melted material. In order to determine why an increase occurs, it could be verified by using additional sensors.

In order to detect the anomaly even more precisely or to exclude errors, additional internal measuring devices for determining the surface topography and checking the surface quality, such as a triangulation sensor for distance measurement, in particular a laser-based triangulation sensor, and/or a camera 300 and/or an interferometer 310, in particular a white light interferometer, can be included in the device and the method. These provide further process data 70, such as image data 80. In this way, process errors and process interruptions can be avoided by identifying and intervening in good time. If there are small deviations, this does not necessarily reduce the component quality as long as the deviations remain within the permissible limits.

In a further embodiment of the method that improves the accuracy of the anomaly determination, this can include a visualization 90 of the process data 40, 70 including the further process data 70 of the additional internal measuring devices for direct quality control by the operator. After the production process, the process data 40, 70 can be visualized, e.g. layer-related, for quality control in order to be able to select out potentially critical components. The data can be shown along the machine path shown, e.g. in the form of a path 100.

In order to increase the application possibilities and in turn improve the accuracy of the anomaly determination, the visualization 90 can take place on the human-machine interface 280 and/or on an external computer and/or a production analysis and management tool and the visualization 90 can take place by means of color markings and/or diagrams and/or camera images 110, wherein the object 290 can be shown in a 2D or 3D image 120. Measured variables can be displayed in color gradations, for example, in order to make critical areas quickly visible. For more precise analysis, visualized and color-coded statistical data 10, 20, 30 can be displayed on a surface together with process data 40, 70, e.g. in diagram form, as well as the camera images 110 from the process. Each layer is displayed individually. A 3D representation 120 of the object 290 including the visualized process data 40, 70 would also be conceivable.

The determination of anomalies can also be improved if the construction platform 250 is moved by means of linear axes or a multi-joint robot 320. The use of linear motors and robot arms with corresponding sensors is low-interference and highly accurate.

In an embodiment of the method that is advantageous for the operability of the device and the executability of the method, the human-machine interface 280 and/or the control and regulation unit 270 learns from the operator's inputs by means of artificial intelligence K1 and/or issues recommendations for action to the operator. In this way, recorded parameters can be traced back to specific error cases and inexperienced operators can be enabled to carry out simple and quick error correction. It also contributes to improving the accuracy of the anomaly determination if the statistical data 10, 20, 30 are improved in the control and regulation unit 270 by means of artificial intelligence Kl by using the process data 40, 70.

Productivity and quality can also be improved if the operator can set a deviation limit using the human-machine interface 280, upon reaching or exceeding which the production of the object 290 is interrupted. In this case, the size of the permitted deviations (as a percentage) can be set by the operator using the controls 140 or the touch screen in order to either let the process continue or to abort it.

The device for carrying out the method for quality-assured additive production of a three-dimensional object 290 comprises, according to Fig. 3, a processing unit 230 for at least one plasticizable or plasticized material, at least one print head 210 with at least one discharge nozzle 220, a construction platform 250 that can be moved in an x, y and z coordinate axis or along several axes 240, 241, 242 by means of at least one motor 200, 201, 202 assigned to each axis, sensors 260 for detecting physical values of the motors 200, 201, 202 and optionally of the at least one print head 210 and the processing unit 230, an internal and/or external control and regulating unit 270 and a human-machine interface 280 for an operator of the device.

It is advantageous for the operability of the device and the executability of the method if the human-machine interface 280 and/or the control and regulation unit 270 contains an artificial intelligence Kl.

Furthermore, it is advantageous for the accuracy of the process if the construction platform 250 is moved in the x, y and z coordinate axes by means of linear axes. These can be driven by linear motors and contain corresponding sensors for current, voltage and mechanical resistance.

It is to be understood that this description is susceptible to various modifications, changes and adaptations which come within the range of equivalents to the appended claims. list of reference symbols

10 Process parameters 20 Process prediction 30 Process model 40 Process data 50 Evaluation of deviations 60 Error case n 61 Error case n+1 62 Error case n+2 63 Error case n+i 70 Additional process data 80 Image data 90 Visualization 100 Path 110 Camera image 120 3D representation of the object, display of a specific layer 130 Deviation of process data values from expected values 140 Control elements 200, 201, 202 Motors 210 Print head 220 Discharge nozzle 230 Preparation unit 240, 241, 242 Axes 250 Construction platform 260 Sensors 270 Control and regulation unit 280 Human-machine interface 290 Object 300 Camera 310 Interferometer 320 Robot

Kl Artificial Intelligence P Process data value A Comparison of statistical data and process data

Claims

patent claims 1. Method for the quality-assured additive production of a three-dimensional object by means of a device comprising - a processing unit (230) for at least one plasticizable or plasticized material, - at least one print head (210) with at least one discharge nozzle (220), - a construction platform (250) which can be moved along several axes (240, 241, 242) by means of at least one motor (200, 201, 202) assigned to each axis, - sensors (260) for detecting physical values of the motors (200, 201, 202) and, if necessary, at least one of the elements from the group comprising the at least one print head (210) and the processing unit (230), - a control and regulation unit (270) and - a human-machine interface (280) for an operator of the device, - wherein the method comprises an evaluation of sensor data from the sensors (260) and a controlled control of the motors (200, 201, 202) and, if necessary, of the at least one element from the group comprising the at least one print head (210) and the processing unit (230), by the control and regulation unit (270), wherein at least one plasticized material from the at least one discharge nozzle (220) is deposited on the construction platform (250) and wherein the construction platform (250) is moved in order to produce the object (290) layer by layer, characterized in that the method further comprises: - a comparison (A) of previously determined statistical data (10, 20, 30) of the motors (200, 201, 202) with process data (40) determined during the manufacture of the object (290) in the control and regulation unit (270) to determine deviations, wherein the statistical data (10, 20, 30) and the process data (40) comprise the recorded physical values of the sensors (260) and wherein the statistical data (10, 20, 30), consisting of process parameters (10) and a process prediction (20), which form a process model (30) which depicts the manufacturing process of the object (290), are obtained from process data (40) of repeated real and/or simulated manufacture of identical objects (290), - an evaluation of the deviations of the statistical data (10, 20, 30) and the process data (40) in the control and regulation unit (270) to detect anomalies of the object to be manufactured (290) and - an evaluation-dependent communication of the control and regulation unit (270) with an artificial intelligence (Kl) and/or via the human-machine interface (280) with the operator for controlling the method, wherein the artificial intelligence, in the event of a deviation of the process data (40) from the statistical data (10, 20, 30), Connection with the control and regulation unit (270) automatically adapts the control of the process and/or issues recommendations for action to the operator. 2. Method according to claim 1, characterized in that the physical values comprise a motor current and/or a motor voltage and/or a dynamic electrical resistance determined from a motor current and a motor voltage and/or a mechanical resistance. 3. Method according to one of claims 1 or 2, characterized in that it comprises a comparison (A) of previously determined statistical data (10, 20, 30) of the print head (210) and/or the processing unit (230) with process data (40, 70) during the manufacture of the object (290) in the control and regulating unit (270) for determining deviations, wherein the statistical data and the process data comprise the recorded physical values of the sensors (260). 4. Method according to one of the preceding claims, characterized in that the statistical data (10, 20, 30) comprise construction times, shift times, material consumption, speeds, energy consumption, current intensities, lengths of contours and volumes of fillings and that the process data (40, 70) comprise shift times, material consumption, temperatures, process pressures and electrical signals which are recorded with further sensors (260) of the device. 5. Method according to one of the preceding claims, characterized in that, in order to evaluate deviations, additional process data (40, 70) are considered in relation to one another. 6. Method according to one of the preceding claims, characterized in that the device comprises additional internal optical measuring devices, such as a triangulation sensor for distance measurement, in particular a laser-based triangulation sensor, and/or a camera (300) and/or an interferometer (310), in particular a white light interferometer, for generating further process data (70), and that the method comprises a determination of the surface topography or the surface quality by means of the further process data (70) of the triangulation sensor and/or the camera and/or the interferometer. 7. Method according to claim 6, characterized in that it comprises a visualization (90) of the process data (40, 70) for quality control by the operator, which (90) is displayed on the human-machine interface (280) and/or on an external computer and/or a production analysis and management tool and that the visualization (90) is carried out by means of colored markings and/or diagrams and/or camera images (110), wherein the object (290) is represented in a 2D or 3D image (120). 8. Method according to one of the preceding claims, characterized in that the human-machine interface (280) and/or the control and regulating unit (270) learns from operator inputs by means of artificial intelligence (Kl) and/or issues recommendations for action to the operator. 9. Method according to one of the preceding claims, characterized in that in the control and regulating unit (270) the statistical data (10, 20, 30) are improved by using the process data (40, 70) by means of artificial intelligence (Kl). 10. Method according to one of the preceding claims, characterized in that the operator can set a deviation limit by means of the human-machine interface (280), upon reaching or exceeding which the production of the article (290) is interrupted. 11. Device for the quality-assured additive production of a three-dimensional object (290), comprising a preparation unit (230) for at least one plasticizable or plasticized material, at least one print head (210) with at least one discharge nozzle (220), a construction platform (250) that can be moved along several axes (200, 201, 202) by means of at least one motor (240, 241, 242) assigned to each axis, sensors for detecting physical values of the motors (240, 241, 242) and, if necessary, at least one of the elements from the group comprising the at least one print head (210) and the preparation unit (230), a control and regulating unit (270) and a human-machine interface (280) for an operator of the device, characterized in that it is set up to carry out the method according to one of claims 1 to 10. 12. Device according to claim 11, characterized in that the human-machine interface (280) and/or the control and regulation unit (270) contains an artificial intelligence (Kl).