DE102024102109A1 - Method for providing a component, computer program and electronically readable data carrier - Google Patents

Abstract

The invention relates to a method for providing a component through a production process by means of process planning, comprising the steps:
- Provision of inventory data, which includes a data set for several additional components, each of which characterizes a production process for the respective additional component; (S1)
- Recording at least one similarity criterion which is used when comparing the component with at least one of the other components stored in the inventory data; (S2)
- Combining the at least one similarity criterion into a reference similarity tuple; (S3)
- Determining at least one further component and/or another production process from the inventory data using the reference similarity tuple, wherein a similarity value is determined for the at least one similarity criterion; (S4)
- Extracting at least one piece of data from the inventory data relating to the determined at least one further component and/or to the determined at least one further production process; (S5) and
- Summarizing and providing at least one piece of data. (S6)
Furthermore, the invention relates to a computer program and a data carrier.

Description

The invention relates to a method for providing a component through a production process, in particular that of a manufacturing plant, by means of process planning according to patent claim 1. Furthermore, the invention relates to a computer program and an electronically readable data carrier according to patent claims 9 and 10.

The current state of the art is that problems that arise during the production of parts in press shops or during the commissioning of tools are reported by workers or machine operators and manually saved in a database. These databases contain fields with numerical, categorical, and text data. For regularly occurring problems, a user of such a database can generate standardized messages. Numerical and categorical data as well as the standardized messages can be evaluated using statistical methods to improve knowledge transfer between production and planning. This way, for example, weak points in a production process can be identified, which can then be used to improve the design of production processes. Reports on typical problems in production can also help determine the need for the development of new solutions.

However, not all problems can be covered by standard text messages. For example, capturing technological advances may require free-text fields in databases to effectively report problems. Currently, new production processes can lead to new challenges that cannot be foreseen in advance.

Databases that can be advantageously used for planning new production processes share information for assessing similarities between parts and/or production processes. In this context, a similarity could be related to a sheet metal part or component, its type, material type, dimensions, and/or thickness. Similarities can also be considered between production processes. For sheet metal parts, similarity criteria could include, for example, the type of press line, the system, and/or the number of operations.

When a planner or user designing a new production process analyzes emerging problems, it can be beneficial to identify problems encountered during the production of a similar production process. Consequently, data from databases containing information about problems and/or similarities must be combined to exploit them to advantage. This is a time-consuming task, especially when multiple similarity criteria are considered and/or data is stored in multiple databases. Furthermore, manual analysis is very demanding if none of the production processes exactly matches the one currently being designed.

The object of the present invention is therefore to provide a method, a computer program and a data carrier by means of which the provision of a component, in particular a component to be newly designed, can be carried out particularly advantageously by means of a production process on the basis of process planning which describes the production process for the creation or provision of the component.

This object is achieved according to the invention by the subject matter of the independent patent claims. Advantageous embodiments and further developments of the invention are the subject matter of the dependent claims, the description, and the drawings.

• In einem ersten Schritt erfolgt ein Bereithalten von Bestandsdaten, beziehungsweise historischen Daten, welche für mehrere weitere Bauteile jeweils einen Datensatz umfassen, der jeweils einen bereits durchgeführten, weiteren Produktionsprozess, beispielsweise dessen Produktionsparameter, Art der Fertigungsanlage und/oder Spezifikationen des verwendeten Halbzeugs für das jeweilige weitere Bauteil charakterisiert.

A first aspect of the invention relates to a method for providing a component, for example a component produced or to be produced by forming, through a production process, in particular, for example, a manufacturing plant, for example, a press, by means of process planning. The method according to the invention comprises the following steps:

• In a first step, inventory data or historical data is made available, which each comprise a data set for several additional components, each of which characterizes an additional production process that has already been carried out, for example its production parameters, type of production plant and/or specifications of the semi-finished product used for the respective additional component.

In a second step, at least one similarity criterion is recorded, for example based on a user specification, wherein the at least one similarity criterion is specified or used when comparing the component with at least one of the other components stored in the inventory data and is then recorded.

In a third step, the at least one similarity criterion is combined into a reference similarity tuple. The reference similarity tuple thus comprises the at least one similarity criterion and, in particular, or multiple similarity criteria if multiple similarity criteria are captured, where the tuple can be used as a vector and thus as a reference similarity vector.

In a fourth step, at least one further component and/or a further production process is determined from the inventory data using the reference similarity tuple, wherein a similarity value is determined for the at least one similarity criterion which is contained in the reference similarity tuple or is described by it.

In a fifth step, at least one piece of data is extracted or extracted, or derived or declared, from the inventory data relating to the identified at least one further component and/or to the identified at least one further production process. The at least one piece of data can describe or characterize a process characteristic or a process characteristic value, and thus in particular a deviation from an expected value or an expectation of the at least one previously performed production process, based on the inventory data.

Finally, in a sixth step, the at least one data part is summarized, in particular if several data parts are present, and the at least one data part or the several summarized data parts are provided.

The respective additional component of the several additional components that is characterized by the inventory data is a component that can be similar in type to the component to be provided, for example, both can be manufactured by forming. The additional production process of the several additional production processes is the production process of the additional component that has already been carried out and thus also planned, so that process planning has already been carried out for the additional production process, whereby this planning can, for example, also relate to a specific production facility. Advantageously, the inventory data includes, for example, all components already produced or manufactured by a manufacturer, so that in particular a particularly large data set of the additional components with the production processes on which they are based or further production processes is available.

The at least one similarity criterion is recorded in particular in such a way that a user, and thus for example the process planner who designs the process for producing the component to be provided, specifies the at least one similarity criterion, which is then recorded. This can, for example, be a specification of a semi-finished product which is formed into the component by forming, if, for example, the process planner wishes to manufacture the component to be provided from a corresponding semi-finished product which has the specification of an already produced component. The similarity value can, for example, be understood as a degree of similarity and, for example, describe a feature between the component and/or its production process and the at least one other specific component or its production process.

In other words, the method according to the invention proposes combining all available data on documentation of problems and experiences, as well as similarities in the inventory data set, and storing or making the result available in a suitable space or location, such as a centrally maintained database. The inventory data is advantageously updated regularly.

A basic idea is that problems observed from current or past production processes may occur again in new production processes if a design or process planning is not adapted accordingly.

For the procedure, a designer or planner determines all similarity criteria that can describe the current design task and/or the number of similar parts of the production process that are considered.

The at least one similarity criterion or a set of similarity criteria, in particular determined by the planner, is then summarized in a tuple or vector by means of the method according to the invention. On the basis of this tuple or vector, a machine learning method, for example in the form of a self-learning algorithm and/or a neural network, is applied in particular to determine possible parts or further components and/or production processes that are similar with regard to the reference similarity tuple or the reference similarity vector. The problems and experiences reported for these further components, which are present in the data part or can be represented by it, concerning the respective further component or the associated production process, which in particular Specifically, they are identified as similar and thus extracted using machine learning methods. The problems and experiences captured in the existing data are summarized and presented, especially to the planner or designer.

An advantage of the method is that the information obtained from the method and in particular from at least one piece of data can help to avoid the repetition of design errors, which in particular are not obvious, and to find particularly good solutions for production processes in the future.

In an advantageous embodiment of the invention, the production process is carried out at least partially by forming, in particular in a press, which can be part of a press line and/or a press shop. In other words, the component is a pressed part or formed part, and thus at least part of the production process represents forming. This results in the advantage that the method can be used particularly advantageously, for example, in automotive production, where many formed parts are used.

In a further advantageous embodiment of the invention, the at least one similarity criterion comprises a specification, for example for a semi-finished product on which the component is based, a make, a part geometry of the component and/or a system type, for example a type of press used in a press line, which, for example, carries out at least part of the production process of the component or of the at least one further component. The make can, for example, describe a type or brand of a product, such as a car, for which the component is used. Additionally or alternatively, the inventory data record and/or the at least one data part comprises a formula, a text, in particular free text, a video, an image and/or an audio recording and/or a three-dimensional model. In other words, the similarity criterion describes a general design of the component or of a semi-finished product on which the component is based, and not specific semi-finished product properties for an individual specimen of the component, which is manufactured in series, in particular using the production process. In addition, the properties of the production process are also aimed more at general properties of the production facility or a type of production facility, rather than at specific process parameters that are used in the production of an individual component. The inventory data set and thus the data part contained therein or the at least one data part that can be extracted from it can comprise virtually any type of media that can be used to characterize or describe the component or its underlying production process. This results in the advantage that, for example, a large selection and thus a derivation of the at least one data part is possible based on general data. A further advantage is that the method can process virtually any type of data and can therefore be used to improve the production process for the component to be provided.

In a further advantageous embodiment of the invention, the respective similarity value describes a distance between the reference similarity tuple or vector and the corresponding component and/or production process in a similarity space, wherein the distance in particular describes a Euclidean distance or is determined as such. In this way, a distance can be determined for virtually any data, for example also for categories, character strings or Boolean values. Depending on the type of data, the distance measure can be adjusted accordingly. In other words, when determining the at least one further component or the at least one further production process, a distance value is determined which in particular thus characterizes the similarity value, for example in such a way that it describes a distance of similarity. This results in the advantage that, for example, in the case of multiple similarity values, an order or hierarchy between the component and at least one further component can be described, so that, for example, prioritization of certain similarity values can take place.

In a further advantageous embodiment of the invention, a ranking is created based on the distance between more than one data piece and/or, in particular, the summarizing and provision of the at least one data piece takes place depending on the ranking. In other words, if more than one data piece is present, the respective data piece can be weighted so that the information contained in the data piece can be sorted according to relevance based on the value or size of the distance determined for the associated similarity criterion between the component and the further component or between the production process and the further production process. This results in the advantage that the method can be used particularly advantageously for providing the component.

In a further advantageous embodiment of the invention, at least the method step of determining the at least one further component and/or extracting the at least one piece of data, but also, for example, the method steps of determining at least one component or detecting the similarity values, are carried out using a machine learning method, in particular with a self-learning algorithm and/or a neural network. In other words, a model that is trained in particular using training data and is then capable of making independent decisions is used to carry out the method and in particular the extraction process. This results in the advantage that the method can also be applied by inexperienced users.

In a further advantageous embodiment of the invention, the machine learning method is based on unsupervised learning or a nearest neighbor algorithm, and in particular a combination of the two. The method is used in particular to find similar components. Unsupervised learning is a machine learning method in which no target values are known in advance and the model is not trained, for example, through rewards. The method attempts to recognize input patterns and, for example, perform a principal component analysis. The nearest neighbor algorithm, in particular a K-nearest neighbor algorithm, is a non-parametric, supervised learning classifier in which classifications and/or predictions are made about data points based on proximity. For example, a nearest neighbor algorithm can be used to find similar components. By using the aforementioned machine learning methods, and in particular in combination with them, machine learning can be used for the method in a particularly advantageous manner.

In a further advantageous embodiment of the invention, when compiling the at least one piece of data, the information contained therein is arranged, in particular by means of a large language model. The large language model is also based on a machine learning method and is characterized by its ability to understand language for general purposes, in particular where possible, but at least to generate it. The large language model is based in particular on an artificial neural network, which mainly comprises transformers. The underlying neural network can in particular also be trained or pre-trained using self-supervised and/or semi-supervised learning. This results in the advantage that the at least one piece of data can be prepared or presented to a user in a particularly advantageous manner, so that instructions for manufacturing the component can be derived particularly easily.

A second aspect of the invention comprises a computer program. The computer program can, for example, be loaded into a memory of the electronic computing device, in particular of a manufacturing system that at least partially provides the component, and comprises program means for executing the steps of the method when the computer program is executed in the electronic computing device or a control device.

Advantages and advantageous embodiments of the first aspect of the invention are to be regarded as advantages and advantageous embodiments of the second aspect of the invention and vice versa.

A third aspect of the invention relates to an electronically readable data carrier. The electronically readable data carrier comprises electronically readable control information stored thereon, which comprises at least one computer program as just presented and is configured such that, when the data carrier is used in an electronic computing device, it can execute a method presented here according to the first aspect of the invention.

Advantages and advantageous embodiments of the third aspect of the invention are to be regarded as advantages and advantageous embodiments of both the second and the first aspect of the invention and vice versa.

Further features of the invention emerge from the claims, the figures, and the description of the figures. The features and combinations of features mentioned above in the description, as well as the features and combinations of features mentioned below in the description of the figures and/or shown alone in the figures, can be used not only in the respective specified combination, but also in other combinations or on their own.

• 1 ein schematisches Ablaufdiagramm für ein Verfahren zum Bereitstellen eines Bauteils durch einen Produktionsprozess mittels einer Prozessplanung.

The invention will now be explained in more detail using a preferred embodiment and with reference to the drawings. It shows:

• 1 a schematic flow diagram for a method for providing a component through a production process by means of process planning.

The only 1 shows a schematic sequence of a method for providing a component, in particular of a first type or a first kind, through a production process, in particular carried out by a manufacturing facility, such as a forming press, using process planning. This means that the method begins with process planning so that the component can be manufactured, in particular in series, on the manufacturing facility using the planned production process and thus provided.

• Im ersten Schritt S1 erfolgt ein Bereithalten von Bestandsdaten, welche für mehrere weitere Bauteile jeweils einen Datensatz umfassen, der jeweils einen bereits durchgeführten, weiteren Produktionsprozess des jeweiligen weiteren Bauteils charakterisiert. Mit anderen Worten handelt es sich bei den Bestandsdaten um historische Daten, welche mehrere weitere Bauteile beziehungsweise Bauteilarten, das heißt, die Bauteile unterscheiden sich beispielsweise jeweils in ihrer Ausgestaltung leicht voneinander, so können die Bauteile unterschiedlicher Karosseriebauteile sein, sodass ein erstes der weiteren Bauteile ein linker Kotflügel und ein weiteres der Bauteile einen rechten Kotflügel darstellt, wobei das zu bereitstellende Bauteil beispielsweise ebenfalls ein Kotflügel jedoch für ein von dem Fahrzeugmodell des weiteren Bauteile abweichendes Fahrzeugmodell sein kann.

The procedure presented here essentially comprises the six steps S1 to S6, which are described below:

• In the first step S1, inventory data is made available, which data set for several further components each characterizes a further production process that has already been carried out for the respective further component. In other words, the inventory data is historical data that characterizes several further components or component types, i.e. the components differ slightly from one another in their design, for example. The components can be different body components, so that a first of the further components represents a left fender and another of the components represents a right fender, whereby the component to be made available can, for example, also be a fender, but for a vehicle model that differs from the vehicle model of the further components.

The inventory data set may in particular include any type of data such as formulas, writing, in particular free text, videos, images, audio recordings and/or three-dimensional models, such as CAD models.

It is advantageous if the inventory data includes all available data for the respective component and the production process underlying the component, which is stored centrally. This means that the inventory data can be kept available in a centrally managed database, which is run, for example, on an electronic computer, in particular a server, which is connected to several different production systems or can access their data.

In the second step S2 of the method, at least one similarity criterion is recorded, which is compared when the component is compared with at least one of the other components stored in the inventory data. The similarity criterion can, in particular, be, for example, a specification of a semi-finished product on which the component is based, or a make, such as a vehicle model, a part geometry, such as the shape of the fender, or even a system type, such as the type of press in which the fender is formed or an entire press line, for example, including a coil system, in which the semi-finished product is cut for forming.

In this case, the at least one similarity criterion is specified by a user and recorded by the electronic computing device or control device executing the method, for example.

In a third step S3, the at least one similarity criterion, i.e. in particular in the case of several similarity criteria, is combined to form a reference similarity tuple or similarity criteria vector.

In a fourth step S4, at least one further component and/or a further production process, which in particular forms the basis for the at least one further component, is determined from the inventory data using the reference similarity tuple, wherein a similarity value is determined for the at least one similarity criterion of the similarity tuple or reference similarity tuple.

In other words, at least one further component is determined or selected based on a maximum distance, which has a feature selected by the similarity criterion or by the similarity criterion, or which is available for selection based on the at least one similarity criterion. The component to be manufactured should have this maximum distance from the feature or the similarity criterion or should not exceed it. If there are multiple similarity criteria, a reference similarity tuple can be provided for each specified or recorded similarity criterion, so that, as a rule, a sufficient number of further components or further production processes are determined in step S4.

If the at least one further component or the associated further production process has been determined, in the fifth step S5 an extraction of at least one data part takes place, which in particular contains anomalies that have occurred in the further production process or have unexpectedly occurred in the further component, from the Inventory data for the identified at least one further component or the at least one further production process.

Finally, in a sixth step S6, the at least one data part is summarized and provided, in particular several data parts are prepared.

The reported problems and/or experiences contained in the existing data are captured using at least one machine learning method. In particular, the extraction, summarization, and determination in steps S3, S4, and S5, as well as the capture of similarity criteria in step S2, can be performed using machine learning methods. It appears promising to formulate an unsupervised learning problem using a nearest neighbor algorithm. This way, no training data or training is required, and unsupervised learning eliminates the need for data labeling. The principle of the nearest neighbor algorithm or method is to find a predefined number of samples from the training dataset that have the shortest distance to a given input, and to predict the label from these samples. Consequently, the prediction of the model provides a list of similar components and/or production processes concerning the component to be manufactured from the inventory data, which are thus similar with respect to the given reference similarity vector or the reference similarity tuple.

In particular, a distance measure can also be calculated for each part or production process, which reflects the distance between the reference similarity vector or the reference similarity tuple and the corresponding component and/or production process in the similarity space. This can be, for example, a Euclidean distance or another distance measure.

Furthermore, based on the distance, a list of similar parts or components, the corresponding problems and associated experiences from the historical data or inventory data can be summarized and displayed to the user or designer or planner for the production process of the component to be manufactured.

The distance makes it possible to rank the identified problems and experiences. For example, the problems and experiences to which the at least one piece of data describes and which are extracted from the existing data and are most similar in terms of the reference similarity vector or the reference similarity tuple can be given the highest rank. This at least one piece of data or the information on problems and experiences contained therein can be summarized using a large language model, for example, and can be highlighted for the user when the at least one piece of data is prepared, for example on a display element. Images and/or videos can be summarized using image recognition algorithms or computer vision, and thus for example using convolutional neural networks (CNN).

For example, it is possible to present the problems and experiences in a list sorted by the calculated ranking. The corresponding information with the highest ranking could be displayed at the top of the list. The colors, style, and/or font size of the text or a widget used to present the data piece to the user in order to output the production of the component to be manufactured can also be varied and depend on the ranking.

The method presented here is suitable for any type of data that specifically reflects problems and/or experiences. The basic idea of the method presented here is to find similar components and production processes based on similarity criteria and, from this, to identify potential problems that could arise in a similar, newly designed production process. In addition to numeric fields, problem descriptions, images, or even videos stored in free text fields can be returned to the user.

List of reference symbols

S1

first step

S2

second step

S3

third step

S4

fourth step

S5

fifth step

S6

sixth step

Claims (10)

Method for providing a component through a production process by means of process planning, comprising the steps: - keeping inventory data available which is relevant for several several further components each comprise a data set that characterizes a further production process already carried out for the respective further component; (S1) - detecting at least one similarity criterion, which is used when comparing the component with at least one of the further components stored in the inventory data; (S2) - summarizing the at least one similarity criterion into a reference similarity tuple; (S3) - determining at least one further component and/or a further production process from the inventory data using the reference similarity tuple, wherein a similarity value is determined for the at least one similarity criterion; (S4) - extracting at least one data part describing a process characteristic from the inventory data for the determined at least one further component and/or for the determined at least one further production process; (S5) and - summarizing and providing the at least one data part. (S6) Procedure according to Claim 1 , characterized in that the production process is carried out at least partly by forming, in particular in a press. Procedure according to Claim 1 or 2 , characterized in that the at least one similarity criterion describes a specification, a make, part geometry and/or a system type and/or the inventory data record and/or the at least one data part comprises a formula, a text, in particular a free text, a video, an image, an audio recording and/or a three-dimensional model. Method according to one of the preceding claims, characterized in that the respective similarity value describes a distance between the reference similarity tuple and the corresponding component and/or production process in the similarity space, wherein the distance describes in particular a Euclidean distance. Procedure according to Claim 4 , characterized in that a ranking is created for more than one data part based on the distance and/or the summarizing and the provision of the at least one data part takes place depending on the ranking. Method according to one of the preceding claims, characterized in that at least the determination of the at least one further component and/or the extraction of the at least one data part is carried out using a method of the machine learner, in particular a self-learning algorithm and/or a neural network. Procedure according to Claim 5 , characterized in that the machine learning method is based on unsupervised learning and/or on a nearest neighbor algorithm and the method is used in particular to find similar components. Method according to one of the preceding claims, characterized in that the at least one data part is arranged during the compilation, in particular by means of a large language model. Computer program which can be loaded directly into a memory of an electronic computing device, with program means for carrying out the steps of the method according to one of the Claims 1 until 8 to execute when the computer program is executed. Electronically readable data carrier with electronically readable control information stored thereon, which contains at least one computer program according to Claim 9 and are designed in such a way that when the data carrier is used in an electronic computing device, they implement a method according to one of the Claims 1 until 8 carry out.