Computer-implemented method and system for data consolidation between at least two factories

Abstract

The invention relates to a method for data management and / or data matching of at least two factories, wherein each factory provides at least one data set from at least one machine, the method comprising the following steps: providing (S1) a trained transformer network based on natural language processing to perform a semantic analysis of the provided data sets depending on at least one keyword and / or label; performing (S2) a data matching between the provided data sets depending on the at least one keyword to identify data points that are identical in content and / or data context or that exhibit significant similarities; creating (S3) a consolidated data set containing the identified identical or similar data points from the at least two factories;Providing (S4) the consolidated data set for use and / or optimization and / or standardization and / or knowledge transfer of at least one function and / or method relating to at least one machine.

Description

The invention relates to a computer-implemented method for data consolidation between at least two plants. The invention further relates to a system for data consolidation between at least two plants.Prior ArtIn today's digital era, the processing and / or analysis of data has taken a central role in almost all areas of life. Companies and organizations rely on efficient methods to collect, integrate, and harmonize data from various sources. In this case, the data alignment and / or the data consolidation represents a decisive phase in order to ensure that information can be linked together precisely and consistently in order to make synergies usable in this way. Up to now, this data consolidation process has largely been carried out manually or with the aid of software packages such as Informativea or Databricks.However, manual and software-based approaches to data alignment are associated with significant challenges. One of the difficulties is to reliably and correctly interpret the terms and definitions of the data sets. This not rarely leads to misunderstandings and errors in data harmonization and / or data consolidation. For example, different data sets could have similar designations such as "distance", but the actual units of measure or units may be significantly different from one another, such as kilometers (km) and kilometers per hour (km / h). In addition, notations and abbreviations of terms such as "D" for distance and "km" for kilometers may also vary.Moreover, manual interventions and the need to read and interpret extensive documentation are time-consuming and error-prone. Tap errors, inaccuracies and difficulties in dealing with capital and small writing contribute to an increased risk of data inconsistencies and errors. Against this background, it is indispensable to develop innovative solutions that optimize the data alignment process and / or data consolidation and allow harmonization and / or data consolidation of information that is as precise, efficient and / or low-error as possible.The object of the invention is thus to specify a correspondingly improved method and / or system.The object is achieved by a computer-implemented method for data consolidation between at least two factories according to the features of claim 1.Disclosure of the InventionIn a first aspect, a computer-implemented method for data consolidation between at least two factories is specified, each factory providing at least one data record of at least one machine and / or a workpiece which in particular operate on identical or similar production lines, the method comprising the following steps: a) providing a trained transformer network which is based on natural speech processing in order to carry out a semantic analysis of the provided data records as a function of at least one keyword and / or label; b) carrying out a data comparison between the provided data records as a function of the at least one keyword in order to identify data points which coincide in terms of content and / or data context or have significant similarities; c) establishing a consolidated data record which has the identified matching or similar data points from the at least two factories; d) providing the consolidated data set for the use and / or optimization and / or standardization and / or realization transmission of at least one function and / or method relating to the at least one machine in order to enable the transfer of realizations and methods between the factories.The respective data sets can have, for example, measured values with respect to at least one measured variable with respect to at least one machine of the respective factory and / or with respect to a workpiece. The data sets can basically have numerical and / or textual and / or pictorial information. The data sets are preferably recorded in a production line which is the same or similar, for example in an engine production or in a specific production process or production step. The data sets may include historical information and / or real-time information. The data sets can have measured values recorded by sensors.The keyword can indicate, for example, a label of the respective data record and preferably have a type and / or a class and / or a group and / or a size and / or a physical unit. The keyword can be used to classify the respective data record preferably unambiguously belonging to a category. One keyword or more keywords are important for the data alignment. These keywords play a role in identifying relevant information in the records.The transformer network is a type of artificial intelligence capable of performing complex natural language processing tasks. It was previously trained to understand the query and perform the data alignment.The present method preferably uses a large language model in the form of the transformer network as a data recovery tool. The method solves several problems, the companies that have several factories at different locations. In this case, a trained transformer network, for example with a ChatGPT architecture or the like, is used to solve the problem of data alignment and / or data consolidation between different data sources from different factories.It is often the case, namely, that machines of the same or similar production line collect and / or provide data or data sets in different factories, which data or data sets are to be consolidated and / or matched as much as possible for the transfer of findings and / or methods between the factories. Due to partly different time zones and / or production conditions and / or machine users in the factories in which the respective machines to be matched with each other are installed, the provided data sets are often marked either as being the same, slightly different or significantly different from the data sets of the at least one machine of the other factory. In order to make reasonable use of such data sets, it is preferred to match the same data from different factories with those of the other factories. This then enables the transfer of findings and / or methods from one factory to another or from one machine to the other.In order to simplify the process of data consolidation and in particular no longer have to be carried out manually, large speech models in the form of transformer networks are suitable. Namely, the architecture based on pre-trained transformer networks provides a human-like chat function that is particularly helpful in data consolidation.The proposed method can preferably interpret a large number of data sets without a re-training being required.In principle, however, the transformer network can also be trained overnight or online in order to improve a transformer network performance, in particular continuously. The present method also has the advantage of saving time and labor over manual data consolidation.In one embodiment, performing a data alignment between the provided data sets in dependence on the at least one keyword comprises: inputting a text-based query comprising the at least one keyword, by which a data alignment task is indicated between the provided data sets, into the trained transformer network; and solving the data alignment task comprising the at least one keyword by the transformer network by identifying the most likely match or similarity of the data points.Performing a data alignment refers to performing a process in which data is compared and matched to determine matches or similarities between the data sets. The data sets provided are preferably those available for the data alignment. The inputting and / or providing of a text-based query in text-based form, which contains the at least one keyword, is used to indicate to the transformer network which type of data alignment task is to be carried out. The transformer network uses the text-based query and identifies the most likely match or similarity between the data points in the provided records. The transformer network uses its trained capabilities to determine the best matching matches or similarities between the data points in the data sets. This is done based on the keywords specified in the text-based query and other relevant information.It is therefore preferably proposed to input an in particular standardized, automated query into the pre-trained transformer network or to provide it to it. For example, if a label (keyword) from one record is to be matched with another, such a query may be: "Provide the best match for the label 'weight'. Here, the list of all designations from the other record 'list' is given. Return only the name of the label with the highest match." The Transformer Network preferably returns the keyword, which holds that for the most likely match. In addition to the list of designations, for example, additional information such as a data description document may be added.In one embodiment, the transformer network comprises a large language model operable in an online mode or an offline mode. In particular, there are extensions of speech models in which the models can access the Internet and the domain knowledge contained therein by, for example, essentially dividing the request of a user into sub-tasks and / or consulting the Internet for solving at least one of these sub-tasks (see, for example, Auto-GPT). The transformer network has a large speech model. This means that it contains a large amount of linguistic information and data to effectively understand and process natural language. A large language model is capable of recognizing complex language patterns and contexts. "online mode" means that the transformer network can operate in online mode. This means that it operates in real time and is able to process requests continuously as they arrive. In online mode, the network may respond to requests and provide immediate results. The transformer network has access to information from the Internet. "Offline mode" means that the transformer network can also be operated in offline mode. In offline mode, the network will preferably operate in a batch processing mode where it processes a lot of data at a time without responding to real-time requests. This may be useful when a large amount of data needs to be processed in a single pass without delays through real-time processing. The transformer network preferably has no Internet access and is operable as a stand-alone solution. This can be advantageous in the case of safety-critical data.In one embodiment, creating a consolidated dataset including the identified matching or similar data points from the at least two factories includes: data preprocessing, in particular filtering and / or recalculating and / or transforming and / or matching and / or unifying, the identified data points by the transformer network.The method can be further developed to classify two data sets as being similar to one another on the basis of matching keywords and / or a correspondence based thereon, and then to check whether both data sets or the data points included therein use the same units. If this is not the case, the transformer network can preferably search for a conversion formula or transformation in a database, such as Excel, or on the Internet, and process a data record using this formula.In one embodiment, creating a consolidated dataset including the identified matching or similar data points from the at least two factories further includes: creating a report about the data preprocessing by the transformer network. Generating a report is advantageous because it makes transparent what has been matched and calculated.In general, it has been described how a label of one record is matched to the proper label of another record. This comparison is preferably carried out for all designations of a first data set, in particular in a computing loop. Alternatively, a query can also be changed to the effect that they are, for example: "Order all the designations of data record A to the designations of data record B".In one embodiment, the method is executable on a computer or on a cloud or on a server.In one embodiment, the respective data records also have documentation and / or metafiles and / or meta information.According to a second aspect, a system for data consolidation between at least two factories is specified, each factory providing at least one data record of at least one machine and / or a workpiece which in particular operate on identical or similar production lines, the system having an evaluation and / or computing device which is designed to carry out at least the following steps: a) providing a trained transformer network which is based on natural speech processing in order to carry out a semantic analysis of the provided data records as a function of at least one keyword and / or label; b) carrying out a data comparison between the provided data records as a function of the at least one keyword in order to identify data points which match or have significant similarities in terms of content and / or data context; c) creating a consolidated dataset having the identified matching or similar data points from the at least two factories; d) providing the consolidated dataset for use and / or optimization and / or standardization and / or discovery transfer of at least one function and / or method relating to the at least one machine to enable transfer of findings and methods between the factories.The system may be a computer or cloud or server.Particularly preferably, a pipeline or a user interface is provided in which the user can designate and / or enter the data sets and possibly supplementary documentation (e.g. label description or some csv files with similar proposals), which are to be matched and / or consolidated according to the invention. The alignment or consolidated data record is preferably returned to the user via the user interface in any desired format. In the simplest form in an Excel file. Such a return could also be used again to be automatically introduced into tools such as Informativea or Databricks, since these tools are explicitly made for the storage of this type of label relationships.It goes without saying that the steps according to the invention and further optional steps do not necessarily have to be carried out in the sequence shown, but can also be carried out in a different sequence. Further intermediate steps can also be provided. The individual steps can also comprise one or more substeps without thereby departing from the scope of the method according to the invention.According to the invention, a computer program with program code is also claimed in order to execute at least parts of the method according to the invention in one of its embodiments when the computer program is executed on a computer. In other words, according to the invention, a computer program (product) comprising instructions which, when the program is executed by a computer, cause the computer to execute the method / steps of the method according to the invention in one of its embodiments.According to the invention, a computer-readable data carrier with program code of a computer program is also proposed in order to execute at least parts of the method according to the invention in one of its embodiments when the computer program is executed on a computer. In other words, the invention relates to a computer-readable (storage) medium comprising instructions which, when executed by a computer, cause the computer to carry out the method / steps of the method according to the invention in one of its embodiments.The described embodiments and developments can be combined with one another as desired.Further possible embodiments, developments and implementations of the invention also include combinations of features of the invention described above or below with respect to the exemplary embodiments, which combinations are not explicitly mentioned.Brief Description of the DrawingsThe accompanying drawings are intended to provide a further understanding of the embodiments of the invention. They illustrate embodiments and, in conjunction with the description, serve to explain principles and concepts of the invention.Other embodiments and many of the advantages mentioned are evident with reference to the drawings. The elements of the drawings shown are not necessarily shown true to scale with respect to one another.The following are shown: FIG. 1 shows a schematic flow diagram of an exemplary embodiment of the present method.In the figures of the drawings, identical reference numerals designate identical or functionally identical elements, components or components, unless indicated to the contrary.FIG. 1 shows a schematic flow diagram of a computer-implemented method for data consolidation between at least two factories, each factory providing at least one dataset of at least one machine in each case.In any given embodiment, the method can be carried out at least partially by a system 1, which can include a plurality of components, not shown in detail, for example one or a plurality of provision devices and / or at least one evaluation and computing device. It is understood that the provision device can be formed jointly with the evaluation and computing device or can be different therefrom. Furthermore, the system can comprise a storage device and / or an output device and / or a display device and / or an input device.According to the invention, the computer-implemented method comprises at least the following steps:In a step S 1, a trained transformer network is provided which is based on natural speech processing in order to perform a semantic analysis of the provided data records depending on at least one keyword and / or label.In a step S 2, a data comparison is carried out between the provided data sets depending on the at least one keyword in order to identify data points that coincide in terms of content and / or data context or have significant similarities.In a step S 3, a consolidated data set is created which has the identified matching or similar data points from the at least two factories.In a step S 4, the consolidated data set is provided for the use and / or optimization and / or standardization and / or realization transmission of at least one function and / or method relating to the at least one machine.

Claims

Computer-implemented method for data consolidation between at least two factories, each factory providing at least one data record of at least one machine and / or a workpiece, the method comprising the following steps: a) providing (S1) a trained transformer network which is based on natural speech processing in order to perform a semantic analysis of the provided data records depending on at least one keyword and / or label; b) performing (S2) a data alignment between the provided data records depending on the at least one keyword in order to identify data points which match in content and / or data context relationship or have significant similarities; c) creating (S3) a consolidated data record which has the identified matching or similar data points from the at least two factories; d) Provision (S 4) of the consolidated data set for the use and / or optimization and / or standardization and / or realization transmission of at least one function and / or method relating to the at least one machine.Method according to claim 1, wherein performing (S2) a data alignment between the provided data sets in dependence on the at least one keyword comprises: - inputting a text-based query comprising the at least one keyword, by which a data alignment task is indicated between the provided data sets, into the trained transformer network; and - solving the data alignment task comprising the at least one keyword by the transformer network by identifying the most likely match or similarity of the data points.The method of claim 1 or 2, wherein the transformer network comprises a large language model operable in an online mode or an offline mode.Method according to one of the preceding claims, wherein the creation (S3) of a consolidated data set comprising the identified matching or similar data points from the at least two factories: - data preprocessing, in particular filtering and / or recalculating and / or transforming and / or matching and / or unifying, the identified data points by the transformer network.The method of claim 4, wherein creating (S3) a consolidated dataset comprising the identified matching or similar data points from the at least two factories further comprises: - creating a report about the data preprocessing by the transformer network.The method of any preceding claim, wherein the method is executable on a computer or on a cloud or on a server.Method according to one of the preceding claims, wherein the respective data records furthermore have documentation and / or metafiles and / or meta information.A system (1) for data consolidation between at least two factories, each factory providing at least one data set of at least one machine and / or a workpiece, the system (1) comprising an evaluation and / or computing device which is configured to carry out at least the following steps: a) providing a trained transformer network which is based on natural speech processing in order to carry out a semantic analysis of the provided data sets in dependence on at least one keyword and / or label; b) carrying out a data comparison between the provided data sets in dependence on the at least one keyword in order to identify data points which coincide in terms of content and / or data context or have significant similarities; c) creating a consolidated data set which has the identified matching or similar data points from the at least two factories; d) Providing the consolidated data set for the use and / or optimization and / or standardization and / or realization transmission of at least one function and / or method relating to the at least one machine.Computer program with program code for executing at least parts of a method according to one of Claims 1 to 7 when the computer program is executed on a computer.Computer-readable data medium with program code of a computer program for executing at least parts of a method according to one of Claims 1 to 7 when the computer program is executed on a computer.

Images