Method for determining a component parts list for manufacturing a product by means of an electronic computing device, computer program product, computer-readable storage medium and electronic computing device

Abstract

The invention relates to a method for generating a component parts list (10) for manufacturing a product (12) by means of an electronic computing device (14), comprising the steps: - Providing a variant parts list (16) of the product (12) by means of the electronic computing device (14); - detecting at least one input (30) of a provided product property (24) of the product (12) by means of an input device (26) of the electronic computing device (14); - Optimizing the variant parts list (16) depending on the at least one product property (24) by means of an optimization algorithm (28) of the electronic computing device (14); and - Generating the component parts list (10) for the product (12) as a function of the optimized variant parts list (16) by means of the electronic computing device (14). Furthermore, the invention relates to a computer program product, a computer-readable storage medium and an electronic computing device (14).

Description

[0001] The invention relates to a method for determining a component parts list for manufacturing a product by means of an electronic computing device according to the applicable patent claim 1. Furthermore, the invention relates to a computer program product, a computer-readable storage medium and an electronic computing device.

[0002] The following invention specifically addresses the problem of efficient and optimized requirements elicitation in the context of technical systems, with a particular focus on an optimal sequence for eliciting individual requirements. Particularly in complex technical systems characterized by a multitude of variable parameters, requirements elicitation can be a major challenge. A rigid, list-like or disordered, and thus suboptimal, sequence of requirements elicitation can lead to inefficiencies, errors, and increased manual effort. Sequential queries or queries that attempt to limit the maximum amount of unknown information at each step are suboptimal and do not achieve a global optimum with regard to minimizing the number of "questions" posed in the entire "question chain."

[0003] Previous requirements elicitation solutions have often relied on manual processes and expert knowledge to determine the order of requirements or tended to perform the order of requirements queries in a fixed sequence. These approaches can be inefficient and error-prone, especially in systems with a large number of variables and parameters. Furthermore, these solutions often failed to adequately consider the complexity and interdependencies between different requirements.

[0004] The object of the present invention is to provide a method, a computer program product, a computer-readable storage medium and an electronic computing device by means of which a component parts list for a product can be generated with reduced effort.

[0005] This object is achieved by a method, a computer program product, a computer-readable storage medium, and an electronic computing device according to the independent patent claims. Advantageous embodiments are specified in the subclaims.

[0006] One aspect of the invention relates to a method for generating a component parts list for manufacturing a product using an electronic computing device. A variant parts list of the product is provided using the electronic computing device. At least one input of a provided product property of the product is recorded using an input device of the electronic computing device. The variant parts list is optimized depending on the at least one product property using an optimization algorithm of the electronic computing device. The component parts list for the product is generated using the electronic computing device depending on the optimized variant parts list.

[0007] In particular, the invention uses a variant parts list as input information to generate a component parts list depending on the corresponding product parameter or the intended product property. In other words, the variant parts list can comprise a plurality of products with different variants and, in particular, the individual components of the different variants are also listed. Based on a product property, a component parts list is then generated using the optimization method in such a way that the variant parts list is, in particular, incrementally reduced until only solutions remain that correspond to the at least one product property.

[0008] In particular, efficiency can be increased and manual effort reduced, particularly by utilizing existing data and delivering precise and comprehensive results. In particular, this results in the advantages over the prior art that improved efficiency can be realized, as well as optimized resource utilization. Furthermore, accurate and consistent requirements gathering and a potentially faster and more reliable process of product configuration and / or adaptation can be realized. Furthermore, the present invention offers valuable insights, particularly into potential areas for product innovation or development, when requirements cannot be covered by existing variants in the bill of materials.

[0009] According to an advantageous embodiment, the variant BOM is optimized using a Monte Carlo algorithm as the optimization algorithm. In particular, a Monte Carlo Tree Search (MCTS) algorithm can be provided to determine optimal query paths. This algorithm goes beyond the short-term maximization of information exclusion and considers the entire query process to achieve a global optimum. This is an important difference from most known methods, which focus on the immediate exclusion of the information space and may not achieve the most efficient overall query chain.

[0010] Furthermore, it can be planned that a configuration rule set for the variant BOM is taken into account during the optimization. The configuration rule set used in this method, in particular, enables even more targeted requirements elicitation. While other solutions often generate independent requirements, this method is able to identify relationships between requirements for different components / assets and consider these relationships when determining the sequence and type of requirement queries. In particular, a product configuration rule set is used to further optimize the feature queries. Information about variants and their interdependencies is taken into account, which then influences the sequence and type of requirement queries.It is stipulated that statements about variants that exclude each other and variants that require the application of another variant are included in the configuration rule set. It is emphasized that these relationships do not necessarily have to be reciprocal. For example, variant A can require the application of variant B without the application of variant B requiring variant A. The variant BOM thus functions as an essential input parameter, and the automated requirements query also functions without the configuration rule set. However, the integration of such a rule set can enable even more targeted requirements elicitation. Relationships between requirements for different components can thus be derived, which then influence the sequence and type of requirement queries.

[0011] A further advantageous embodiment provides that at least one feature relationship of components within the variant BOM is taken into account during optimization. The feature relationships can, in particular, involve corresponding component dependencies. For example, if a corresponding component is used for the product, it is imperative that another component is also used accordingly. Furthermore, components can also be mutually exclusive, for example, so that these feature relationships can also be used accordingly. This allows the component BOM to be advantageously generated.

[0012] It is also advantageous if at least a bundling of components within the variant BOM is taken into account during optimization. This bundling is particularly a so-called feature clustering. Feature clustering is used to reduce requirement queries. Component characteristics that can be meaningfully queried simultaneously are bundled. This creates a clustering according to characteristics in combination with their component relationships. This bundling reduces the number of requirement queries by grouping characteristics that can meaningfully be queried together. For example, the "maximum operating temperature" could be viewed as a generic characteristic that is relevant for many components. A single query for the "maximum ambient temperature at the customer's facility location" as an exemplary product characteristic could provide information about all components that have this characteristic.However, it can also be provided that there are features that need to be queried alone, such as the pump performance of a specific pump. In particular, the invention implements a method for feature bundling that groups components according to common features in order to efficiently query requirements. Other prior art methods often lack such a structured and systematic approach to feature bundling and can therefore be inefficient or carry the risk of inconsistent requirements elicitation.

[0013] Furthermore, it has proven advantageous to specify an upper confidence bound (UCB) for the optimization process. Especially in the Monte Carlo method, the upper confidence bound (UCB) plays a central role in deciding which nodes to explore. A key feature of this upper confidence bound is its ability to quickly differentiate between node exploration after just a few simulations. Nodes that appear less promising are given lower priority after multiple runs, for example, because a corresponding value of R(F) is lower in these cases. On the other hand, promising nodes are highlighted by a higher average value of R(F) and are given priority in subsequent simulations.The sophisticated action of the different components for the upper confidence interval leads to a self-regulating strategy where, after a sufficient number of simulations, a strong focus is placed on the most promising question nodes, while less promising paths are given less consideration.

[0014] A further advantageous embodiment provides that the upper confidence interval is calculated using the formula: UCB(F)=R(F)+C∗ln nN where R(F) represents an average success of a simulation starting from an initial node F, C represents an exploration bonus, n represents a total number of simulations executed by a parent node, and N represents the number of simulations for the specific node.

[0015] It is also advantageous if an exploration bonus of 1.4 is specified. In particular, multiplying the expression by the exploration bonus C allows for fine-tuning of this exploration bonus. A higher value for C would more strongly encourage the exploration of new paths, while a lower value of C would more heavily weight the exploitation of known paths. The importance of the value for C, for example, was established through a series of careful experiments and trials in which different values ​​for C were tested in several simulated scenarios. By systematically varying C and subsequently observing the effects on the algorithm's results, the value of C = 1.4 in particular can be identified as a generically applicable and essentially optimal balance between exploration and exploitation. This value has been determined to be the most performant in a wide variety of applications.However, it is further suggested that fine-tuning the value to the specific variant BOM and the requirements of the respective context may lead to even more optimal results. This may require additional experimentation and investigation or the use of domain-specific knowledge to precisely tune the value of C to the individual BOM and the specific use case.

[0016] It has also proven advantageous to generate a control signal for a functional unit based on the generated component parts list. For example, the generated component parts list can be used to automatically order components. Furthermore, the generated component parts list can be used, for example, to control a corresponding manufacturing machine and thus actually manufacture the product.

[0017] It is also advantageous if the variant bill of materials is optimized based on a large number of intended product properties. In particular, multiple product properties can be specified. Based on the large number of intended product properties, a component bill of materials can then be reliably generated, which in turn corresponds to the product properties. It can also be provided, for example, that if the product properties do not lead to a suitable solution, these properties can be adjusted accordingly.

[0018] It is also advantageous if the optimization algorithm determines an optimization score for a potential component BOM. This optimization score can also be referred to as a score. In particular, each potential component BOM that matches the product characteristics can be evaluated. This can, for example, display which component BOM has the highest score and thus, for example, suggest it to a user.

[0019] It has also proven advantageous to select the component BOM from the potential component BOMs that has the highest optimization score. This can be done automatically. In other words, the optimization score can be determined from multiple component BOMs. The highest optimization score can then be selected automatically, and, for example, corresponding component orders and machine control can be generated based on the component BOM with the highest optimization score.

[0020] The method presented is, in particular, a computer-implemented method. Therefore, a further aspect of the invention relates to a computer program product with program code means that, when the program code means are processed by the electronic computing device, cause an electronic computing device to perform a method according to the preceding aspect.

[0021] Furthermore, the invention therefore also relates to a computer-readable storage medium with at least one computer program product according to the preceding aspect.

[0022] Yet another aspect of the invention relates to an electronic computing device for determining a component parts list for manufacturing a product, comprising at least one input device, wherein the electronic computing device is configured to perform a method according to the preceding aspect. In particular, the method is performed by means of the electronic computing device.

[0023] Advantageous embodiments of the method are to be regarded as advantageous embodiments of the computer program product, the computer-readable storage medium, and the electronic computing device. The electronic computing device, in particular, has physical features enabling the execution of corresponding method steps.

[0024] A computing unit / electronic computing device can be understood, in particular, as a data processing device that contains a processing circuit. The computing unit can therefore, in particular, process data to perform computing operations. This may also include operations for performing indexed access to a data structure, for example, a look-up table (LUT).

[0025] The computing unit can in particular contain one or more computers, one or more microcontrollers and / or one or more integrated circuits, for example one or more application-specific integrated circuits (ASICs), one or more field-programmable gate arrays (FPGAs), and / or one or more single-chip systems (SoCs). The computing unit can also contain one or more processors, for example one or more microprocessors, one or more central processing units (CPUs), one or more graphics processing units (GPUs), and / or one or more signal processors, in particular one or more digital signal processors (DSPs). The computing unit can also contain a physical or virtual network of computers or other of the aforementioned units.

[0026] In various embodiments, the computing unit includes one or more hardware and / or software interfaces and / or one or more memory units.

[0027] A memory unit can be a volatile data memory, for example a dynamic random access memory (DRAM) or a static random access memory (SRAM), or a non-volatile data memory, for example a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory or flash EEPROM, a ferroelectric random access memory (FRAM), a magnetoresistive random access memory,MRAM (magnetoresistive random access memory) or phase-change random access memory, PCR_AM (phase-change random access memory).

[0028] For use cases or application situations that may arise during the method and which are not explicitly described here, it may be provided that, in accordance with the method, an error message and / or a request to enter user feedback is issued and / or a default setting and / or a predetermined initial state is set.

[0029] Regardless of the grammatical gender of a particular term, persons with male, female or other gender identity are included.

[0030] Further features and combinations of features of the invention will become apparent from the figures and their description, as well as from the claims. In particular, further embodiments of the invention do not necessarily have to contain all features of one of the claims. Further embodiments of the invention may have features or combinations of features that are not mentioned in the claims.

[0031] Showing: Fig. 1 a schematic flow diagram according to an embodiment of the method; and Fig. 2 another schematic flow diagram of the process.

[0032] The invention is explained in more detail below with reference to specific embodiments and associated schematic drawings. In the figures, identical or functionally equivalent elements may be provided with the same reference numerals. The description of identical or functionally equivalent elements may not necessarily be repeated for different figures.

[0033] Fig. Figure 1 shows a schematic flow diagram according to an embodiment of the method. In particular, Fig. 1 shows a method for generating a component parts list 10 for manufacturing a product 12 using an electronic computing device 14. A variant parts list 16 is provided. The variant parts list 16 can, for example, describe a configuration rule set 18 as well as a feature bundling 20 and a feature relationship 22. Alternatively, only the variant parts list 16 can be used. At least one input 30 of a designated product property 24 of the product 12 is recorded using an input device 26 of the electronic computing device 14. The variant parts list 16 is optimized as a function of the at least one product property 24 using an optimization algorithm 28 of the electronic computing device 14. The component parts list 10 for the product 12 is then generated as a function of the optimized variant parts list 16 using the electronic computing device 14.

[0034] In particular, it can be provided that the variant parts list 16 is optimized by means of a Monte Carlo algorithm as the optimization algorithm 28.

[0035] Furthermore, it can be provided, in particular, that an upper confidence interval for the optimization is specified within the optimization. In particular, the upper confidence interval can be determined using the formula: UCB(F)=R(F)+C∗ln nN where R(F) represents the average success of a simulation starting from an initial node F, C represents an exploration bonus, n represents the total number of simulations executed by a parent node, and N represents the number of simulations for the specific node. In particular, it can be provided that, for example, the exploration bonus C is specified as 1.4.

[0036] Fig.2 shows a schematic flow diagram according to one embodiment of the method. In particular, a Monte Carlo algorithm is shown. For example, a corresponding question or product property 24 can be provided. In a first subdivision step, the present exemplary embodiment shows that, for example, two potential solutions and two non-potential solutions are present. In the second substep, it is shown that the solutions that do not offer any solutions are no longer pursued. This is indicated in particular by an X. The solutions that are pursued further are indicated in particular by a corresponding check mark.

[0037] In the second layer, it can then be determined, for example, that additional components are required for this purpose. Additional components can then also be used for this purpose. For example, it is shown that the upper decision path offers a solution. Furthermore, it is shown that a corresponding optimization score of 32 can be assigned. In the present exemplary embodiment, for example, the upper optimization score of 32 can achieve a higher score than the lower one, since, in particular, fewer components are required to arrive at a solution.

[0038] The invention thus solves the problem identified by a series of technical features based on the principles of the variant parts list 16, feature clustering and the use of an algorithm for the optimized sequence of requirement queries.

[0039] The variant BOM 16, together with the associated configuration rules 18, is the sole input for this method. A variant BOM 16 is a type of BOM that maps different variants of a product 12 in a single, hierarchically structured BOM. It is a consolidated list containing all technical characteristics of the components / assets used in the context of the BOM. This includes characteristics that refer to specific positions in the BOM as well as asset-specific characteristics. It is a compilation of all technical variants of a product 12 or a system that can be selected from a predefined technical spectrum.

[0040] Feature clustering / feature bundling 22 is used to reduce requirement queries. Asset characteristics that can be meaningfully queried simultaneously are bundled. This results in clustering by characteristics in combination with their asset relationships. This bundling reduces the number of requirement queries by grouping characteristics that can meaningfully be queried together. For example, the "maximum operating temperature" could be viewed as a generic characteristic that is relevant for many assets and, through a single query for the "maximum ambient temperature at the customer's installation site," provides information about all assets that have this characteristic. However, there may also be characteristics that must be queried alone, such as the pump performance of a specific pump.

[0041] The configuration rule set 18 of the plant / product 12 is used to further optimize the feature queries. Information about variants and their interdependencies is taken into account, which then influences the sequence and type of requirement queries. It is specified that statements about variants that exclude each other and variants that require the application of another variant are contained in the configuration rule set 18. It is emphasized that these relationships do not necessarily have to be reciprocal: A variant A can require the application of a variant B without the application of B requiring variant A. The variant bill of materials 16 acts as an essential input, and the automated requirements elicitation functions even without the configuration rule set 18. However, the integration of such a rule set can enable even more targeted requirements elicitation.Relationships between requirements for different assets can be derived, which then influence the sequence and type of requirement queries.

[0042] All preparatory steps have now been completed to determine the optimal query sequence for requirements queries. This phase is based on the use of the Monte Carlo Tree Search (MCTS) algorithm. The MCTS algorithm performs an exploration phase in which it simulates various chains of requirements queries. Each query represents a bundle of features that can be queried together. In these simulations, the simulated answers to the questions are based on the value spectrum of the feature in the variant BOM 16. If there is no variance of the feature, but only a specific value, the response simulations are divided into the options that this value matches the requirements or that it does not match them.

[0043] Random combinations of questions and answers are simulated. After each simulated answer, the variant BOM 16 is analyzed to determine any restrictions on the variant selection that are determined by this answer. For example, if the maximum outside temperature is simulated with 40°C as the answer, all assets that do not meet this answer, i.e., cannot withstand an outside temperature of 40°C, are removed from the BOM and no longer considered in the next question-answer simulation.

[0044] The requirements elicitation therefore continuously refines the variant BOM 16. Based on the responses received, the selection of technical variants is gradually narrowed down by excluding inconsistent variants. If, for example, an answer excludes two variants, the characteristics of these variants are no longer queried in the follow-up questions because they are no longer relevant.

[0045] Since it is almost impossible to simulate all possible question-answer chains due to the enormous number of combinations, random samples are simulated. The success of the simulated question-answer chains, measured by the reduction of the unknown information space, is then "backpropagated" back to the initial question node (backpropagation). Success is measured by the number of reduced feature clusters after complete simulation of the simulation depth n. The more simulations performed, the more accurately the probability of success of a question can be estimated, a principle known as the law of large numbers.

[0046] Within the MCTS algorithm, the Upper Confidence Bound (UCB) formula plays a central role in deciding which question nodes to explore. The expression: UCB(F)=R(F)+C∗ln nN represents the basis, where R(F) stands for the average success of the simulations starting from the initial question node F.

[0047] The specific part of the UCB equation known as C∗ln nN has a special significance in the context of decision making within the MCTS algorithm.

[0048] In this equation, C is a fixed value that controls the balance between exploring new paths and exploiting known, successful paths. The value n represents the total number of simulations run by the parent node, and N represents the number of simulations for the specific node under consideration.

[0049] The entire form of expression C∗ln nN serves to incorporate uncertainty about a particular node into the decision about which nodes to explore next. The expression encourages exploration of nodes that have been visited less frequently by adding a bonus to nodes that have been simulated less often.

[0050] The term Inn takes into account how often the parent node has been visited. As the parent node has been visited more frequently, the value of In n increases, resulting in an increase in the exploration bonus.

[0051] Dividing by N ensures that nodes that have themselves been visited less frequently receive a higher bonus to encourage exploration of these less explored nodes.

[0052] Multiplying this expression by C allows for fine-tuning of this exploration bonus. A higher value of C would more strongly encourage the exploration of new paths, while a lower value of C would more heavily weight the exploitation of known paths.

[0053] The significance of the value for C was established through a series of careful experimental investigations in which different values ​​for C were tested in numerous simulated scenarios. By systematically varying C and subsequently observing the effects on the algorithm's results, the value C=1.4 was identified as a generically applicable and optimal balance between exploration and exploitation. This value was determined to be the most performant in a variety of applications. However, it is envisaged that fine-tuning the value to the specific variant BOM 16 and the requirements of the respective context may lead to even more optimal results. This may require additional experimental investigations or the use of domain-specific knowledge to precisely tune the value of C to the individual BOM and the specific use case.

[0054] A core feature of the UCB formula is its ability to quickly differentiate node exploration after a few simulations. Nodes that appear less promising receive lower priority after multiple runs because the value of R(F) is lower in these cases. On the other hand, promising nodes are highlighted by a higher average value of R(F) and are favored in subsequent simulations. The sophisticated interaction of the various components of the formula leads to a self-regulating strategy in which, after a sufficient number of simulations, a strong focus is placed on the most promising query nodes, while less promising paths are given less consideration.

[0055] This designed interplay of factors allows the algorithm to balance both the immediate and long-term benefits of a question within the context of the entire question series, thus intelligently optimizing the path through the query requirements. This process optimizes the order of the query requirements, making the entire query sequence more efficient and effective.

[0056] The identified question is finally asked to the respondent, and the respondent provides a real answer. Based on this, the variant BOM 16 is restricted again, and the MCTS simulation starts again to identify the next question with the highest probability of achieving a global optimum in terms of reducing the unknown information across the next questions.

[0057] This process is repeated until all unknown information has been identified. The final result is a reduced bill of materials (BOM) that contains only the technical variants that meet the respondent's requirements, i.e., the specified product feature. It may also be the case that the reduced BOM can no longer represent the product, and new variants may need to be created to meet customer requirements. This could mean developing new technical solutions or modifying existing solutions to meet specific requirements. In such situations, the method also serves as a tool for identifying areas where further product development or innovation is required.

[0058] This combination of feature bundling, application of configuration knowledge and an optimized sequence of requirement queries based on the MCTS algorithm enables very efficient and targeted requirements elicitation. List of reference symbols 10 Component parts list 12 Product 14 electronic computing device 16 Variant parts list 18 Configuration rules 20 Feature relationship 22 Feature bundling 24 Input 24 intended product feature 26 Input device 28 Optimization algorithm 30 Input 32 optimization score

Claims

[1] Method for generating a component parts list (10) for manufacturing a product (12) by means of an electronic computing device (14), comprising the steps: - Providing a variant parts list (16) of the product (12) by means of the electronic computing device (14); - detecting at least one input (30) of a provided product property (24) of the product (12) by means of an input device (26) of the electronic computing device (14); - Optimizing the variant parts list (16) depending on the at least one product property (24) by means of an optimization algorithm (28) of the electronic computing device (14); and - Generating the component parts list (10) for the product (12) as a function of the optimized variant parts list (16) by means of the electronic computing device (14). [2] Method according to claim 1, characterized bythat the variant parts list (16) is optimized using a Monte Carlo algorithm as the optimization algorithm (28). [3] Method according to claim 1 or 2, characterized by that a configuration rule set (18) for the variant parts list (16) is taken into account during optimization. [4] Method according to one of the preceding claims, characterized by that at least one feature relationship (20) of components within the variant parts list (16) is taken into account during optimization. [5] Method according to one of the preceding claims, characterized by that at least one bundling (22) of components within the variant parts list (16) is taken into account during the optimization. [6] Method according to one of the preceding claims, characterized by that an upper confidence interval for the optimization is specified within the optimization. [7] Method according to claim 6, characterized bythat the upper confidence interval is calculated using the formula: UCB(F)=R(F)+C∗ln nN where R(F) represents an average success of a simulation starting from an initial node F, C represents an exploration bonus, n represents a total number of simulations executed by a parent node, and N represents the number of simulations for the specific node. [8] Method according to claim 7, characterized by that an exploration bonus of 1.4 is given. [9] Method according to one of the preceding claims, characterized by that a control signal for a functional unit is generated on the basis of the generated component parts list (10). [10] Method according to one of the preceding claims, characterized by that the variant parts list (16) is optimized on the basis of a large number of intended product properties (24). [11] Method according to one of the preceding claims, characterized by that the optimization algorithm (28) determines an optimization score (32) for a potential component parts list (10). [12] Method according to claim 11, characterized by that the component parts list (10) is selected from the potential component parts lists (10) which has the highest optimization score (32). [13] Computer program product with program code means which cause an electronic computing device (14) to carry out a method according to one of claims 1 to 12 when the program code means are processed by the electronic computing device (14). [14] A computer-readable storage medium comprising at least one computer program product according to claim 13. [15] Electronic computing device (14) for determining a component parts list (10) for producing a product (12), with at least one input device (26), wherein the electronic computing device (14) is designed to carry out a method according to one of claims 1 to 12.

Images