Method and system for the automated design of an electromechanical component

Abstract

The invention relates to a method for the automated construction and / or design of at least one electromechanical component (10) or a mechanical component (15), in particular for an assembly or an object, comprising the following steps - Creating (S10) a training data set (420) wherein the training data set (420) contains images and / or drawings and / or sensor signals and / or interface requirements and / or assembly accesses and / or other data of electromechanical components (10) and / or mechanical components (15); - Training (S20) an AI model (350) with the training dataset (420); - Input (S30) first input data (210) for a design task and passing the input data (210) to the trained AI model (350); - Generating (S40) a design draft (370) from the AI ​​model (350); - Input (S50) of requirements in the form of second input data (220) to optimize and improve the design (370); - Repeat (S60) the process steps (S30) and (S40) in one iteration until a target design (375) is obtained; - Output (S70) of the target design (375) and / or store the target design (375) in a database.

Description

[0001] The invention relates to a method for the automated construction and / or design of at least one electromechanical component or a mechanical component, in particular for an assembly or an object.

[0002] In mechanical and automotive engineering, a large number of components are manufactured with mechanical, mechatronic, electrical, electronic, hydraulic, chemical, etc. These components are subject to constant, time-consuming and costly further development due to new model series, environmental regulations, etc.

[0003] An electromechanical component is a device or system that combines electrical and mechanical elements to perform a specific function or task. In these components, there is usually an interaction between electrical signals or energy and mechanical movement or force. Electromechanical components are found in various industries and applications, such as robotics, automation, consumer electronics, automotive systems, and industrial machinery. An example of an electromechanical component is an electric motor, which typically consists of a rotor and a stator, with rotary motion generated by the interaction of magnetic fields. Electric motors are used in a wide variety of applications, from small appliances to large industrial machinery.Another example is sensors, which often contain both electrical and mechanical components to detect and measure physical quantities such as temperature, pressure, position, and motion. A strain gauge, for example, measures mechanical strain by converting it into a corresponding change in electrical resistance.

[0004] The selection or design of a part or component for an object such as a vehicle or a production facility depends on selected boundary conditions and parameters. For a mechanical or mechatronic component, these include the relative speed between two moving components, the operating temperature, the service life, the component material, certain mechanical loads, etc. A component can be designed or constructed in such a way that different parameters such as dimensions, shape, material or electrical resistance can be varied while adhering to selected boundary conditions. The design of a component or the construction of a more complex unit is usually based on the specialist knowledge and experience of experts, such as engineers. CAD software programs such as CATIA are used for this purpose.CATIA (Computer-Aided Three-Dimensional Interactive Application) is a powerful and comprehensive computer-aided design (CAD) software developed by Dassault Systèmes. It is widely used in various industries, including aerospace, automotive, industrial design, and mechanical engineering. CAD programs like CATIA support engineers and designers in creating, analyzing, and visualizing complex 2D and 3D models, which can range from simple components to complex assemblies and entire systems. However, CATIA is a very feature-rich CAD software that requires appropriate training and expertise to utilize its full potential.

[0005] Although modern CAD programs like CATIA streamline and accelerate the design process by providing a range of features, the design process still involves a significant amount of work. This is particularly true for the integration of electrical components into a part based on an electrical schematic. While CATIA has dedicated modules for integrating electrical components, such as routing electrical cables, harnesses, and connectors, as well as positioning electrical components within a part or assembly, aspects such as heat generation, weight, and electromagnetic compatibility (EMC) have not yet been adequately addressed.

[0006] US 2018 / 0285517 A1 discloses a computer-aided design (CAD) model of a component to be manufactured, which is intended to create an optimized design of the component based on user specifications such as materials, constraints, stresses, loading conditions, and preferred manufacturing processes. Based on this information, an iterative optimization is performed, and upon completion of the optimization process, the optimized design is presented to the user in a CAD file.

[0007] US 2021 / 0216674 A1 discloses a system for automatically generating optimal design concepts from a set of system design requirements. Based on user-defined design requirements, an optimized geometry is generated by chaining iterative applications. The system uses the result of the initial design process as part of the design requirements for the next iteration.

[0008] The object underlying the invention is to create possibilities for an improved automated construction and / or design process of an electromechanical part or a mechanical component, which is characterized by high reliability, safety and accuracy and is easy to implement.

[0009] The present invention can improve the design of electromechanical parts and mechanical components by allowing the AI ​​model to generate a large number of design alternatives more quickly. This makes it possible to find completely new solutions and optimize designs and / or geometries according to various criteria such as installation space, contact media, location, manufacturing processes, joining techniques, mechanical / thermal / chemical / electrical loads, assembly access, performance, efficiency, weight, or costs. In particular, significant time and cost savings can be achieved, as conventional design processes often involve manual iterations and physical prototypes, which can be time-consuming and expensive. The AI ​​model can be used in particular for complex design tasks with complicated geometries, as it can operate in multi-dimensional design spaces.In addition, the AI ​​model can generate design drafts that meet a wide range of technical constraints, manufacturability requirements, and functional specifications. For example, constraints such as environmental compatibility or energy consumption can be better considered. The AI ​​model can thus help improve the performance and efficiency of electromechanical parts and mechanical components.

[0010] Furthermore, the AI ​​model can learn from existing designs and historical data, enabling continuous development. Furthermore, since the AI ​​model can be integrated into a common platform, it facilitates the exchange of design ideas, thus promoting design innovation and creativity. The AI ​​model of the invention is thus a powerful tool that complements and expands the capabilities of engineers, enabling them to create better designs more efficiently.

[0011] This object is achieved according to the invention with respect to a method by the features of patent claim 1, with respect to a system by the features of patent claim 10, and with respect to a computer program product by the features of patent claim 15. The further claims relate to preferred embodiments of the invention.

[0012] According to a first aspect, a method for the automated construction and / or design of at least one electromechanical part or mechanical component, in particular for an assembly or an object, is provided. The method comprises the following method steps: - Creating a training data set, wherein the training data set contains images and / or drawings and / or sensor signals and / or interface requirements and / or assembly access and / or further data of electromechanical components and / or mechanical components; - Training an AI model with the training dataset; - Entering initial input data for a design task and passing the input data to the trained AI model; - Generating a design draft from the AI ​​model; - Entering requirements in the form of second input data to optimize and improve the design; - Performing an iteration until a target design is obtained; - Outputting the target design and / or storing the target design in a database.

[0013] In a further development, the AI ​​model is intended to use neural networks and / or recurrent neural networks and / or convolutional neural networks and / or a reinforcement learning algorithm.

[0014] In an advantageous embodiment, the reinforcement learning algorithm is designed as a Markov decision method, temporal difference learning (TD learning), Q-learning, SARSA, Monte Carlo simulation or actor-critic.

[0015] In a further embodiment, it is provided that at least one parameter and / or a parameter value and / or a property of the part or component represents a dimension, a material, a shape, a characteristic value, a measured value, a safety factor, a thermal resistance, an electrical resistance, an inductance, a conductivity, a strength, a fracture behavior, a color, a density, a plastic deformation, a tensile, compressive, flexural strength, a material property, a surface property or a heat treatment.

[0016] In a further development, it is provided that the target design is transferred to a simulation module, whereby the simulation module comprises a software application that contains simulation tools such as finite element analysis (FEA) or computational fluid dynamics (CFD).

[0017] In particular, it is planned to integrate the target design into the training data set for further training of the AI ​​model.

[0018] Advantageously, the training data set is transformed by a transformation module into a transformed data set with a data format that can be processed by the AI ​​model.

[0019] A further development will include an AI agent that will review the design drafts generated by the AI ​​model and make suggestions for improvement.

[0020] In particular, the AI ​​agent includes artificial intelligence algorithms such as neural networks and reinforcement learning and is also trained with the training data set or the transformed data set in a training phase.

[0021] According to a second aspect, the invention provides a system for the automated construction and / or design of at least one electromechanical part or one mechanical component, in particular for an assembly or an object. The system comprises an input module, a design module, a database, and an output module. The database contains a training data set with images and / or drawings and / or sensor signals and / or interface requirements, and / or assembly accesses, and / or further data of electromechanical parts and / or mechanical components. The design module comprises an AI model that can be trained using the training data set. The input module is configured to forward input first input data and second input data for a design task to the trained AI model.The trained AI model is designed to iteratively generate a plurality of design drafts to create a target design draft. The output module is designed to output the target design draft or store it in the database.

[0022] In a further development, it is provided that the AI ​​model uses neural networks and / or recurrent neural networks and / or convolutional neural networks and / or a reinforcement learning algorithm, whereby the reinforcement learning algorithm is designed as a Markov decision method, temporal difference learning (TD learning), Q-learning, SARSA, Monte Carlo simulation or actor-critic.

[0023] In an advantageous embodiment, it is provided that the target design is transferred to a simulation module, wherein the simulation module comprises a software application that contains simulation tools such as finite element analysis (FEA) or computational fluid dynamics (CFD).

[0024] In a further embodiment, it is provided that the training data set is transformed by a transformation module into a transformed data format that can be processed by the AI ​​model, wherein the transformation module has at least one encoder and at least one decoder.

[0025] In a further development, an AI agent is planned that reviews the design drafts generated by the AI ​​model and makes suggestions for improvement. The AI ​​agent includes artificial intelligence algorithms such as neural networks and reinforcement learning and is also trained with the training data set or the transformed data set in a training phase.

[0026] According to a third aspect, the invention provides a computer program product comprising executable program code configured to carry out the method according to the first aspect when executed.

[0027] The invention is explained in more detail below with reference to embodiments shown in the drawing.

[0028] It shows: Fig. 1 is a block diagram illustrating an embodiment of a system according to the invention; Fig. 2 a flow chart to explain the individual method steps of a method according to the invention; Fig. 3 is a block diagram of a computer program product according to an embodiment of the third aspect of the invention.

[0029] Additional features, aspects and advantages of the invention or embodiments thereof will become apparent from the detailed description taken in conjunction with the claims.

[0030] The Fig. The system 100 according to the invention shown in Figure 1 comprises an input module 200, a design module 300, a database 400 and an output module 500. In addition, a simulation module 700 can be provided.

[0031] The input module 200, the design module 300, the database 400, the output module 500, and the simulation module 700 can be implemented as standalone computing units or as a cloud-based solution. In particular, they can each be equipped with a processor and / or a memory unit.

[0032] In the context of the invention, a “processor” can be understood to mean, for example, a machine or an electronic circuit. A processor can in particular be a main processor (Central Processing Unit, CPU), a microprocessor or a microcontroller, e.g. an application-specific integrated circuit or a digital signal processor, optionally in combination with a memory unit for storing program instructions, etc. A processor can also be a virtualized processor, a virtual machine or a soft CPU. It can also be, for example, a programmable processor that is equipped with configuration steps for carrying out the aforementioned method according to the invention or is configured with configuration steps such that the programmable processor implements the inventive features of the method, the modules or other aspects and / or sub-aspects of the invention.In particular, the processor can contain highly parallel computing units and powerful graphics modules.

[0033] In the context of the invention, a "storage unit" or "storage module" and the like can be understood to mean, for example, a volatile memory in the form of random access memory (RAM), a permanent memory such as a hard disk or a data storage device, or, for example, a removable storage module. The storage module can also be a cloud-based storage solution.

[0034] In the context of the invention, a "module" can be understood, for example, as a processor and / or a memory unit for storing program instructions. For example, the processor is specifically configured to execute the program instructions in such a way that the processor and / or the control unit performs functions to execute or implement the method according to the invention or a step of the method according to the invention.

[0035] The term "database" refers to both a storage algorithm and the hardware in the form of a storage unit. In particular, the design module 300, test database 200, and parameter database 500 can be implemented as a cloud computing infrastructure.

[0036] In particular, it is provided that the design module 300, the database 400 and the simulation module 700 are integrated in a cloud computing infrastructure.

[0037] A cloud computing infrastructure offers the ability to expand or reduce resources as needed, allowing computing power, storage space, or network resources to be easily adapted to changing requirements. This scalability enables cost optimization and efficient resource allocation without large investments in hardware. Furthermore, users can access applications and data from anywhere with internet access. This accessibility enables collaboration and seamless integration across different devices and locations. Cloud computing also offers flexibility in software deployment, allowing companies to deploy and update applications quickly and without disruption.

[0038] In particular, cloud computing infrastructure is known for its high computing efficiency, as cloud providers maintain large data centers with a vast pool of computing resources. This resource pooling allows multiple users and applications to share and efficiently utilize the same underlying infrastructure. By consolidating resources and dynamically allocating them as needed, cloud providers can achieve higher resource utilization and avoid resource waste. Furthermore, cloud computing relies heavily on virtualization technologies such as virtual machines (VMs) or containers. Virtualization allows the partitioning of physical resources into virtual instances, allowing multiple workloads to run concurrently on the same hardware. This virtualized environment enables efficient resource allocation, scalability, and isolation, leading to greater computing efficiency.Furthermore, cloud infrastructure offers elasticity, allowing users to scale resources up or down as needed. This scalability ensures that computing resources can be quickly provisioned and removed to meet workload demands. By dynamically adjusting resources to current needs, cloud computing optimizes resource utilization, avoids congestion, and reduces unnecessary costs. Furthermore, cloud providers use load balancing mechanisms to distribute the workload evenly across multiple servers or instances. Load balancing ensures that computing resources are used efficiently and avoids resource bottlenecks or overloading of individual servers. Through intelligent routing and load balancing, cloud infrastructure optimizes resource utilization and ensures high computing efficiency.The hardware utilizes energy-efficient processors and high-density memory systems. These hardware optimizations reduce power consumption, increase performance, and improve overall data processing efficiency.

[0039] Since the design of complex components and their visualization are very computationally intensive, high IT efficiency with minimization of resource and energy consumption as well as computing time while maximizing the output is an important aspect of the present invention.

[0040] According to the invention, the design module 300 comprises an AI model 350 with artificial intelligence algorithms. The AI ​​model 350 comprises, in particular, neural networks, but can also comprise other types of machine learning, such as reinforcement learning.

[0041] A neural network consists of neurons arranged in multiple layers and connected to each other in different ways. A neuron is able to receive information at its input from outside or from another neuron, evaluate this information in a specific way, and pass it on in a modified form at the neuron's output to another neuron or output it as the final result. So-called hidden neurons are located between the input neurons and the output neurons. Depending on the type of network, there can be several layers of hidden neurons. They ensure the forwarding and processing of information. The output neurons ultimately provide a result and pass it on to the outside world. The arrangement and interconnection of the neurons gives rise to different types of neural networks, such as feed-forward networks, recurrent networks, or convolutional neural networks.The networks can be trained through unsupervised or supervised learning.

[0042] The AI ​​model 350 specifically includes a convolutional neural network (CNN) for processing 2D and 3D images of component designs. A convolutional neural network (CNN) is well-suited for image analysis tasks and consists of multiple layers of interconnected neurons, including convolutional layers for feature extraction and pooling layers for spatial subsampling. Using these layers, the AI ​​model 350 learns hierarchical representations of image features.

[0043] Generative Adversarial Networks (GANs) can also be used. GANs are deep learning models consisting of two neural networks, a generator and a discriminator. GANs are designed to generate new data that resembles a specific training dataset.

[0044] The AI ​​model 350 may also incorporate reinforcement learning (RL) algorithms. Reinforcement learning can be used to train the AI ​​model to make appropriate decisions in an environment. The AI ​​model learns to perform actions that maximize a reward signal over a specified period of time. By interacting with the environment, the AI ​​model tries different actions and learns from the feedback it receives in the form of rewards or punishments. The goal of the AI ​​model is to find an optimal strategy that maximizes the expected long-term gain. RL enables the AI ​​model to adapt and optimize its behavior in dynamic environments because, by constantly interacting with the environment, it can adjust its decision-making process based on the observed rewards and changes in the environment.This adaptability enables the AI ​​model to cope with situations where fluctuations, uncertainties or temporal changes occur in the environment.

[0045] A Markov decision process is preferred as a reinforcement learning algorithm. However, temporal difference learning (TD learning) can also be used. With TD learning, actions are adjusted not only when the agent or AI model receives the reward, but after each action based on an estimate of the expected reward. Algorithms such as Q-learning and SARSA, actor-critic, or Monte Carlo simulations are also conceivable, enabling dynamic programming and strategy adaptation through iterative procedures.

[0046] In a further development, the AI ​​model 350 can also include deep reinforcement learning algorithms, which combine reinforcement learning with deep neural networks. Deep RL algorithms such as Deep Q-Networks (DQN) or Proximity Policy Optimization (PPO) use deep neural networks as function approximators to handle high-dimensional state and action spaces.

[0047] According to the invention, the AI ​​model 350 is trained to construct an electromechanical part 10 or a mechanical component 15 for an assembly or an object.

[0048] An example of an electromechanical component 10 is a solenoid in an automatic transmission system, which controls gear shifting by actuating various valves and mechanisms. The parameters of a solenoid in an automatic transmission system vary depending on the application and design. A key parameter of a solenoid is the coil resistance, which determines the amount of electrical current required to energize the solenoid. Coil resistance is measured in ohms and affects the overall performance and efficiency of the solenoid. Coil inductance is a measure of the solenoid's ability to store magnetic energy. It is usually measured in henries (H) and affects the response time and magnetic field strength of the solenoid. Stroke length refers to the maximum linear displacement of the magnetic core when energized.It determines the range of motion available for operating valves or engaging and disengaging mechanical components. Another important parameter for the correct operation of a solenoid coil is the force generated by the magnet when energized. This is often specified as the attraction force required to move the magnetic core and the release force required to return it to its original position.

[0049] An example of a mechanical component 15 is a spring component, which is characterized by parameters such as a spring constant, number of coils, etc. Another example is a bearing such as a plain or rolling bearing. In a plain bearing, the parts that move relative to one another either touch each other or are separated from one another by a sliding film. In a rolling bearing, rolling elements such as balls or rollers, which perform a rolling movement, are located between an inner ring and an outer ring. Another example is a housing, which is characterized by parameters such as tightness, flexural rigidity, pressure resistance, interfaces to other components and installation space specifications and the necessary assembly access.

[0050] A part 10 or a component 15 can thus be described by parameters, parameter values, and properties. A parameter and / or a parameter value and / or a property of the part 10 or the component 15 can represent, for example, a dimension, a material, a shape, a characteristic value, a measured value, a safety factor, a thermal resistance, an electrical resistance, an inductance, a conductivity, a strength, a fracture behavior, a color, a density, a plastic deformation, a tensile, compressive, or flexural strength, a material property, a surface property, a heat treatment, a manufacturing process, or assembly access.

[0051] The component 10 or component 15 can be intended for an assembly, a system, or a vehicle. The vehicle can be, for example, a motor vehicle, an autonomously driving vehicle, an agricultural vehicle such as a combine harvester, a robot in production or in service and maintenance facilities, a watercraft, or a flying object such as a drone. The assembly or system can be, for example, a braking system, an electric motor or combustion engine for a motor vehicle, an energy generation system such as a photovoltaic system or a wind turbine, or a production system for manufacturing a product such as a production system for manufacturing a motor vehicle. In particular, the system can be a scientific analysis instrument such as a spectrometer or a medical device.

[0052] Training the AI ​​model 350 requires a comprehensive training dataset 420 containing images and data of known electromechanical components 10 or mechanical components 10. This training dataset 420 is stored in the database 400 and may contain 3D models, CAD data, digital twins, technical drawings, specifications, interface requirements, assembly access, production processes, and other relevant information about the electromechanical components 10 or mechanical components 10. In particular, the training dataset 420 encompasses a broad spectrum of electromechanical components 10 or mechanical components 15 to ensure broad applicability of the AI ​​model 350, so that it can be used for a variety of design tasks.

[0053] This training dataset 420 is processed in a transformation module 320 and converted into a transformed dataset 430 with a data format suitable for training the AI ​​model 350. This transformation of the training dataset 420 can include converting 3D models into a standardized representation, extracting key features from technical drawings, and coding specifications in a structured manner. Depending on the type of data, algorithms for feature extraction, geometric analysis, or natural language processing can be used.

[0054] In particular, the transformation module 320 may comprise at least one encoder and at least one decoder, which are trained by a further training data set and / or an attention algorithm in order to transform the training data set 420 into the transformed data set 430 for the AI ​​model 350.

[0055] The AI ​​model 350 is trained using the thus transformed data set 430. Various machine learning algorithms are used to train the AI ​​model 350, in particular deep learning architectures such as convolutional neural networks (CNNs) or generative adversarial networks (GANs) to learn the patterns and properties of the electromechanical components 10 and the mechanical components 15 from the collected data, and reinforcement learning algorithms to optimize the designs based on specific performance criteria or constraints.

[0056] The trained AI model 350 is then used according to the invention to generate new design drafts of electromechanical components 10 or mechanical components 15 based on user-specific requirements or constraints. The user-specific requirements are input as initial input data 210 via the input module 200. The input module 200 can be integrated into a computer, a tablet, a smartphone, etc., and has a user interface such as a touchscreen, a keyboard, a camera, or a microphone.

[0057] The first input data 210 can be text messages, images, 3D data, digital twins, CAD data, interface requirements, assembly access, graphics, diagrams, video sequences, holograms, mathematical formulas, PowerPoint presentations, which specify the type of the electromechanical component 10 or the mechanical component 15 and can also contain specifications such as installation space, force introduction points, contact surfaces, sealing surfaces, force curves, etc.

[0058] The first input data 210 is transferred to the transformation module 320 and transformed into a data format suitable for processing by the AI ​​model 350. The AI ​​model 350 then uses this transformed input data to generate a design draft 370, which is output by the output module 500 as output data 570. In particular, the output module 500 also includes a transformation module 520 for transforming the data sets of the design draft 370 into a data format that can be displayed on a user interface such as the input module 200. In particular, it can be provided that the input module 200 and the output module 800 are integrated into the same hardware device, such as a computer, a tablet, a smartphone, etc.

[0059] A user can now input changes and further requirements in the form of second input data 220 to the design 370 via the input module 200 in order to iteratively improve it until a target design 375 is created. In particular, it is provided that the AI ​​model 350 has optimization algorithms such as genetic algorithms, particle swarm optimization, or Bayesian optimization for improving existing design 370 in order to propose design alternatives or optimize parameters.

[0060] The second input data 220 for the optimization of a design 370 may, for example, contain the following input: a) Minimize the cross-sectional area of ​​the component down to a cross-section of 0.5 mm. b) Reduce the length of the component to 2 cm.

[0061] In particular, it can be provided that an electrical circuit diagram is entered as second input data 220 and the AI ​​model 350 creates an ideal arrangement of the electrical and electronic elements with regard to weight, heat generation, electromagnetic compatibility, etc. In a further embodiment, it can also be provided that the AI ​​model 350 itself creates an electrical circuit diagram based on predetermined boundary conditions for a specific component or assembly.

[0062] Within the scope of the invention, it may also be provided to validate and further refine the created design drafts 375. For this purpose, the target design draft 375 is transferred to the simulation module 700. The simulation module 700 comprises a software application 750, which contains, for example, simulation tools such as finite element analysis (FEA) or computational fluid dynamics (CFD) to simulate and evaluate the performance of the generated target design draft under various operating conditions.

[0063] Finite element analysis (FEA) is a computational method for analyzing and simulating the behavior of complex structures or systems under various physical conditions. It is used to predict how objects will respond to different loadings, thermal conditions, and environmental conditions. In FEA, a complex structure is decomposed into smaller, finite-sized elements or domains. Each element is defined by a set of mathematical equations that describe its behavior, such as stiffness, elasticity, and thermal properties. The behavior of the entire structure is then determined by solving these equations. Finite element analysis makes it possible to simulate and optimize the target design, predict structural integrity, assess failure risks, and improve performance. This can minimize the need for physical prototypes and testing.

[0064] Computational fluid dynamics (CFD) is a numerical method for simulating and analyzing fluid flow, heat transfer, and other related phenomena, such as those found in an automatic transmission. Flow equations, such as the Navier-Stokes equations, are solved using computational algorithms and techniques.

[0065] The simulation and evaluation results can, in turn, be output via the output module 500 as output data 570. These simulation and evaluation results, as well as the target design draft 375, can also be transferred to the database 400 to be available to the AI ​​model 350 as additional training data sets, thus contributing to the improvement of the AI ​​model 350. In this way, the AI ​​model 350 can be successively improved so that the generated design drafts 370 increasingly conform to the required specifications and standards.

[0066] Furthermore, it may be provided to integrate the AI ​​model 350 into existing CAD software or design platforms so that users can interact with the AI ​​model 350 via a suitable user interface and use the design drafts 375, recommendations or optimizations generated by it in their design processes.

[0067] In a further embodiment, the input module 200 may include an AI agent 250 that reviews the design drafts 370 generated by the AI ​​model 350 and makes suggestions for improvement. The AI ​​agent 250 also includes artificial intelligence algorithms such as neural networks and reinforcement learning and is also trained in a training phase using the training data set 420 or the transformed data set 430 or another data set.

[0068] Within the scope of the invention, it is provided to vary the AI ​​model 350 according to the respective requirements and the available data and resources with regard to the specific techniques and algorithms used. Thus, it can be provided that a first AI model is designed and trained for the design of mechanical components, while another second AI model is specialized in the design of electrical and electronic components and a third AI model is specialized in the package arrangement of various electrical components. A package arrangement of various electrical components is a design or assembly scheme in which different types of electronic components required for a specific purpose or function are arranged, e.g., on a circuit board or a circuit, or e.g.Electrical components are positioned relative to one another in a given installation space, aligned in space and, for example, independently connected electrically using additional components (such as busbars). The components can be, for example, resistors, capacitors, transistors, inductors, relays, contactors, circuit boards, fuses, cable harnesses or other components that fulfil an electrical function. In addition, the AI ​​can then develop proposed solutions for the mechanical connection of the electrical components, such as a holder or defined connection points, which then have to be represented in a housing. The correct arrangement of the components is crucial for the flawless functioning of the electronic devices or systems.

[0069] The three AI models can then initially work independently of each other and interact with each other in a later phase of the design process. Within the scope of the invention, however, additional AI models can also be provided and trained with specific training data in order to deliver optimal results for a specific requirement. In particular, it is envisaged that the AI ​​models contain interfaces that enable interaction between the various AI models.

[0070] Image recognition algorithms, for example, are particularly important for the design of mechanical components. They use an appropriately trained training dataset to correctly recognize different object categories. The neural network model undergoes a training phase in which it learns to recognize objects from provided annotated images. The training data is fed into the network, the weights and biases of the neurons are adjusted using backpropagation, and the model parameters are optimized using algorithms such as stochastic gradient descent or its variants. Transfer learning can also be used to further improve the model's performance. Transfer learning uses pre-trained models trained on large datasets such as ImageNet.The knowledge and properties of the previously trained model can be used as a starting point for the current task, resulting in faster convergence and improved accuracy.

[0071] In Fig. 2 shows the method steps for the automated construction and / or design of at least one electromechanical component 10 or one mechanical component 15, in particular for an assembly or an object.

[0072] In a step S10, a training data set 420 is created, wherein the training data set 420 contains images and / or drawings and / or sensor signals and / or interface requirements and / or assembly access and / or further data of electromechanical components 10 and / or mechanical components 15.

[0073] In a step S20, an AI model 350 is trained with the training data set 420.

[0074] In a step S30, first input data 210 for a design task are entered and the input data 210 are passed on to the trained AI model 350.

[0075] In a step S40, a design draft 370 is generated by the AI ​​model 350.

[0076] In a step S50, requirements are entered in the form of second input data 220 for optimizing and improving the design 370.

[0077] In a step S60, steps S40 and S50 are repeated in an iteration until a target design draft 375 is obtained.

[0078] In a step S70, the target design draft is output or stored in a database 400.

[0079] Fig.3 schematically illustrates a computer program product 900 comprising executable program code 950 configured to perform the method according to the first aspect of the present invention when executed.

[0080] The present invention can improve the design of electromechanical parts and mechanical components by allowing the AI ​​model to generate a large number of design alternatives more quickly. This enables entirely new solutions to be found and designs and geometries to be optimized according to various criteria such as installation space, contact media, location, manufacturing processes, joining techniques, mechanical / thermal / chemical / electrical loads, assembly access, performance, efficiency, weight, or cost. In particular, significant time and cost savings can be achieved, as conventional design processes often involve manual iterations and physical prototypes, which can be time-consuming and expensive. The AI ​​model can be used particularly for complex design tasks with complicated geometries, as it can operate in multidimensional design spaces.In addition, the AI ​​model can generate design drafts that meet a wide range of technical constraints, manufacturability requirements, and functional specifications. For example, constraints such as environmental compatibility or energy consumption can be better considered. The AI ​​model can thus contribute to improving the performance and efficiency of electromechanical parts and mechanical components.

[0081] Furthermore, the AI ​​model can learn from existing designs and historical data, enabling continuous development. Furthermore, since the AI ​​model can be integrated into a common platform, it facilitates the exchange of design ideas, thus promoting design innovation and creativity. The AI ​​model of the invention is thus a powerful tool that complements and expands the capabilities of engineers, enabling them to create better designs more efficiently. Reference symbol 10 components 15 components 100 systems 200 input module 210 first input data 220 second input data 250 AI agents 300 construction module 320 Transformation Module 350 AI model 370 Construction design 375 Target construction design 400 database 420 training data set 430 transformed data sets 500 output module 520 Transformation Module 570 output data 700 Simulation module 750 software applications 900 Computer program product 950 program code QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 2018 / 0285517 A1

[0006] US 2021 / 0216674 A1

[0007]

Claims

[1] Method for the automated construction and / or design of at least one electromechanical component (10) or one mechanical component (15), in particular for an assembly or an object, comprising the following steps - Creating (S10) a training data set (420), wherein the training data set (420) contains images and / or drawings and / or sensor signals and / or interface requirements and / or assembly access and / or further data of electromechanical components (10) and / or mechanical components (15); - Training (S20) an AI model (350) with the training data set (420); - entering (S30) first input data (210) for a design task and passing the input data (210) to the trained AI model (350); - generating (S40) a design draft (370) from the AI ​​model (350); - inputting (S50) requirements in the form of second input data (220) for optimizing and improving the design (370); - repeating (S60) the method steps (S40) and (S50) in one iteration until a target design (375) is obtained; - Outputting (S70) the target design (375) and / or storing the target design (375) in a database (400). [2] The method of claim 1, wherein the AI ​​model (350) uses neural networks and / or recurrent neural networks and / or convolutional neural networks and / or a reinforcement learning algorithm. [3] The method according to claim 2, wherein the reinforcement learning algorithm is designed as a Markov decision method, temporal difference learning (TD learning), Q-learning, SARSA, Monte Carlo simulation or Actor-Critic. [4] Method according to one of claims 1 to 3, wherein at least one parameter and / or a parameter value and / or a property of the part (10) or the component (15) represents a dimension, a material, a shape, a characteristic value, a measured value, a safety factor, a thermal resistance, an electrical resistance, an inductance, a conductivity, a strength, a fracture behavior, a color, a density, a plastic deformation, a tensile, compressive, flexural strength, a material property, a surface property or a heat treatment. [5] The method of any one of claims 1 to 4, wherein the target design (375) is passed to a simulation module (700), the simulation module (700) comprising a software application (750) containing simulation tools such as finite element analysis (FEA) or computational fluid dynamics (CFD). [6] Method according to one of claims 1 to 5, wherein the target design draft (375) is integrated into the training data set (420) for further training of the AI ​​model (350). [7] Method according to one of claims 1 to 6, wherein the training data set (420) is transformed by a transformation module (320) into a transformed data set (430) having a data format that can be processed by the AI ​​model (350). [8] Method according to one of claims 1 to 7, wherein an AI agent (250) is provided which checks the design drafts (370) generated by the AI ​​model (350) and makes suggestions for improvement. [9] Method according to claim 8, wherein the AI ​​agent (250) comprises artificial intelligence algorithms such as neural networks and reinforcement learning and is also trained with the training data set (420) or the transformed data set (430) in a training phase. [10] System (100) for the automated construction and / or design of at least one electromechanical component (10) or a mechanical component (15), in particular for an assembly or an object, comprising an input module (200), a design module (300), a database (400), and an output module (500); wherein the database (400) contains a training data set (420) with images and / or drawings and / or sensor signals and / or interface requirements and / or assembly access and / or further data of electromechanical components (10) and / or mechanical components (15); wherein the design module (300) contains an AI model (350) that can be trained with the training data set (420); wherein the input module (200) is configured to forward input first input data (210) and second input data (220) for a design task to the trained AI model (350);wherein the trained AI model (350) is configured to iteratively generate a plurality of design drafts (370) to create a target design draft (375); and wherein the output module is configured to output the target design draft (375) or to store it in the database (400); [11] The system (100) of claim 10, wherein the AI ​​model (350) uses neural networks and / or recurrent neural networks and / or convolutional neural networks and / or a reinforcement learning algorithm; and wherein the reinforcement learning algorithm is configured as a Markov decision method, temporal difference learning (TD learning), Q-learning, SARSA, Monte Carlo simulation, or actor-critic. [12] The system (100) of claim 10 or 11, wherein the target design (375) is passed to a simulation module (700), the simulation module (700) comprising a software application (750) containing simulation tools such as finite element analysis (FEA) or computational fluid dynamics (CFD). [13] The system (100) of any one of claims 10 to 12, wherein the training data set (420) is transformed by a transformation module (320) into a transformed data set (430) having a data format that can be processed by the AI ​​model (350); and wherein the transformation module (320) comprises at least one encoder and at least one decoder. [14] System (100) according to one of claims 10 to 13, wherein an AI agent (250) is provided which reviews the design drafts (370) generated by the AI ​​model (350) and makes suggestions for improvement; and wherein the AI ​​agent (250) comprises artificial intelligence algorithms such as neural networks and reinforcement learning and is also trained with the training data set (420) or the transformed data set (430) in a training phase. [15] A computer program product (900) comprising an executable program code (950) configured to carry out the method according to any one of claims 1 to 9 when executed.

Images