Methods for generating custom connectors using an AI generator

By using large language models and generative design algorithms, the problem of time-consuming and labor-intensive custom connector design has been solved, enabling rapid and low-cost custom connector design and manufacturing to meet customers' needs for rapid time-to-market.

CN123125597APending Publication Date: 2026-11-17HARTING INT INNOVATION AG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202580028613.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-06-14
Filing Date
2025-04-15
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies are time-consuming and labor-intensive in custom connector design, requiring a lot of manual work and making rapid iteration difficult, thus failing to meet customers' requirements for rapid time-to-market.

Method used

A method for generating custom connectors using Large Language Models (LLM) is employed. This method converts human natural language input into computer-interpretable specifications, combines generative design algorithms, optimizes 2D and/or 3D model datasets, and leverages machine learning and AI-driven generative design algorithms to rapidly explore the design space and optimize connector design.

Benefits of technology

It reduces manual engineering time, improves design efficiency and consistency, shortens the time from concept to product, reduces material waste and costs, and enables faster iteration and higher quality custom connector designs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123125597A_ABST
    Figure CN123125597A_ABST
Patent Text Reader

Abstract

This invention provides a method for generating custom connectors using an AI generator, comprising the following steps: - providing some input information to an LLM, wherein the input information includes human natural language information; - processing the input information into a set of computer-interpretable specifications using the LLM; - aligning the set of specifications with physical requirements; - using a generative design algorithm to generate an optimized 2D and / or 3D model dataset containing the set of specifications and the physical requirements; - outputting a 2D and / or 3D dataset provided by the custom connector model.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] This invention relates to the easy creation of custom connectors based on customer specifications.

[0002] The goal is to provide shorter time-to-market and faster iterations based on customer feedback.

[0003] Today, the process is based on an initial design by educated engineers who refine and iterate using software simulations and / or testing of physical prototypes. This process is very time-consuming and requires a significant amount of manual work.

[0004] This invention addresses the challenge of automatically generating custom connectors.

[0005] The present invention solves this problem by using the method according to claim 1.

[0006] According to the present invention, this is achieved by the method for generating custom connectors using an AI generator as described in claim 1, the product data file as described in claim 10, the computer program product as described in claim 11, the computer system as described in claim 12, and the method for generating custom connectors as described in claim 13.

[0007] For example, advantageous embodiments may be obtained from the dependent claims. The content of the claims becomes part of the specification through explicit inclusion.

[0008] This invention provides a method for generating custom connectors using an AI generator, comprising the following steps: - Provide some input information to the Large Language Model (LLM), wherein the input information includes human natural language information; - The input information is processed using the LLM into a set of specifications that can be interpreted by a computer; - Align the set of specifications with the physical requirements; - Use generative design algorithms to generate an optimized 2D and / or 3D model dataset containing the specification set and the physical requirements; - Output 2D and / or 3D datasets of custom connector models.

[0009] In the context of this invention, the described method is a computer-implemented method designed to be executed by a computer.

[0010] This invention offers significant advantages in the field of custom connector design and manufacturing. By leveraging artificial intelligence and machine learning techniques, this approach streamlines the process of creating custom connectors tailored to specific customer requirements. Using a Large Language Model (LLM) to interpret human natural language input allows for a more intuitive and accessible interface, enabling customers to describe their needs without requiring extensive technical knowledge.

[0011] Automating input information into computer-interpretable specifications significantly reduces the time and effort traditionally required to manually translate customer requirements. This automation minimizes errors that can occur during manual interpretation and specification creation. Alignment between specifications and physical requirements ensures that the resulting design is not only theoretically sound but also practically feasible for manufacturing.

[0012] By employing generative design algorithms, this invention rapidly explores a vast design space, considering the impractical need for extensive iterations and optimizations for manual evaluation by human designers. This results in innovative connector designs optimized simultaneously for multiple criteria, such as performance, cost, and manufacturability. The output of optimized 2D and / or 3D model datasets provides a direct link to the manufacturing process, potentially reducing the time from concept to production.

[0013] Furthermore, this approach enhances the iterative design process, allowing for rapid adjustments based on customer feedback or changing requirements. The speed and efficiency of the AI-driven approach enable faster prototyping cycles, thereby accelerating the overall product development timeline. This acceleration in design and production provides a competitive advantage in the fast-paced electronics and connectivity market.

[0014] This invention also facilitates cost reduction in the custom connector development process by minimizing the need for extensive manual engineering time and physical prototyping. By digitally simulating and optimizing the design before physical fabrication, this method reduces material waste and associated costs. Furthermore, the consistency and repeatability of the AI-driven process lead to higher quality standards and more reliable results in custom connector design.

[0015] A prime example of a connector type that can be easily customized is the PushPull connector system from manufacturer Harting. This can be found on the following website: https: / / www.harting.com / de-DE / s / splp-harting-pushpull and from the following manual: "HARTING Device Connectivity," Publication No. MO / 2016-03-01 / 5.7598 41 007 0201, Version 07 5, pp. 66-137. This connector provides a series of easy-to-handle housings for transmitting data, signals, and power across a variety of applications. The modular concept includes standard RJ45 and USB interfaces, as well as connector faces for signal, power transmission, and optical interfaces. Various metal or plastic materials can be used. The system provides connectors for signal, power, and optical interfaces in easy-to-use housings suitable for different applications. It is designed as a solution for Ethernet applications and is characterized by a robust housing.

[0016] The PushPull system comprises cable assemblies, connector assemblies with crimped terminals, and device connection technology. It provides solutions for demanding outdoor applications, such as camera systems used to monitor process plants, telecommunications equipment, and transportation technologies. These applications expose the connectors to environmental factors such as sun, wind, rain, and salt spray in coastal areas.

[0017] PushPull connector systems include a variety of connector types, including those for display interfaces and robotic applications.

[0018] Another interesting type of connector for customization is the cable-to-board connector, especially those with angled pins shown in the left column of page 864 of the aforementioned manual, "HARTING Device Connections." The method of the present invention is particularly useful for these connectors because it allows for optimization not only of the plug's pins but also of their corresponding pins (receptacle holes) on the board.

[0019] It has been realized that a method is needed to overcome one or more of these problems.

[0020] Input information can be written submissions from the chatbot, voice-to-text input via a loudspeaker, or any other written or digital input.

[0021] Large language models (LLMs) are a type of artificial intelligence (AI) model designed to understand, generate, and sometimes translate human language. They are “large” in the sense that they consist of many parameters, ranging from hundreds of millions to billions, enabling them to capture vast amounts of information about language patterns, grammar, context, and semantics.

[0022] Several current large language models (LLMs) can be applied to generate custom connectors, including both open-source and proprietary versions. These models include the chatGPT family from OpenAI, BERT from Google, and the LLaMA model from MetaAI. These models can be adapted and fine-tuned for specific tasks related to generating custom connectors, where the choice of model depends on factors such as task-specific requirements, available computational resources, and licensing considerations.

[0023] LLMs can be fine-tuned specifically for connector design tasks through a multi-stage process that enhances their domain expertise and technical accuracy. As used in this paper, the term "fine-tuning" can refer to the process of adapting a pre-trained language model to a specific task or domain by further training it on a smaller, task-specific dataset. Fine-tuning can involve adjusting the model's parameters to optimize performance with respect to the target task while preserving the general language understanding acquired during pre-training. This process can modify the weights of some or all layers in the neural network, potentially leading to improved accuracy, relevance, and specificity for the intended application. Fine-tuning can allow the model to learn domain-specific vocabulary, conventions, and reasoning patterns, which can enhance its ability to generate more accurate and context-appropriate responses in the target domain.

[0024] This fine-tuning begins with a comprehensive dataset of connector specifications, technical documents, industry standards (such as IEC, UL, CSA, and MIL-STD), and historical design cases representing successful custom connector implementations. The training corpus should include detailed connector datasheets, engineering drawings, material property databases, electrical performance characteristics, and documented design fundamentals from previous projects. Domain-specific pre-training can be achieved by exposing the model to connector engineering literature, technical manuals, and industry-specific terminology. This phase helps the LLM develop specialized vocabulary and conceptual understanding of connector design principles, manufacturing constraints, and performance requirements. The model can then be supervised fine-tuned using paired examples of natural language requests and their corresponding technical specifications, where human experts provide the fundamental factual translation from customer requirements to precise engineering parameters.

[0025] Reinforcement learning from human feedback (RLHF) can further refine the model's capabilities by incorporating evaluations from experienced connector design engineers. In this approach, multiple specification outputs are generated for the same input, and engineers rank or score them based on technical accuracy, completeness, and feasibility. These rankings train a reward model that guides the LLM toward producing technically more reasonable specifications.

[0026] To enhance contextual understanding, the model can be trained using multimodal inputs, which include not only textual descriptions but also images such as connectors, circuit diagrams, or application environments. This helps the LLM better understand the physical and operational context in which the connector will function. Fine-tuning should also include adversarial training examples presenting challenging or ambiguous requirements, teaching the model to recognize when additional information is needed and to generate appropriate clarifying questions.

[0027] A continuous learning mechanism can be implemented to allow the model to improve over time based on feedback from actual implementation results. When the generated specifications lead to a successful connector design, this positive outcome strengthens the model's approach. Conversely, when specifications require revision, these corrections are fed back into the training process. The fine-tuned LLM should also be integrated with a knowledge retrieval system, such as internet access, to obtain the latest standard definitions, material properties, and manufacturing capabilities, ensuring that the generated specifications remain up-to-date with industry developments and technological advancements.

[0028] The resulting specification set can be reviewed by human experts to ensure it is technically sound and meets all required standards and user needs. LLMs may not fully grasp the engineering complexities or safety standards involved in designing such components.

[0029] Hint patterns are structured methods for specifying inputs to an LLM to elicit a specific type of response. Several hint patterns can be used for custom connector designs: Interview-style prompts are particularly effective for gathering comprehensive specifications. This pattern involves the LLM asking a series of questions to gather all necessary information before moving to the next step in the design process. Mind chain prompts guide the LLM through a logical sequence of questions, where each subsequent question builds on previous answers to create a complete specification profile. This ensures that all interdependent parameters are properly considered. Tree-based prompts allow the LLM to branch its questioning based on previous responses. For example, if the user specifies an outdoor connector, the LLM would follow up with environmental resistance questions that wouldn't be relevant to indoor applications. Checklist prompts ensure the LLM systematically covers all required specifications by working through a predefined list of necessary parameters, asking about each parameter until the checklist is complete.

[0030] For effective interview patterns, LLM can implement these patterns, for example, as follows: The initial broad question: "What type of connector do you want to design?" " The specific questions that followed: How many pins or contacts will this connector require? " - "What voltage and current ratings are required?" " - "What environmental conditions must this connector withstand?" " LLM tracking has provided information on which specifications are still missing and continues to provide prompts until all the necessary information has been collected.

[0031] This can be combined with a verification prompt where LLM repeatedly returns the understood specifications and requests confirmation: "I understand you need a waterproof circular connector with 8 pins rated for 24V DC. Is this correct, or do you want to modify any specifications?" ".

[0032] Prioritization prompts ensure that critical parameters are addressed before moving on to optional features: "First, let's establish the necessary electrical requirements before discussing physical dimensions and special features."

[0033] Contextual hints allow LLMs to tailor their questions to industry standards or typical use cases: "Since you mentioned aerospace applications, I will need to inquire about vibration resistance and temperature range specifications that meet industry standards."

[0034] The completion check prompt helps LLM verification that all necessary information has been gathered: "I have gathered information on electrical ratings, physical dimensions, and environmental requirements. Are there any special characteristics or compliance standards that we have not yet discussed before proceeding?" ".

[0035] The interview process continues until the LLM determines that it has gathered all the specifications needed for the next design phase, at which point it can transition from information gathering to information processing and design generation.

[0036] By implementing these prompting patterns in an interview-like manner, LLM creates a conversational interface that systematically gathers all the necessary specifications while retaining flexibility for the user's level of technical expertise and specific requirements.

[0037] The interview process used to generate custom connectors can also be conducted via voice interaction, allowing users to speak directly to the AI ​​tool without typing. This voice-based interface enhances user convenience and accessibility, especially when typing is impractical or for users who prefer verbal communication.

[0038] In voice-enabled systems, AI tools can incorporate speech recognition technology to convert spoken words into text. This speech-to-text conversion allows Large Language Models (LLMs) to process spoken input in a manner similar to written text. The system can then leverage Natural Language Processing (NLP) techniques to interpret the user's spoken commands and queries.

[0039] To respond to users, AI tools can employ text-to-speech (TTS) technology. This allows the system to convert its text-based responses into audible speech, creating fully interactive voice conversations. TTS systems can be designed to produce natural-sounding speech with appropriate intonation and cadence to enhance understanding. The voice interface can support the same prompting patterns and interview techniques described earlier. For example, the AI ​​tool could verbally ask an initial, broad question: “What type of connector would you like to design?” Then, the user can respond verbally, and the conversation can continue with follow-up questions and clarifications. Multilingual support can also be applied to accommodate users who prefer to speak languages ​​other than English. Voice-enabled AI tools can also provide visual feedback or display key information on the screen, creating a multimodal interactive experience. This can help users verify that their verbal input has been correctly interpreted and allow them to visually review the collected specifications.

[0040] In situations involving precise technical terminology or complex specifications, the system can offer a hybrid approach. Users can state their general requirements and then refine or confirm specific details via on-screen prompts or by spelling out key information. By providing both text-based and voice-based interaction options, the AI ​​generator for custom connectors can adapt to a wider range of user preferences and use cases, potentially improving accessibility and user experience in the connector design process.

[0041] Examples of a set of specifications that can be output from an LLM based on user input include: Connector type: Circular DIN Number of pins: 7 Rated voltage: 250 V Rated current: 5A per pin Contact resistance: ≤5mOhm Insulation resistance: ≥100 MOhm at 500 V DC Operating temperature range: -40℃ to 85℃ Inlet protection rating: IP67 (dustproof and waterproof) Casing material: Nickel-plated brass Contact material: Gold-plated copper alloy Termination type: welding cup Compliance: RoHS Directive, CE.

[0042] Information generated by LLM can be organized and stored in a structured data format similar to a table or matrix. This format can utilize predefined technical terms as categories, ensuring that information is presented in a standardized manner. By adopting such a structure, the system can facilitate easier parsing, comparison, and analysis of connector specifications across different designs. The standardized format can include columns for each technical parameter, such as connector type, pin count, rated voltage, rated current, resistance value, temperature range, protection rating, material, and compliance standards. Each row in this tabular structure can represent a unique set of connector specifications based on user input, allowing for efficient storage and retrieval of multiple design outputs. This data organization method enhances the system's ability to quickly access and compare different connector designs, streamlining the decision-making process for engineers and designers. Furthermore, the standardized format improves interoperability with other design tools and databases, enabling seamless integration into a wider range of engineering workflows and systems.

[0043] In another embodiment, the LLM is configured to validate input information based on at least one of the following parameters: - Value range; - Threshold; - Symbol; - contradiction.

[0044] The verification process implemented by LLM serves as a critical quality control mechanism to ensure the technical feasibility and consistency of custom connector specifications. When verifying value ranges, LLM checks whether specified parameters such as rated voltage, current capacity, or dimensional measurements fall within physically possible and industry-standard boundaries. Threshold verification ensures that minimum requirements for performance characteristics, such as insulation resistance or contact durability, are met or exceeded. Sign verification prevents logical errors by confirming that direction values ​​are correctly specified (e.g., positive values ​​for resistance measurements). Inconsistency detection may represent the most complex aspect of verification, where LLM identifies and flags mutually exclusive requirements, such as specifications for both high current capacity and a small form factor, which would create thermal management issues or material choices incompatible with stated environmental resistance requirements. This comprehensive verification framework significantly reduces design iterations by identifying potential problems before the physical modeling phase begins.

[0045] The LLM can be configured and / or trained to correct incorrect and / or incomplete input information. The LLM includes a set of training data comprising multiple pre-designed plug connectors and / or is trained using this set of training data. The LLM can be configured to access lookup table information and / or other external information.

[0046] LLM can also be configured to refine the specification set by entering new or modified prompts.

[0047] In one embodiment, the AI ​​generator is configured to execute locally, or to outsource tasks to the cloud if the local instance's capacity utilization exceeds a certain threshold, where the threshold considers at least one of the following metrics: percentage of total available CPU utilization, GPU memory consumption, RAM usage, number of floating-point operations per second (FLOPS), and total duration of the task execution. The capacity utilization threshold that triggers cloud outsourcing can be defined along multiple dimensions. Primarily, it can be expressed as a percentage of total available computing resources, typically varying from 75% to 90% of CPU utilization, GPU memory consumption, or RAM usage. For example, a cloud outsourcing agreement can be initiated when a local system detects that CPU utilization has been consistently above 85% for a predetermined period (e.g., 5 minutes). The threshold can also be defined in absolute terms, such as when processing demands exceed a specific number of floating-point operations per second (FLOPS), typically in the trillions of FLOPS range for complex generative design tasks. Time-based thresholds can also be implemented, where tasks estimated to require more than a certain duration (e.g., 2 hours) on local hardware are automatically routed to cloud resources to maintain system responsiveness for other users or tasks.

[0048] When thresholds are exceeded, the system employs a resource allocation algorithm to determine which specific components of the workflow should be outsourced. For data security reasons, LLM processing may remain local, while computationally intensive simulations and generative design iterations are distributed across cloud resources. This hybrid approach optimizes both performance and cost-effectiveness, as cloud resources typically incur usage-based costs. The system maintains continuous monitoring of both local and cloud resource utilization, dynamically adjusting the distribution of computational tasks based on real-time performance metrics, the priority levels of different design projects, and cost considerations. Furthermore, the system implements a data synchronization protocol to ensure all design information remains consistent across the distributed computing environment, with appropriate encryption and security measures applied to data transmitted between the local and cloud systems.

[0049] Training data, including pre-designed plug connectors, serves as the foundational knowledge base for the LLM, enabling it to identify common patterns, specifications, and design constraints in connector engineering. This dataset typically includes comprehensive technical documentation of existing connector designs, encompassing dimensional specifications, electrical parameters, mechanical properties, material composition, and performance characteristics across a wide range of operating conditions. When processing user input, the LLM leverages this pre-existing knowledge to identify similarities between the requested custom connector and previously designed connectors, allowing it to suggest appropriate specifications even when user input is ambiguous or incomplete. For example, if a user requests a "waterproof connector for marine applications" without specifying an ingress protection rating, the LLM can use its training data to identify that marine environments typically require at least IP67 or IP68 protection and suggest that specification accordingly.

[0050] The lookup table integration provides a structured reference mechanism to complement the neural network capabilities of the LLM with precise deterministic data retrieval. These tables are meticulously organized databases containing standardized specifications, industry-standard parameters, material properties, and regulatory compliance requirements for connector components. When the LLM processes user requests, it can query these lookup tables to retrieve the exact values ​​of specific parameters, ensuring the technical accuracy of the generated specifications. For example, when a user specifies a connector for a particular application, the LLM can access a lookup table containing standard pin configurations, rated voltages, and current capacities suitable for that application area. The lookup tables also contain relational data that helps the LLM understand the interdependencies between different specifications, such as how increasing current capacity might necessitate a larger conductor cross-section or how certain material choices might affect temperature resistance.

[0051] The system can systematically compare structured data tables obtained from user interviews with lookup tables to validate and refine specifications. This comparison process may involve matching each parameter from the user-provided data against the corresponding entry in the lookup table. The system can check whether the user-specified values ​​fall within the acceptable range defined in the lookup table, or whether they align with the standardized options for that particular connector type or application. If the user-provided data differs from standard values, the system can flag these differences for further review or automatically suggest the closest matching standard specification. This comparison step can also help identify any missing key parameters that the user may have overlooked, prompting the system to request additional information or suggest default values ​​based on similar connector designs in the lookup table. By cross-referencing user input with standardized data, the system can ensure that the final connector specification is both technically feasible and conforms to industry standards.

[0052] The collaborative interaction between pre-trained connector knowledge and lookup table functionality creates a robust system for specification refinement. When the LLM encounters incomplete information, it first attempts to infer the missing specification based on its pre-trained understanding of connector design principles and typical configurations. It then validates these inferences against lookup tables to ensure they comply with industry standards and physical feasibility constraints. This validation process involves cross-referencing multiple tables to check for consistency across interdependent parameters. For example, if the LLM infers a contact material based on the application environment, it will check a material compatibility table to ensure that the material selection is suitable for the voltage, current, and temperature conditions specified elsewhere in the requirements. This multi-layered approach to specification completion ensures that even with minimal user input, the resulting connector design will meet all necessary functional requirements and comply with relevant industry standards.

[0053] LLMs can also be configured to refine the specification set by inputting new or modified prompts. The computational requirements for locally executing AI generators can be substantial, especially when dealing with complex connector designs with numerous parameters and constraints. Systems typically require high-performance computing hardware with multi-core processors, ample RAM, and dedicated graphics processing units (GPUs) to efficiently handle the computational demands of both large language models and generative design algorithms. For example, an LLM component might require 16-32 GB of RAM for optimal performance, while generative design algorithms (especially when performing finite element analysis or other simulation tasks) may require additional dedicated computing resources. Storage requirements are also critical, as the system must maintain a database of connector specifications, material properties, and historical design data.

[0054] In another embodiment, the LLM is configured to output processed information to the user from human natural language information including a set of specifications. This output can be presented during an interview mode, where the LLM displays the explained specifications on the screen as it collects the explained specifications through text-based or voice-based interaction. For text-based interviews, the specifications can appear alongside the dialogue in a structured format, allowing the user to visually confirm the accuracy of the explained requirements in real time. In voice-based interaction, the system can display the identified specifications on the screen while simultaneously providing verbal confirmation via text-to-speech technology. This multimodal presentation enables the user to view and verify the processed specifications during the interview process, facilitating immediate correction or adjustment before proceeding to the design phase.

[0055] In one embodiment, physical requirements are aligned through lookup tables and / or algorithmic and / or ML-guided parameter optimization.

[0056] The algorithm can include physical factors, such as material shrinkage behavior and heat transfer between contacts.

[0057] In a preferred embodiment of the method of the present invention for generating custom connectors using an AI generator, it includes the step of aligning physical requirements via a lookup table, wherein alignment includes using the lookup table to map a set of specifications toward the physical requirements.

[0058] Using lookup tables to align a set of specifications with physical requirements can involve mapping user-provided requirements or an already defined set of specifications. A lookup table is essentially a data structure commonly used in computation that stores known values ​​in an organized manner, allowing a program to quickly and efficiently find specific data. Lookup tables can function as follows.

[0059] First, the lookup table can be configured to capture a range of possible specifications for different parts or attributes of the product in question. This table will be indexed in a way that allows for quick retrieval of information based on a given input.

[0060] When users provide their requirements to the LLM, the LLM will parse the input text to extract keywords, phrases, and values ​​related to the specifications as described above.

[0061] The parsed keywords and values ​​are then matched against keys, indexes, value ranges, or thresholds in a lookup table. The matching process may involve some natural language processing (NLP) capabilities to handle synonyms, acronyms, or jargon commonly used in the industry.

[0062] After identifying the correct key in the lookup table, the matching physical component or corresponding specification of the connector can be retrieved. These retrieved values ​​are standardized specifications or components that match the user's requirements or description.

[0063] Then, the lookup table or AI generator compiles all the selected specifications and presents them to the user in a consistent format, such as a report or structured digital document or dataset that can be provided to the generative design algorithm.

[0064] The final step in a parameter-optimized alignment process with lookup tables or ML guidance can include human expert knowledge reviewing the output to ensure that all requirements are met and that specifications align with user needs and industry standards.

[0065] Examples of functions for looking up tables can be found as follows: Manual input: "I need a connector that can handle at least 500 volts, is resistant to salt water corrosion, and is suitable for outdoor marine environments."

[0066] Using LLM analysis: The system identified "500 volts", "salt water corrosion resistance" and "suitable for outdoor marine environments" as key requirements.

[0067] Lookup table: "500V" -> Connector category: High voltage, Specification: Rated voltage ≥ 500V "Saltwater corrosion resistant" -> Material: Corrosion resistant, Specifications: Marine grade stainless steel or nickel-plated. "Outdoor marine environment" -> Weatherproof rating, specification: IP67 or higher.

[0068] In a preferred embodiment of the method of the present invention for generating custom connectors using an AI generator, it includes a step of aligning physical requirements through ML-guided parameter optimization, wherein the ML-guided parameter optimization utilizes a training dataset comprising various design parameters and their corresponding performance metrics, so that the trained model can predict the optimal set of design parameters.

[0069] Machine learning (ML)-guided parameter optimization uses advanced techniques such as regression, classification, and even deep learning models to predict and optimize parameters for a given set of requirements. This approach can serve as an alternative to lookup tables when designing custom parts or products, such as electrical connectors. The ML process typically involves the following steps.

[0070] First, the assembly includes a comprehensive dataset of various design parameters and their corresponding performance metrics. This dataset can contain historical design data from previous connectors, simulation results, experimental data, or any combination thereof.

[0071] Then, a machine learning model can be selected for parameter optimization. The selection of an appropriate machine learning model for connector design optimization involves a system evaluation process that considers multiple factors, including the nature of the design parameters, available training data, computational resources, and the specific optimization objective.

[0072] One efficient approach to model selection is to leverage readily available AutoML (Automated Machine Learning) platforms that can systematically evaluate multiple model architectures and hyperparameter configurations. These platforms provide controlled test connector design datasets to automate workflows for testing various models, including input requirements and output connector designs. They typically implement a range of algorithms, including gradient boosting machines, neural networks, support vector machines, and ensemble methods, evaluating each based on performance metrics relevant to connector design optimization.

[0073] When selecting a model through AutoML or manual evaluation, several key criteria should be assessed. Predictive accuracy is paramount, especially for critical parameters such as electrical performance characteristics and mechanical integrity. The model's ability to generalize across different connector types and use cases should be evaluated using cross-validation techniques. Computational efficiency becomes particularly important when the model needs to be integrated into interactive design workflows that provide valuable real-time feedback. The model's interpretability may be necessary when design engineers need to understand the reasoning behind recommendations for specific parameters. Furthermore, the model's ability to handle multi-objective optimization scenarios is crucial, as connector design often involves balancing competing requirements, such as size constraints on thermal performance.

[0074] To test model candidates, a comprehensive evaluation protocol should be implemented. This includes dividing the available connector design data into appropriate training, validation, and test sets, and considering stratification to ensure representative sampling across different connector types and applications. Performance metrics should be tailored to connector design requirements, such as mean absolute percentage error using dimensional parameters, classification metrics for classification decisions (e.g., material selection), and custom metrics reflecting domain-specific constraints (e.g., manufacturing feasibility or electrical compliance).

[0075] Several model architectures have demonstrated particular promise for connector design optimization. For more complex optimization scenarios, deep learning methods may be appropriate. Specifically, neural network architectures with dedicated layers for handling both continuous parameters (size, electrical ratings) and categorical features (material type, connector style) can be effective.

[0076] Ensemble approaches that combine multiple model types often yield superior results by leveraging the strengths of different methods. For example, a stacked ensemble might use a random forest for initial parameter range estimation, followed by a more specialized neural network for fine-tuning specific dimensions. The model selection process should also consider integration capabilities with existing engineering software and simulation tools. Models whose predictions can be exported in formats compatible with CAD systems, finite element analysis software, or electromagnetic simulation tools will streamline the overall design workflow. Furthermore, models that support active learning—where they can suggest which design variants should be physically prototyped or simulated next to maximize information gain—can significantly accelerate the optimization process.

[0077] Ultimately, the selected model should not only be validated on historical data, but also through its recommended finite physical prototyping or high-fidelity simulation. This final validation step ensures that the model's optimization recommendations are effectively translated into real-world connector performance before being fully integrated into the production design workflow.

[0078] Once the model type, architecture, and hyperparameters are selected, the ML model can be trained on the prepared dataset. The training process may involve using optimization algorithms to adjust the weights and biases within the model to minimize the difference between the model's predictions and the actual results. Different ML models can be used for this purpose, such as decision trees for classification tasks, support vector machines for regression, or neural networks for complex relationships within the data.

[0079] For training, a number of prior connector designs can be used. Once the model has been trained, it needs to be validated and tested against a separate dataset not used during training to ensure that the model generalizes well to new, unseen data. Another number of prior connector designs can be used to examine the training results of ML-guided parameter optimization.

[0080] The model can be refined by tuning its hyperparameters (settings that manage the model's architecture and learning process). Hyperparameter tuning is essential for finding the ideal configuration that provides the best predictive accuracy.

[0081] Then, the identification of the most relevant features (design parameters) that significantly affect performance can occur, for example, by comparing different designs with manual input and assigning ratings to each design. Appropriate feature selection can improve model performance by eliminating irrelevant or redundant data. This can also be accomplished using generative design algorithms.

[0082] For new requirements or specifications, a trained model can predict the optimal set of design parameters. This may involve iteratively running the model while adjusting the input requirements to optimize desired performance metrics (such as minimizing weight while maximizing strength or conductivity).

[0083] In another embodiment, the AI ​​generator is configured to provide the manufacturing station with a dataset of 2D and / or 3D models of optimized information. The manufacturing station can then be configured to generate corresponding custom connectors. The manufacturing station may include at least lathes and / or milling machines and / or injection molding machines and / or painting units.

[0084] The output data for connector optimization parameters can include a comprehensive set of specifications defining the connector's physical and electrical characteristics. This set can include dimensional information such as overall length, diameter, pin pitch, and contact depth; material selection for conductor, insulator, and housing components; electrical specifications including rated voltage, current capacity, and impedance; environmental ratings such as IP protection ratings; and mechanical properties such as mating force and durability cycle counts. In some cases, the output can specify standard plug and socket configurations (if applicable) or provide custom geometries for proprietary designs. The dataset can also include tolerances for critical dimensions, surface finish requirements, and any special features such as keying or locking mechanisms. Depending on the application, additional parameters such as operating temperature range, chemical resistance, or shielding effectiveness may be included. The specific dimensions and values ​​of these parameters will vary based on the particular connector design and intended use, ranging from sub-millimeter precision for micro-connectors to centimeter precision for high-power industrial connectors.

[0085] Physics-based simulations play a crucial role in validating the optimized features of custom connectors to meet user requirements. These simulations provide detailed insights into connector performance under various conditions, allowing for comprehensive analysis even before physical prototyping. For electrical connectors, finite element analysis (FEA) can be used to simulate electrical, thermal, and mechanical properties. In the realm of electrical performance, FEA can be used to calculate voltage distribution, current density, and electric field strength within the connector. This analysis helps verify that clearances and creepage distances meet safety standards, which is particularly important for high-voltage applications. Simulations can model the connector's geometry, material properties, and boundary conditions to predict potential areas of electrical stress concentration or breakdown. Thermal simulations using FEA techniques can be applied to evaluate heat dissipation characteristics. These simulations can model the heat generated by current and predict the temperature distribution across the connector under various operating conditions. This analysis can be particularly valuable for high-current applications where thermal management is critical. Mechanical simulations using FEA can evaluate the structural integrity of the connector under various loads, including insertion and withdrawal forces, vibration, and shock. These simulations help predict stress concentrations, deformation, and potential failure points, ensuring that the connector meets durability and reliability requirements. In addition, multiphysics simulations that couple electrical, thermal, and mechanical analyses, taking into account phenomena such as thermal expansion and its effect on electrical contact resistance, can provide a more comprehensive understanding of connector behavior.

[0086] Results from these physics-based simulations can be used to iteratively improve the optimized features of a connector design. This iterative process can involve feedback loops, where simulation results inform design modifications, which are then resimulated to assess their impact. For example, if a thermal simulation indicates localized hot spots, the design can be adjusted to enhance heat dissipation in those areas, perhaps by modifying the connector's geometry or material composition. The modified design can then be resimulated to validate the improvements. Similarly, if an electrical simulation reveals regions of high electric field strength, the design can be refined to redistribute the electric field more uniformly. This iterative approach can also involve sensitivity analysis, where key design parameters are systematically changed to understand their impact on performance metrics. Machine learning techniques can be integrated into this process to more efficiently guide the exploration of the design space. For example, surrogate models or meta-models trained on simulation data can quickly predict the performance of new design changes, allowing for faster iteration cycles. As the design evolves, increasingly detailed and computationally intensive simulations can be employed to fine-tune specific aspects of the connector. This iterative simulation-driven optimization process can continue until the design meets or exceeds all specified requirements, thus balancing various performance criteria and manufacturing constraints. Then, before proceeding to physical prototyping and testing, the final optimized design can be validated through a comprehensive set of simulations.

[0087] Provide optimization parameters in a readable and understandable format for engineers, stakeholders, or generative design algorithms.

[0088] One example scenario where ML-guided parameter optimization can be used instead of lookup tables might be when designing custom connectors, where ideal specifications (such as size, material, pin count, electrical ratings, etc.) need to be balanced against a set of constraints (such as cost, weight, environmental factors, etc.). The ML model will process the requirements and, by identifying patterns in the data, propose an optimal design that meets the desired criteria.

[0089] This approach is more flexible and scalable than lookup tables, especially for complex optimization tasks involving many interactive factors. ML models can handle nonlinear relationships between parameters and outcomes, and they can adapt to new data as design and technology evolve, while lookup tables are static and may not capture the complexity or interdependencies between design variables. However, the effectiveness of ML-guided optimization depends heavily on the quality and quantity of the training data and the appropriateness of the chosen ML method for the specific engineering problem at hand.

[0090] Generative design, utilizing generative design algorithms, is a method used in computer-aided design (CAD) that leverages algorithms and computational processes to generate a wide variety of design alternatives based on user-defined goals and constraints. Optimization algorithms, often including machine learning techniques, can be used to explore possible design spaces and create optimized 2D or 3D models, or 2D and / or 3D datasets tailored to specification sets and physical requirements.

[0091] Generative design algorithms can be or include iterative processes and / or steps. After providing an initial 2D and / or 3D dataset, the 2D and / or 3D dataset can be evaluated through evaluation steps. Optimizing the generative design algorithm can also be configured to iteratively improve the output of the 2D and / or 3D dataset internally.

[0092] Iterative steps may include evaluating certain thresholds and / or specifications and returning the information to the generative design algorithm to further refine the first 2D and / or 3D dataset or before outputting the first 2D and / or 3D dataset.

[0093] The generation process begins with specifications and physical requirements that describe the desired characteristics of the final design. These may include constraints such as material properties, strength, weight, cost, manufacturing methods, environmental conditions, and any specific design variables to be considered.

[0094] Generative design algorithms can use input data to explore the possible design space. They typically generate numerous iterations of potential designs by changing defined design variables within given constraints. This iterative process is fast and can explore far more options than human designers would typically.

[0095] Simulations can be used to evaluate each design to determine how well it meets specified objectives. Algorithms that can be used include finite element analysis (FEA) for evaluating strength, durability, and / or electrical properties; computational fluid dynamics (CFD) for evaluating airflow or fluid movement; thermal analysis for checking heat resistance; and other parameters.

[0096] Algorithms can apply optimization techniques to refine the design. They can learn from each simulation, identifying which parameters lead to better results and adjusting accordingly. Effective generative designs can also employ evolutionary algorithms or other machine learning methods, which mimic natural selection by preserving "successful" designs and iterating over them.

[0097] It can present users with a set of best-performing designs, often showing multiple solutions that balance the trade-offs between different standards and physical requirements in different ways.

[0098] Users can view the generated designs and select the final design, adjust input parameters and constraints to refine the results, or even combine elements from different designs to create hybrid solutions.

[0099] Once the user is satisfied with the results, the selected design can be further detailed and prepared for prototyping or production, including generating the necessary 2D and / or 3D model datasets.

[0100] In one embodiment, the physical requirements include at least one of the following parameters: • This includes connector types that are standard or specifically designed, such as BNC, SMA, N-type for RF connections, or USB, HDMI, and RJ45 for data and audio / video applications; • Gender, where connectors can typically be described as male (pins) or female (sockets), indicating how they mate with the corresponding components.

[0101] • Contact materials, including those used for the conductive parts of the connector, may be gold-plated, silver-plated, or nickel-plated to ensure good conductivity and corrosion resistance.

[0102] • Insulating material, which can be a non-conductive part that provides necessary electrical isolation between pins or between the connector and the housing. Materials typically include plastics such as PTFE (polytetrafluoroethylene) or durable polymers.

[0103] Size and dimensions can include physical dimensions, including diameter, length, and profile, which are crucial for ensuring that the connector fits into the corresponding device or port; for RF connectors, this can also include the interface dimensions defined by standards.

[0104] • The number of contacts may include the number of individual conductive elements within the connector that can be configured to carry signals or electricity.

[0105] • Rated voltage and current may include the maximum voltage and current that the connector can safely handle.

[0106] Impedance (for RF connectors) can include the characteristic impedance of the RF connector, which can be 50 or 75 ohms for most applications and is crucial for maintaining signal integrity and minimizing reflections.

[0107] • Frequency range (for RF connectors) can include the frequency range on which the connector performs well, resulting in minimal loss and distortion.

[0108] Durability can include the number of mating cycles a connector can withstand while still maintaining its mechanical and electrical integrity.

[0109] • Environmental resistance can include the connector’s ability to withstand environmental conditions such as extreme temperatures, humidity, UV exposure, and chemical exposure.

[0110] • Cable attachment mechanisms and strain relief can include how connectors attach to and support cables to prevent damage caused by bending or pulling.

[0111] • Interface standards may include multimedia connectors that may comply with standards such as HDMI, DisplayPort, or USB-C, which can carry audio, video, and data simultaneously.

[0112] • Bandwidth can include the maximum rate at which data can be transmitted for connectors carrying high-definition video or high-speed data.

[0113] • Audio / video channel support can include the number and type of supported audio and video channels (e.g., an HDMI connector supports multiple audio channels and 4K video).

[0114] • Data transfer rate can include the maximum speed at which data associated with the USB or Thunderbolt connector can be transferred through the connector.

[0115] • Support for power delivery may include connectors (such as USB-C) that can also deliver power, so the rated voltage and current for power delivery are important aspects.

[0116] • Digital rights management (DRM) may include some connectors for transmitting protected content, and DRM support may be necessary (e.g., HDMI connectors support HDCP for protected content transmission).

[0117] • Signal conversion or adaptation may include some multimedia connectors containing active or passive components to convert from one signal type to another or adapt impedance.

[0118] • Physical configuration can include whether the connector is straight, angled, or rotatable, which can be important for cable management and equipment placement.

[0119] • The locking mechanism can be integrated with the multimedia connector to ensure a secure connection that will not be accidentally disconnected.

[0120] • Color coding may be used by multimedia connectors to help identify the correct port for connection, especially in complex setups.

[0121] • EMI / RFI shielding can include the shielding necessary to prevent electromagnetic interference (EMI) and radio frequency interference (RFI) that may degrade signal quality.

[0122] In one embodiment, generative design algorithms use derivation functions to enhance their ability to quickly converge to an optimal or near-optimal solution from the space of all possible designs.

[0123] Accelerating generative design algorithms can involve enhancing their ability to converge quickly to an optimal or near-optimal solution from a space of all possible designs. One way to achieve this is by using a derivation function, which, in the context of the optimization algorithm, refers to the mathematical process of finding the gradient or derivative of a function. This approach largely involves gradient-based optimization methods.

[0124] The derivation function (also known as the gradient-based method) calculates the gradient (partial derivative) of the objective function with respect to the design variables. The objective function specifies what you are trying to optimize (e.g., minimizing the weights, maximizing the strength).

[0125] The gradient can represent the direction of the steepest increase in the objective function. For optimization, if you are minimizing, you are interested in the opposite direction (the steepest decrease), or if you are maximizing, you are interested in the direction of the current state.

[0126] The algorithm can then use this gradient information to update the design variables, thereby pushing the design toward an improved solution. This update may involve a simple step in the opposite direction of the gradient (in the case of minimization), or it may require more sophisticated techniques (such as row search or trust region methods) to select the appropriate step size.

[0127] By leveraging gradient information, generative design algorithms can take informed and efficient steps toward optimization, rather than employing random or exhaustive search methods. However, the direct use of derived functions assumes that the objective function is differentiable with respect to the design variables. This makes methods using derived functions more efficient.

[0128] If the derivation function is applied effectively, it can significantly accelerate the convergence of the algorithm by more intelligently guiding the search process through the design space.

[0129] Sometimes, derivative-free methods such as evolutionary algorithms can be combined with gradient-based methods, leveraging both robustness to landscape features and efficient gradient-guided refinement.

[0130] The present invention also provides a product data file generated using an AI generator according to at least one of the foregoing embodiments.

[0131] The present invention also provides a computer program product that can be loaded into the computer memory of a device according to at least one of the AI ​​generators in the foregoing embodiments.

[0132] The present invention also provides a computer system configured to execute an AI generator according to at least one of the foregoing embodiments.

[0133] The present invention also provides a method for producing it, comprising the following steps: • Perform a method according to one of the foregoing embodiments; • Generate the connector based on the connector model generated in step 1.

[0134] In another embodiment, a finite element model validity check is applied to 2D and / or 3D custom connector models for a plausibility check.

[0135] In one embodiment, if the 2D and / or 3D custom connector model conforms to the set of mechanical specifications, the finite element model validity check outputs check information.

[0136] This invention is not limited to the described embodiments. Within the scope of this invention, unless otherwise indicated, all described and / or drawn features may be combined with each other as desired.

[0137] The invention is described below by way of example only with reference to the accompanying drawings. It shows: · Figure 1 This is a method of using an AI generator to generate custom connectors.

[0138] Figure 1 A method for generating custom connectors using an AI generator is disclosed. The method includes the following steps: Providing some input information to an LLM (S1), wherein the input information includes human natural language information. Processing the input information into a computer-interpretable set of specifications using the LLM (S2). Aligning the set of specifications to physical requirements (S3).

[0139] Generative design algorithms are used to generate optimized 2D and / or 3D model datasets containing specification sets and physical requirements (S4). The output is a 2D and / or 3D dataset provided by a custom connector model (S5).

[0140] Reference symbol Steps S1-S5

Claims

1. A method for generating custom connectors using an AI generator, Includes the following steps: - Provide some input information to the LLM, including human natural language information (S1). - The input information is processed into a set of specifications that can be interpreted by a computer using the LLM (S2); - Align the set of specifications with the physical requirements (S3); - Use generative design algorithms to generate optimized 2D and / or 3D model datasets containing specification sets and physical requirements (S4); - Output 2D and / or 3D datasets of the custom connector model (S5).

2. The method for generating custom connectors using an AI generator according to claim 1, The LLM is configured to validate the input information based on at least one of the following parameters: • Range of values; • Threshold; • Symbols; • contradiction.

3. The method for generating custom connectors using an AI generator according to at least one of the preceding claims, in The AI ​​generator is configured to execute locally or outsource the task to the cloud if the capacity utilization exceeds a certain threshold, wherein the threshold considers at least one of the following metrics: percentage of total available CPU utilization, GPU memory consumption, RAM usage, number of floating-point operations per second, and total duration of the task.

4. A method for generating custom connectors using an AI generator according to at least one of the preceding claims, comprising the following steps: - The LLM outputs processed information from the human natural language information that includes the set of specifications to the user.

5. A method for generating custom connectors using an AI generator according to at least one of the preceding claims, comprising the following steps: - Align the physical requirements by a lookup table, wherein alignment includes using the lookup table to map the set of specifications toward the physical requirements.

6. A method for generating a custom connector using an AI generator according to at least one of the preceding claims, comprising the following steps: - Alignment with the physical requirements through ML-guided parameter optimization, wherein the ML-guided parameter optimization utilizes a training dataset including various design parameters and their corresponding performance metrics, so that the trained model can predict the optimal set of design parameters.

7. A method for generating custom connectors using an AI generator according to at least one of the preceding claims, comprising the following steps: - Provide the optimized 2D and / or 3D model dataset to the manufacturing station.

8. The method for generating custom connectors using an AI generator according to at least one of the preceding claims, The physical requirements mentioned above include at least one of the following parameters: • Connector type, gender, contact material, insulation material, size and dimensions, number of contacts, rated voltage and current, impedance, frequency range, durability, environmental resistance, cable attachment mechanism and strain relief, interface standard, bandwidth, audio / video channel support, data transmission rate, power delivery support, digital rights management (DRM), signal conversion or adaptation, physical configuration, locking mechanism, color coding, EMI / RFI shielding.

9. A method for generating custom connectors using an AI generator according to at least one of the preceding claims, wherein... The generative design algorithm uses derivation functions to enhance its ability to quickly converge to the optimal or near-optimal solution from the space of all possible designs.

10. A product data file generated using an AI generator according to at least one of claims 1 to 9.

11. A computer program product which can be loaded into the computer memory of a device according to at least one of claims 1 to 9 of an AI generator.

12. A computer system configured to execute an AI generator according to any one of claims 1 to 9.

13. A method for producing a custom connector, comprising the following steps: Perform the method according to at least one of claims 1 to 9; The connector is generated based on the connector model created in step 1.

14. The method for producing a custom connector according to claim 13, wherein... The finite element model validity check is applied to the 2D and / or 3D custom connector model for a rationality check.

15. The method for producing a custom connector according to claim 14, wherein... If the 2D and / or 3D custom connector model conforms to the set of mechanical specifications, the finite element model validity check outputs check information.