Light-weight natural interaction control method and system for industrial simulation

Abstract

The invention discloses a lightweight natural interaction control method and system for industrial simulation, and belongs to the technical field of industrial simulation. According to an existing industrial simulation technology, the operation barrier of mobile equipment is high, the interaction mode threshold is high, and the simulation operation requirement of a user in the moving process cannot be met. According to the lightweight natural interaction control method for industrial simulation, the simulation capability registration module, the lightweight interaction unit and the simulation instruction analysis unit are constructed, so that a user does not need to spend a lot of time to learn industrial simulation operation logic and directly describe simulation requirements in an oral manner, and an industrial simulation result can be obtained; therefore, the problems that an existing industrial simulation technology is high in operation barrier and low in interaction efficiency can be effectively solved, the technical threshold for operating professional simulation software on mobile equipment is effectively lowered, the simulation technology can be used in a mobile scene, the simulation operation requirement of a user in the moving process can be met, and the user experience is improved. And the simulation technology can be popularized conveniently.

Description

Technical Field

[0001] This invention relates to a lightweight natural interactive control method and system for industrial simulation, belonging to the field of industrial simulation technology. Background Technology

[0002] Traditional industrial simulation software (such as Ansys) has long relied on high-performance computers for local operation. Users need to operate it in a fixed location equipped with specialized hardware and master complex graphical interface interaction skills, such as parameter configuration, model building, and result analysis. Even though some simulation tools have moved to the cloud, their user interfaces are still limited to desktop computers or web pages, continuing the traditional "mouse and keyboard + graphical interface" interaction mode to complete simulation tasks.

[0003] From the perspective of mobile device applications, this industrial simulation mode, which relies on complex graphical interactions, has several significant limitations. Firstly, the small screen size of mobile devices makes it difficult to fully display the complex functions of traditional industrial simulation interfaces. Traditional industrial simulation interfaces contain numerous function buttons, parameter setting areas, and graphical display areas; on the small screen of a mobile device, the layout of this content is difficult, making it impossible for users to clearly and comprehensively view and operate it. Secondly, in mobile scenarios such as outdoors or during commutes, users find it difficult to accurately complete tasks such as model adjustments and parameter settings through touch operations. Touch operations themselves have low precision, and the unstable operating environment during movement further affects operational accuracy, leading to low simulation efficiency.

[0004] Therefore, the existing industrial simulation technology has the following problems: 1. High barriers to operation on mobile devices: Existing simulation tools are not optimized for interaction with lightweight devices such as mobile phones and tablets. The operating mode and interface layout of professional computers are directly applied to mobile devices, making it impossible for users to complete simulation tasks on mobile devices without a professional computer or complex interface. This greatly limits the use of simulation technology in mobile scenarios and fails to meet the simulation operation needs of users during mobile operations.

[0005] 2. High barrier to entry in interaction methods: Traditional simulation operation modes rely on graphical interfaces and manual operation, requiring users to spend a significant amount of time learning the operational logic. For non-professional users, mastering this complex operational logic is quite difficult, which not only increases the learning cost but also limits the popularization and application scope of simulation technology.

[0006] The information disclosed in this background section is only for understanding the background of the inventive concept, and therefore may include information that does not constitute prior art. Summary of the Invention

[0007] To address the aforementioned problems, the first objective of this invention is to provide a lightweight, natural interactive control method and system for industrial simulation. By constructing a simulation capability registration module, a lightweight interaction unit, and a simulation instruction parsing unit, users can obtain industrial simulation results simply by verbally describing their simulation needs without spending a significant amount of time learning the operational logic of industrial simulation. This effectively solves the problems of high operational barriers and low interaction efficiency in existing industrial simulation technologies, significantly lowers the technical threshold for operating professional simulation software on mobile devices, enables the use of simulation technology in mobile scenarios, meets the simulation operation needs of users during mobile operations, facilitates the popularization of simulation technology, and expands the application scope of industrial simulation.

[0008] To address the aforementioned problems, the second objective of this invention is to provide another lightweight, natural interactive control method and system for industrial simulation. This method enables the mobile application of simulation technology. Users can directly input colloquial descriptions using a lightweight interactive unit to generate professional simulation target commands, thereby achieving mobile simulation. Users can perform simulation operations anytime, anywhere, breaking the dependence of traditional simulation on fixed locations and specialized hardware, and greatly expanding the application scenarios of simulation technology. Simultaneously, it lowers the barrier to entry for simulation technology, allowing more non-professional users to easily use simulation tools, promoting the popularization and application of simulation technology.

[0009] To achieve one of the above objectives, the first technical solution of the present invention is as follows: A lightweight, natural interactive control method for industrial simulation includes the following: By using a pre-built simulation capability registration module, the executable operations of the target simulation software are obtained, and the executable operations are vectorized to establish a simulation method vector library. Using pre-built lightweight interactive units, natural speech information or natural text commands from users regarding their industrial simulation needs are collected, and the natural speech information or natural text commands are standardized to obtain simulation command data. A pre-built simulation instruction parsing unit is used to perform structured processing on the simulation instruction data, and simulation target instructions are generated based on the simulation method vector library. The simulation target command is input into the target simulation software to execute the user's industrial simulation requirements and obtain industrial simulation results, thereby realizing lightweight natural interactive control based on industrial simulation.

[0010] This invention acquires the executable operations of the target simulation software by constructing a simulation capability registration module, a lightweight interaction unit, and a simulation instruction parsing unit. These executable operations are then vectorized to establish a simulation method vector library. Simultaneously, the user's natural requirement information is standardized to obtain simulation instruction data. This data is then structured and, based on the simulation method vector library, simulation target instructions are generated. These target instructions are then input into the target simulation software to execute the user's industrial simulation requirements, yielding industrial simulation results. This allows simulation technology to be directly applied to mobile devices, enabling its use in mobile scenarios and meeting users' simulation operation needs during movement. Furthermore, users do not need to spend significant time learning industrial simulation operation logic; they can directly describe their simulation requirements in conversational language, facilitating the popularization of simulation technology and expanding the application scope of industrial simulation.

[0011] Furthermore, this invention can effectively solve the problems of high operational barriers and low interaction efficiency in existing mobile devices due to the complexity of simulation operation logic, screen and input limitations. This effectively reduces the technical threshold for operating professional simulation software on mobile devices, allowing users to submit simulation tasks and obtain results without relying on complex graphical interfaces, thus promoting the mobile and lightweight application of simulation technology.

[0012] Furthermore, this invention enables the mobile application of simulation technology. Users can directly input colloquial descriptions using a lightweight interactive unit to generate professional simulation target commands, thus achieving mobile simulation. This allows users to perform simulation operations anytime, anywhere, breaking the dependence of traditional simulation on fixed locations and specialized hardware, and greatly expanding the application scenarios of simulation technology. At the same time, it lowers the barrier to entry for simulation technology, enabling more non-professional users to easily use simulation tools, promoting the popularization and application of simulation technology.

[0013] As a preferred technical measure: The method for establishing a simulation method vector library by acquiring the executable operations of the target simulation software through a pre-built simulation capability registration module and vectorizing these executable operations is as follows: Step 11: Determine the target simulation software to be adapted and sort out the executable operations of the target simulation software, including but not limited to: model parameter setting, mesh generation, static analysis, simulation task start, result data query, stress calculation and temperature calculation; Step 12: Based on the target simulation software and executable operations, establish a proprietary vocabulary library for the simulation field, which includes at least megapascals, revolutions per minute, and rendering data formats; Step 13: Based on the proprietary vocabulary of the simulation field and each executable operation, generate several structured description files. Each structured description file corresponds to an executable operation, which includes the method name, function description, parameter information and execution content. Step 14: Through a vectorization processing mechanism, each structured description file is converted into a simulation method vector that can be recognized by the target simulation software, and a simulation method vector library is built for instruction matching.

[0014] As a preferred technical measure: Step 14: Through a vectorization mechanism, the structured description file is converted into a vector form that can be recognized by the target simulation software. The method for constructing the simulation method vector library is as follows: Extract the method name, function description, parameter description, and execution content from the structured description file, and generate a text file with four fields; A domain-pretrained word vector model is used to vectorize text with four fields, resulting in basic vectors for each field, including method name vector, function description vector, parameter description vector, and execution content vector. The mean aggregation algorithm is used to process several parameter description vectors to calculate the overall parameter vector; Assign preset weights to each field based on its importance in operation recognition; Based on preset weights, the method name vector, function description vector, overall parameter vector, and execution content vector are weighted and concatenated to generate a simulation method vector corresponding to a certain executable operation. All executable simulation method vectors are stored uniformly, and a vector index is created to obtain a simulation method vector library.

[0015] As a preferred technical measure: The method for collecting user natural speech information or natural text commands regarding industrial simulation needs using pre-built lightweight interactive units, and then standardizing the natural speech information or natural text commands to obtain simulation command data is as follows: Step 21: Collect natural speech information or natural text commands from users regarding their industrial simulation needs through a lightweight interactive unit; Step 22: To address the professional terminology in the industrial simulation field, establish a dedicated vocabulary database for the simulation field, which includes at least simulation operation terms, unit terms, and data format terms, and input it into the speech recognition module as a custom hot word list. When the user outputs natural speech information, the speech recognition module matches professional terms through a custom hot word list, recognizes colloquial expressions in the natural speech information, obtains natural text instructions, and then executes step 23. If the user outputs a natural text instruction, then proceed to step 23; Step 23: Standardize the natural text instructions, extract core information, and obtain simulation instructions; Step 24: The standardized simulation instructions are encapsulated into a structured, lightweight data format to form simulation instruction data in a text-based open-source data exchange format.

[0016] As a preferred technical measure: Step 23: Standardize the natural text instructions, extract core information, and obtain the simulation instructions using the following method: Redundant information is removed from the natural text instructions to obtain the first-stage instructions; The unit format in the first-stage instructions is standardized and converted into standard units to obtain the second-stage instructions. The capitalization of the second-stage instructions is standardized to obtain the third-stage instructions; Extract core information from the third-stage instructions to obtain key elements of the instructions, including the target object, operation parameters, and core operations. The target object, operation parameters, and core operations are summarized to obtain simulation instructions.

[0017] As a preferred technical measure: The method for generating simulation target instructions by using a pre-built simulation instruction parsing unit to perform structured processing on simulation instruction data and based on the simulation method vector library is as follows: Step 31: Deploy the simulation instruction parsing unit on the cloud server; Step 32: The simulation instruction parsing unit receives simulation instruction data through a long connection protocol and extracts the core instructions from the simulation instruction data as the original instruction text. Step 33: Construct a prompt word template to convert the original instruction text into a structured instruction consistent with the structured description file format, ensuring field alignment; Step 34: Through a vectorization processing mechanism, the structured instructions are converted into a method name vector, a function description vector, a parameter information vector, and an execution content vector; then, the mean aggregation algorithm is used to process several parameter description vectors to calculate the overall parameter vector. The method name vector, function description vector, parameter vector, and execution content vector are weighted and concatenated according to preset weights to calculate the user instruction vector. Step 35: Using the cosine similarity algorithm, the user instruction vector is matched with the simulation method vector in the simulation method vector library to obtain the optimal executable operation; Step 36: Bind the parameter values ​​in the user instruction vector to the optimal executable operation, supplement the application programming interface information of the target simulation software, and generate simulation target instructions that can be executed by the target simulation software.

[0018] As a preferred technical measure: Step 33, constructing prompt word templates, the method for converting the original instruction text into structured instructions consistent with the structured description file format is as follows: Step 331: Based on the original instruction text, set up the simulation role; The simulation role is an industrial simulation instruction parsing engine, whose task is to convert natural language instructions into standardized structured data; Step 332: Based on the structured description file, set the output format requirements information. The output format requirements information is to output in a text-based open-source data exchange format, and its fields completely match the four core fields of the structured description file. The four core fields include method name, function description, parameter information, and execution content. Step 333: Establish parsing rules based on the simulation role and output format requirements; The parsing rules include prioritizing the matching of domain-specific terms, extracting parameter values, and preserving the original terminology. Step 334: Based on the simulation role, output format requirements, and parsing rules, establish a prompt word template; Step 335: Input the original instruction text into the prompt word template to generate a structured instruction that is consistent with the structured description file format.

[0019] As a preferred technical measure: The method for inputting simulation target commands into the target simulation software to execute the user's industrial simulation requirements and obtain industrial simulation results is as follows: Step 41: Adapt the interface protocol between the simulation target instruction and the target simulation software, and at the same time, eliminate invalid simulation target instructions through verification and generate a simulation instruction request packet. Step 42: Based on the network transmission protocol, transmit the simulation instruction request packet to the target simulation software and perform end-to-end encryption on the simulation instruction request packet; Step 43: The target simulation software schedules resources according to the priority of the simulation instruction request packet, executes the user's industrial simulation requirements, and collects process information in real time. Step 44: The target simulation software collects simulation result data, encapsulates it in a standard format, and sends it back to the lightweight interactive unit so that the industrial simulation results can be displayed.

[0020] As a preferred technical measure: Step 41 involves adapting the interface protocol between the simulation target instructions and the target simulation software, and simultaneously filtering out invalid simulation target instructions through verification. The method for generating the simulation instruction request packet is as follows: Step 411: The simulation instruction parsing unit retrieves the interface protocol specification of the target simulation software. The interface protocol specification is either an application programming interface protocol or a script protocol. Step 412: According to the interface protocol specification, convert the generated simulation target instructions into a format that can be recognized by the target simulation software, which includes the following: If the target simulation software's interface protocol specification is an application programming interface: The target method and binding parameters in the simulation target instruction are mapped to application programming interface request parameters, and the interface address, request method, and authentication token are supplemented to generate a simulation instruction request packet that conforms to the network transmission protocol. If the target simulation software's interface protocol specification is script execution: The target simulation instructions are converted into script statements, and a simulation instruction request packet is generated, so that the parameter assignment is consistent with the software script syntax. Step 413: Based on the authentication token, verify the user's operation permissions to see if they have the execution permission for the simulation control. If the user has the execution permission, proceed to step 414; if the user does not have the execution permission, use "no permission" as the simulation result data, and then proceed to step 44. Step 414: Check whether the computing resources of the target simulation software meet the task requirements. If the computing resources meet the task requirements, proceed to step 415; if the computing resources do not meet the task requirements, use the insufficient resources as the simulation result data, and then proceed to step 44. Step 415: According to the structured description file, the simulation instruction request packet is validated to check whether all required parameters have been bound to valid values. If all required parameters have been bound to valid values, proceed to step 42; if any required parameter has not been bound to a valid value, the missing parameter valid value is used as the simulation result data, and then proceed to step 44.

[0021] To achieve one of the above objectives, the second technical solution of the present invention is as follows: A lightweight natural interactive control system for industrial simulation, including a cloud server and a mobile device terminal; The cloud server is equipped with a simulation capability registration module and a simulation instruction parsing unit; The simulation capability registration module is used to obtain the executable operations of the target simulation software, and to vectorize the executable operations to establish a simulation method vector library. The simulation instruction parsing unit is used to perform structured processing on simulation instruction data and generate simulation target instructions based on the simulation method vector library; The mobile device features a lightweight interaction unit; The lightweight interactive unit is used to collect users' natural speech information or natural text commands regarding industrial simulation needs, and to standardize the natural speech information or natural text commands to obtain simulation command data; and to display the industrial simulation results generated by the target simulation software after executing the simulation target commands. Mobile devices and cloud servers communicate via standardized network protocols, forming a remotely decoupled architecture that enables lightweight, natural interactive control for industrial simulation.

[0022] Compared with existing technical solutions, the present invention has the following beneficial effects: This invention acquires the executable operations of the target simulation software by constructing a simulation capability registration module, a lightweight interaction unit, and a simulation instruction parsing unit. These executable operations are then vectorized to establish a simulation method vector library. Simultaneously, the user's natural requirement information is standardized to obtain simulation instruction data. This data is then structured and, based on the simulation method vector library, simulation target instructions are generated. These target instructions are then input into the target simulation software to execute the user's industrial simulation requirements, yielding industrial simulation results. This allows simulation technology to be directly applied to mobile devices, enabling its use in mobile scenarios and meeting users' simulation operation needs during movement. Furthermore, users do not need to spend significant time learning industrial simulation operation logic; they can directly describe their simulation requirements in conversational language, facilitating the popularization of simulation technology and expanding the application scope of industrial simulation.

[0023] Furthermore, this invention can effectively solve the problems of high operational barriers and low interaction efficiency in existing mobile devices due to the complexity of simulation operation logic, screen and input limitations. This effectively reduces the technical threshold for operating professional simulation software on mobile devices, allowing users to submit simulation tasks and obtain results without relying on complex graphical interfaces, thus promoting the mobile and lightweight application of simulation technology.

[0024] Furthermore, this invention enables the mobile application of simulation technology. Users can directly input colloquial descriptions using a lightweight interactive unit to generate professional simulation target commands, thus achieving mobile simulation. This allows users to perform simulation operations anytime, anywhere, breaking the dependence of traditional simulation on fixed locations and specialized hardware, and greatly expanding the application scenarios of simulation technology. At the same time, it lowers the barrier to entry for simulation technology, enabling more non-professional users to easily use simulation tools, promoting the popularization and application of simulation technology. Attached Figure Description

[0025] Figure 1 This is a flowchart of a lightweight natural interactive control method for industrial simulation according to the present invention. Figure 2 This is a framework diagram of a lightweight natural interactive control system for industrial simulation according to the present invention. Figure 3 This is a schematic diagram of the matching process in the language processing and simulation method of this invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application. This invention covers any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of the invention as defined by the claims.

[0027] like Figure 1 As shown, this is the first specific embodiment of the lightweight natural interactive control method for industrial simulation of the present invention: A lightweight, natural interactive control method for industrial simulation includes the following: By using a pre-built simulation capability registration module, the executable operations of the target simulation software are obtained, and the executable operations are vectorized to establish a simulation method vector library. Using pre-built lightweight interactive units, natural speech information or natural text commands from users regarding their industrial simulation needs are collected, and the natural speech information or natural text commands are standardized to obtain simulation command data. A pre-built simulation instruction parsing unit is used to perform structured processing on the simulation instruction data, and simulation target instructions are generated based on the simulation method vector library. The simulation target command is input into the target simulation software to execute the user's industrial simulation requirements and obtain industrial simulation results, thereby realizing lightweight natural interactive control based on industrial simulation.

[0028] A first specific embodiment of the lightweight natural interactive control system for industrial simulation of the present invention: A lightweight natural interactive control system for industrial simulation, including a cloud server and a mobile device terminal; The cloud server is equipped with a simulation capability registration module and a simulation instruction parsing unit; The simulation capability registration module is used to obtain the executable operations of the target simulation software, and to vectorize the executable operations to establish a simulation method vector library. The simulation instruction parsing unit is used to perform structured processing on simulation instruction data and generate simulation target instructions based on the simulation method vector library; The mobile device features a lightweight interaction unit; The lightweight interactive unit is used to collect users' natural speech information or natural text commands regarding industrial simulation needs, and to standardize the natural speech information or natural text commands to obtain simulation command data; and to display the industrial simulation results generated by the target simulation software after executing the simulation target commands. Mobile devices and cloud servers communicate via standardized network protocols, forming a remotely decoupled architecture that enables lightweight, natural interactive control for industrial simulation.

[0029] like Figure 2 As shown, this is the second specific embodiment of the lightweight natural interactive control system for industrial simulation of the present invention: A lightweight natural interactive control system for industrial simulation includes a three-layer decoupled architecture comprising a lightweight interactive unit, a simulation instruction parsing unit, and a target simulation software terminal. Each layer communicates through a standardized network protocol, such as the full-duplex communication protocol WebSocket.

[0030] The lightweight interaction unit serves as the user interaction entry point. Its core design principle is lightweight processing. It is responsible for command acquisition, preprocessing, and result display, without undertaking complex calculations. It includes an interaction and preprocessing module, a lightweight command encapsulation module, and a communication module. Furthermore, the lightweight interaction unit can be obtained by loading a mini-program or software developed based on this invention onto existing mobile devices such as smartphones or tablets.

[0031] In this embodiment, the interaction and preprocessing module is used to receive and preliminarily process the user's natural interaction commands (voice / text), solving the problem of low accuracy in recognizing technical terms in the simulation field. Its processing flow is as follows: Step 1: Collect commands, supporting microphone input (voice) or keyboard input (text). Voice commands will preferentially call the third-party speech recognition module SDK (such as iFlytek's speech recognition module). Step 2: Enhance the domain vocabulary and set up a "simulation domain-specific vocabulary library", which includes at least terms such as Meshing, Static Structural, and MPa. Call the speech recognition module SDK to load this vocabulary library as a custom hot word list to improve the accuracy of professional command recognition. Step 3 involves standardizing the instructions and preprocessing the recognized text, including unifying capitalization, removing filler words like "that thing" and "troublesome," and extracting core instructions. For example, "That thing, run this model for me and see how much stress there is" is standardized to "Run the model and check the stress."

[0032] In this embodiment, the lightweight instruction encapsulation module encapsulates standardized text instructions into a structured, transmittable, lightweight data format, avoiding waste of mobile device computing power. It uses a text-based open-source data exchange format (JSON) to encapsulate the instructions, including key fields such as a unique instruction identifier, user information, and the original instruction. The unique instruction identifier is used for tracking; the user information is the user identifier. The standardized core instructions include running the model and / or viewing stress.

[0033] In this embodiment, the communication module is used to achieve bidirectional data transmission with the cloud server, ensuring the real-time performance and reliability of instructions and results. The method is as follows: Step 1. Send the command by uploading the encapsulated JSON command to the specified interface in the cloud using the full-duplex communication protocol WebSocket.

[0034] Step 2. Transmit interactive information by establishing a long connection through the full-duplex communication protocol WebSocket to handle subsequent process information after the command is sent.

[0035] Step 3. While transmitting the interaction process data, to facilitate user observation, it typically includes rendering data of model points, lines, and surfaces, as well as user commands. The rendering data of model points, lines, and surfaces is transmitted from the server to the lightweight interaction unit in the form of a ZIP compressed package. User commands are encapsulated by the interaction and preprocessing module and the lightweight command encapsulation module, and then sent back to the server.

[0036] In this embodiment, the display module presents the entire simulation process information using a simplified interface adapted to mobile screens. Specific display content includes a task list, progress visualization, and result presentation. The task list categorizes user-submitted simulation tasks into "Pending / In Progress / Completed." Progress visualization displays real-time progress via progress bars and text prompts (such as "60% solution in progress"). Result presentation displays the 3D model, contour plots, and results export after the server sends a compressed package containing the results data.

[0037] The 3D model display unit renders a 3D model by parsing point, line, and surface data from a ZIP archive. It supports user-friendly mobile gestures such as single-finger rotation and two-finger zoom / pan, allowing users to view the model from various angles. The cloud map display unit maps calculation results (such as stress magnitude and temperature distribution) onto the model using different colors, creating a cloud map. The results export unit allows users to easily capture and save the current results for sharing or reporting.

[0038] In this embodiment, the simulation instruction parsing unit serves as the core hub of the system, undertaking three core tasks: simulation capability registration, natural language parsing, and instruction generation and forwarding. It includes a simulation capability registration module, an instruction parsing module, and a data storage module.

[0039] The simulation capability registration module defines a unified functional call specification. By semantically describing the capabilities of registered tools and combining this with runtime parameter binding technology, it enables dynamic scheduling and execution of services. Its implementation method is as follows: Step 1: Use the semantic description registration tool to create a description file for each simulation operation (such as "applying force load"), specifying the operation method name, semantic information, parameter information, etc.

[0040] Step 2: Using parameter binding technology, the parameter values ​​parsed from natural language are "precisely mapped" to the required parameters during the instruction execution phase. For example, if the user instruction is "apply 100 Newtons of force to surface A", the system automatically associates "surface A" with surface number 5 through the part surface ID mapping table, and binds "100" to the attribute value of that part surface.

[0041] In this embodiment, the instruction parsing module extracts the "raw_text" from the instructions sent by the lightweight interaction unit, uses the large model and the prompt word template Prompt to convert the text content into structured instructions, and finally obtains vectorized instructions. The parsing process is as follows: The first step is to convert the text content into structured instructions, which include the following: Obtain the raw text raw_tex and set the parsing role according to the instruction text; in this embodiment, the parsing role is a professional instruction parsing engine, which is responsible for converting natural language instructions into standardized structured data.

[0042] The command parsing engine parses user-input natural language commands into structured data in JSON format for task description. This structured data includes the method name, an overall description of the method's functionality, the name of parameter A, a detailed description of parameter A, the parameter values ​​extracted from the command, and a description of the specific operations performed by the method.

[0043] The parsing rules in this embodiment are as follows: S1. Identify the operation instructions (simulation methods) and match the most suitable method name (operation name) according to the instruction intent. S2. Accurately extract parameter values ​​from the instruction text while preserving the original meaning; S3. Generate accurate technical descriptions based on method functionality; S4. If a specific term is encountered (such as "A side"), the original text is retained, and the mapping is handled by subsequent processes to complete the value mapping.

[0044] For example: The user inputs a natural language command to apply a force of 100 Newtons to surface A. The command is parsed to obtain parsed data; this data includes the force applied to the specified surface, surface A as the identifier, and the force magnitude as 100 Newtons. Based on the parsed data, the current simulation command is obtained: apply a force of 100 Newtons to surface A. This command is processed to obtain a structured description file in JSON format, ensuring structural integrity and that all fields are filled in.

[0045] The second step involves using a weighted feature vector concatenation algorithm to convert each field of the structured description file into a high-dimensional vector (denoted as the method vector). The structured description files include method description files and user-structured instruction vectors.

[0046] In this embodiment, the specific steps of vectorizing the method description file, vectorizing user-structured instructions, and vector matching and association are as follows: S1. Perform vectorized computation on the method description file, which includes the following: Step 1: Vectorize the field-level text. For the "Method Name", "Method Description", "Execution Content" and "Description" fields in "Configuration Parameters" of the structured description file, use a domain-pre-trained word vector model (such as Word2Vec or BERT word vector model trained on simulated corpus) to vectorize the text, thereby obtaining the basic vectors of each field, which include the method name vector, method description vector, execution content vector, and description parameter vector description.

[0047] The method name vector is d represents the vector dimension, such as 300 dimensions; the method description vector is... The execution content vector is The description vector of the i-th parameter is , This represents the total number of parameters.

[0048] Step two: Use the mean aggregation algorithm to aggregate the parameter vectors. Since the "configuration parameter params" field contains descriptions of multiple parameters, the base vectors of all parameters need to be aggregated to obtain the overall vector of the parameter set. The formula for the mean aggregation algorithm is as follows: Where n represents the number of parameters under the params field of the configuration parameters. For example, in the “applyForce” method, n=3. This formula balances the semantic weight of each parameter by taking the average value, preventing a single parameter from having an excessive impact on the overall vector. Let i be the description vector of the i-th parameter.

[0049] Step 3: Perform a weighted concatenation operation on the method vectors. Given the varying importance of different fields in method identification, different weights need to be assigned to each field vector. These weights can be determined through expert annotation and cross-validation in the simulation field, satisfying the following expression: in, Weight for method name, Describe the weights for the method. To implement content weighting, To describe the parameter weights.

[0050] Finally, the method vector is obtained by concatenation. The calculation formula is as follows: in, A vector of method names. Describe the vector for the method. To execute content vectors, The parameter vector.

[0051] The method name is the most direct identification criterion, therefore it has the highest weight; the others serve as descriptive aids to identification, and thus have relatively lower weights. Therefore, the values ​​for each weight can be... , .

[0052] All method vectors , stored in the simulation method vector library, and a vector index that can be quickly retrieved is constructed.

[0053] S2. Perform vectorization calculation on the user's structured instructions. After the user instructions are structurally converted by the instruction parsing unit to obtain JSON format data, the same vectorization algorithm as the "vectorization calculation of the method description file" is used to calculate the user instruction vector (i.e., steps one to three above). This ensures that the vector spaces of the two are consistent, satisfying the similarity calculation conditions.

[0054] S3. Perform vector matching and method association, and calculate the user command vector using the cosine similarity algorithm. , and the simulation method vectors in the simulation method vector library similarity The calculation formula is as follows: in, It is the dot product of two vectors; , Let be the magnitudes of the two vectors, respectively. The range of values ​​for ) is The closer the value is to 1, the higher the semantic matching degree between the two. This represents the total number of simulation method vectors.

[0055] In this embodiment, the instruction forwarding module matches the corresponding operation in the simulation capability registration module based on the intent recognition result, and accurately forwards the instruction that actually needs to be executed to the corresponding target simulation software.

[0056] In this embodiment, the data storage module uses the simulation instruction parsing unit to store the returned simulation data information.

[0057] In this embodiment, the target simulation software can be various existing simulation software (such as Ansys software in the field of industrial simulation, and ABAQUS software widely used in structural mechanics and related fields). These software programs, after passing through the simulation capability registration module, can receive instructions sent by the cloud service area, execute corresponding simulation operations, and return simulation progress and results.

[0058] This invention addresses the high operational barriers and low interaction efficiency inherent in existing mobile devices due to screen and input limitations by designing a system architecture that decouples the lightweight interaction unit from the simulation instruction parsing unit. Applying this invention effectively lowers the technical threshold for operating professional simulation software on mobile devices, allowing users to submit simulation tasks and obtain results without relying on complex graphical interfaces, thus promoting the mobile and lightweight application of simulation technology.

[0059] like Figure 3 As shown, a specific embodiment of the present invention is used for simulation testing of an automobile engine: The lightweight natural interactive control method of industrial simulation according to the present invention is used to simulate and test an automobile engine, which includes the following: First, preliminary work needs to be done. Before the system officially runs, the initialization configuration of the simulation capability registration module in the simulation instruction parsing unit needs to be completed to realize the simulation capability registration of the simulation instruction parsing unit. The specific operation is as follows: The simulation methods are structured. Technical personnel write structured description files for the executable operations (such as "model parameter settings", "simulation task startup", "result data query") of the target simulation software (such as engineering simulation software Ansys, Meta Universe platform). Each file must clearly define the English name of the method (such as engine simulation parameter name), functional description, required / optional parameters (including parameter name, description, and value range) and execution content.

[0060] In this embodiment, the structured description file includes the method name, method description information, configuration parameters, and execution content.

[0061] The method description describes performing a car engine simulation test based on specified initial speed, ambient temperature, and runtime, and then outputting the test results. Configuration parameters (params) include initial speed, ambient temperature, and runtime. The initial engine speed is used to characterize the initial engine speed in the car engine simulation, measured in revolutions per minute (rpm), with a range of 500-5000. A placeholder is set to receive the specific value from the user. The ambient temperature is used to characterize the simulation environment temperature, measured in degrees Celsius, with a range of -40 to 85 degrees Celsius. The runtime is used to characterize the simulation runtime, measured in seconds (s), with a range of 10-3600. Based on the configuration parameters, the execution content is set, and the car engine is started for simulation testing.

[0062] The method vector library construction process in this embodiment is described in the structured conversion process of the instruction parsing module.

[0063] Then, the system developed using the method of this invention is run to perform simulation testing on an automobile engine, which includes the following steps: Step 1: The user initiates a simulation operation request via a mobile device. The lightweight interaction unit then collects, preprocesses, and sends the instructions. The specific process is as follows: Step 1. In the interaction and preprocessing module of the mobile device, the user can input voice commands through the microphone (such as "set the initial speed of the car engine simulation model to 1500 rpm, the ambient temperature to 25℃, the running time to 120 seconds, and perform a simulation test"), or input text commands with the same content through the keyboard.

[0064] Step 2. Preprocess the instructions, which includes the following: If the command is a voice command, the module calls a third-party speech recognition SDK and loads a built-in domain-specific vocabulary (including terms such as revolutions per minute, simulation test, and initial speed) as a custom hot word list to improve the accuracy of professional terminology recognition and convert speech into text.

[0065] Standardization processing is performed on the converted text (or text directly entered by the user): words such as “that thing” and “trouble” are removed, capitalization and punctuation are standardized, core instruction content is extracted, and the standardized car engine simulation instruction is obtained, which has an initial speed of 1500 rpm, an ambient temperature of 25℃, and a running time of 120 seconds.

[0066] Step 3. Encapsulate and send instructions. Use a lightweight instruction encapsulation module to encapsulate the standardized core instructions into transmittable data in a preset JSON format.

[0067] Subsequently, the communication module sends the encapsulated instructions to the designated interface of the simulation instruction parsing unit in real time through the data transmission protocol, ensuring the real-time performance and reliability of data transmission.

[0068] Step 2: After receiving the instructions from the lightweight interaction unit, the simulation instruction parsing unit completes instruction parsing, method matching, and forwarding. The specific process is as follows: S1. Receiving instructions and extracting fields: The simulation instruction parsing unit communication interface receives JSON instructions sent by the lightweight interaction unit and extracts the raw_text field.

[0069] S2. The structured instructions are converted. The instruction parsing module passes the raw_text field to the preset large model prompt template Prompt (the template contains the set roles, JSON output format requirements, parsing rules and examples), driving the large model to output structured instructions. The structured instructions include method name, method description, and configuration parameters.

[0070] S3. Using the same vectorization algorithm as the method description file in the "Preliminary Work," the above structured instructions are converted into user instruction vectors. The cosine similarity algorithm is used to calculate the similarity between the user instruction vectors and all method vectors in the simulation method vector library, selecting the simulation method (operation instruction) with the highest similarity. The instruction forwarding module operates according to the target simulation software API interface bound to the matched simulation method.

[0071] Step 3: The adapter interface of the target simulation software receives the instructions from the simulation instruction parsing unit, converts them into operation commands that the simulation software can recognize (such as starting the simulation, setting parameters, obtaining results, etc.), and the simulation execution module executes the specific simulation tasks according to the instructions and collects process information.

[0072] Step 4: The result feedback module of the target simulation software sends the simulation data back to the simulation instruction parsing unit. The simulation instruction parsing unit then sends the data back to the lightweight interactive unit. At the same time, it also uses the data storage module to back up the simulation data for easy viewing on other devices.

[0073] Step 5: After the lightweight interactive unit receives the simulation command and the parsing unit forwards the data, it visualizes the progress and results. The specific process is as follows: S1. The communication module receives process data from the cloud, parses it, and then transmits it to the display module.

[0074] S2. The display module shows the simulation progress in real time in the task list using a progress bar and text prompts (such as "60% of the car engine simulation calculation is in progress").

[0075] S3. The analysis results are compressed into a 3D model using points, lines, and surfaces. Simulation results (such as temperature and stress distribution of key engine components) are overlaid onto the model as color cloud maps, and key indicators (such as "maximum stress: 250MPa" and "average temperature: 82℃") are displayed as digital cards.

[0076] S4. Finally, take a screenshot and export the simulation results to complete the simulation test of the car engine.

[0077] This invention employs a lightweight natural interaction architecture and mobile device adaptation technology to construct a lightweight natural interaction module suitable for mobile devices. By leveraging natural language processing technologies such as speech recognition and text parsing, user-inputted voice / text commands are converted into a standardized protocol format, thereby breaking through the complex operation mode of traditional simulation software that relies on a "mouse and keyboard + graphical interface".

[0078] Meanwhile, this invention utilizes a cloud-based dynamic function mapping and intelligent intent and method matching mechanism to propose a cloud-based instruction parsing technology based on vectorized representation and intent recognition. By semantically describing and structurally modeling the functional interfaces of the target simulation software (defining a unified instruction method interface specification and vectorized storage format), and combining runtime parameter binding and vector search algorithms, dynamic mapping from natural language instructions to specific simulation functions is achieved.

[0079] Therefore, this invention can be conveniently applied to industrial design and R&D scenarios as well as scientific research fields. When engineers are conducting field research or on-site investigations, they can remotely control the simulation software in the R&D department in real time via mobile devices to perform simulation analysis and optimization of product designs, thereby accelerating product development. When researchers are conducting field investigations or collaborating in different locations, they can use mobile devices to remotely operate and monitor complex simulation tasks, improving the efficiency and convenience of scientific research.

[0080] A server embodiment employing the method of the present invention: A server comprising: One or more processing units; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processing units, the one or more processing units implement the aforementioned lightweight natural interactive control method for industrial simulation.

[0081] The storage device can be internal memory, external memory, cache memory, or other special-purpose memory. The processing unit has signal processing capabilities and can be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, an off-the-shelf programmable gate array, or other programmable logic device.

[0082] An embodiment of a device employing the method of the present invention: An electronic device is provided with a computer-readable storage medium storing a computer program, which, when executed by a processing unit, implements the aforementioned lightweight natural interactive control method for industrial simulation.

[0083] Computer-readable storage media are physical carriers capable of storing computer-recognizable data, instructions, or programs. They must meet the core characteristic of being "readable by a computer," meaning the data exists in the form of electrical, magnetic, or optical signals and can be converted into binary information that a computer can process through appropriate devices. Physical carriers include magnetic storage media, optical storage media, semiconductor storage media, or other storage media.

[0084] Those skilled in the art will understand that the embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of entirely hardware, entirely software, or a combination of hardware and software embodiments. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0085] The models, modules, or units in this application are objects that objectively describe the form and structure in physical or virtual form. They are not equivalent to objects and are not limited to physical or virtual forms. They can be data processing functions, software programs, processing modes, usage methods, operation methods, workflows, application processes, electronic hardware, circuit modules, processing systems, system imitations, or simulation objects.

[0086] Finally, it should be noted that the above embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions, and are not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that within the technical scope disclosed in the present invention, modifications or equivalent substitutions can be made to the technical solutions of the foregoing embodiments. These modifications or substitutions will not substantially deviate from the spirit and scope of the embodiments of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the scope of the claims.

Claims

1. A lightweight natural interaction control method for industrial simulation, characterized by comprising the following steps: acquiring executable operations of target simulation software through a pre-constructed simulation capability registration module, and performing vectorization processing on the executable operations to establish a simulation method vector library; collecting natural speech information or natural text instructions of users about industrial simulation requirements through a pre-constructed lightweight interaction unit, and performing standardization processing on the natural speech information or natural text instructions to obtain simulation instruction data; performing structuralization processing on the simulation instruction data through a pre-constructed simulation instruction analysis unit, and generating simulation target instructions according to the simulation method vector library; inputting the simulation target instructions into the target simulation software end to execute the industrial simulation requirement task of the user, obtaining the industrial simulation result, and realizing the lightweight natural interaction control based on the industrial simulation.

2. The lightweight natural interaction control method for industrial simulation according to claim 1, characterized by comprising the following steps: acquiring executable operations of target simulation software through a pre-constructed simulation capability registration module, and performing vectorization processing on the executable operations to establish a simulation method vector library, the method comprising the following steps: Step 11: determining the target simulation software to be adapted, and sorting out the executable operations of the target simulation software, which include but are not limited to model parameter setting, mesh division, statics analysis, simulation task starting, result data query, stress calculation, and temperature calculation; Step 12: establishing a simulation field-specific vocabulary library according to the target simulation software and the executable operations, which at least includes mega-pascal, revolutions per minute, and rendering data format; Step 13: generating a plurality of structural description files based on the simulation field-specific vocabulary library and each executable operation, each structural description file corresponding to an executable operation, which includes method name, function description, parameter information, and execution content; Step 14: converting each structural description file into a simulation method vector recognizable by the target simulation software through a vectorization processing mechanism, and constructing a simulation method vector library for instruction matching.

3. The lightweight natural interaction control method for industrial simulation according to claim 2, characterized by comprising the following steps: Step 14: converting the structural description file into a vector form recognizable by the target simulation software through a vectorization processing mechanism to construct the simulation method vector library, the method comprising the following steps: acquiring the method name, function description, parameter description, and execution content in the structural description file, and forming a text with four fields; performing vectorization processing on the text with four fields using a field pre-training word vector model to obtain basic vectors of each field, including method name vector, function description vector, parameter description vector, and execution content vector; performing processing on a plurality of parameter description vectors through a mean aggregation algorithm to calculate a parameter overall vector; allocating a preset weight according to the importance of each field in operation recognition; performing weighted splicing on the method name vector, function description vector, parameter overall vector, and execution content vector according to the preset weight to generate a simulation method vector corresponding to a certain executable operation. ​ The simulation method vector of all executable operations is stored uniformly, and a vector index is established to obtain a simulation method vector library.

4. The lightweight natural interaction control method for industrial simulation according to claim 1, characterized in that: The natural speech information or natural text instruction of the user about the industrial simulation demand is collected by using the pre-constructed lightweight interaction unit, and the natural speech information or natural text instruction is standardized to obtain the simulation instruction data. Step 21, the natural speech information or natural text instruction of the user about the industrial simulation demand is collected by the lightweight interaction unit; Step 22, a simulation field specific vocabulary library is established for the professional terms in the industrial simulation field, which at least includes simulation operation terms, unit terms and data format terms, and is transmitted to the voice recognition module as a custom hotword table; When the user outputs natural speech information, the voice recognition module matches the professional terms through the custom hotword table, recognizes the colloquial expressions in the natural speech information, obtains the natural text instruction, and then executes step 23; When the user outputs natural text instruction, step 23 is executed; Step 23, the natural text instruction is standardized to extract core information and obtain the simulation instruction; Step 24, the standardized simulation instruction is packaged into a structured and lightweight data format to form simulation instruction data based on the text-based open source data exchange format.

5. The lightweight natural interaction control method for industrial simulation according to claim 1, characterized in that: Step 23, the natural text instruction is standardized to extract core information and obtain the simulation instruction, and the method is as follows: The redundant information in the natural text instruction is removed to obtain the first stage instruction; The unit format in the first stage instruction is unified and converted into a standard unit to obtain the second stage instruction; The case of the second stage instruction is unified to obtain the third stage instruction; The core information is extracted from the third stage instruction to obtain the instruction key elements, including the target object, operation parameters and core operation; The target object, operation parameters and core operation are summarized to obtain the simulation instruction.

6. The lightweight natural interaction control method for industrial simulation according to claim 1, characterized in that: The simulation instruction data is structured by using the pre-constructed simulation instruction analysis unit, and the simulation target instruction is generated according to the simulation method vector library, and the method is as follows: Step 31, the simulation instruction analysis unit is deployed on the cloud server; Step 32, the simulation instruction analysis unit receives the simulation instruction data through the long connection protocol, and extracts the core instruction from the simulation instruction data as the original instruction text; Step 33, a prompt word template is constructed to convert the original instruction text into a structured instruction consistent with the structured description file format, so that the fields are aligned; Step 34, the structured instruction is converted into a method name vector, a function description vector, a parameter information vector and an execution content vector by a vectorization processing mechanism; then a parameter overall vector is calculated by processing a plurality of parameter description vectors through a mean aggregation algorithm. The method name vector, the function description vector, the parameter integral vector and the execution content vector are weighted and spliced according to preset weights, and a user instruction vector is calculated; Step 35, by a cosine similarity algorithm, the user instruction vector is matched with the simulation method vector in the simulation method vector library, and an optimal executable operation is obtained; Step 36, the parameter value in the user instruction vector is bound with the optimal executable operation, and the application programming interface information of the target simulation software is supplemented, so that a simulation target instruction capable of being executed by the target simulation software is generated.

7. The lightweight natural interaction control method for industrial simulation according to claim 6, characterized in that: Step 33, a prompt word template is constructed, and the original instruction text is converted into a structured instruction consistent with the structured description file format, in the following manner: Step 331, based on the original instruction text, a simulation character is set; The simulation character is an industrial simulation instruction analysis engine, and its task is to convert the natural language instruction into standardized structured data; Step 332, according to the structured description file, output format requirement information is set, and the output format requirement information is output in a text-based open source data exchange format, and the fields thereof are completely matched with the four core fields of the structured description file; the four core fields include method name, function description, parameter information and execution content; Step 333, according to the simulation character and the output format requirement information, an analysis rule is established; The analysis rule includes preferentially matching simulation field specific vocabulary, extracting parameter value and retaining original term expression; Step 334, based on the simulation character, the output format requirement and the analysis rule, a prompt word template is established; Step 335, the original instruction text is input into the prompt word template, and a structured instruction consistent with the structured description file format is generated.

8. The lightweight natural interaction control method for industrial simulation according to claim 1, characterized in that: The simulation target instruction is input into the target simulation software end, the user's industrial simulation demand task is executed, and an industrial simulation result is obtained, in the following manner: Step 41, the interface protocol of the simulation target instruction and the target simulation software is adapted, invalid simulation target instructions are excluded through verification, and a simulation instruction request package is generated; Step 42, based on a network transmission protocol, the simulation instruction request package is transmitted to the target simulation software, and the simulation instruction request package is end-to-end encrypted; Step 43, the target simulation software schedules resources according to the priority of the simulation instruction request package, executes the user's industrial simulation demand task, and collects process information in real time; Step 44, the target simulation software collects simulation result data, encapsulates and returns the data to the lightweight interaction unit in a standard format, so that the industrial simulation result can be displayed.

9. The lightweight natural interaction control method for industrial simulation according to claim 8, characterized in that: Step 41, the interface protocol of the simulation target instruction and the target simulation software is adapted, invalid simulation target instructions are excluded through verification, and a simulation instruction request package is generated, in the following manner: Step 411, the simulation instruction analysis unit calls the interface protocol specification of the target simulation software, and the interface protocol specification is an application programming interface protocol or a script protocol; Step 412, according to the interface protocol specification, the generated simulation target instruction is converted into a format that can be recognized by the target simulation software, which includes the following contents: If the interface protocol specification of the target simulation software is an application programming interface: The target method and binding parameters in the simulation target instruction are mapped to the application programming interface request parameters, and the interface address, request method and authentication token are supplemented to generate a simulation instruction request package conforming to the network transmission protocol; If the interface protocol specification of the target simulation software is a script execution: The target simulation instruction is converted into a script statement to generate a simulation instruction request package, so that the parameter assignment is consistent with the software script syntax; Step 413, based on the authentication token, verify the user's operation authority, whether it has the execution authority of the simulation control, if the user has the execution authority, then execute step 414; if the user does not have the execution authority, then the simulation result data is no authority, and then step 44 is executed; Step 414, check whether the computing resources of the target simulation software meet the task requirements, if the computing resources meet the task requirements, then execute step 415; if the computing resources do not meet the task requirements, then the simulation result data is insufficient resources, and then step 44 is executed; Step 415, according to the structured description file, the simulation instruction request package is checked, and whether all the required parameters have been bound with valid values is checked, if all the required parameters have been bound with valid values, then step 42 is executed; if there is a required parameter that has not been bound with a valid value, then the simulation result data is missing parameter valid value, and then step 44 is executed.

10. A lightweight natural interaction control system for industrial simulation, characterized in that: It comprises a cloud server and a mobile device; The cloud server is provided with a simulation capability registration module and a simulation instruction analysis unit; The simulation capability registration module is used to obtain executable operations of the target simulation software, and to perform vectorization processing on the executable operations to establish a simulation method vector library; The simulation instruction analysis unit is used to structure the simulation instruction data, and to generate simulation target instructions according to the simulation method vector library; The mobile device is provided with a lightweight interaction unit; The lightweight interaction unit is used to collect natural voice information or natural text instructions of the user about the industrial simulation demand, and to standardize the natural voice information or natural text instructions to obtain simulation instruction data; and to display the industrial simulation result generated by the target simulation software after executing the simulation target instruction; The mobile device and the cloud server communicate through a standardized network protocol to form a remote decoupling architecture, realizing lightweight natural interaction control of industrial simulation.

Images