A device model file self-correction conversion method and system
By employing a heterogeneous computing architecture and a large language model self-correction mechanism, the problem of syntax and physical errors in cross-platform conversion of device model files was solved, achieving efficient and automated model file conversion and repair, and improving conversion quality and efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN HUADA EMPYREAN TECH CO LTD
- Filing Date
- 2026-04-30
- Publication Date
- 2026-07-31
AI Technical Summary
Existing technologies suffer from several drawbacks in cross-platform conversion of device model files, including difficulty in covering structural syntactic differences, limitations of standardized regularization methods, difficulty in detecting errors in physical meaning, high debugging costs, and a lack of closed-loop feedback mechanisms, leading to model conversion failures or simulation anomalies.
Employing a heterogeneous computing architecture (GPU+CPU) and a simulation feedback-driven large language model self-correction mechanism, the system achieves automated device model file conversion and repair by parsing the logical topology, performing simulation verification, semantic analysis, and providing repair prompts.
It achieves unattended, high-fidelity, and high-pass-rate automatic conversion and repair of model files, breaking through the limitations of regular expression methods, accurately identifying physical errors and optimizing resource scheduling, thereby improving conversion efficiency and quality controllability.
Smart Images
Figure CN122491209A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the interdisciplinary field of integrated circuit design automation (EDA) and artificial intelligence, specifically to a self-correcting conversion method and system for device model files. Background Technology
[0002] Device model files are standardized text / data files that describe the electrical behavior of semiconductor devices in integrated circuit design, simulation, and chip manufacturing. They are core input files that EDA simulators must read. In advanced integrated circuit design flows, device model files (such as SPICE models) are a crucial bridge connecting process manufacturing and circuit simulation. As process nodes continue to shrink, model complexity increases dramatically, often containing hundreds of physical parameters. When it is necessary to migrate a model from a source foundry to a target foundry or different simulation platforms, format conversion and parameter adaptation are essential.
[0003] Existing technologies primarily rely on human experience or regular expression-based scripts for model transformation. However, this process suffers from the following significant problems: 1) Structural syntactic differences are difficult to cover with rules: Hspice and Spectre differ fundamentally in their model file organization. For example, Hspice often uses inline .MODEL statements to define device parameters, while Spectre requires the use of independent model blocks and supports hierarchical referencing; copyright information in the file header, process corner identifiers, sub-circuit calling methods, and the nesting relationships of parameter blocks are often incompatible between the two platforms. These differences cannot be resolved by simple string replacement.
[0004] 2) Limitations of Regular Expression Methods: Traditional regular expression matching relies on preset, fixed templates. However, in actual engineering, model files written by different manufacturers or even different engineers exhibit highly personalized writing styles in terms of parameter order, line break style, comment placement, non-standard abbreviations, and nesting levels. Regular expressions lack the ability to understand contextual semantics, cannot handle dynamic nested structures, and are prone to conversion failures or the introduction of hidden errors.
[0005] 3) Errors in physical meaning are difficult to detect: Even if the model file syntax is completely correct, if the order of magnitude, unit system or sign of a certain physical parameter (such as carrier mobility μ_eff, threshold voltage V_th0) is wrong, it will cause the simulation to fail to converge or the output to be abnormal. However, such errors cannot be detected by syntax checking.
[0006] 4) High debugging cost: It may take several hours to several days to locate the cause of a single convergence failure, especially in large-scale models (such as FinFET, GAA) where the parameters are strongly coupled and manual troubleshooting is extremely inefficient.
[0007] 5) Lack of closed-loop feedback mechanism: Current LLM-assisted modeling tools are only used for initial generation and cannot form an automated iterative closed loop of "generation-verification-repair" with the simulator, resulting in uncontrollable generation quality.
[0008] Therefore, there is an urgent need for a cross-platform conversion system and method for device model files that can automatically generate, verify, iteratively repair, and quantifiablely accept them, in order to meet the needs of modern PDK development and multi-platform compatible deployment. Summary of the Invention
[0009] To address the shortcomings of existing technologies, this application aims to provide a device model file self-correction conversion method and system. Based on a heterogeneous computing architecture (GPU+CPU) and a simulation feedback-driven large language model self-correction mechanism, it achieves unattended, high-fidelity, and high-pass-rate automatic conversion and repair of model files.
[0010] To achieve the above objectives, the device model file self-correcting conversion method provided in this application includes: The logical topology of the source device model file is parsed, the syntax required by the target platform is reconstructed, and an initial device model file conforming to the target platform specification is generated. Load the initial device model and perform simulation to verify syntax and parameter integrity; Perform semantic analysis on error logs to extract structured repair prompts; The structured repair prompt information is corrected, and the initial device model file is updated according to the corrected parameter definition or structure adjustment code to generate a new version of the device model file; Perform automated verification of conversion quality and output the final device model file.
[0011] Furthermore, the step of parsing the logical topology of the source model file, reconstructing the syntax required by the target platform, and generating an initial device model file that conforms to the target platform specification also includes: calculating the syntax structure difference entropy of the source device model file. H struct The formula for estimating conversion difficulty and optimizing resource scheduling is as follows: , in, For the source device model file, the first Normalized frequency of occurrence of class struct elements m This represents the total number of categories of different types of structural elements in the source device model file.
[0012] Furthermore, the source device model file includes: device model data from the source process plant; the target platform specification includes: the syntax, unit system, parameter naming rules, and file structure requirements of the target simulator.
[0013] Furthermore, the step of loading the initial device model and performing simulation to verify syntax and parameter integrity also includes: Load the initial device model and call the simulator to perform simulation to verify the following: Check the initial device model file for syntax errors, including mismatched brackets, missing keywords, and illegal characters; Verify that all required parameters are correctly defined and that their numerical formats are valid; Monitor the simulator logs to capture runtime exceptions, including but not limited to: "Parameter undefined", "Syntax error near line X", "Newton iteration failed", and "Floatingpoint exception". Furthermore, the step of performing semantic analysis on the error logs to extract structured repair prompts also includes: Parse the simulation log file to extract structured error information, including error type, error parameter name, line number, and context description; For parameters with anomalous numerical values but valid syntax, calculate their physical plausibility error metric. : , in, These are the converted parameter values. This is a typical reference value for this parameter at the target process node; Based on the calculated physical rationality error metric To determine whether the quality of the model's numerical conversion meets the standards.
[0014] Furthermore, the step of correcting the structured repair prompt information and updating the initial device model file according to the corrected parameter definition or structure adjustment code to generate a new version of the device model file also includes: The structured repair prompt information is input into the LLM inference error root cause, which includes: incorrect parameter values or misaligned structure nesting; Based on the corrected parameter definitions or structural adjustment code output by LLM, the device model file is automatically updated, generating a new version of the device model file.
[0015] Furthermore, it also includes: Record the parameter vector to be corrected after two consecutive repairs. and Calculate the total relative change : , in, n The total number of variables. i Indicates the first i One variable, k This represents the number of iterations in the current period; If the number of simulation failures exceeds the preset number and If the convergence threshold is less than the preset threshold, the model is determined to be unable to be automatically repaired, the iteration is terminated and an alarm is issued; otherwise, the verification is re-executed to form a closed-loop iteration until the simulator is successfully loaded and there are no abnormal errors.
[0016] Furthermore, the step of performing automated conversion quality verification and outputting the final device model file also includes: Statistical parameter pass rate and test case pass rate; The parameter pass rate and the test case pass rate are combined into a comprehensive quality index for model transformation. Based on the comprehensive quality index of the model conversion, the conversion quality is determined to be up to standard, and the final device model file is output.
[0017] To achieve the above objectives, this application also provides a device model file self-correcting conversion system to implement the device model file self-correcting conversion method described above.
[0018] To achieve the above objectives, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the device model file self-correction conversion method as described above.
[0019] Compared with the prior art, the device model file self-correction conversion method and system of this application have the following advantages: By leveraging the semantic understanding capabilities of the Large Language Model (LLM), we can adaptively handle the nonlinear differences between different simulation platforms in terms of file headers, nested structures, and parameter description syntax, effectively addressing non-standard writing styles and overcoming the limitations of regular expression methods. Automatic repair of parameter and structural errors: This is achieved through quantitative indicators of physical rationality error measurement. It accurately identifies unreasonable physical parameters and guides LLM correction based on simulation logs; By avoiding infinite loops through the convergence criterion Δ, the system's stability and reliability are improved, and it possesses a robust iterative control mechanism. Based on structural entropy Dynamically allocate GPU inference resources, support conversion difficulty prediction and resource scheduling, and improve overall throughput efficiency; The comprehensive indicator Q enables engineering-level delivery judgment and provides quantifiable quality standards. GPUs accelerate LLM inference, CPUs execute verification tasks in parallel, and through efficient collaboration of heterogeneous resources, the system can be scaled to batch conversion of large-scale model libraries.
[0020] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing this application. Attached Figure Description
[0021] The accompanying drawings are provided to further illustrate the present application and form part of the specification. Together with the embodiments of the present application, they serve to explain the present application but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the device model file self-correction conversion method according to an embodiment of this application; Figure 2 This is a diagram illustrating the overall architecture of the device model file self-correction conversion system according to an embodiment of this application. Figure 3 This is a schematic diagram of an electronic device structure according to an embodiment of this application.
[0022] The above figures are only used to illustrate the technical solution of this application and are not intended to limit the scope of protection of this application. Detailed Implementation
[0023] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application.
[0024] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.
[0025] The term "comprising" and its variations as used in this application are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.
[0026] It should be noted that the terms "first" and "second" may be used in this application only to distinguish different devices, components or parts, and are not used to define the order of functions performed by these devices, components or parts or their interdependence.
[0027] It should be noted that the terms "one" and "more" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless explicitly stated otherwise in the context, they should be understood as "one or more". "More" should be understood as two or more.
[0028] The device model file self-correction conversion method and system of this application are aimed at the migration of semiconductor device models (including BSIM, HiSIM, etc.) across platforms or process nodes. Based on heterogeneous clusters and simulation feedback, it uses the Large Language Model (LLM) and circuit simulator to automatically complete model conversion, verification and repair.
[0029] The core of the device model file self-correction conversion method and system in this application lies in building a collaborative closed-loop system composed of a GPU-based Large Language Model (LLM) inference engine and a CPU-based EDA simulator. Through a four-stage process of "generation-verification-repair-checking", the automatic optimization and delivery of model files are realized.
[0030] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0031] Example 1 The technical solution of this application will be described in detail below with reference to specific embodiments. Figure 1 This is a flowchart of the device model file self-correction conversion method according to an embodiment of this application. The following will be combined with... Figure 1 The self-correction conversion method for device model files in this application is described in further detail.
[0032] First, in step 101, the source device model file and the target platform specification are obtained. The logical topology structure is parsed from the source device model file, the syntax required by the target platform is reconstructed, and an initial device model file conforming to the target platform specification is generated.
[0033] In this embodiment of the application, the source device model file includes device model data from the source process plant, and the target platform specification includes the target simulator's syntax, unit system, parameter naming rules, and file structure requirements.
[0034] The dedicated fine-tuning LLM deployed on the GPU is invoked to parse the logical topology of the source device model file based on contextual semantics and physical knowledge base. The logical topology includes file header metadata, Include and sub-circuit nesting relationships, and parameter scope. The syntax required by the target platform is reconstructed to generate a V1.0 model file (initial device model file) that conforms to the target platform specification.
[0035] In one embodiment of this application, the syntactic structure difference entropy of the source device model file is calculated. H struct Based on the estimated conversion difficulty and optimized resource scheduling, the calculation formula is as follows: , in f j For the first in the source file j Normalized frequency of occurrence of class structure elements (including .MODEL, .SUBCKT, include, parameter block starter). H struct The higher the value, the more complex the structure, and the more LLM inference resources the system will allocate to ensure the quality of the generated data.
[0036] In step 102, the initial device model is loaded, and the simulator is called to verify the syntax and parameter integrity.
[0037] In this embodiment, the system automatically generates a standardized testbench template for loading the initial device model and performing basic simulation tasks.
[0038] In this embodiment, an initial device model is loaded, and a simulator is invoked to perform a lightweight simulation to complete the following verifications: 1) Check the model file for syntax errors, including mismatched brackets, missing keywords, and illegal characters; 2) Are all required parameters correctly defined and in valid numerical formats? 3) Monitor the emulator logs and capture runtime exceptions, including but not limited to: "Parameter undefined", "Syntax error near line X", "Newton iteration failed", and "Floatingpoint exception".
[0039] This stage uses the successful completion of model loading and the return to a normal exit status by the simulator as the criterion for "passing".
[0040] Step 103: Perform semantic analysis on the error log to extract structured repair prompts.
[0041] In this embodiment of the application, if the simulator reports an error, the system parses the simulation log file and extracts structured repair prompt information, including the error type, the name of the error parameter, the line number, and the context description.
[0042] For parameters with abnormal numerical values but valid syntax, the system calculates a physical reasonableness error metric. : in These are the converted parameter values. This is a typical reference value for this parameter at the target process node (obtained from the built-in semiconductor physics knowledge base). If Greater than the preset physical threshold If the value is 1.5 (e.g., 1.5), it is considered an order-of-magnitude anomaly and needs to be corrected first.
[0043] Based on the initial device model parameter values, the current converted parameter values, the error context, and the aforementioned quantification indicators, the system constructs a structured repair prompt message.
[0044] In this embodiment, the large language model requires sufficient structured information to make decisions. Physical rationality error measurement. It is only used to determine whether the transformation result is within an acceptable range. The repair prompt information is used to provide information to the large model, and the large model makes decisions and modifications accordingly.
[0045] In step 104, the structured repair prompt information is corrected, and the initial device model file is updated according to the corrected parameter definition or structure adjustment code to generate a new version of the device model file.
[0046] In this embodiment, structured repair prompts are input into a Large Language Model (LLM). The LLM, combining its built-in semiconductor physics knowledge and understanding of cross-platform syntax, infers the root cause of the error. The root cause of the error includes incorrect parameter values or misaligned structural nesting.
[0047] The LLM outputs the corrected parameter definitions or structural adjustment codes, and the system automatically updates the device model file, generating a new version of the device model file.
[0048] In this embodiment of the application, to prevent invalid loops or divergent repairs, the system introduces an iterative convergence criterion: recording the parameter vector to be corrected after two consecutive repairs. and Calculate the total relative change : , in, n The total number of variables. i Indicates the first i One variable,k This represents the number of iterations in the current period.
[0049] If the number of simulation failures exceeds the preset number and Less than the preset convergence threshold If the value is 0.01 (e.g., 0.01), the model is determined to be unable to be automatically repaired, the iteration is terminated, and an alarm is issued.
[0050] Otherwise, return to step 102 to re-execute the verification, forming a closed-loop iteration until the simulator is successfully loaded and there are no abnormal errors.
[0051] In step 105, the quality assessment module is started to perform automated conversion quality verification and output the final device model file.
[0052] In this embodiment of the application, after the initial device model passes the verification in step 102, the system starts the quality assessment module and calculates the following indicators: Parameter pass rate: ,in, The number of parameters passed. This represents the total number of parameters. For parameter pass rate; Test case pass rate: ,in, The number of passable test cases. This represents the total number of use cases. This represents the pass rate for test cases.
[0053] Further integration into a comprehensive quality index for model transformation : , This is the weighting coefficient. If... Greater than or equal to the preset quality threshold If the value is 0.996 (e.g., 0.996), the conversion quality is deemed satisfactory, and the final model file is output; otherwise, a new round of iteration of step 104 is triggered.
[0054] In this embodiment, the entire system is deployed on a heterogeneous computing cluster: the GPU cluster unit is responsible for LLM high-concurrency inference, and the CPU cluster unit runs multiple verification tasks in parallel to achieve a high-throughput, low-latency automated conversion pipeline.
[0055] Example 2 In this embodiment of the application, the device model file self-correction conversion method of this application is described using a BSIM4 nMOS model (containing 1024 parameters) provided by a user-input source process manufacturer and a Cadence Spectre target platform as an example.
[0056] Step 1: The Large Language Model (LLM) identifies the Hspice-style .MODEL inline structure in the source device model file, converts it into an independent model block required by Spectre, rewrites the parameter description syntax, and generates an initial device model file that conforms to the target platform specification; simultaneously, it calculates... It was determined to be a high-complexity model and was allocated high-priority GPU resources.
[0057] Step two: When calling the target platform Spectre to load the initial device model, the error "Parameter U0 not defined" was encountered.
[0058] Step 3: Analyze the log and find that U0 is missing. Combined with the original value of 0.045, construct the prompt message: "Original U0=0.045, the definition of U0 is missing after conversion, please complete it."
[0059] Step four: LLM completes U0=0.044 and outputs the new version; verification still fails, reporting the error "mobility too high". The system calculates... After ruling out orders of magnitude errors, the unit system was checked, and the final value was corrected to U0 = 44 (units). ), verification passed; Step 5, statistics are obtained , Therefore The decision was approved. The system outputs the final model file and records the conversion log.
[0060] Example 3 This application provides a device model file self-correcting conversion system for implementing the device model file self-correcting conversion method described in the above embodiments.
[0061] Figure 2 This is a diagram illustrating the overall architecture of the device model file self-correction conversion system according to an embodiment of this application. Figure 2 As shown, the device model file self-correction conversion system of this application embodiment consists of a heterogeneous cluster composed of CPU cluster units and GPU cluster units, which, along with simulation feedback, realizes the self-correction conversion of device model files, including: The heterogeneous collaborative scheduling module 201 dynamically allocates tasks to GPU cluster units or CPU node cluster units based on the structural complexity (Hstruct) of the input source device model file, thereby achieving heterogeneous collaborative optimal scheduling of resources.
[0062] The GPU LLM inference module 202 generates an initial device model based on the source device model file and specifications, and sends it to the CPU simulation verification module 203 for loading, syntax checking and verification.
[0063] The CPU simulation verification module 203 performs syntax checks and verifications on the initial device model generated by the GPU LLM inference module 202.
[0064] The iterative convergence control module 204 monitors the change (Δ) in the self-feedback iteration process of the GPU LLM inference module 202 in real time, and stops the iteration if convergence is achieved.
[0065] Error parsing and quantization module 205 parses the simulation log file, extracts structured repair prompts, calculates the physical error (εp), and generates precise repair instructions to be sent back to GPU LLM inference module 202.
[0066] The quality assessment module 206 calculates the comprehensive score (Q) after all iterations, and outputs the final model if it meets the standard.
[0067] The device model file self-correction conversion system of this application embodiment is based on the large language model self-correction mechanism driven by the simulator's running log. It converts the parameter missing, syntax error or numerical anomaly information returned by the EDA simulator into structured prompts to guide the large language model (LLM) to correct model parameters or cross-platform syntax structures.
[0068] The device model file self-correction conversion method and system of this application, based on a heterogeneous computing architecture (GPU+CPU) and a simulation feedback-driven large language model self-correction mechanism, solves the problem of simulation loading failure or abnormal operation caused by unreasonable physical meaning of parameters or incompatibility of cross-platform syntax structure in the existing device model conversion process, and realizes unattended, high-fidelity, and high-pass-rate automatic conversion and repair of model files.
[0069] Example 4 In embodiments of this application, an electronic device is also provided. Figure 3 This is a schematic diagram of the electronic device structure according to an embodiment of this application, such as... Figure 3 As shown, the electronic device of this application includes a processor 401 and a memory 402, wherein, The memory 402 stores a computer program, which, when read and executed by the processor 401, performs the steps described above in the embodiment of the device model file self-correction conversion method.
[0070] Example 5 In the embodiments of this application, a computer-readable storage medium is also provided, which stores a computer program, wherein the computer program is configured to execute the steps in the embodiments of the device model file self-correction conversion method as described above when running.
[0071] In this embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.
[0072] It will be understood by those skilled in the art that the above descriptions are merely preferred embodiments of this application and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A self-correcting conversion method for device model files, comprising: The logical topology of the source device model file is parsed, the syntax required by the target platform is reconstructed, and an initial device model file conforming to the target platform specification is generated. Load the initial device model and perform simulation to verify syntax and parameter integrity; Perform semantic analysis on error logs to extract structured repair prompts; The structured repair prompt information is corrected, and the initial device model file is updated according to the corrected parameter definition or structure adjustment code to generate a new version of the device model file; Perform automated verification of conversion quality and output the final device model file.
2. The device model file self-correcting conversion method according to claim 1, characterized in that, The steps of parsing the logical topology of the source model file, reconstructing the syntax required by the target platform, and generating an initial device model file that conforms to the target platform specification also include: calculating the syntax structure difference entropy of the source device model file. H struct The formula for estimating conversion difficulty and optimizing resource scheduling is as follows: ， in, For the source device model file, the first Normalized frequency of occurrence of class struct elements m This represents the total number of categories of different types of structural elements in the source device model file.
3. The device model file self-correction conversion method according to claim 1, wherein The source device model file includes: device model data from the source process manufacturer; the target platform specification includes: the syntax, unit system, parameter naming rules, and file structure requirements of the target simulator.
4. The device model file self-correction conversion method according to claim 1, wherein The step of loading the initial device model and performing simulation to verify syntax and parameter integrity further includes: Load the initial device model and call the simulator to perform simulation to verify the following: Check the initial device model file for syntax errors, including mismatched brackets, missing keywords, and illegal characters; Verify that all required parameters are correctly defined and that their numerical formats are valid; Monitor the simulator logs and capture runtime exceptions, including but not limited to: "Parameter undefined", "Syntax error near line X", "Newton iteration failed", and "Floating point exception".
5. The device model file self-correction conversion method according to claim 1, wherein The step of performing semantic analysis on the error log and extracting structured repair prompt information further includes: Parse the simulation log file to extract structured error information, including error type, error parameter name, line number, and context description; For numerically abnormal but syntactically legal parameters, compute their physical reasonableness error metric : ， wherein, is the converted parameter value, is a typical reference value for this parameter at the target process node; based on the computed physical plausibility error measure judges whether the model value conversion quality is up to standard.
6. The device model file self-correction conversion method according to claim 1, wherein The step of correcting the structured repair prompt information and updating the initial device model file and generating a new version of the device model file based on the corrected parameter definition or structure adjustment code further includes: The structured repair prompt information is input into the LLM inference error root cause, which includes: incorrect parameter values or misaligned structure nesting; Based on the corrected parameter definitions or structural adjustment code output by LLM, the device model file is automatically updated, generating a new version of the device model file.
7. The device model file self-correction conversion method according to claim 6, wherein Also includes: Record the parameter vector to be corrected after two consecutive repairs. and Calculate the total relative change : ， in, n The total number of variables. i Indicates the first i One variable, k This represents the number of iterations in the current period; If the number of simulation failures exceeds the preset number and If the convergence threshold is less than the preset threshold, the model is determined to be unable to be automatically repaired, the iteration is terminated and an alarm is issued; otherwise, the verification is re-executed to form a closed-loop iteration until the simulator is successfully loaded and there are no abnormal errors.
8. The device model file self-correcting conversion method according to claim 1, characterized in that, The steps of performing automated conversion quality verification and outputting the final device model file also include: Statistical parameter pass rate and test case pass rate; The parameter pass rate and the test case pass rate are combined into a comprehensive quality index for model transformation. Based on the comprehensive quality index of the model conversion, the conversion quality is determined to be up to standard, and the final device model file is output.
9. A device model file self-correcting conversion system, characterized in that, Implement the device model file self-correction conversion method according to any one of claims 1-8.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor is configured to execute the computer program stored in the memory to implement the device model file self-correction conversion method according to any one of claims 1-8.