Circuit design method and system based on AI

Abstract

The invention discloses an AI-based circuit design method and system, and belongs to the crossing field of electrical engineering, electronic engineering, automatic control and artificial intelligence, and the method comprises a demand analysis module which is used for receiving a project demand inputted by a user, analyzing a keyword and a function target of the project demand, and determining a design direction; the multi-modal design module is used for receiving a design direction and a project demand, selecting a corresponding design module according to the design direction, and designing a circuit according to the project demand by the multi-modal design module to obtain a normalized text, a logic function expression or an SPICE netlist; the closed-loop verification and iterative optimization module is used for performing simulation verification and automatic iterative optimization processing on the SPICE netlist, and is matched with the multi-modal design module to obtain the SPICE netlist with a simulation verification report; a component type selection module; according to the method, the design efficiency and consistency are remarkably improved.

Description

Technical Field

[0001] This invention relates to the interdisciplinary fields of electrical engineering, electronic engineering, automation control and artificial intelligence, and specifically to AI-based circuit design methods and systems. Background Technology

[0002] Existing technologies include AI-assisted circuit design (such as using AI agents to optimize SoC circuits), the application of mathematical models in circuit simulation (such as SPICE netlists), and domain-specific automation tools. For example, in the field of relay control, there is PLC-based logic programming software; in the field of electronic circuits, there are EDA tools for schematic design and simulation. However, these tools and methods are often isolated, serving different design stages and technical fields.

[0003] Existing technologies suffer from significant "island effects" and "discontinuity problems." First, the lack of a unified design framework leads to incompatibility of AI design methodologies across different circuit domains (high-voltage, low-voltage, and logic control), requiring separate model construction for each. Second, design and verification are disconnected; AI outputs are mostly graphical or proprietary code, making it difficult to directly convert them into models that can be simulated and verified using standardized tools (such as SPICE). Third, inconsistent output formats and poor readability make it difficult for human engineers to quickly understand, verify, and modify the AI's design intent, especially in complex patent disclosure and team collaboration scenarios, resulting in low efficiency. Finally, for specialized applications such as high voltage, existing AI designs often neglect systemic safety and protection mechanisms.

[0004] More importantly, existing technologies generally lack automated closed-loop capabilities for design and verification. After the design is generated, simulation, analysis, and debugging need to be performed manually. If the results are inconsistent, the design parameters need to be manually modified and the process repeated, resulting in a long development cycle and a high dependence on the individual experience of engineers. This open-loop design model cannot achieve true end-to-end automation.

[0005] Based on this, the present invention designs an AI-based circuit design method and system to solve the above problems. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, this invention provides an AI-based circuit design method and system.

[0007] To achieve the above objectives, the present invention provides the following technical solution: AI-based circuit design systems include: Requirements Analysis Module: This module receives project requirements input by the user, analyzes the keywords and functional goals of the project requirements, and determines the design direction. Multimodal design module: Used to receive design direction and project requirements, select the corresponding design module according to the design direction, and design the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist; Closed-loop verification and iterative optimization module: Used to perform simulation verification and automatic iterative optimization of SPICE netlist. When there is a deviation between the simulation results and the project requirements, the deviation parameters are automatically identified and the multimodal design module is driven to regenerate the optimized SPICE netlist until the performance indicators fully meet the project requirements, thus forming a closed loop with the multimodal design module to obtain a SPICE netlist with a simulation verification report. Component Selection Module: Used when the multimodal design module generates standardized text to select appropriate component function types; used when the closed-loop verification and iterative optimization module generates the SPICE netlist for the simulation verification report to select and replace different component performance parameters or SPICE models; used when the output generation module prepares output to match appropriate specific component models from the engineering knowledge base module based on the parameters of the output standardized text, logical function expression, or SPICE netlist with simulation verification report, and generate a planning table. Engineering Knowledge Base Module: It contains a structured component database, which provides component parameter information, SPICE models and selection decision support when designing circuits in the multimodal design module and adjusting parameters in the closed-loop verification and iterative optimization module. Output generation module: Outputs standardized text, logical function expressions, or SPICE netlists with simulation verification reports. The SPICE netlists adopt standardized node naming rules, sub-circuit packaging specifications, and topology description formats. The visualization tool can parse the SPICE netlists with simulation verification reports in standardized topology description formats and generate visualized circuit schematics.

[0008] Furthermore, the design directions include electrical circuits, relay circuits, and electronic circuits.

[0009] Furthermore, the multimodal design module includes an electrical circuit engine module, a relay circuit engine module, and an electronic circuit engine module; When the design direction is electrical circuit, the electrical circuit engine module is activated. When the design direction is the relay circuit direction, the relay circuit engine module is activated. When the design direction is electronic circuits, the electronic circuit engine module is activated.

[0010] Furthermore, the specific operation of the electrical circuit engine module is as follows: (1) Transform project requirements into a computable mathematical model; (2) Based on the “mathematics-component mapping knowledge base”, match the corresponding components, combine the matched corresponding components with the “component function set” to obtain the circuit topology that conforms to the mathematical model; (3) Output standardized text.

[0011] Mathematical forms include PID control, integral, and derivative.

[0012] The standardized text includes the operating status of electrical circuits, current / voltage paths, protection mechanisms, and safe operating sequences.

[0013] Furthermore, the relay circuit engine module operates as follows: (1) Analyze project requirements according to the industrial "logic design method" and generate a work cycle diagram and component status table; (2) Analyze the characteristic number of the control scenario required by the project, then introduce intermediate relays and derive the logic function expression.

[0014] Furthermore, the specific operation of the electronic circuit engine module is as follows: (1) The three-level requirement decomposition method was used to analyze the project requirements and determine the state machine model and component parameters; (2) Generate a SPICE netlist for simulation based on the state machine model and component parameters.

[0015] Furthermore, the specific operations of the three-level requirement decomposition method for analyzing project requirements and determining the state machine model are as follows: ① First-level decomposition: AI decomposes project requirements into three levels of system modules; ② Second-level decomposition: AI decomposes the 3-level module into 2-level functional circuits and defines signal interfaces; ③ Third-level decomposition: AI decomposes the level 2 modules into level 1 basic elements and constructs a state machine model that describes the dynamic behavior of the system modules.

[0016] Level 3 system modules include power supply, sensing, or control system modules; Level 2 functional circuits include filters or comparators.

[0017] Furthermore, the specific operations of the closed-loop verification and iterative optimization module are as follows: Step 1: Receive the SPICE netlist generated by the electronic circuit engine module, and call the backend SPICE simulator to perform simulation analysis and processing to obtain a simulation verification report; Step 2: The AI ​​reads and parses the simulation output data file to extract performance metrics; Performance metrics such as voltage, current, rise time, and frequency response; Step 3: AI compares the extracted performance indicators with the indicators in the project requirements and determines whether there is a deviation. If the deviation is found, proceed to step 4. If the deviation is found, generate a simulation verification report. Then, combine the simulation verification report with the SPICE netlist to form a SPICE netlist with the simulation verification report. The SPICE netlist with the simulation verification report is output to the component selection module and the output generation module. Step 4: Generate a simulation deviation report. The AI ​​identifies the circuit parameters that need to be adjusted based on the simulation deviation report and optimizes them based on its domain knowledge. Step 5: The electronic circuit engine module regenerates a modified SPICE netlist based on the optimized and adjusted data, and then executes Step 1.

[0018] Circuit parameters such as resistance values, capacitance values, and op-amp gain.

[0019] A design method utilizing an AI-based circuit design system includes the following steps: Step 1: Receive project requirements from users, analyze the keywords and functional goals of the project requirements, and determine the design direction; Step 2: Receive the design direction and project requirements, select the corresponding design module according to the design direction, and the multimodal design module designs the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist. When generating standardized text, the component selection module is called to select the appropriate component function type. Step 3: The SPICE netlist is processed for simulation verification and combined with the multimodal design module to obtain a SPICE netlist with a simulation verification report. The component selection module is called to select and replace different component performance parameters or SPICE models. Step 4: Receive normalized text, logical function expressions, or SPICE netlists with simulation verification reports; extract the component names from the normalized text, logical function expressions, or SPICE netlists with simulation verification reports; determine the component information in the component database based on the performance, cost, and safety constraints within the project requirements; and compile the information into a plan table. Step 5: Output standardized text, logic function expressions, or SPICE netlists with simulation verification reports. The component selection module is invoked to match suitable specific component models from the engineering knowledge base module based on the parameters of the output standardized text, logic function expressions, or SPICE netlists with simulation verification reports, and generate a planning table. The visualization tool can parse the SPICE netlists with standardized topology description formats and generate visualized circuit schematics.

[0020] Compared with the prior art, the beneficial effects of this invention are as follows: This application significantly improves design efficiency and consistency: By using a unified AI framework, design experience in electrical, relay, and electronic fields is systematized, realizing the automation of cross-domain design processes, shortening design time, and ensuring the consistency of design style and quality.

[0021] 2. Achieve integrated design and verification: By outputting structured files that can be directly used for simulation (SPICE netlist), plotting (logical expressions), and review (normalized text), the barrier from abstract design to quantifiable verification is completely broken down, greatly shortening the product iteration cycle.

[0022] 3. Lowering the threshold for complex system design: AI automates complex intellectual labor, including logical deduction, multi-level decomposition, and state machine modeling, allowing engineers to focus more on top-level innovation while also reducing the difficulty for junior engineers to participate in complex designs.

[0023] 4. Enhanced Design Reliability and Safety: AI can systematically and comprehensively integrate protection mechanisms (such as overvoltage / overcurrent protection), safety sequences, and interlocking logic during the design process. Compared to manual design, it is better able to avoid safety hazards caused by negligence, especially in high-risk fields such as high voltage and industrial control. The built-in closed-loop verification and iterative optimization mechanism ensures that the final output circuit design is theoretically correct and meets performance specifications, greatly reducing the debugging costs and failure risks of later physical prototypes.

[0024] 5. It achieves true end-to-end automation: from abstract requirement input to verified, deliverable circuit model output, the entire process is completed automatically by the AI ​​platform, boosting R&D efficiency to a new level.

[0025] 6. Accumulate and reuse engineering knowledge: The structured knowledge base and rule engine solidify expert experience and engineering specifications into the platform, ensuring that every design is based on best practices and guaranteeing high quality and reliability. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0027] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0029] The terms "left," "right," "front," "back," "up," and "down" used in the following description refer to the orientation from the perspective of the front view.

[0030] Example 1: In some embodiments, please refer to the accompanying drawings. Figure 1 AI-based circuit design systems include: Requirements Analysis Module: This module receives project requirements input by the user, analyzes the keywords and functional goals of the project requirements, and determines the design direction. Multimodal design module: Used to receive design direction and project requirements, select the corresponding design module according to the design direction, and design the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist; The requirements analysis module automatically determines the design direction based on project requirements keywords and functional objectives, and drives the design engine corresponding to electrical circuits, relay circuits or electronic circuits in the multimodal design module according to the design direction, so as to realize a multimodal unified design framework across the fields of high power, low power and logic control. Closed-loop verification and iterative optimization module: Used to perform simulation verification and automatic iterative optimization of SPICE netlist. When there is a deviation between the simulation results and the project requirements, the deviation parameters are automatically identified and the multimodal design module is driven to regenerate the optimized SPICE netlist until the performance indicators fully meet the project requirements, thus forming a closed loop with the multimodal design module to obtain a SPICE netlist with a simulation verification report. Component Selection Module: Used when the multimodal design module generates standardized text to select appropriate component function types; used when the closed-loop verification and iterative optimization module generates the SPICE netlist for the simulation verification report to select and replace different component performance parameters or SPICE models; used when the output generation module prepares output to match appropriate specific component models from the engineering knowledge base module based on the parameters of the output standardized text, logical function expression, or SPICE netlist with simulation verification report, and generate a planning table. Engineering Knowledge Base Module: It contains a structured component database, which provides component parameter information, SPICE models and selection decision support when designing circuits in the multimodal design module and adjusting parameters in the closed-loop verification and iterative optimization module. Output generation module: Outputs standardized text, logical function expressions, or SPICE netlists with simulation verification reports. The SPICE netlists adopt standardized node naming rules, sub-circuit packaging specifications, and topology description formats. The visualization tool can parse the SPICE netlists with simulation verification reports in standardized topology description formats and generate visualized circuit schematics.

[0031] The design directions include electrical circuits, relay circuits, and electronic circuits.

[0032] The specific operations of the requirements analysis module are as follows: (1) The system has a built-in "keyword-design direction" mapping knowledge base.

[0033] (2) When keywords such as “motor”, “sequential start”, “interlock” and “PLC” are detected, the mapping knowledge base determines that it is “relay circuit direction”.

[0034] (3) "High voltage", "heating film", "power conversion" and "PID control" were detected, and the mapping knowledge base determined it to be "electrical circuit direction".

[0035] (4) The following were detected: “signal processing”, “filtering”, “amplification”, “sensor”, “ADC”, “microcontroller”. The mapping knowledge base determined that it was “electronic circuit direction”.

[0036] (5) Analyze the functional objectives and constraints of the project requirements.

[0037] For example, in the same "control", is it necessary to adjust the analog signal precisely (electronic circuit direction) or to have reliable switching logic (relay circuit direction)? Example: ① The requirement is "Design a system to sequentially control the start and stop of three water pumps". The AI ​​will identify that the core function is "sequential logic" and thus determine that "the direction of the relay circuit" is the optimal solution.

[0038] ② The requirement is "Design a circuit that can amplify weak sensor signals and filter out high-frequency noise". The AI ​​will identify that the core function is "analog signal processing", thus determining the "electronic circuit direction".

[0039] (6) Based on the combined keyword mapping results and functional semantic analysis results, the AI's decision logic will provide a final design direction. This design direction is then used to activate the multimodal design module, electrical circuit engine module, relay circuit engine module, or electronic circuit engine module.

[0040] The decision-making logic of AI specifically refers to a computational model based on weighted scoring and rule-based reasoning. Its workflow is as follows: Weighted scoring phase: The AI ​​first scans user needs for keywords. The system's built-in "keyword-design direction" mapping knowledge base assigns a basic score to each keyword; (For example, "motor" is "relay circuit direction" +10 points, "filter" is "electronic circuit direction" +15 points).

[0041] Rule-based reasoning stage: The AI ​​then performs functional semantic analysis and applies preset engineering rules to calculate the score.

[0042] For example, the rule: IF Function Target IN ("Timing Control", "Sequential Logic") AND Component IN ("Contactor", "Button") THEN "Relay Direction" weight * 1.5. This rule amplifies the possibilities of relay direction.

[0043] Final Decision: The AI ​​aggregates all weighted and adjusted scores, and the design direction with the highest score (such as "electronic circuits") is determined as the final decision. This decision result is then passed to the multimodal design module to activate the corresponding design engine.

[0044] AI's decision-making logic is essentially an expert system that simulates the thought process of a senior engineer when making technology selections at the start of a project.

[0045] The multimodal design module includes an electrical circuit engine module, a relay circuit engine module, and an electronic circuit engine module; When the design direction is electrical circuit, the electrical circuit engine module is activated. When the design direction is the relay circuit direction, the relay circuit engine module is activated. When the design direction is electronic circuits, the electronic circuit engine module is activated.

[0046] The specific operation of the electrical circuit engine module is as follows: (1) Transform project requirements into a computable mathematical model; (2) Based on the “mathematics-component mapping knowledge base”, match the corresponding components, combine the matched corresponding components with the “component function set” to obtain the circuit topology that conforms to the mathematical model; (3) Output standardized text.

[0047] Mathematical forms include PID control, integral, and derivative.

[0048] The standardized text includes the operating status of electrical circuits, current / voltage paths, protection mechanisms, and safe operating sequences.

[0049] The AI ​​transforms the circuit topology into normalized text using graph traversal algorithms and predefined grammar-based text generation rules. Internally, the AI ​​constructs a graph representation of the circuit (nodes represent components, edges represent connections), and then "serializes" it into normalized text through the following steps: 1. Identify the main circuit and branches: The AI ​​starts with the power supply and uses a depth-first search or breadth-first search algorithm to traverse the entire circuit diagram to identify the main series paths (main circuit) and parallel branches (branches).

[0050] 2. Applying the serial representation method: (1) For a series path without branches, AI will output the names of components in sequence according to the current flow direction and connect them with the "→" symbol.

[0051] (2) Syntax: Component 1+ → Component 2 → ... → Component n → Component 1- 3. Apply parallel representation: (1) When traversing to a branch node, the AI ​​will recognize that it is a parallel structure. It will recursively traverse each parallel branch, separate the result of each branch with the "||" symbol, and finally enclose the entire parallel structure with "【】" brackets.

[0052] (2) Grammar: ... → [(Branch A) || (Branch B) || ...] → ... (3) Among them, the format of branch A is → component A1 → component A2... 4. Applying coupling and interlocking representation: (1) For coupling elements such as transformers and mutual inductors, AI will define them as Component 1 Part 1 - Component 1 Part 2 and reference each part in the circuit path.

[0053] (2) For logical relationships such as relay interlocking, AI will label them as component 1-component 2.

[0054] 5. Structured information generation: In addition to the circuit path, AI will also extract other key information from its internal data structure and fill it in according to the preset chapter format (such as operating status, control port, protection details, security sequence).

[0055] AI is "reading aloud" a circuit diagram: starting from the beginning (power supply), it follows the circuit, reading through series circuits and saying "there are several branches here, the first is..., the second is..." when it encounters parallel circuits. Finally, it records the entire "reading" process using predefined grammar to form standardized text.

[0056] "Reading aloud" is essentially a structured graph data serialization process.

[0057] Internally, AI represents circuit topology as a graph data structure, where components are nodes and electrical connections are edges. The process of converting this graph data structure into standardized text is the true meaning of "recording this reading process," which follows a strict set of grammar generation rules: Input: The graphical data structure of the circuit.

[0058] Processing: The AI ​​starts a graph traversal algorithm (such as depth-first search) from the power node.

[0059] Output generation: During the traversal, the algorithm generates a corresponding text string based on our predefined syntax templates (e.g., → for concatenation and || for parallel connection) for each node (component) or edge (connection) it visits.

[0060] Final result: After the entire traversal is completed, the concatenated strings form a complete, standardized text describing the circuit path.

[0061] Therefore, the act of "recording" is essentially the transformation (or "serialization") of a non-linear graph structure into a linear text string.

[0062] The specific operation of the relay circuit engine module is as follows: (1) Analyze project requirements according to the industrial "logic design method" and generate a work cycle diagram and component status table; (2) Analyze the characteristic number of the control scenario required by the project, then introduce intermediate relays and derive the logic function expression.

[0063] The specific operation of the relay circuit engine module is as follows: Step 1: Generate the work cycle diagram (motion beat diagram and state definition): (1) AI first parses the user’s timing control requirements (such as “After A takes action, B will take action after a 5-second delay”) into a state sequence.

[0064] (2) AI assigns a unique program number to each state (or step) and draws the transition relationship and triggering conditions between states (such as button press, sensor trigger, timer expiration) to form a visual work cycle diagram (or internal equivalent state transition table).

[0065] Step 2: Component State Table Generation (Generate I / O List and State Table): (1) AI identifies all the execution components (such as motor contactor KM1, solenoid valve YV1) and detection components (such as button SB1, limit switch SQ1, timer KT1) in the project requirements.

[0066] (2) Based on the work cycle diagram generated in the previous step, AI will create an "Execution Element Status Table" to define which execution element's coil is energized (1) or de-energized (0) under each program number.

[0067] (3) At the same time, AI creates a "Detection Element Status Table" to define whether the contact status of each detection element under each program number is closed (1) or open (0).

[0068] Step 3: Analyze the number of features in the control scenarios of project requirements (feature code analysis and grouping): (1) AI regards the contact state combination (such as 01101) of each row (i.e. each program number) in the "Detection Element Status Table" as a unique "feature number" (i.e. feature code) of that state.

[0069] (2) The AI ​​will traverse the feature codes of all states and find those states with the same number of features to form a group of states with the same number of features. These states look the same to the external input signals. If no special method is used to distinguish them, the circuit will be confused logically.

[0070] Step 4: Introduce an intermediate relay: (1) For the state group with the same number of features found in the previous step, AI will introduce one or more intermediate relays (KA) as “memory elements”.

[0071] (2) The AI ​​energizes and latches the coil of the intermediate relay when it enters the first confusing state, and de-energizes it when it leaves the confusing state. By checking the contact state of the intermediate relay (whether it is KA or -KA), the AI ​​can create a new and unique internal signature for these states with the same signature, thereby solving the logic confusion problem.

[0072] Step 5: Derive the logical function expression: (1) After all states have a unique feature code, the AI ​​will deduce the energizing conditions for each coil of the actuator (such as KM1, KM2...) and each coil of the intermediate relay (such as KA1, KA2...).

[0073] (2) The energizing condition is a Boolean logic expression that describes "under what conditions the coil should be energized". AI generates a series of logic function expressions by combining the relevant detection element contacts and intermediate relay contacts (series represents *, parallel represents +), such as KM1=KA1*SQ2*(-KA2).

[0074] It fully replicates the rigorous thought process that senior electrical engineers use when designing complex relay control systems.

[0075] The specific operation of the electronic circuit engine module is as follows: (1) The three-level requirement decomposition method was used to analyze the project requirements and determine the state machine model and component parameters; (2) Generate a SPICE netlist for simulation based on the state machine model and component parameters.

[0076] The specific steps for using the three-level requirement decomposition method to analyze and determine the state machine model for project requirements are as follows: ① First-level decomposition: AI decomposes project requirements into three levels of system modules; ② Second-level decomposition: AI decomposes the 3-level module into 2-level functional circuits and defines signal interfaces; ③ Third-level decomposition: AI decomposes the level 2 modules into level 1 basic elements and constructs a state machine model that describes the dynamic behavior of the system modules.

[0077] Level 3 system modules include power supply, sensing, or control system modules; Level 2 functional circuits include filters or comparators.

[0078] Furthermore, the specific operations of the closed-loop verification and iterative optimization module are as follows: Step 1: Receive the SPICE netlist generated by the electronic circuit engine module, and call the backend SPICE simulator to perform simulation analysis and processing to obtain a simulation verification report; Step 2: The AI ​​reads and parses the simulation output data file to extract performance metrics; Performance metrics such as voltage, current, rise time, and frequency response; Step 3: AI compares the extracted performance indicators with the indicators in the project requirements and determines whether there is a deviation. If the deviation is found, proceed to step 4. If the deviation is found, generate a simulation verification report. Then, combine the simulation verification report with the SPICE netlist to form a SPICE netlist with the simulation verification report. The SPICE netlist with the simulation verification report is output to the component selection module and the output generation module. 1. Simulation Verification Report (1) Definition: A summary document automatically generated by the AI ​​system when the extracted performance indicators are compared with the indicators in the project requirements without deviation. This proves that the current version of the SPICE netlist design is qualified and meets the requirements.

[0079] (2) Specific content: ①A. Simulation Summary: Includes the type of simulation performed (e.g., .DC TEMP, .TRAN), the simulation range (e.g., temperature 25-100 degrees), and the filename and version number of the SPICE netlist.

[0080] ②B. Performance Indicator Comparison Table: For example:

[0081] ③C. Key Waveforms / Charts (Optional): Includes links to or embedded screenshots of key chart files from the simulation output, such as a graph of LED current versus temperature, for engineers to visualize and review.

[0082] ④D. Conclusion: such as "Simulation verification shows that the current design meets all key performance indicators". Purpose: The simulation verification report is the "certificate of conformity" of the design. It is delivered to the user or downstream process along with the final SPICE netlist as a "SPICE netlist with simulation verification report".

[0083] Step 4: Generate a simulation deviation report. The AI ​​identifies the circuit parameters that need to be adjusted based on the simulation deviation report and optimizes them based on its domain knowledge. Simulation Deviation Report Definition: A diagnostic document generated by the AI ​​system when there are discrepancies between the extracted performance metrics and those in the project requirements. Its purpose is to quantify the discrepancy and provide input for the next iteration of optimization.

[0084] 2. Specific content: (1) A. Simulation summary: including the type of simulation performed (e.g., .DC TEMP, .TRAN), simulation range (e.g., temperature 25-100 degrees), and the file name and version number of the SPICE netlist.

[0085] (2) B. Performance Deviation Details Table: For example:

[0086] (3) C. Optimization suggestions (AI's core capabilities): Based on the built-in domain knowledge (knowledge base and rule engine), AI will analyze the reasons for deviations and propose circuit parameters that need to be adjusted.

[0087] ①Example: Analysis of deviation causes: The alarm threshold temperature is too low.

[0088] ②Preliminary diagnosis: The comparator's reference voltage V_ref (node ​​3) is too low.

[0089] ③ Optimization suggestion: Increase V_ref. Consider reducing the resistance of voltage divider resistor R3.

[0090] ④ Suggested parameter adjustment: {"Component":"R3","Action":"Reduce resistance","Strategy":"Iterative search algorithm"}; 3. Function: This report is a "diagnosis" and "optimization prescription". It is directly input into the closed-loop verification and iterative optimization module to guide the AI ​​on how to make the next parameter adjustment, thereby starting a new round of design-simulation cycle.

[0091] The "Simulation Verification Report" is the endpoint, declaring success; the "Simulation Deviation Report" is the starting point, initiating optimization. Step 5: The electronic circuit engine module regenerates a modified SPICE netlist based on the optimized and adjusted data, and then executes Step 1.

[0092] Circuit parameters such as resistance values, capacitance values, and op-amp gain.

[0093] A design method utilizing an AI-based circuit design system includes the following steps: Step 1: Receive project requirements from users, analyze the keywords and functional goals of the project requirements, and determine the design direction; Step 2: Receive the design direction and project requirements, select the corresponding design module according to the design direction, and the multimodal design module designs the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist. When generating standardized text, the component selection module is called to select the appropriate component function type. Step 3: The SPICE netlist is processed for simulation verification and combined with the multimodal design module to obtain a SPICE netlist with a simulation verification report. The component selection module is called to select and replace different component performance parameters or SPICE models. Step 4: Receive normalized text, logical function expressions, or SPICE netlists with simulation verification reports; extract the component names from the normalized text, logical function expressions, or SPICE netlists with simulation verification reports; determine the component information in the component database based on the performance, cost, and safety constraints within the project requirements; and compile the information into a plan table. Step 5: Output standardized text, logic function expressions, or SPICE netlists with simulation verification reports. The component selection module is invoked to match suitable specific component models from the engineering knowledge base module based on the parameters of the output standardized text, logic function expressions, or SPICE netlists with simulation verification reports, and generate a planning table. The visualization tool can parse the SPICE netlists with standardized topology description formats and generate visualized circuit schematics.

[0094] Visualization tools: SPICE Vision, LTspice, TINA Pro; SPICE Vision is a tool for SPICE circuit simulation, whose main functions include visualization, debugging, and analysis of SPICE netlists. The following is specific information.

[0095] LTspice is a circuit simulation software developed by Analog Devices, primarily used for the design and analysis of analog electronic circuits.

[0096] TINA Pro includes virtual instruments, circuit simulation, and parameter optimization; Virtual Instruments: Integrates 8 virtual measurement tools (such as oscilloscopes and signal analyzers), supporting dynamic demonstrations and real-time signal comparison.

[0097] Circuit simulation: Supports DC analysis, transient analysis, Fourier analysis, etc., and the results can be directly embedded into the circuit diagram.

[0098] Parameter optimization: Optimizing circuit design through symbolic analysis and worst-case analysis. In step two, the multimodal design module generates standardized text during the process topology design, and the component selection module is called to select the appropriate component function type. In step three, the component selection module is invoked when the SPICE netlist used in the closed-loop verification and iterative optimization module generates the simulation verification report to select and replace different component performance parameters or SPICE models. In step five: when the output generation module is ready to output, the component selection module is called to match the appropriate specific component model from the engineering knowledge base module based on the parameters of the output normalized text, logical function expression or SPICE netlist with simulation verification report, and generate a plan table. This application significantly improves design efficiency and consistency: by using a unified AI framework, design experience in electrical, relay, and electronic fields is systematized, realizing the automation of cross-domain design processes, shortening design time, and ensuring consistency in design style and quality.

[0099] 2. Achieve integrated design and verification: By outputting structured files that can be directly used for simulation (SPICE netlist), plotting (logical expressions), and review (normalized text), the barrier from abstract design to quantifiable verification is completely broken down, greatly shortening the product iteration cycle.

[0100] 3. Lowering the threshold for complex system design: AI automates complex intellectual labor, including logical deduction, multi-level decomposition, and state machine modeling, allowing engineers to focus more on top-level innovation while also reducing the difficulty for junior engineers to participate in complex designs.

[0101] 4. Enhanced Design Reliability and Safety: AI can systematically and comprehensively integrate protection mechanisms (such as overvoltage / overcurrent protection), safety sequences, and interlocking logic during the design process. Compared to manual design, it is better able to avoid safety hazards caused by negligence, especially in high-risk fields such as high voltage and industrial control. The built-in closed-loop verification and iterative optimization mechanism ensures that the final output circuit design is theoretically correct and meets performance specifications, greatly reducing the debugging costs and failure risks of later physical prototypes.

[0102] 5. It achieves true end-to-end automation: from abstract requirement input to verified, deliverable circuit model output, the entire process is completed automatically by the AI ​​platform, boosting R&D efficiency to a new level.

[0103] 6. Accumulate and reuse engineering knowledge: The structured knowledge base and rule engine solidify expert experience and engineering specifications into the platform, ensuring that every design is based on best practices and guaranteeing high quality and reliability.

[0104] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AI-based circuit design system, characterized in that: include: Requirements Analysis Module: This module receives project requirements input by the user, analyzes the keywords and functional goals of the project requirements, and determines the design direction. Multimodal design module: Used to receive design direction and project requirements, select the corresponding design module according to the design direction, and design the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist; Closed-loop verification and iterative optimization module: Used to perform simulation verification and automatic iterative optimization of SPICE netlist. When there is a deviation between the simulation results and the project requirements, the deviation parameters are automatically identified and the multimodal design module is driven to regenerate the optimized SPICE netlist until the performance indicators fully meet the project requirements, thus forming a closed loop with the multimodal design module to obtain a SPICE netlist with a simulation verification report. Component selection module: This module is called when the multimodal design module generates standardized text to select the appropriate component function type. It is used to select and replace different component performance parameters or SPICE models when the SPICE netlist in the closed-loop verification and iterative optimization module generates the simulation verification report; it is also used to match the appropriate specific component model from the engineering knowledge base module and generate the plan table based on the parameters of the output normalized text, logical function expression or SPICE netlist with simulation verification report when the output generation module prepares the output. Engineering Knowledge Base Module: It contains a structured component database, which provides component parameter information, SPICE models and selection decision support when designing circuits in the multimodal design module and adjusting parameters in the closed-loop verification and iterative optimization module. Output generation module: Outputs standardized text, logical function expressions, or SPICE netlists with simulation verification reports. The SPICE netlists adopt standardized node naming rules, sub-circuit packaging specifications, and topology description formats. The visualization tool can parse the SPICE netlists with simulation verification reports in standardized topology description formats and generate visualized circuit schematics.

2. The AI-based circuit design system according to claim 1, characterized in that, The design directions include electrical circuits, relay circuits, and electronic circuits.

3. The AI-based circuit design system according to claim 2, characterized in that, The multimodal design module includes an electrical circuit engine module, a relay circuit engine module, and an electronic circuit engine module; When the design direction is electrical circuit, the electrical circuit engine module is activated. When the design direction is the relay circuit direction, the relay circuit engine module is activated. When the design direction is electronic circuits, the electronic circuit engine module is activated.

4. The AI-based circuit design system according to claim 3, characterized in that, The specific operation of the electrical circuit engine module is as follows: (1) Transform project requirements into a computable mathematical model; (2) Match the corresponding components based on the "mathematics-component mapping knowledge base", combine the matched corresponding components with the "component function set" to obtain the circuit topology that conforms to the mathematical model; (3) Output standardized text.

5. The AI-based circuit design system according to claim 4, characterized in that, The specific operation of the relay circuit engine module is as follows: (1) Analyze project requirements according to the industrial "logic design method" and generate a work cycle diagram and component status table; (2) Analyze the characteristic number of the control scenario required by the project, then introduce intermediate relays and derive the logic function expression.

6. The AI-based circuit design system according to claim 5, characterized in that, The specific operation of the electronic circuit engine module is as follows: (1) The three-level requirement decomposition method was used to analyze the project requirements and determine the state machine model and component parameters; (2) Generate a SPICE netlist for simulation based on the state machine model and component parameters.

7. The AI-based circuit design system according to claim 6, characterized in that, The specific steps for using the three-level requirement decomposition method to analyze and determine the state machine model for project requirements are as follows: ① First-level decomposition: AI decomposes project requirements into three levels of system modules; ② Second-level decomposition: AI decomposes the 3-level module into 2-level functional circuits and defines signal interfaces; ③ Third-level decomposition: AI decomposes the level 2 modules into level 1 basic elements and constructs a state machine model that describes the dynamic behavior of the system modules.

8. The AI-based circuit design system according to claim 7, characterized in that, The specific operations of the closed-loop verification and iterative optimization module are as follows: Step 1: Receive the SPICE netlist generated by the electronic circuit engine module, and call the backend SPICE simulator to perform simulation analysis and processing to obtain a simulation verification report; Step 2: The AI ​​reads and parses the simulation output data file to extract performance metrics; Step 3: AI compares the extracted performance indicators with the indicators in the project requirements and determines whether there is a deviation. If the deviation is found, proceed to step 4. If the deviation is found, generate a simulation verification report. Then, combine the simulation verification report with the SPICE netlist to form a SPICE netlist with the simulation verification report. The SPICE netlist with the simulation verification report is output to the component selection module and the output generation module. Step 4: Generate a simulation deviation report. The AI ​​identifies the circuit parameters that need to be adjusted based on the simulation deviation report and optimizes them based on its domain knowledge. Step 5: The electronic circuit engine module regenerates a modified SPICE netlist based on the optimized and adjusted data, and then executes Step 1.

9. A design method utilizing the AI-based circuit design system of claim 8, characterized in that, Includes the following steps: Step 1: Receive project requirements from users, analyze the keywords and functional goals of the project requirements, and determine the design direction; Step 2: Receive the design direction and project requirements, select the corresponding design module according to the design direction, and the multimodal design module designs the circuit according to the project requirements to obtain standardized text, logic function expressions or SPICE netlist. When generating standardized text, the component selection module is called to select the appropriate component function type. Step 3: The SPICE netlist is processed for simulation verification and combined with the multimodal design module to obtain a SPICE netlist with a simulation verification report. The component selection module is called to select and replace different component performance parameters or SPICE models. Step 4: Receive normalized text, logical function expressions, or SPICE netlists with simulation verification reports; extract the component names from the normalized text, logical function expressions, or SPICE netlists with simulation verification reports; determine the component information in the component database based on the performance, cost, and safety constraints within the project requirements; and compile the information into a plan table. Step 5: Output standardized text, logic function expressions, or SPICE netlists with simulation verification reports. The component selection module is invoked to match suitable specific component models from the engineering knowledge base module based on the parameters of the output standardized text, logic function expressions, or SPICE netlists with simulation verification reports, and generate a planning table. The visualization tool can parse the SPICE netlists with standardized topology description formats and generate visualized circuit schematics.

Images