A Method and System for Linked Generation of Traffic Simulation Files Based on Natural Language Parsing

By correcting lane attribute conflicts through natural language parsing and fault-tolerance mechanisms, and synchronizing heterogeneous log timestamps, the problems of logical conflicts and evaluation errors in traffic simulation systems are solved, enabling the system to operate coherently and evaluate accurately.

CN123116583APending Publication Date: 2026-11-17XIONGAN QIANFANG SHUCHENG ZHILIAN TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611281358.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-24
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing traffic simulation systems lack a logical conflict correction mechanism when faced with natural language input, causing the model building process to stop. Furthermore, the misalignment of heterogeneous log time dimensions leads to evaluation errors, making it impossible to maintain the system's operational status.

Method used

Traffic simulation files are generated by parsing natural language syntax. Fault tolerance is achieved using a circuit breaker token bucket and hash value verification mechanism to correct lane attribute conflicts. Heterogeneous log timestamps are synchronized using a linear interpolation algorithm to ensure the continuity of the simulation process and the accuracy of the evaluation.

Benefits of technology

It effectively avoids simulation model interruptions caused by logical conflicts, ensures the continuity of the traffic simulation system and the accuracy of evaluation indicators, provides a fault-tolerance mechanism to isolate parsing errors, unifies the time dimension of heterogeneous logs, and improves the reliability of simulation results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116583A_ABST
    Figure CN123116583A_ABST
Patent Text Reader

Abstract

This invention relates to the field of traffic simulation technology, and discloses a method and system for generating traffic simulation files based on natural language grammar parsing. The method includes receiving external text sequences, calculating domain lexical density, allocating a circuit breaker token bucket; extracting intersection topology parameters, generating a basic road network configuration within a temporary isolation sandbox, and recording file hash values; parsing timing instructions to correct lane physical attribute conflicts and generate timing configurations; determining the input traffic overflow state and triggering a spatiotemporal peak-shaving algorithm to allocate discrete delays and generate routing configurations; verifying the file status hash values ​​within the sandbox and combining them with the token bucket state to synthesize the simulation master control file and start the underlying process; aligning the underlying heterogeneous logs, calculating the average delay index, and generating an evaluation report. This invention ensures the continuity of the road network scenario through internal logic path correction, maintains the stable operation of the simulation master process through an isolation verification takeover mechanism, and ensures evaluation accuracy by aligning the time dimension.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of traffic simulation technology, specifically to a method and system for generating traffic simulation files based on natural language syntax parsing. Background Technology

[0002] Natural language input is gradually being introduced into the construction of simulation scenarios in the field of traffic engineering. Existing file conversion methods rely on fixed syntax mapping rules. When there is a matching conflict between the controlled phase command described by the external text and the underlying physical road network, the existing system lacks an automatic logical correction path for the lane steering matrix. The basic configuration conflict directly leads to the suspension of the model construction process.

[0003] Traffic simulation requires the collaborative generation of various configuration files, such as road network, timing, and routing. In the multi-stage parsing and generation process, the existing business architecture has not established a temporary data isolation and verification mechanism. Local parsing anomalies in the front-end syntax generate erroneous parameters. These erroneous parameters are injected directly into the main business line without being intercepted, causing a global interruption of the simulation main process. When the existing system encounters data gaps, it lacks business fault tolerance and baseline configuration takeover capabilities, making it unable to maintain the basic operating state of the system.

[0004] In the evaluation index calculation phase after the simulation runs, the data output by the underlying simulation process exhibits heterogeneous characteristics. The discrete event-driven logs and the fixed-period time-step logs have not yet established a unified benchmark node in the time dimension. The raw log streams that have not been synchronized with timestamps will cause data misalignment when summarizing the micro-vehicle crossing times at intersections, resulting in a comparison error between the actual travel time and the theoretical free-flow time, causing the average delay evaluation index of the road network to deviate from the true calculation benchmark. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for generating traffic simulation files based on natural language syntax parsing. This method solves the problems of business flow interruption and evaluation errors caused by the lack of logical conflict correction, fault tolerance takeover mechanism, and misalignment of heterogeneous log time dimensions during the natural language to traffic simulation file conversion process.

[0006] To achieve the above objectives, the first aspect of this invention provides a method for generating traffic simulation files in conjunction with natural language syntax parsing, the method comprising the following steps:

[0007] Receive a sequence of natural language text input from outside, calculate the domain lexical density of the natural language text sequence, and when the domain lexical density meets a preset threshold, allocate a circuit breaker token bucket to the current task to trigger the data to flow backward.

[0008] Parse natural language text sequences to extract intersection topology parameters, instantiate global topology intermediate state objects, generate basic road network configuration files in temporary isolation sandboxes and record file status hash values;

[0009] Parse natural language text sequences to obtain signal timing instructions, compare the signal timing instructions with the lane-level steering mapping relationship inside the global topology intermediate object, modify lane attributes and phase control type when there are lane physical attribute conflicts, calculate the maximum theoretical capacity boundary value and update it synchronously inside the global topology intermediate object, generate timing configuration file and record file status hash value;

[0010] Parse natural language text sequences to obtain expected input traffic, determine whether the expected input traffic exceeds the corresponding maximum theoretical throughput boundary value, trigger a spatiotemporal peak shaping algorithm to allocate discrete delay amount if it exceeds the state, generate a routing configuration file based on the discrete delay amount and record the file status hash value;

[0011] Verify the file status hash values ​​of the basic road network configuration file, timing configuration file, and routing configuration file in the temporary isolation sandbox. Combine the comparison results of the file status hash values ​​with the status of the circuit breaker token bucket to synthesize the simulation master control file and start the traffic simulation underlying process.

[0012] The event-driven logs and time-step logs output by the underlying process of traffic simulation are analyzed to calculate the average delay index and generate a formatted evaluation report.

[0013] Furthermore, the steps for calculating the domain lexical density of a natural language text sequence include: converting the natural language text sequence into a continuous character sequence, segmenting it into independent lexical units and removing stop words to obtain a target text sequence composed of lexical units; counting the number of lexical units in the target text sequence that accurately match the traffic engineering domain lexical database; and calculating the domain lexical density based on the number of lexical units and the total number of lexical units in the target text sequence.

[0014] Furthermore, the steps of extracting intersection topology parameters and instantiating a global topology intermediate object include: converting the natural language text sequence into a target text sequence, and reorganizing the target text sequence into a structured instruction sequence that conforms to predefined entity relationships; using a finite state machine parser to trigger internal state transition mapping based on the direction and numerical terms in the structured instruction sequence, extracting intersection topology parameters containing physical lane quantity parameters and vehicle steering action definition parameters; initializing a lane-level steering mapping matrix within the global topology intermediate object, writing the independent index number of the physical lane and the corresponding allowed vehicle steering actions of the physical lane into the lane-level steering mapping matrix to form a key-value binding mapping, thus constituting a lane-level steering mapping relationship.

[0015] Furthermore, the steps for modifying lane attributes and phase control types in the presence of lane physical attribute conflicts include: parsing signal timing instructions to extract the controlled phase direction; comparing the controlled phase direction with the corresponding driving direction enumeration value in the lane-level steering mapping relationship; and determining that a physical lane support conflict is generated as a lane physical attribute conflict when an independent physical lane index binding relationship is detected where the controlled phase direction lacks a corresponding driving direction enumeration value in the lane-level steering mapping relationship.

[0016] Within the global topology intermediate object, the independent index number of the physical lane adjacent to the missing physical lane is retrieved, and the single driving direction enumeration value corresponding to the independent index number of the adjacent physical lane is modified to a hybrid steering attribute. The control command for the original protected phase is reconstructed into a permissible avoidance phase control command. Through lane attribute modification and command reconstruction, when the controlled phase defined by the natural language command conflicts with the actual physical lane distribution at the intersection, the hybrid steering attribute of the adjacent lane is dynamically adjusted to provide physical support and maintain the coherence of the underlying road network logic of the simulation.

[0017] Further steps for calculating the maximum theoretical capacity boundary value include: extracting the green light duration allocated to each physical lane; querying the capacity standard chart based on lane width and speed limit parameters to obtain the basic saturation flow rate of a single physical lane; and calculating the maximum theoretical capacity boundary value of each approach lane of the intersection based on the green light duration, total cycle time, and basic saturation flow rate.

[0018] Furthermore, the steps for triggering the spatiotemporal peak shaving algorithm to allocate discrete delays under the over-state include: summing the expected input flow into a total expected input flow; calculating the overflow ratio by dividing the difference between the total expected input flow and the maximum theoretical capacity boundary value by the maximum theoretical capacity boundary value; extracting the predefined peak dissipation time window constant and traffic flow travel delay coefficient; combining the peak dissipation time window constant, overflow ratio, and traffic flow travel delay coefficient with independent pseudo-random variables, substituting them into a nonlinear flow delay model based on negative exponential decay characteristics, and calculating the nonlinear delay offset as the discrete delay.

[0019] Furthermore, the steps to verify the file status hash values ​​of the basic road network configuration file, timing configuration file, and routing configuration file in the temporary isolation sandbox, and to synthesize the simulation master control file by combining the comparison results of the file status hash values ​​with the status of the circuit breaker token bucket, and to start the traffic simulation underlying process include: reading the historical hash value sequence recorded inside the global topology intermediate state object in the system memory pool; and performing a bit-by-bit comparison operation between the file status hash values ​​of each configuration file generated in real time and the historical hash value sequence.

[0020] If a difference is detected, execute a file system-level destruction command to forcibly clear the temporary configuration file in the temporary isolation sandbox and decrement the remaining retry count in the circuit breaker token bucket; if the remaining retry count is detected to be zero, extract the preset security template parameters to forcibly overwrite the official execution directory and generate a baseline configuration file.

[0021] Based on the baseline configuration file, the simulation master control file is synthesized, and the traffic simulation underlying process is started. By comparing hash sequences and monitoring the token bucket status, when an anomaly is detected in the parsed configuration file, the sandbox is cleared and retried. After the number of retries is exhausted, the preset safety template parameters are extracted and the official execution directory is forcibly overwritten, thereby providing a fault-tolerant takeover mechanism to prevent the simulation master process from being interrupted due to configuration file parsing errors.

[0022] Furthermore, the steps for analyzing the event-driven logs and time-step logs output by the underlying process of traffic simulation include: extracting the preset evaluation time window constant; using a linear interpolation algorithm to align the timestamps in the two heterogeneous data streams, event-driven logs and time-step logs, to the discrete time grid nodes defined by the preset evaluation time window constant. Since event-driven logs have discrete instantaneous characteristics, while time-step logs have fixed periodic characteristics, the timestamps of the two heterogeneous logs are synchronously mapped to the discrete time grid nodes using a linear interpolation algorithm to establish a unified data evaluation time benchmark.

[0023] Furthermore, the steps for calculating the average delay index include: at each discrete-time grid node, summarizing the actual crossing time of individual vehicles within the corresponding preset evaluation time window constant, as well as the theoretical crossing time at free-flow speed; calculating the average delay index within each preset evaluation time window constant based on the actual crossing time and the theoretical crossing time at free-flow speed; comparing the average delay index with the traffic service level rating table using threshold mapping, determining the service level letter rating, and generating a formatted evaluation report.

[0024] A second aspect of this invention provides a traffic simulation file linkage generation system based on natural language syntax parsing, the system comprising:

[0025] The orchestration module is configured to receive natural language text sequences from external input, calculate the domain lexical density of the natural language text sequence, and allocate a circuit breaker token bucket to the current task when the domain lexical density meets a preset threshold, triggering the data to flow backward.

[0026] The road network generation module is configured to parse natural language text sequences to extract intersection topology parameters, instantiate a global topology intermediate state object, generate a basic road network configuration file in a temporary isolation sandbox, and record the file status hash value.

[0027] The signal timing module is configured to parse natural language text sequences to obtain signal timing instructions, compare the signal timing instructions with the lane-level steering mapping relationship inside the global topology intermediate object, modify lane attributes and phase control type when there is a conflict in lane physical attributes, calculate the maximum theoretical capacity boundary value and update it synchronously inside the global topology intermediate object, generate a timing configuration file and record the file status hash value.

[0028] The traffic flow generation module is configured to parse natural language text sequences to obtain the expected input flow, determine whether the expected input flow exceeds the corresponding maximum theoretical capacity boundary value, trigger a spatiotemporal peak shaving algorithm to allocate discrete delays if the flow exceeds the limit, generate a routing configuration file based on the discrete delays and record the file status hash value.

[0029] The simulation master control module is configured to verify the file status hash values ​​of the basic road network configuration file, timing configuration file, and routing configuration file in the temporary isolation sandbox, combine the comparison results of the file status hash values ​​with the status of the circuit breaker token bucket to synthesize the simulation master control file, and start the traffic simulation underlying process.

[0030] The evaluation and analysis module is configured to parse the event-driven logs and time-step logs output by the underlying process of traffic simulation, calculate the average delay index, and generate a formatted evaluation report.

[0031] This invention provides a method and system for generating traffic simulation files in conjunction with natural language parsing. It offers the following advantages:

[0032] 1. This invention compares the controlled phase direction in the signal timing command with the lane-level steering mapping relationship. In the case of a conflict caused by the lack of physical lane support, it modifies the adjacent physical lanes to a hybrid steering attribute and reconstructs the phase control command. When the attribute modification and command reconstruction operations do not match the external natural language input and the basic road network physical structure, they provide an internal logic correction path to avoid the failure of the underlying simulation model construction due to the conflict of basic physical configuration, thus ensuring the continuity of road network scene generation.

[0033] 2. This invention performs hash sequence comparison on the generated configuration file in a temporary isolation sandbox, combines the circuit breaker token bucket mechanism to perform abnormal retries, and forcibly extracts the preset security template parameters to overwrite the formal execution directory when the retry count reaches zero. The file verification and fault tolerance process effectively isolates the data risks caused by front-end syntax parsing errors. In the event of a serious exception, the baseline configuration is forcibly retrieved to take over the subsequent generation tasks, preventing local parsing errors from causing a global interruption of the simulation main control process and maintaining the basic operating capabilities of the system.

[0034] 3. This invention introduces a linear interpolation algorithm to timestamp and synchronize the event-driven logs with discrete instantaneous characteristics and the time-step logs with fixed periodic characteristics at the underlying output. The logs are then uniformly mapped to discrete time grid nodes defined by a preset evaluation time window. The time grid alignment operation eliminates the time dimension misalignment between heterogeneous log data streams, establishes a standardized benchmark for summarizing the actual crossing time of individual micro-vehicles, and ensures the accuracy of the final calculation output of the average delay evaluation index. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the system timing interaction of the present invention;

[0036] Figure 2 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0037] Figure 3 Generate an interactive schematic diagram for the basic road network configuration of this invention;

[0038] Figure 4 This is a timing diagram illustrating the timing and routing configuration generation of the present invention;

[0039] Figure 5 This is a schematic diagram of the verification and fault tolerance process of the simulation main control module of the present invention;

[0040] Figure 6 Figure 1 is a schematic diagram of the mapping between the nonlinear delay offset and the average delay index data generated by the spatiotemporal peak shaving algorithm of the present invention. Figure 2 shows the distribution of the nonlinear delay offset and departure time series based on the nonlinear traffic congestion model, and Figure 3 shows the mapping comparison between the average delay index on the discrete time grid node and the threshold of the traffic service level table. Detailed Implementation

[0041] The technical solutions in 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] See appendix Figure 1 This invention provides a traffic simulation file linkage generation system based on natural language syntax parsing. The system includes:

[0043] The orchestration module is configured to receive natural language text sequences from external input, perform domain lexical density pre-validation, and initialize distributed circuit breaker token buckets for legitimate tasks.

[0044] The road network generation module is configured to extract basic intersection parameters through preset grammar rules, instantiate a global topology intermediate object, and serialize and generate a basic road network configuration file in a temporary isolation sandbox.

[0045] The signal timing module is configured to read the global topology intermediate state object to perform cross-domain physical lane support verification, perform attribute reconstruction triggered by physical lane support conflict, calculate the maximum theoretical capacity boundary value, and backpropagate the maximum theoretical capacity boundary value to the global topology intermediate state object.

[0046] The traffic flow generation module is configured to calculate the expected input flow based on natural language description, read the maximum theoretical capacity boundary value to perform threshold comparison, and trigger a spatiotemporal peak shaving algorithm to perform departure time delay allocation for overflowing vehicles.

[0047] The simulation master control module is configured to initiate file status checks and hash comparisons based on the model context protocol, execute a two-phase commit transaction mechanism and rollback of isolated files, trigger a forced overwrite of baseline parameters when retry resources are exhausted, and synthesize the master control file to start the underlying simulation process.

[0048] The evaluation and analysis module is configured to parse multi-source heterogeneous raw logs in parallel after the simulation process ends, perform time series alignment, and calculate traffic delay indicators.

[0049] See appendix Figure 2 This invention provides a method for generating traffic simulation files in conjunction with natural language parsing, the method comprising the following steps:

[0050] S1, relying on a distributed architecture, the orchestration module, as the entry node, receives natural language text sequences from external input, performs an initial screening of the legality of input instructions, and calculates the domain lexical density of the natural language text sequence; when the domain lexical density meets a preset threshold, it allocates a circuit breaker token bucket containing the initial number of retries to the current task, and triggers the data to flow sequentially after verification;

[0051] S2, each microservice module is physically isolated from each other, and establishes asynchronous communication connections at the logical level based on the model context protocol; the road network generation module extracts intersection topology parameters through a finite state machine, establishes the initial structure of the global topology intermediate object, and instantiates the intersection topology parameters into a global topology intermediate object residing in the system memory pool. The global topology intermediate object serves as the unique state benchmark for cross-module data interaction; the road network generation module generates a road network configuration file in a temporary isolation sandbox based on the global topology intermediate object and records the file state hash value;

[0052] S3, the signal timing module parses the signal timing command, performs status verification based on the global topology intermediate state object, and compares the lane-level steering mapping relationship between the signal timing command and the global topology intermediate state object; in the case of lane physical attribute conflict, it modifies the lane attributes and phase control type, calculates the maximum theoretical capacity boundary value of each approach lane of the intersection based on the timing parameters, and synchronously updates the maximum theoretical capacity boundary value as the physical capacity parameter to the global topology intermediate state object, generates a timing configuration file in the temporary isolation sandbox and records the file status hash value;

[0053] S4, the traffic flow generation module parses the traffic flow instructions to obtain the expected input flow in each direction, reads the updated global topology intermediate state object parameters to constrain the flow allocation strategy, and forms a mesh data flow mechanism; it determines whether the expected input flow exceeds the corresponding maximum theoretical capacity boundary value, calculates the overflow ratio in the case of exceeding the limit, allocates discrete delays to vehicle departure timestamps based on the overflow ratio, and generates a routing configuration file in a temporary isolation sandbox and records the file status hash value.

[0054] S5, during the period when the underlying simulation files are uniformly written to the temporary isolation sandbox, the simulation master control module monitors the generation status of the components within the sandbox, verifies the status hash value of each configuration file in the sandbox, and executes the transaction commit command to migrate the component to the execution directory when the component is confirmed to be complete and logically self-consistent. When a logical break is detected, i.e., the resolution of any microservice module fails, the simulation master control module triggers a linkage rollback to destroy the sandbox files and reduces the remaining retries in the circuit breaker token bucket. When the remaining retries reach zero, the simulation master control module uses the token status maintained by the orchestration module to intervene and control, intercepts subsequent resolution requests, extracts the baseline parameters of the pre-set security template to take over the system generation process, forcibly overwrites the sandbox directory to generate the baseline configuration file, and starts the traffic simulation underlying process.

[0055] S6, the evaluation and analysis module collects the event-driven logs and time-step logs output by the underlying process of traffic simulation, aligns the heterogeneous data streams to the preset evaluation time window, calculates the average delay index, and outputs a formatted evaluation report.

[0056] See appendix Figure 3 In specific implementations, step S1 provided by the present invention may include the following steps:

[0057] S11, the orchestration module receives the natural language text sequence transmitted from the external system, converts the natural language text sequence into a continuous character sequence, and uses a natural language processing tool to perform word segmentation on the continuous character sequence, dividing it into independent lexical units. The natural language processing tool is implemented using an open-source word segmentation component based on a hidden Markov model or a conditional random field. After word segmentation is completed, the orchestration module calls the stop word list to remove prepositions, conjunctions and regular modifiers from the natural language text sequence, and obtains the target text sequence composed of words.

[0058] S12, the orchestration module loads a pre-built traffic engineering vocabulary in memory. This vocabulary is based on national standards for road traffic engineering terminology and industry specifications. It stores specialized technical entity entries covering road channelization design, signal phase attributes, and traffic flow theory. The orchestration module iterates through the target text sequence, counting the number of words in the sequence that precisely match the traffic engineering vocabulary. Based on this count and the total number of words in the target text sequence, the orchestration module calculates the domain lexical density. The formula for calculating the domain lexical density is:

[0059] ;

[0060] In the formula, Domain lexical density; This represents the total number of specialized terms matched to the traffic engineering vocabulary database in the target text sequence. This represents the total number of words in the target text sequence after word segmentation and filtering.

[0061] The orchestration module compares the domain lexical density with a set minimum density threshold. The minimum density threshold is set based on the empirical value of the lower limit of text density extracted from historical test samples with complete topological parameters. In this embodiment, the value of the minimum density threshold ranges from 0.15 to 0.30. When the value of the domain lexical density is lower than the minimum density threshold, the orchestration module determines that the input natural language text sequence lacks the intention of traffic simulation engineering instructions, actively intercepts the current task request and blocks the flow of data to downstream microservice modules.

[0062] S13, when the domain lexical density is greater than or equal to the minimum density threshold, the orchestration module determines that the natural language text sequence is a legitimate task instruction. The orchestration module reads the globally unique identifier corresponding to the legitimate task instruction and instantiates a circuit breaker token bucket for the legitimate task instruction in the distributed shared memory pool.

[0063] The orchestration module allocates the maximum allowed retries as the initial capacity to the circuit breaker token bucket based on the preset system fault tolerance level. The maximum allowed retries are set according to the concurrent load pressure. In this embodiment, the maximum allowed retries range from 3 to 5. The maximum allowed retries define the upper limit of error messages that the microservice module can send to external input terminals. During the initialization phase, the orchestration module sets the remaining available retries in the circuit breaker token bucket as the maximum allowed retries, synchronously records the system timestamp of task creation, and completes the pre-transaction state configuration. After the state configuration is completed, the orchestration module passes the target text sequence to the downstream road network generation module.

[0064] In specific implementations, step S2 provided by this invention may include the following steps:

[0065] S21, the road network generation module receives the target text sequence transmitted by the orchestration module and uses a pre-built semantic dependency analysis algorithm to reorganize the target text sequence into a structured instruction sequence that conforms to predefined entity relationships. The pre-built semantic dependency analysis algorithm is implemented based on an open-source natural language processing toolkit and is pre-trained using a manually labeled traffic engineering instruction dataset to accurately identify the dependency relationships between traffic entities in the external text. The road network generation module calls the pre-built Backus paradigm-based grammar rules and finite state machine parser to perform word-by-word reading and scanning of the structured instruction sequence. The Backus paradigm-based grammar rules define the syntactic hierarchy spanning tree structure of road edges and lane attributes at intersection nodes through terminal symbols and non-terminal symbols. The finite state machine parser has pre-built state nodes with corresponding spatial orientation attributes and numerical attributes.

[0066] The finite state machine parser triggers internal state transition mapping based on the direction and numerical terms in the structured instruction sequence, completing continuous jumps from the initial parsing state to the parameter extraction termination state. The road network generation module gradually extracts the intersection center coordinate parameters and the direction attributes of each approach lane based on the state transition path, and simultaneously extracts the physical lane quantity parameters and the vehicle turning action definition parameters allowed for each physical lane corresponding to each approach lane.

[0067] S22, the road network generation module instantiates a global topology intermediate object in the memory pool based on the extracted topology parameters. The global topology intermediate object is constructed as a nested data structure based on extensible markup language syntax, containing multi-level attributes. The global topology intermediate object internally constructs a node coordinate matrix that records the absolute spatial position of intersections, as well as an edge set attribute that records the connection relationship of each road branch. The road network generation module initializes a lane-level steering mapping matrix within the global topology intermediate object, and writes the independent index number of each physical lane and the vehicle steering action allowed for the corresponding physical lane into the lane-level steering mapping matrix to form a key-value binding mapping. The key-value binding mapping uses the independent index number of the physical lane as the primary key and the enumerated value of the driving direction as the key value. The global topology intermediate object maps the connectivity constraints of the traffic simulation physical road network through the lane-level steering mapping matrix.

[0068] S23, the road network generation module reads the attribute data from the global topology intermediate object, and serializes it into a basic road network configuration file in a temporary isolation sandbox according to the extensible markup language data format preset by the underlying simulation engine. The basic road network configuration file maps the physical space topology structure defined by the global topology intermediate object with static nodes and edge elements. The road network generation module calls a secure hash algorithm to perform data digest extraction operation on the generated basic road network configuration file to generate a file status hash value. The road network generation module writes the generated file status hash value back into the global topology intermediate object for parameter binding and recording. The global topology intermediate object remains in a resident state in the memory pool, triggering data flow to downstream microservice modules.

[0069] See appendix Figure 4 In specific implementations, step S3 provided by the present invention may include the following steps:

[0070] S31, the signal timing module receives a target text sequence containing system control parameters, parses and extracts the total cycle duration and controlled phase direction of the signal control scheme, and reads the lane-level steering mapping matrix inside the global topology intermediate object residing in the system memory pool across microservice nodes. The signal timing module compares the controlled phase direction with the driving direction enumeration value bound to the corresponding physical lane independent index number in the lane-level steering mapping matrix. The driving direction enumeration value covers the standard motor vehicle driving trajectory options for left turn, straight, right turn, and U-turn. During the comparison process, if it is detected that the controlled phase direction is missing the independent physical lane index binding relationship of the corresponding driving direction enumeration value in the lane-level steering mapping matrix, the signal timing module determines that a physical lane support conflict has occurred.

[0071] S32, after a physical lane support conflict is triggered, the signal timing module searches for the independent index number of the physical lane adjacent to the missing physical lane within the global topology intermediate state object. The signal timing module modifies the single driving direction enumeration value corresponding to the independent index number of the adjacent physical lane to a mixed steering attribute that includes the controlled phase direction. The signal timing module simultaneously modifies the signal control scheme, refactoring the control command for the original protected phase into a permissible avoidance phase control command applicable to the mixed lane. The signal timing module updates the modified mixed steering attribute and permissible avoidance phase control command into the global topology intermediate state object and performs a topology semantic downgrade reconstruction operation.

[0072] S33, the signal timing module extracts the green light duration allocated to each physical lane based on the reconstructed signal control scheme. The signal timing module also queries the system's preset capacity standard chart based on lane width and speed limit parameters to obtain the basic saturation flow rate for each physical lane. Based on the green light duration, total cycle time, and basic saturation flow rate, the signal timing module calculates the maximum theoretical capacity boundary value for each approach lane of the intersection. The formula for calculating the maximum theoretical capacity boundary value is as follows:

[0073] ;

[0074] In the formula, For the intersection The maximum theoretical capacity boundary value of each approach lane; The index number of the approach lane at the intersection; The index number of the physical lane; For the first The total number of physical lanes included in each entrance lane; For the first The basic saturation flow rate of each physical lane; To be assigned to the The green light duration for each physical lane; This represents the total cycle time of the signal control scheme.

[0075] S34, the signal timing module extracts the signal control parameters containing the green light duration and permission avoidance phase control instructions from the global topology intermediate state object. Based on the extensible markup language data format preset by the underlying simulation engine, the signal timing module serializes and generates a timing configuration file in a temporary isolation sandbox. The signal timing module calls the standard message digest algorithm to perform a hash operation on the timing configuration file, extracts the file status hash value of the timing configuration file, and appends the file status hash value of the timing configuration file to the global topology intermediate state object to perform a cross-node transaction state binding operation.

[0076] In specific implementations, step S4 provided by this invention may include the following steps:

[0077] S41, the traffic flow generation module receives and parses the target text sequence, extracts the macro total demand flow value of each approach lane of the intersection and the turning ratio coefficient of the corresponding driving direction, the traffic flow generation module uses the macro total demand flow value multiplied by the turning ratio coefficient of the corresponding driving direction to calculate the expected input flow of each approach lane for a single driving direction, the traffic flow generation module sums up the expected input flow of each single driving direction and performs a summation calculation to obtain the total expected input flow of each approach lane of the intersection.

[0078] S42, the traffic flow generation module reads the maximum theoretical capacity boundary value stored internally in the global topology intermediate state object across nodes. The traffic flow generation module performs a numerical comparison calculation between the total expected input flow and the maximum theoretical capacity boundary value. When the value of the total expected input flow is less than or equal to the maximum theoretical capacity boundary value, the traffic flow generation module directly generates the initial departure time sequence based on the expected input flow. When the value of the total expected input flow is greater than the maximum theoretical capacity boundary value, the traffic flow generation module triggers the spatiotemporal peak shaving algorithm, and calculates the overflow ratio of the approach lane by dividing the difference between the total expected input flow and the maximum theoretical capacity boundary value by the maximum theoretical capacity boundary value.

[0079] S43, the traffic flow generation module executes a spatiotemporal peak shaving algorithm, introducing a nonlinear flow retardation model based on negative exponential decay characteristics. The traffic flow generation module extracts a predefined peak dissipation time window constant and a traffic flow retardation coefficient. The peak dissipation time window constant is set based on the statistical mean of intersection queue clearing time. In this embodiment, the value range of the peak dissipation time window constant is 200s to 300s. The traffic flow retardation coefficient is set according to the road geometry level, representing the departure queuing friction under congestion conditions. In this embodiment, the value range of the traffic flow retardation coefficient is 0.5 to 0.8. The traffic flow generation module combines the peak dissipation time window constant, overflow ratio, and traffic flow retardation coefficient with independent pseudo-random variables, and substitutes them into the nonlinear flow retardation model to calculate the nonlinear delay offset. In this embodiment, the value range of the independent pseudo-random variables is 0 to 1. The calculation formula for the nonlinear delay offset is:

[0080] ;

[0081] In the formula, This is a non-linear delay offset; The peak dissipation time window constant; This represents the overflow ratio. This is the traffic flow obstruction coefficient; They are independent pseudo-random variables.

[0082] S44, the traffic flow generation module serializes and generates a routing configuration file in a temporary isolation sandbox based on the departure time sequence containing discrete distribution characteristics and the lane index mapping relationship in the global topology intermediate object. The routing configuration file records the entry time and path attributes of individual micro-vehicles in the simulated traffic network. The traffic flow generation module calls the standard message digest algorithm to calculate the file status hash value of the routing configuration file. The traffic flow generation module records the file status hash value of the routing configuration file into the global topology intermediate object, and together with the file status hash values ​​of the basic road network configuration file and the timing configuration file, they form a historical hash value sequence. The system maintains the data state of the global topology intermediate object in the memory pool and receives transaction verification instructions from the simulation master control module.

[0083] See appendix Figure 5 In specific implementations, step S5 provided by the present invention may include the following steps:

[0084] S51, the simulation master control module scans the temporary isolation sandbox, reads the basic road network configuration file, timing configuration file, and routing configuration file. The simulation master control module calls the standard message digest algorithm to re-perform hash operation on the basic road network configuration file, timing configuration file, and routing configuration file to obtain the real-time generated file status hash value. The simulation master control module reads the historical hash value sequence recorded inside the global topology intermediate state object in the system memory pool across microservice nodes, and performs a bit-by-bit comparison operation between the real-time generated file status hash value and the historical hash value sequence. When it is detected that the real-time generated file status hash value and the historical hash value sequence are completely consistent, the simulation master control module executes the commit instruction of the two-phase commit transaction mechanism to migrate the basic road network configuration file, timing configuration file, and routing configuration file in the temporary isolation sandbox to the formal execution directory.

[0085] S52, when a difference is detected between the real-time generated file status hash value and the historical hash value sequence during the bit-by-bit comparison operation, the simulation master control module determines that a logical break has occurred due to the failure of the microservice module to parse. The simulation master control module triggers the linkage rollback mechanism and executes the file system-level destruction command to forcibly clear the temporary configuration file in the temporary isolation sandbox.

[0086] The simulation master control module sends a synchronization update command to the system memory pool, calls the orchestration module to decrement the remaining retry count in the circuit breaker token bucket, the orchestration module updates the capacity status of the circuit breaker token bucket, and sends a correction prompt carrying diagnostic information of the parsing failure node to the external input terminal. The orchestration module opens the task pipeline and receives the natural language text sequence parsing request adjusted by the external input terminal based on the correction prompt.

[0087] S53, the simulation master control module periodically polls the circuit breaker token bucket capacity status maintained by the orchestration module. When it detects that the remaining retry count in the circuit breaker token bucket has returned to zero, the simulation master control module intercepts subsequent parsing requests initiated by external input terminals. The simulation master control module extracts preset security template parameters from the local repository. The preset security template parameters are derived from the stable operating conditions of typical trunk line intersections.

[0088] The preset safety template parameters internally include a standard crossroads topology, a preset fixed signal control cycle, and a preset off-peak demand flow value. In this embodiment, the value range of the preset fixed signal control cycle is 90s to 150s, and the value range of the preset off-peak demand flow value is 400pcu / h to 800pcu / h. The simulation main control module uses the preset safety template parameters to forcibly overwrite the formal execution directory and generate a baseline configuration file, and then takes over the scenario generation process based on the baseline configuration file.

[0089] S54, the simulation master control module synthesizes the simulation master control file based on the basic road network configuration file, timing configuration file, routing configuration file or the baseline configuration file generated by forced overwrite in the formal execution directory. The simulation master control file encapsulates the physical time step, random number seed and absolute storage path of output log required for the operation of the traffic simulation underlying process.

[0090] The physical time step is set according to the update frequency of the micro car-following model. In this embodiment, the physical time step ranges from 0.1s to 1s, and the random number seed ranges from 1 to 9999. The simulation main control module calls the command line interface of the traffic simulation underlying process and passes in the path address of the simulation main control file to trigger the process creation function to start the traffic simulation underlying process. The traffic simulation underlying process occupies computing resources to perform iterative calculations of the vehicle micro car-following and lane-changing models until the set simulation termination time is reached.

[0091] In specific implementations, step S6 provided by the present invention may include the following steps:

[0092] S61, after the iterative calculation of the traffic simulation underlying process terminates, the evaluation and analysis module scans the output log storage path defined in the simulation master control file, and reads the event-driven log and time-step log output by the traffic simulation underlying process in parallel. The event-driven log records the instantaneous timestamp and instantaneous speed attributes of individual micro-vehicles crossing virtual loop detectors, and the time-step log records the spatial coordinate system position and acceleration attributes of individual micro-vehicles running in the road network within a fixed period. The evaluation and analysis module parses the event-driven log and time-step log to extract structured time-series data of vehicle trajectories.

[0093] S62, the evaluation and analysis module extracts the preset evaluation time window constant. The preset evaluation time window constant is set according to the statistical stationarity period of the macro traffic flow state. In this embodiment, the value range of the preset evaluation time window constant is 300s to 900s. The evaluation and analysis module uses a linear interpolation algorithm to align the timestamps in the two heterogeneous data streams, event-driven logs and time-step logs, to the discrete time grid nodes defined by the preset evaluation time window constant. At each discrete time grid node, the evaluation and analysis module summarizes the actual crossing time of individual micro vehicles within the corresponding preset evaluation time window constant, as well as the theoretical crossing time under free-flow speed.

[0094] S63, the evaluation and analysis module calculates the average delay index within each preset evaluation time window constant based on the actual crossing time at the intersection and the theoretical crossing time under free-flow velocity. The formula for calculating the average delay index is as follows:

[0095] ;

[0096] In the formula, The average delay index is defined as the average delay within a preset evaluation time window constant. For individual micro-vehicles; The total number of individual micro-vehicles that leave the traffic simulation physical road network within a preset evaluation time window constant; For the first The actual time it takes for an individual vehicle to cross an intersection; For the first The theoretical time span of a microscopic vehicle at free-flow velocity.

[0097] S64, the evaluation and analysis module compares the obtained average delay index with the defined traffic service level table by threshold mapping. The traffic service level table discretizes the intersection delay values ​​and maps them into a six-level evaluation system consisting of letters A to F. The evaluation and analysis module determines the service level letter rating corresponding to each preset evaluation time window constant, generates a formatted evaluation report, and checks the historical action records of the simulation main control module.

[0098] When the system detects that the underlying traffic simulation process is forced to start based on preset safety template parameters, the evaluation and analysis module adds a downgraded status warning label to the header area of ​​the formatted evaluation report, and then outputs the formatted evaluation report to an external display terminal.

[0099] See appendix Figure 6 To aid in understanding the technical solution of this invention, the following provides application examples of a method and system for generating traffic simulation files based on natural language syntax parsing.

[0100] The orchestration module receives the natural language text sequence from external input and converts it into a character sequence to perform word segmentation and filtering. The orchestration module uses a vocabulary database in the field of traffic engineering to count the number of professional words hit in the target text sequence to calculate the domain lexical density. If the value of the domain lexical density is greater than or equal to the minimum density threshold, the target text sequence is determined to be a valid task instruction. The orchestration module instantiates a circuit breaker token bucket for the valid task instruction and allocates the maximum allowed number of retries as the initial capacity. The orchestration module completes the configuration of the preceding transaction state and passes the target text sequence to the downstream road network generation module.

[0101] The road network generation module parses the target text sequence to extract the intersection center coordinate parameters, the physical lane quantity parameters corresponding to each approach lane, and the steering action definition parameters. Based on the extracted topology parameters, the road network generation module instantiates a global topology intermediate state object and initializes a lane-level steering mapping matrix to bind the lane independent index number and driving direction enumeration value. The road network generation module reads the global topology intermediate state object and serializes it in a temporary isolation sandbox to generate a basic road network configuration file. The road network generation module performs a data digest extraction operation on the basic road network configuration file to obtain the file status hash value and writes it back into the global topology intermediate state object for parameter binding and recording.

[0102] The signal timing module parses the target text sequence to extract the controlled phase direction and compares it with the lane-level steering mapping matrix inside the global topology intermediate object. When a physical lane support conflict is triggered when a missing corresponding binding relationship is detected, the signal timing module reconstructs the adjacent physical lanes into mixed steering attributes and modifies the control command to a permitted avoidance phase control command, updating it inside the global topology intermediate object. The signal timing module calculates the maximum theoretical capacity boundary value of each approach lane of the intersection based on the green light duration and the basic saturation flow rate. The signal timing module generates a timing configuration file in a temporary isolation sandbox, extracts the file status hash value, and appends it to the global topology intermediate object.

[0103] The traffic flow generation module parses the target text sequence to obtain the expected input flow and summarizes it into the total expected input flow. The traffic flow generation module reads the maximum theoretical capacity boundary value and performs a numerical threshold comparison calculation on the total expected input flow. When the value of the total expected input flow is greater than the maximum theoretical capacity boundary value, the overflow ratio is calculated and the spatiotemporal peak shaving algorithm is triggered. The traffic flow generation module uses a nonlinear flow stagnation model to calculate the nonlinear delay offset to generate a departure time series containing discrete distribution characteristics. The traffic flow generation module generates a routing configuration file in a temporary isolation sandbox, extracts the file status hash value, and records it inside the global topology intermediate state object.

[0104] The simulation master control module reads the configuration file in the temporary isolation sandbox and performs a hash operation to obtain the real-time generated file status hash value. The simulation master control module performs a bitwise comparison operation on the real-time generated file status hash value with the historical hash value sequence recorded in the system memory pool. If a difference is detected, the linkage rollback mechanism is triggered to forcibly clear the temporary configuration file and decrement the remaining retry count in the circuit breaker token bucket. If the remaining retry count is detected to be zero, the preset security template parameters are extracted and the baseline configuration file is forcibly overwritten in the formal execution directory to generate a baseline configuration file. The simulation master control module synthesizes the simulation master control file based on the file in the formal execution directory or the baseline configuration file forcibly overwritten and starts the traffic simulation underlying process.

[0105] After the traffic simulation underlying process iterative calculation terminates, the evaluation and analysis module reads the output event-driven log and time step log and extracts the structured time series data of the vehicle's driving trajectory. The evaluation and analysis module uses a linear interpolation algorithm to align the timestamps in the two types of heterogeneous data streams to the discrete time grid nodes defined by the preset evaluation time window constant. The evaluation and analysis module summarizes the actual crossing time and the theoretical crossing time to calculate the average delay index. The evaluation and analysis module performs threshold mapping comparison between the average delay index and the traffic service level table to determine the service level letter rating and generates a formatted evaluation report to be output to the external display terminal.

[0106] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result using substantially the same method falls within the scope of the present invention.

Claims

1. A method for automatic generation of a traffic simulation file based on natural language syntax parsing, characterized in that, The method includes the following steps: Receive an externally input natural language text sequence, calculate the domain lexical density of the natural language text sequence, and when the domain lexical density meets a preset threshold, allocate a circuit breaker token bucket to the current task to trigger data flow. Parse the natural language text sequence to extract intersection topology parameters, instantiate a global topology intermediate state object, generate a basic road network configuration file in a temporary isolation sandbox and record the file status hash value; Parse the natural language text sequence to obtain signal timing instructions, compare the signal timing instructions with the lane-level steering mapping relationship inside the global topology intermediate object, modify lane attributes and phase control type when there is a lane physical attribute conflict, calculate the maximum theoretical capacity boundary value and update it synchronously inside the global topology intermediate object, generate a timing configuration file and record the file status hash value; The natural language text sequence is parsed to obtain the expected input traffic. It is determined whether the expected input traffic exceeds the corresponding maximum theoretical throughput boundary value. If the expected input traffic exceeds the boundary value, a spatiotemporal peak shaping algorithm is triggered to allocate discrete delay. A routing configuration file is generated based on the discrete delay and the file status hash value is recorded. Verify the file status hash values ​​of the basic road network configuration file, the timing configuration file, and the routing configuration file in the temporary isolation sandbox. Combine the comparison results of the file status hash values ​​with the status of the circuit breaker token bucket to synthesize the simulation master control file and start the traffic simulation underlying process. The event-driven logs and time-step logs output by the underlying traffic simulation process are analyzed to calculate the average delay index and generate a formatted evaluation report.

2. The method of claim 1, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The steps for calculating the domain lexical density of the natural language text sequence include: The natural language text sequence is converted into a continuous character sequence, segmented into independent lexical units, and stop words are removed to obtain the target text sequence composed of words; The number of words in the target text sequence that precisely match the vocabulary of the traffic engineering field is counted. The domain lexical density is calculated based on the number of words and the total number of words in the target text sequence.

3. The method of claim 1, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The steps of extracting intersection topology parameters and instantiating a global topology intermediate object include: The natural language text sequence is converted into a target text sequence, and the target text sequence is reorganized into a structured instruction sequence that conforms to predefined entity relationships; The finite state machine parser triggers internal state transition mapping based on the direction and numerical terms in the structured instruction sequence, and extracts the intersection topology parameters, which include physical lane quantity parameters and vehicle steering action definition parameters. The lane-level steering mapping matrix is ​​initialized within the global topology intermediate state object. The independent index number of the physical lane and the vehicle steering action allowed for the corresponding physical lane are written into the lane-level steering mapping matrix to form a key-value binding mapping, thus constituting the lane-level steering mapping relationship.

4. The method of claim 3, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The steps of comparing the signal timing command with the lane-level steering mapping relationship within the global topology intermediate state object, and modifying lane attributes and phase control types in the presence of lane physical attribute conflicts, include: The controlled phase direction is extracted by parsing the signal timing command; Compare the driving direction enumeration value corresponding to the controlled phase direction and the lane-level steering mapping relationship; If the controlled phase direction is found to be missing an independent physical lane index binding relationship corresponding to the driving direction enumeration value in the lane-level steering mapping relationship, a physical lane support conflict is determined as the lane physical attribute conflict. Within the global topology intermediate state object, retrieve the independent index number of the physical lane adjacent to the missing physical lane, and modify the single driving direction enumeration value corresponding to the independent index number of the adjacent physical lane to a hybrid steering attribute. The control commands for the original protected phase are refactored into permissioned avoidance phase control commands.

5. The method of claim 1, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The steps for calculating the boundary value of the maximum theoretical capacity include: Extract the green light duration allocated to each physical lane; Based on lane width and speed limit parameters, query the traffic capacity standard chart to obtain the basic saturation flow rate of a single physical lane; Based on the green light duration, total cycle time, and basic saturation flow rate, the maximum theoretical capacity boundary value of each approach lane of the intersection is calculated.

6. The method of claim 1, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The step of triggering the spatiotemporal peak clipping algorithm to allocate discrete delays under the out-of-state condition includes: The expected input traffic is aggregated into the total expected input traffic; The overflow ratio is calculated by dividing the difference between the total expected input flow and the maximum theoretical capacity boundary value by the maximum theoretical capacity boundary value. Extract the predefined peak dissipation time window constant and traffic flow travel delay coefficient; The peak dissipation time window constant, the overflow ratio, and the traffic flow travel stagnation coefficient are combined with independent pseudo-random variables and substituted into a nonlinear flow stagnation model based on negative exponential decay characteristics to calculate and obtain the nonlinear delay offset as the discrete delay.

7. The method of claim 1, wherein the natural language grammar-based parsing of the traffic simulation file association generation is based on a set of rules. The steps for verifying the file status hash values ​​of the basic road network configuration file, the timing configuration file, and the routing configuration file in the temporary isolation sandbox, combining the comparison results of the file status hash values ​​with the circuit breaker token bucket status to synthesize the simulation master control file, and starting the traffic simulation underlying process include: Read the historical hash value sequence recorded inside the global topology intermediate state object in the system memory pool; A bitwise comparison operation is performed between the file status hash value of each configuration file generated in real time and the historical hash value sequence; If a discrepancy is detected, a file system-level destruction command is executed to forcibly clear the temporary configuration file in the temporary isolation sandbox, and the remaining retry count in the circuit breaker token bucket is decremented by one. If the remaining number of retries is detected to be zero, the preset security template parameters are extracted, the official execution directory is forcibly overwritten, and a baseline configuration file is generated. The simulation master control file is synthesized based on the baseline configuration file, and the underlying process of the traffic simulation is started.

8. The method for generating traffic simulation files based on natural language syntax parsing according to claim 1, characterized in that, The steps for parsing the event-driven logs and time-step logs output by the underlying process of the traffic simulation include: Extract the preset evaluation time window constant; Using a linear interpolation algorithm, the timestamps in the two heterogeneous data streams, the event-driven log and the time-step log, are aligned to the discrete time grid nodes defined by the preset evaluation time window constant.

9. The method for generating traffic simulation files based on natural language syntax parsing according to claim 8, characterized in that, The steps for calculating the average delay index include: At each of the discrete time grid nodes, the actual crossing time of the micro vehicle within the corresponding preset evaluation time window constant, and the theoretical crossing time under free flow velocity are summarized. Based on the actual crossing time at the intersection and the theoretical crossing time at the free-flow velocity, calculate the average delay index within each preset evaluation time window constant; The average delay index is compared with the traffic service level rating table using threshold mapping to determine the service level letter rating and generate a formatted evaluation report.

10. A traffic simulation file linkage generation system based on natural language syntax parsing, characterized in that, The system using the traffic simulation file generation method based on natural language syntax parsing as described in any one of claims 1 to 9 includes: The orchestration module is configured to receive a natural language text sequence from an external input, calculate the domain lexical density of the natural language text sequence, and when the domain lexical density meets a preset threshold, allocate a circuit breaker token bucket to the current task to trigger data flow. The road network generation module is configured to parse the natural language text sequence to extract intersection topology parameters, instantiate a global topology intermediate state object, generate a basic road network configuration file in a temporary isolation sandbox, and record the file status hash value. The signal timing module is configured to parse the natural language text sequence to obtain signal timing instructions, compare the signal timing instructions with the lane-level steering mapping relationship inside the global topology intermediate object, modify lane attributes and phase control type when there is a lane physical attribute conflict, calculate the maximum theoretical capacity boundary value and synchronously update it inside the global topology intermediate object, generate a timing configuration file and record the file status hash value. The traffic flow generation module is configured to parse the natural language text sequence to obtain the expected input flow, determine whether the expected input flow exceeds the corresponding maximum theoretical capacity boundary value, trigger a spatiotemporal peak shaving algorithm to allocate discrete delay amount if the condition exceeds the limit, generate a routing configuration file based on the discrete delay amount and record the file status hash value. The simulation master control module is configured to verify the file status hash values ​​of the basic road network configuration file, the timing configuration file, and the routing configuration file in the temporary isolation sandbox, combine the comparison results of the file status hash values ​​with the circuit breaker token bucket status to synthesize the simulation master control file, and start the traffic simulation underlying process. The evaluation and analysis module is configured to parse the event-driven logs and time-step logs output by the underlying traffic simulation process, calculate the average delay index, and generate a formatted evaluation report.