A city planning method based on digital twinning

Abstract

This invention discloses a digital twin-based urban planning method, specifically relating to the field of digital cities, to address the problem that operational observations during planning evaluation are easily affected by external disturbances, leading to unclear basis and difficulty in verification of proposed solutions. The method extracts planning action identifiers and scopes of action from planning action information, generating a snapshot of the city's state before the action. It collects operational observations within the scope of action to form an observation segment set, calculates boundary attenuation consistency and rhythm template deviation strength to determine a replacement segment set, and generates a baseline observation sequence based on the snapshot of the city's state before the action to replace it, thus obtaining a purification channel dataset. Based on the purification channel dataset, it generates an indicator judgment table and proposed solutions, and generates a comparison suggestion based on the observation segment set to form a difference description, which is then written into the evaluation task. The proposed solutions are imported into a digital twin simulation and a consistency check is performed. If the check passes, the proposed solutions are released and a purification certificate is generated; if the check fails, a supplementary data collection task is generated and the proposed solutions are frozen.

Description

Technical Field

[0001] This invention relates to the field of digital cities, and more specifically, to a digital twin-based urban planning method. Background Technology

[0002] In urban planning, an increasing trend is to continuously aggregate information on roads, buildings, municipal facilities, population activities, and operational status into a digital twin city synchronized with reality. Planners first simulate and evaluate plans within the twin city, generating adjustment suggestions, and then gradually implement these plans in the construction and governance of the real city. After implementation, the platform continues to collect feedback information from the city's operation and management process to determine whether the plan has achieved its intended goals, triggering the next round of optimization and adjustments. In this way, urban planning transforms from a one-off process into a continuously iterative, closed-loop process. The platform must be able to both simulate plans and integrate post-implementation feedback into planning decisions, ensuring that the plan more closely reflects real-world operation.

[0003] However, existing technologies, such as the patent "Intelligent City Planning System and Method Based on Big Data Analysis" (application number: Chinese Patent Application No. 202410774536.2), disclose a closed-loop approach of implementing a planning scheme, obtaining monitoring information, using feedback results to evaluate the effect, and promoting further planning. This approach is prone to a subtle problem in actual operation: the platform directly uses the feedback information after implementation as the basis for the next round of prediction and evaluation, without distinguishing between changes caused by planning actions and changes caused by environmental fluctuations or external factors. Because the feedback information itself already contains the impact of planning measures, the system may mistakenly treat the changes caused by planning as naturally occurring patterns in subsequent iterations, thus continuously reinforcing existing practices during deduction and recommendation, leading to a trend of increasing deviation with each iteration. This deviation often doesn't appear suddenly, but accumulates gradually in multiple closed loops. On the surface, the platform keeps optimizing based on feedback, but in reality, it may lead the planning direction astray, causing subsequent spatial configuration and governance decisions to become increasingly dependent on the polluted input data. When the results do not meet expectations, it is difficult to quickly explain where the problem started and how it was transmitted to the final conclusion.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a digital twin-based urban planning method. This method extracts planning action identifiers and their scope of action from planning action information, generating a snapshot of the city's state before the action. It collects operational observations within the scope of action to form an observation segment set, calculates boundary attenuation consistency and rhythm template deviation strength to determine a replacement segment set, and generates a baseline observation sequence based on the snapshot of the city's state before the action to replace it, thus obtaining a purification channel dataset. Based on the purification channel dataset, it generates an indicator judgment table and scheme recommendations, and based on the observation segment set, it generates a comparison recommendation to form a difference description, which is then written into the evaluation task. The scheme recommendations are imported into the digital twin simulation and a consistency check is performed. If the check passes, the scheme recommendations are released and a purification certificate is generated; if the check fails, a supplementary collection task is generated and the scheme recommendations are frozen, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] S1: Receive planning action information, extract planning action identifier and scope of action, save a snapshot of the city status before the action, and bind the planning action identifier to the snapshot of the city status before the action.

[0008] S2: Collect operational observations and time markers within the scope of action, generate an observation segment set based on the planned action markers, output a replacement segment set based on the observation segment set, and replace the operational observations corresponding to the replacement segment set with the baseline observation sequence generated by the city status snapshot before the action to obtain the purification channel dataset;

[0009] S3: Generate effect judgment results and form solution suggestions based on the purification channel dataset, generate comparison suggestions based on the observation segment set, output difference descriptions, bind the difference descriptions to the planning action identifiers and replacement segment sets and write them into the evaluation task;

[0010] S4: The proposed solution and explanation of differences are entered into the digital twin simulation to generate expected changes. The expected changes and the effect judgment results are checked for consistency. If the check fails, a supplementary collection task is issued and the proposed solution is frozen. If the check passes, the proposed solution is released and a purification certificate is generated.

[0011] Furthermore, the planning action information is parsed to obtain the planning action identifier and scope of action. The scope of action is normalized by coordinates and standardized by boundary sequence to generate a scope of action hash value. The scope of action filters urban objects and performs coverage determination according to object type to solidify the object status table and form a snapshot of the urban status before the action. The snapshot of the urban status before the action is written into the snapshot index and boundary sequence verification is performed. The snapshot index records the planning action identifier, scope of action hash value, boundary sequence summary, snapshot version number and snapshot reference.

[0012] Furthermore, the operational observations and time stamps are sorted by time stamp and duplicate reported segments are removed. The operational observations are mapped to a finite state sequence according to the indicator definition in the planning and management rule base. The finite state sequence is divided into observation segment sets by state switching points. The observation segment set records the start and end time stamps of the segment, the finite state sequence, the observation summary, and the segment spatial coverage reference.

[0013] Furthermore, multiple spatial zones are generated inward from the boundary of the effective range. The observation segment set is aggregated according to the spatial zones to form a finite state sequence and a sequence of observation values. The contact point records are extracted from the zonal sequence and the contact point sequence and amplitude of adjacent spatial zones are compared to obtain the boundary attenuation consistency. The template fragment sequence is extracted from the urban state snapshot before the action and aligned with the current rhythm fragment sequence extracted from the observation segment set. The rhythm template deviation intensity is obtained by normalization based on the misaligned fragments and missing fragments.

[0014] Furthermore, the observation segment set generates rhythm median boundaries and boundary median boundaries based on the rhythm template deviation intensity and boundary attenuation consistency. Observation segments that satisfy the condition that the rhythm template deviation intensity is higher than the rhythm median boundary and the boundary attenuation consistency is lower than the boundary median boundary form a replacement segment set. The city status snapshot before the action establishes a caliber mapping table and generates a benchmark observation sequence. The running observation values ​​within the coverage area of ​​the replacement segment set are replaced by the benchmark observation sequence to form a purification channel dataset and written into the replacement trace index.

[0015] Furthermore, the purification channel dataset extracts continuous indicator sequences according to indicator caliber and aligns them with the segment boundaries of the observation segment set. The planning and management rule base executes judgment conditions on the continuous indicator sequences to generate a set of violation segments. If the set of violation segments is empty, it records the compliance judgment conclusion; if the set of violation segments is not empty, it records the non-compliance judgment conclusion. The segment with the longest duration is selected as the main evidence segment. The indicator judgment table solidifies the indicator caliber, judgment rule reference, judgment conclusion, and main evidence segment reference to form the effect judgment result.

[0016] Furthermore, the planning management rule base provides a suggested action type mapping table and suggested spatial location reference rules. When the indicator judgment table shows a non-compliance judgment conclusion, an adjustment item list is generated. The adjustment item list includes the adjustment object identifier, suggested action type, suggested spatial location reference, and suggested reason evidence reference. The suggested reason evidence reference is bound to the main evidence section reference and the replacement trace index reference. The adjustment item list is sorted by constraint level and main evidence section start time identifier to form a scheme suggestion.

[0017] Furthermore, the observation segment set extracts the comparison indicator sequence according to the indicator caliber and executes the judgment conditions to generate the comparison violation segment set, forming the comparison main evidence segment reference and generating isomorphic comparison suggestions. The scheme suggestions and comparison suggestions extract the difference key set by using the adjustment object identifier and the suggested action type as the alignment key. The difference key set generates difference descriptions and binds the planning action identifier and the replacement segment set index to the evaluation task.

[0018] Furthermore, the city status snapshot before the action is loaded as the initial state of the simulation according to the snapshot index. The list of adjustment items in the scheme suggestion is translated into the simulation action configuration and the adjustment object identifier and the suggested spatial location reference are bound. The simulation trajectory outputs the simulation indicator sequence according to the indicator caliber of the indicator judgment table and is aligned with the reference time window of the main evidence section. The simulation indicator sequence generates the expected change and is paired with the change amount of the corresponding time window in the purification channel dataset and saved.

[0019] Furthermore, the expected changes and the amount of purification changes are subjected to consistency verification and a consistency verification result is generated. The consistency verification result includes the conclusions of consistency in direction, consistency in segment, and consistency in location, and is written into the evaluation task. If there are any items that fail the verification, a supplementary sampling task is generated and the plan suggestion is frozen. The supplementary sampling task is written with the planning action identifier, scope of action, supplementary sampling time window, and supplementary sampling object reference. When all consistency verification results pass, the plan suggestion is released and a purification certificate is generated.

[0020] The technical effects and advantages of the digital twin-based urban planning method of this invention are as follows:

[0021] 1. The identification and scope of planning actions are fixed by the city status snapshot and snapshot index before the action. The replacement trace index synchronously records the source of the replacement segment set and the benchmark observation sequence, so that the purification channel dataset, indicator judgment table, difference description and purification certificate form a continuous evidence chain. The planning evaluation conclusion can be traced back to the reference and replacement boundary of the main evidence segment segment, reducing the drift of the scope and the mixing of snapshots when multiple planning actions are superimposed in the same area, and improving the repeatability and consistency of review and verification.

[0022] 2. Under the joint constraints of boundary attenuation fit and rhythm template deviation intensity, the observation segment set generates a replacement segment set. The baseline observation sequence is generated and replaced by combining the pre-action urban state snapshot with the template fragment sequence. This makes the purification channel dataset closer to the spatial propagation pattern and temporal rhythm structure of the planning action's impact. External disturbances are less likely to be mistakenly included in the planning effect judgment. The scheme recommendation is based on the adjustment item list, which is attached to the adjustment object identifier and the suggested spatial location reference. The planning action identifier runs through the judgment and recommendation expression, improving the pertinence and stability of the planning scheme evaluation.

[0023] 3. The digital twin simulation uses a snapshot of the city's state before the action as the initial state and applies proposed solutions to form a simulation trajectory. Consistency verification aligns the expected changes with the changes in the purification channel dataset under the indicator caliber and the reference time window of the main evidence segment. Consistency in direction, segment, and location jointly constrains the release decision. If an item fails to pass, a supplementary data collection task is triggered and the proposed solution is frozen. Purification certificates are generated and archived through items, so that the release of planning recommendations has an interpretable simulation basis and a verifiable judgment path, reducing the planning risks caused by implementation deviations. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a digital twin-based urban planning method according to the present invention. Detailed Implementation

[0025] 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 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.

[0026] Example 1: Figure 1 This invention presents a digital twin-based urban planning method, comprising:

[0027] S1: Receive planning action information, extract planning action identifiers and scope of action, save a snapshot of the city status before the action, and bind the planning action identifier to the snapshot of the city status before the action.

[0028] S2: Collect operational observations and time stamps within the scope of action, generate an observation segment set based on the planned action stamps, output a replacement segment set based on the observation segment set, and replace the operational observations corresponding to the replacement segment set with the baseline observation sequence generated by the city status snapshot before the action to obtain the purification channel dataset.

[0029] S3: Generate effect judgment results and form solution suggestions based on the purification channel dataset, generate comparison suggestions based on the observation segment set, output difference descriptions, bind the difference descriptions to the planning action identifiers and replacement segment sets, and write them into the evaluation task.

[0030] S4: The proposed solution and explanation of differences are entered into the digital twin simulation to generate expected changes. The expected changes and the effect judgment results are checked for consistency. If the check fails, a supplementary collection task is issued and the proposed solution is frozen. If the check passes, the proposed solution is released and a purification certificate is generated.

[0031] When urban planning methodologies enter the implementation and evaluation phase, operational observations accumulate over time. Without stable definitions of planning action identifiers and their scope of application, observation segments are prone to misclassification, replacement segment sets are difficult to locate accurately, and the cleanup channel dataset lacks a reliable benchmark. Planning action information originates from the operational side, and discrepancies often arise in field formats and spatial boundary representations. If pre-action city state snapshots are not strictly bound to planning action identifiers, the mixing of snapshots will directly impact the benchmark observation sequence generation process, leading to distorted cleanup judgments. However, once planning action identifiers and their scope of application are fixed, and pre-action city state snapshots are solidified, the observation segment set possesses clear boundaries, allowing replacement actions to be reviewed under the same benchmark.

[0032] Complete and detailed implementation of step S1:

[0033] S101 Planning Action Information Structured Analysis and Conflict Constraints.

[0034] The planning action information enters the parsing process, first extracting the planning action identifier and the original boundary text of the scope. The planning action identifier undergoes a complete consistency check, which includes the combination of the planning action identifier character sequence and the publishing source identifier. This combination is used to check if a record with the same planning action identifier already exists in the snapshot index. If a match is found, the scope hash value in the record is compared with the scope hash value obtained in this parsing. If they are completely consistent, repeated snapshot generation stops to avoid the same planning action repeatedly occupying index resources. If they are inconsistent, a conflict record is written and processing stops. The conflict record contains the planning action identifier, the publishing source identifier, two sets of scope hash values, and a boundary sequence summary, used for manual verification to ensure the planning action has not been mistakenly reused.

[0035] Upon completion, the planned action identifier and the original boundary text of the scope of action are obtained. Conflict paths are blocked, and the snapshot index will not show the same name in different domains.

[0036] S102 Unified expression of scope and generation of scope hash value.

[0037] The original boundary text of the scope of action enters the unified expression process, first undergoing coordinate normalization. Coordinate normalization uses a preset coordinate reference and a preset grid step size to convert the coordinates of each boundary vertex into grid index coordinates. The grid index coordinate conversion adopts a uniform down-rounding rule, with both positive and negative coordinates being rounded down in the same direction, ensuring that the same scope of action yields consistent grid index results in different computing environments.

[0038] After completing the grid index coordinate transformation, boundary sequence standardization is performed. Boundary sequence standardization includes closure verification and start point specification. Closure verification ensures that the first and last vertices are consistent. Start point specification selects the lexicographically smallest vertex as the start point of the boundary sequence and maintains the vertex order according to the boundary direction, ensuring that there is only one standard expression for the same boundary sequence.

[0039] The scope hash value generation employs a multinomial cumulative encoding approach. It iterates through each grid index vertex in the standard boundary sequence, multiplying the vertex's horizontal index by a preset coefficient and its vertical index by another preset coefficient, then summing the two to obtain the vertex's encoded value. This vertex encoded value is then multiplied by a recursive power of a preset base and accumulated, with each accumulation process modulo a preset modulus, to obtain the scope hash value. The scope hash value is used only for index location and collision detection and is not included in the planning metric determination.

[0040] Upon completion, a standard boundary sequence and a scope hash value are obtained, with a fixed spatial scope and stable input for index positioning.

[0041] S103 captures city state snapshots and solidifies object state tables before actions.

[0042] Before an action is performed, a city state snapshot is captured, using the scope of effect as the filtering domain. Spatial measures are selected based on the geometric type of the city object to ensure consistent filtering for planar, linear, and point objects. Planar objects use area coverage, calculated by dividing the intersection area of ​​the object's geometry and the scope by the object's own area. Linear objects use length coverage, calculated by dividing the intersection length of the object's geometry and the scope by the object's own length. Point objects use a fall-in criterion, determined by the point coordinates falling within the scope. When the coverage reaches a preset threshold or the fall-in criterion is true, the city object is written to the object state table.

[0043] The object status table is fixed in row records, with each row containing an object identifier, spatial location, business attribute, and observable indicator reference. The spatial location uses a coordinate reference consistent with the scope of application. Business attribute field names are derived from the caliber field names in the planning management rule base. Business attributes include the observation benchmark level, which is the indicator value converted from the object observation status corresponding to the snapshot generation time identifier using the caliber mapping table, avoiding attribute drift caused by synonymous fields. The observable indicator reference stores the indicator identifier and collection caliber identifier of the running observation values, and includes the rhythm window definition corresponding to the collection caliber identifier in the planning management rule base. This ensures that the historical rhythm skeleton with the same caliber can be constructed from the running observation values ​​according to time windows, guaranteeing that the correct data caliber can be found during the benchmark observation sequence generation process.

[0044] Once the object status table is completed, a snapshot of the city status before the action is generated is created. This snapshot is written to the snapshot storage area, and the snapshot generation time identifier and snapshot version number are recorded.

[0045] S104 Snapshot Index Write and Collision Check.

[0046] Snapshot index writing uses a composite index key, which is generated by combining the planned action identifier encoding value and the scope hash value. The planned action identifier encoding value is generated by encoding the planned action identifier character sequence in a fixed order. The encoding process converts each character into a fixed code point value, recursively accumulates it according to a preset base, and takes the remainder when divided by a preset modulus to obtain the planned action identifier encoding value. The composite index key is generated by combining the planned action identifier encoding value and the scope hash value according to a preset combination rule, and takes the remainder when divided by a preset modulus to obtain the index key value.

[0047] To avoid hash collisions causing incorrect index pointers, boundary sequence verification is performed before writing. If the index key value already exists in the snapshot index, the standard boundary sequence summary of the existing record is read and compared with the current standard boundary sequence summary for vertex-by-vertex consistency verification. If they are completely consistent, the snapshot version number is updated and a reference to the city state snapshot before the new action is written. If they are inconsistent, a collision record is written and writing stops. The collision record contains the index key value, two sets of planning action identifiers, two sets of scope hash values, and two sets of standard boundary sequence summaries.

[0048] Once the write is complete, the snapshot index record contains the planning action identifier, scope hash value, standard boundary sequence summary, snapshot generation time identifier, snapshot version number, and reference to the city status snapshot before the action.

[0049] Through the above verification, the index key value positioning remains reliable, and the city status snapshot before the action will not be mistakenly referenced due to collision.

[0050] The pre-action city status snapshot and planning action identifier are bound together, and the scope of effect is fixed by a standard boundary sequence. The scope hash value and snapshot index together provide a stable location entry point. The object status table covers area objects, line objects, and point objects, and the caliber references required for generating the benchmark observation sequence are also fixed. After collision verification and conflict constraints are completed, the attribution of the observation segment set within the scope of effect is clear. The replacement segment set can point to the same pre-action city status snapshot, and the review and verification can locate evidence item by item along the snapshot index and boundary sequence summary.

[0051] The planning action identifiers and scope of action have been fixed, and the pre-action city state snapshots are now searchable. When running observations enter the purification process, the time order must first be restored to a single sequence, and the observation changes must be converged to the observation segment set to avoid interference from out-of-order and duplicate reporting in segment boundary determination. However, the core of the purification process is not to delete observation information, but to identify and replace the segment set and use the pre-action city state snapshot to generate a baseline observation sequence to complete the replacement. This ensures that the purified channel dataset retains the true timeline and true spatial coverage references, while maintaining the traceability of evidence.

[0052] Complete and detailed implementation of step S2:

[0053] S201 Operation: Observation value processing and observation segment set segmentation.

[0054] Time stamps and operational observations are arranged in ascending order of time stamp. Time alignment uses the sampling period configured for the data collection caliber identifier in the planning and management rule base. Time stamps are merged into the alignment scale of the sampling period. When multiple records appear for the same time stamp, field completeness is calculated first. Field completeness is scored based on the filling of key fields, which include operational observations, observation location coordinates, indicator identifiers, and data collection caliber identifiers. The record with the highest field completeness is retained. Among records with the same field completeness, the record with the latest reception time is retained, and the remaining records are deleted to form a continuous observation stream. The continuous observation stream performs finite state sequence mapping. The boundary set of the finite state sequence comes from the indicator caliber definition in the planning and management rule base. The boundary set includes the boundary point order and boundary closure rules. The boundary closure rules stipulate that boundary points fall within the upper interval, ensuring that boundary points do not repeatedly jump between adjacent states. After the finite state sequence is completed, the observation segment set is obtained by dividing the state-maintained segment into stable segments and the state-transition point into change segment boundaries. Each observation segment records the start and end time identifiers, the finite state sequence within the segment, and the observation summary within the segment. The observation summary is expressed by the duration of the state within the segment and the number of state transitions. The spatial coverage reference of the segment is also recorded. The spatial coverage reference directly points to the set of observation locations within the effective range, avoiding spatial drift.

[0055] S202 Boundary Attenuation Matching Calculation.

[0056] The effective range boundary generates multiple spatial zones inwards. The width of each spatial zone is taken as the spatial resolution of the observation location set. The spatial resolution is calculated by taking the median distance from each observation location in the observation location set to its nearest neighbor. When the observation location set is insufficient to form a distance set, the spatial resolution is taken as the preset grid step size used for the unified expression of the effective range. The spatial zone numbering starts from the boundary zone and increases inwards. Observations for each observation segment are aggregated and run according to the spatial zones to obtain a finite state sequence and a sequence of observations for each zone. To avoid misjudgments caused by single touch points, a touch point record is formed for each state transition in the zone sequence. The touch point record includes the touch point time identifier, touch point direction, and touch point amplitude. The touch point direction is determined by the direction of change of the state number, and the touch point amplitude is determined by the absolute value of the difference between the observation value before and after the touch point.

[0057] Boundary attenuation consistency is assessed using adjacent pairs of spatial zones as the unit of evaluation. The contact sequences of the inner and outer spatial zones within the same observation segment are compared sequentially. The order of contact occurrence is determined by comparing contact time markers; if the difference in contact time markers is less than the sampling period, they are considered to have occurred simultaneously. In cases of simultaneous occurrence, the order is determined by the magnitude of the contact amplitude. A contact occurring first in the inner spatial zone with an amplitude not less than that of the outer spatial zone is counted as one instance of consistency; a contact occurring first in the outer spatial zone or with a stronger amplitude is counted as one instance of non-consistency. The number of consistency and non-consistency occurrences for all adjacent spatial zone pairs are normalized. The normalization result is the proportion of the number of consistency occurrences to the sum of the number of consistency and non-consistency occurrences, yielding the boundary attenuation consistency. Boundary attenuation consistency is expressed dimensionlessly, allowing direct comparison and ranking between observation segments. Segments whose spatial propagation patterns are closer to the planned area of ​​effect, appearing first and then attenuating towards the boundary, will receive higher consistency values.

[0058] S203 rhythm template deviation strength calculation.

[0059] Before the action, the city status snapshot reads the snapshot generation time stamp and uses the indicator identifier and collection caliber identifier in the observable indicator reference of the object status table as the search key. It then retrieves continuous historical observation segments defined by the rhythm window of the planning management rule base before the snapshot generation time stamp from the running observation values. After deleting duplicate reported segments, a historical observation sequence is formed. A historical rhythm skeleton with the same caliber is extracted from the historical observation sequence, and the rhythm skeleton is expressed as a template segment sequence. The template segment sequence consists of continuous segments, and the segment boundaries are determined by the peak-valley sequence switching points. The peak-valley sequence switching points are determined by the change in the sign of adjacent differences in the historical observation sequence. A peak point is determined when the sign changes from rising to falling, and a valley point is determined when the sign changes from falling to rising. When the difference is zero, the nearest non-zero difference sign is used. The segment is encoded using trend direction and persistence pattern. The trend direction is determined by the magnitude relationship between the observation value at the start and end of the segment, and the persistence pattern is jointly encoded by the segment duration and the number of state transitions.

[0060] The current rhythmic segment sequence is extracted from the observation segment, and the segment boundary extraction method is consistent with that of the template segment sequence to ensure consistent alignment. Alignment uses the segment boundary of the template segment sequence as the time frame, projecting the current rhythmic segment sequence onto the same segment boundary. Misaligned segments are recorded when the peak-valley sequence is inconsistent with any of the trend directions; missing segments are recorded when the template segment exists but the current segment does not; and newly added segments are recorded when the current segment exists but the template segment does not. The rhythmic template deviation intensity is obtained by normalizing the number of misaligned segments and the number of missing segments. The normalization method uses the total number of segments as the denominator, with misaligned and missing segments counted separately to ensure that the dimensionless result falls within a fixed range. The higher the rhythmic template deviation intensity, the more the rhythmic structure deviates from the normal rhythmic skeleton reflected in the pre-action city state snapshot.

[0061] S204 bidirectional comprehensive analysis generates a set of replacement segments.

[0062] Each observation segment within the observation segment set simultaneously possesses both boundary attenuation fit and rhythm template deviation strength. These two indicators represent two chains of evidence for spatial propagation patterns and temporal rhythm structures. The comprehensive analysis employs an in-task median boundary method. First, the rhythm template deviation strengths of all observation segments are sorted from smallest to largest, and the value at the middle of the sort is taken as the rhythm median boundary. When the number of observation segments is even, the arithmetic mean of the two middle values ​​is taken as the rhythm median boundary. Then, the boundary attenuation fit of all observation segments is sorted from smallest to largest, and the value at the middle of the sort is taken as the boundary median boundary. Again, in the case of an even number of observation segments, the arithmetic mean of the two middle values ​​is taken.

[0063] The selection rule for the replacement segment set adopts a simultaneous admission requirement. An observation segment is included in the replacement segment set if its rhythm template deviation intensity is higher than the rhythm median limit, while its boundary attenuation fit is lower than the boundary median limit. To ensure the replacement segment set covers continuous disturbances without excessive fragmentation, a segment merging rule is applied within the replacement segment set. When the interval between adjacent observation segments falls into a stable segment and the interval duration is shorter than the typical length of the stable segment, they are merged into a single replacement segment. The start and end times of the replacement segment are identified by the earliest start and latest end times of the merged segment. After the replacement segment set is completed, the replacement range possesses both temporal continuity and evidentiary consistency.

[0064] S205 baseline observation sequence generation and purification channel dataset formation.

[0065] The object status table in the pre-action city status snapshot contains observable indicator references, which provide the indicator caliber identifiers for the operational observations. The benchmark observation sequence generation first establishes a caliber mapping table. This table aligns the business attributes in the object status table with the indicator caliber identifiers for the operational observations. The mapping rules are taken from the indicator caliber mapping relationships in the planning management rule base, avoiding name conflicts between object status table fields and operational observation fields. After the caliber mapping table is completed, the template fragment sequence is projected onto the time-stamped range of the replacement segment set. The trend direction and persistence pattern in the template fragment sequence are used to generate the time change skeleton within the replacement segment. The observation benchmark level in the object status table is converted from the object observation status corresponding to the snapshot generation time stamp through the caliber mapping table. The observation benchmark level is used to determine the starting and ending levels of the skeleton, resulting in a benchmark observation sequence with the same caliber and time axis as the operational observations.

[0066] Within the time-stamped range covered by the replacement segment set, the observed values ​​are replaced with the observation values ​​corresponding to the baseline observation sequence. After replacement, the cleaned-up channel dataset is obtained. The replacement trace index is written synchronously during replacement, recording the planning action identifier, observation segment identifier, start and end time identifiers of the replacement segment, reference to the baseline observation sequence, and reference to the caliber mapping table. The replacement trace index ensures that the cleanup action has a verifiable path, and the cleaned-up channel dataset maintains an interpretable source.

[0067] The observed data are time-aligned and segmented to form an observation segment set. Boundary attenuation fit and rhythm template deviation are calculated on the observation segment set. The replacement segment set is generated using a median boundary comprehensive discrimination. The baseline observation sequence is generated and replaced using both the object state table and the template fragment sequence. The cleaned channel dataset and the replacement trace index are simultaneously stored. The cleaned channel dataset and the observation segment set coexist, and the discrepancies indicate that both types of evidence fragments can be used simultaneously, providing the evaluation task with clear replacement boundaries and replacement sources.

[0068] The planning action identifier runs through both the snapshot index and the replacement trace index. The observation segment set retains the original evidence fragments, while the cleanup channel dataset retains the stable caliber after replacement. When the planning evaluation enters the conclusion output stage, a one-to-one correspondence must be established between the conclusion and the evidence fragments; otherwise, the review and verification can only see the conclusion but cannot locate the source. However, there are structural differences between the cleanup channel dataset and the observation segment set. The conclusions need to generate scheme recommendations and control recommendations with isomorphic structures, and then present the source of the differences with a difference description so that the evaluation task can form a complete chain of evidence.

[0069] Complete and detailed implementation of step S3:

[0070] The S301 purification channel dataset generates an index judgment table and effect judgment results.

[0071] The planning management rule base includes rule identifiers, indicator definitions, trigger conditions, judgment conditions, indicator definition mapping relationships, suggested action type mapping tables, suggested spatial location reference rules, sampling periods, no-change judgment definitions, coverage thresholds, offset thresholds, rhythm window definitions, and constraint levels. The indicator definitions include boundary sets, which in turn include boundary point order and boundary closure rules. The purification channel dataset extracts continuous indicator sequences according to indicator definitions. These sequences are aligned with the segment boundaries of the observation segment set to avoid treating internal segment breakpoints as judgment boundaries. Each rule identifier applies judgment conditions to the indicator sequence, returning a set of non-compliant segments. The set of non-compliant segments consists of continuous time segments, each representing a situation where the judgment condition is not met. When the set of non-compliant segments is empty, the judgment conclusion is recorded as compliant. When the set of non-compliant segments is not empty, the judgment conclusion is recorded as non-compliant, and the longest-lasting time segment is selected from the set as the main evidence segment. The main evidence segment is expressed using the segment start and end time identifiers and the observation segment identifier. The indicator judgment table is fixed row by row. The row content includes indicator definition, judgment rule reference, judgment conclusion, main evidence section reference, and replacement trace index reference. The effect judgment result is formed by summarizing the indicator judgment table. The summarization rule adopts the strictest principle. If any indicator definition has a non-compliant judgment conclusion, the effect judgment result is recorded as non-compliant, and the corresponding main evidence section reference set is retained.

[0072] The S302 effect assessment results were translated into solution recommendations and a list of adjustment items was finalized.

[0073] The indicator judgment table provides clear triggering criteria using rule identifiers as indexes. The adjustment item list needs to translate non-compliance conclusions into executable expressions at the planning action level, avoiding the solution recommendations remaining at the level of abstract evaluation. The planning management rule base maintains a suggested action type mapping table and suggested spatial location reference rules for each rule identifier. For each rule identifier, when a non-compliance judgment conclusion appears in the indicator judgment table, an adjustment item is generated. The adjustment item is written to the adjustment object identifier, which is jointly determined from the object status table and the suggested spatial location reference rules. It is then written to the suggested action type, which is determined by the suggested action type mapping table. Finally, it is written to the suggested spatial location reference, which is obtained from the hierarchical structure of the sub-partitions within the scope of effect. Finally, it is written to the suggested reason evidence reference, which is bound to the main evidence section reference and the replacement trace index reference. The adjustment item list is sorted by constraint level and the start time identifier of the main evidence section. The constraint level is given by the planning management rule base. After sorting, the solution recommendations are formed. The solution recommendations maintain structural stability to facilitate alignment with the comparison recommendations.

[0074] The constraint level is defined as a sorting label bound to the rule identifier in the planning management rule base. It is used to reflect the intensity of the impact on planning decisions and the priority of review and handling when the rule is violated. The constraint level directly participates in the sorting of the adjustment item list, ensuring that the proposed solutions under the same planning action identifier are output first for the items that must be processed, and then for the optimization items.

[0075] The constraint level construction logic is completed through a deterministic mapping from rule attributes to level labels. The planning management rule base maintains the rule source attribute, constraint type attribute, affected object attribute, affected scope attribute, and reversibility attribute for each rule identifier. The rule source attribute distinguishes between statutory clauses, competent authority regulations, and internal management standards. The constraint type attribute distinguishes between prohibitive constraints, mandatory constraints, threshold constraints, and suggestive constraints. The affected object attribute identifies the set of objects involved in public safety, ecological red lines, infrastructure protection, public services, and operational efficiency. The affected scope attribute is derived from the scope level and suggested spatial location. The reversibility attribute identifies the difficulty of restoring the object's state after adjustment. During mapping, the baseline level is first determined by the rule source attribute, then the baseline level is adjusted upwards or downwards by the constraint type attribute, and finally, the final level is determined by the affected object attribute and reversibility attribute. The final constraint level is written into the rule identifier metadata and fixed in the rule base version. When the evaluation task references the rule identifier, the constraint level is referenced simultaneously to ensure stable sorting criteria and reviewability.

[0076] S303 observation segment set generates control suggestions and maintains isomorphic structure.

[0077] The observation segment set retains evidence of operational observations before replacement. Without generating isomorphic comparison suggestions, the explanation of differences can only remain as textual interpretations, unable to pinpoint the root cause of the differences. The observation segment set extracts a comparison indicator sequence according to the indicator caliber, and this sequence is also aligned with the segment boundaries of the observation segment set. The planning management rule base applies the same judgment conditions as the indicator judgment table to the comparison indicator sequence, outputting a set of comparison violation segments, forming the comparison judgment conclusion and the main evidence segment reference. The comparison suggestion generation process is consistent with the scheme suggestion, still driven by the suggestion action type mapping table and suggestion spatial location reference rules to generate an adjustment item list, only replacing the suggestion reason evidence reference with the main evidence segment reference, and pointing the evidence fragment to the observation segment identifier and segment start and end time identifier. The comparison suggestion and the scheme suggestion maintain an isomorphic field set, ensuring that the difference comparison can be aligned item by item using a fixed alignment key.

[0078] A comparison of the S304 proposal with the control proposal was used to generate a description of the differences, which was then incorporated into the evaluation task.

[0079] The discrepancy description needs to point to evidence fragments from both the cleanup channel dataset and the observation segment set to clearly express the impact of the replaced segment set on the conclusion. The alignment key is fixed as a combination of the adjustment object identifier and the recommended action type. The alignment key is used to establish a one-to-one correspondence between the proposed adjustment items and the control proposed adjustment items. After alignment, discrepancy key extraction is performed. The discrepancy key set consists of alignment keys that exist only in the proposed or controlled suggestions. A discrepancy entry is generated for each discrepancy key, recording the discrepancy type, discrepancy entry location, and discrepancy evidence reference. The discrepancy evidence reference is simultaneously bound to the main evidence segment reference and the control main evidence segment reference, and also bound to the planning action identifier and the replacement segment set index, ensuring that the discrepancy entry can trace back the replacement scope. The discrepancy description is written into the evaluation task. The evaluation task records the version and generation time of the discrepancy description. The evaluation task saves references to the indicator judgment table, the adjustment item list, and the replacement trace index, allowing review and verification to locate the basis of the conclusion and the source of the discrepancy along the evaluation task.

[0080] The purification channel dataset uses a set of violation segments to fix the judgment conclusions and main evidence segments. The proposed solutions translate the non-compliance judgment conclusions into adjustment object identifiers and suggested action types through an adjustment item list. The comparison suggestions generate alignable results on the observation segment set with an isomorphic structure. The difference description extracts the difference key set with a fixed alignment key and binds it to the evidence fragments on both sides. The evaluation task obtains a verifiable link expression, and the planning action identifier runs through the entire process of conclusions, suggestions, and difference interpretations.

[0081] The planning action identification has been integrated into the pre-action city status snapshot and replacement trace index; the proposed solutions have been fixed by the adjustment item list; the indicator judgment table has solidified the references to the main evidence sections and judgment rules; and the difference description has included evidence fragments from the purification channel dataset and the observation section set in the evaluation task. However, if the planning simulation does not compare the expected changes under the same indicator caliber with the purification changes, the proposed solutions may still deviate from the scenario. Freezing the proposed solutions and supplementary data collection tasks must be based on the consistency verification results to ensure that the review and verification have a repeatable derivation chain.

[0082] Complete and detailed implementation of step S4:

[0083] S401 simulation scenario construction and input loading.

[0084] Digital twin simulations need to reproduce the implementation semantics corresponding to the planning action identifiers. The simulation inputs must come from the city state snapshot and scheme suggestions before the action to avoid the initial simulation state from being disconnected from observation evidence.

[0085] The snapshot index locates the pre-action city state snapshot based on the planned action identifier and scope of action, and loads the pre-action city state snapshot as the initial state for simulation. The list of adjustment items in the proposed scheme is translated item by item into simulation action configurations. The simulation action configurations use the adjustment object identifier to lock the object instance, the suggested spatial location reference to lock the internal sub-regions within the scope of action, and the suggested action type to lock the action semantics. The simulation time window is taken from the time identifier range covered by the main evidence segment reference, and the start and end time identifiers of the main evidence segment are used as the alignment endpoints of the simulation output. The simulation trajectory outputs a simulation indicator sequence according to the indicator criteria in the indicator judgment table. The sampling time of the simulation indicator sequence is consistent with the simulation time window, ensuring that the simulation output can be reproduced and verified within the evaluation task.

[0086] The expected changes in S402 are aligned with the changes in purification.

[0087] Consistency verification relies on comparing changes under the same indicator caliber. The inference side and the observation side must use the same time window endpoints to ensure that the direction is consistent and the magnitude trend is interpretable.

[0088] For each indicator, the indicator value at the start and end of the extrapolated time window is read from the extrapolated indicator sequence. The extrapolated change is obtained by subtracting the starting indicator value from the ending indicator value, while maintaining the same indicator dimension. For each indicator, the purification indicator value at the start and end of the same time window is read from the purification channel dataset. The purification change is obtained by subtracting the starting purification indicator value from the ending purification indicator value, while maintaining the same indicator dimension. Extrapolated changes and purification changes are stored in pairs using the indicator indicator as an index. The evaluation task saves the time stamp of the aligned endpoints. The extrapolated trajectory and the purification channel dataset can be verified along the same endpoints, avoiding distortion of changes caused by endpoint offset.

[0089] S403 consistency check calculation and consistency check result fixation.

[0090] Consistency verification needs to cover consistency in direction, segment, and location simultaneously. A single dimension cannot eliminate the deviation between the inferred trajectory and the observed evidence in spatiotemporal location.

[0091] The consistency of direction is determined by the sign relationship between the projected change and the purification change. A record is passed when both projected and purification changes are positive, passed when both are negative, and failed when their signs are opposite. A record is passed when both projected and purification changes are zero, failed when both are zero and non-zero, and failed when both are zero. The determination of zero uses the planning management rule base's criteria for determining no change, ensuring consistent judgment boundaries across different indicators.

[0092] The segment consistency determination is completed by assessing the coverage of the main evidence segment based on the coverage range of the replacement segment set. When the start and end time markers of the main evidence segment overlap, the endpoint of the main evidence segment is expanded using the sampling period of the observation segment set, ensuring the length of the main evidence segment is positive. The intersection of the time marker range covered by the replacement segment set and the main evidence segment is taken. The intersection length is measured according to the sampling period, and the length of the main evidence segment is also measured according to the sampling period. The segment coverage ratio is obtained by dividing the intersection length by the length of the main evidence segment. Records are approved when the segment coverage ratio reaches the coverage threshold configured for the indicator caliber in the planning management rule base; records are not approved when the coverage threshold is not reached.

[0093] The consistency determination is based on the offset between the deduced evidence time and the observed evidence time. The deduced evidence time is the earliest time identifier of the judgment state switch in the deduced trajectory. The judgment state switch is determined by the judgment conditions referenced in the indicator judgment table. The judgment conditions are executed hourly on the deduced indicator sequence. The moment of the first switch from compliance to non-compliance or from non-compliance to compliance is recorded as the deduced evidence time. The observed evidence time is the start time identifier of the main evidence segment. The evidence offset is the absolute value of the time difference between the deduced evidence time and the observed evidence time. The record passes when the evidence offset does not exceed the offset threshold configured for the indicator caliber in the planning management rule base; the record fails when it exceeds the offset threshold.

[0094] The consistency verification results are fixed with the index caliber as the row key. The row content includes the direction consistency conclusion, the segment consistency conclusion, and the location consistency conclusion. The main evidence segment reference, the replacement segment set index, and the inference evidence time of the failed items are also written into the evaluation task. The review and verification can directly locate the source of the conflict.

[0095] S404 Freeze Release Decision and Cleanup Certificate Generation.

[0096] Before implementation, the proposed solutions need to undergo consistency verification and screening. Items that fail the verification mean that the inferred trajectory cannot reproduce the observation conclusions on key evidence fragments. The supplementary sampling task and the proposed freezing plan need to fix the scope of disputes to a verifiable time window and set of objects.

[0097] If any indicator fails the consistency check, the supplementary sampling task includes a planned action identifier, scope of action, supplementary sampling time window, and supplementary sampling object reference. The supplementary sampling time window is obtained by merging the time identifier ranges of the main evidence segment and the control main evidence segment, and absorbing adjacent stable segments within the observation segment set to form a continuous window. Stable segments are determined by the state maintenance segments of a finite state sequence. The supplementary sampling object reference is extracted from the object state table, taking the set of object identifiers consistent with the spatial coverage reference of the main evidence segment, and retaining the reference caliber of observable indicators to avoid mismatch between the supplementary sampling object and the judgment indicator caliber. The freeze scheme recommendation includes the freeze status and the index of failed entries in the evaluation task. The scheme recommendation text remains unchanged, and the review and verification can locate the conflict link by comparing the freeze reasons and difference explanations.

[0098] When all consistency verification results pass, a proposed plan is released, and a purification certificate is generated. The purification certificate records the planning action identifier, scope of application, pre-action city status snapshot index, replacement segment set index, baseline observation sequence index, indicator judgment table reference, difference description reference, consistency verification result reference, and simulation task key reference. After the purification certificate is archived in the evaluation task, the planning action identifier can be linked to the simulation trajectory and evidence fragments along the purification certificate. Review and verification can reproduce the simulation input, change calculation, and consistency judgment process under the same caliber.

[0099] Digital twin simulation constructs a simulation scenario using a snapshot of the city's state before the action and proposed solutions. The simulated changes and the purified changes are aligned under the same indicator caliber and the same time window endpoint. Consistency verification solidifies the consistency of direction, segment, and location into searchable entries. Supplementary data collection tasks and frozen solution recommendations fix the scope of disputes to the main evidence segment and the comparison main evidence segment. The purification certificate links the snapshot index, the replacement segment set index, the indicator judgment table, the difference explanation, and the consistency verification results into a complete review path.

[0100] Specifically, the above are merely preferred embodiments of this application and are not intended to limit this application.

[0101] In the description of this specification, references to terms such as "an embodiment," "example," and "specific example" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0102] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A digital twin-based urban planning method, characterized in that, Including the following steps: S1: Receive planning action information, extract the planning action identifier and scope of action, save a snapshot of the city's state before the action, and bind the planning action identifier to the snapshot of the city's state before the action, including: The planning action information is parsed to obtain the planning action identifier and scope of action. The scope of action is normalized by coordinates and standardized by boundary sequence to generate a scope of action hash value. The scope of action filters urban objects and completes the coverage determination according to object type to solidify the object status table and form a snapshot of the urban status before the action. The snapshot of the urban status before the action is written into the snapshot index and boundary sequence verification is performed. The snapshot index records the planning action identifier, scope of action hash value, boundary sequence summary, snapshot version number and snapshot reference. S2: Collect operational observations and time stamps within the scope of action, generate an observation segment set based on the planned action stamps, and output a replacement segment set based on the observation segment set. The operational observations corresponding to the replacement segment set are replaced by the baseline observation sequence generated from the city status snapshot before the action, resulting in the purification channel dataset, including: The running observations and time stamps are sorted by time stamp and duplicate reported segments are removed. The running observations are mapped to a finite state sequence according to the indicator definition of the planning and management rule base. The finite state sequence is divided into observation segment sets by state switching points. The observation segment set records the start and end time stamps of the segment, the finite state sequence, the observation summary and the segment spatial coverage reference. The action range boundary generates multiple spatial zones inward. The observation segment set is aggregated according to the spatial zones to form a finite state sequence and a sequence of observation values. The contact point records are extracted from the zonal sequence and the contact point sequence and amplitude of adjacent spatial zones are compared to obtain the boundary attenuation matching degree. The template segment sequence is extracted from the city state snapshot before the action and aligned with the current rhythm segment sequence extracted from the observation segment set. The rhythm template deviation intensity is obtained by normalization based on the misaligned segment and missing segment. The observation segment set generates rhythm median boundary and boundary median boundary according to the rhythm template deviation intensity and boundary attenuation matching degree. The observation segment that satisfies the rhythm template deviation intensity being higher than the rhythm median boundary and the boundary attenuation matching degree being lower than the boundary median boundary forms the replacement segment set. The city status snapshot before the action establishes a caliber mapping table and generates a benchmark observation sequence. The running observation values ​​within the coverage area of ​​the replacement segment set are replaced by the benchmark observation sequence to form a purification channel dataset and written into the replacement trace index. S3: Generate effect judgment results and form solution suggestions based on the purification channel dataset, generate comparison suggestions based on the observation segment set, output difference descriptions, bind the difference descriptions to the planning action identifiers and replacement segment sets and write them into the evaluation task; S4: The proposed solution and explanation of differences are entered into the digital twin simulation to generate expected changes. The expected changes and the effect judgment results are checked for consistency. If the check fails, a supplementary collection task is issued and the proposed solution is frozen. If the check passes, the proposed solution is released and a purification certificate is generated.

2. The urban planning method based on digital twins according to claim 1, characterized in that, Step S3 includes: The purification channel dataset extracts continuous indicator sequences according to indicator caliber and aligns them with the segment boundaries of the observation segment set. The planning and management rule base executes judgment conditions on the continuous indicator sequences to generate a set of violation segments. If the set of violation segments is empty, the compliance judgment conclusion is recorded. If the set of violation segments is not empty, the non-compliance judgment conclusion is recorded, and the segment with the longest duration is selected as the main evidence segment. The indicator judgment table solidifies the indicator caliber, judgment rule reference, judgment conclusion, and main evidence segment reference to form the effect judgment result.

3. The urban planning method based on digital twins according to claim 2, characterized in that, Step S3 also includes: The planning management rule base provides a mapping table of suggested action types and rules for suggesting spatial locations. When a non-compliance judgment is found in the indicator judgment table, an adjustment item list is generated. The adjustment item list includes the adjustment object identifier, suggested action type, suggested spatial location reference, and suggested reason evidence reference. The suggested reason evidence reference is bound to the main evidence section reference and the replacement trace index reference. The adjustment item list is sorted by constraint level and the start time identifier of the main evidence section to form a scheme suggestion.

4. The urban planning method based on digital twins according to claim 3, characterized in that, Step S3 also includes: The observation segment set extracts the reference index sequence according to the index caliber and generates a reference violation segment set by applying the judgment conditions. It forms the reference of the main evidence segment and generates the isomorphic reference suggestion. The scheme suggestion and the reference suggestion extract the difference key set by using the adjustment object identifier and the suggested action type as the alignment key. The difference key set generates the difference description and binds the planning action identifier and the replacement segment set index to write it into the evaluation task.

5. A digital twin-based urban planning method according to claim 4, characterized in that, Step S4 includes: The city status snapshot before the action is loaded as the initial state of the simulation according to the snapshot index. The list of adjustment items in the scheme suggestion is translated into the simulation action configuration and the adjustment object identifier and the suggested spatial location reference are bound. The simulation trajectory outputs the simulation indicator sequence according to the indicator caliber of the indicator judgment table and is aligned with the reference time window of the main evidence section. The simulation indicator sequence generates the expected change and is paired with the change amount of the corresponding time window in the purification channel dataset and saved.

6. The urban planning method based on digital twins according to claim 5, characterized in that, Step S4 also includes: The expected changes and the purification changes are subjected to consistency verification and a consistency verification result is generated. The consistency verification result includes the direction consistency conclusion, the segment consistency conclusion, and the location consistency conclusion, and is written into the evaluation task. If there are any items that fail the verification, a supplementary sampling task is generated and the plan suggestion is frozen. The supplementary sampling task is written with the planning action identifier, the scope of action, the supplementary sampling time window, and the supplementary sampling object reference. When all consistency verification results pass, the plan suggestion is released and a purification certificate is generated.

Images