An intelligent city building index supervision system and method based on an AI algorithm

Abstract

The application relates to the technical field of smart city supervision, and discloses a smart city building index supervision system and method based on an AI algorithm. The system comprises a building index collection unit, which acquires building energy consumption, structural safety and environmental protection indexes in real time; a supervision strategy generation unit compares the indexes with multi-dimensional threshold values, generates difference signals according to the results, and outputs corresponding supervision strategies; an index fluctuation analysis module continuously receives strategy execution records and index change processes, extracts strategy lag characteristic values, and calculates index recovery efficiency characteristic values; an index correlation module adjusts data transmission frequency and optimizes strategy execution intensity according to the characteristic values; a three-dimensional model mapping unit matches the building indexes with a city three-dimensional geographic information model, and generates a spatial distribution atlas; and a strategy effectiveness judgment unit counts abnormal signal frequencies in a preset period, and generates a supervision failure identifier when the abnormal signal frequencies exceed a stability threshold value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart city supervision, in particular to a smart city building index supervision system and method based on AI algorithm. BACKGROUND

[0002] With the continuous acceleration of urbanization, the number of urban buildings continues to grow, and the problems of energy consumption, structural safety and environmental protection during the operation of buildings are increasingly prominent. Traditional building index supervision methods rely on manual inspection and discrete data collection, which has the disadvantages of data acquisition lag, limited supervision range, and untimely strategy adjustment.

[0003] In terms of building energy consumption supervision, existing systems can only statistically analyze a single type of energy consumption data, lack comprehensive analysis of multiple energy consumption indicators such as water, electricity and gas, and are difficult to accurately locate energy consumption anomalies. Building structural safety supervision mostly uses periodic detection methods, which have long detection cycles and are greatly influenced by human factors, making it difficult to capture potential safety hazards such as structural deformation and crack expansion in real time, and leading to missed or misjudged safety accidents.

[0004] In the field of building environmental protection supervision, existing technologies do not comprehensively monitor building construction dust, noise pollution and building waste disposal indicators, and have low data transmission efficiency, making it difficult for environmental protection supervision departments to timely grasp pollution dynamics and take effective control measures. In addition, traditional supervision strategy formulation relies on experience and lacks dynamic correlation with real-time monitoring data, making the strategy less targeted and effective, and difficult to meet the needs of fine management of smart cities.

[0005] The combination of urban building indexes and geographic spatial information is not close enough, and the distribution characteristics of building indexes in urban space cannot be intuitively presented, which is not conducive to the supervision department to grasp the overall situation of urban buildings from a macro perspective, and also affects the optimal allocation of supervision resources. With the development of artificial intelligence technology, how to use AI algorithms to improve the intelligent level of building index supervision and realize the automation, precision and efficiency of the supervision process has become a problem to be solved in the current smart city construction. SUMMARY

[0006] The present application aims to provide a smart city building index supervision system based on AI algorithm to solve the problems raised in the background.

[0007] To achieve the above-mentioned purpose, the present application provides a smart city building index supervision system based on AI algorithm, which comprises:

[0008] a building index collection unit for real-time acquisition of building indexes, including building energy consumption indexes, building structural safety indexes and building environmental protection indexes;

[0009] The regulatory policy generation unit performs multi-dimensional threshold comparison on the building indicators, generates a difference signal according to the comparison result, and outputs a corresponding building regulatory policy based on the difference signal;

[0010] The indicator fluctuation analysis module continuously receives the building regulatory policy execution records and the building indicator change process sent by the regulatory policy generation unit, extracts a policy lag characteristic value by analyzing the building regulatory policy execution records, and calculates an indicator recovery efficiency characteristic value by analyzing the building indicator change process;

[0011] The indicator correlation module obtains the policy lag characteristic value and the indicator recovery efficiency characteristic value from the indicator fluctuation analysis module, adjusts the data transmission frequency of the building indicator acquisition unit according to the policy lag characteristic value, and optimizes the execution strength of the building regulatory policy according to the indicator recovery efficiency characteristic value;

[0012] The three-dimensional model mapping unit performs spatial matching on the building indicators and the city three-dimensional geographic information model to generate a building indicator spatial distribution atlas;

[0013] The policy effectiveness judgment unit obtains a policy execution abnormal signal through the indicator fluctuation analysis module, counts the frequency of the abnormal signal in a preset regulatory period, and generates a building regulatory failure identifier when the frequency of the abnormal signal exceeds a preset stability threshold.

[0014] Preferably, the regulatory policy generation unit obtains the building energy consumption indicators, building structure safety indicators, and building environmental protection indicators through the building indicator acquisition unit, compares the building energy consumption indicators with a preset energy consumption standard range, generates an energy consumption abnormality identifier if the building energy consumption indicators exceed the energy consumption standard range, compares the building structure safety indicators with a preset safety threshold, generates a structure risk identifier if the building structure safety indicators are lower than the safety threshold, compares the building environmental protection indicators with a preset environmental protection threshold, and generates an environmental protection defect identifier if the building environmental protection indicators do not meet the environmental protection threshold.

[0015] The regulatory policy generation unit triggers a building structure reinforcement strategy when generating a structure risk identifier, starts an energy consumption hierarchical control strategy when generating an energy consumption abnormality identifier, and executes an environmental protection facility optimization strategy when generating an environmental protection defect identifier.

[0016] Preferably, the process of the regulatory policy generation unit executing building energy consumption control includes:

[0017] The median value of the energy consumption standard range is selected as the energy consumption reference value, the absolute difference between the current building energy consumption indicators and the energy consumption reference value is calculated, the absolute difference is input into an energy consumption dynamic adjustment model to generate a power adjustment instruction, and the operation parameters of the building energy system are controlled;

[0018] The process of performing building structure safety reinforcement comprises: obtaining a real-time monitoring value of a structure safety index, calculating a deviation degree of the real-time monitoring value from a safety critical value, and generating a structure reinforcement scheme of different levels according to the deviation degree.

[0019] The process of performing building environmental protection index optimization comprises: establishing an environmental protection index dynamic balance curve, identifying a deviation amount of a current environmental protection index in the dynamic balance curve, and generating an operation parameter correction instruction of a pollution control device according to the deviation amount.

[0020] Preferably, the index fluctuation analysis module obtains the occurrence time points of the energy consumption anomaly identifier, the structure risk identifier and the environmental protection defect identifier through the supervision strategy generation unit, records the actual building index values corresponding to the occurrence time points, and performs double comparison of the actual building index values with the standard upper limit value and the standard lower limit value of the corresponding index; the index fluctuation analysis module calculates the minimum deviation amount of the actual building index value from the standard upper limit value and the minimum deviation amount of the actual building index value from the standard lower limit value, and selects the smaller value of the two minimum deviation amounts as a key deviation characteristic quantity; a strategy response delay coefficient is generated through weighted calculation of the key deviation characteristic quantity and the standard upper limit value and the standard lower limit value, and a strategy lag alarm signal is generated when the strategy response delay coefficient is greater than a preset delay tolerance threshold value;

[0021] The index fluctuation analysis module records the time span from the generation of the anomaly identifier to the recovery of the building index to the normal value, compares the time span with a preset standard recovery time, and generates an index recovery lag signal if the time span exceeds the standard recovery time.

[0022] Preferably, the index correlation module receives the strategy lag alarm signal and the strategy response delay coefficient sent by the index fluctuation analysis module, calculates a data transmission acceleration factor according to the proportional relationship between the strategy response delay coefficient and the delay tolerance threshold value, and sends the data transmission acceleration factor to the building index collection unit;

[0023] The building index collection unit shortens the length of the dynamic time interval based on the data transmission acceleration factor; the index correlation module simultaneously receives the index recovery lag signal sent by the index fluctuation analysis module, generates a strategy execution intensity improvement instruction and sends it to the supervision strategy generation unit, and the supervision strategy generation unit increases the action amplitude of the building supervision strategy according to the strategy execution intensity improvement instruction.

[0024] Preferably, the strategy effectiveness judging unit continuously acquires the strategy lag alarm signals and the index recovery delay signals generated by the index fluctuation analysis module, accumulates the number of signal occurrences in a single supervision period to form an abnormal signal total amount, calculates the ratio of the abnormal signal total amount to the total length of the supervision period to obtain a system failure rate, and sends a building supervision failure identifier to the three-dimensional model mapping unit when the system failure rate reaches a preset failure threshold.

[0025] Preferably, the process of spatial matching performed by the three-dimensional model mapping unit includes: extracting spatial coordinate data of the building in the three-dimensional geographic information model, converting the building energy consumption index into heat distribution gradient data, converting the building structure safety index into structure stress distribution data, and converting the building environmental protection index into pollutant diffusion trajectory data; and fusing the heat distribution gradient data, the structure stress distribution data, and the pollutant diffusion trajectory data to generate a multi-dimensional building index spatial distribution graph.

[0026] The three-dimensional model mapping unit transmits the building index spatial distribution graph to the index correlation module in real time, and the index correlation module adjusts the priority of the supervision strategy based on the spatial clustering features in the building index spatial distribution graph.

[0027] Preferably, the process of handling the structure risk identifier by the index fluctuation analysis module includes:

[0028] The process of handling the structure risk identifier by the index fluctuation analysis module includes:

[0029] Preferably, the process of handling the building index spatial distribution graph by the index correlation module includes:

[0030] The process of handling the building index spatial distribution graph by the index correlation module includes:

[0031] Preferably, the application also includes an AI algorithm-based smart city building index supervision method, applied to the AI algorithm-based smart city building index supervision system described above, the method comprising:

[0032] Real-time energy consumption parameters, structural stress parameters and environmental emission parameters of the building are collected by distributed sensing equipment; the collected building parameters are compared with preset energy consumption benchmark values, structural safety critical values and environmental protection standard values in multiple dimensions; energy consumption control strategies, structural reinforcement strategies and environmental protection management strategies are generated according to the comparison results; the change trajectory of the building parameters during strategy execution is monitored in real time, and the parameter recovery rate and strategy response delay characteristic quantity are calculated; the collection frequency of the building parameters is dynamically adjusted based on the strategy response delay characteristic quantity; the change trajectory of the building parameters is converted into a spatial heat distribution model through three-dimensional space mapping technology; the parameter abnormal aggregation area is identified in the spatial heat distribution model, and a spatial collaborative supervision scheme is generated; the frequency of strategy execution abnormal events in a preset period is counted, and the system self-correction mechanism is triggered when the frequency exceeds the stability threshold.

[0033] Compared with the prior art, the application has the following beneficial effects:

[0034] The AI algorithm-based smart city building index supervision system breaks through the limitations of scattered and lagging data collection in traditional supervision by real-time acquisition of multi-dimensional indexes such as building energy consumption, structural safety and environmental protection through the building index collection unit, enabling the supervision department to timely grasp the dynamic information of building operation. The supervision strategy generation unit compares the building indexes with multiple thresholds and generates corresponding supervision strategies, changing the previous way of relying on experience to develop strategies, making the development of strategies more scientific and targeted, and enabling different control measures to be taken according to different index abnormal situations.

[0035] The index fluctuation analysis module continuously analyzes the supervision strategy execution records and the building index change process, extracts the strategy lag characteristic value and calculates the index recovery efficiency characteristic value, providing a basis for the self-optimization of the system. The index correlation module adjusts the data transmission frequency and optimizes the strategy execution intensity according to these characteristic values, so that data collection can meet the supervision needs and avoid unnecessary resource waste, while making the execution of the supervision strategy more practical and avoiding the problem of excessive or insufficient strategy execution.

[0036] The three-dimensional model mapping unit performs spatial matching of the building indexes with the city three-dimensional geographic information model, generates a building index spatial distribution atlas, and intuitively presents the distribution of various indexes in the city space, which helps the supervision department to grasp the overall operation situation of the city buildings and provides an intuitive reference for the rational allocation of supervision resources.

[0037] The policy effectiveness judging unit can discover problems existing in the supervision process in time by counting the abnormal signal frequency and generating a supervision failure identifier, and promote the supervision department to adjust and improve the failed supervision policy, thereby forming a continuous improvement supervision closed loop. Through the deep application of the AI algorithm, the whole system realizes the change of the building index supervision from passive response to active early warning and from scattered management to comprehensive control, improves the intelligent level and management efficiency of the intelligent city building supervision, and adapts to the requirements of the modern city fine and dynamic management. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A working principle diagram of the intelligent city building index supervision system based on the AI algorithm is described.

[0039] Figure 2 A working principle diagram of the supervision policy generating unit is described.

[0040] Figure 3 A working principle diagram of the index fluctuation analysis module is described.

[0041] Figure 4 A working principle diagram of the policy effectiveness judging unit is described. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0043] Please refer to Figure 1 The present application provides an intelligent city building index supervision system and method based on an AI algorithm, which comprises:

[0044] The multiple functional units work cooperatively to realize real-time monitoring and analysis of building indicators through AI algorithms. The building indicator collection unit obtains building energy consumption indicators, building structure safety indicators, and building environmental protection indicators in real time through distributed sensors and transmits the data to the regulatory strategy generation unit. The regulatory strategy generation unit compares the received building indicators with multiple-dimensional threshold values, generates difference signals by comparing each indicator with the preset standard threshold value, and outputs the corresponding building regulatory strategy according to the difference signals. This process involves calculating the deviation degree of the indicator value from the preset range, and dynamically starting the strategy execution based on the difference signal as the trigger mechanism. The indicator fluctuation analysis module continuously receives the building regulatory strategy execution records and building indicator change process data sent by the regulatory strategy generation unit. The module extracts the strategy lag characteristic value by analyzing the building regulatory strategy execution records, specifically including the record strategy start time and the interval between the effective time, and calculates the strategy delay coefficient based on the time interval. At the same time, the indicator recovery efficiency characteristic value is calculated by analyzing the building indicator change process, for example, the ratio of the time required for the indicator to recover from an abnormal value to a normal value to the preset recovery time. The indicator correlation module connects the indicator fluctuation analysis module and receives the strategy lag characteristic value and the indicator recovery efficiency characteristic value. The module adjusts the data transmission frequency of the building indicator collection unit according to the strategy lag characteristic value, for example, calculates the shortening ratio of the data transmission interval based on the delay coefficient; at the same time, optimizes the execution strength of the building regulatory strategy according to the indicator recovery efficiency characteristic value, including increasing or decreasing the strategy action amplitude. The three-dimensional model mapping unit spatially matches the building indicators with the city three-dimensional geographic information model; the unit converts the indicator values into spatial distribution data and generates a building indicator spatial distribution atlas through geographic coordinate superposition. The strategy effectiveness judgment unit obtains the strategy execution abnormal signal through the indicator fluctuation analysis module, and counts the frequency of abnormal signals within the preset regulatory period; the unit calculates the ratio of the abnormal signal frequency to the system stability threshold value, and generates a building regulatory failure identifier when the ratio exceeds the threshold value, which is used to mark the system failure state.

[0045] Embodiment 1: refer to Figure 2 The regulatory strategy generation unit establishes a real-time communication link with the building indicator collection unit through a special data interface. The collection of building energy consumption indicators relies on a group of intelligent electricity meters installed in the building power distribution system. These meters collect current, voltage, and power factor data at a millisecond level, and generate comprehensive energy consumption values updated at a minute level after aggregation by an edge computing node. The regulatory strategy generation unit is provided with a dynamic energy consumption standard library, which automatically adjusts the upper and lower threshold values of the energy consumption standard range according to different building types and climate seasons. When the real-time energy consumption value exceeds the upper limit of the standard range for two consecutive sampling periods, the system starts a three-level discrimination mechanism: the first time the standard is exceeded triggers data review, the second time the standard is exceeded starts an abnormality checking algorithm, and the third time the standard is exceeded generates an energy consumption abnormality identifier. The identifier contains the exact over-standard percentage, the over-standard starting timestamp, and the corresponding power grid load phase characteristics.

[0046] The building structure safety index is derived from a distributed fiber-optic sensor array embedded in the building load-bearing structure, which monitors concrete strain, steel structure displacement, and foundation settlement in real time. The regulatory strategy generation unit uses a hierarchical threshold comparison mechanism to process this index: the primary comparison compares single-point sensor data with the material elastic limit threshold in real time, and if the single-point data is abnormal, it triggers a secondary regional linkage analysis, and the third comparison integrates all abnormal regional data and the overall safety threshold for comprehensive judgment. The safety threshold is dynamically generated by the building structure digital twin model. When the system detects that more than three key monitoring points are simultaneously below 90% of the threshold value, a structure risk identification containing a stress distribution hot zone is generated. This identification automatically associates with the coordinates of vulnerable components in the building BIM model and marks the vector difference between the actual safety value and the threshold value.

[0047] The building environmental protection index is collected by a chemical sensor group deployed at the building exhaust port, monitoring parameters including PM2.5, nitrogen oxides, and volatile organic compound concentrations. The regulatory strategy generation unit uses a composite comparison strategy: first, compare the concentration of a single pollutant with the national standard threshold, and then calculate the comprehensive environmental protection index through a multi-pollutant factor coupling algorithm. The system sets a dual judgment window: when the concentration of a single pollutant exceeds 110% of the standard value for three consecutive samplings, or the comprehensive environmental protection index is below the preset compliance threshold, an environmental protection defect identification containing a pollutant concentration deviation spectrum is generated. The identification records the main over-discharge pollutant species, concentration peak value, and percentage deviation from the standard value.

[0048] In the identification processing logic, the system sets a strategy trigger priority mechanism. When a structure risk identification is generated, an emergency response channel is immediately activated: send an encrypted instruction package to the building intelligent reinforcement system through a dedicated safety protocol, which contains the structure reinforcement type code (C type for local reinforcement, D type for overall strengthening), implementation coordinate grid, and material quantity parameters. The system synchronously locks other strategy execution sequences for this building, ensuring that the structure reinforcement strategy exclusively occupies system resource channels.

[0049] The energy consumption anomaly identification triggers a hierarchical response mechanism: the system first analyzes the over-standard feature spectrum, and when it identifies that the basic energy consumption is over-standard, it starts a first-level control strategy, which adjusts the lighting circuit brightness parameters and air conditioning set temperature; when it detects peak energy consumption caused by production equipment, it activates a second-level control strategy, which includes a device power limitation time table and a peak-shaving operation scheme; a third-level control is only started under critical load conditions of the power grid, implementing freezer unit shutdown and elevator group control strategies. All control strategies are directly written into the control system registers through the API interface of the building energy management system.

[0050] After the environmental defect identifier is generated, the system invokes the pollutant source tracing analysis model: it locates the main pollution source equipment using a diffusion trajectory back-calculation algorithm, and then generates equipment-level optimization strategies. For boiler emissions exceeding standards, the strategy involves adjusting the air-fuel ratio and increasing the catalyst quantity; for pollution diffusion in the ventilation system, the strategy includes adjusting the exhaust vent vector angle and increasing the electrostatic precipitator field strength. The strategy instructions are transmitted to the equipment controller via an Industrial Internet of Things (IIoT) protocol, and the controller adjusts the actuator status in real time based on the instruction parameters.

[0051] The entire strategy execution process incorporates a closed-loop verification mechanism. Once the regulatory strategy generation unit issues a strategy command, a timeout retransmission mechanism is activated: if no confirmation signal from the device is received within a preset time window, the system automatically upgrades the strategy strength level and retransmits the command. Simultaneously, a real-time strategy-indicator mapping table is established, with each strategy command bound to the expected change curve of the target indicator. Strategy parameters are dynamically adjusted by comparing the actual data with the expected curve. For structural reinforcement strategies, the system additionally includes a deformation monitoring and synchronous tracking program, increasing the sampling frequency of structural safety indicators to the second level during strategy execution.

[0052] The lifecycle management of anomaly markers adopts an event-driven architecture. Energy consumption anomaly markers are automatically cleared after the corresponding indicator returns to the standard range for six consecutive sampling periods. An energy consumption recovery analysis report is generated before clearing. Structural risk markers need to be cleared after the safety indicator has been stably exceeding the critical value by 120% for 24 consecutive hours. A secondary verification scan of structural integrity is triggered before clearing. The clearing conditions for environmental defect markers must be met simultaneously: the concentration of a single pollutant returns to the standard range and the comprehensive environmental index is higher than the threshold line for 12 consecutive hours. An emission control effectiveness assessment data package is automatically generated when clearing.

[0053] At the system resource scheduling level, a dynamic priority queue is set up. Structural risk indicators always occupy the highest priority channel, and their policy instructions can interrupt the execution of other policies; energy consumption anomaly indicators are assigned medium to high priority based on the severity of exceeding standards; environmental defect indicators are assigned resource weights based on daily total emissions. All policy instructions are appended with timestamps and digital signatures, enabling operation traceability through blockchain nodes. Policy execution logs contain complete process parameters, which are compressed and encrypted and stored in a distributed database, providing a complete data traceability chain for the indicator fluctuation analysis module.

[0054] Example 2: See Figure 3When implementing building energy consumption control, the regulatory strategy generation unit first extracts the median value from the energy consumption standard range as the energy consumption benchmark value. This benchmark value is obtained through statistical analysis of historical energy consumption data, comprehensively considering the building's functional type, usage period, and external environmental factors. The system uses a sliding window algorithm to dynamically update the benchmark value, with a window period set to 24 hours, ensuring that the benchmark value reflects the latest energy consumption characteristics. The absolute difference between the current building energy consumption index and the energy consumption benchmark value is calculated by comparing real-time sampled data with the benchmark value point by point. The difference result is converted into a standard deviation through normalization. When this deviation is input into the energy consumption dynamic adjustment model, the model first performs feature decomposition on the deviation to identify whether the main source of the deviation is base load fluctuation or peak load impact. For base load fluctuation, the model generates a gradual adjustment command; for peak load impact, the model generates a rapid suppression command. The generation process of power adjustment commands adopts a hierarchical decision-making mechanism, with lower-level commands directly controlling equipment operating parameters and higher-level commands coordinating energy consumption distribution among multiple devices.

[0055] The process of structural safety reinforcement is based on real-time monitoring values ​​of structural safety indicators, obtained from strain sensor arrays installed at key parts of the building. The system first filters the raw monitoring data to eliminate transient interference signals, then calculates the deviation between the real-time monitoring values ​​and the safety threshold. The deviation is calculated as a percentage, reflecting the degree of proximity of the current structural state to the safety boundary. When the deviation exceeds a preset threshold, the system initiates the structural reinforcement scheme generation process. This process first analyzes the spatial distribution characteristics of the deviation, identifies high stress concentration areas, and then matches predefined reinforcement templates based on the area characteristics. The reinforcement scheme is graded based on the cumulative effect of the deviation; lower-level schemes address localized stress anomalies, while higher-level schemes address overall structural stability issues. After the scheme is generated, the system automatically verifies the feasibility of implementation conditions, including material availability and construction space limitations, and finally outputs an executable set of reinforcement instructions.

[0056] The core of optimizing building environmental protection indicators is establishing a dynamic balance curve for these indicators. This curve, constructed by integrating historical environmental data and real-time monitoring values, reflects the normal fluctuation range of pollutant concentrations. The system employs a dynamic time warping algorithm to align real-time data with the curve's baseline, identifying the current deviation of environmental indicators. The calculation of the deviation considers not only the absolute value but also analyzes its changing trend, distinguishing between instantaneous fluctuations and continuous deviations. The operational parameter correction commands for pollution control equipment are generated based on the spatiotemporal characteristics of the deviation: for local instantaneous deviations, the commands focus on adjusting equipment operating parameters; for wide-area continuous deviations, the commands include equipment collaborative control strategies. Before issuing commands, the system simulates the correction effect, predicting the trajectory of pollutant concentration changes to ensure that the expected effect of the correction commands matches the actual needs.

[0057] The processing logic of the indicator fluctuation analysis module revolves around strategy response delay and indicator recovery efficiency. The module first obtains the occurrence times of energy consumption anomaly indicators, structural risk indicators, and environmental defect indicators from the regulatory strategy generation unit, recording the actual building indicator values ​​at the time each indicator is triggered. These values ​​are then compared twice with the corresponding indicator's standard upper and lower limits; the comparison results are used to calculate key offset feature quantities. The extraction process for key offset feature quantities employs a minimum value selection principle to ensure that the feature quantities reflect the most pressing abnormal states. The formula for generating the strategy response delay coefficient is: in, This is the strategy response delay coefficient. This represents the minimum deviation of the actual building indicator value from the upper limit of the standard. This is the minimum offset from the lower limit of the standard value. and These are the weighting factors for the upper and lower limits of the standard, respectively. The weighting factors are dynamically adjusted according to the indicator type, with structural safety indicators having a higher weight than energy consumption and environmental protection indicators. When the strategy response delay coefficient exceeds the delay tolerance threshold, the module generates a strategy lag alarm signal, which includes the delay time, deviation type, and impact range.

[0058] The evaluation of indicator recovery efficiency is based on the time span from the generation of anomaly markers to the restoration of building indicators to normal values. The system presets a standard recovery time as a benchmark, with the time span calculated to the second, and deducting inherent delays in strategy transmission and execution. Conditions for generating indicator recovery hysteresis signals include exceeding the time span limit and abnormal recovery trajectories. Anomaly detection of recovery trajectories employs a curve fitting algorithm, comparing the deviation of the actual recovery trajectory from the ideal recovery model. After signal generation, the module automatically correlates with similar historical events, analyzing the repetitiveness and regularity of hysteresis phenomena to provide data support for subsequent strategy optimization.

[0059] At the data processing level, the module adopts a streaming computing architecture to process input data streams and output analysis results in real time. Raw data is cleaned, aligned, and transformed before being stored in a time-series database, while analysis results are written to a high-performance cache for use by other modules. The module internally employs an event queue management strategy to delay alarm signals and indicator recovery lag signals, ensuring that high-priority events are processed promptly. Detailed logs are recorded for all analysis processes, including data sources, processing steps, and output results, forming a complete audit trail. Log data is transmitted encrypted to a central storage system, supporting post-event querying and analysis.

[0060] The system resource allocation employs a dynamic load balancing mechanism. When multiple building indicators simultaneously exhibit abnormalities, the module allocates computing resources based on the strategy response latency coefficient and the severity of the indicator recovery hysteresis signal. Structural safety-related analysis tasks receive priority resource allocation, while energy consumption and environmental protection tasks are allocated on demand. The resource scheduling algorithm monitors the system load status in real time and automatically activates a degradation processing mode under overload conditions to ensure the continuous operation of core functions. The interface between the module and the regulatory strategy generation unit uses a bidirectional communication protocol, supporting real-time feedback of analysis results and dynamic adjustment of strategy parameters.

[0061] Closed-loop management of anomalies is achieved through a state machine. Each anomaly flag undergoes multiple state transitions from generation to clearance, including pending processing, analysis in progress, strategy execution, and recovery verification. State transitions trigger corresponding data processing logic and update the event's time attributes. The module periodically generates anomaly event statistical reports, summarizing the frequency of occurrence, processing time, and recovery effectiveness of various flags. The report data is used to optimize system parameters and algorithm configurations. The report output format is compatible with standard data analysis tools, supporting further visualization and in-depth analysis.

[0062] Example 3: See Figure 4 The indicator correlation module operates on an event-driven architecture. When it receives a strategy lag alarm signal from the indicator fluctuation analysis module, the signal data packet contains two core parameters: the strategy response latency coefficient τ and the latency tolerance threshold τ_max. The system calculates the data transmission acceleration factor ε using a scaling algorithm. in, Indicates the maximum allowed data transmission delay (milliseconds). This represents the real-time policy response latency coefficient. The tolerance margin constant set for the system (default is 0.85). The acceleration factor ε is limited to the range of 0.5 to 3.0, and a critical value is automatically applied when the calculation result exceeds the boundary. The calculation results are sent to the control center of the building index acquisition unit in real time. The center reconstructs the data acquisition timing based on the ε value: assuming the original sampling interval is T seconds, the new interval T_new = T / ε. This reconstruction process adopts a phased smooth transition strategy to avoid timing breaks caused by sudden changes in data sampling frequency. For distributed sensor networks, each node dynamically fine-tunes its execution parameters based on its own communication quality, and the central node aggregates the actual sampling rate of the entire network every 30 seconds for closed-loop verification.

[0063] The strategy execution intensity optimization is implemented for the indicator recovery hysteresis signal. The signal data packet embeds the indicator recovery time span Δt, the standard recovery time Δt_s, and the hysteresis factor κ (κ=Δt / Δt_s). The logic of the indicator association module to generate the strategy execution intensity enhancement instruction includes three levels of judgment: when 1.2≤κ<1.5, a light instruction with an intensity increase of 20% is generated; when 1.5≤κ<2.0, a medium instruction with an increase of 50% is generated; and when κ≥2.0, a heavy instruction with an increase of 80% is generated. The instruction encoding adopts the binary marking method, with the first bit marking the energy consumption strategy type (01), safety strategy (10), or environmental protection strategy (11), and the last 16 bits storing the hexadecimal increase parameter. The instruction is transmitted directly to the strategy executor of the regulatory strategy generation unit through a high-priority channel. After decoding, the executor implements the intensity adjustment: for the energy consumption graded control strategy, the increase is reflected in the increased proportion of the power limit threshold reduction; the structural reinforcement strategy is reflected in the increase of the thickness of the reinforcement material; and the environmental protection strategy is transformed into an increased adjustment range of the operating parameters of the pollution control equipment. All adjustment operations are subject to a mandatory upper limit to prevent secondary system oscillations caused by over-adjustment.

[0064] The strategy effectiveness assessment unit continuously monitors the signal stream output by the indicator fluctuation analysis module using a preset monitoring period C (default 24 hours) as the time window. The signal capture mechanism employs timestamp alignment technology, marking the precise nanosecond-level occurrence time of each strategy lag alarm signal or indicator recovery delay signal. The calculation formula for the system failure rate μ abandons the simple counting method and adopts a weighted cumulative model: in, The total number of signals, This represents the duration (in minutes) of the i-th signal. The weighting coefficients for this signal type are: (1.0 for structural risk signals, 0.7 for energy consumption anomalies, and 0.5 for environmental defects). The calculation results are compared with the failure threshold. (Default value is 0.35) During comparison, a moving average filter is used to eliminate instantaneous fluctuation interference. When μ > 0.35 for three consecutive regulatory periods... At that time, the system generates a building regulatory failure identifier containing a unique geocode for the building. This identifier is transmitted to the spatial annotation engine of the 3D model mapping unit using a three-layer encryption protocol.

[0065] Upon receiving the identifier, the spatial labeling engine activates the geographic matching process: first, it parses the geocode to obtain the building's 3D coordinates (x, y, z), and then locates the corresponding spatial voxel in the building indicator spatial distribution map. The labeling operation is implemented in two dimensions: visual and data structure. The visual layer reduces the target building's transparency by 40% and overlays a red pulsed halo; the data layer writes a failure label into the spatial voxel metadata and simultaneously creates an independent monitoring channel to continuously collect the building's core indicators. The labeling results are synchronized in real time to the strategy decision tree of each functional unit. The monitoring strategy generation unit activates backup control strategies for the failed building, while the indicator fluctuation analysis module increases the building's signal weight by 1.5 times.

[0066] The core data storage and verification mechanism employs distributed ledger technology. Each operation record of the building indicator collection unit adjusting the sampling frequency, the response delay time of the regulatory strategy generation unit receiving intensity enhancement instructions, and the raw data of the strategy effectiveness judgment unit calculating the failure rate are all generated into independent data blocks. These blocks are linked through a Merkle tree structure, and every ten blocks form a verification batch uploaded to the blockchain node. When the system performs cross-module data verification (such as verifying the intensity enhancement effect of the strategy execution by the indicator association module), trusted data traceability is achieved through smart contract calls, eliminating the risk of single-point data tampering.

[0067] A dynamic resource scheduling mechanism ensures stability during high-load operation. When the strategy validity judgment unit detects that the μ value is close to... In such cases, the system automatically triggers resource reallocation: downgrading the indicator analysis tasks of non-critical buildings to sampling mode; limiting the rendering frame rate of 3D models to a basic level; and compressing the precision of historical data storage. Simultaneously, a priority channel is established to handle failed building data streams, ensuring that this data occupies at least 40% of the system's computing resources. The resource scheduling status panel displays the resource occupancy rate of each unit in real time, and automatically triggers a circuit breaker mechanism to suspend the lowest priority tasks when abnormal limits are exceeded.

[0068] The fault recovery process employs a phased rollback strategy. This rollback occurs when the failure rate of a building is μ < 0.8 for two consecutive monitoring periods. Upon completion, the system initiates a recovery process: the first phase retains the annotations but restores the basic strategy channels; the second phase gradually increases the sampling frequency to the standard value; the third phase removes visual annotations but retains metadata tags; and the final phase completely removes the failure markers after the building indicators have met the standards for 72 consecutive hours. Each phase transition requires dual verification: continuous compliance of data indicators and stable system resource utilization within safe thresholds. The data trajectory during the recovery process is stored separately for subsequent building vulnerability prediction models.

[0069] The system parameters are dynamically optimized using a reinforcement learning framework. Each time a building monitoring failure indicator is generated, the system automatically initiates a parameter optimization program: collecting raw data streams and failure judgment process records within the monitoring cycle C, and inputting them into a deep Q-network model to generate parameter tuning suggestions. The core optimization objects include the fluctuation range of the delay tolerance threshold, the baseline value of the tolerance margin δ, and the weight coefficients of various signals. The optimization results were verified through A / B testing: the target building complex was randomly divided into an experimental group and a control group. The experimental group was run with the new parameters, and the difference in system failure rates between the two groups was compared in the next monitoring cycle. The verification period lasted for three monitoring cycles before the significantly effective new parameters were officially written into the system configuration library.

[0070] Example 4: The spatial matching process of the 3D model mapping unit begins with the extraction of building spatial coordinate data. Taking a commercial complex as an example, the system obtains the precise 3D coordinates of the building (longitude 121.4737°, latitude 31.2304°, altitude 24.5 meters) from the urban geographic information database, and simultaneously loads the structural topology data from the building's BIM model. The conversion of building energy consumption indicators is achieved through thermodynamic simulation: the electricity data of individual metering items (lighting system 38.6 kWh per hour, air conditioning system 72.3 kWh per hour, elevator system 15.2 kWh per hour) are input into the computational fluid dynamics model, and the building surface temperature distribution matrix is ​​output. After rasterization, this matrix generates thermal distribution gradient data, with each raster cell (0.5m × 0.5m) recording the temperature gradient value and heat flow direction.

[0071] The conversion of structural safety indicators relies on real-time data collected by a fiber optic sensor network. The system inputs data from this location and eight adjacent monitoring points into the finite element analysis module, calculating a spherical region with a stress diffusion range of 3.2 meters in radius, forming a structural stress distribution data package containing the stress peak location and attenuation curve. Each spatial point in the data package contains stress components and a composite vector along three axes, stored in a spatiotemporal database in JSON format.

[0072] The conversion process of environmental indicators involves the fusion of multi-source data. An emission monitoring system at a manufacturing plant recorded an instantaneous PM2.5 concentration of 78 μg / m³. Simultaneously, the system acquired real-time wind speed data (2.8 m / s, northwest direction) from a weather station. A pollutant dispersion model, based on a Gaussian plume algorithm and incorporating the building's 3D morphological features, generated pollutant dispersion trajectory data including concentration contour lines and deposition trajectories. The data was updated every minute, with each calculation consuming approximately 120 MB of memory and 0.8 seconds of CPU time.

[0073] The generation of the multidimensional spatial distribution map of building indicators employs a layered rendering technique. The system first maps thermal distribution gradient data to a red spectrum (the higher the temperature, the darker the hue), converts structural stress distribution data into a blue transparent layer (the greater the stress, the lower the transparency), and presents pollutant diffusion trajectory data as a green dynamic particle flow. These three types of data are aligned using spatial coordinates in the 3D engine to form a composite layer with depth information. The table below shows a fragment of multidimensional spatial distribution data for a building complex in a certain area:

[0074] Table 1: A data segment showing the spatial distribution of multidimensional indicators of a building complex in a certain area.

[0075] Building No. Mean value of thermal gradient (kW / m2) Maximum stress value (MPa) PM2.5 diffusion radius (m) Type of spatial overlap B-1027 0.86 98.4 32.7 Thermal-stress B-1028 1.12 45.2 28.9 Thermal-environmental B-1029 0.67 112.8 41.5 Triple overlap

[0076] The indicator association module executes a spatial feature recognition algorithm when processing the map. For thermal distribution gradient data, the system automatically delineates high-energy-consumption monitoring areas when the temperature gradient of three or more consecutive grid cells exceeds 0.75 kW / m². Analysis of an office building cluster showed a thermal accumulation zone of 56 m² in the core tube area, which the system marked as a Class A monitoring object. The identification of structural stress anomalies uses the spatial derivative method; when the stress value at a point differs from the average stress of its eight surrounding points by more than 15%, it is determined to be a critical point for safety control. Such anomalies were detected in the suspension cable anchorage area of ​​a bridge structure, and the system immediately triggered a local scan command to obtain higher-precision data.

[0077] The analysis of intersection areas of pollutant diffusion trajectories employs graph theory algorithms. The system abstracts each diffusion trajectory as a directed edge, with nodes representing spatial coordinates. When three or more trajectories intersect within a 5-meter diameter area, they are marked as the core area for environmental remediation. Analysis of a chemical industrial park revealed two intersection areas covering emission outlets 3 and 7, respectively. The system automatically generated a correlation report and sent it to the environmental regulatory department.

[0078] The generation of multi-indicator collaborative monitoring strategies is based on spatial overlap analysis. When there is more than 50% overlap between high-energy-consumption monitoring areas and key safety control points, the system generates a joint "energy consumption-safety" strategy: prioritizing the implementation of structural reinforcement and energy consumption adjustment plans. In a data center case, server rack areas simultaneously experienced heat accumulation and abnormal floor stress. The system's joint strategy included: firstly, reinforcing the steel beams under the racks within 30 minutes; and secondly, reducing the air conditioning supply temperature by 2°C. For triple-overlapping areas (heat, stress, and environmental protection), the strategy generation logic is more complex: analysis of a steel plant's smelting workshop showed a triple-overlapping area of ​​120㎡. The system activated a three-level response mechanism—immediately halting production for structural testing, simultaneously activating the backup environmental purification system, and implementing total energy consumption control after production resumes.

[0079] The interaction between the 3D model mapping unit and the regulatory strategy generation unit adopts an event bus architecture. Each time the spatial distribution map is updated, the system publishes an event message containing the following fields: building ID, indicator type, spatial coordinate set, and timestamp. After subscribing to relevant events, the regulatory strategy generation unit integrates spatial feature data into the strategy decision tree. For example, a thermal gradient update event for a hospital building triggers a strategy adjustment: the air conditioning system renovation plan, originally scheduled for the next day, is moved forward to the current time period, and the frequency of temperature monitoring in the operating room area is increased.

[0080] The historical data backtracking function enables dynamic strategy optimization. The system stores a snapshot of the spatial distribution map for 30 consecutive days. When a certain type of spatial feature is detected to recur, historical data is automatically retrieved for comparison and analysis. For example, if a commercial district experiences heat accumulation in the same location for three consecutive weeks, the system will find a flaw in the air conditioning load allocation algorithm for that area through backtracking and generate a permanent renovation suggestion: re-divide the air conditioning zones and add terminal regulating valves.

[0081] An anomaly handling mechanism ensures the reliability of spatial matching. When a coordinate transformation deviation exceeds 0.3 meters, the system automatically triggers a LiDAR calibration procedure; when data layers are misaligned, an automatic registration algorithm based on feature points is initiated; when the real-time rendering frame rate is below 15fps, the modeling accuracy of non-critical areas is dynamically reduced. In a coordinate offset fault that occurred after a system upgrade, the calibration procedure completed the coordinate remapping of all buildings within 142 seconds, with the error controlled within ±0.05 meters.

[0082] The resource allocation for spatial analysis tasks is dynamically prioritized. Structural safety-related computational tasks have the highest priority, exclusively using 40% of the GPU computing resources; environmental indicator update tasks are set to medium priority, sharing 30% of the resources; thermal analysis, as a basic task, uses the remaining resources. When a sudden structural alarm is triggered, the system immediately preempts resources: suspending all non-safety-related computational tasks to ensure stress analysis is completed within 500 milliseconds. In a subway station structural monitoring case, this system mechanism ensured that after detecting signs of column crack propagation, a full-station stress field simulation and reinforcement plan were completed within 1.2 seconds.

[0083] Example 5: The index association module's analysis of the spatial distribution map of building indicators begins with spatial topology analysis. The module's built-in point cloud clustering algorithm processes thermal distribution gradient data: first, the building surface is divided into grid cells with 0.2-meter precision; the standard deviation of thermal values ​​in five adjacent grid cells is calculated; when the standard deviation exceeds a set floating threshold three times consecutively, it is determined to be a valid cluster area. The spatial feature engine scans the entire building complex, merging thermal cluster areas within three meters to form a polygonal boundary of the high-energy-consumption monitoring area. During the analysis of an industrial park, four dispersed high-temperature thermal points were identified and connected within 15 seconds, generating a monitoring area of ​​82 square meters. The system automatically labels the equipment number list and monitoring level code of the corresponding buildings in this area.

[0084] Anomaly detection in structural stress distribution data employs a multi-scale analysis method. Spatially, the system establishes an octree-based hierarchical index, first filtering candidate points at a 1-meter resolution level, then switching to a 0.5-meter resolution for precise judgment. The judgment rules include two core conditions: the stress value at the current point exceeds 1.7 times the regional average, and its gradient change rate differs from that of adjacent points by more than 25%. In the inspection of a stadium roof structure, the system detected a standard-compliant anomaly at a cable net node in the southeast corner. After marking it as a critical safety control point, a special monitoring instruction was generated: a millimeter-wave radar scan was initiated within a three-meter radius of the point, increasing the data sampling frequency from minutes to seconds.

[0085] The identification of intersection areas of pollutant diffusion trajectories relies on a spatiotemporal correlation engine. The system discretizes each diffusion trajectory into a time-space sequence of points, establishing a four-dimensional index structure (X coordinate, Y coordinate, Z height, timestamp). When the Euclidean distance between the point sets of different trajectories in three-dimensional space is less than a preset threshold and the time difference is within an allowable range, a trajectory intersection event is generated. In a wastewater treatment plant monitoring case, two ammonia diffusion trajectories from the sludge drying workshop and the biological treatment tank spatially intersect at a height of 12 meters. The system marks this overlapping area as the core area for environmental governance, and the core area automatically expands to a spherical space with a diameter of 10 meters at the intersection point, encompassing all seven sets of environmental facility identification codes within this area.

[0086] The generation mechanism of the multi-indicator collaborative supervision strategy follows the spatial overlap feature analysis rules. The module configures a strategy type discrimination matrix: when there is any form of spatial overlap between the high-energy-consumption supervision area and the key safety control point, the strategy combination mapping protocol is activated. The protocol first calculates the area ratio of the overlapping area; when the ratio exceeds 30%, a "safety-first" strategy sequence is generated: the first stage freezes the energy adjustment plan and forcibly initiates the structural reinforcement process; the second stage resumes energy consumption optimization after the safety alarm is lifted. In a commercial building, the air conditioning room area simultaneously triggers heat accumulation and abnormal floor vibration. The system-generated collaborative strategy includes three steps: complete the reinforcement of the room's supporting structure within 1 hour; begin replacing the high-efficiency motor 2 hours after reinforcement; and install a heat recovery device within 72 hours.

[0087] For areas with overlapping high energy consumption, safety-critical points, and environmental protection core areas, the system adopts a progressive response strategy. A 3D scan of a chemical plant's reactor area revealed these overlapping features. The strategy engine completed a tiered response plan within 200 milliseconds: immediately implementing equipment shutdown protection; simultaneously activating the emergency cooling system to reduce heat load; the structural reinforcement team arriving on-site within 40 minutes; and the environmental purification unit operating at maximum power. During strategy execution, a chain of triggers was set up, unlocking subsequent operations only when the previous stage's task completion reached 90%. Task progress was confirmed in real-time via IoT devices.

[0088] The encoding and transmission of policy instructions employ layered compression technology. The policy logic unit converts multi-indicator collaborative monitoring policies into binary instruction packets. Each packet contains three data segments: the first segment, a 16-bit segment, stores a policy type marker (0001 for a single-indicator independent policy, 0010 for dual-indicator collaborative policy, and 0011 for triple-indicator collaborative policy); the middle segment stores the coordinate set of the action area, using the Douglas-Pock algorithm to compress spatial point data; and the last segment contains specific execution parameters, such as structural reinforcement thickness or equipment power adjustment. The instruction packets are transmitted to the monitoring policy generation unit via a dedicated message queue. The transmission protocol employs a triple verification mechanism, requiring the receiver to return an acknowledgment frame containing a checksum after each data packet transmission.

[0089] The strategy fusion execution process implements dynamic priority scheduling. Upon receiving the collaborative strategy, the monitoring strategy generation unit activates the strategy coordinator: first, it parses the header of the instruction packet to determine the strategy type; then, it retrieves all relevant strategies in the current execution queue; and finally, it intelligently terminates conflicting strategies. In a case involving a conflict between an energy optimization strategy and a sudden structural reinforcement strategy in an office area, the system automatically saves the energy consumption adjustment parameters and terminates execution, while simultaneously opening an independent control channel for the new strategy. The adjustment of the strategy's impact is based on real-time building indicator data, scanning the strategy's execution effect every five minutes and fine-tuning parameter values. All adjustment operations are recorded in a distributed log, forming a chain of evidence for strategy optimization.

[0090] Changes in spatial characteristics trigger a self-correction mechanism for the strategy. The 3D model mapping unit pushes an updated spatial distribution map package to the indicator association module every ten minutes. When a change in the spatial attributes of a monitored area is detected, the strategy adaptive engine initiates a strategy reassessment process within 0.5 seconds. For example, if a heat-concentrating area in a shopping mall is deformed due to tenant renovations, the system detects the boundary change, recalculates the area parameters, and generates a correction instruction containing a new air conditioning zoning scheme to replace the original strategy.

[0091] A closed-loop feedback channel is established for quality monitoring. During the execution of multi-indicator collaborative strategies, the regulatory strategy generation unit continuously transmits strategy-effectiveness time slot data back to the indicator association module: the time difference between the issuance of structural reinforcement commands and the detection of stress changes by sensors; the delay period between the transmission of energy consumption adjustment commands and the response of electricity meter readings; and the correlation curve between environmental protection equipment parameter adjustments and emission concentration changes. The feedback data stream is updated at 5-second intervals, and the indicator association module dynamically optimizes spatial clustering algorithm parameters accordingly, such as adjusting the detection threshold for thermal fluctuations from 0.25 kW to 0.18 kW and expanding the judgment radius of safety critical points from 0.5 meters to 0.8 meters. This optimization process runs independently in the data processing memory area without interfering with the online business system, and the results are automatically synchronized to the main system parameter database after 24-hour verification.

[0092] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0093] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart city building indicator monitoring system based on AI algorithms, characterized in that, include: The building indicator acquisition unit is used to acquire building indicators in real time, including building energy consumption indicators, building structural safety indicators, and building environmental protection indicators. The regulatory strategy generation unit performs multi-dimensional threshold comparison of building indicators, generates difference signals based on the comparison results, and outputs corresponding building regulatory strategies based on the difference signals. The indicator fluctuation analysis module continuously receives building regulatory strategy execution records and building indicator change processes sent by the regulatory strategy generation unit. It extracts strategy lag characteristic values ​​by analyzing the building regulatory strategy execution records and calculates indicator recovery efficiency characteristic values ​​by analyzing the building indicator change process. The indicator association module obtains the strategy lag characteristic value and the indicator recovery efficiency characteristic value from the indicator fluctuation analysis module. Based on the strategy lag characteristic value, it adjusts the data transmission frequency of the building indicator collection unit, and at the same time optimizes the execution intensity of the building supervision strategy based on the indicator recovery efficiency characteristic value. The 3D model mapping unit spatially matches building indicators with the city's 3D geographic information model to generate a spatial distribution map of building indicators. The strategy effectiveness judgment unit obtains abnormal signals of strategy execution through the indicator fluctuation analysis module, counts the frequency of abnormal signals within the preset supervision period, and generates a building supervision failure mark when the frequency of abnormal signals exceeds the preset stability threshold. The indicator fluctuation analysis module obtains the occurrence time points of energy consumption anomaly indicators, structural risk indicators, and environmental defect indicators through the regulatory strategy generation unit, records the actual building indicator values ​​corresponding to the occurrence time points, and performs a double comparison between the actual building indicator values ​​and the standard upper limit and standard lower limit values ​​of the corresponding indicators. The indicator fluctuation analysis module calculates the minimum deviation of the actual building indicator values ​​from the standard upper limit value and the minimum deviation from the standard lower limit value, and selects the smaller of the two minimum deviation values ​​as the key deviation feature. The strategy response delay coefficient is generated by weighting the key deviation feature value with the standard upper limit and standard lower limit values. When the strategy response delay coefficient is greater than the preset delay tolerance threshold, a strategy lag alarm signal is generated. The indicator fluctuation analysis module also records the time span from the generation of the anomaly indicator to the recovery of the building indicator to the normal value, compares this time span with the preset standard recovery time, and generates an indicator recovery hysteresis signal if the time span exceeds the standard recovery time. The spatial matching process performed by the three-dimensional model mapping unit includes: extracting the spatial coordinate data of the building in the three-dimensional geographic information model, converting the building energy consumption index into thermal distribution gradient data, converting the building structural safety index into structural stress distribution data, and converting the building environmental protection index into pollutant diffusion trajectory data. A multidimensional spatial distribution map of building indicators is generated by integrating thermal distribution gradient data, structural stress distribution data, and pollutant diffusion trajectory data. The three-dimensional model mapping unit transmits the spatial distribution map of building indicators to the indicator association module in real time. The indicator association module adjusts the priority of regulatory strategies based on the spatial clustering characteristics in the spatial distribution map of building indicators.

2. The smart city building indicator monitoring system based on AI algorithm according to claim 1, characterized in that, The regulatory strategy generation unit acquires building energy consumption indicators, building structural safety indicators, and building environmental protection indicators through the building indicator collection unit. It compares the building energy consumption indicators with preset energy consumption standard ranges. If the building energy consumption indicators exceed the energy consumption standard range, an energy consumption anomaly indicator is generated. It compares the building structural safety indicators with preset safety thresholds. If the building structural safety indicators are below the safety thresholds, a structural risk indicator is generated. It compares the building environmental protection indicators with preset environmental compliance thresholds. If the building environmental protection indicators do not meet the environmental compliance thresholds, an environmental defect indicator is generated. When a structural risk indicator is generated, the regulatory strategy generation unit immediately triggers a building structural reinforcement strategy; when an energy consumption anomaly indicator is generated, it initiates an energy consumption tiered control strategy; and when an environmental defect indicator is generated, it executes an environmental protection facility optimization strategy.

3. The smart city building indicator monitoring system based on AI algorithm according to claim 1, characterized in that, The process of the regulatory strategy generation unit in implementing building energy consumption regulation includes: selecting the median value of the energy consumption standard range as the energy consumption benchmark value, calculating the absolute difference between the current building energy consumption index and the energy consumption benchmark value, inputting the absolute difference into the energy consumption dynamic adjustment model to generate power adjustment instructions, and controlling the operating parameters of the building energy system; the process of implementing building structural safety reinforcement includes: obtaining real-time monitoring values ​​of structural safety indicators, calculating the deviation between the real-time monitoring values ​​and the safety critical values, and generating structural reinforcement schemes of different levels according to the magnitude of the deviation; the process of implementing building environmental protection indicator optimization includes: establishing a dynamic balance curve of environmental protection indicators, identifying the offset of the current environmental protection indicators in the dynamic balance curve, and generating operating parameter correction instructions for pollution control equipment according to the offset.

4. The smart city building indicator monitoring system based on AI algorithm according to claim 1, characterized in that, The indicator association module receives the strategy lag alarm signal and strategy response delay coefficient sent by the indicator fluctuation analysis module, calculates the data transmission acceleration factor based on the ratio of the strategy response delay coefficient to the delay tolerance threshold, and sends the data transmission acceleration factor to the building indicator acquisition unit. The building index acquisition unit shortens the duration of the dynamic time interval based on the data transmission acceleration factor; The indicator association module simultaneously receives the indicator recovery hysteresis signal sent by the indicator fluctuation analysis module, generates a strategy execution intensity enhancement instruction, and sends it to the regulatory strategy generation unit. The regulatory strategy generation unit increases the effectiveness of the building regulatory strategy according to the strategy execution intensity enhancement instruction.

5. The smart city building indicator monitoring system based on AI algorithm according to claim 1, characterized in that, The strategy effectiveness judgment unit continuously acquires strategy lag alarm signals and indicator recovery delay signals generated by the indicator fluctuation analysis module, and accumulates the number of signal occurrences within a single regulatory cycle to form the total number of abnormal signals. The system failure rate is obtained by calculating the ratio of the total number of abnormal signals to the total duration of the monitoring cycle. When the system failure rate reaches the preset failure threshold, the strategy effectiveness judgment unit sends a building monitoring failure identifier to the three-dimensional model mapping unit. After receiving the building monitoring failure identifier, the three-dimensional model mapping unit marks the monitoring failure area of ​​the corresponding building in the building indicator spatial distribution map.

6. The smart city building indicator monitoring system based on AI algorithm according to claim 2, characterized in that, The process of the indicator fluctuation analysis module for processing structural risk identifiers includes: identifying the spatial location of the building corresponding to the structural risk identifier and retrieving the structural deformation record data of that location within the historical monitoring period; calculating the average rate of change and acceleration characteristic value of the structural deformation record data; predicting the development trend of structural risk based on the combination relationship of the average rate of change and acceleration characteristic value; and when the development trend of structural risk exceeds the preset safety evolution trajectory, the indicator fluctuation analysis module sends a structural reinforcement instruction to the regulatory strategy generation unit.

7. A smart city building indicator monitoring system based on AI algorithm according to claim 1, characterized in that, The process of the indicator association module in processing the spatial distribution map of building indicators includes: identifying spatial clustering areas of thermal distribution gradient data in the map and marking them as high-energy-consumption monitoring areas; identifying spatial anomalies in structural stress distribution data and marking them as key points for safety control; identifying intersection areas of pollutant diffusion trajectory data and marking them as core areas for environmental governance; generating a multi-indicator collaborative monitoring strategy based on the spatial overlap relationship between high-energy-consumption monitoring areas, key points for safety control, and core areas for environmental governance; and sending the multi-indicator collaborative monitoring strategy to the monitoring strategy generation unit for strategy fusion and execution.

8. A smart city building indicator monitoring method based on AI algorithm, applied to the smart city building indicator monitoring system based on AI algorithm as described in claim 7, characterized in that, include: Real-time energy consumption parameters, structural stress parameters, and environmental emission parameters of buildings are collected through distributed sensing devices. The collected building parameters are compared with preset energy consumption benchmark values, structural safety critical values ​​and environmental protection standard values ​​in multiple dimensions; based on the comparison results, energy consumption control strategies, structural reinforcement strategies and environmental governance strategies are generated. Real-time monitoring of building parameters during strategy execution, and calculation of parameter recovery rate and strategy response delay characteristics; The system dynamically adjusts the collection frequency of building parameters based on the strategy response delay characteristic; it transforms the building parameter change trajectory into a spatial thermal distribution model through three-dimensional spatial mapping technology; it identifies abnormal parameter aggregation areas in the spatial thermal distribution model and generates a spatial collaborative monitoring scheme; it statistically analyzes the frequency of abnormal events in strategy execution within a preset period, and triggers a system self-correction mechanism when the frequency exceeds a stability threshold.

Images