Urban area comprehensive energy consumption analysis system and method based on multi-source data fusion, and storage medium

Abstract

The application discloses a kind of urban area comprehensive energy consumption analysis system and method based on multi-source data fusion and storage medium, it is related to urban energy management technical field, the system includes: data fusion unit cross-departmental collection heterogeneous data, generates full-dimension fusion feature library by joint modeling;Benchmark construction unit constructs differentiating energy consumption benchmark system accordingly;Factor identification unit filters energy consumption driving factor to generate correlation model and factor library;Supervision execution unit identifies abnormal energy consumption and early warning;Linkage optimization unit builds auxiliary decision engine, iteratively updates algorithm parameters, synchronously plans control requirement to supervision unit;The application builds cross-departmental heterogeneous data acquisition system, constructs fusion feature library, guarantees data security privacy, breaks data island, provides solid data support for energy consumption analysis;At the same time, construct whole-process urban area energy consumption analysis and control system, realize planning approval and energy consumption supervision linkage optimization, form closed loop.

Description

Technical Field

[0001] This invention relates to the field of urban energy management technology, specifically to an urban area integrated energy consumption analysis system, method, and storage medium based on multi-source data fusion. Background Technology

[0002] Under the development trend of green and low-carbon development and refined governance in cities, comprehensive energy consumption management in urban areas has become a core link in improving urban energy efficiency and implementing the dual control targets for energy consumption. The urban energy consumption system involves multiple government management departments and energy operation units, and the data related to energy consumption are complex and stored in a scattered manner, forming a heterogeneous data pattern across fields. At the same time, regional energy consumption is affected by multiple dynamic factors such as spatial form, population flow, meteorological environment, and industrial structure adjustment, which makes energy consumption analysis work need to take into account the integration of multi-dimensional characteristics and adapt to dynamic changes. In addition, the privacy protection and secure sharing of government data have become important prerequisites for cross-departmental data utilization. How to achieve effective integration of multi-source data on the basis of ensuring data security, and build an analysis system that fits the actual energy consumption patterns of cities to provide accurate basis for energy consumption supervision, has become a technical problem that urgently needs to be solved in the field of urban energy management. It also puts forward higher requirements for the comprehensiveness, accuracy and intelligence of energy consumption analysis technology.

[0003] Traditional urban energy consumption analysis technologies suffer from numerous technical shortcomings in practical applications, making it difficult to meet the current needs of refined urban energy consumption management. These technologies generally suffer from data silos, failing to achieve effective integration and deep correlation of heterogeneous data across departments. They rely solely on data from a single dimension or domain for analysis, resulting in energy consumption analysis results that cannot comprehensively reflect the true characteristics and inherent patterns of regional energy consumption. Furthermore, the energy consumption assessment benchmarks constructed by traditional technologies are mostly static standards, unable to adapt in real time to the impact of external dynamic factors such as changes in urban spatial form, population migration, and seasonal climate fluctuations, making it difficult to provide accurate quantitative assessment criteria. In addition, traditional energy consumption supervision is mostly limited to post-event data statistics and monitoring, lacking accurate identification and quantitative analysis of energy consumption driving factors, failing to effectively trace the root causes and responsible parties of abnormal energy consumption, and being disconnected from the urban planning approval process, failing to achieve pre-emptive energy consumption control from the planning source and form a full-process control system. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a comprehensive energy consumption analysis system, method, and storage medium for urban areas based on multi-source data fusion. The core of this invention revolves around the need for refined energy consumption management in urban areas, constructing an analysis system composed of five major units: data fusion, benchmark construction, factor identification, regulatory execution, and coordinated optimization. It achieves secure fusion of heterogeneous data across departments through a coupled privacy computing framework, constructs a full-dimensional fusion feature library, generates a differentiated energy consumption benchmark system based on a spatially coupled dynamic energy consumption benchmark algorithm, and filters energy consumption driving factors using a spatial-energy consumption feedback driving factor quantification algorithm. This enables accurate identification, tiered early warning, and accountability for abnormal energy consumption. Simultaneously, it embeds a planning energy consumption management auxiliary decision-making engine into the government approval process, iteratively optimizing algorithm parameters with regulatory data to achieve closed-loop management of energy consumption from planning to operational supervision. This effectively breaks down data silos, improves the refinement, intelligence, and collaboration of energy consumption management, and provides comprehensive technical support for the green and low-carbon development of cities.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: Firstly, a comprehensive urban area energy consumption analysis system based on multi-source data fusion. This system includes: a data fusion unit: used for cross-departmental collection of heterogeneous data, joint modeling through a privacy computing framework coupled with horizontal and vertical federation, completing spatiotemporal alignment and feature association, and generating a full-dimensional fusion feature library; the full-dimensional fusion feature library generated by the data fusion unit uses a 100m × 100m standard geographic grid as the spatial carrier and hourly timestamps as the time dimension, integrating three major categories of features: spatial, temporal, and attribute features. The spatial features include unique geographic codes for the grid, spatial coordinates, land use, plot ratio, building density, green space ratio, road network density, and mixed land use. The data includes: degree, 500-meter coverage of public transportation stations, density of public service facilities, GIS geospatial topology, and indicators of the control detailed planning plots; time-series features include hourly individual metering data for electricity, hourly individual household water and gas consumption data, hourly pipeline operation loss data, hourly population flow data, hourly traffic flow data, hourly meteorological and environmental data, hourly public transportation operation data, daily regional energy consumption summary data, and monthly energy consumption statistics; attribute-based features include: regional function type, building construction year, building floor structure, green building star rating, type of energy-consuming equipment, type of leading industry, number of enterprises above designated size, energy consumption per unit of output value, regional socio-economic data, government approval-related data, and basic information of energy-consuming entities. All features have undergone spatiotemporal alignment and standardization, and are associated and bound by grid and timestamp to form a complete feature set for each grid and each time node, constituting a full-dimensional fusion feature library. The benchmark construction unit is used to classify regional functional types based on the full-dimensional fusion feature library, extract temporal, attribute, and spatial morphological features, and construct regional type-specific energy consumption benchmark prediction models using a spatially coupled dynamic energy consumption benchmark algorithm. Model parameters are updated through a rolling time window, generating a differentiated energy consumption benchmark system adapted to different functional regions. The specific method for updating model parameters through a rolling time window is as follows: the benchmark construction unit independently configures a rolling time window for each subdivided functional region, and the window length can be dynamically set according to the regional functional attributes. For daily, weekly, monthly, or quarterly data, the baseline building unit uses the historical fusion data of the corresponding region over the past three years as the basic training set to complete the initial training of the model and form the initial parameter set. After each rolling time window period, the baseline building unit automatically adds the actual operating data generated in the region during the current period, the synchronously updated time series and attribute feature data, and the spatial morphology dynamic change data to the training set. Based on the incremental training strategy, the model parameters are iteratively updated without the need to retrain on the full historical data. The updated parameter set maps in real time the impact of external factors such as changes in regional functional attributes, seasonal climate fluctuations, population migration, industrial structure adjustment, and improvement of supporting facilities on regional energy consumption, maintaining the dynamic consistency between model parameters and the actual energy consumption patterns of the city.The differentiated energy consumption baseline system uses a 100m x 100m standard geographic grid as the spatial unit and the hour as the smallest time granularity. It outputs standardized energy consumption baseline content with multiple time dimensions and multiple types of thresholds for each subdivided functional area. Specifically, the differentiated energy consumption baseline system includes hourly, daily, monthly, and annual energy consumption baselines. The hourly energy consumption baseline matches the peak and off-peak energy consumption characteristics of electricity, water, and gas; the daily energy consumption baseline adapts to the summary of daily energy consumption patterns; the monthly energy consumption baseline corresponds to the periodic fluctuations in energy consumption caused by seasonal climate changes; and the annual energy consumption baseline directly connects to the regional energy consumption dual control indicators and annual energy-saving target requirements. The differentiated energy consumption baseline system also includes... It includes a Level 1 warning threshold, a Level 2 warning threshold, and a regional energy consumption control red line. The Level 1 warning threshold corresponds to an actual energy consumption deviation of 10%-20% from the baseline, the Level 2 warning threshold corresponds to an actual energy consumption deviation of 20%-50% from the baseline, and the regional energy consumption control red line corresponds to a critical value where the actual energy consumption deviation from the baseline exceeds 50%. This provides a unified, accurate, and directly applicable quantitative benchmark for subsequent abnormal energy consumption identification, tiered supervision and early warning, and targeted control. The factor identification unit is used to calibrate energy consumption data deviation based on the aforementioned full-dimensional fusion feature library and the differentiated energy consumption baseline system. It employs a space-energy consumption mutual feedback driving factor quantification algorithm to calculate the influence weight of spatial morphology indicators on energy consumption, screens energy consumption driving factors, and generates spatial morphology- An energy consumption characteristic correlation model and driving factor library; the specific method of the factor identification unit for energy consumption data deviation calibration is as follows: based on the differentiated energy consumption baseline system, the actual energy consumption data of each region is standardized, and energy consumption deviations caused by differences in regional functional types, extreme weather, temporary large-scale events, statutory holidays, and temporary population fluctuations are eliminated to obtain a standardized energy consumption intensity index that can be compared horizontally across regions and cycles; the regulatory execution unit is used to identify abnormal energy consumption behaviors exceeding the baseline and trace the responsible parties based on the full-dimensional fusion feature library, the differentiated energy consumption baseline system, and the energy consumption driving factors as control guidelines, and generate a graded regulatory early warning signal; the regulatory execution unit generates graded regulatory... The specific methods for issuing early warning signals are as follows: based on the magnitude, duration, and scope of abnormal energy consumption exceeding the baseline, they are divided into three levels: general early warning, important early warning, and emergency early warning. General early warnings are pushed to community grid workers and corresponding responsible entities; important early warnings are pushed to street offices and district-level industry authorities; and emergency early warnings are pushed to municipal authorities and government law enforcement units. Each early warning signal includes the location of the abnormality, the responsible entity, the root cause of the abnormality, and control recommendations. The linkage optimization unit is used to build a planning energy consumption control auxiliary decision-making engine embedded in the approval process, supported by the aforementioned spatial morphology-energy consumption characteristic correlation model and driving factor library. Iterative updates of the algorithm parameters of each unit are made using regulatory operation data as incremental samples, and planning control requirements are synchronized to the regulatory execution unit.The planning energy consumption control auxiliary decision-making engine built by the aforementioned linkage optimization unit is deeply embedded in the online government approval system of the natural resources and planning departments, connecting with the approval processes of detailed control planning, urban renewal projects, and state-owned construction land transfer. The input indicators of the planning energy consumption control auxiliary decision-making engine include plot ratio, building density, land use layout, road network planning, public transportation facilities, and public service facility layout. The output includes energy intensity calculation results, energy-saving potential calculation results, and planning optimization suggestions for the corresponding area of ​​the planning scheme.

[0006] Furthermore, the system includes: a data fusion unit: used for cross-departmental collection of heterogeneous data, joint modeling through a privacy computing framework coupled with horizontal and vertical federation, completing spatiotemporal alignment and feature association, and generating a full-dimensional fusion feature library; a benchmark construction unit: used for dividing regional functional types based on the full-dimensional fusion feature library, extracting temporal, attribute, and spatial morphological features, constructing regional type-specific energy consumption benchmark prediction models using a spatially coupled dynamic energy consumption benchmark algorithm, updating model parameters through a rolling time window, and generating a differentiated energy consumption benchmark system adapted to different functional regions; and a factor identification unit: used for energy consumption data deviation calibration based on the full-dimensional fusion feature library and the differentiated energy consumption benchmark system, employing spatial-energy consumption mutual... The feedback-driven factor quantification algorithm calculates the influence weight of spatial morphology indicators on energy consumption, filters energy consumption driving factors, and generates a spatial morphology-energy consumption feature correlation model and a driving factor library. The regulatory execution unit, based on the comprehensive feature library, uses the differentiated energy consumption baseline system as the evaluation criterion and energy consumption driving factors as control guidelines to identify abnormal energy consumption behaviors exceeding the baseline, trace the responsible parties, and generate tiered regulatory early warning signals. The linkage optimization unit, supported by the spatial morphology-energy consumption feature correlation model and driving factor library, builds a planning energy consumption control auxiliary decision-making engine embedded in the approval process, iteratively updates the algorithm parameters of each unit using regulatory operation data as incremental samples, and synchronizes planning control requirements to the regulatory execution unit.

[0007] Furthermore, the heterogeneous data collected by the data fusion unit specifically includes: regional industrial project approval data, energy consumption dual control indicator management data, and energy consumption filing data of enterprises above designated size from the Development and Reform Commission; building completion acceptance data, green building rating data, public building energy consumption statistics data, and municipal building operation and maintenance data from the Housing and Urban-Rural Development Commission; urban lighting, sanitation facilities, and municipal public utility operation energy consumption data from the Urban Management Commission; public transportation operation data, urban road network traffic data, and transportation hub operation and maintenance data from the Transportation Commission; sub-metering time series data, household water and gas consumption details, and pipeline operation loss data from power supply, water supply, and gas companies; urban GIS geospatial data and control detailed planning plot indicator data from the Natural Resources Commission; regional socio-economic data from the Statistics Commission; hourly meteorological environment data from the Meteorological Commission; population flow dynamic perception data collected by mobile phone signaling; and traffic flow data collected by traffic checkpoints and smart intersections.

[0008] Furthermore, in the data fusion unit, the privacy computing framework coupled with horizontal and vertical federation is deployed in the secure isolation zone of the government cloud. The framework independently deploys local computing nodes for each data provider, and all raw data is stored in the internal government network of each department. Each local node completes local data standardization preprocessing and local model calculation. Intermediate parameters for model training are transmitted between nodes through a national-level encrypted channel. The specific method for joint modeling of the privacy computing framework coupled with horizontal and vertical federation is as follows: Horizontal federated learning divides the entire administrative region into functional attributes such as residential streets, industrial parks, core business districts, transportation hubs, cultural and tourism areas, and municipal public areas. Local computing nodes of the same type of functional area are included in the same federated learning cluster to complete joint modeling. Vertical federated learning completes cross-departmental heterogeneous data sample ID matching for the same administrative region and the same energy user through an encrypted sample alignment algorithm. After matching, the local nodes of each department output the encrypted feature gradient of the corresponding sample to complete joint training.

[0009] Furthermore, the specific method for dividing the functional types of the reference construction unit into regions is as follows: the entire standard grid and administrative regions are divided into primary types: residential areas, commercial and business areas, industrial parks, municipal public areas, transportation hubs, and cultural and tourism leisure areas. Each primary type is further subdivided into secondary types. Among them, industrial parks are subdivided into high-tech industrial parks, traditional manufacturing industrial parks, and logistics and warehousing parks; residential areas are subdivided into old residential areas, newly built commercial housing areas, and affordable housing areas; commercial and business areas are subdivided into core business districts, regional commercial centers, and community commercial clusters; and transportation hubs are subdivided into rail transit hubs, highway passenger hubs, and highway passenger hubs. Transportation hubs and port shipping hubs; the temporal, attribute, and spatial morphological features extracted from the benchmark construction unit specifically include: temporal and attribute features such as the region's statutory planning function positioning, building construction year, floor structure, green building star rating, energy-consuming equipment type, average daily resident population, peak passenger flow, diurnal population difference, dominant industry type, number of large-scale enterprises, energy consumption per unit output value, hourly temperature, air humidity, wind speed, precipitation, and sunshine duration; spatial morphological features such as grid content area ratio, building density, green space ratio, road network density, land use mix, 500-meter coverage rate of public transportation stations, and density of supporting public service facilities.

[0010] Furthermore, in the benchmark construction unit, the mathematical expression of the spatially coupled dynamic energy consumption benchmark algorithm is: in, For the first Each city grid The dynamic energy consumption baseline value at any given time. For the first Each city grid Reliable fusion of multi-source features at any given time. This represents the total number of dimensions for both temporal and attribute features. For the first Each time series and attribute feature has a preset weight, and all weights sum to 1. For the first Each grid in The first moment Standardized values ​​of time series and attribute features, This is the weight vector of spatial morphology indicators. For the first Each grid in Standardized vector of spatial morphological indicators at time. For dynamic rolling calibration factor, This is the constraint coefficient for government administration and control.

[0011] Furthermore, in the factor identification unit, the mathematical expression for the space-energy consumption feedback-driven factor quantization algorithm is: in, For the first The comprehensive contribution of urban spatial morphology indicators to regional energy consumption. The total number of urban grids, The total time step within the statistical period. For the first Each city grid Reliable fusion of multi-source features at any given time. The actual total energy consumption of grid i at time t. The standardized value of the m-th spatial morphological index at time t and grid i. For the first Each grid in The dynamic energy consumption baseline value at any given time. To standardize the energy consumption deviation rate, Let be the marginal impact coefficient of the m-th spatial morphology index on actual energy consumption. The causal constraint coefficient, Weighting for the feasibility of government policies.

[0012] Furthermore, the specific method by which the regulatory enforcement unit identifies abnormal energy consumption behavior exceeding the baseline and traces the responsible parties is as follows: The regulatory enforcement unit uses the early warning threshold and control red line of the differentiated energy consumption baseline system as the evaluation criteria, and monitors the actual energy consumption data of the region in real time through a time-series anomaly detection algorithm. Combined with meteorological, population, and activity-related data in the full-dimensional fusion feature library, it distinguishes between normal energy consumption fluctuations caused by extreme weather and temporary large-scale events and abnormal energy consumption behavior caused by illegal energy use, equipment malfunctions, and pipeline losses. Then, it combines energy consumption driving factors to match the characteristic patterns of abnormal energy consumption, clarifies the specific root cause of abnormal energy consumption, and through the energy-consuming entities and operation and maintenance related information associated in the full-dimensional fusion feature library, it corresponds to specific energy-consuming enterprises, property management companies, operation and maintenance units, and corresponding industry regulatory authorities to complete the tracing of responsible parties.

[0013] Furthermore, the specific method by which the linkage optimization unit iteratively updates the algorithm parameters of each unit using regulatory operation data as incremental samples is as follows: the actual regional energy consumption operation data, rectification effect data, and energy-saving renovation achievements output by the regulatory execution unit are used as incremental training samples, and the algorithm models of the data fusion unit, benchmark construction unit, and factor identification unit are automatically input every month to complete the incremental training and optimization of the algorithm parameters; the specific method by which the linkage optimization unit synchronizes the planning and control requirements to the regulatory execution unit is as follows: the regional energy consumption control red line and energy consumption benchmark requirements determined in the planning approval process are automatically and synchronously updated to the early warning threshold and control standard library of the regulatory execution unit.

[0014] On the other hand, based on the comprehensive energy consumption analysis method for urban areas based on multi-source data fusion, the specific steps of this method are as follows: S1, Data Fusion: Heterogeneous data is collected across departments through the data fusion unit, and joint modeling is performed based on a privacy computing framework that couples horizontal and vertical federations to complete spatiotemporal alignment and feature association, generating a full-dimensional fusion feature library; S2, Benchmark Construction: Based on the full-dimensional fusion feature library, the benchmark construction unit divides the regional functional types, extracts temporal, attribute, and spatial morphological features, and uses a spatially coupled dynamic energy consumption benchmark algorithm to construct energy consumption benchmark prediction models for different regional types. The model parameters are iteratively updated through a rolling time window to generate a differentiated energy consumption benchmark system; S3, Factor Identification: Based on the full-dimensional fusion feature library and the differentiated energy consumption benchmark system, the factor identification unit completes energy consumption data deviation calibration and adopts... The algorithm for quantifying the energy consumption impact weight of spatial morphology indicators is used to calculate the energy consumption driving factors, screen energy consumption driving factors, and generate a spatial morphology-energy consumption feature correlation model and driving factor library; S4, Regulatory Early Warning: Based on the full-dimensional integrated feature library, differentiated energy consumption baseline system, and energy consumption driving factors, the regulatory execution unit identifies abnormal energy consumption behavior and traces the responsible entities, generates graded regulatory early warning signals according to level and pushes them to the corresponding entities; S5, Linked Optimization: Based on the spatial morphology-energy consumption feature correlation model and driving factor library, the linked optimization unit builds a planning energy consumption control auxiliary decision engine embedded in the government approval process. At the same time, the regulatory operation data is used as an incremental sample to iteratively update the algorithm parameters of each unit every month, and the energy consumption control requirements determined by the planning approval are synchronized to the regulatory execution unit to achieve closed-loop control of energy consumption throughout the entire process.

[0015] On the other hand, a computer-readable storage medium stores program code that can be called by a processor to execute an urban area integrated energy consumption analysis system based on multi-source data fusion.

[0016] Compared with existing technologies, this urban area comprehensive energy consumption analysis system, method, and storage medium based on multi-source data fusion has the following beneficial effects: First, by building a cross-departmental heterogeneous data acquisition system and relying on a coupled privacy computing framework to carry out joint modeling, this invention achieves secure fusion and spatiotemporal alignment of multi-source data, constructing a fusion feature library covering multi-dimensional features. This approach not only ensures the privacy and security of the original data from each department and the data ownership, but also effectively breaks down the data silo problem, achieving comprehensive integration and standardized association of various features such as spatial, temporal, and attribute features. This ensures the integrity and accuracy of the data source for energy consumption analysis. At the same time, by using a standardized geographic grid as the spatial carrier and fine-grained timestamps as the time dimension, it completes the deep binding of features with spatiotemporal dimensions, providing a unified and traceable data analysis foundation for subsequent energy consumption benchmark construction and factor identification. This provides solid underlying data support for urban area energy consumption analysis, improving the reliability of analysis results from the source.

[0017] Second, this invention constructs a full-process urban area energy consumption analysis and control system, from the dynamic construction of differentiated energy consumption benchmarks to the precise quantitative identification of energy consumption driving factors, and then to hierarchical and classified regulatory early warning and accountability, ultimately achieving the linkage optimization of planning approval and energy consumption supervision. Relying on dynamically updated benchmark models and incremental training strategies, the energy consumption benchmarks can adapt to various dynamic changes in urban development in real time, ensuring the scientific nature and adaptability of benchmark evaluation. At the same time, the energy consumption control auxiliary decision engine is embedded in the government approval process, realizing the synchronous implementation of energy consumption control requirements from the planning stage to the supervision stage. Combined with the continuous iterative optimization of algorithm parameters, a closed-loop system for energy consumption control is formed, which greatly improves the refinement and intelligence level of urban energy consumption control and promotes the transformation of urban energy consumption management from passive monitoring to proactive prediction.

[0018] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 This is a framework diagram of an urban area integrated energy consumption analysis system based on multi-source data fusion;

[0021] Figure 2 A flowchart of a method for comprehensive energy consumption analysis in urban areas based on multi-source data fusion;

[0022] Figure 3 This is a schematic diagram of data transmission between modules of an urban area integrated energy consumption analysis system based on multi-source data fusion. Detailed Implementation

[0023] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0024] Example 1:

[0025] The urban area comprehensive energy consumption analysis system based on multi-source data fusion of the present invention, such as... Figure 1As shown, it includes a data fusion unit, a benchmark construction unit, a factor identification unit, a regulatory execution unit, and a linkage optimization unit. The data fusion unit, benchmark construction unit, factor identification unit, and regulatory execution unit are sequentially connected by signals. The linkage optimization unit is unidirectionally connected to the regulatory execution unit. At the same time, the linkage optimization unit is also bidirectionally connected to the data fusion unit, benchmark construction unit, and factor identification unit. The units work together to realize the intelligent collection, analysis, monitoring, and planning optimization of urban area comprehensive energy consumption, providing full-process quantitative support and practical guidance for urban energy consumption dual control and energy conservation and consumption reduction.

[0026] Data Fusion Unit: The data fusion unit is used to collect heterogeneous data across departments. It performs joint modeling through a privacy computing framework that couples horizontal and vertical federation, completes spatiotemporal alignment and feature association, and generates a full-dimensional fusion feature library. This provides a full-dimensional and standardized data source for data analysis and model building in subsequent units.

[0027] The heterogeneous data collected by this unit covers multiple departments and types of data related to urban energy management. Specifically, it includes regional industrial project approval data, energy consumption dual control indicator management data, and energy consumption registration data of large-scale enterprises from the Development and Reform Commission; building completion acceptance data, green building rating data, public building energy consumption statistics data, and municipal building operation and maintenance data from the Housing and Urban-Rural Development Commission; energy consumption data of urban lighting, sanitation facilities, and other municipal public facilities from the Urban Management Commission; public transportation operation data, urban road network traffic data, and transportation hub operation and maintenance data from the Transportation Commission; sub-metering time series data, detailed water and gas usage data for individual households, and pipeline operation loss data from power supply, water supply, and gas companies; urban GIS geospatial data and control detailed planning plot indicator data from the Natural Resources Commission; regional socio-economic data from the Statistics Commission; hourly meteorological and environmental data from the Meteorological Commission; dynamic population flow perception data collected based on mobile phone signaling; and traffic flow data collected from traffic checkpoints and smart intersections.

[0028] The privacy-preserving computing framework, which couples horizontal and vertical federation, is deployed in a secure isolation zone of the government cloud. Each data provider independently deploys a local computing node, while all raw data is stored on each department's own internal government network. Each local node performs local data standardization preprocessing and local model computation. Intermediate parameters for model training are transmitted between nodes via a nationally encrypted channel, achieving secure data fusion that is "usable but not visible." Horizontal federated learning divides the entire administrative region into functional categories such as residential streets, industrial parks, core business districts, transportation hubs, cultural and tourism areas, and municipal public areas. Local computing nodes in the same functional area are included in the same federated learning cluster to complete joint modeling. Vertical federated learning uses an encrypted sample alignment algorithm to match cross-departmental heterogeneous data sample IDs within the same administrative region and for the same energy user. After matching, each department's local nodes output the encrypted feature gradients of the corresponding samples to complete joint training.

[0029] The comprehensive feature library is constructed using a 100m x 100m standard geographic grid as the spatial carrier and hourly timestamps as the time dimension, integrating three major categories of features: spatial, temporal, and attribute-based. Spatial features include unique geographic codes for the grid, spatial coordinates, land use, plot ratio, building density, green space ratio, road network density, land use mix, 500-meter coverage of public transportation stops, density of public service facilities, GIS geospatial topology, and indicators of controlled detailed planning plots. Temporal features include hourly individual metered electricity data, hourly individual household water and gas consumption data, hourly pipeline operation loss data, hourly population flow data, hourly traffic flow data, hourly meteorological and environmental data, hourly public transportation operation data, daily regional energy consumption summary data, and monthly energy consumption statistics. Attribute-based features include regional functional type, building construction year, building floor structure, green building star rating, type of energy-consuming equipment, type of leading industry, number of large-scale enterprises, energy consumption per unit of output value, regional socio-economic data, government approval-related data, and basic information of energy-consuming entities. All features have undergone spatiotemporal alignment and standardization, and are associated and bound by grid and timestamp to form a complete feature set for each grid and each time node, ultimately constituting a full-dimensional fused feature library, such as... Figure 3 As shown.

[0030] Benchmark Building Unit: The benchmark building unit is used to classify regional functional types based on a full-dimensional fusion feature library, extract temporal, attribute and spatial morphological features, and construct regional type energy consumption benchmark prediction models using a spatially coupled dynamic energy consumption benchmark algorithm. The model parameters are updated through a rolling time window to generate a differentiated energy consumption benchmark system adapted to different functional areas, providing a unified quantitative evaluation standard for energy consumption anomaly identification and supervision.

[0031] This unit adopts a combination of primary classification and secondary subdivision to divide the regional functional types. The standard grid of the entire area and the administrative region are divided into primary types: residential area, commercial and business area, industrial park, municipal public area, transportation hub area, and cultural and tourism leisure area. Each primary type is further subdivided into secondary types. Among them, industrial parks are subdivided into high-tech industrial parks, traditional manufacturing industrial parks, and logistics and warehousing parks; residential areas are subdivided into old residential areas, newly built commercial housing areas, and affordable housing areas; commercial and business areas are subdivided into core business districts, regional commercial centers, and community commercial clusters; and transportation hub areas are subdivided into rail transit hubs, highway passenger transport hubs, and port and shipping hubs.

[0032] The temporal, attribute, and spatial morphological features extracted from the full-dimensional integrated feature library each have a clear direction. The temporal and attribute features include the region's statutory planning function positioning, building construction year, floor structure, green building star rating, energy-consuming equipment type, average daily resident population, peak passenger flow, diurnal population difference, dominant industry type, number of large-scale enterprises, energy consumption per unit output value, hourly temperature, air humidity, wind speed, precipitation, and sunshine duration. The spatial morphological features include grid content ratio, building density, green space ratio, road network density, land use mix, 500-meter coverage rate of public transportation stations, and density of supporting public service facilities.

[0033] An energy consumption benchmark prediction model is constructed using a spatially coupled dynamic energy consumption benchmark algorithm. The mathematical expression of the spatially coupled dynamic energy consumption benchmark algorithm is as follows: in, For the first Each city grid The dynamic energy consumption baseline value at any given time. For the first Each city grid Reliable fusion of multi-source features at any given time. This represents the total number of dimensions for both temporal and attribute features. For the first Each time series and attribute feature has a preset weight, and all weights sum to 1. For the first Each grid in The first moment Standardized values ​​of time series and attribute features, This is the weight vector of spatial morphology indicators. For the first Each grid in Standardized vector of spatial morphological indicators at time. For dynamic rolling calibration factor, The algorithm serves as a constraint coefficient for government administration and control. The specific steps are as follows: The spatially coupled dynamic energy consumption benchmark algorithm combines multi-dimensional features extracted from a comprehensive feature library, including temporal, attribute, and spatial morphology features, to conduct targeted modeling of energy consumption characteristics in different sub-functional areas of the city, fully adapting to the functional attributes and energy consumption patterns of each area. During the modeling process, the reliability of multi-source feature fusion for each city grid at the corresponding time node is first considered. Weights are then rationally allocated and standardized for various temporal and attribute features. Simultaneously, corresponding weight vectors are matched for spatial morphology indicators to accurately reflect the impact of spatial morphology on regional energy consumption. The algorithm also introduces a dynamic rolling calibration factor to adapt to the impact of dynamic changes such as seasonal climate fluctuations, population migration, and industrial restructuring on energy consumption, and combines this with the government administration and control constraint coefficient to align with regional energy consumption dual control and annual energy-saving targets and other government management requirements. The model uses a 100m x 100m standard geographic grid as the basic spatial unit and an hour as the smallest time granularity. It calculates the appropriate energy consumption benchmark value for each grid and each time node. The initial training of the model is completed based on the historical fusion data of each subdivided functional area over the past 3 years. Subsequently, the parameters can be iteratively updated through an incremental training strategy with a rolling time window, so that the model always fits the actual energy consumption pattern of the city and ensures the accuracy and dynamic adaptability of the energy consumption benchmark prediction.

[0034] The model parameters are dynamically updated using a rolling time window. An independent rolling time window is configured for each subdivided functional area, with the window length dynamically set to daily, weekly, monthly, or quarterly based on the area's functional attributes. Initial model training is performed using the historical fusion data from the past three years as the training set, forming the initial parameter set. After each rolling time window period, the actual operational data generated in the current period, synchronously updated time-series and attribute feature data, and dynamic spatial morphology change data are automatically added to the training set. Iterative updates of the model parameters are completed based on an incremental training strategy, eliminating the need for retraining on the entire historical data set. The updated parameter set can map in real time the impact of external factors such as changes in regional functional attributes, seasonal climate fluctuations, population migration, industrial restructuring, and the improvement of supporting facilities on regional energy consumption, maintaining dynamic consistency between model parameters and actual urban energy consumption patterns.

[0035] Based on the updated model, a differentiated energy consumption baseline system is generated. This system uses a 100m×100m standard geographic grid as the spatial unit and the hour as the smallest time granularity. For each type of subdivided functional area, it outputs standardized energy consumption baseline content with multiple time dimensions and multiple types of thresholds. The differentiated energy consumption baseline system specifically includes hourly, daily, monthly, and annual energy consumption baselines. The hourly baseline matches peak and off-peak energy consumption characteristics for electricity, water, and gas; the daily baseline adapts to summaries of daily energy consumption patterns; the monthly baseline corresponds to the cyclical fluctuations in energy consumption caused by seasonal climate changes; and the annual baseline directly aligns with regional energy consumption control indicators and annual energy-saving targets. Simultaneously, the system includes a first-level warning threshold, a second-level warning threshold, and a regional energy consumption control red line. The first-level warning threshold corresponds to an actual energy consumption deviation of 10%-20% from the baseline; the second-level warning threshold corresponds to an actual energy consumption deviation of 20%-50% from the baseline; and the regional energy consumption control red line corresponds to a critical value where actual energy consumption deviates from the baseline by more than 50%.

[0036] Factor identification unit: The factor identification unit is used to calibrate energy consumption data deviation based on a full-dimensional fusion feature library and a differentiated energy consumption baseline system. It uses a space-energy consumption mutual feedback driving factor quantification algorithm to calculate the influence weight of spatial morphology indicators on energy consumption, screen energy consumption driving factors, and generate a spatial morphology-energy consumption feature correlation model and driving factor library, providing clear directional basis for energy consumption management and planning optimization.

[0037] This unit uses a differentiated energy consumption baseline system as a benchmark to standardize the actual energy consumption data of each region, eliminating energy consumption deviations caused by differences in regional functional types, extreme weather, temporary large-scale events, statutory holidays, and temporary population fluctuations. This results in a standardized energy consumption intensity index that can be compared horizontally across regions and cycles, eliminating interference factors for the subsequent calculation of the influence weight of spatial morphology indicators.

[0038] The spatial-energy consumption feedback driving factor quantification algorithm is used to calculate the comprehensive contribution of each spatial morphology index to regional energy consumption. The mathematical expression of the spatial-energy consumption feedback driving factor quantification algorithm is as follows: in, For the first The comprehensive contribution of urban spatial morphology indicators to regional energy consumption. The total number of urban grids, The total time step within the statistical period. For the first Each city grid Reliable fusion of multi-source features at any given time. The actual total energy consumption of grid i at time t. The standardized value of the m-th spatial morphological index at time t and grid i. For the first Each grid in The dynamic energy consumption baseline value at any given time. To standardize the energy consumption deviation rate, Let be the marginal impact coefficient of the m-th spatial morphology index on actual energy consumption. The causal constraint coefficient, The weighting of government implementation is as follows: The spatial-energy consumption feedback driving factor quantification algorithm uses a full-dimensional fusion feature library as its data foundation, combined with a differentiated energy consumption baseline system, to complete the deviation calibration of the actual energy consumption data of the region. It eliminates energy consumption deviations caused by irrelevant factors such as differences in regional functional types, extreme weather, temporary large-scale events, and statutory holidays, resulting in standardized energy consumption data that can be compared horizontally across regions and cycles. In the comprehensive contribution calculation process, it covers all geographic grids in the city, statistically analyzes energy consumption-related data throughout the entire cycle, comprehensively considers the reliability of multi-source feature fusion at the corresponding time node for each grid, and the standardized deviation rate of each grid's actual energy consumption relative to the benchmark energy consumption. Simultaneously, it accurately calculates the marginal impact coefficient of each spatial morphology indicator on actual energy consumption, intuitively reflecting the driving effect of indicator changes on energy consumption. The algorithm also introduces a causal constraint coefficient to eliminate interference from indicators that have no actual causal relationship with energy consumption. It combines the feasibility weight of government affairs to consider the operability of each indicator in actual urban energy consumption control and planning adjustment. By comprehensively calculating the multi-dimensional data of the entire grid and the entire statistical period, it finally obtains the comprehensive contribution value of each spatial form indicator to regional energy consumption. This quantifies the actual impact of each spatial form indicator on energy consumption and provides a scientific quantitative basis for subsequent screening of energy consumption impact indicators.

[0039] Based on the comprehensive contribution values ​​of various spatial morphology indicators, we sort and filter them to identify energy consumption drivers that have a significant impact on regional energy consumption. At the same time, based on the correlation between each indicator and energy consumption data, we generate a spatial morphology-energy consumption characteristic correlation model. We store the energy consumption drivers and the correlation model in a unified manner to build a driving factor library, which provides data and model support for subsequent energy consumption management and planning optimization.

[0040] Regulatory Execution Unit: The regulatory execution unit is used to identify abnormal energy consumption behaviors that exceed the baseline and trace the responsible parties by using a multi-dimensional integrated feature library as the data foundation, a differentiated energy consumption baseline system as the evaluation basis, and energy consumption driving factors as the control guidelines. It generates hierarchical regulatory early warning signals to achieve real-time monitoring and precise supervision of energy consumption in urban areas.

[0041] This unit uses the early warning thresholds and control red lines of the differentiated energy consumption baseline system as evaluation criteria. It monitors the actual energy consumption data of the region in real time through a time-series anomaly detection algorithm. Combined with meteorological, population, and activity-related data in the full-dimensional fusion feature library, it distinguishes between normal energy consumption fluctuations caused by extreme weather and temporary large-scale events and abnormal energy consumption behaviors caused by illegal energy use, equipment malfunctions, and pipeline losses. Then, by combining energy consumption driving factors to match the characteristic patterns of abnormal energy consumption, it clarifies the specific root causes of abnormal energy consumption. By connecting the energy-consuming entities and operation and maintenance information in the full-dimensional fusion feature library, it identifies specific energy-consuming enterprises, property management companies, operation and maintenance units, and corresponding industry regulatory authorities, thus completing the accurate tracing of the responsible parties.

[0042] Based on the magnitude, duration, and scope of abnormal energy consumption exceeding the baseline, regulatory early warning signals are categorized into three levels: general warning, important warning, and emergency warning. Different levels of warning signals are pushed to the corresponding responsible units and regulatory departments: general warnings are pushed to community grid workers and corresponding responsible entities; important warnings are pushed to street offices and district-level industry regulatory departments; and emergency warnings are pushed to municipal-level regulatory departments and government law enforcement units. Each warning signal includes the location of the anomaly, the responsible entity, the root cause of the anomaly, and control recommendations, achieving closed-loop regulatory guidance for abnormal energy consumption issues.

[0043] Linked Optimization Unit: The linked optimization unit is used to build a planning energy consumption control auxiliary decision engine embedded in the approval process, supported by the spatial form-energy consumption characteristic correlation model and driving factor library. It uses regulatory operation data as incremental samples to iteratively update the algorithm parameters of each unit, and synchronizes the planning control requirements to the regulatory execution unit to realize two-way linkage and dynamic optimization between urban energy consumption supervision and planning approval.

[0044] The planning energy consumption control auxiliary decision-making engine built in this unit is deeply embedded in the online government approval system of the natural resources and planning departments, directly connecting with the approval processes of detailed control plans, urban renewal projects, and state-owned construction land transfers. The engine's input indicators include plot ratio, building density, land use layout, road network planning, public transportation infrastructure, and public service facility layout. Its outputs include energy intensity calculation results, energy-saving potential calculation results, and planning optimization suggestions for the corresponding area, providing a scientific basis for planning approval in terms of energy consumption and controlling regional energy consumption levels from the source.

[0045] Using regulatory operation data as incremental samples, the algorithm parameters of each unit are iteratively updated. The actual energy consumption operation data, rectification effect data, and energy-saving renovation achievements output by the regulatory execution unit are used as incremental training samples. The algorithm models of the data fusion unit, benchmark construction unit, and factor identification unit are automatically input every month to complete the incremental training and optimization of the algorithm parameters of each unit. This allows each algorithm model to continuously adapt to the dynamic changes in urban energy consumption and improve the accuracy of analysis and supervision.

[0046] Simultaneously, the planning and control requirements are synchronized to the regulatory execution unit. The regional energy consumption control red line and energy consumption benchmark requirements determined in the planning approval process are automatically updated to the early warning threshold and control standard library of the regulatory execution unit. This ensures that the energy consumption requirements of the planning approval are consistent with the evaluation standards of daily energy consumption supervision, forming a closed-loop management of the entire process of planning, supervision and optimization.

[0047] This invention presents a comprehensive urban energy consumption analysis system based on multi-source data fusion. It achieves secure fusion of heterogeneous data from multiple sources across departments through a privacy-preserving computing framework coupled horizontally and vertically, constructing a comprehensive energy consumption feature library. Combined with a spatially coupled dynamic energy consumption benchmark algorithm, it generates a differentiated energy consumption benchmark system, providing a quantitative standard for energy consumption supervision. A spatial-energy consumption feedback-driven factor quantification algorithm accurately identifies energy consumption driving factors, providing clear direction for control and optimization. Real-time and precise energy consumption supervision is achieved through tiered supervision and early warning. A planning-based energy consumption control auxiliary decision engine enables bidirectional linkage between planning and supervision. Simultaneously, incremental training continuously optimizes the algorithm parameters of each unit, ensuring the system always adapts to the actual energy consumption patterns of the city. This system realizes intelligent and refined management of comprehensive urban energy consumption from data collection, benchmark construction, factor identification to supervision execution and planning optimization. It provides a feasible technical solution for cities to implement dual energy consumption control targets and promote energy conservation and consumption reduction, effectively improving the scientific and intelligent level of urban energy management.

[0048] Example 2:

[0049] This embodiment is applied to the dynamic monitoring and anomaly handling of energy consumption in the core commercial and business district of an city. Addressing the characteristics of this area—high population flow, significant energy consumption peak and valley features, high proportion of commercial energy consumption, and the significant impact of temporary activities on energy consumption—it utilizes a comprehensive urban regional energy consumption analysis system based on multi-source data fusion to conduct full-process energy consumption analysis and control. The specific implementation process is as follows: Figure 2 The following is stated:

[0050] Data Fusion: Through the system's data fusion unit, heterogeneous data is collected across departments, including: project approval data from the Development and Reform Commission; completion and acceptance data, green building rating data, and public building energy consumption statistics from the Housing and Urban-Rural Development Commission; energy consumption data for urban lighting and sanitation facilities from the Urban Management Commission; public transportation operation data and urban road network traffic data from the Transportation Commission; electricity metering time series data, detailed water and gas usage data, and pipeline operation loss data from power, water, and gas companies; GIS geospatial data and control detailed planning plot index data from the Natural Resources Commission; hourly meteorological environmental data from the Meteorological Commission; dynamic perception data of customer flow from mobile phone signaling; and traffic flow data from traffic checkpoints and smart intersections. This comprehensive integration of multi-dimensional data related to commercial operations, municipal facilities, traffic flow, and energy supply in the commercial district allows energy consumption analysis to cover various scenarios and influencing factors related to energy use in the commercial district. This unit relies on a privacy-preserving computing framework that couples horizontal and vertical federation to conduct joint modeling. The framework is deployed in a secure isolation zone of the government cloud. After local computing nodes of each data provider complete local data processing, intermediate parameters for model training are transmitted through a national-level encrypted channel. This achieves effective fusion computing of cross-departmental data while ensuring the privacy of data from each department is not leaked. Horizontal federated learning incorporates the core business district into a commercial and business area-type federated learning cluster, while vertical federated learning completes encrypted matching of cross-departmental heterogeneous data sample IDs within the business district. This enables effective linkage analysis between data from similar commercial areas and cross-departmental data within the same business district. Ultimately, it completes the spatiotemporal alignment and feature association of the entire dataset. Using a 100m × 100m standard geographic grid as the spatial carrier and hourly timestamps as the time dimension, it integrates the spatial, temporal, and attribute features of the business district to generate a full-dimensional fusion feature library exclusive to the core commercial and business district. It focuses on improving the temporal features of the business district, such as hourly customer flow, traffic flow, and energy consumption data, so that the feature library can accurately match the high-frequency changes in customer flow and energy consumption peaks and valleys in the business district. This provides comprehensive, accurate, and dynamic data support for subsequent refined energy consumption analysis of the business district.

[0051] Benchmark Construction: Through the system's benchmark construction unit, based on a full-dimensional fusion feature library, the functional type of this area is divided into core business district sub-types under the commercial and business district. Benchmark modeling is carried out in accordance with the commercial energy consumption characteristics and customer flow changes of the core business district, extracting the temporal and attribute features and spatial morphological features of the business district. The temporal and attribute features include the legal planning functional positioning of the business district, the construction year of commercial buildings, floor structure, green building star rating, energy-consuming equipment type, hourly customer flow data, diurnal population difference, hourly temperature, air humidity, sunshine duration, etc. The spatial morphological features include the area ratio of the business district grid, building density, land use mix, 500-meter coverage of public transportation stations, density of public service facilities, etc. Comprehensive extraction of various core features affecting the energy consumption of the core business district, such as customer flow, buildings, meteorology, and spatial morphology, allows the construction of the benchmark model to accurately match the actual energy consumption scenario of the business district. This unit uses a spatially coupled dynamic energy consumption benchmark algorithm to construct an energy consumption benchmark prediction model specifically for the core business district. The mathematical expression of the spatially coupled dynamic energy consumption benchmark algorithm is: in, For the first Each city grid The dynamic energy consumption baseline value at any given time. For the first Each city grid Reliable fusion of multi-source features at any given time. This represents the total number of dimensions for both temporal and attribute features. For the first Each time series and attribute feature has a preset weight, and all weights sum to 1. For the first Each grid in The first moment Standardized values ​​of time series and attribute features, This is the weight vector of spatial morphology indicators. For the first Each grid in Standardized vector of spatial morphological indicators at time. For dynamic rolling calibration factor, To constrain government control, and to ensure the baseline model accurately couples the comprehensive impact of spatial morphology and various temporal attributes on the energy consumption of business districts, a daily rolling time window is configured for this sub-region, taking into account the significant peak-valley fluctuations in energy consumption within the core business district. Initial model training is completed using historical fusion data from the past three years as the training set, ensuring sufficient historical energy consumption data to support the model's initial parameters. After each daily rolling time window period, the actual operational data and feature update data of the current business district are automatically added to the training set. Incremental training iteratively updates the model parameters without requiring retraining on the entire historical data set, thus improving efficiency. While improving parameter update efficiency, the model parameters are updated in real time to match changes in customer flow and seasonal fluctuations in energy consumption patterns within the business district. This results in the generation of a differentiated energy consumption baseline system adapted to the core business district. The system uses a 100m x 100m standard geographic grid as the spatial unit and the hour as the smallest time granularity. It focuses on optimizing the hourly energy consumption baseline to match the peak and off-peak energy consumption characteristics of electricity, water, and gas in the business district. At the same time, it sets a first-level warning threshold, a second-level warning threshold, and a regional energy consumption control red line to meet the business district's needs for refined and high-frequency energy consumption supervision, providing a precise quantitative standard for energy consumption assessment of each grid and each hour within the business district.

[0052] Factor Identification: The system's factor identification unit, based on a comprehensive feature library and a differentiated energy consumption baseline system, calibrates energy consumption data deviations. It standardizes actual energy consumption data for the business district, eliminating deviations caused by extreme weather, large-scale promotional activities, statutory holidays, and temporary population fluctuations. This yields standardized energy intensity indicators that can be compared across regions and time periods, eliminating interference from various temporary external factors and making the comparative analysis of business district energy consumption data more objective and reliable. This unit employs a space-energy consumption feedback-driven factor quantification algorithm to calculate the comprehensive contribution of spatial morphology indicators such as 500-meter coverage of public transportation stations, land use mix, road network density, and green space ratio to energy consumption. The mathematical expression for the space-energy consumption feedback-driven factor quantification algorithm is: in, For the first The comprehensive contribution of urban spatial morphology indicators to regional energy consumption. The total number of urban grids, The total time step within the statistical period. For the first Each city grid Reliable fusion of multi-source features at any given time. For the first Each grid in Actual energy consumption monitoring value at any given time For the first Each grid in The dynamic energy consumption baseline value at any given time. To standardize the energy consumption deviation rate, Let be the marginal impact coefficient of the m-th spatial morphology index on actual energy consumption. The causal constraint coefficient, To enhance the feasibility of government policies, this study precisely quantifies the impact of various spatial form indicators on the formation of energy consumption in business districts. It identifies core driving factors that significantly influence energy consumption in core business districts, focuses on key spatial factors for energy consumption control in business districts, and generates a spatial form-energy consumption characteristic correlation model and driving factor library adapted to the core business district. This establishes a clear logical connection between the optimization and adjustment of business district spatial form and energy consumption control, providing clear guidance for targeted energy consumption control and spatial planning optimization in business districts.

[0053] Regulatory early warning: Through the system's regulatory execution unit, based on a full-dimensional integrated feature library, a differentiated energy consumption baseline system, and energy consumption driving factors, real-time monitoring and identification of energy consumption anomalies in business districts are carried out, providing comprehensive data support, accurate evaluation standards, and clear control directions for the identification of energy consumption anomalies in business districts. This unit monitors hourly actual energy consumption data of the business district using a time-series anomaly detection algorithm, achieving high-frequency, refined, real-time dynamic monitoring of energy consumption data. Combined with meteorological, passenger flow, and commercial activity-related data from a comprehensive feature library, it accurately distinguishes between normal energy consumption fluctuations caused by promotional activities and peak holiday passenger flow, and abnormal energy consumption behaviors caused by commercial building equipment malfunctions, excessive energy consumption of lighting systems, gas pipeline leaks, and water supply pipeline damage. This accurately identifies the true causes of energy consumption fluctuations in the business district, avoiding false alarms caused by various temporary commercial activities, and improving the accuracy of energy consumption anomaly identification. Furthermore, by combining energy consumption driving factors with the characteristic patterns of abnormal energy consumption, it clarifies the specific root causes of abnormal energy consumption, allowing for precise identification of the core issues in handling energy consumption anomalies in the business district. Through comprehensive integration of energy-consuming entities and operation and maintenance information from the feature library, the responsibility for abnormal energy consumption is traced back to specific commercial operators, property management units, municipal facility maintenance units, and district-level commerce, housing and construction, and urban management departments, completing the precise positioning of responsible entities and ensuring that the responsibility for energy consumption management in the business district is accurately assigned to specific units and departments. Subsequently, based on the magnitude, duration, and scope of the abnormal energy consumption exceeding the baseline, three warning levels were established: general, important, and urgent. General warnings were pushed to community grid workers and corresponding commercial operators and maintenance units; important warnings were pushed to street offices and district-level commerce and housing departments; and urgent warnings were pushed to municipal-level commerce and housing departments and government law enforcement units. This ensures that warning instructions can accurately reach the corresponding management level according to the severity of the anomaly, achieving a hierarchical response and handling of abnormal energy consumption in the business district. Each warning signal includes the location of the abnormal grid, the responsible entity, the specific root cause of the anomaly, and actionable control suggestions, enabling the receiving entities to quickly grasp the anomaly and carry out precise handling work, thereby improving the efficiency of handling abnormal energy consumption in the business district.

[0054] Linked Optimization: Through the system's linked optimization units, supported by a spatial form-energy consumption characteristic correlation model and a driving factor library, a planning energy consumption control auxiliary decision-making engine is built. This engine is deeply embedded in the online government approval system of the natural resources and planning departments, connecting with the approval process of urban renewal projects within core business districts. This allows energy consumption control requirements to be brought forward to the approval stage of business district renovation and construction planning, controlling energy consumption levels from the source during the renovation and upgrading process, and improving the energy-saving scientific nature of business district spatial planning. When the business district undertakes approval work for the renovation of old commercial buildings and the construction of new commercial supporting facilities, the engine takes the plot ratio, building density, public service facility layout, and public transportation support optimization indicators in the renovation and construction plan as inputs, and outputs the energy intensity calculation results, energy-saving potential calculation results, and targeted planning optimization suggestions for the corresponding area. This ensures that the planning scheme for the renovation and upgrading of the business district fully integrates energy consumption control requirements, achieving effective energy consumption control while optimizing spatial form. Meanwhile, this unit uses the actual energy consumption operation data, rectification results data of illegal energy use, and energy-saving renovation achievements of the business district output by the regulatory execution unit as incremental training samples. It automatically inputs the algorithm models of the data fusion unit, benchmark construction unit, and factor identification unit every month to complete the incremental training and optimization of the algorithm parameters of each unit. This ensures that the algorithm models of each unit can continuously adapt to the dynamic changes in the energy consumption patterns of the business district, and ensures that the entire system's ability to analyze and control the energy consumption of the business district is always adapted to the actual changes in the energy consumption of the business district. This unit will also automatically and synchronously update the core business district energy consumption control red lines and energy consumption benchmark requirements determined in the urban renewal project approval process to the early warning thresholds and control standard library of the regulatory execution unit. This allows the energy consumption control requirements determined in the planning approval process to be directly implemented in the daily energy consumption supervision work of the business district, realizing the closed-loop control of the core business district energy consumption from planning approval, daily monitoring, abnormal early warning to rectification and optimization, and adapting to the dynamic changes in the energy consumption characteristics of the core business district.

[0055] In summary, this embodiment, taking into account the characteristics of large fluctuations in passenger flow and significant peak and valley energy consumption in core commercial and business districts, relies on this energy consumption analysis system to carry out refined energy consumption supervision and coordinated management. Each unit completes the fusion of multi-source data according to the process and focuses on improving the time-series characteristics, constructs an energy consumption benchmark model adapted to the business district and configures a daily rolling window, eliminates temporary factor deviations and selects core energy consumption driving factors, accurately identifies abnormal energy consumption and completes responsibility tracing and hierarchical early warning. At the same time, the planning energy consumption management engine is embedded in the urban renewal approval process, and the algorithm parameters are iterated monthly with the supervision data to achieve effective linkage between pre-planning and daily supervision, forming a closed-loop management system that adapts to the dynamic energy consumption characteristics and refined management needs of the business district.

[0056] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A comprehensive energy consumption analysis system for urban areas based on multi-source data fusion, characterized in that, The system includes: Data fusion unit: used to collect heterogeneous data across departments, and to perform joint modeling through a privacy computing framework that couples horizontal and vertical federation, to complete spatiotemporal alignment and feature association, and generate a full-dimensional fusion feature library; The benchmark construction unit is used to divide the functional types of regions based on the full-dimensional fusion feature library, extract temporal, attribute and spatial morphological features, construct regional type energy consumption benchmark prediction models using spatial coupling dynamic energy consumption benchmark algorithm, update model parameters through rolling time windows, and generate a differentiated energy consumption benchmark system adapted to different functional regions. Factor identification unit: used to calibrate energy consumption data deviation based on the full-dimensional fusion feature library and the differentiated energy consumption baseline system, use the space-energy consumption mutual feedback driving factor quantification algorithm to calculate the influence weight of spatial morphology indicators on energy consumption, screen energy consumption driving factors, and generate spatial morphology-energy consumption feature association model and driving factor library; The regulatory enforcement unit is used to identify abnormal energy consumption behaviors that exceed the baseline and trace the responsible parties, and generate hierarchical regulatory early warning signals, based on the full-dimensional integrated feature library, the differentiated energy consumption baseline system, and the energy consumption driving factors. Linkage Optimization Unit: This unit, supported by the spatial morphology-energy consumption characteristic correlation model and driving factor library, builds a planning energy consumption control auxiliary decision-making engine embedded in the approval process. It iteratively updates the algorithm parameters of each unit using regulatory operation data as incremental samples, synchronizing planning control requirements to the regulatory execution unit. In the aforementioned benchmark construction unit, the mathematical expression of the spatially coupled dynamic energy consumption benchmark algorithm is: in, For the first Each city grid Dynamic energy consumption baseline value at any given time. For the first Each city grid Reliable fusion of multi-source features at any given time. This represents the total number of dimensions for both temporal and attribute features. For the first Each time series and attribute feature has a preset weight, and all weights sum to 1. For the first Each grid in The first moment Standardized values ​​of time series and attribute features, This is a weight vector for spatial morphology indicators. For the first Each grid in Standardized vector of spatial morphological indicators at time. For dynamic rolling calibration factor, This is the government control constraint coefficient. In the factor identification unit, the mathematical expression of the space-energy consumption feedback-driven factor quantization algorithm is: in, For the first The comprehensive contribution of urban spatial morphology indicators to regional energy consumption. The total number of urban grids, The total time step within the statistical period. For the first Each city grid Reliable fusion of multi-source features at any given time. The actual total energy consumption of grid i at time t. The standardized value of the m-th spatial morphological index at time t and grid i. For the first Each grid in Dynamic energy consumption baseline value at any given time. To standardize the energy consumption deviation rate, Let be the marginal impact coefficient of the m-th spatial morphology index on actual energy consumption. The causal constraint coefficient, Weighting for the feasibility of government policies.

2. The urban area comprehensive energy consumption analysis system based on multi-source data fusion according to claim 1, characterized in that, The heterogeneous data collected by the data fusion unit specifically includes: regional industrial project approval data, energy consumption dual control indicator management data, and energy consumption filing data of enterprises above designated size from the Development and Reform Commission; building completion acceptance data, green building rating data, public building energy consumption statistics data, and municipal building operation and maintenance data from the Housing and Urban-Rural Development Commission; urban lighting, sanitation facilities, and municipal public utility operation energy consumption data from the Urban Management Commission; public transportation operation data, urban road network traffic data, and transportation hub operation and maintenance data from the Transportation Commission; sub-metering time series data, detailed water and gas usage data for individual households, and pipeline operation loss data from power supply, water supply, and gas companies; urban GIS geospatial data and control detailed planning plot indicator data from the Natural Resources Commission; regional socio-economic data from the Statistics Commission; hourly meteorological and environmental data from the Meteorological Commission; population flow dynamic perception data collected by mobile phone signaling; and traffic flow data collected by traffic checkpoints and smart intersections.

3. The urban area comprehensive energy consumption analysis system based on multi-source data fusion according to claim 1, characterized in that, In the data fusion unit, the privacy computing framework coupled with horizontal and vertical federation is deployed in the secure isolation zone of the government cloud. The framework independently deploys local computing nodes for each data provider. All raw data is stored in the internal government network of each department. Each local node completes local data standardization preprocessing and local model calculation. Intermediate parameters for model training are transmitted between nodes through a national-level encrypted channel. The specific method of joint modeling by the privacy computing framework coupled with horizontal and vertical federation is as follows: Horizontal federated learning divides the entire administrative region into functional attributes such as residential streets, industrial parks, core business districts, transportation hubs, cultural and tourism areas, and municipal public areas. Local computing nodes of the same type of functional area are included in the same federated learning cluster to complete joint modeling. Vertical federated learning completes cross-departmental heterogeneous data sample ID matching for the same administrative region and the same energy user through an encrypted sample alignment algorithm. After matching, the local nodes of each department output the encrypted feature gradient of the corresponding sample to complete joint training.

4. The urban area comprehensive energy consumption analysis system based on multi-source data fusion according to claim 1, characterized in that, The specific method for dividing the functional types of the benchmark construction unit is as follows: the entire standard grid and administrative regions are divided into primary types: residential living area, commercial and business area, industrial park, municipal public area, transportation hub area, and cultural tourism and leisure area. Each primary type is further subdivided into secondary types. Among them, industrial parks are subdivided into high-tech industrial parks, traditional manufacturing industrial parks, and logistics and warehousing parks; residential living areas are subdivided into old residential areas, newly built commercial housing areas, and affordable housing areas; commercial and business areas are subdivided into core business districts, regional commercial centers, and community commercial clusters; and transportation hub areas are subdivided into rail transit hubs, highway passenger transport hubs, and port and shipping hubs. The temporal, attribute, and spatial morphological features extracted from the benchmark construction unit specifically include: temporal and attribute features such as the region's statutory planning function positioning, building construction year, floor structure, green building star rating, energy-consuming equipment type, average daily resident population, peak passenger flow, diurnal population difference, dominant industry type, number of large-scale enterprises, energy consumption per unit output value, hourly temperature, air humidity, wind speed, precipitation, and sunshine duration; and spatial morphological features such as grid content area ratio, building density, green space ratio, road network density, land use mix, 500-meter coverage rate of public transportation stations, and density of supporting public service facilities.

5. The urban area comprehensive energy consumption analysis system based on multi-source data fusion according to claim 1, characterized in that, The specific method by which the regulatory enforcement unit identifies abnormal energy consumption behavior exceeding the baseline and traces the responsible parties is as follows: The regulatory enforcement unit uses the early warning threshold and control red line of the differentiated energy consumption baseline system as the evaluation criteria. It monitors the actual energy consumption data of the region in real time through a time-series anomaly detection algorithm. Combined with meteorological, population, and activity-related data in the full-dimensional fusion feature library, it distinguishes between normal energy consumption fluctuations caused by extreme weather and temporary large-scale events and abnormal energy consumption behavior caused by illegal energy use, equipment malfunctions, and pipeline losses. Then, it combines energy consumption driving factors to match the characteristic patterns of abnormal energy consumption, clarifies the specific root cause of abnormal energy consumption, and uses the energy-consuming entities and operation and maintenance related information associated in the full-dimensional fusion feature library to correspond to specific energy-consuming enterprises, property management companies, operation and maintenance units, and corresponding industry regulatory authorities to complete the tracing of responsible parties.

6. The urban area comprehensive energy consumption analysis system based on multi-source data fusion according to claim 1, characterized in that, The specific method by which the linkage optimization unit iteratively updates the algorithm parameters of each unit using regulatory operation data as incremental samples is as follows: the actual regional energy consumption operation data, rectification effect data, and energy-saving renovation achievements output by the regulatory execution unit are used as incremental training samples, and the algorithm models of the data fusion unit, benchmark construction unit, and factor identification unit are automatically input every month to complete the incremental training and optimization of the algorithm parameters; the specific method by which the linkage optimization unit synchronizes the planning and control requirements to the regulatory execution unit is as follows: the regional energy consumption control red line and energy consumption benchmark requirements determined in the planning approval process are automatically and synchronously updated to the early warning threshold and control standard library of the regulatory execution unit.

7. A method for comprehensive energy consumption analysis of urban areas based on multi-source data fusion, applicable to the comprehensive energy consumption analysis system for urban areas based on multi-source data fusion as described in any one of claims 1-6, characterized in that, The specific steps of this method are as follows: S1. Data Fusion: Through cross-departmental collection of heterogeneous data by the data fusion unit, joint modeling is carried out based on the privacy computing framework coupled with horizontal and vertical federation to complete spatiotemporal alignment and feature association, and generate a full-dimensional fusion feature library. S2. Benchmark Construction: Based on a full-dimensional fusion feature library, the functional types of regions are divided by benchmark construction units. Temporal, attribute and spatial morphological features are extracted. A spatially coupled dynamic energy consumption benchmark algorithm is used to construct energy consumption benchmark prediction models for different regions. The model parameters are iteratively updated through a rolling time window to generate a differentiated energy consumption benchmark system. S3. Factor Identification: Based on the full-dimensional fusion feature library and differentiated energy consumption baseline system, the factor identification unit completes the energy consumption data deviation calibration, uses the space-energy consumption mutual feedback driving factor quantification algorithm to calculate the energy consumption influence weight of spatial morphology indicators, screens energy consumption driving factors, and generates a spatial morphology-energy consumption feature correlation model and driving factor library. S4. Regulatory Early Warning: Based on the full-dimensional integrated feature library, differentiated energy consumption baseline system, and energy consumption driving factors, the regulatory execution unit identifies abnormal energy consumption behavior and traces the responsible parties, generates graded regulatory early warning signals according to the level, and pushes them to the corresponding entities. S5. Linked Optimization: Through the linked optimization unit, relying on the spatial form-energy consumption characteristic correlation model and driving factor library, a planning energy consumption control auxiliary decision engine embedded in the government approval process is built. At the same time, the regulatory operation data is used as an incremental sample to iteratively update the algorithm parameters of each unit every month, and the energy consumption control requirements determined by the planning approval are synchronized to the regulatory execution unit to achieve closed-loop control of energy consumption throughout the entire process.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code, which can be called by a processor to execute the urban area comprehensive energy consumption analysis method based on multi-source data fusion as described in claim 7.

Images