Highway-based energy consumption monitoring data visualization method and system

Abstract

The application provides a kind of energy consumption monitoring data visualization method and system based on highway, and the energy consumption monitoring data integration body is obtained by integrating original monitoring record;The mapping relationship is established by analyzing the correlation between traffic flow change mode and energy consumption mode in it, and the road section energy consumption benchmark index is generated;The multi-scale deviation degree is calculated by the weighted comparison operation of the energy consumption monitoring record collected in real time and it under different time granularity, and the energy consumption deviation index set is obtained;The energy consumption deviation index set is converted into visual encoding vector containing encoding instruction by establishing the mapping strategy between deviation index and visual performance parameter, and the encoding instruction is generated;Based on the visual encoding vector, the energy consumption state visualization rendering operation is driven on three-dimensional road network model, and the energy consumption distribution characteristics and abnormal fluctuation are presented.The application can intuitively present energy consumption spatial distribution and dynamic change trend, and improve the identification efficiency of energy consumption abnormal interval.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for visualizing energy consumption monitoring data based on highways. Background Technology

[0002] With the development of intelligent transportation technology, the collection, analysis, and graphical display of energy consumption data along highways can provide an intuitive presentation and monitoring management of energy consumption status. However, current visualization methods lack a unified spatiotemporal data correlation framework, making it difficult to effectively integrate multi-source monitoring data. This results in an inaccurate reflection of the dynamic correlation between energy consumption and influencing factors such as traffic flow. Furthermore, traditional visualization methods often use fixed thresholds to divide energy consumption intervals, making it difficult to capture abnormal energy consumption fluctuations at different time scales. They also fail to adequately depict the dynamic changes in energy consumption deviating from the baseline, impacting the efficiency of managers' rapid identification and decision-making in addressing energy consumption anomalies. Summary of the Invention

[0003] In view of this, the present invention provides a method and system for visualizing energy consumption monitoring data based on highways. The technical solution of the embodiments of the present invention is implemented as follows:

[0004] On one hand, embodiments of the present invention provide a method for visualizing energy consumption monitoring data based on highways. The method includes: integrating the original monitoring records collected by various types of monitoring equipment deployed along the highway by establishing a unified time reference system and spatial coordinate system to obtain an integrated energy consumption monitoring data with spatiotemporal correlation; establishing a mapping relationship by analyzing the correlation between traffic flow change patterns and energy consumption patterns in the integrated energy consumption monitoring data to generate a road segment energy consumption benchmark index; calculating multi-scale deviation by performing a weighted comparison operation between the road segment energy consumption benchmark index and the real-time collected measured energy consumption records at different time granularities to obtain a set of energy consumption deviation indexes covering different time granularities; converting the set of energy consumption deviation indexes into a visualization encoding vector containing amplitude and frequency domain features by establishing a mapping strategy between the deviation indexes and visual performance parameters, generating encoding instructions that can drive the display of a three-dimensional model; and driving the three-dimensional road network model to perform energy consumption status visualization rendering operations based on the visualization encoding vector, presenting energy consumption distribution characteristics and abnormal fluctuations through a combination of thermal layer overlay display and abnormal interval pulse warnings.

[0005] On the other hand, embodiments of the present invention provide a computer system including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement the steps in the above-described method.

[0006] This invention provides a method for visualizing energy consumption monitoring data based on highways. It integrates multiple types of monitoring data into a unified spatiotemporal reference system to obtain an integrated energy consumption monitoring data body. It analyzes the correlation between traffic flow and energy consumption patterns to generate road segment energy consumption benchmark indicators. Weighted comparisons at different time granularities are used to calculate multi-scale deviations, resulting in a set of energy consumption deviation indicators. A mapping strategy is established to convert these into visual encoding vectors containing amplitude and frequency domain features, generating encoding instructions. Finally, this drives a 3D road network model to present energy consumption distribution and fluctuations through thermal layer overlay and abnormal interval pulse warnings. This method not only ensures the spatiotemporal correlation of multi-source monitoring data through a unified spatiotemporal reference system, avoiding analytical biases caused by data asynchrony and spatial inconsistencies, but also lays a high-quality data foundation for subsequent energy consumption feature extraction and pattern recognition. It also establishes a correlation between traffic flow and energy consumption patterns. The energy consumption benchmark index generated by the dynamic modeling dynamically reflects the energy consumption benchmark level under different traffic conditions, overcoming the limitation of static benchmarks in adapting to complex road conditions and improving the accuracy of energy consumption deviation calculation. Moreover, by calculating multi-scale deviation through multi-time granularity weighted comparison, it comprehensively captures the energy consumption anomaly characteristics at different time scales, avoiding the blind spots of single-time granularity analysis in identifying short-term disturbances or long-term trends, and enhancing the sensitivity to energy consumption anomalies. Through the visualization encoding vector containing amplitude and frequency domain features, it realizes the multi-dimensional expression of energy consumption deviation information, enriching the information content of the visualization presentation. At the same time, through the three-dimensional rendering method combining thermal layer overlay and dynamic pulse warning, it intuitively presents the spatial distribution and dynamic change trend of energy consumption, improves the identification efficiency of energy consumption anomaly intervals, and avoids the problem of insufficient dynamic anomaly capture by traditional static visualization. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the implementation process of a method for visualizing energy consumption monitoring data based on highways, provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of the hardware entity of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0009] This invention provides a method for visualizing energy consumption monitoring data based on highways, which can be executed by a processor of a computer system. The computer system can refer to a server in the highway's backend system.

[0010] Figure 1 This is a schematic diagram illustrating the implementation process of a method for visualizing energy consumption monitoring data based on highways, as provided in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0011] Step S100: By establishing a unified time reference system and spatial coordinate system, the original monitoring records collected by various types of monitoring equipment deployed along the highway are integrated to obtain an integrated energy consumption monitoring data with spatiotemporal correlation.

[0012] A unified time reference system is a standard established to ensure the comparability and consistency of data collected by different monitoring devices across time, guaranteeing accurate temporal alignment of all data. A spatial coordinate system is used to determine the specific spatial location of various monitoring devices and points along the highway. It employs a common geographic coordinate system, such as latitude and longitude, to accurately locate each monitoring point. Raw monitoring records are unprocessed energy consumption data collected by various types of monitoring devices along the highway. These devices include, but are not limited to, power monitoring instruments installed on highway streetlights to monitor their energy consumption; power monitoring devices installed at toll stations to monitor their power consumption; and energy meters installed in service areas to record their energy consumption.

[0013] Specifically, a unified time reference system can be established first, selecting a standard time source, such as Coordinated Universal Time (UTC), and synchronizing the clocks of all monitoring devices with this standard time source. For the spatial coordinate system, the Global Positioning System (GPS) is used to locate each monitoring device, obtaining its precise latitude and longitude coordinates, and these coordinates are then associated with the corresponding monitoring devices. Next, the collected raw monitoring records are classified and organized according to the time reference system and the spatial coordinate system. For example, monitoring data from different locations at the same time point are sorted according to spatial coordinates; monitoring data from different times at the same location are arranged in chronological order. Finally, the organized data is integrated to form a spatiotemporally correlated energy consumption monitoring data consolidation system.

[0014] Step S200: Establish a mapping relationship by analyzing the correlation between traffic flow change patterns and energy consumption patterns in the energy consumption monitoring data integration system, and generate road segment energy consumption benchmark indicators.

[0015] Traffic flow variation patterns refer to the patterns and characteristics of traffic flow changes on highways over different time periods, including flow magnitude, fluctuations, and the timing of peak and trough periods. Energy consumption patterns refer to the patterns and characteristics of energy consumption on highways over different time periods, such as the timing of peak and trough energy consumption and the rate of change in energy consumption. Correlation is the relationship between traffic flow variation patterns and energy consumption patterns; by analyzing the correlation between the two, the degree and pattern of the impact of traffic flow changes on energy consumption can be determined. Mapping relationships represent the correlation between traffic flow variation patterns and energy consumption patterns in the form of mathematical models or functions, so as to predict energy consumption based on changes in traffic flow. Segment energy consumption benchmark indicators are determined based on historical data and analysis results, and are used to measure the normal energy consumption level of a highway segment, including energy consumption reference ranges under different traffic flow levels and predicted curves of energy consumption change trends.

[0016] In one implementation, step S200 may include the following steps S210 to S260:

[0017] Step S210: Perform time series segmentation processing on the integrated energy consumption monitoring data to obtain multiple time window units with periodic variation characteristics. Each time window unit contains traffic flow data sequence and energy consumption data sequence within the corresponding time period.

[0018] Time series segmentation involves dividing continuous energy consumption monitoring data into multiple non-overlapping time periods, each called a time window unit, according to certain time intervals. The periodic variation characteristic is that traffic flow and energy consumption exhibit recurring patterns within a certain time range, such as peak traffic flow and energy consumption during daily morning and evening rush hours. The traffic flow data series consists of traffic flow data recorded chronologically within each time window unit. The energy consumption data series consists of energy consumption data recorded chronologically within the same time window unit.

[0019] Step S220: Perform pattern recognition processing on the traffic flow data sequence in each time window unit to extract traffic flow change pattern features. The traffic flow change pattern features include the distribution of peak traffic times, the duration of the traffic rise phase, and the decay rate of the traffic fall phase.

[0020] Pattern recognition processing identifies representative patterns and features from traffic flow data sequences. The peak traffic flow time distribution shows the distribution of traffic flow reaching its peak within each time window, reflecting the regularity of peak traffic periods. The duration of the traffic flow rise phase is the time it takes for traffic flow to rise from its initial peak to reach its peak, reflecting the rate of traffic flow increase. The traffic flow decline phase decay rate is the speed at which traffic flow decreases from its peak, reflecting the speed at which traffic flow dissipates after the peak.

[0021] In one implementation, step S220 may include the following steps S221 to S226:

[0022] Step S221: Perform multi-resolution decomposition processing on the traffic flow data sequence, decompose the original traffic signal into multi-level signal components containing different fluctuation periods, and generate a multi-scale traffic signal set. The frequency range of the multi-level signal components decreases exponentially with the increase of the decomposition level.

[0023] Multi-resolution decomposition (MLD) can break down complex traffic flow data sequences into multiple signal components with different fluctuation periods, thereby enabling better analysis of the signal's local and global characteristics. The raw traffic flow signal is an unprocessed sequence of traffic flow data, containing fluctuation information of various frequencies and amplitudes. Multi-level signal components are signals in different frequency bands obtained through MLD processing; each level of signal component represents traffic flow changes at different fluctuation periods. The multi-scale traffic flow signal set is a collection of signal components from all levels, containing information about traffic flow data at different scales. The frequency range decreases exponentially with increasing decomposition levels: as the decomposition levels increase, the frequency range contained in each level of signal component gradually narrows, and the rate of narrowing increases exponentially.

[0024] When performing multi-resolution decomposition, wavelet transform algorithms, such as Daubechies wavelet and Symlets wavelet, can be used. Wavelet decomposition is applied to traffic flow data sequences, breaking them down into approximate and detail signals at different levels. The approximate signal represents the low-frequency components of the signal, corresponding to the long-term trend of traffic flow; the detail signal represents the high-frequency components of the signal, corresponding to short-term fluctuations in traffic flow. As the decomposition level increases, the approximate signal is further decomposed into finer-grained signal components, while the frequency range of the detail signal gradually shrinks. In this way, the original traffic flow signal is decomposed into multi-level signal components containing different fluctuation periods, generating a multi-scale traffic flow signal set.

[0025] Step S222: Select the highest-level component from the multi-scale flow signal set for trend extraction processing, generate the principal component of flow trend reflecting the macro-change law through the neighborhood data weighted average method, and merge the remaining components into the fluctuation residual component characterizing the local disturbance characteristics.

[0026] The highest-level component is the signal component with the narrowest frequency range and the longest fluctuation period in the multi-scale traffic flow signal set, reflecting the macroscopic trend of traffic flow changes. Trend extraction processing extracts features that represent the long-term variation patterns of traffic flow from the highest-level component. The neighborhood data weighted averaging method eliminates noise and local fluctuations in the data by weighting and averaging each data point and its neighborhood data points, thus obtaining a smoother trend curve. The principal component of traffic flow trend, obtained through trend extraction processing, is the main feature vector that reflects the macroscopic variation patterns of traffic flow. The fluctuation residual component is obtained by merging all components in the multi-scale traffic flow signal set except for the highest-level component, representing the local disturbances and short-term fluctuations of traffic flow.

[0027] Step S223: Calculate the energy proportion parameters of each level of signal components, construct the frequency domain feature spectrum of the flow signal, the horizontal axis of the frequency domain feature spectrum corresponds to the decomposition level, and the vertical axis corresponds to the cumulative distribution value of the energy proportion parameters.

[0028] The energy proportion parameter represents the percentage of energy of each signal component at each level within the total energy, reflecting the contribution of that level's signal component to the overall traffic flow signal. The frequency domain characteristic spectrum is a graphical representation used to describe the characteristics of a signal in the frequency domain. It allows a visual representation of the proportion of signal components in different frequency bands within the total signal. The horizontal axis corresponds to the decomposition levels, representing the different levels in the multi-resolution decomposition process, with each level corresponding to signal components in different frequency bands. The vertical axis corresponds to the cumulative distribution value of the energy proportion parameter, representing the cumulative value of the energy proportion parameter for each level of the frequency domain characteristic spectrum, reflecting the proportion of signal component energy from the lowest level to the current level within the total energy.

[0029] When calculating the energy proportion parameter, the energy of each level of signal component is first calculated. The signal energy can be obtained by integrating the square of the signal. For discrete signal sequences, the energy can be approximated by summing the squares of each element in the signal sequence. Then, the energy of each level of signal component is divided by the total energy of all levels of signal components to obtain the energy proportion parameter of that level of signal component. Next, the energy proportion parameters of each level of signal component are accumulated sequentially according to the decomposition level to obtain the cumulative distribution value of the energy proportion parameter. Finally, the frequency domain characteristic spectrum of the flow signal is plotted with the decomposition level as the horizontal axis and the cumulative distribution value of the energy proportion parameter as the vertical axis.

[0030] Step S224: Collect traffic flow data sequences of adjacent road segments within the same time window unit, calculate the synchronicity coefficient of traffic flow changes between road segments through cross-covariance analysis, and generate a road segment coupling feature vector containing spatial correlation strength.

[0031] Cross-covariance analysis (CCOVA) is a statistical method for measuring the correlation between two time series. By calculating the cross-covariance of two traffic flow data series, the degree of synchronicity between them can be obtained. The synchronicity coefficient, obtained through CCOVA, reflects the degree of synchronicity in traffic flow changes between adjacent road segments. The larger the synchronicity coefficient, the more synchronized the traffic flow changes between the two road segments. The road segment coupling feature vector is a vector containing information on the spatial correlation strength between road segments, which can be used to describe the mutual influence relationship of traffic flow between adjacent road segments.

[0032] Specifically, the mean of the two traffic flow data sequences is first calculated. Then, for each time point, the product of the difference between the two sequences at that time point and their respective means is calculated, and the products are summed over all time points. Finally, the summation is divided by the length of the two sequences to obtain the cross-covariance value. To obtain the synchronicity coefficient, the cross-covariance value is normalized, converting it into a value between -1 and 1. Based on the calculated synchronicity coefficient, a road segment coupling feature vector containing the spatial correlation strength is constructed.

[0033] Step S225: Perform pattern clustering on the fluctuation residual components, classify residual segments with the same fluctuation pattern by similarity measure, and generate a set of fluctuation pattern labels containing cluster center vectors.

[0034] Similarity metrics are indicators used to measure the degree of similarity between two residual segments. Similarity measurement methods include Euclidean distance and cosine similarity. Similar fluctuation patterns are residual segments with similar fluctuation shapes, amplitudes, and frequencies. The cluster center vector is the representative vector of each cluster, representing the mean or median of all residual segments in that cluster, reflecting the typical characteristics of that fluctuation pattern. The fluctuation pattern label set is a set of labels corresponding to each cluster, with each label representing a fluctuation pattern.

[0035] Step S226: Integrate the frequency domain feature spectrum, road segment coupling feature vector, and fluctuation pattern label set to construct a three-dimensional traffic flow change pattern feature tensor. The first dimension of the feature tensor corresponds to the time window unit number, the second dimension corresponds to the signal decomposition level, and the third dimension corresponds to the spatial correlation parameters.

[0036] The frequency domain feature spectrum, constructed in step S223, reflects the characteristics of traffic flow signals in the frequency domain. The road segment coupling feature vector, generated in step S224, contains information on the spatial correlation strength of traffic flow between adjacent road segments. The fluctuation pattern label set, obtained in step S225, represents different patterns of local traffic flow fluctuations. The three-dimensional traffic flow change pattern feature tensor is a three-dimensional data structure that integrates the information from the frequency domain feature spectrum, road segment coupling feature vector, and fluctuation pattern label set, providing a more comprehensive description of traffic flow change patterns. The first dimension corresponds to the time window unit index, representing different time window units, each corresponding to traffic flow data within a certain time period. The second dimension corresponds to the signal decomposition level, representing different levels in multi-resolution decomposition processing, each level corresponding to signal components in different frequency bands. The third dimension corresponds to spatial correlation parameters, representing spatial correlation information between road segments, such as synchronization coefficients.

[0037] When constructing the feature tensor of the three-dimensional traffic flow change pattern, the frequency domain feature spectrum, road segment coupling feature vector, and fluctuation pattern label set are first converted to a unified data format to ensure that they can be represented in the same tensor. Then, according to the correspondence between the time window unit number, signal decomposition level, and spatial correlation parameters, the corresponding feature information is filled into the corresponding positions of the feature tensor.

[0038] Step S230: Perform trend fitting processing on the energy consumption data sequence in each time window unit to generate an energy consumption trend curve, and extract energy consumption pattern features based on the energy consumption trend curve. The energy consumption pattern features include the time distribution characteristics of energy consumption peaks and valleys and the fluctuation amplitude parameter of energy consumption change rate.

[0039] Trend fitting is a process that uses mathematical models or methods to fit energy consumption data sequences, resulting in a curve that reflects their long-term trend. The energy consumption trend curve, obtained through trend fitting, smooths out short-term fluctuations and noise in the energy consumption data, showcasing the overall trend of energy consumption. Energy consumption pattern characteristics are extracted from the energy consumption trend curve and are used to describe the characteristic parameters of energy consumption patterns. The temporal distribution characteristics of energy consumption peaks and troughs represent the distribution of specific times when energy consumption reaches its peak and trough within each time window, reflecting the regularity of peak and trough periods. The fluctuation range parameter of the energy consumption change rate is the range of fluctuations in the rate of change of energy consumption over a period of time, reflecting the stability and drastic changes in energy consumption.

[0040] When performing trend fitting, methods such as polynomial fitting, exponential fitting, and spline fitting can be used. For example, suppose the energy consumption data sequence is y=[y1, y2, ..., yn], and the time points are x=[x1, x2, ..., xn]. A polynomial model can be chosen, such as y=a0+a1x+a2x. 2 +...+a m x m Where m is the order of the polynomial, a0, a1, ..., a m Let a0, a1, ..., am be the coefficients to be determined. Using the least squares method, the sum of squared errors between the fitted curve and the actual data points is minimized to solve for the values ​​of the coefficients a0, a1, ..., am, thus obtaining the energy consumption trend curve. Based on the obtained energy consumption trend curve, energy consumption pattern characteristics are extracted. For the time distribution characteristics of energy consumption peaks and troughs, the time points corresponding to the maximum and minimum values ​​in the curve can be determined by traversing the trend curve. For the fluctuation amplitude parameter of the energy consumption change rate, the energy consumption change rate curve can be obtained by calculating the first derivative of the trend curve, and then the difference between the maximum and minimum values ​​of this curve can be calculated to obtain the fluctuation amplitude parameter.

[0041] Step S240: Calculate the mutual information entropy value between traffic flow change pattern characteristics and energy consumption pattern characteristics, and construct the correlation evaluation matrix.

[0042] Mutual information entropy reflects the degree of interdependence between two variables. The general formula for calculating mutual information entropy can be used, and will not be elaborated upon here. Traffic flow change pattern characteristics include the distribution of peak traffic times, the duration of the traffic rise phase, and the rate of decline in traffic flow. Energy consumption pattern characteristics include the temporal distribution of energy consumption peaks and troughs, and the fluctuation range of the energy consumption change rate. The correlation evaluation matrix is ​​a two-dimensional matrix used to represent the correlation between traffic flow change pattern characteristics and energy consumption pattern characteristics. Rows correspond to traffic flow change pattern characteristics, columns correspond to energy consumption pattern characteristics, and the values ​​of the matrix elements are the mutual information entropy values ​​between the corresponding two characteristics.

[0043] Step S250: Determine the weighting coefficients of the influence of traffic flow change pattern characteristics on energy consumption pattern characteristics based on the correlation evaluation matrix, and establish a traffic flow-energy consumption correlation mapping model.

[0044] When determining the influence weight coefficients, a normalization method can be used to normalize the mutual information entropy values ​​in the correlation assessment matrix so that their sum is 1. For example, first calculate the sum of the elements in each column of the correlation assessment matrix. Then, divide each element in the matrix by the sum of the elements in that column to obtain the normalized weight coefficients. Based on the determined influence weight coefficients, a traffic flow-energy consumption correlation mapping model is established. A linear combination method can be used, multiplying the traffic flow change pattern features by the corresponding influence weight coefficients and then summing them to obtain the predicted values ​​of the energy consumption pattern features. For example, suppose the traffic flow change pattern features are x1, x2, ..., x... n The corresponding influence weight coefficients are w1, w2, ..., w n Then the predicted value y of the energy consumption pattern characteristics can be expressed as: y = w1x1 + w2x2 + ... + w n x n .

[0045] Step S260: Perform energy consumption simulation calculations on historical traffic flow data from the same period using a traffic flow-energy consumption correlation mapping model to generate road segment energy consumption benchmark indicators. The road segment energy consumption benchmark indicators include energy consumption reference ranges and trend prediction curves under different traffic flow levels.

[0046] Energy consumption simulation calculations predict corresponding energy consumption based on a traffic flow-energy consumption correlation mapping model and historical traffic flow data from the same period. The road segment energy consumption benchmark index, determined based on the energy consumption simulation calculation results, is used to measure the normal energy consumption level of a specific section of a highway. The energy consumption reference range under different traffic flow levels represents the normal energy consumption range for that road segment under different traffic flow volumes, serving as a reference standard for judging whether current energy consumption is normal. The trend prediction curve, plotted based on the energy consumption simulation calculation results, is used to predict the energy consumption trend of that road segment over a future period.

[0047] When performing energy consumption simulation calculations, historical traffic flow data from the same period is first input into the traffic flow-energy consumption correlation mapping model. The corresponding predicted energy consumption values ​​are then obtained based on the model's output. Next, statistical analysis is performed on the predicted energy consumption values ​​under different traffic flow levels to determine the reference range for energy consumption under each level. For example, traffic flow is divided into low, medium, and high levels, and the minimum, maximum, and average predicted energy consumption values ​​for each level are statistically analyzed. The minimum and maximum values ​​are then used as the reference range for energy consumption under that traffic flow level. Time series analysis methods, such as moving averages and exponential smoothing, can be used to generate the trend prediction curve.

[0048] Step S300: Perform a weighted comparison operation between the road section energy consumption benchmark index and the real-time collected energy consumption measurement records at different time granularities to calculate the multi-scale deviation, and obtain a set of energy consumption deviation indexes covering different time granularities.

[0049] Real-time energy consumption records are collected in real-time by monitoring equipment along the highway, representing current energy consumption data. Time granularity refers to the time interval used when analyzing energy consumption data; different time granularities provide different levels of energy consumption information, such as hourly, daily, and weekly levels. Weighted comparison assigns different weights to different data points when comparing the road segment's energy consumption benchmark and the real-time energy consumption records, reflecting their importance and reliability. Multi-scale deviation measures the degree of deviation between the real-time energy consumption records and the road segment's energy consumption benchmark at different time granularities, reflecting whether current energy consumption deviates from normal levels. The energy consumption deviation index set is a collection of energy consumption deviation indicators at different time granularities, providing a comprehensive understanding of the highway's energy consumption deviation across different time scales.

[0050] In one implementation, step S300 may include the following steps S310 to S360:

[0051] Step S310: Perform spatiotemporal coupling feature decomposition on the road segment energy consumption benchmark index, extract the road segment spatial correlation parameters and time dynamic evolution features, and generate a benchmark coupling feature matrix through the neighboring road segment energy consumption transmission model. The row dimension of the benchmark coupling feature matrix corresponds to the road segment spatial unit, the column dimension corresponds to the time evolution stage, and the matrix elements include the spatial correlation strength and the time evolution rate.

[0052] Spatiotemporal coupling feature decomposition processing decomposes the temporal and spatial information contained in the road segment energy consumption benchmark index, extracting spatial correlation parameters and temporal dynamic evolution characteristics. Road segment spatial correlation parameters represent the degree of spatial interrelationship between different road segments, such as the energy consumption transmission relationship between adjacent road segments and the impact of traffic flow between road segments. Temporal dynamic evolution characteristics describe the changing patterns and trends of road segment energy consumption at different time stages, such as increases, decreases, and periodic changes in energy consumption. The neighboring road segment energy consumption transmission model is a mathematical model used to describe the energy consumption transmission relationship between adjacent road segments. This model can be used to calculate the spatial correlation strength and temporal evolution rate between road segments.

[0053] When performing spatiotemporal coupling feature decomposition, the energy consumption benchmark indicators of road segments are first analyzed to identify spatial and temporal information. For extracting spatial correlation parameters, spatial statistical analysis methods, such as spatial autocorrelation analysis and spatial interpolation, can be used. For example, spatial autocorrelation analysis can be used to calculate the correlation between energy consumption of adjacent road segments to obtain the spatial correlation strength between road segments. For extracting temporal dynamic evolution features, time series analysis methods, such as trend analysis and periodic analysis, can be used. For example, trend analysis can be used to fit the curve of road segment energy consumption over time to obtain the temporal evolution rate of energy consumption. Then, the extracted spatial correlation parameters and temporal dynamic evolution features are input into the neighboring road segment energy consumption transmission model. The neighboring road segment energy consumption transmission model can be a physics-based model, such as a variant of the heat conduction model, or a data-driven model, such as a neural network model. Through model calculation, the spatial correlation strength and temporal evolution rate of each road segment at different temporal evolution stages are obtained. Finally, this information is filled into the benchmark coupling feature matrix.

[0054] Step S320: Perform the same spatiotemporal coupling feature decomposition process as the road section energy consumption benchmark index on the measured energy consumption records to obtain the measured coupling feature matrix. The spatial unit division and temporal evolution stage division of the measured coupling feature matrix are consistent with the benchmark coupling feature matrix.

[0055] When executing the spatiotemporal coupling feature decomposition process, first ensure that the spatial unit division and temporal evolution stage division of the measured energy consumption records are consistent with the baseline coupling feature matrix. For spatial unit division, the measured energy consumption records are allocated to each road segment spatial unit according to the same road segment division method as the baseline coupling feature matrix. For temporal evolution stage division, the measured energy consumption records are divided into each temporal evolution stage according to the same time interval division method as the baseline coupling feature matrix. Then, the measured energy consumption records in each spatial unit and temporal evolution stage are analyzed to extract road segment spatial correlation parameters and temporal dynamic evolution characteristics. For the extraction of spatial correlation parameters, spatial statistical analysis methods, such as spatial autocorrelation analysis and spatial interpolation, are used. For the extraction of temporal dynamic evolution characteristics, time series analysis methods, such as trend analysis and periodic analysis, are used. Finally, the extracted measured spatial correlation strength and measured temporal evolution rate are filled into the measured coupling feature matrix.

[0056] In one implementation, step S320 may include the following steps S321 to S326:

[0057] Step S321: Analyze the spatial distribution characteristics of the measured energy consumption records, extract the location of monitoring equipment, road segment topology, and communication delay parameters between equipment, and generate a set of measured spatial characteristic parameters. The set of measured spatial characteristic parameters is used to guide the division of spatial units.

[0058] The spatial distribution characteristics of measured energy consumption records refer to their spatial location along the highway, which is closely related to the deployment location of monitoring equipment and the road segment topology. The deployment location of monitoring equipment refers to the specific geographical location of the monitoring equipment installed along the highway to collect energy consumption data; this location information can be obtained through the Global Positioning System (GPS). Road segment topology refers to the connections and relative positions between different highway segments, such as adjacency and upstream / downstream relationships. Inter-device communication delay parameters are the time delays incurred when data is transmitted between monitoring devices, affecting the real-time performance and accuracy of the energy consumption data. The measured spatial characteristic parameter set is a collection containing information such as the deployment location of monitoring equipment, road segment topology, and inter-device communication delay parameters.

[0059] When analyzing the spatial distribution characteristics of measured energy consumption records, the deployment location information of monitoring equipment is first collected, which can be obtained by querying equipment installation records or using GPS positioning technology. Then, the topological relationships of road segments are analyzed, and relevant information can be obtained from highway design drawings, Geographic Information System (GIS) data, etc. For extracting communication delay parameters between devices, test signals are sent between monitoring devices, and the time difference of signal transmission is recorded. The collected monitoring equipment deployment locations, road segment topological relationships, and inter-device communication delay parameters are then organized and integrated to generate a set of measured spatial characteristic parameters. For example, the set of measured spatial characteristic parameters can be represented as a list containing multiple elements, each corresponding to relevant information for a monitoring device or road segment. When dividing spatial units, based on the set of measured spatial characteristic parameters, factors such as the distribution density of monitoring equipment, the connectivity of road segments, and communication delays are considered to rationally divide spatial units, ensuring that the energy consumption data within each spatial unit has good consistency and correlation.

[0060] Step S322: Based on the measured spatial characteristic parameter set, replicate the spatial unit division scheme of the benchmark coupling feature matrix, including the number of road segments, the neighborhood association radius and the spatial grid resolution, to generate the measured spatial framework.

[0061] The neighborhood association radius is the association range between each road segment and its adjacent road segments, used to account for the mutual influence between road segments. Spatial grid resolution refers to the size and precision of spatial cells, affecting the accuracy of spatial data representation and analysis. The measured spatial framework is generated based on the spatial cell partitioning scheme of the measured spatial characteristic parameter set and the benchmark coupling characteristic matrix, and is used for spatial organization and analysis of measured energy consumption records.

[0062] When generating the measured spatial framework, relevant information is first obtained from the measured spatial characteristic parameter set to understand the distribution of monitoring equipment and the topology of road segments. Then, according to the spatial unit division scheme of the reference coupling feature matrix, the number of road segment segments, the neighborhood association radius, and the spatial grid resolution are determined. For the number of road segment segments, the division number of the reference coupling feature matrix is ​​directly adopted, dividing the highway into the corresponding number of segments. For the neighborhood association radius, the neighborhood range of each road segment is determined based on the settings of the reference coupling feature matrix. For the spatial grid resolution, the resolution of the reference coupling feature matrix is ​​also copied, dividing the space into grids of the appropriate size. Finally, based on the determined parameters, the measured spatial framework is generated.

[0063] Step S323: Perform spatial interpolation processing on the measured energy consumption records to convert discrete monitoring point data into continuously spatially distributed energy consumption surface data. The spatial resolution of the energy consumption surface data is consistent with the grid resolution of the measured spatial frame.

[0064] When performing spatial interpolation, various interpolation algorithms can be used, such as inverse distance weighted interpolation, kriging interpolation, and spline interpolation. For example, using the inverse distance weighted interpolation algorithm, the location of the point to be interpolated is first determined, i.e., each grid point in the measured spatial frame. Then, for each point to be interpolated, the surrounding monitoring points are determined, and the distance between the point to be interpolated and each monitoring point is calculated. Next, the weight of each monitoring point is calculated based on the distance. The formula for calculating the weight is: weight = 1 / distance. p Where p is the power exponent, with a value of, for example, 2. Finally, the energy consumption data of the monitoring points are weighted and averaged to obtain the energy consumption value of the point to be interpolated. By interpolating all grid points in the measured spatial frame, the energy consumption surface data of the continuous spatial distribution is obtained.

[0065] Step S324: Calculate the energy consumption surface data gradient of adjacent spatial units, and generate the initial value of spatial correlation strength through spatial covariance analysis within the sliding window. The initial value of spatial correlation strength reflects the spatial coupling degree of energy consumption of adjacent road segments.

[0066] Adjacent spatial units are geographically adjacent units within a measured spatial framework, and their energy consumption data may be mutually influential and correlated. The gradient of energy consumption surface data is the rate of change of energy consumption surface data in space, reflecting the variation of energy consumption among adjacent spatial units. The finite difference method can be used to calculate the gradient of energy consumption surface data for adjacent spatial units. For two-dimensional energy consumption surface data, the first-order differences in the x and y directions are calculated separately to obtain the gradient components of the energy consumption surface data in both directions. Then, the magnitude of the gradient is calculated based on the gradient components to obtain the gradient of the energy consumption surface data. Next, a sliding window is used to perform spatial covariance analysis on the energy consumption data of adjacent spatial units. For example, the size and step size of the sliding window are first determined. Then, the sliding window is moved spatially, and the covariance of the energy consumption data of adjacent spatial units within the window is calculated each time it moves to a new position. Finally, the calculated covariance is used as the initial value for the spatial correlation strength.

[0067] Step S325: Combine time series analysis to extract the time evolution rate of energy consumption surface data, calculate the rate parameter by the ratio of energy consumption change to time interval between adjacent time steps, and generate a time evolution feature sequence.

[0068] The temporal evolution rate of energy consumption surface data is the speed at which the data changes over time, reflecting the degree of change in energy consumption between different points in time. Adjacent time steps are the time intervals between two adjacent time points in a time series. The change in energy consumption is the difference in energy consumption surface data within adjacent time steps. The rate parameter is calculated by the ratio of the change in energy consumption to the time interval within adjacent time steps, quantifying the temporal evolution rate of the energy consumption surface data. The temporal evolution feature sequence is a sequence containing the rate parameter for each time point, reflecting the temporal evolution characteristics of the energy consumption surface data over the entire time range.

[0069] In time series analysis, energy consumption data are first arranged chronologically to form a time series. Then, for each time point, the change in energy consumption compared to adjacent time points is calculated. Specifically, this is done by subtracting the energy consumption data from the previous time point's data. Next, rate parameters are calculated based on the time intervals between adjacent time steps. The rate parameters for each time point are then arranged sequentially to generate a time evolution characteristic sequence.

[0070] Step S326: Integrate the initial value of spatial correlation strength and the temporal evolution feature sequence to generate a measured coupling feature matrix that is completely aligned with the structural and reference coupling feature matrix. Each element of the measured coupling feature matrix contains the measured spatial correlation strength and the measured temporal evolution rate.

[0071] The measured coupling feature matrix is ​​a matrix used to comprehensively represent the spatial and temporal characteristics of road segment energy consumption. Its structure is completely aligned with the benchmark coupling feature matrix, ensuring the comparability of measured data and benchmark data in the spatiotemporal dimensions. The measured spatial correlation strength is the spatial correlation strength information contained in each element of the measured coupling feature matrix, obtained through further processing and integration based on the initial values ​​of spatial correlation strength. The measured temporal evolution rate is the temporal evolution rate information contained in each element of the measured coupling feature matrix, derived from the temporal evolution feature sequence.

[0072] When integrating the initial values ​​of spatial correlation strength with the temporal evolution feature sequence, the size and dimension of the measured coupling feature matrix are first determined based on the structure of the reference coupling feature matrix. Then, the information in the initial values ​​of spatial correlation strength and the temporal evolution feature sequence is matched and filled according to the corresponding spatial units and temporal evolution stages. For example, for each element of the measured coupling feature matrix, its corresponding spatial unit and temporal evolution stage are found. In the initial values ​​of spatial correlation strength, the spatial correlation strength of the adjacent spatial units corresponding to that spatial unit is found and filled into the matrix element as the measured spatial correlation strength. In the temporal evolution feature sequence, the temporal evolution rate corresponding to that temporal evolution stage is found and filled into the matrix element as the measured temporal evolution rate.

[0073] Step S330: Calculate the coupling difference tensor between the reference coupling feature matrix and the measured coupling feature matrix, and generate a coupling difference data structure containing the spatial difference field and the temporal difference sequence through joint modeling of spatial correlation strength deviation and temporal evolution rate deviation.

[0074] The coupling difference tensor is a three-dimensional tensor used to describe the differences between the reference coupling feature matrix and the measured coupling feature matrix, comprehensively reflecting the spatial and temporal differences in road segment energy consumption. The spatial association strength deviation is the difference between the spatial association strength in the reference coupling feature matrix and the measured coupling feature matrix; it reflects the change in the spatial coupling degree of energy consumption between adjacent road segments. The temporal evolution rate deviation is the difference between the temporal evolution rate in the reference coupling feature matrix and the measured coupling feature matrix; it reflects the difference in the rate of change of road segment energy consumption over time. Joint modeling comprehensively considers the spatial association strength deviation and the temporal evolution rate deviation to establish a unified model to describe their relationship. The spatial difference field is a two-dimensional field used to represent the spatial distribution of road segment energy consumption differences and can be generated from the spatial association strength deviation. The temporal difference sequence is a one-dimensional sequence used to represent the temporal variation of road segment energy consumption differences and can be generated from the temporal evolution rate deviation. The coupled difference data structure is a composite data structure that includes spatial difference fields and temporal difference sequences, which can more comprehensively describe the differences between the benchmark coupled feature matrix and the measured coupled feature matrix.

[0075] In one implementation, step S330 may include the following steps S331 to S336:

[0076] Step S331: Construct a coupled feature alignment index. The row dimension of the coupled feature alignment index corresponds to the spatial unit number, and the column dimension corresponds to the time evolution stage. Each index item includes the baseline spatial correlation strength, the baseline time evolution rate, the measured spatial correlation strength, and the measured time evolution rate.

[0077] The coupling feature comparison index is a two-dimensional index structure used to organize and store key information in the benchmark coupling feature matrix and the measured coupling feature matrix. The time evolution stage divides time into different intervals to represent the change process of energy consumption data over time. The benchmark spatial correlation strength is the spatial correlation strength information contained in each element of the benchmark coupling feature matrix, reflecting the degree of spatial coupling between road segments under the benchmark condition. The benchmark time evolution rate is the time evolution rate information contained in each element of the benchmark coupling feature matrix, reflecting the rate of change of road segment energy consumption over time under the benchmark condition.

[0078] When constructing the coupling feature alignment index, the row and column dimensions of the index are first determined. The row dimension corresponds to the spatial unit number, and the number of rows is determined based on the number of spatial units in the measured spatial framework. The column dimension corresponds to the temporal evolution stage, and the number of columns is determined based on the number of temporal evolution stages. Then, for each element of the index, i.e., each index entry, the corresponding baseline spatial correlation strength, baseline temporal evolution rate, measured spatial correlation strength, and measured temporal evolution rate are extracted from the baseline coupling feature matrix and the measured coupling feature matrix. For example, for the (i, j)th element of the index, where i is the spatial unit number and j is the temporal evolution stage, the spatial correlation strength and temporal evolution rate of the (i, j)th element are extracted from the baseline coupling feature matrix as the baseline spatial correlation strength and baseline temporal evolution rate, and the spatial correlation strength and temporal evolution rate of the (i, j)th element are extracted from the measured coupling feature matrix as the measured spatial correlation strength and measured temporal evolution rate. This information is stored in the index entry.

[0079] Step S332: Traverse the coupling feature comparison index, calculate the difference between the baseline spatial correlation strength and the measured spatial correlation strength in each index entry, and generate a spatial correlation strength deviation. The spatial correlation strength deviation retains the positive or negative sign to indicate the strengthening or weakening trend.

[0080] After constructing the coupling feature comparison index, it is traversed. Each index entry contains two key pieces of information: the baseline spatial correlation strength and the measured spatial correlation strength. When calculating the difference, the baseline spatial correlation strength is subtracted from the measured spatial correlation strength, and the result is the spatial correlation strength deviation. Retaining the positive and negative signs is significant: a positive sign indicates that the measured spatial correlation strength is stronger than the baseline, meaning the energy consumption spatial coupling between adjacent road segments is stronger; a negative sign indicates that the measured spatial correlation strength is weaker than the baseline, meaning the energy consumption spatial coupling between adjacent road segments is weaker.

[0081] Step S333: Perform the same difference calculation on the reference time evolution rate and the measured time evolution rate in the index item to generate the time evolution rate deviation. When the reference time evolution rate is zero, set the time evolution rate deviation to the absolute value of the measured time evolution rate.

[0082] After obtaining the spatial correlation strength deviation, it is also necessary to analyze the temporal evolution rate. Similarly, based on the coupling feature comparison index, the baseline temporal evolution rate and the measured temporal evolution rate in each index entry are processed. The baseline temporal evolution rate reflects the rate at which road segment energy consumption changes over time under baseline conditions, while the measured temporal evolution rate is the currently measured rate of change of road segment energy consumption over time.

[0083] When performing the difference calculation, similar to calculating the spatial correlation strength deviation, the measured time evolution rate is subtracted from the baseline time evolution rate to obtain the time evolution rate deviation. This deviation value reflects the change in the measured time evolution rate relative to the baseline. When the baseline time evolution rate is zero, it may mean that under the baseline setting, the energy consumption of this road segment does not show a significant trend over time. In this case, directly performing the difference calculation may not accurately reflect the actual situation, so the time evolution rate deviation is set as the absolute value of the measured time evolution rate. The purpose of this is to highlight the impact of the measured time evolution rate on the overall energy consumption over time, because even if the baseline rate is zero, the existence of the measured rate indicates a new dynamic change in energy consumption over time. By performing this calculation for each index item, complete time evolution rate deviation data can be obtained, which helps to analyze the differences in energy consumption changes over time in road segments.

[0084] Step S334: Construct a spatial difference field based on the spatial correlation strength deviation, and convert the discrete spatial deviation values ​​into a continuously distributed spatial difference surface. The contour lines of the spatial difference surface represent the spatial gradient changes of the deviation.

[0085] Spatial correlation strength deviation reflects the change in the spatial coupling degree of energy consumption between adjacent road segments relative to the baseline. To more intuitively illustrate the spatial distribution of this change, a spatial difference field needs to be constructed based on the spatial correlation strength deviation. Discrete spatial deviation values ​​are calculated through the previous steps, with each value corresponding to the spatial correlation strength deviation of a spatial unit, but these values ​​are discretely distributed across each spatial unit.

[0086] Contour lines on a spatial difference surface represent lines connecting points on the surface with equal deviation values. The density of contour lines reflects the spatial gradient variation of the deviation. If the contour lines are relatively dense, it indicates that the spatial correlation strength deviation varies drastically within the region, meaning that the deviation difference between adjacent spatial units is large; if the contour lines are relatively sparse, it indicates that the spatial correlation strength deviation varies relatively gently within the region, meaning that the deviation difference between adjacent spatial units is small.

[0087] Step S335: Sort the time evolution rate deviations according to the time series to generate a time difference sequence. The time difference sequence contains the rate deviation value of each time evolution stage and the deviation change rate of adjacent stages.

[0088] The time evolution rate deviation reflects the difference in the rate of change of road segment energy consumption over time relative to a baseline. Arranging the time evolution rate deviations in chronological order of the time evolution stages yields a time difference sequence. This sequence includes not only the rate deviation value for each time evolution stage but also the deviation change rate between adjacent stages. The deviation change rate between adjacent stages is obtained by calculating the difference in the time evolution rate deviations between two adjacent time evolution stages, reflecting the degree of change in the time evolution rate deviation between adjacent time stages.

[0089] Step S336: Fuse the spatial difference field and the temporal difference sequence to construct a three-dimensional coupled difference tensor. The first dimension of the three-dimensional coupled difference tensor is the spatial unit, the second dimension is the temporal stage, and the third dimension is the deviation type.

[0090] After obtaining the spatial difference field and temporal difference sequence, to comprehensively consider the differences in road segment energy consumption relative to the baseline in both space and time, they need to be fused to construct a three-dimensional coupled difference tensor. The spatial difference field displays the spatial distribution of correlation strength deviations in road segment energy consumption; it is a two-dimensional information structure describing the differences between different spatial units. The first dimension of the three-dimensional coupled difference tensor corresponds to the spatial unit, meaning that different spatial locations can be distinguished in this dimension, and each spatial unit has a corresponding position in the tensor. The second dimension corresponds to the time stage, reflecting the energy consumption differences at different points in time. The third dimension corresponds to the deviation type, including two different types of deviation information: spatial correlation strength deviation and temporal evolution rate deviation.

[0091] Step S340: Based on the spatial correlation strength deviation in the coupling difference tensor, a dynamic correlation comparison window is constructed. The dynamic correlation comparison window dynamically adjusts the spatial coverage and time duration according to the energy consumption transmission coefficient between road segments. For road segments whose transmission coefficient exceeds the preset transmission threshold, the spatial coverage and time duration of the corresponding dynamic correlation comparison window are reduced by a preset ratio.

[0092] The coupling difference tensor contains important information such as spatial correlation strength deviation and temporal evolution rate deviation. The purpose of constructing a dynamic correlation comparison window based on the spatial correlation strength deviation is to analyze the correlation of road segment energy consumption more accurately.

[0093] The energy transfer coefficient between road segments reflects the degree of mutual influence on energy consumption between adjacent road segments. Due to differences in factors such as geographical location, traffic flow, and equipment operating conditions, the energy transfer coefficient will vary between different road segment pairs.

[0094] The dynamic correlation comparison window has the characteristic of dynamic adjustment; its spatial coverage and duration change according to the energy transfer coefficient between road segments. The preset transfer threshold is a pre-set standard value used to determine the strength of energy transfer between road segments.

[0095] When the transmission coefficient of a road segment pair exceeds a preset transmission threshold, it indicates that the energy consumption transmission between the two road segments is relatively strong, and their energy consumption correlation is closer. In this case, to more focusedly analyze the energy consumption differences between the two road segments, the spatial coverage of the corresponding dynamic correlation comparison window is reduced by a preset ratio, making the window more concentrated in a smaller area of ​​the two road segments and their surroundings; at the same time, the duration is also reduced by a preset ratio to more precisely capture the energy consumption changes of the two road segments over a shorter period of time. Conversely, if the transmission coefficient of a road segment pair does not exceed the preset transmission threshold, it indicates that the energy consumption transmission between them is relatively weak, and the spatial coverage and duration of the dynamic correlation comparison window can remain relatively large to more comprehensively analyze the energy consumption correlation of the area over a longer period of time.

[0096] Step S350: The coupling difference tensor is divided into blocks according to the dynamic correlation comparison window to generate difference feature blocks corresponding to different correlation scales. Each difference feature block contains the coupling difference statistics of spatial sub-region and temporal sub-stage.

[0097] After constructing the dynamic correlation comparison window, it is used to further process the coupled difference tensor. The coupled difference tensor integrates the differences in road segment energy consumption relative to the baseline in space and time, and the purpose of block processing is to divide this huge tensor data according to different correlation scales in order to analyze energy consumption differences in more detail.

[0098] The dynamic correlation comparison window provides a clear scope and temporal definition for block processing. Based on the spatial coverage and temporal duration of the window, the coupled difference tensor is divided into multiple sub-blocks. Each sub-block corresponds to a spatial sub-region and a temporal sub-stage; this sub-block is the difference feature block.

[0099] The difference feature block contains statistical measures of coupling differences between spatial sub-regions and temporal sub-stages. These statistics can include the mean, variance, maximum, and minimum values ​​of spatial correlation strength deviation and temporal evolution rate deviation. By calculating these statistics, the energy consumption differences within the spatial sub-region and temporal sub-stage can be understood macroscopically.

[0100] Step S360: Perform nonlinear correlation aggregation processing on the differential feature blocks through a cross-scale feature fusion network to generate a set of energy consumption deviation indicators covering the dynamic correlation scale. Each indicator in the set of energy consumption deviation indicators includes the spatial difference contribution rate, the temporal difference contribution rate, and the comprehensive deviation quantification value.

[0101] During processing, cross-scale feature fusion networks consider the mutual influence and correlation between feature blocks at different scales. For example, energy consumption differences at a large scale may be affected by local energy consumption changes at a small scale, while energy consumption anomalies at a small scale may also be reflected at a large scale. Through the nonlinear processing of the network, these complex relationships can be captured, effectively integrating information from different scales.

[0102] In one implementation, step S360 may include the following steps S361 to S366:

[0103] Step S361: Extract the associated weight vectors corresponding to each differential feature block from the cross-scale feature fusion network. The associated weight vectors are jointly determined based on the information entropy and mutual information value of the feature block. The weight coefficients of the associated weight vectors corresponding to feature blocks whose information entropy exceeds a preset entropy threshold and whose mutual information value exceeds a preset mutual information threshold are increased according to a preset increase rule.

[0104] When processing differential feature blocks, cross-scale feature fusion networks assign a correlation weight vector to each differential feature block. This vector reflects the importance of the feature block in the entire fusion process. The correlation weight vector is determined based on the information entropy and mutual information value of the feature block.

[0105] Information entropy is an indicator that measures the uncertainty of information. For differential feature blocks, information entropy reflects the complexity and uncertainty of the energy consumption difference information contained in the feature block. The higher the information entropy, the more complex the information in the feature block, and it may contain more unique energy consumption difference information. Mutual information value is used to measure the correlation between two feature blocks. In this embodiment of the invention, it reflects the degree of correlation between differential feature blocks at different scales.

[0106] The preset entropy threshold and preset mutual information threshold are pre-defined standard values ​​used to filter out feature blocks with high importance. When the information entropy of a feature block exceeds the preset entropy threshold and its mutual information value exceeds the preset mutual information threshold, it indicates that the feature block contains rich and unique information and has a strong correlation with other feature blocks, making it significant for the overall energy consumption deviation assessment. For such feature blocks, the weight coefficient of their corresponding correlation weight vector is increased according to a preset increase rule. The preset increase rule can be a fixed percentage increase or dynamically adjusted based on the degree to which the information entropy and mutual information values ​​exceed the threshold.

[0107] Step S362: Perform element-wise multiplication of the coupled difference statistics of each difference feature block with the corresponding associated weight vector to generate a weighted difference feature block. The value of each statistic in the weighted difference feature block is the product of the original statistic value and the weight coefficient.

[0108] Each differential feature block contains coupled differential statistics between spatial sub-regions and temporal sub-stages. These statistics, such as mean and variance, are important data describing the energy consumption differences within that feature block. The association weight vector assigns a corresponding weight to each statistic, reflecting the importance of that feature block in the overall fusion process.

[0109] Element-wise multiplication involves multiplying each statistic value in the differential feature block by the corresponding weight coefficient in the associated weight vector. This operation yields a weighted differential feature block. In the weighted differential feature block, each statistic value is no longer the original value, but rather the result after weight adjustment—that is, the product of the original statistic value and the weight coefficient.

[0110] Step S363: Perform cross-scale feature transfer processing on the weighted difference feature block. The time granularity fluctuation difference information corresponding to the basic correlation scale of the basic correlation scale feature block is transferred to the fusion correlation scale feature block through the feature mapping function, and a fusion correlation scale feature block with supplemented detailed difference information is generated. The supplemented detailed difference information includes the time granularity fluctuation frequency and amplitude features corresponding to the basic correlation scale.

[0111] After obtaining the weighted differential feature blocks, cross-scale feature transfer processing is needed to further explore the information correlation between feature blocks of different scales. The basic correlation scale feature blocks and the fusion correlation scale feature blocks are differential feature blocks at different scales, representing energy consumption differences at different spatial ranges and temporal granularities, respectively.

[0112] The basic correlation scale corresponds to a smaller spatial range and a finer temporal granularity. Feature blocks at this scale may contain more detailed information on energy consumption fluctuation differences, such as the fluctuation frequency and amplitude characteristics at the time granularity corresponding to the basic correlation scale. In contrast, feature blocks at the fusion correlation scale typically correspond to a larger spatial range and a coarser temporal granularity, and their information is relatively macroscopic.

[0113] Feature mapping functions are tools for achieving cross-scale feature transfer. They can transfer information from the basic correlated scale feature block—that is, the temporal granularity fluctuation difference information corresponding to the basic correlated scale—to the fused correlated scale feature block. Through this transfer, the fused correlated scale feature block obtains supplementary detailed difference information, forming a fused correlated scale feature block with supplementary detailed difference information.

[0114] Step S364: Based on the fusion of related scale feature blocks with supplementary detail difference information, calculate the difference contribution of each related scale feature block, assign the contribution ratio of each feature block to the overall deviation, and generate a contribution allocation vector containing the contribution of time granularity fluctuation difference items corresponding to the basic related scale.

[0115] The difference contribution is the magnitude of each associated scale feature block's contribution to the overall deviation. To calculate the difference contribution, it's necessary to comprehensively consider various information contained in the fused associated scale feature blocks that supplement detailed difference information, such as spatial association strength deviation, temporal evolution rate deviation, and temporal granularity fluctuation differences corresponding to the supplemented basic association scale. Using a pre-defined algorithm or model, this information is analyzed and calculated to determine the difference contribution of each feature block. Then, the contribution ratio of each feature block to the overall deviation is allocated based on its difference contribution. This contribution ratio reflects the importance of each feature block in the overall energy consumption deviation assessment.

[0116] The generated contribution allocation vector includes the contribution term for the fluctuation difference at the time granularity corresponding to the basic correlation scale. This is because, in the previous steps, the fluctuation difference information at the time granularity corresponding to the basic correlation scale was supplemented into the feature block of the fused correlation scale, and this information also contributes to the overall deviation.

[0117] Step S365: Weight the difference statistics of the contribution allocation vector and the corresponding feature block, and focus on integrating the time granularity fluctuation difference information of the basic association scale during the summation process to generate a comprehensive deviation quantification value that reflects both the long-term trend of the integrated association scale and the time granularity fluctuation characteristics of the basic association scale.

[0118] After obtaining the contribution assignment vector, a weighted summation operation is performed between it and the difference statistics of the corresponding feature blocks. The difference statistics of the corresponding feature blocks contain coupling difference information between spatial sub-regions and temporal sub-stages, such as spatial correlation strength deviation and temporal evolution rate deviation.

[0119] The weighted summation process involves multiplying each element in the contribution allocation vector by the corresponding element in the difference statistics of the corresponding feature block, and then summing all the products. In this summation process, the focus is on fusing the supplementary time-granularity fluctuation difference information corresponding to the basic correlation scale. This is because this fluctuation difference information was supplemented into the fused correlation scale feature blocks in the previous steps, providing more microscopic and detailed information on energy consumption changes.

[0120] Step S366: Integrate the comprehensive deviation measurement value, spatial difference contribution rate, temporal difference contribution rate and corresponding correlation scale identifier to generate a set of energy consumption deviation indicators covering dynamic correlation scales. The indicators in the energy consumption deviation indicator set are sorted from smallest to largest according to the correlation scale.

[0121] After obtaining the comprehensive deviation quantification value, spatial difference contribution rate, and temporal difference contribution rate, this important information is fused with the corresponding correlation scale identifier to generate a set of energy consumption deviation indicators covering dynamic correlation scales.

[0122] The comprehensive deviation quantification is a numerical value that comprehensively measures the degree of deviation of highway energy consumption from the baseline, integrating differences in spatial and temporal dimensions. The spatial difference contribution rate reflects the proportion of the spatial correlation strength deviation in the overall energy consumption deviation, highlighting the importance of spatial factors in energy consumption deviation. The temporal difference contribution rate represents the proportion of the temporal evolution rate deviation in the overall energy consumption deviation, emphasizing the impact of temporal factors on energy consumption deviation.

[0123] The corresponding scale identifier is used to distinguish energy consumption deviation information at different scales, and can be the scale number or other identifier. By combining the comprehensive deviation quantification value, spatial difference contribution rate, temporal difference contribution rate and corresponding scale identifier, a complete energy consumption deviation index is formed.

[0124] Step S400: By establishing a mapping strategy between deviation indicators and visual performance parameters, the energy consumption deviation indicator set is converted into a visual encoding vector containing amplitude and frequency domain features, generating encoding instructions that can drive the display of 3D models.

[0125] The energy consumption deviation index set contains energy consumption deviation information at different correlation scales. To present this abstract data in an intuitive way, a mapping strategy needs to be established to link the deviation index with visual performance parameters. Visual performance parameters are used to control the display effect of 3D models, such as color, transparency, and dynamic change frequency.

[0126] The mapping strategy is a rule that maps the values ​​of deviation indicators to the values ​​of visual performance parameters. Using this strategy, the corresponding visual performance parameter values ​​can be determined based on the magnitude and characteristics of the energy consumption deviation indicator. For example, a higher overall deviation indicator value may correspond to a more vibrant color or a faster frequency of dynamic changes.

[0127] In one implementation, step S400 may specifically include the following steps S410 to S460:

[0128] Step S410: Perform spatiotemporal feature fusion processing on the energy consumption deviation index set, extract the spatial distribution features of the spatial difference contribution rate and the time series features of the time difference contribution rate, and generate a deviation spatiotemporal feature matrix through spatiotemporal correlation modeling. The row dimension of the deviation spatiotemporal feature matrix corresponds to the spatial location of the road segment, the column dimension corresponds to the time sampling point, and the matrix elements contain the comprehensive deviation quantification value.

[0129] The energy consumption deviation index set contains energy consumption deviation information at different correlation scales. Among them, the spatial difference contribution rate reflects the influence of spatial factors on energy consumption deviation, while the temporal difference contribution rate reflects the role of temporal factors. In order to analyze this information more comprehensively, the energy consumption deviation index set needs to be subjected to spatiotemporal feature fusion processing.

[0130] First, the spatial distribution characteristics of the spatial difference contribution rate are extracted from the energy consumption deviation index set. These characteristics describe the distribution of the spatial difference contribution rate across different road segments, revealing which road segments have a greater spatial contribution to energy consumption deviation. Simultaneously, the time-series characteristics of the temporal difference contribution rate are extracted. These characteristics reflect the changing trend of the temporal difference contribution rate over time, demonstrating the impact of dynamic changes in energy consumption over time on deviation. Then, the spatial distribution characteristics and time-series characteristics are combined through spatiotemporal correlation modeling (such as spatiotemporal regression models or spatiotemporal autocorrelation analysis). The matrix elements contain a comprehensive deviation metric, integrating the influence of spatial and temporal factors on energy consumption deviation.

[0131] Step S420: Construct a deviation-visual parameter association tensor, and associate the deviation spatiotemporal feature matrix with the preset visual performance parameter library to generate a three-dimensional association data structure containing spatial visual channels, temporal visual channels and dynamic visual channels.

[0132] Association mapping is the process of converting elements in the deviation spatiotemporal feature matrix to parameters in a pre-defined visual performance parameter library. Through this mapping, the corresponding visual performance parameter values ​​can be determined based on information such as the comprehensive deviation quantification value, spatial difference contribution rate, and temporal difference contribution rate in the deviation spatiotemporal feature matrix.

[0133] The constructed deviation-visual parameter correlation tensor is a three-dimensional correlation data structure containing spatial visual channels, temporal visual channels, and dynamic visual channels. The spatial visual channel controls visual representations related to spatial location, such as the color of different road segments or the scaling of spatial dimensions; the temporal visual channel handles time-related visual changes, such as changes in transparency over time or the flashing frequency of dynamic elements; and the dynamic visual channel reflects the dynamic characteristics of energy consumption deviation, such as the periodic scaling of elements or changes in color intensity.

[0134] In one implementation, step S420 may include the following steps S421 to S426:

[0135] Step S421: Parse the preset visual performance parameter library, extract the color channel parameter range, transparency parameter range, dynamic change frequency parameter range, and spatial size scaling parameter range, and generate a visual parameter attribute set. The visual parameter attribute set contains the value range and data type of each parameter.

[0136] The pre-defined visual representation parameter library is a collection of various visual representation parameters. To effectively utilize these parameters for associative mapping, they need to be parsed. The color channel parameter range specifies the range of color values, such as the RGB value range, determining the types of colors that can be used in visualization. The transparency parameter range defines the transparency range of an element, from completely transparent to completely opaque; different levels of transparency can be used to represent different degrees of energy consumption deviation. The dynamic change frequency parameter range defines the frequency range of dynamic elements, such as the speed of flashing or the frequency of scaling, which can reflect the dynamic characteristics of energy consumption deviation. The spatial size scaling parameter range specifies the scaling ratio of elements in space, used to highlight or reduce the energy consumption deviation of different road sections.

[0137] Step S422: Based on the set of visual parameter attributes, construct a three-dimensional visual channel framework. The three-dimensional visual channel framework includes a spatial visual channel, a temporal visual channel, and a dynamic visual channel. Each channel corresponds to different types of visual performance parameters.

[0138] The spatial vision channel is primarily responsible for handling visual representations related to spatial location. In highway energy consumption visualization, it corresponds to the spatial display effects of different road sections, such as the color of the road section and the scaling of its spatial dimensions. Through the spatial vision channel, the visual presentation of different road sections in the 3D model can be adjusted according to their energy consumption deviations, allowing users to intuitively see the differences between different road sections.

[0139] The temporal visual channel focuses on time-related visual changes, allowing control over elements' transparency and the flashing frequency of dynamic elements. It displays the dynamic characteristics of energy consumption deviation over time, helping users understand how this deviation evolves. The dynamic visual channel showcases the dynamic features of energy consumption deviation, such as periodic scaling of elements and changes in color intensity. It enhances the dynamic effects of visualization, making energy consumption deviation information more vividly presented.

[0140] Each channel corresponds to different types of visual performance parameters. For example, the spatial visual channel may correspond to color channel parameters and spatial size scaling parameters, the temporal visual channel may correspond to transparency parameters and the time-related part of the dynamic change frequency parameters, and the dynamic visual channel may correspond to the overall dynamic change frequency parameters as well as other parameters related to dynamic effects.

[0141] Step S423: Perform channel mapping processing on the deviation spatiotemporal feature matrix, mapping the spatial difference contribution rate to the spatial visual channel, the temporal difference contribution rate to the temporal visual channel, and the comprehensive deviation quantification value to the dynamic visual channel to generate the initial correlation tensor.

[0142] Channel mapping processing maps these key information from the spatiotemporal feature matrix of deviation to different channels of the 3D vision channel framework. Mapping the spatial difference contribution rate to the spatial vision channel means adjusting the visual representation parameters corresponding to the spatial vision channel, such as the color or spatial size scaling of the road segment, according to the magnitude of the spatial difference contribution rate of different road segments. For example, road segments with a larger spatial difference contribution rate may correspond to more vibrant colors or larger spatial sizes to highlight their spatial energy consumption deviation.

[0143] By mapping the contribution rate of time difference to the time visual channel, and adjusting the visual performance parameters of the time visual channel based on changes in the contribution rate of time difference, such as the change in the transparency of elements over time or the flashing frequency of dynamic elements, the dynamic changes in energy consumption deviation over time can be displayed.

[0144] The overall deviation quantification is mapped to the dynamic visual channel. The larger the overall deviation quantification, the stronger the visual performance of the dynamic visual channel may be, such as a larger periodic scaling of elements or more obvious changes in color intensity.

[0145] This channel mapping process generates an initial correlation tensor. The initial correlation tensor is a preliminary correlation data structure that establishes a preliminary correspondence between the energy consumption deviation information in the deviation spatiotemporal feature matrix and the visual representation parameters in the 3D vision channel framework.

[0146] Step S424: Calculate the correlation strength of each channel parameter in the initial correlation tensor, and generate the channel correlation weight matrix through the covariance analysis between channels. The matrix element values ​​reflect the degree of mutual influence between different visual channel parameters.

[0147] While the initial correlation tensor establishes a preliminary association between the spatiotemporal feature matrix of deviation and the visual performance parameters, there may be mutual influences between parameters of different visual channels. To more accurately grasp these relationships, it is necessary to calculate the correlation strength of each channel parameter in the initial correlation tensor.

[0148] In this embodiment of the invention, the correlation strength between different visual channel parameters is calculated through inter-channel covariance analysis. Based on the results of the covariance analysis, a channel correlation weight matrix is ​​generated. The channel correlation weight matrix is ​​a two-dimensional matrix, and its element values ​​reflect the degree of mutual influence between different visual channel parameters. The rows and columns of the matrix correspond to different visual channel parameters, respectively. The larger the element value, the stronger the correlation between the two parameters and the greater the mutual influence; the smaller the element value, the weaker the correlation.

[0149] Step S425: Adjust the initial correlation tensor according to the channel correlation weight matrix to enhance the mapping strength of strong correlation parameter pairs and weaken the mapping strength of weak correlation parameter pairs, thereby generating an optimized correlation tensor.

[0150] The channel correlation weight matrix reflects the degree of mutual influence between parameters of different visual channels. For strongly correlated parameter pairs, i.e., parameter pairs with larger element values ​​in the channel correlation weight matrix, their mapping strength in the initial correlation tensor is enhanced. For example, if the color parameter in the spatial visual channel and the color intensity change parameter in the dynamic visual channel are strongly correlated, then when adjusting the initial correlation tensor, the mapping ratio between these two parameters is increased, so that they can change more closely in the visualization, enhancing the consistency of the display effect.

[0151] For weakly correlated parameter pairs—that is, parameter pairs with smaller element values ​​in the channel correlation weight matrix—their mapping strength in the initial correlation tensor is reduced. This avoids some unnecessary interference, making the visualization clearer and simpler.

[0152] By adjusting the initial correlation tensor in this way, an optimized correlation tensor is generated. While retaining the basic mapping relationship of the initial correlation tensor, the optimized correlation tensor further considers the correlation between parameters of different visual channels, resulting in a more coordinated and accurate visualization that better reflects energy consumption deviation information.

[0153] Step S426: Perform dimension normalization on the optimized correlation tensor so that the dimension of the spatial visual channel is consistent with the number of road segment spatial locations, the dimension of the temporal visual channel is consistent with the number of time sampling points, and the dimension of the dynamic visual channel is consistent with the feature dimension of the comprehensive deviation quantification value, thereby generating a standard deviation-visual parameter correlation tensor.

[0154] After obtaining the optimized correlation tensor, dimensionality normalization is required to ensure that it accurately matches the display requirements of the 3D model. Although the optimized correlation tensor has considered the correlation between parameters of different visual channels, its dimensions may not be consistent with the actual number of road segment spatial locations, the number of time sampling points, and the feature dimensions of the comprehensive deviation quantification value.

[0155] The dimension of the spatial visual channel needs to match the number of spatial locations of the road segments to ensure that each road segment has corresponding visual display parameters. If the dimension of the spatial visual channel is smaller than the number of spatial locations of the road segments, the energy consumption deviation information of some road segments will not be accurately displayed; if the dimension is larger than the number of spatial locations of the road segments, it will cause data redundancy.

[0156] The consistency between the dimensionality of the temporal visual channel and the number of time sampling points ensures accurate representation of energy consumption deviations at different time points. Each time sampling point should have corresponding visual representation parameters to reflect the dynamic changes in energy consumption deviation over time. The consistency between the dimensionality of the dynamic visual channel and the feature dimension of the comprehensive deviation quantification ensures that all features of the comprehensive deviation quantification are accurately reflected in the dynamic visual channel. By performing dimensionality normalization on the optimized correlation tensor, the dimensions of each channel are adjusted to match the corresponding actual number or feature dimensions. Finally, a standard deviation-visual parameter correlation tensor is generated.

[0157] Step S430: Based on the spatial distribution characteristics in the deviation-visual parameter correlation tensor, generate spatial visual parameter mapping rules. The spatial visual parameter mapping rules dynamically adjust the mapping ratio between color channel parameters and spatial size scaling parameters according to the gradient change of the spatial difference contribution rate.

[0158] The deviation-visual parameter correlation tensor contains correlation information from spatial, temporal, and dynamic visual channels. Its spatial distribution characteristics reflect the spatial differences in energy consumption deviation across different road segments. Based on these spatial distribution characteristics, spatial visual parameter mapping rules are generated to accurately convert the contribution rate of spatial differences into a visually appealing spatial representation.

[0159] In one implementation, step S430 may include the following steps S431 to S436:

[0160] Step S431: Extract spatial visual channel data from the deviation-visual parameter correlation tensor to generate spatial feature slices. The spatial feature slices contain the spatial difference contribution rate of each road segment's spatial location and the corresponding initial values ​​of the visual parameter mapping.

[0161] The deviation-visual parameter association tensor is a three-dimensional data structure containing association information of spatial visual channels, temporal visual channels, and dynamic visual channels. To generate spatial visual parameter mapping rules, spatial visual channel data must first be extracted from it.

[0162] Spatial visual channel data contains information related to spatial display, such as the spatial difference contribution rate of different road segments and the corresponding initial visual parameter mapping values. After extracting this data, spatial feature slices are generated. A spatial feature slice is a two-dimensional data structure, with each row corresponding to a road segment's spatial location, containing the spatial difference contribution rate of that road segment and the corresponding initial visual parameter mapping values.

[0163] Step S432: Perform spatial clustering on the spatial feature slices, and divide the spatial locations of road segments with similar spatial difference contribution rates into the same spatial cluster. Each spatial cluster contains multiple continuous road segment spatial units.

[0164] Spatial clustering is a method of grouping data points with similar characteristics in a dataset into one class. In this embodiment of the invention, road segments with similar spatial difference contribution rates are divided into the same spatial cluster.

[0165] Road segments with similar spatial contribution rates indicate that their spatial energy consumption deviations are relatively similar and may be affected by similar factors. Dividing these road segments into the same spatial cluster, each cluster contains multiple consecutive road segment spatial units. Consecutive road segment spatial units are geographically adjacent, and their energy consumption deviations may be correlated.

[0166] Spatial clustering can employ algorithms such as K-means clustering and DBSCAN clustering. Through spatial clustering, the spatial locations of road segments in the spatial feature slices are grouped to form multiple spatial clusters.

[0167] Step S433: Calculate the mean and variance of the spatial difference contribution rate for each spatial cluster, and generate the statistical features of the spatial clusters. The statistical features are used to guide the mapping range division of spatial visual parameters.

[0168] After spatial clustering, multiple spatial clusters were obtained, each containing the spatial locations of road segments with similar spatial difference contribution rates. To further understand the characteristics of each spatial cluster, it is necessary to calculate the mean and variance of its spatial difference contribution rate.

[0169] The mean spatial difference contribution rate is the average of the spatial difference contribution rates of all road segments within a spatial cluster, reflecting the overall level of spatial difference contribution of the cluster. Variance measures the dispersion of the spatial difference contribution rate within the spatial cluster; a larger variance indicates greater differences in the spatial difference contribution rates of different road segments within the cluster, while a smaller variance indicates that the spatial difference contribution rates of different road segments are relatively similar.

[0170] The calculated mean and variance are combined to generate spatial cluster statistical features. If a spatial cluster has a high mean and low variance in its spatial difference contribution rate, it indicates that the overall spatial difference contribution of road segments within the cluster is large and relatively consistent. When generating spatial visual parameter mapping rules, a relatively large and uniform visual parameter mapping range can be assigned to this spatial cluster, such as using darker colors and larger spatial size scaling ratios. Conversely, if a spatial cluster has a low mean and high variance in its spatial difference contribution rate, it indicates that the spatial difference contribution of road segments within the cluster is small but the differences are large. It may be necessary to assign it a relatively small but more variable visual parameter mapping range.

[0171] Step S434: Based on the statistical characteristics of spatial clusters, the range of values ​​for the spatial difference contribution rate is divided into multiple continuous sub-intervals, each sub-interval corresponding to a combination of color channel parameters and spatial size scaling parameters.

[0172] After obtaining the statistical characteristics of spatial clusters, these characteristics can be used to divide the range of values ​​for the spatial difference contribution rate. The range of values ​​for the spatial difference contribution rate is the interval between the minimum and maximum values ​​of the spatial difference contribution rate of all road segments.

[0173] Based on the statistical characteristics of spatial clusters, this range of values ​​is divided into multiple continuous sub-intervals. The division can be based on features such as the mean and variance of the spatial clusters. For example, for spatial clusters with a high mean and low variance, their corresponding spatial difference contribution rate range can be divided into a relatively narrow but higher-value sub-interval; for spatial clusters with a low mean and high variance, their corresponding spatial difference contribution rate range can be divided into a wider and lower-value sub-interval. Each sub-interval corresponds to a combination of color channel parameters and spatial scaling parameters. Color channel parameters determine the color display of the road segment in the 3D model; for example, different combinations of RGB values ​​can produce different colors. Spatial scaling parameters control the spatial size of the road segment in the 3D model.

[0174] Step S435: Establish a mapping table between the spatial difference contribution rate sub-intervals and the visual parameter combinations. The mapping table includes the upper and lower limits of the sub-intervals, color channel parameter values, and spatial size scaling parameter values.

[0175] After dividing the range of spatial difference contribution rate into multiple consecutive sub-intervals and determining the corresponding combination of color channel parameters and spatial size scaling parameters for each sub-interval, a mapping table needs to be established to record these correspondences.

[0176] The mapping table is a two-dimensional table, with each row corresponding to a sub-interval of spatial difference contribution rate. The columns of the table include the upper and lower limits of the sub-interval, the color channel parameter values, and the spatial size scaling parameter values. The upper and lower limits of the sub-interval clearly define the value range of that sub-interval, allowing the sub-interval to be determined based on the spatial difference contribution rate of a road segment.

[0177] The color channel parameter value represents the color display parameter corresponding to the sub-interval, such as RGB value or color code. The spatial scaling parameter value specifies the spatial scaling ratio of the road segment corresponding to the sub-interval in the 3D model.

[0178] Step S436: Smooth the mapping table to generate the final spatial visual parameter mapping rules.

[0179] Because the mapping table is divided into sub-intervals based on the contribution rate of spatial differences, abrupt changes in visual presentation may occur at the boundaries of these sub-intervals. For example, the color channel parameters and spatial scaling parameters corresponding to two adjacent sub-intervals may differ significantly. When the contribution rate of spatial differences of a road segment just crosses the boundary of a sub-interval, its visual presentation will change abruptly, which may affect the user's intuitive understanding of energy consumption deviation information.

[0180] The purpose of smoothing is to eliminate such abrupt changes, making the visual display smoother and more continuous across the entire range of spatial difference contribution rates. Various smoothing methods can be employed, such as linear interpolation and spline interpolation. By smoothing the mapping table, the final spatial visual parameter mapping rule is generated. This rule ensures that the corresponding visual display effect transitions smoothly when the spatial difference contribution rate changes across different road segments, improving the quality of the visualization and the user experience.

[0181] Step S440: Based on the time series features in the deviation-visual parameter correlation tensor, establish a time-visual parameter mapping rule. The time-visual parameter mapping rule dynamically adjusts the mapping relationship between the transparency parameter and the dynamic change frequency parameter according to the rate of change of the contribution rate of time difference.

[0182] The deviation-visual parameter correlation tensor contains not only information from the spatial visual channel but also the temporal series characteristics of the temporal visual channel. The temporal difference contribution rate reflects the temporal variation of energy consumption deviation, and its rate of change reflects the speed of this variation.

[0183] Based on the time-series features in the deviation-visual parameter correlation tensor, a time-visual parameter mapping rule is established to accurately convert the contribution rate and rate of change of time differences into a visual time display effect. Transparency and dynamic change frequency parameters are important visual parameters used to control the temporal display effect of elements.

[0184] The time-visual parameter mapping rule dynamically adjusts the mapping relationship between the transparency parameter and the dynamic change frequency parameter based on the rate of change of the contribution rate of time difference. When the rate of change of the contribution rate of time difference is large, it indicates that the energy consumption deviation changes rapidly over time. In this case, the change amplitude of the transparency parameter can be increased to make the transparency of the element change significantly in a short period of time. At the same time, the dynamic change frequency parameter can be increased to speed up the flashing or change frequency of dynamic elements, so as to highlight this rapid change feature.

[0185] Step S450: Integrate spatial visual parameter mapping rules and temporal visual parameter mapping rules to construct a deviation-visual parameter dynamic mapping model. The deviation-visual parameter dynamic mapping model includes a spatial mapping weight vector, a temporal mapping weight vector, and dynamic adjustment coefficients.

[0186] After generating spatial visual parameter mapping rules and temporal visual parameter mapping rules respectively, in order to integrate the energy consumption deviation information in the spatial and temporal dimensions for visualization, it is necessary to fuse these two rules and construct a deviation-visual parameter dynamic mapping model.

[0187] The spatial visual parameter mapping rule dynamically adjusts the mapping ratio between color channel parameters and spatial size scaling parameters based on the gradient change of the contribution rate of spatial differences, primarily focusing on displaying the energy consumption deviation of different road segments in space. The temporal visual parameter mapping rule dynamically adjusts the mapping relationship between transparency parameters and dynamic change frequency parameters based on the rate of change of the contribution rate of temporal differences, emphasizing the display of the evolution of energy consumption deviation over time.

[0188] The deviation-visual parameter dynamic mapping model includes a spatial mapping weight vector, a temporal mapping weight vector, and dynamic adjustment coefficients. The spatial mapping weight vector controls the weight of spatial visual parameters in the overall visualization, determining the degree of influence of spatial factors on the final visual effect. The temporal mapping weight vector controls the weight of temporal visual parameters, reflecting the importance of temporal factors in the visualization.

[0189] The dynamic adjustment coefficient is used to dynamically adjust the relative magnitudes of the spatial and temporal mapping weight vectors based on the actual energy consumption deviation. For example, when the spatial difference contribution rate changes significantly while the temporal difference contribution rate changes relatively small, the dynamic adjustment coefficient can increase the weight of the spatial mapping weight vector to make the spatial visual display effect more prominent; conversely, when the temporal difference contribution rate changes more significantly, the weight of the temporal mapping weight vector is increased to emphasize the display effect of the temporal dimension.

[0190] Step S460: The energy consumption deviation index set is processed by visual parameter encoding through the deviation-visual parameter dynamic mapping model to generate a visual encoding vector containing amplitude and frequency domain features, and the visual encoding vector is converted into encoding instructions that can drive the display of the three-dimensional model according to the three-dimensional model driving protocol.

[0191] After constructing the deviation-visual parameter dynamic mapping model, it is necessary to use this model to encode the visual parameters of the energy consumption deviation index set. The energy consumption deviation index set contains energy consumption deviation information under different correlation scales, such as spatial difference contribution rate, temporal difference contribution rate, and comprehensive deviation quantification value.

[0192] Visual parameter encoding processing transforms data from the energy consumption deviation index set into specific visual performance parameter values ​​using a deviation-visual parameter dynamic mapping model. The spatial mapping weight vector, temporal mapping weight vector, and dynamic adjustment coefficients determine how energy consumption deviation information in the spatial and temporal dimensions is combined with visual parameters. For example, based on the spatial mapping weight vector and spatial visual parameter mapping rules, the spatial difference contribution rate is converted into color channel parameters and spatial size scaling parameters; based on the temporal mapping weight vector and temporal visual parameter mapping rules, the temporal difference contribution rate is converted into transparency parameters and dynamic change frequency parameters.

[0193] This encoding process generates a visual encoded vector containing amplitude and frequency domain features. Amplitude features correspond to the intensity of visual representations, such as color depth and transparency; frequency domain features are related to the frequency of dynamic element changes, such as flashing speed and periodic scaling frequency. Finally, according to the 3D model-driven protocol, the visual encoded vector is converted into encoded instructions that can drive the display of the 3D model. The 3D model-driven protocol specifies the interface and conversion rules between the encoded vector and the 3D model display. By converting the visual encoded vector into encoded instructions, energy consumption deviation information can be accurately transmitted to the 3D model, driving the 3D model to display according to the encoded instructions, thus presenting abstract energy consumption deviation data to the user in an intuitive 3D visualization.

[0194] Step S500: Based on the visualization encoding vector, drive the three-dimensional road network model to perform energy consumption status visualization rendering operation, and present the energy consumption distribution characteristics and abnormal fluctuations by combining thermal layer overlay display with abnormal interval pulse warning.

[0195] In the 3D road network model, different colors are assigned to different road segments based on the amplitude characteristics in the visualized encoding vector. By overlaying heatmaps, users can intuitively see the energy consumption distribution of different highway segments and quickly identify which segments have more severe energy consumption deviations. Anomaly zone pulse warnings are used to highlight areas where energy consumption fluctuates abnormally. Abnormal fluctuations occur when energy consumption deviations exceed preset thresholds; these areas may indicate equipment malfunctions, abnormal traffic flow, or other issues. Pulse warning effects are set in abnormal zones based on the frequency domain characteristics in the visualized encoding vector. Pulse warnings can manifest as periodic scaling of elements, changes in color intensity, and other dynamic effects that attract user attention and quickly draw their focus to areas of abnormal energy consumption.

[0196] In one implementation, step S500 may include the following steps S510~S560:

[0197] Step S510: Call the preset three-dimensional road network model of the expressway. The three-dimensional road network model includes the three-dimensional geometric model and spatial coordinate information of geographical elements such as expressway sections, bridges, tunnels and service areas.

[0198] The pre-built 3D highway network model is a pre-constructed 3D model used to simulate the actual geographical layout of highways. This model encompasses several key geographical elements of highways, including highway sections, bridges, tunnels, and service areas. The 3D geometric model accurately describes the shape, size, and structure of these geographical elements, enabling the model to realistically represent the actual appearance of highways. Spatial coordinate information determines the specific location of each geographical element in 3D space. These coordinates allow for accurate association of energy consumption data with the corresponding geographical elements.

[0199] When the model is invoked, the system reads relevant data from the database storing the model and loads it into the visualization environment. For example, for a highway segment, the model may contain geometric information such as the segment's length, width, and slope, as well as its specific coordinates in the Earth coordinate system. The 3D geometric models of bridges and tunnels will describe their unique structural features in detail, such as the span of bridges and the location of piers, and the length and cross-sectional shape of tunnels. Service area models will display information such as their building layout and parking lot locations.

[0200] Step S520: Analyze the visualization encoding vector and extract color channel parameters, transparency parameters, dynamic change frequency parameters, spatial size scaling parameters, and spatial position encoding information.

[0201] Color channel parameters determine the colors displayed for different road segments or areas in the 3D road network model. Different colors can represent different levels of energy consumption deviation; for example, high energy consumption deviation might correspond to red, and low energy consumption deviation to green. Transparency parameters control the transparency of displayed elements. Adjusting transparency allows different levels of information to be overlaid, enhancing the visual hierarchy. Dynamic change frequency parameters control the rate of change of dynamic elements, such as the flashing frequency of pulse warning effects in abnormal zones. Spatial size scaling parameters allow for size adjustments to elements in the 3D model, while spatial location encoding information specifies the exact location within the 3D road network model corresponding to each energy consumption data point. This information allows for accurate mapping of energy consumption data to the corresponding geographic features in the model.

[0202] During the parsing process, these parameters are extracted from the visual encoding vector according to predefined encoding rules. For example, if the encoding vector adopts a set format, different types of parameters are arranged in a certain order, and color channel parameters, transparency parameters, etc. are extracted in sequence according to this order.

[0203] Step S530: Locate the corresponding road segment in the 3D road network model based on the spatial location coding information, and generate the location coordinates of the energy consumption visualization area.

[0204] The spatial location encoding information has been extracted from the visualization encoding vector in the previous steps, providing a corresponding location identifier in the 3D road network model for each energy consumption data point. Based on this encoding information, a positioning operation is performed in the 3D road network model to find the specific road segment location corresponding to each energy consumption data point.

[0205] In a 3D road network model, each geographic feature has its precise spatial coordinates. By matching the spatial location encoding information with the coordinate information in the model, the specific location of the road segment corresponding to the energy consumption data in the model can be determined.

[0206] Step S540: Based on the color channel parameters and transparency parameters, a basic thermal layer is generated at the coordinates of the energy consumption visualization area. The color distribution of the basic thermal layer reflects the spatial distribution characteristics of the energy consumption deviation.

[0207] After determining the coordinates of the energy consumption visualization area, the basic heatmap layer is generated by combining the previously extracted color channel parameters and transparency parameters. Color channel parameters and transparency parameters are key factors controlling the display effect of the heatmap.

[0208] The color channel parameters assign corresponding colors to each energy consumption visualization area based on the different levels of energy consumption deviation. For example, through pre-defined mapping rules, areas with high energy consumption deviation are associated with red, areas with medium energy consumption deviation with yellow, and areas with low energy consumption deviation with green. In this way, the color distribution can intuitively reflect the spatial distribution characteristics of energy consumption deviation. Users can quickly understand the energy consumption of different sections of the highway by observing the color changes.

[0209] The transparency parameter is used to adjust the display transparency of the heatmap. Appropriate transparency settings allow the heatmap to blend seamlessly with other elements in the 3D road network model (such as roads and bridges), avoiding obstructions or jarring appearances. For example, setting the transparency to a moderate value allows users to clearly see the overall structure of the 3D road network model while viewing the heatmap, enhancing the visualization effect and readability.

[0210] Step S550: Based on the dynamic change frequency parameter and spatial size scaling parameter, overlay an abnormal interval pulse warning effect on the basic thermal layer. The abnormal interval pulse warning effect is achieved by periodic scaling of visual elements and color intensity changes.

[0211] The dynamically changing frequency parameter controls the rate at which the pulse warning effect changes in the abnormal range. By adjusting this parameter, the flashing frequency of the pulse can be set. For example, when the energy consumption deviation is very high, a higher flashing frequency can be set to attract the user's attention; when the energy consumption deviation is relatively low but still exceeds the normal range, the flashing frequency can be appropriately reduced.

[0212] In one implementation, step S550 may specifically include the following steps S551 to S556:

[0213] Step S551: Analyze the energy consumption deviation index set, identify abnormal intervals where the energy consumption deviation exceeds the preset threshold, and generate a set of abnormal interval location coordinates and corresponding deviation index values.

[0214] During the analysis process, the system iterates through each indicator value in the energy consumption deviation indicator set and compares it with a preset threshold. If an indicator value exceeds the threshold, the area corresponding to that indicator is determined to be an abnormal interval. Simultaneously, the system records the location coordinates of this abnormal interval in the 3D road network model, forming a set of abnormal interval location coordinates. Furthermore, it records the specific deviation indicator value corresponding to this abnormal interval; this value can be used to subsequently determine the specific parameters for the pulse warning effect, such as the pulse intensity and the degree of color change.

[0215] Step S552: For each coordinate point in the set of abnormal interval location coordinates, extract the dynamic change frequency parameter and spatial size scaling parameter from the visualization encoding vector according to the corresponding deviation index value.

[0216] For each coordinate point in the set of coordinates of the abnormal interval, the system extracts the corresponding dynamic frequency parameter and spatial scaling parameter from the visualization encoding vector based on its corresponding deviation index value. A larger deviation index value indicates a more severe energy consumption anomaly in that abnormal interval, thus requiring a stronger pulse warning effect. For example, when the deviation index value is high, the dynamic frequency parameter extracted from the visualization encoding vector will correspond to a higher flashing frequency, and the spatial scaling parameter will correspond to a larger magnification factor to highlight the abnormal interval. During the extraction process, the system matches the deviation index value with the parameters in the visualization encoding vector according to predefined mapping rules. If the mapping rules specify a concrete functional relationship between the deviation index value and the dynamic frequency parameter and spatial scaling parameter, the system will perform accurate extraction based on this relationship.

[0217] Step S553: ​​Calculate the pulse period based on the dynamically changing frequency parameter. The pulse period is the reciprocal of the dynamically changing frequency parameter.

[0218] Step S554: Determine the pulse amplitude based on the spatial size scaling parameter. The pulse amplitude corresponds to the spatial size scaling parameter.

[0219] The spatial scaling parameter controls the visual size change of the abnormal region, while the pulse amplitude is a parameter describing the scaling of visual elements in the pulse warning effect. There is a correspondence between the pulse amplitude and the spatial scaling parameter, which can be linear or non-linear, depending on the predefined mapping rules.

[0220] After obtaining the spatial scaling parameter corresponding to each abnormal interval, the corresponding pulse amplitude is determined based on this correspondence. For example, if the spatial scaling parameter is large, it means that the abnormal interval needs to be displayed more prominently, so the corresponding pulse amplitude will also be large, that is, the scaling of the visual element during the pulse process will be greater. Conversely, if the spatial scaling parameter is small, the pulse amplitude will also be reduced accordingly.

[0221] Step S555: Construct a pulse warning animation model. The pulse warning animation model realizes the periodic scaling of visual elements. The scaling amplitude is controlled by the pulse amplitude, and the scaling frequency is controlled by the pulse period.

[0222] After determining the pulse period and pulse amplitude for each abnormal interval, a pulse warning animation model needs to be constructed to achieve the pulse warning effect for the abnormal interval. The pulse warning animation model is a model used to control the dynamic changes of visual elements, capable of periodically scaling the visual elements according to preset parameters.

[0223] When constructing a pulse warning animation model, corresponding animation keyframes are set according to the pulse period and pulse amplitude. For example, within one pulse period, multiple keyframes are set, each keyframe corresponding to a size of the visual element. By performing interpolation calculations between these keyframes, a smooth scaling effect of the visual element is achieved.

[0224] Step S556: Apply a pulse warning animation model to the abnormal area location of the basic heat map to generate an abnormal area warning sign with dynamic pulse effect, and adjust the color intensity of the warning sign to make it consistent with the color distribution of the basic heat map.

[0225] After constructing the pulse warning animation model, it is applied to the abnormal interval locations on the basic heatmap layer to generate abnormal interval warning signs with dynamic pulse effects. First, the system accurately locates the position of each abnormal interval on the basic heatmap layer based on the previously determined set of abnormal interval location coordinates. Then, the corresponding pulse warning animation model is applied to the visual elements at these locations, causing these elements to periodically scale according to a preset pulse period and pulse amplitude, thereby creating a dynamic pulse warning effect.

[0226] Step S560: Simultaneously render the basic thermal layer and the abnormal interval pulse warning effect to generate a 3D visualization scene that presents the energy consumption distribution characteristics and abnormal fluctuations, and refresh the 3D visualization scene according to the real-time energy consumption data update frequency.

[0227] During the rendering process, the basic heatmap layer and visual elements of the abnormal zones are reasonably combined and displayed according to the rendering algorithm. For example, it ensures that the pulse warning effect of the abnormal zone is not completely obscured by the basic heatmap layer, while also coordinating with the color and transparency of the basic heatmap layer. Through synchronous rendering, the generated 3D visualization scene can intuitively and clearly display the energy consumption status of the highway, enabling users to quickly understand the distribution characteristics and abnormal fluctuations of energy consumption.

[0228] It is understood that the various algorithms involved in the above descriptions of the embodiments of the present invention can all be obtained from relevant content in the prior art. To save space, they will not be elaborated on in the embodiments of the present invention. In addition, those skilled in the art can supplement the details based on common knowledge in the art when implementing the solutions of the present invention. For example, based on common knowledge in the art, normalization can be used to eliminate dimensional conflicts before feature fusion (e.g., normalizing the coupling difference statistics in the difference feature blocks to convert the spatial correlation strength deviation and time evolution rate deviation into dimensionless standardized statistics), interpolation can be used to eliminate dimensional differences, thresholds can be reasonably set based on historical data, experience or business scenario requirements, the model can be trained based on a general model training method, the number of layers in the model structure can be set based on actual needs, the activation function can be selected, etc. The present invention will not provide redundant descriptions of the implementation process in excessive detail.

[0229] Figure 2 A hardware entity diagram of a computer system provided as an embodiment of the present invention, such as... Figure 2 As shown, the hardware entity of the computer system 1000 includes a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can run on the processor 1001, and the processor 1001 executes the program to implement the steps in the method of any of the above embodiments.

Claims

1. A method for visualizing energy consumption monitoring data based on highways, characterized in that, The method includes: By establishing a unified time reference system and spatial coordinate system, the original monitoring records collected by various types of monitoring equipment deployed along the highway are integrated to obtain an integrated energy consumption monitoring data with spatiotemporal correlation. By analyzing the correlation between traffic flow change patterns and energy consumption patterns in the energy consumption monitoring data integration, a mapping relationship is established to generate road segment energy consumption benchmark indicators. The energy consumption benchmark index of the road segment and the real-time energy consumption measurement records are weighted and compared at different time granularities to calculate the multi-scale deviation, resulting in a set of energy consumption deviation indexes covering different time granularities. Specifically, this includes: performing spatiotemporal coupling feature decomposition on the energy consumption benchmark index of the road segment to extract spatial correlation parameters and temporal dynamic evolution features; generating a benchmark coupling feature matrix through a neighboring road segment energy consumption transmission model, where the row dimension of the benchmark coupling feature matrix corresponds to the spatial unit of the road segment, the column dimension corresponds to the temporal evolution stage, and the matrix elements include spatial correlation strength and temporal evolution rate; performing the same spatiotemporal coupling feature decomposition process as the energy consumption benchmark index on the energy consumption measurement records to obtain a measured coupling feature matrix, where the spatial unit division and temporal evolution stage division of the measured coupling feature matrix are consistent with the benchmark coupling feature matrix; calculating the coupling difference tensor between the benchmark coupling feature matrix and the measured coupling feature matrix, and using the spatial correlation strength deviation and... The joint modeling of temporal evolution rate deviation generates a coupled difference data structure containing spatial difference fields and temporal difference sequences. Based on the spatial correlation strength deviation in the coupled difference tensor, a dynamic correlation comparison window is constructed. The dynamic correlation comparison window dynamically adjusts its spatial coverage and temporal duration according to the energy consumption transmission coefficient between road segments. For road segments whose transmission coefficient exceeds a preset transmission threshold, the spatial coverage and temporal duration of the corresponding dynamic correlation comparison window are reduced by a preset ratio. The coupled difference tensor is segmented according to the dynamic correlation comparison window to generate difference feature blocks corresponding to different correlation scales. Each difference feature block contains the coupling difference statistics of spatial sub-regions and temporal sub-stages. The difference feature blocks are nonlinearly correlated and aggregated through a cross-scale feature fusion network to generate a set of energy consumption deviation indicators covering the dynamic correlation scale. Each indicator in the set of energy consumption deviation indicators contains a spatial difference contribution rate, a temporal difference contribution rate, and a comprehensive deviation quantification value. By establishing a mapping strategy between deviation indicators and visual performance parameters, the energy consumption deviation indicator set is converted into a visual encoding vector containing amplitude and frequency domain features, generating encoding instructions that can drive the display of 3D models. Based on the aforementioned visualization encoding vector, the three-dimensional road network model is driven to perform energy consumption status visualization rendering operations. The energy consumption distribution characteristics and abnormal fluctuations are presented by combining thermal layer overlay display with abnormal interval pulse warning.

2. The method according to claim 1, characterized in that, The process involves establishing a mapping relationship between traffic flow change patterns and energy consumption patterns within the integrated energy consumption monitoring data, and generating benchmark indicators for road segment energy consumption, including: The energy consumption monitoring data collection is subjected to time series segmentation processing to obtain multiple time window units with periodic variation characteristics. Each time window unit contains traffic flow data sequence and energy consumption data sequence within the corresponding time period. Pattern recognition processing is performed on the traffic flow data sequence in each time window unit to extract traffic flow change pattern features, which include the distribution of peak traffic times, duration of the traffic rise phase, and attenuation rate of the traffic fall phase. The energy consumption data sequence in each time window unit is subjected to trend fitting processing to generate an energy consumption trend curve, and energy consumption pattern features are extracted based on the energy consumption trend curve. The energy consumption pattern features include the time distribution characteristics of energy consumption peaks and valleys and the fluctuation amplitude parameter of energy consumption change rate. Calculate the mutual information entropy value between the traffic flow change pattern characteristics and the energy consumption pattern characteristics, and construct a correlation evaluation matrix; Based on the correlation evaluation matrix, the influence weight coefficients of traffic flow change pattern characteristics on energy consumption pattern characteristics are determined, and a traffic flow-energy consumption correlation mapping model is established. The energy consumption simulation calculation is performed on the historical traffic flow data of the same period by the traffic flow-energy consumption correlation mapping model to generate the road segment energy consumption benchmark index. The road segment energy consumption benchmark index includes the energy consumption reference range and the change trend prediction curve under different traffic flow levels.

3. The method according to claim 2, characterized in that, The step of performing pattern recognition processing on the traffic flow data sequence in each time window unit to extract traffic flow change pattern features includes: The traffic flow data sequence is subjected to multi-resolution decomposition processing, which decomposes the original traffic flow signal into multi-level signal components containing different fluctuation periods, generating a multi-scale traffic flow signal set. The frequency range of the multi-level signal components decreases exponentially as the decomposition level increases. The highest-level component is selected from the multi-scale flow signal set for trend extraction processing. The principal component of flow trend reflecting the macro-change law is generated by the neighborhood data weighted average method, and the remaining components are merged into the fluctuation residual component characterizing the local disturbance characteristics. Calculate the energy proportion parameters of each level of signal components, and construct the frequency domain feature spectrum of the flow signal. The horizontal axis of the frequency domain feature spectrum corresponds to the decomposition level, and the vertical axis corresponds to the cumulative distribution value of the energy proportion parameters. Traffic flow data sequences of adjacent road segments within the same time window are collected. The synchronicity coefficient of traffic flow changes between road segments is calculated through cross-covariance analysis, and a road segment coupling feature vector containing the spatial correlation strength is generated. The fluctuation residual components are subjected to pattern clustering processing. Residual segments with the same fluctuation pattern are classified by similarity measurement to generate a set of fluctuation pattern labels containing cluster center vectors. By integrating the frequency domain feature spectrum, road segment coupling feature vector, and fluctuation pattern label set, a three-dimensional traffic flow change pattern feature tensor is constructed. The first dimension of the feature tensor corresponds to the time window unit number, the second dimension corresponds to the signal decomposition level, and the third dimension corresponds to the spatial correlation parameter.

4. The method according to claim 1, characterized in that, The measured energy consumption records are subjected to the same spatiotemporal coupling feature decomposition process as the road section energy consumption benchmark index to obtain the measured coupling feature matrix, including: The spatial distribution characteristics of the measured energy consumption records are analyzed, and the locations of monitoring equipment, road segment topology, and communication delay parameters between equipment are extracted to generate a set of measured spatial characteristic parameters. This set of measured spatial characteristic parameters is used to guide the division of spatial units. Based on the measured spatial characteristic parameter set, the spatial unit division scheme of the benchmark coupling feature matrix is ​​copied, including the number of road segment segments, the neighborhood association radius and the spatial grid resolution, to generate the measured spatial framework. Spatial interpolation processing is performed on the measured energy consumption records to convert discrete monitoring point data into continuously spatially distributed energy consumption surface data. The spatial resolution of the energy consumption surface data is consistent with the grid resolution of the measured spatial frame. Calculate the energy consumption surface data gradient of adjacent spatial units, and generate an initial value of spatial correlation strength through spatial covariance analysis within a sliding window. The initial value of spatial correlation strength reflects the spatial coupling degree of energy consumption of adjacent road segments. By combining time series analysis to extract the time evolution rate of energy consumption surface data, the rate parameter is calculated by the ratio of energy consumption change to time interval between adjacent time steps, and a time evolution characteristic sequence is generated. By integrating the initial value of spatial correlation strength and the temporal evolution feature sequence, a measured coupling feature matrix is ​​generated that is completely aligned with the reference coupling feature matrix. Each element of the measured coupling feature matrix contains the measured spatial correlation strength and the measured temporal evolution rate.

5. The method according to claim 4, characterized in that, The calculation of the coupling difference tensor between the reference coupling feature matrix and the measured coupling feature matrix, through joint modeling of spatial correlation strength bias and temporal evolution rate bias, generates a coupling difference data structure containing spatial difference fields and temporal difference sequences, including: Construct a coupled feature alignment index, wherein the row dimension of the coupled feature alignment index corresponds to the spatial unit number, the column dimension corresponds to the time evolution stage, and each index item includes the baseline spatial correlation strength, the baseline time evolution rate, the measured spatial correlation strength, and the measured time evolution rate. The coupling feature comparison index is traversed, and the difference between the baseline spatial correlation strength and the measured spatial correlation strength in each index entry is calculated to generate a spatial correlation strength deviation. The spatial correlation strength deviation retains the positive or negative sign to indicate the strengthening or weakening trend. Perform the same difference calculation on the reference time evolution rate and the measured time evolution rate in the index item to generate the time evolution rate deviation. When the reference time evolution rate is zero, set the time evolution rate deviation to the absolute value of the measured time evolution rate. A spatial difference field is constructed based on the spatial correlation strength deviation, and discrete spatial deviation values ​​are converted into continuously distributed spatial difference surfaces. The contour lines of the spatial difference surfaces represent the spatial gradient changes of the deviation. The time evolution rate deviations are sorted according to the time series to generate a time difference sequence, which includes the rate deviation value of each time evolution stage and the deviation change rate of adjacent stages. By fusing the spatial difference field and the temporal difference sequence, a three-dimensional coupled difference tensor is constructed. The first dimension of the three-dimensional coupled difference tensor is the spatial unit, the second dimension is the temporal stage, and the third dimension is the deviation type.

6. The method according to claim 5, characterized in that, The step of performing nonlinear correlation aggregation processing on the differential feature blocks through a cross-scale feature fusion network to generate a set of energy consumption deviation indicators covering dynamic correlation scales includes: The association weight vector corresponding to each differential feature block is extracted from the cross-scale feature fusion network. The association weight vector is determined jointly based on the information entropy and mutual information value of the feature block. The weight coefficient of the association weight vector corresponding to the feature block whose information entropy exceeds a preset entropy threshold and whose mutual information value exceeds a preset mutual information threshold is increased according to a preset increase rule. The coupled difference statistics of each difference feature block are multiplied element-wise with the corresponding associated weight vector to generate a weighted difference feature block. The value of each statistic in the weighted difference feature block is the product of the original statistic value and the weight coefficient. The weighted difference feature block is subjected to cross-scale feature transfer processing. The time granularity fluctuation difference information corresponding to the basic correlation scale of the basic correlation scale feature block is transferred to the fusion correlation scale feature block through the feature mapping function, and a fusion correlation scale feature block with supplemented detail difference information is generated. The supplemented detail difference information includes the time granularity fluctuation frequency and amplitude features corresponding to the basic correlation scale. Based on the fusion-related scale feature blocks with the supplementary detail difference information, the difference contribution of each related scale feature block is calculated, the contribution ratio of each feature block to the overall deviation is assigned, and a contribution allocation vector containing the contribution of the time granularity fluctuation difference contribution item corresponding to the basic related scale is generated. The contribution allocation vector and the difference statistics of the corresponding feature blocks are weighted and summed. During the summation process, the time granularity fluctuation difference information corresponding to the basic correlation scale is integrated and supplemented to generate a comprehensive deviation quantification value that reflects both the long-term trend of the integrated correlation scale and the time granularity fluctuation characteristics corresponding to the basic correlation scale. By integrating the comprehensive deviation quantification value, spatial difference contribution rate, temporal difference contribution rate, and corresponding correlation scale identifier, an energy consumption deviation index set covering dynamic correlation scale is generated. The indicators in the energy consumption deviation index set are sorted from smallest to largest according to the correlation scale.

7. The method according to claim 1, characterized in that, The method involves establishing a mapping strategy between deviation indices and visual performance parameters, converting the energy consumption deviation index set into a visual encoding vector containing amplitude and frequency domain features, and generating encoding instructions that can drive the display of 3D models, including: The set of energy consumption deviation indicators is subjected to spatiotemporal feature fusion processing to extract the spatial distribution features of spatial difference contribution rate and the time series features of temporal difference contribution rate. A deviation spatiotemporal feature matrix is ​​generated through spatiotemporal correlation modeling. The row dimension of the deviation spatiotemporal feature matrix corresponds to the spatial location of the road segment, the column dimension corresponds to the time sampling point, and the matrix elements contain the comprehensive deviation quantification value. Construct a deviation-visual parameter correlation tensor, and map the deviation spatiotemporal feature matrix with a preset visual performance parameter library to generate a three-dimensional correlation data structure containing spatial visual channels, temporal visual channels and dynamic visual channels. Based on the spatial distribution characteristics in the deviation-visual parameter correlation tensor, a spatial visual parameter mapping rule is generated. The spatial visual parameter mapping rule dynamically adjusts the mapping ratio between color channel parameters and spatial size scaling parameters according to the gradient change of the spatial difference contribution rate. Based on the time series features in the deviation-visual parameter correlation tensor, a time-visual parameter mapping rule is established. The time-visual parameter mapping rule dynamically adjusts the mapping relationship between the transparency parameter and the dynamic change frequency parameter according to the rate of change of the contribution rate of time difference. By integrating the spatial visual parameter mapping rules and the temporal visual parameter mapping rules, a deviation-visual parameter dynamic mapping model is constructed. The deviation-visual parameter dynamic mapping model includes a spatial mapping weight vector, a temporal mapping weight vector, and a dynamic adjustment coefficient. The energy consumption deviation index set is processed by visual parameter encoding through the deviation-visual parameter dynamic mapping model to generate a visual encoding vector containing amplitude and frequency domain features. The visual encoding vector is then converted into encoding instructions that can drive the display of a 3D model according to the 3D model driving protocol.

8. The method according to claim 7, characterized in that, The construction of the deviation-visual parameter correlation tensor involves mapping the deviation spatiotemporal feature matrix to a preset visual performance parameter library to generate a three-dimensional correlation data structure containing spatial visual channels, temporal visual channels, and dynamic visual channels, including: The preset visual performance parameter library is analyzed, and the range of color channel parameters, transparency parameters, dynamic change frequency parameters, and spatial size scaling parameters are extracted to generate a set of visual parameter attributes. The set of visual parameter attributes includes the value range and data type of each parameter. Based on the set of visual parameter attributes, a three-dimensional visual channel framework is constructed. The three-dimensional visual channel framework includes a spatial visual channel, a temporal visual channel, and a dynamic visual channel. Each channel corresponds to different types of visual performance parameters. The deviation spatiotemporal feature matrix is ​​processed by channel mapping, which maps the spatial difference contribution rate to the spatial visual channel, the temporal difference contribution rate to the temporal visual channel, and the comprehensive deviation quantification value to the dynamic visual channel to generate an initial correlation tensor. The correlation strength of each channel parameter in the initial correlation tensor is calculated, and a channel correlation weight matrix is ​​generated through covariance analysis between channels. The element values ​​of the channel correlation weight matrix reflect the degree of mutual influence between different visual channel parameters. The initial correlation tensor is adjusted based on the channel correlation weight matrix to generate an optimized correlation tensor; The optimized correlation tensor is subjected to dimension normalization processing so that the dimension of the spatial visual channel is consistent with the number of road segment spatial locations, the dimension of the temporal visual channel is consistent with the number of time sampling points, and the dimension of the dynamic visual channel is consistent with the feature dimension of the comprehensive deviation quantification value, thereby generating a standard deviation-visual parameter correlation tensor.

9. A computer system comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

Images