A data processing method and system for comprehensive energy intelligent management

Abstract

The application provides a kind of data processing method and system for integrated energy wisdom management, it is related to data processing technical field, the application is unified to the operation data of electric power, heat and gas equipment Collection, time alignment and identification processing, construct multi-source heterogeneous time series data set;Based on sliding time window and difference calculation, the mean, first-order change rate and second-order acceleration are extracted, the data evolution stability index is calculated, and the evaluation is triggered, adaptive reinforcement sampling and priority scheduling strategy;Further based on adaptive window index and window adjustment coefficient calculation adaptive time window length, after evaluation, window constraint and multi-scale reconstruction are carried out;By calculating the data consistency index and comparing with the data reliability threshold, the consistency of multi-source data is identified and optimized.The method can realize the dynamic monitoring, stability evaluation and consistency optimization of the operation state of multi-energy complementary energy equipment, and provide high-precision and reliable data support for integrated energy scheduling and control.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a data processing method and system for integrated energy intelligent management. Background Technology

[0002] With the widespread application of integrated energy systems in multi-energy complementary scenarios at the park level, various energy systems, including electricity, heat, and gas, are gradually achieving coordinated operation. Their operational status is highly dependent on the real-time acquisition and fusion processing of multi-source data. However, in practical applications, different energy subsystems exhibit significant differences in data acquisition methods, sampling frequencies, time scales, and data accuracy, leading to considerable challenges in time alignment and unified modeling of multi-source heterogeneous time-series data. On the one hand, traditional data processing methods often employ fixed time windows or single-scale analysis, which are difficult to adapt to the dynamic changes that frequently occur during energy system operation, such as load fluctuations, equipment start-ups and shutdowns, and sudden disturbances, easily resulting in the loss of key abnormal information or the accumulation of data redundancy. On the other hand, multi-source data is susceptible to sensor errors, communication delays, and data gaps during synchronization, leading to inconsistencies or even conflicts in the same physical quantity between different devices, thereby reducing the quality of data fusion and affecting the accuracy of subsequent optimization scheduling and control decisions.

[0003] While some existing technologies process data by introducing sliding windows or statistical feature analysis, they generally lack adaptive time window adjustment mechanisms based on data evolution characteristics, making it difficult to achieve dynamic balance in scenarios with high and low volatility. Furthermore, for the issue of consistency among multi-source data, most methods rely solely on simple threshold filtering or single data source selection strategies, failing to fully consider differences in data reliability and the collaborative relationships among multiple sources. This results in low data utilization efficiency and makes it difficult to establish a stable and reliable data support system. Summary of the Invention

[0004] The purpose of this invention is to provide a data processing method and system for integrated energy intelligent management to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A data processing method for integrated smart energy management includes the following steps: Step 1: Collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations within the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset; Step 2: Perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter, and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. Step 3: Calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy will be given. Step 4: Calculate the adaptive time window length and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, apply adaptive window constraints and multi-scale reconstruction strategies. Step 5: Calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If unqualified, consistency optimization and conflict resolution strategies are given.

[0006] Further, step one includes: S11. In the multi-energy complementary energy stations at the park level under the comprehensive energy group, the operating status of power equipment is monitored in real time. Power quality analyzers, smart meters, voltage transformers and current transformers are used to measure the grid connection nodes and feeder circuits, and power equipment data are collected, including voltage data, current data, active power data and grid frequency data. S12. In the district heating and heat pump coupling system of the integrated energy park, the operating status of the thermal equipment is monitored in real time. The supply and return water pipe network and heat exchange unit are detected by platinum resistance temperature sensor, ultrasonic flow meter and pressure transmitter, and thermal equipment data are collected, including: supply water temperature data, return water temperature data, flow data and pipeline pressure data. S13. In the natural gas distributed energy station and biomass fuel energy supply device in the integrated energy park, the operating status of the gas and combustion system is monitored in real time. The gas transmission and distribution pipeline and combustion emission process are detected by gas turbine flow meter, gas pressure sensor and flue gas analyzer. Data of gas equipment are collected, including gas flow data, gas pressure data, flue gas temperature data and flue gas component concentration data. S14. Perform unified identification and time alignment processing on power equipment data, thermal equipment data and gas equipment data. Obtain unified timestamp data through time synchronization method, and generate equipment identification data ID and physical quantity type label in combination with equipment coding rules, and construct a unified multi-source heterogeneous time series dataset.

[0007] Furthermore, step two includes: S21. Based on a unified multi-source heterogeneous time-series dataset, real-time monitoring is performed on the active power data, voltage data, and current data of power equipment, the supply and return water temperature data, flow data, and heat exchange efficiency data of thermal equipment, and the pipeline pressure data, instantaneous flow data, and valve opening data of gas equipment. A unified timestamp alignment method is used to synchronously calibrate the original operating data of each energy device at each sampling time, and obtain the current value of each energy parameter at time t. S22. Based on the current time value, the sliding time window statistical method is used to calculate the mean of the continuous historical data within the preset time window interval before the current time t, and obtain the average value within the corresponding time window. S23. Based on the current time value, the adjacent sampling time difference calculation method is used to perform difference calculation between the current time value and the previous time value to obtain the first-order rate of change of each energy parameter. S24. Based on the first-order rate of change, a second-order difference calculation method is further adopted to perform difference operations on the first-order rate of change at adjacent times to obtain the second-order acceleration of each energy parameter.

[0008] Furthermore, step three includes: S31. By obtaining the current time value and the average value of the time window. The first-order rate of change and the second-order acceleration of change of each energy parameter are dimensionlessly processed using the extreme value normalization method, and then the data evolution stability index is calculated using a weighted fusion algorithm.

[0009] Furthermore, step three also includes: S32. By setting a preset data evolution stability threshold and comparing the data evolution stability index with the data evolution stability threshold, the first evaluation result is obtained, including: When the data evolution stability index is less than or equal to the data evolution stability threshold, it indicates that the current energy data evolution state is stable and should be continuously monitored. When the data evolution stability index exceeds the data evolution stability threshold, it indicates that the current energy data evolution state is unstable, posing a risk of abnormal load fluctuations or sudden equipment operation changes. This triggers the first early warning instruction and generates the first strategy: An adaptive sampling control method is used to dynamically adjust the raw data stream, shortening the data analysis time window by 20%–50%. Furthermore, by increasing the sampling frequency by 30%–80% and introducing a local feature enhancement algorithm, the changing features within key time intervals are extracted to enhance the accuracy of identifying abnormal fluctuations. The corresponding energy equipment operation data and its associated time segments are then marked with fluctuation status. A key data tracking and processing mechanism and a priority scheduling mechanism are initiated, and the marked data is passed as priority input to step four to drive the generation of the adaptive time window and refined time alignment processing.

[0010] Furthermore, step four includes: S41. Based on the fluctuation state marker data, the marked energy equipment operation data and its associated time segments are segmented and extracted. The sliding time window statistical analysis method is used to calculate the dispersion of the collected data within the window and obtain the local standard deviation. The extreme value statistical method is used to extract the maximum and minimum values ​​of the collected data within the same time window, and the difference is calculated to obtain the data change amplitude. The time series change direction determination method is used to compare the data change trends of adjacent sampling points within the window point by point, and the number of times the change direction changes per unit time is counted to obtain the data change frequency. S42. The obtained local standard deviation, data variation amplitude and data variation frequency are processed into dimensionless values ​​using the extreme value normalization method, and then the weighted fusion algorithm is used to calculate and obtain the adaptive window index. S43. Based on the data evolution stability index, extract historical stability statistics within the corresponding time window, use historical data regression analysis and proportional mapping calculation methods to perform scale mapping processing on the data evolution stability index within the preset reference interval, and combine the correlation between the time window length and data stability in historical operating data to perform fitting calculations and obtain the window adjustment coefficient. S44. After dimensionless processing of the obtained adaptive window index and window adjustment coefficient, the length of the adaptive time window is calculated using the inverse proportional mapping calculation method.

[0011] Furthermore, step four also includes: S45. By using a preset window threshold range and comparing the adaptive time window length with the window threshold range, a second evaluation result is obtained, including: When the adaptive time window length is within the window threshold range, it indicates that the current data is in a reasonable fluctuation state and should be continuously monitored. When the adaptive time window length When the window threshold range is reached, it indicates that the current data is not in a reasonable fluctuation state and there is a risk of sudden fluctuation or fluctuation redundancy. This triggers a second early warning instruction and generates a second strategy: An adaptive sampling control method and a time window constraint method are used to perform boundary convergence processing on the time window and adjust the interval window. When the adaptive time window length is less than the minimum adaptive time window length, the time window is increased to the minimum adaptive time window length, and the data sampling frequency is increased by 20%–40% to enhance the local time scale alignment accuracy. When the adaptive time window length is greater than the maximum adaptive time window length, the time window is compressed to the maximum adaptive time window length, and the data sampling frequency is reduced by 20%–30%, and aggregation processing is performed. A unified multi-scale reconstruction mechanism is initiated to perform hierarchical time alignment and scale reconstruction processing on data of different energy types to eliminate data deviations caused by inconsistent time scales. The multi-source data after window adjustment and multi-scale alignment processing is then input into step five for comprehensive energy optimization scheduling and control decision analysis.

[0012] Furthermore, step five includes: S51. Based on the multi-source data after window adjustment and multi-scale alignment processing, and based on the device identification data ID and physical quantity type label of the current time value, the same physical quantity matching method and multi-source index mapping method are used to extract data with the same physical attributes as the current time value from different devices and obtain the corresponding data. S52. Based on a unified multi-source heterogeneous time-series dataset, the completeness, stability, and timeliness of the corresponding device data are quantified using methods such as missing rate statistics, sliding time window variance calculation, and timestamp delay analysis. The reliability parameters of the current device data and the reliability parameters of the compared device data are calculated and obtained through extreme value normalization and weighted fusion algorithms. S53. After dimensionless processing of the obtained current time value, corresponding data, current device data reliability parameters and comparison device data reliability parameters, a data consistency index is constructed using a weighted difference accumulation method.

[0013] Furthermore, step five also includes: S54. By setting a preset data credibility threshold and comparing the data consistency index with the data credibility threshold, a third evaluation result is obtained, including: When the data consistency index is less than or equal to the data credibility threshold, it indicates that the current multi-source data consistency status is qualified and can be directly used as the data input for integrated energy optimization scheduling and control decision-making, and continuously monitored. When the data consistency index exceeds the data credibility threshold, it indicates that the current multi-source data consistency status is unqualified, posing a risk of data conflict or multi-source measurement bias. This triggers a third early warning instruction and generates a third strategy: A consistency optimization function is constructed using a weighted least squares optimization method to resolve conflicts in the multi-source data, find the optimal consistency value, and replace conflicting data with the optimal consistency value. Simultaneously, the weight of data sources with unqualified consistency status is reduced by 30%–60%, and the data reconstruction frequency for conflict time segments is increased by 20%–40%. Local resampling and smoothing are performed on conflicting data to improve data consistency. Furthermore, abnormal data sources are marked and included in the subsequent priority verification queue.

[0014] Furthermore, a data processing system for integrated intelligent energy management includes: The multi-source data acquisition module is used to collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations in the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset. The time series feature extraction module is used to perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. The data evolution stability analysis module is used to calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy is given. The adaptive time window analysis module is used to calculate the length of the adaptive time window and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, adaptive window constraints and multi-scale reconstruction strategies are applied. The multi-source data consistency assessment module is used to calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If it is qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If it is not qualified, consistency optimization and conflict resolution strategies are given.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves effective alignment and conflict identification of the same physical quantity data between different devices by uniformly identifying and aligning data from multiple sources such as electricity, heat, and gas, and by combining multi-source index mapping, matching of similar physical quantities, and data consistency index construction methods. Furthermore, through weighted least squares optimization and conflict resolution strategies, abnormal data is corrected and reconstructed, which can effectively reduce multi-source measurement bias, improve data consistency and reliability, and thus provide a high-quality data foundation for subsequent integrated energy optimization and scheduling.

[0016] This invention also constructs a multi-dimensional time-series feature system that includes current value, moving average, first-order rate of change, and second-order acceleration of change, and introduces a data evolution stability index to quantitatively evaluate the state of energy data, thereby achieving accurate characterization of load fluctuations and sudden changes in equipment operation. When an unstable state is detected, an adaptive enhanced sampling and local feature enhancement strategy is adopted to effectively improve the data resolution and anomaly detection capability in key time intervals, thereby improving the system's response speed and identification accuracy to sudden anomalies.

[0017] This invention also introduces an adaptive time window index and a window adjustment coefficient to establish a dynamic mapping relationship between the time window length and data fluctuation characteristics, thereby achieving adaptive adjustment of the time window. When data fluctuations are abnormal, the invention uses window constraints and a multi-scale reconstruction mechanism to perform hierarchical alignment and scale unification processing on multi-source data, effectively eliminating analytical biases caused by inconsistent time scales. At the same time, by combining consistency assessment and optimization strategies, the invention achieves adaptive control and collaborative optimization throughout the entire data processing process, thereby improving the stability and reliability of integrated energy system operation analysis. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating the overall execution of the method of the present invention; Figure 2 This is a schematic diagram of the overall method flow of the present invention; Figure 3 This is a schematic diagram of the overall system flow of the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0021] Example 1 Please see Figures 1 to 3 This invention provides a technical solution: a data processing method for integrated energy intelligent management, the specific steps of which include: Step 1: Collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations within the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset; Step 2: Perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter, and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. Step 3: Calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy will be given. Step 4: Calculate the adaptive time window length and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, apply adaptive window constraints and multi-scale reconstruction strategies. Step 5: Calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If unqualified, consistency optimization and conflict resolution strategies are given.

[0022] Figure 1The left side shows the underlying hardware architecture of the park-level multi-energy complementary energy station, which includes mutually coupled power equipment, thermal equipment and gas equipment. This physical entity strictly corresponds to the data acquisition source of the "unified multi-source heterogeneous time series dataset" in step one of the claims of this invention. The diagram shows a logical space technology roadmap extending from the physical space to the right. This roadmap consists of four core process frames, precisely expressing the unique execution logic of this invention: The first rectangle, "Multi-source Data Acquisition and Unified Alignment," corresponds to steps one and two, presenting the basic data processing stage where the system uniformly identifies and synchronizes the parameters of the underlying heterogeneous devices, and extracts the first-order rate of change and second-order acceleration through a sliding time window; the second rectangle, "Evolutionary Stability Assessment and Enhanced Sampling," corresponds to step three, reflecting the system's calculation of the data evolution stability index and the response mechanism that directly triggers the priority scheduling strategy when an unstable state is determined; the third rectangle, "Adaptive Window Constraint Multi-scale Reconstruction," corresponds to step four, demonstrating the reconstruction process where the system calculates the adaptive time window length to eliminate load fluctuation deviations; and the fourth rectangle, "Data Consistency Optimization and Scheduling Output," corresponds to step five, clarifying the entire data closed-loop control process of calculating the data consistency index, resolving multi-source data conflicts, and ultimately using high-quality qualified data as input for comprehensive energy optimization scheduling decisions. This diagram clearly reveals the complete technical link of this invention from multi-source data perception to intelligent scheduling output.

[0023] In this embodiment, by constructing a unified multi-source heterogeneous time-series data processing flow covering power, heat and gas systems, a closed-loop management of the entire process from data acquisition and feature extraction to stability analysis and consistency assessment is achieved, which significantly improves the systematicness and synergy of integrated energy system data processing and provides a high-quality data foundation for subsequent optimized scheduling.

[0024] Example 2 Please see Figures 1 to 3 In this embodiment, as explained in Embodiment 1, specifically, step one includes: S11. In the multi-energy complementary energy stations at the park level under the comprehensive energy group, the operating status of power equipment is monitored in real time. Power quality analyzers, smart meters, voltage transformers and current transformers are used to measure the grid connection nodes and feeder circuits, and power equipment data are collected, including voltage data, current data, active power data and grid frequency data. S12. In the district heating and heat pump coupling system of the integrated energy park, the operating status of the thermal equipment is monitored in real time. The supply and return water pipe network and heat exchange unit are detected by platinum resistance temperature sensor, ultrasonic flow meter and pressure transmitter, and thermal equipment data are collected, including: supply water temperature data, return water temperature data, flow data and pipeline pressure data. S13. In the natural gas distributed energy station and biomass fuel energy supply device in the integrated energy park, the operating status of the gas and combustion system is monitored in real time. The gas transmission and distribution pipeline and combustion emission process are detected by gas turbine flow meter, gas pressure sensor and flue gas analyzer. Data of gas equipment are collected, including gas flow data, gas pressure data, flue gas temperature data and flue gas component concentration data. S14. Perform unified identification and time alignment processing on power equipment data, thermal equipment data and gas equipment data. Obtain unified timestamp data through time synchronization method, and generate equipment identification data ID and physical quantity type label in combination with equipment coding rules, and construct a unified multi-source heterogeneous time series dataset.

[0025] In this embodiment, multiple types of sensors are used to collect refined data from different energy devices, and unified identification and time alignment are performed to effectively solve the problems of inconsistent data formats and asynchronous time from multiple sources, thereby improving the accuracy of data fusion and the comparability of cross-system data.

[0026] Example 3 Please see Figures 1 to 3 In the explanation of Example 2, this embodiment specifically includes the following steps: S21. Based on a unified multi-source heterogeneous time-series dataset, real-time monitoring is performed on the active power, voltage, and current data of power equipment; the supply and return water temperature, flow rate, and heat exchange efficiency data of thermal equipment; and the pipeline pressure, instantaneous flow rate, and valve opening data of gas equipment. A unified timestamp alignment method is used to synchronously calibrate the original operating data of each energy device at each sampling time, obtaining the current value of each energy parameter at time t, denoted as... ; S22, Based on the current time value The sliding time window statistical method is used to calculate the mean of continuous historical data within a preset time window interval before the current time t, and the average value within the corresponding time window is obtained, denoted as . ; S23, Based on the current time value The method of calculating the difference between adjacent sampling times is used to calculate the value at the current time. The value compared to the previous time step is denoted as Perform difference operations to obtain the first-order rate of change of each energy parameter, denoted as ; S24, Based on first-order rate of change Furthermore, a quadratic difference calculation method is employed to perform differential operations on the first-order rates of change at adjacent time points to obtain the second-order acceleration of each energy parameter, denoted as... .

[0027] In this embodiment, multi-dimensional time-series features such as mean, first-order rate of change, and second-order acceleration of change are extracted by using a sliding time window and differential calculation method, so as to achieve a comprehensive characterization of energy data in terms of "level, rate of change, and trend of change" and enhance the dynamic perception of the system's operating status.

[0028] Example 4 Please see Figures 1 to 3 In the explanation of Example 3, this embodiment specifically includes the following steps: S31, using the obtained current time value Average value of time window First-order rate of change of each energy parameter and the second-order acceleration of various energy parameters The data evolution stability index is calculated and obtained by using the extreme value normalization method for dimensionless processing, followed by a weighted fusion algorithm. This index is denoted as... The formula is as follows:

[0029] In the formula, N represents the number of sampling points in the time window, and w1, w2 and w3 represent weighting coefficients.

[0030] The deviation of the current value from the mean has a high weight; the deviation between the current value and the average value of the time window reflects whether the data deviates from the normal operating range and is a direct basis for judging whether the system has experienced abnormal fluctuations; when the deviation increases, it indicates that the system's operating state may drift or change abruptly, so it is given the highest weight to highlight its core impact on stability. The first-order rate of change represents the impact of data evolution stability and has the second-highest weight. The first-order rate of change reflects the speed characteristics of data change and can characterize the trend of load fluctuations or equipment operation changes. When the rate of change is too large, it indicates that the system is in a state of rapid change and there may be disturbances or anomalies. Therefore, it is given the second-highest weight to reflect the significant impact of dynamic changes on stability. The second-order acceleration characterizes the impact of the change and has a medium weight. The second-order acceleration reflects the degree of abrupt change in the trend of change, that is, whether the change is accelerating or decelerating. When the acceleration is large, it indicates that there is a sudden shock or discontinuous change in the system. Therefore, it is given a certain weight to enhance the ability to identify abrupt behavior. However, its impact is relatively indirect, so its weight is slightly low. By constructing a multi-dimensional fusion index composed of a deviation term (steady-state deviation), a velocity term (dynamic change), and an acceleration term (abrupt change characteristics), the data evolution stability index is established. It can characterize the evolution of energy data from three levels: "static deviation, dynamic change, and abrupt change trend." When the data evolution stability index... When the value increases, it indicates that the deviation of the system is intensifying and the changes are becoming more drastic, resulting in a decrease in the stability of data evolution.

[0031] In this embodiment, by constructing a multi-dimensional data evolution stability index that integrates deviation terms, rate of change, and acceleration, a comprehensive quantitative assessment of the stability of energy data is achieved. This effectively identifies operational deviations, increased fluctuations, and abrupt changes, improving the accuracy and foresight of anomaly detection.

[0032] Example 5 Please see Figures 1 to 3 In the explanation of Example 4, specifically, step three further includes: S32. By setting a preset data evolution stability threshold, denoted as Dth, and setting the data evolution stability index... A comparative analysis was conducted with the data evolution stability threshold Dth to obtain the first evaluation results, including: When data evolution stability index When the value is ≤ data evolution stability threshold Dth, it indicates that the current energy data evolution state is stable and should be continuously monitored. When data evolution stability index When the data evolution stability threshold Dth is reached, it indicates that the current energy data evolution state is unstable, with risks of abnormal load fluctuations or sudden changes in equipment operation. This triggers the first early warning instruction and generates the first strategy: An adaptive sampling control method is used to dynamically adjust the raw data stream, shortening the data analysis time window by 20%–50%. Furthermore, by increasing the sampling frequency by 30%–80% and introducing a local feature enhancement algorithm, the changing features within key time intervals are extracted to enhance the accuracy of identifying abnormal fluctuations. The corresponding energy equipment operation data and its associated time segments are then marked with fluctuation status. A key data tracking and processing mechanism and a priority scheduling mechanism are initiated, and the marked data is passed as priority input to step four to drive the generation of the adaptive time window and fine-grained time alignment.

[0033] Method for obtaining the data evolution stability threshold Dth: The calibration process of the data evolution stability threshold Dth aims to determine a critical data evolution stability index that can effectively distinguish between "stable energy data evolution state" and "abnormal energy data fluctuation state". First, a standard database of energy equipment operation data is constructed. This database is formed by continuously collecting operational status data from the power, heat, and gas equipment in the integrated energy park. The database has a sample size of no less than 1000 sets, covering various states such as stable operation, light load fluctuation, and abnormal equipment fluctuation. Each sample set contains the current value. Average value of time window First-order rate of change of each energy parameter and the second-order acceleration of various energy parameters .

[0034] Subsequently, the indicators were dimensionless by using the extreme value normalization method, and the data evolution stability index of each sample was calculated using a weighted fusion algorithm. Value. Evolutionary stability index for the data in the "stable evolutionary state" group and the "fluctuating abnormal state" group. Statistical distribution analysis was performed on the values, and probability density function curves were plotted, forming two probability peaks with partially overlapping intervals. The stability index of the data evolution in the stable evolution state is shown in the figure. The values ​​are generally close to 1, while the data evolution stability index in abnormal fluctuation states... The overall value is relatively high.

[0035] The classification performance was evaluated based on the ROC curve, and the point with the largest Youden Index was selected as the optimal split point, corresponding to the data evolution stability index. The numerical value is the data evolution stability threshold Dth. The preferred data evolution stability threshold Dth is 1.2. When the data evolution stability index... When the data evolution stability index is ≤1.2, the energy data evolution is stable and can be continuously monitored; when the data evolution stability index is ≤1.2, the energy data evolution is stable and can be continuously monitored. When the value is greater than 1.2, the data evolution is determined to be abnormal, and an adaptive enhanced sampling and priority scheduling strategy is triggered.

[0036] In this embodiment, a data evolution stability threshold calibration method based on statistical distribution and ROC curve optimization is used to achieve a scientific division between stable and abnormal states. An adaptive enhanced sampling and priority scheduling mechanism is triggered when an anomaly occurs, thereby improving the ability to capture key abnormal data and the response speed.

[0037] Example 6 Please see Figures 1 to 3 In the explanation of Example 5, specifically, step four includes: S41. Based on the fluctuation state labeling data, the marked energy equipment operation data and their associated time segments are segmented and extracted. A sliding time window statistical analysis method is used to calculate the dispersion of the data collected within the window, obtaining the local standard deviation, denoted as... The extreme value statistics method is used to extract the maximum and minimum values ​​of the collected data within the same time window, and the difference is calculated to obtain the data change range, denoted as . A time-series change direction determination method is used to compare the data change trends of adjacent sampling points within the window point by point, and the number of times the change direction switches per unit time is counted to obtain the data change frequency, which is denoted as . ; S42. The obtained local standard deviation , data change range and data change frequency The extreme value normalization method is used to perform dimensionless processing, and then a weighted fusion algorithm is used to calculate and obtain the adaptive window index, denoted as . The formula is as follows:

[0038] In the formula, a1, a2, and a3 represent weighting coefficients; The local standard deviation represents the impact of window adjustment and has a high weight. The standard deviation reflects the dispersion of data within the window and is the core indicator for measuring the stability of fluctuations. When the dispersion increases, the time window should be shortened to improve the accuracy of analysis. Therefore, it is given the highest weight. The magnitude of change represents the impact of data variation and has the second highest weight. The magnitude of change reflects the difference between extreme values ​​within the window and reflects the size of the fluctuation range. When the magnitude is large, it indicates that there is strong fluctuation and the time resolution should be increased. Therefore, it is given the second highest weight. The frequency of change represents the impact of change and has an auxiliary weight. The frequency of change reflects the frequency of data fluctuations and can characterize the characteristics of high-frequency disturbances, but its impact on the window length is relatively indirect, so it is given an auxiliary weight. Adaptive window index The adaptive window index comprehensively reflects the "fluctuation intensity + fluctuation range + fluctuation frequency" of the data. An increase in the adaptive window index indicates that the data is fluctuating wildly and is complex, requiring a smaller time window to improve analytical accuracy; when the adaptive window index increases... A smaller value indicates stable data, and the window can be expanded to reduce computational overhead.

[0039] S43, Data-based Evolution Stability Index The historical stability statistics within the corresponding time window were extracted, and historical data regression analysis and proportional mapping calculation methods were used to calculate the data evolution stability index. Scale mapping is performed within a preset reference interval, and the window adjustment coefficient is obtained by fitting the relationship between time window length and data stability in historical operating data, denoted as . ; S44, through the obtained adaptive window index and window adjustment coefficient After dimensionless processing, the length of the adaptive time window is calculated using an inverse proportional mapping method, denoted as . The formula is as follows:

[0040] In the formula, This represents a minimal constant to prevent the denominator from being zero.

[0041] The physical principle behind the formula: This formula is based on the inverse proportional coupling relationship between the local fluctuation characteristics of data and the time analysis scale. Adaptive window exponent. The adaptive window exponent characterizes the overall volatility of data within the current time interval, and its value is determined by the degree of dispersion, the magnitude of change, and the frequency of change. A large time window indicates that the data exhibits high dispersion, significant fluctuations, or frequent changes within that time interval. Using a large time window in this case will average out key change information, thus reducing anomaly detection capabilities. Therefore, by adapting the time window length... With adaptive window index By establishing an inverse proportional relationship, the window length adaptively decreases as the fluctuation intensity increases, thereby improving the accuracy of characterizing local abrupt changes. Conversely, when the adaptive window exponent... When the time window is small, it indicates that the data changes steadily and the fluctuations are weak. In this case, using a larger time window will not introduce significant information loss. On the contrary, it will help reduce computational complexity and improve the overall analysis stability. Therefore, the model automatically expands the length of the time window through inverse proportional mapping.

[0042] Among them, the window adjustment coefficient It is used to adjust the overall scale of the time window to adapt to different operating conditions or historical statistical characteristics; the small constant ε is used to prevent numerical divergence when the denominator approaches zero.

[0043] Therefore, this formula essentially realizes an adaptive mechanism of "data fluctuation intensity driving dynamic adjustment of time resolution", enabling the time window to maintain a matching relationship with the data evolution characteristics.

[0044] In this embodiment, by introducing an adaptive window exponent and an inverse proportional mapping mechanism, the time window length is dynamically adjusted according to the intensity of data fluctuations, so that the data analysis resolution can adaptively match the actual operating state, thereby reducing computational redundancy while ensuring accuracy.

[0045] Example 7 Please see Figures 1 to 3 In the explanation of Example Six, specifically, step four further includes: S45. Pass through the preset window threshold range, denoted as And adapt the time window length With window threshold range A comparative analysis was conducted to obtain the second evaluation results, including: When the adaptive time window length ∈ window threshold range When the time is specified, it indicates that the current data is within a reasonable fluctuation range and will be continuously monitored; When the adaptive time window length Window threshold range When this occurs, it indicates that the current data is not in a reasonable fluctuation state, and there is a risk of sudden fluctuation or fluctuation redundancy. This triggers a second early warning instruction and generates a second strategy: an adaptive sampling control method and a time window constraint method are used to perform boundary convergence processing on the time window and adjust the interval window. When the adaptive time window length... When the minimum adaptive time window length is denoted as Wmin, the time window is increased to the minimum adaptive time window length Wmin, and the data sampling frequency is increased by 20% to 40% on the original basis to enhance the local time scale alignment accuracy; when the adaptive time window length is... >When the maximum adaptive time window length is denoted as Wmax, the time window is compressed to the maximum adaptive time window length Wmax, and the data sampling frequency is reduced by 20% to 30%, and aggregation processing is performed; a unified multi-scale reconstruction mechanism is launched to perform hierarchical time alignment and scale reconstruction processing on data of different energy types to eliminate data deviations caused by inconsistent time scales; and the multi-source data after window adjustment and multi-scale alignment processing is input into step five for comprehensive energy optimization scheduling and control decision analysis.

[0046] Window threshold range Method of obtaining: Window threshold range The calibration process aims to determine the adaptive time window length. The reasonable range is determined to distinguish between "reasonable fluctuation state" and "fluctuation mutation or redundancy state". First, historical operating data and time-series data of multi-source energy equipment are collected to build a sample database with a sample size of no less than 1,000 groups, covering data under different seasons, different load levels and different sampling frequencies.

[0047] For each sample group, the local standard deviation is calculated using a sliding time window. , data change range and data change frequency And calculate the adaptive window exponent. Furthermore, by combining the relationship between historical time window length and data stability, an adaptive window exponent is calculated using an inverse proportional mapping. For adaptive window index Conduct statistical distribution analysis and determine a reasonable window range by combining equipment engineering experience and historical abnormal fluctuations.

[0048] The preferred minimum adaptive time window length, Wmin, is 8 seconds, and the preferred maximum adaptive time window length, Wmax, is 30 seconds. When the adaptive time window length... When the minimum adaptive time window length Wmin is less than the minimum adaptive time window length, high-precision analysis is performed using the minimum window; when the adaptive time window length is less than the minimum adaptive time window length Wmin, high-precision analysis is performed using the minimum adaptive time window length Wmin. When the maximum adaptive time window length Wmax is reached, the maximum window is used for low-frequency aggregation analysis; otherwise, segmentation processing is performed according to the calculation window.

[0049] In this embodiment, by setting a time window threshold range and combining it with a multi-scale reconstruction mechanism, boundary constraints and hierarchical alignment are performed when the window is abnormal, which effectively avoids the time scale mismatch problem and improves the fusion consistency of multi-source data at different sampling frequencies.

[0050] Example 8 Please see Figures 1 to 3 In the explanation of Example 7, specifically, step five includes: S51. Based on multi-source data after window adjustment and multi-scale alignment processing, and based on the current time value... The device identification data ID and physical quantity type label are used, and a similar physical quantity matching method and a multi-source index mapping method are employed to extract values ​​from different devices at the current time. Data with the same physical properties are retrieved, and denoted as [data type]. ; S52. Based on a unified multi-source heterogeneous time-series dataset, using missing rate statistics, sliding time window variance calculation, and timestamp delay analysis, the completeness, stability, and timeliness of the corresponding device data are quantified. Furthermore, through extreme value normalization and weighted fusion algorithms, the reliability parameter of the current device data is calculated and denoted as... And the data reliability parameter of the comparison device is denoted as ; S53, using the obtained current time value Corresponding data Current device data reliability parameters And the reliability parameters of the comparison device data After dimensionless processing, a weighted difference accumulation method is used to construct the data consistency index, denoted as CDI, as shown in the following formula:

[0051] In the formula, n represents the number of multi-source device data that participate in the consistency analysis for the same physical quantity at the same timestamp t.

[0052] The physical principle behind the formula: This formula constructs a consistency metric based on the energy accumulation mechanism of differences between multi-source data. For current time values ​​with the same physical attributes at the same timestamp... Corresponding data The difference It characterizes the degree of deviation between measurement results from different devices, and transforms the deviation into a non-negative energy form through squaring, thereby avoiding the cancellation of positive and negative values ​​and amplifying the effect of large deviations.

[0053] Based on this, the current device data reliability parameter is introduced. And the reliability parameters of the comparison device data The physical meaning of applying a two-sided weighting to the discrepancies is that the higher the integrity, stability, and timeliness of the data source, the more reliable its measurement results, and the more significant its impact on the consistency assessment should be; conversely, for low-reliability data, its contribution to the overall consistency assessment should be weakened. Therefore, by constructing weighting coefficients in a product form, we can achieve a stronger expression of discrepancies in high-quality data and a suppression of interference from low-quality data.

[0054] By summing all data pairs involved in the consistency analysis, a cumulative value of overall difference energy is formed. When the data consistency index (CDI) is small, it indicates that the deviation between data sources is small and the overall consistency is high. When the data consistency index (CDI) increases, it indicates that there is significant deviation or conflict between multiple data sources, reflecting problems such as measurement error, equipment malfunction, or time alignment error.

[0055] Therefore, this formula essentially achieves a comprehensive quantitative assessment of the "degree of difference multiplied by data credibility" of multi-source data, and can effectively characterize the consistency status of multi-source data.

[0056] In this embodiment, by constructing a data consistency index based on differential energy accumulation and credibility weighting, a quantitative assessment of the consistency of multi-source physical quantity data of the same type is achieved, which can effectively identify measurement deviations, equipment malfunctions and data conflicts.

[0057] Example 9 Please see Figures 1 to 3 In the explanation of Embodiment Eight, specifically, step five further includes: S54. By setting a preset data credibility threshold, denoted as Cth, and comparing the data consistency index CDI with the data credibility threshold Cth, a third evaluation result is obtained, including: When the data consistency index CDI ≤ data credibility threshold Cth, it indicates that the current multi-source data consistency status is qualified and can be directly used as data input for integrated energy optimization scheduling and control decisions, and continuously monitored. When the Data Consistency Index (CDI) exceeds the Data Reliability Threshold (Cth), it indicates that the current multi-source data consistency status is unqualified, posing a risk of data conflict or multi-source measurement bias. This triggers a third early warning instruction and generates a third strategy: A consistency optimization function is constructed using a weighted least squares optimization method to resolve conflicts in the multi-source data, find the optimal consistency value, and replace conflicting data with the optimal consistency value. Simultaneously, the weight of data sources with unqualified consistency status is reduced by 30%–60%, and the data reconstruction frequency for conflict time segments is increased by 20%–40%. Local resampling and smoothing are performed on conflicting data to improve data consistency. Furthermore, abnormal data sources are marked and included in the subsequent priority verification queue.

[0058] The method for obtaining the data credibility threshold Cth: The calibration process of the data credibility threshold Cth aims to distinguish the critical value between "qualified" and "unqualified" multi-source data consistency status. First, a multi-source heterogeneous energy data database is constructed. This database is formed by collecting data on the same physical quantities from electricity, heat, and gas equipment under different operating conditions. The database sample size is no less than 1500 sets, and each set includes data integrity indicators (missing rate), stability indicators (sliding time window variance), and timeliness indicators (timestamp delay).

[0059] The current time value of each sample is calculated using a weighted fusion algorithm. Corresponding data Current device data reliability parameters And the reliability parameters of the comparison device data Calculate the Data Consistency Index (DCI). Perform statistical distribution analysis on the DCI values ​​of the "Data Consistent" group and the "Data Conflicting" group, and plot probability density function curves. Use ROC curves to evaluate classification performance, and select the DCI value corresponding to the maximum Youden index as the data confidence threshold Cth.

[0060] The preferred data reliability threshold is Cth = 0.015. When the data consistency index (DCI) ≤ 0.015, the multi-source data is deemed to be in a qualified consistency state and can be directly input into the integrated energy optimization and scheduling system; when the data consistency index (DCI) > 0.015, consistency optimization and conflict resolution strategies are triggered.

[0061] In this embodiment, a conflict resolution mechanism based on weighted least squares and a dynamic weight adjustment strategy are used to optimize and reconstruct inconsistent data and correct its credibility. At the same time, local resampling and smoothing processes are combined to improve data consistency and overall reliability.

[0062] Example 10 Please refer to a data processing system for integrated intelligent energy management. Figures 1 to 3 Specifically, including: The multi-source data acquisition module is used to collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations in the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset. The time series feature extraction module is used to perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. The data evolution stability analysis module is used to calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy is given. The adaptive time window analysis module is used to calculate the length of the adaptive time window and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, adaptive window constraints and multi-scale reconstruction strategies are applied. The multi-source data consistency assessment module is used to calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If it is qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If it is not qualified, consistency optimization and conflict resolution strategies are given.

[0063] In this embodiment, the functions of data acquisition, feature extraction, stability analysis, window adjustment, and consistency evaluation are decoupled through modular system architecture design, thereby achieving high scalability and maintainability of the system and facilitating its application in integrated energy scenarios of different scales.

[0064] It should be noted that all calculation formulas in this application employ regression analysis, including but not limited to machine learning algorithms, to deeply analyze the collected parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their effectiveness and accuracy. In all calculation formulas in this application, the parameters in each formula undergo dimensionless processing within a consistent range to ensure that different physical quantities are compared on the same scale; dimensionless processing techniques include, but are not limited to, min-max-normalization and Z-score standardization. The algorithm of this invention is implemented as a Python script. Before executing the core logic, the program first executes a data loading module (e.g., using the widely used pandas library in Python) configured to read the aforementioned spreadsheet file and load its contents into the program's working memory (e.g., a DataFrame data structure). Subsequent algorithm steps will directly query and retrieve the required configuration parameters from this in-memory data structure.

[0065] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data processing method for integrated energy intelligent management, characterized in that, The specific steps include: Step 1: Collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations within the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset; Step 2: Perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter, and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. Step 3: Calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy will be given. Step 4: Calculate the adaptive time window length and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, apply adaptive window constraints and multi-scale reconstruction strategies. Step 5: Calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If unqualified, consistency optimization and conflict resolution strategies are given.

2. The data processing method for integrated energy intelligent management according to claim 1, characterized in that: Step one includes: S11. In the multi-energy complementary energy stations at the park level under the comprehensive energy group, the operating status of power equipment is monitored in real time. Power quality analyzers, smart meters, voltage transformers and current transformers are used to measure the grid connection nodes and feeder circuits, and power equipment data are collected, including voltage data, current data, active power data and grid frequency data. S12. In the district heating and heat pump coupling system of the integrated energy park, the operating status of the thermal equipment is monitored in real time. The supply and return water pipe network and heat exchange unit are detected by platinum resistance temperature sensor, ultrasonic flow meter and pressure transmitter, and thermal equipment data are collected, including: supply water temperature data, return water temperature data, flow data and pipeline pressure data. S13. In the natural gas distributed energy station and biomass fuel energy supply device in the integrated energy park, the operating status of the gas and combustion system is monitored in real time. The gas transmission and distribution pipeline and combustion emission process are detected by gas turbine flow meter, gas pressure sensor and flue gas analyzer. Data of gas equipment are collected, including gas flow data, gas pressure data, flue gas temperature data and flue gas component concentration data. S14. Perform unified identification and time alignment processing on power equipment data, thermal equipment data and gas equipment data. Obtain unified timestamp data through time synchronization method, and generate equipment identification data ID and physical quantity type label in combination with equipment coding rules, and construct a unified multi-source heterogeneous time series dataset.

3. The data processing method for integrated energy intelligent management according to claim 2, characterized in that: Step two includes: S21. Based on a unified multi-source heterogeneous time-series dataset, real-time monitoring is performed on the active power data, voltage data, and current data of power equipment, the supply and return water temperature data, flow data, and heat exchange efficiency data of thermal equipment, and the pipeline pressure data, instantaneous flow data, and valve opening data of gas equipment. A unified timestamp alignment method is used to synchronously calibrate the original operating data of each energy device at each sampling time, and obtain the current value of each energy parameter at time t. S22. Based on the current time value, the sliding time window statistical method is used to calculate the mean of the continuous historical data within the preset time window interval before the current time t, and obtain the average value within the corresponding time window. S23. Based on the current time value, the adjacent sampling time difference calculation method is used to perform difference calculation between the current time value and the previous time value to obtain the first-order rate of change of each energy parameter. S24. Based on the first-order rate of change, a second-order difference calculation method is further adopted to perform difference operations on the first-order rate of change at adjacent times to obtain the second-order acceleration of each energy parameter.

4. The data processing method for integrated energy intelligent management according to claim 3, characterized in that: Step three includes: S31. By obtaining the current time value and the average value of the time window. The first-order rate of change and the second-order acceleration of change of each energy parameter are dimensionlessly processed using the extreme value normalization method, and then the data evolution stability index is calculated using a weighted fusion algorithm.

5. The data processing method for integrated energy intelligent management according to claim 4, characterized in that: Step three also includes: S32. By setting a preset data evolution stability threshold and comparing the data evolution stability index with the data evolution stability threshold, the first evaluation result is obtained, including: When the data evolution stability index is less than or equal to the data evolution stability threshold, it indicates that the current energy data evolution state is stable and should be continuously monitored. When the data evolution stability index exceeds the data evolution stability threshold, it indicates that the current energy data evolution state is unstable, posing a risk of abnormal load fluctuations or sudden equipment operation changes. This triggers the first early warning instruction and generates the first strategy: An adaptive sampling control method is used to dynamically adjust the raw data stream, shortening the data analysis time window by 20%–50%. Furthermore, by increasing the sampling frequency by 30%–80% and introducing a local feature enhancement algorithm, the changing features within key time intervals are extracted to enhance the accuracy of identifying abnormal fluctuations. The corresponding energy equipment operation data and its associated time segments are then marked with fluctuation status. A key data tracking and processing mechanism and a priority scheduling mechanism are initiated, and the marked data is passed as priority input to step four to drive the generation of the adaptive time window and refined time alignment processing.

6. The data processing method for integrated energy intelligent management according to claim 5, characterized in that: Step four includes: S41. Based on the fluctuation state marker data, the marked energy equipment operation data and its associated time segments are segmented and extracted. The sliding time window statistical analysis method is used to calculate the dispersion of the collected data within the window and obtain the local standard deviation. The extreme value statistical method is used to extract the maximum and minimum values ​​of the collected data within the same time window, and the difference is calculated to obtain the data change amplitude. The time series change direction determination method is used to compare the data change trends of adjacent sampling points within the window point by point, and the number of times the change direction changes per unit time is counted to obtain the data change frequency. S42. The obtained local standard deviation, data variation amplitude and data variation frequency are processed into dimensionless values ​​using the extreme value normalization method, and then the weighted fusion algorithm is used to calculate and obtain the adaptive window index. S43. Based on the data evolution stability index, extract historical stability statistics within the corresponding time window, use historical data regression analysis and proportional mapping calculation methods to perform scale mapping processing on the data evolution stability index within the preset reference interval, and combine the correlation between the time window length and data stability in historical operating data to perform fitting calculations and obtain the window adjustment coefficient. S44. After dimensionless processing of the obtained adaptive window index and window adjustment coefficient, the length of the adaptive time window is calculated using the inverse proportional mapping calculation method.

7. The data processing method for integrated energy intelligent management according to claim 6, characterized in that: Step four also includes: S45. By using a preset window threshold range and comparing the adaptive time window length with the window threshold range, a second evaluation result is obtained, including: When the adaptive time window length is within the window threshold range, it indicates that the current data is in a reasonable fluctuation state and should be continuously monitored. When the adaptive time window length When the window threshold range is reached, it indicates that the current data is not in a reasonable fluctuation state and there is a risk of sudden fluctuation or fluctuation redundancy. This triggers a second early warning instruction and generates a second strategy: An adaptive sampling control method and a time window constraint method are used to perform boundary convergence processing on the time window and adjust the interval window. When the adaptive time window length is less than the minimum adaptive time window length, the time window is increased to the minimum adaptive time window length, and the data sampling frequency is increased by 20%–40% to enhance the local time scale alignment accuracy. When the adaptive time window length is greater than the maximum adaptive time window length, the time window is compressed to the maximum adaptive time window length, and the data sampling frequency is reduced by 20%–30%, and aggregation processing is performed. A unified multi-scale reconstruction mechanism is initiated to perform hierarchical time alignment and scale reconstruction processing on data of different energy types to eliminate data deviations caused by inconsistent time scales. The multi-source data after window adjustment and multi-scale alignment processing is then input into step five for comprehensive energy optimization scheduling and control decision analysis.

8. The data processing method for integrated energy intelligent management according to claim 7, characterized in that: Step five includes: S51. Based on the multi-source data after window adjustment and multi-scale alignment processing, and based on the device identification data ID and physical quantity type label of the current time value, the same physical quantity matching method and multi-source index mapping method are used to extract data with the same physical attributes as the current time value from different devices and obtain the corresponding data. S52. Based on a unified multi-source heterogeneous time-series dataset, the completeness, stability, and timeliness of the corresponding device data are quantified using methods such as missing rate statistics, sliding time window variance calculation, and timestamp delay analysis. The reliability parameters of the current device data and the reliability parameters of the compared device data are calculated and obtained through extreme value normalization and weighted fusion algorithms. S53. After dimensionless processing of the obtained current time value, corresponding data, current device data reliability parameters and comparison device data reliability parameters, a data consistency index is constructed using a weighted difference accumulation method.

9. The data processing method for integrated energy intelligent management according to claim 8, characterized in that: Step five also includes: S54. By setting a preset data credibility threshold and comparing the data consistency index with the data credibility threshold, a third evaluation result is obtained, including: When the data consistency index is less than or equal to the data credibility threshold, it indicates that the current multi-source data consistency status is qualified and can be directly used as the data input for integrated energy optimization scheduling and control decision-making, and continuously monitored. When the data consistency index exceeds the data credibility threshold, it indicates that the current multi-source data consistency status is unqualified, posing a risk of data conflict or multi-source measurement bias. This triggers a third early warning instruction and generates a third strategy: A consistency optimization function is constructed using a weighted least squares optimization method to resolve conflicts in the multi-source data, find the optimal consistency value, and replace conflicting data with the optimal consistency value. Simultaneously, the weight of data sources with unqualified consistency status is reduced by 30%–60%, and the data reconstruction frequency for conflict time segments is increased by 20%–40%. Local resampling and smoothing are performed on conflicting data to improve data consistency. Furthermore, abnormal data sources are marked and included in the subsequent priority verification queue.

10. A data processing system for integrated energy intelligent management, applied to the data processing method for integrated energy intelligent management as described in any one of claims 1 to 9, characterized in that, include: The multi-source data acquisition module is used to collect power equipment data, heating equipment data, and gas equipment data from multiple complementary energy stations in the park, and perform unified identification and time alignment processing to construct a unified multi-source heterogeneous time-series dataset. The time series feature extraction module is used to perform time alignment and synchronization calibration on the operating status parameters of each energy device in the unified multi-source heterogeneous time series dataset, extract the current time value, and obtain the average value, the first-order rate of change of each energy parameter and the second-order acceleration of each energy parameter within the corresponding time window through sliding time window statistics and adjacent time series difference calculation. The data evolution stability analysis module is used to calculate the data evolution stability index and compare it with the data evolution stability threshold to determine whether the current energy data evolution state is stable. If it is unstable, an adaptive enhanced sampling and priority scheduling strategy is given. The adaptive time window analysis module is used to calculate the length of the adaptive time window and compare it with the window threshold range to determine whether the current data is in a reasonable fluctuation state. If not, adaptive window constraints and multi-scale reconstruction strategies are applied. The multi-source data consistency assessment module is used to calculate the data consistency index and compare it with the data credibility threshold to determine whether the current multi-source data consistency status is qualified. If it is qualified, it is directly used as the data input for integrated energy optimization scheduling and control decision-making. If it is not qualified, consistency optimization and conflict resolution strategies are given.

Images