A comprehensive energy optimization scheduling system and method based on big data analysis

Abstract

This application provides a comprehensive energy optimization scheduling system and method based on big data analysis. It acquires historical and real-time monitoring data from multiple energy sources; identifies pattern similarity characteristics between renewable energy output curves and load demand curves from historical operating data; and divides the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on these pattern similarity characteristics. It determines the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, for each non-uniform scheduling period. When the real-time power prediction deviation in a future non-uniform scheduling period exceeds the corresponding power deviation confidence interval, it reallocates power resources for the future non-uniform scheduling period, generating power adjustment instructions for global energy optimization scheduling; and performs comprehensive energy optimization scheduling based on big data analysis using these power adjustment instructions. The technical solution provided in this application can achieve optimized scheduling of comprehensive energy resources under the influence of power imbalance.

Description

Technical Field

[0001] This application relates to the field of energy dispatching technology, and more specifically, to a comprehensive energy optimization dispatching system and method based on big data analysis. Background Technology

[0002] Energy dispatch is a key part of power system operation. It refers to the real-time or forward-looking allocation and coordinated control of various power generation resources through scientific prediction and optimization decision-making, under the premise of ensuring the safety and stability of the power grid. Its goal is to achieve real-time dynamic balance between power generation and power consumption load and ensure power supply reliability. With the large-scale grid connection of new energy sources, their volatility and uncertainty have increased the difficulty of dispatch. Modern energy dispatch relies on advanced information and communication systems to achieve economical, safe and low-carbon operation of the power grid.

[0003] In existing energy dispatching, firstly, based on load and renewable energy forecasts, a day-ahead power generation plan is formulated, arranging unit start-ups and shutdowns with the goal of optimal economic efficiency. During the day, rolling adjustments are made based on ultra-short-term forecasts. In real-time operation, the dispatch center, through an automatic generation control system, issues commands at a frequency of seconds to minutes to dynamically adjust the output of frequency-regulating units and balance grid power fluctuations in real time to complete energy dispatching. However, in comprehensive energy optimization dispatching based on big data analysis, a rigid division of dispatching periods of equal length is typically used. This rigid division ignores the non-linear changes in renewable energy output and diverse load demands over time. The spatiotemporal coupling characteristics of renewable energy are uniform, highly volatile, and interconnected. Due to the intermittent and random nature of renewable energy output and the periodic and event-driven fluctuations in load demand, the two exhibit complex patterns of high coordination or significant divergence in specific periods. Using uniform time periods cannot accurately acquire and utilize these short-term, high-resolution variation patterns and correlations, resulting in a low degree of matching between energy dispatch strategies and actual source-load conditions. During periods of severe fluctuation, large power imbalances are prone to occur, leading to a decrease in the dispatch accuracy of integrated energy optimization dispatch. Therefore, how to achieve optimized dispatch of integrated energy under the influence of power imbalance has become a challenge for the industry. Summary of the Invention

[0004] This application provides a comprehensive energy optimization scheduling system and method based on big data analysis, which can achieve optimized scheduling of comprehensive energy under the influence of power imbalance.

[0005] In a first aspect, this application provides a comprehensive energy optimization scheduling method based on big data analysis, comprising the following steps: Obtain historical operational data and real-time monitoring data of various energy sources from the integrated energy management platform; Big data analysis is used to identify the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data, and the energy optimization scheduling cycle is divided into multiple non-uniform scheduling periods based on the pattern similarity characteristics. Determine the predicted values ​​of renewable energy output and load demand, as well as the associated confidence intervals of power deviation, for each non-uniform scheduling period; During the rolling execution of intraday energy dispatch, when the real-time power prediction deviation in the future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, the power of the adjustment resources in the future non-uniform dispatch period is redistributed with the preset benchmark dispatch strategy as the boundary condition, and a power adjustment instruction with global energy optimization dispatch is generated. The power adjustment command is sent to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

[0006] In some embodiments, identifying pattern similarity features between renewable energy output curves and load demand curves from historical operating data through big data analysis specifically includes: Extract the renewable energy output curve and load demand curve from the historical operating data; The time dimension features of the renewable energy output curve and the load demand curve are extracted to obtain the feature set of the renewable energy output curve and the feature set of the load demand curve. A time-series similarity measurement algorithm is used to perform feature matching calculations on the standardized renewable energy output curve feature set and the load demand curve feature set to obtain the pattern similarity value between the renewable energy output curve and the load demand curve; Cluster analysis was performed based on the pattern similarity measure between the renewable energy output curve and the load demand curve to obtain the pattern similarity characteristics between the renewable energy output curve and the load demand curve.

[0007] In some embodiments, dividing the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on the pattern similarity characteristics specifically includes: Based on the pattern similarity features, the feature similarity of each time node within the energy optimization scheduling cycle is calculated to obtain the feature similarity sequence of time nodes within the scheduling cycle. By combining the preset feature similarity threshold with the feature similarity sequence, the time nodes within the scheduling period are continuously merged to obtain a homogeneous cluster of time-series features within the energy optimization scheduling period. By mapping each homogeneous cluster of time-series characteristics to an independent scheduling unit of the energy optimization scheduling cycle, multiple non-uniform scheduling periods are obtained.

[0008] In some embodiments, determining the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, for each non-uniform scheduling period specifically includes: Historical operation data and real-time monitoring data matched for each non-uniform scheduling period are extracted to obtain the energy characteristic sample set for each non-uniform scheduling period; Using the energy characteristic sample set of each non-uniform scheduling period as input, the predicted values ​​of renewable energy output and load demand in each non-uniform scheduling period are calculated by a preset time series prediction model. Based on the predicted values ​​of renewable energy output and load demand during each non-uniform scheduling period, the power prediction residual sequence for the corresponding non-uniform scheduling period is calculated. Based on the power prediction residual sequence of each non-uniform scheduling period, the confidence interval of power deviation associated with renewable energy output and load demand is determined for each non-uniform scheduling period.

[0009] In some embodiments, determining the power deviation confidence interval related to renewable energy output and load demand within each non-uniform scheduling period based on the power forecast residual sequence of each non-uniform scheduling period specifically includes: Statistical distribution fitting was performed on the power prediction residual sequence for each non-uniform scheduling period to obtain the power residual distribution model for each non-uniform scheduling period. Based on the power residual distribution model for each non-uniform scheduling period and combined with the preset confidence level, the confidence interval of power deviation associated with renewable energy output and load demand in each non-uniform scheduling period is determined.

[0010] In some embodiments, sending the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis specifically includes: The power adjustment command is parsed using a standardized format to obtain a standardized power adjustment command; The standardized power adjustment command is encapsulated using the industrial communication protocol of the integrated energy management platform to obtain the encapsulated power adjustment command message. The encapsulated power adjustment command message is sent to the corresponding actuator through the communication link of the integrated energy management platform to perform power adjustment operations, thereby completing the integrated energy optimization scheduling based on big data analysis.

[0011] In some embodiments, historical operating data of multiple energy sources are obtained from a historical database in an integrated energy management platform.

[0012] Secondly, this application provides a comprehensive energy optimization scheduling system based on big data analysis, used to execute a comprehensive energy optimization scheduling method based on big data analysis. The system includes: The acquisition module is used to acquire historical operating data and real-time monitoring data of various energy sources from the integrated energy management platform; The processing module is used to identify the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data through big data analysis, and to divide the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on the pattern similarity characteristics. The processing module is also used to determine the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, for each non-uniform scheduling period. The processing module is also used to, during the rolling execution of intraday energy dispatch, when the real-time power prediction deviation in the future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, redistribute the power of the adjustment resources in the future non-uniform dispatch period with a preset benchmark dispatch strategy as the boundary condition, and generate a power adjustment instruction for global energy optimization dispatch. The execution module is used to send the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described integrated energy optimization scheduling method based on big data analysis.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned comprehensive energy optimization scheduling method based on big data analysis.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The integrated energy optimization scheduling system and method based on big data analysis provided in this application first obtains historical operating data and real-time monitoring data of multiple energy sources from the integrated energy management platform. Second, through big data analysis, it identifies the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data, and divides the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on these pattern similarity characteristics. Further, it determines the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, within each non-uniform scheduling period. Then, during the rolling execution of intraday energy scheduling, when the real-time power prediction deviation in a future non-uniform scheduling period exceeds the corresponding power deviation confidence interval, it reallocates the power of the regulating resources in the future non-uniform scheduling period using a preset benchmark scheduling strategy as boundary conditions, generating a power adjustment instruction for global energy optimization scheduling. Finally, it sends the power adjustment instruction to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

[0016] Therefore, this application can achieve optimized scheduling of integrated energy under the influence of power imbalance. First, by acquiring historical operating data and real-time monitoring data of multiple energy sources from the integrated energy management platform, it can provide comprehensive and continuous data source support for subsequent full-process scheduling optimization, avoiding analytical bias caused by single data dimensions. Second, through big data analysis, it identifies the pattern similarity characteristics between renewable energy output curves and load demand curves from historical operating data, and divides the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on these characteristics. This avoids the problem that using uniform time period division cannot accurately acquire and utilize these short-term high-resolution variation patterns and correlation patterns, leading to a low matching degree between energy scheduling strategies and actual source loads, thus reducing the scheduling accuracy of integrated energy optimization scheduling. This improves the targeting of subsequent prediction and scheduling, reduces redundant calculations, and improves scheduling efficiency. Furthermore, it determines the predicted values ​​of renewable energy output and load demand and the associated power deviation confidence intervals in each non-uniform scheduling period to solve the deficiency of traditional single-point prediction lacking quantitative error boundaries, clarifying the reliability and error of the predicted values. The fluctuation range provides a precise reference for subsequent real-time deviation determination, reducing scheduling risks caused by prediction errors. Then, during the rolling execution of intraday energy scheduling, when the real-time power prediction deviation in a future non-uniform scheduling period exceeds the corresponding power deviation confidence interval, the power of the regulating resources in the future non-uniform scheduling period is redistributed using a preset benchmark scheduling strategy as boundary conditions. This generates power adjustment instructions for global energy optimization scheduling, thereby constructing a closed-loop control link for real-time deviation correction. This ensures that the power allocation of local regulating resources not only meets the power supply and demand balance requirements within the period but also anchors the economic and stability goals of global energy optimization, avoiding over- or under-regulation of regulating resources and improving the utilization efficiency of regulating resources. Finally, the power adjustment instructions are sent to the execution agency of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis. This enables the precise execution of optimization instructions, thereby increasing the renewable energy consumption rate of the integrated energy system and reducing the operation and maintenance costs of regulating resources. In summary, the technical solution provided in this application can achieve optimized scheduling of integrated energy under the influence of power imbalance. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of a comprehensive energy optimization scheduling method based on big data analysis, as shown in some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of pattern similarity features according to some embodiments of this application; Figure 3 This is a schematic diagram of the structure of a comprehensive energy optimization scheduling system based on big data analysis, according to some embodiments of this application; Figure 4This is a schematic diagram of the structure of a computer device that implements a comprehensive energy optimization scheduling method based on big data analysis, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The figure is an exemplary flowchart of a comprehensive energy optimization scheduling method based on big data analysis, according to some embodiments of this application. The figure mainly includes the following steps: In step S101, historical operation data and real-time monitoring data of multiple energy sources are obtained from the integrated energy management platform.

[0020] It should be noted that the integrated energy management platform in this application refers to a comprehensive information management platform that integrates the Internet of Things, big data, cloud computing, intelligent control and energy optimization algorithms to build a comprehensive information management platform for all aspects of the production, transmission, distribution, storage and use of diverse energy sources, including but not limited to electricity, heat, cooling and gas, for various energy-consuming entities such as industrial parks, industrial enterprises and regional energy service providers. This integrated energy management platform can realize real-time collection and intelligent decision analysis of multi-dimensional operation data of diverse energy sources.

[0021] In practice, historical operating data of various energy sources are obtained from the historical database of the integrated energy management platform, and real-time monitoring data of various energy sources in the integrated energy management platform are obtained through the data processing layer of the integrated energy management platform. The various energy sources include, but are not limited to, wind power, photovoltaic power, hydropower, and thermal power.

[0022] It should be noted that, in this application, historical operating data refers to various operation-related data collected by the integrated energy management platform from the multi-energy production links during its historical operation. The historical operating data includes, but is not limited to, historical output sequences of renewable energy (i.e., output power data of wind turbine generators and photovoltaic arrays in the same historical period), historical demand sequences of multi-load (i.e., power consumption data of electrical load, heat load, and cooling load in the same historical period), and historical operating status and efficiency data of equipment. In this application, real-time monitoring data refers to various operation-related data collected by the integrated energy management platform from the multi-energy production links in real time. Real-time monitoring data includes, but is not limited to, real-time output sequences of renewable energy (i.e., output power data of wind turbine generators and photovoltaic arrays), real-time demand sequences of multi-load (i.e., power consumption data of electrical load, heat load, and cooling load), and real-time operating status and efficiency data of equipment.

[0023] In step S102, the pattern similarity characteristics between the renewable energy output curve and the load demand curve are identified from the historical operating data through big data analysis, and the energy optimization scheduling cycle is divided into multiple non-uniform scheduling periods based on the pattern similarity characteristics.

[0024] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart illustrating the determination of pattern similarity features according to some embodiments of this application. In this embodiment, the identification of pattern similarity features between renewable energy output curves and load demand curves from historical operating data through big data analysis can be achieved through the following steps: In step S1021, the renewable energy output curve and load demand curve are extracted from the historical operating data; In step S1022, time-dimensional features are extracted from the renewable energy output curve and the load demand curve to obtain a feature set of the renewable energy output curve and a feature set of the load demand curve. In step S1023, a time-series similarity measurement algorithm is used to perform feature matching calculation on the standardized renewable energy output curve feature set and the load demand curve feature set to obtain the pattern similarity value between the renewable energy output curve and the load demand curve. In step S1024, cluster analysis is performed based on the pattern similarity measure between the renewable energy output curve and the load demand curve to obtain the pattern similarity characteristics between the renewable energy output curve and the load demand curve.

[0025] In specific implementation, firstly, a timestamp-aligned preprocessing method is used to normalize the renewable energy output time-series data and load demand time-series data in the historical operation data according to a unified time granularity, forming renewable energy output curves and load demand curves. The renewable energy output curve refers to a time-series data sequence reflecting the continuous change of renewable energy output power over time, with time as the horizontal axis and renewable energy output power as the vertical axis. The load demand curve refers to a time-series data sequence reflecting the continuous change of demand power over time, with time as the horizontal axis and user-side energy consumption power as the vertical axis. Secondly, the renewable energy output curve and the load demand curve are traversed through a preset fixed time window and step size. For the time-series data within each window, time-domain features including mean, variance, and peak percentage are calculated sequentially. The time-domain features of each window are arranged in time sequence to form feature vectors. All feature vectors are combined to obtain the renewable energy output curve feature set and the load demand curve feature set, respectively. The renewable energy output curve feature set refers to the feature set composed of the time-domain feature vectors of each sliding window of the renewable energy output curve arranged in time sequence. The load demand curve feature set refers to the feature set composed of the load demand curve feature vectors of each sliding window of the renewable energy output curve arranged in time sequence. The process begins by determining the feature set of each sliding window's time-domain feature vectors on the curve, arranged according to the time series. Then, using Z-score normalization, the sample mean and standard deviation are calculated for each feature dimension in both the renewable energy output curve feature set and the load demand curve feature set. The normalization transformation of each feature value is then performed using the formula (original feature value - sample mean) / sample standard deviation, resulting in the standardized renewable energy output curve feature set and load demand curve feature set. Next, the Dynamic Time Warping (DTW) time series similarity measurement algorithm is selected, using the two standardized feature sets as input. A time series distance matrix is ​​constructed based on the feature vector dimensions, where each element represents the Euclidean distance between the feature vectors of the corresponding time nodes of the two curves. Then, dynamic programming is used to calculate the cumulative distance at each time node according to the DTW recursive formula to determine the time series matching path with the smallest cumulative distance. The total cumulative distance of this time series matching path is then normalized to obtain the pattern similarity value between the renewable energy output curve and the load demand curve. The pattern similarity value, which characterizes the degree of similarity between the features of the renewable energy output curve and the load demand curve, will not be elaborated further here.Finally, the obtained pattern similarity values ​​and their corresponding curve samples (i.e., the combination of renewable energy output curves and load demand curves within a specific time period, specifically defined by a sliding time window) are used as input. The K-means clustering algorithm is employed. First, the number of clusters is preset according to actual engineering needs. A corresponding number of curve samples are randomly selected as initial cluster centers. Then, the Euclidean distance from each curve sample to each initial cluster center is calculated, and the curve samples are assigned to the nearest cluster. After the initial cluster division, the mean of the pattern similarity values ​​of all samples within each cluster is recalculated, and this mean is used as the new cluster center. The operations of sample distance calculation, sample clustering, and cluster center updating are repeated until the change in cluster centers between two adjacent iterations is less than a preset convergence threshold (e.g., 0.01), at which point the iteration stops. Finally, the common time-domain features of all curves within the same cluster are combined to form the pattern similarity features of the renewable energy output curve and the load demand curve.

[0026] It should be noted that, in this application, pattern similarity features refer to a set of features characterizing the similar temporal changes of renewable energy output curves and load demand curves. In integrated energy dispatch, pattern similarity features, by extracting the common temporal change patterns of renewable energy output and load demand in different time periods, can provide a homogeneous discrimination benchmark for the non-uniform division of dispatch cycles, so that each dispatch period after division has consistent power change characteristics, avoiding the defect of traditional uniform time period division ignoring load and output fluctuation patterns. At the same time, based on the pattern similarity features, historical operating samples that match the characteristics of the period to be predicted can be screened, providing targeted input for the prediction model of renewable energy output and load demand, effectively improving the accuracy of prediction values ​​and the reliability of power deviation confidence intervals.

[0027] In some embodiments, dividing the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on the pattern similarity characteristics is achieved through the following steps: Based on the pattern similarity features, the feature similarity of each time node within the energy optimization scheduling cycle is calculated to obtain the feature similarity sequence of time nodes within the scheduling cycle. By combining the preset feature similarity threshold with the feature similarity sequence, the time nodes within the scheduling period are continuously merged to obtain a homogeneous cluster of time-series features within the energy optimization scheduling period. By mapping each homogeneous cluster of time-series characteristics to an independent scheduling unit of the energy optimization scheduling cycle, multiple non-uniform scheduling periods are obtained.

[0028] In specific implementation, firstly, a cosine similarity algorithm is used. The feature vector in the pattern similarity features is taken as the baseline feature vector, and the joint feature vector of renewable energy output-load demand collected at each time node within the energy optimization scheduling cycle is taken as the matching vector. The inner product of the baseline feature vector and each matching vector is calculated, and then this inner product is divided by the product of the magnitude of the baseline feature vector and the magnitude of the corresponding matching vector to obtain the feature similarity of each time node relative to the baseline feature. The feature similarity of each time node is then arranged sequentially according to the time order within the energy optimization scheduling cycle to obtain the feature similarity sequence of time nodes within the scheduling cycle. This feature similarity sequence refers to a numerical sequence arranged chronologically, reflecting the degree of similarity between the features of each time node within the scheduling cycle and the baseline feature. Then, a preset feature similarity threshold is obtained. The specific feature similarity threshold can be set according to actual needs and is not limited here. The feature similarity between the current time node and the next adjacent time node is extracted sequentially, and this feature similarity is compared with the feature similarity threshold. If the feature similarity is higher than the feature similarity threshold... If the similarity of two time nodes is less than the threshold, the merging operation of the current feature set is terminated, and a new feature set merging operation is started from the adjacent time node. This process is repeated until all time nodes within the scheduling cycle are traversed, resulting in homogeneous clusters of time-series features within the energy optimization scheduling cycle. These homogeneous clusters of time-series features refer to a set of nodes with consistent renewable energy output-load demand time-series characteristics. Finally, a set-to-unit mapping method is used to divide each homogeneous cluster of time-series features. Each independent scheduling unit corresponds to an independent scheduling unit within the energy optimization scheduling cycle. The timestamps of the first and last time nodes within each homogeneous cluster of time-series characteristics are used as the start and end times of the corresponding independent scheduling unit. The product of the number of time nodes within the cluster and the time granularity of the energy optimization scheduling cycle is used as the duration of the corresponding independent scheduling unit. The duration of different independent scheduling units varies due to the difference in the number of nodes within the cluster, thus obtaining multiple non-uniform scheduling periods. The set-unit mapping method refers to the mapping method that directly maps a set of nodes with homogeneous characteristics to the basic unit of energy optimization scheduling, which will not be elaborated here.

[0029] It should be noted that, in this application, non-uniform scheduling periods refer to energy optimization scheduling periods with unequal time interval lengths formed by mapping homogeneous clusters of time-series characteristics. The determination of non-uniform scheduling periods enables renewable energy output and load demand within the same time period to have consistent time-series variation characteristics, providing homogeneous sample intervals for the prediction of renewable energy output and load demand, improving the accuracy of prediction values ​​and the reliability of power deviation confidence intervals. During the rolling execution of intraday energy scheduling, appropriate regulation resource response strategies are matched for scheduling periods of different lengths and characteristics, avoiding over-regulation or under-regulation of regulation resources under uniform periods, reducing the loss of regulation resources and improving the renewable energy absorption rate.

[0030] In step S103, the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, are determined for each non-uniform scheduling period.

[0031] In some embodiments, the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, for each non-uniform scheduling period are determined by the following steps: Historical operation data and real-time monitoring data matched for each non-uniform scheduling period are extracted to obtain the energy characteristic sample set for each non-uniform scheduling period; Using the energy characteristic sample set of each non-uniform scheduling period as input, the predicted values ​​of renewable energy output and load demand in each non-uniform scheduling period are calculated by a preset time series prediction model. Based on the predicted values ​​of renewable energy output and load demand during each non-uniform scheduling period, the power prediction residual sequence for the corresponding non-uniform scheduling period is calculated. Based on the power prediction residual sequence of each non-uniform scheduling period, the confidence interval of power deviation associated with renewable energy output and load demand is determined for each non-uniform scheduling period.

[0032] In specific implementation, firstly, historical operational data and real-time monitoring data matching each non-uniform scheduling period are extracted from the historical operational database to obtain an energy characteristic sample set for each non-uniform scheduling period. This energy characteristic sample set refers to a set of samples consisting of historical operational data and real-time monitoring data matching the non-uniform scheduling period. Secondly, using the energy characteristic sample set for each non-uniform scheduling period as input, an autoregressive integral moving average model is adopted as the preset time series prediction model. The autoregressive order, differencing order, and moving average order of the model are determined by performing stationarity tests, autocorrelation, and partial autocorrelation analyses on the time series data in the energy characteristic sample set. Then, the model parameters are fitted and solved using the least squares method. The fitted model is then used to calculate the predicted values ​​of renewable energy output and load demand for each non-uniform scheduling period. The autoregressive integral moving average model is a model constructed using autoregressive terms, differencing, and moving average terms, used for predicting stationary and non-stationary time series data. The known time-series analysis model defines the predicted values ​​of renewable energy output and load demand during each non-uniform scheduling period as numerical results calculated by the time-series prediction model, reflecting the changing trends of renewable energy output and load demand during the non-uniform scheduling period. Then, based on the predicted values ​​of renewable energy output and load demand during each non-uniform scheduling period, the actual values ​​of renewable energy output and load demand in the energy feature sample set are subtracted from the corresponding predicted values ​​to obtain the power prediction residual values ​​at each time node. These residual values ​​are then arranged chronologically within the non-uniform scheduling period to obtain the power prediction residual sequence for the corresponding non-uniform scheduling period. This power prediction residual sequence is a numerical sequence reflecting the changes in prediction error, formed by arranging the differences between the actual and predicted values ​​in the energy feature sample set in chronological order. Finally, the power deviation confidence intervals associated with renewable energy output and load demand during each non-uniform scheduling period are determined based on the power prediction residual sequences for each non-uniform scheduling period.

[0033] In some embodiments, the determination of the power deviation confidence interval related to renewable energy output and load demand within each non-uniform scheduling period, based on the power prediction residual sequence of each non-uniform scheduling period, is achieved through the following steps: Statistical distribution fitting was performed on the power prediction residual sequence for each non-uniform scheduling period to obtain the power residual distribution model for each non-uniform scheduling period. Based on the power residual distribution model for each non-uniform scheduling period and combined with the preset confidence level, the confidence interval of power deviation associated with renewable energy output and load demand in each non-uniform scheduling period is determined.

[0034] In specific implementation, firstly, a well-known normal distribution fitting method is adopted to perform a Shapiro-Wilke normality test on the power prediction residual sequences for each non-uniform scheduling period. Using the power prediction residual sequences as input, the test statistic is calculated and compared with the corresponding critical value to verify whether the residual sequences conform to a normal distribution. After passing the test, the least squares method is used to fit the normal distribution parameters of the residual sequences, solving for the sample mean and sample standard deviation of the residual sequences. Then, a power residual distribution model for each non-uniform scheduling period is constructed using the sample mean and sample standard deviation. The statistical distribution fitting refers to the processing of well-known data to construct the statistical model. In the process, the power residual distribution model refers to a statistical model that characterizes the probability distribution of residual values ​​based on the normal distribution characteristics of the power prediction residual sequence. Secondly, the power residual distribution model of each non-uniform scheduling period is used as input. Combined with the confidence level preset by the integrated energy scheduling project (such as 95%), the normal distribution interval estimation method is adopted to query the standard normal distribution quantile at the corresponding confidence level. The confidence interval of the residual is calculated by using the calculation formula "upper boundary of residual confidence interval = sample mean + quantile × sample standard deviation, lower boundary of residual confidence interval = sample mean - quantile × sample standard deviation".

[0035] It should be noted that, in this application, the residual confidence interval refers to the power deviation confidence interval associated with renewable energy output and load demand during each non-uniform scheduling period. This confidence interval can quantitatively characterize the reliability of the prediction value, clearly delineate the reasonable fluctuation range and abnormal deviation range of the prediction error, and serve as a criterion for determining whether the real-time power prediction deviation exceeds a reasonable threshold during the intraday rolling execution of energy dispatch. When the real-time deviation is within the confidence interval, the dispatch system can maintain the baseline dispatch strategy to avoid losses caused by frequent start-stop of regulating resources due to small errors. When the real-time deviation exceeds the boundary of the confidence interval, the system triggers the power reallocation process of regulating resources to promptly correct the power imbalance problem caused by the prediction deviation.

[0036] In step S104, during the rolling execution of intraday energy dispatch, when the real-time power prediction deviation during a future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, the power of the adjustment resources during the future non-uniform dispatch period is redistributed using a preset benchmark dispatch strategy as boundary conditions, generating a power adjustment instruction for global energy optimization dispatch.

[0037] In some embodiments, during the rolling execution of intraday energy dispatch, when the real-time power prediction deviation during a future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, the power of the adjustment resources for the future non-uniform dispatch period is reallocated using a preset benchmark dispatch strategy as boundary conditions. The generation of power adjustment instructions for global energy optimization dispatch is achieved through the following steps: During the rolling execution of intraday energy dispatch, real-time power prediction deviation values ​​are collected for future non-uniform dispatch periods. The real-time power prediction deviation value of the future non-uniform scheduling period is compared with the corresponding power deviation confidence interval by a threshold to obtain the power deviation exceeding the limit judgment result. Based on the power deviation exceeding the limit judgment result, the preset benchmark scheduling strategy is retrieved to obtain the boundary constraint conditions of the benchmark scheduling strategy. Based on the boundary constraints of the baseline scheduling strategy, the full parameters of the adjustment resources for future non-uniform scheduling periods are extracted to obtain the scheduling feasible region of the adjustment resources. Based on the aforementioned scheduling feasible domain, power adjustment commands that are linked to global energy optimization scheduling are determined.

[0038] In specific implementation, firstly, during the rolling execution of intraday energy dispatch, real-time monitoring values ​​of renewable energy output and load demand for future non-uniform dispatch periods are acquired. These real-time monitoring values ​​are then compared with the predicted renewable energy output and load demand for the corresponding non-uniform dispatch periods to obtain the real-time power prediction deviation value for the future non-uniform dispatch periods. This real-time power prediction deviation value refers to the difference between the real-time monitoring values ​​and the corresponding predicted values ​​for renewable energy output and load demand during the future non-uniform dispatch periods, used to characterize the degree of deviation between the predicted and actual values. Secondly, the real-time power prediction deviation value for the future non-uniform dispatch periods is compared one by one with the upper and lower boundary values ​​of the corresponding power deviation confidence interval to determine whether the real-time power prediction deviation value exceeds the value range of the power deviation confidence interval, obtaining a power deviation exceeding the limit judgment result. This power deviation exceeding the limit judgment result is a binary judgment conclusion used to characterize whether the real-time power prediction deviation value exceeds the power deviation confidence interval, including both deviation exceeding the limit and deviation not exceeding the limit. Further, based on the power deviation exceeding the limit judgment result… When the power deviation exceeds the limit, the corresponding benchmark scheduling strategy is retrieved from the preset integrated energy scheduling strategy library based on the pattern similarity characteristics of future non-uniform scheduling periods. The capacity constraints, response speed constraints, and power adjustment amplitude constraints of the regulating resources in this benchmark scheduling strategy are extracted to obtain the boundary constraints of the benchmark scheduling strategy. These boundary constraints refer to the set of constraint parameters extracted from the benchmark scheduling strategy that limit the power adjustment range of the regulating resources. Then, based on the boundary constraints of the benchmark scheduling strategy, the rated power, current remaining capacity, minimum regulating power, and maximum regulating power parameters of adjustable energy storage devices, controllable generator sets, and flexible loads within the future non-uniform scheduling period are collected. These parameters are then filtered and corrected in conjunction with the boundary constraints to obtain the scheduling feasible region of the regulating resources. This scheduling feasible region refers to the set of power adjustment ranges that the regulating resources can achieve under the boundary constraints of the benchmark scheduling strategy. Finally, based on the scheduling feasible region, a power adjustment command connected to the global energy optimization scheduling is determined.

[0039] In some embodiments, determining the power adjustment command that is linked to global energy optimization scheduling based on the scheduling feasible region is achieved through the following steps: A power redistribution optimization model is constructed by taking the global energy optimization scheduling objective as the optimization objective function and combining it with the scheduling feasible region. An intelligent optimization algorithm is used to solve the power redistribution optimization model to obtain the optimal allocation scheme of regulating resource power during future non-uniform scheduling periods; A set of scheduling instructions is generated based on the optimal allocation scheme of regulating resource power during future non-uniform scheduling periods, resulting in power adjustment instructions that are connected with global energy optimization scheduling.

[0040] In specific implementation, firstly, the global energy optimization scheduling objective is decomposed into an economic objective and a stability objective. The economic objective minimizes the sum of the electricity purchase cost and the operation and maintenance cost of regulating resources within the integrated energy system, while the stability objective minimizes the real-time power supply-demand balance deviation of the system. A multi-objective optimization objective function is constructed using a weighted summation method. Then, the rated power constraints, power regulation rate constraints, and remaining capacity constraints of the regulating resources within the scheduling feasible region are used as constraints on the optimization model. The objective function and constraints are combined to construct a power redistribution optimization model for future non-uniform scheduling periods. This power redistribution optimization model is a mathematical programming model used to solve for the optimal power allocation scheme of regulating resources, with global energy optimization scheduling as the objective and the scheduling feasible region as the constraint. Secondly, the particle swarm optimization algorithm is used to solve the power redistribution optimization model. The specific steps are as follows: the position vector of each particle in the particle swarm corresponds to a set of regulating resource power allocation schemes; the calculated value of the optimization objective function is used as the fitness value of the particle; the position and velocity parameters of the particle swarm are initialized first; then, the velocity and position of the particles are updated iteratively. The process involves dynamically adjusting particle velocity based on the positions corresponding to the individual's optimal fitness value and the global optimal fitness value. Iteration stops when the number of iterations reaches a preset maximum or the fitness value converges to a preset threshold. The particle position vector corresponding to the global optimal fitness value is selected as the solution result, yielding the optimal allocation scheme for regulating resource power during future non-uniform scheduling periods. Here, the particle swarm optimization algorithm refers to a swarm intelligence optimization algorithm that achieves optimization by simulating bird flock foraging behavior. The optimal allocation scheme for regulating resource power refers to the set of power adjustment amounts and timing sequences for each regulating resource that satisfy the scheduling feasible region constraint and optimize the global optimization objective function value. Finally, based on the optimal allocation scheme for regulating resource power during future non-uniform scheduling periods, the power adjustment amounts and timing sequences in the optimal allocation scheme are decomposed and encapsulated according to the instruction format requirements of the integrated energy management platform's execution mechanism. This forms an independent control parameter for each regulating resource allocation, creating a structured scheduling instruction set. This scheduling instruction set is the power adjustment instruction that connects with the global energy optimization scheduling.

[0041] It should be noted that the power adjustment command in this application refers to a set of resource control commands that can be issued to the execution agency based on the optimal allocation scheme of the power of the regulating resources. It is determined that when the real-time power prediction deviation exceeds the power deviation confidence interval during the future non-uniform scheduling period, the power adjustment command can accurately allocate the power of regulating resources such as adjustable energy storage devices, controllable generator sets, and flexible loads based on the boundary constraints of the benchmark scheduling strategy and the feasible region of the regulating resource scheduling.

[0042] In step S105, the power adjustment command is sent to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

[0043] In some embodiments, the process of sending the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis is achieved through the following steps: The power adjustment command is parsed using a standardized format to obtain a standardized power adjustment command; The standardized power adjustment command is encapsulated using the industrial communication protocol of the integrated energy management platform to obtain the encapsulated power adjustment command message. The encapsulated power adjustment command message is sent to the corresponding actuator through the communication link of the integrated energy management platform to perform power adjustment operations, thereby completing the integrated energy optimization scheduling based on big data analysis.

[0044] In specific implementation, firstly, the power adjustment amount, adjustment timing, and corresponding actuator address in the power adjustment command are extracted and formatted uniformly to eliminate format differences between commands from different sources, resulting in a standardized power adjustment command. This standardized power adjustment command refers to a unified format control command that conforms to the actuator command receiving standard of the integrated energy management platform. Secondly, based on the industrial communication protocol (such as IEC61850) preset by the integrated energy management platform, the standardized power adjustment command is used as the message data field. A protocol header, actuator address code, and data checksum are added to complete message assembly, resulting in an encapsulated power adjustment command message. The encapsulated power adjustment command message conforms to the industrial communication protocol specification in real time, contains standardized power adjustment commands, and is a structured data frame that can be transmitted. Finally, through the dedicated industrial communication link (such as an optical fiber communication link or an industrial Ethernet link) of the integrated energy management platform, the encapsulated power adjustment command message is transmitted to the corresponding actuator address and sent to the corresponding actuator. The actuator extracts the standardized power adjustment command from the message data field and performs power adjustment operations including renewable energy output adjustment, energy storage charging and discharging control, and flexible load regulation according to the command requirements, thus completing the integrated energy optimization scheduling based on big data analysis.

[0045] It should be noted that the integrated energy optimization dispatch based on big data analysis in this application refers to the control process that achieves power supply and demand balance, renewable energy consumption improvement and dispatch economic optimization of the integrated energy management platform, which will not be elaborated here.

[0046] Furthermore, in another aspect of this application, in some embodiments, this application provides a comprehensive energy optimization and scheduling system based on big data analysis, referencing... Figure 3The figure is a schematic diagram of the structure of a comprehensive energy optimization scheduling system based on big data analysis according to some embodiments of this application. The comprehensive energy optimization scheduling system based on big data analysis includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to acquire historical operation data and real-time monitoring data of multiple energy sources from the integrated energy management platform; The processing module 202 in this application is mainly used to identify the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data through big data analysis, and to divide the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on the pattern similarity characteristics. The processing module 202 is also used to determine the predicted values ​​of renewable energy output and load demand and the associated power deviation confidence intervals during each non-uniform scheduling period. In addition, the processing module 202 is also used to redistribute power of the adjustment resources in the future non-uniform scheduling period when the real-time power prediction deviation in the future non-uniform scheduling period exceeds the corresponding power deviation confidence interval during the rolling execution of intraday energy scheduling, using a preset benchmark scheduling strategy as the boundary condition, and generate a power adjustment instruction with global energy optimization scheduling. The execution module 203 in this application is mainly used to send the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

[0047] In addition, this application also provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-described integrated energy optimization scheduling method based on big data analysis.

[0048] In some embodiments, reference Figure 4 The figure is a schematic diagram of the structure of a computer device implementing a comprehensive energy optimization scheduling method based on big data analysis, according to some embodiments of this application. The comprehensive energy optimization scheduling method based on big data analysis in the above embodiments can... Figure 4 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0049] The processor 301 can be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the comprehensive energy optimization scheduling method based on big data analysis in this application.

[0050] The communication bus 302 can be used to transmit information between the aforementioned components.

[0051] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0052] The memory 303 stores program code for executing the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, the determination of the integrated energy optimization scheduling method based on big data analysis can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0053] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0054] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0055] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0056] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-mentioned comprehensive energy optimization scheduling method based on big data analysis.

[0057] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0058] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A comprehensive energy optimization scheduling method based on big data analysis, characterized in that, Includes the following steps: Obtain historical operational data and real-time monitoring data of various energy sources from the integrated energy management platform; Big data analysis is used to identify the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data, and the energy optimization scheduling cycle is divided into multiple non-uniform scheduling periods based on the pattern similarity characteristics. Determine the predicted values ​​of renewable energy output and load demand, as well as the associated confidence intervals of power deviation, for each non-uniform scheduling period; During the rolling execution of intraday energy dispatch, when the real-time power prediction deviation in the future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, the power of the adjustment resources in the future non-uniform dispatch period is redistributed with the preset benchmark dispatch strategy as the boundary condition, and a power adjustment instruction with global energy optimization dispatch is generated. The power adjustment command is sent to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

2. The method as described in claim 1, characterized in that, The identification of pattern similarities between renewable energy output curves and load demand curves from historical operational data through big data analysis specifically includes: Extract the renewable energy output curve and load demand curve from the historical operating data; The time dimension features of the renewable energy output curve and the load demand curve are extracted to obtain the feature set of the renewable energy output curve and the feature set of the load demand curve. A time-series similarity measurement algorithm is used to perform feature matching calculations on the standardized renewable energy output curve feature set and the load demand curve feature set to obtain the pattern similarity value between the renewable energy output curve and the load demand curve; Cluster analysis was performed based on the pattern similarity measure between the renewable energy output curve and the load demand curve to obtain the pattern similarity characteristics between the renewable energy output curve and the load demand curve.

3. The method as described in claim 1, characterized in that, Based on the aforementioned pattern similarity characteristics, the energy optimization scheduling cycle is divided into multiple non-uniform scheduling periods, specifically including: Based on the pattern similarity features, the feature similarity of each time node within the energy optimization scheduling cycle is calculated to obtain the feature similarity sequence of time nodes within the scheduling cycle. By combining the preset feature similarity threshold with the feature similarity sequence, the time nodes within the scheduling period are continuously merged to obtain a homogeneous cluster of time-series features within the energy optimization scheduling period. By mapping each homogeneous cluster of time-series characteristics to an independent scheduling unit of the energy optimization scheduling cycle, multiple non-uniform scheduling periods are obtained.

4. The method as described in claim 1, characterized in that, Determining the predicted values ​​of renewable energy output and load demand, as well as the associated confidence intervals for power deviations, during each non-uniform scheduling period specifically includes: Historical operation data and real-time monitoring data matched for each non-uniform scheduling period are extracted to obtain the energy characteristic sample set for each non-uniform scheduling period; Using the energy characteristic sample set of each non-uniform scheduling period as input, the predicted values ​​of renewable energy output and load demand in each non-uniform scheduling period are calculated by a preset time series prediction model. Based on the predicted values ​​of renewable energy output and load demand during each non-uniform scheduling period, the power prediction residual sequence for the corresponding non-uniform scheduling period is calculated. Based on the power prediction residual sequence of each non-uniform scheduling period, the confidence interval of power deviation associated with renewable energy output and load demand is determined for each non-uniform scheduling period.

5. The method as described in claim 4, characterized in that, Based on the power forecast residual sequence for each non-uniform scheduling period, the confidence intervals for power deviations related to renewable energy output and load demand within each non-uniform scheduling period are determined, specifically including: Statistical distribution fitting was performed on the power prediction residual sequence for each non-uniform scheduling period to obtain the power residual distribution model for each non-uniform scheduling period. Based on the power residual distribution model for each non-uniform scheduling period and combined with the preset confidence level, the confidence interval of power deviation associated with renewable energy output and load demand in each non-uniform scheduling period is determined.

6. The method as described in claim 1, characterized in that, Sending the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis specifically includes: The power adjustment command is parsed using a standardized format to obtain a standardized power adjustment command; The standardized power adjustment command is encapsulated using the industrial communication protocol of the integrated energy management platform to obtain the encapsulated power adjustment command message. The encapsulated power adjustment command message is sent to the corresponding actuator through the communication link of the integrated energy management platform to perform power adjustment operations, thereby completing the integrated energy optimization scheduling based on big data analysis.

7. The method as described in claim 1, characterized in that, Historical operational data of various energy sources are obtained from the historical database of the integrated energy management platform.

8. A comprehensive energy optimization scheduling system based on big data analysis, used to execute the comprehensive energy optimization scheduling method based on big data analysis as described in any one of claims 1 to 7, characterized in that, The system includes: The acquisition module is used to acquire historical operating data and real-time monitoring data of various energy sources from the integrated energy management platform; The processing module is used to identify the pattern similarity characteristics between the renewable energy output curve and the load demand curve from the historical operating data through big data analysis, and to divide the energy optimization scheduling cycle into multiple non-uniform scheduling periods based on the pattern similarity characteristics. The processing module is also used to determine the predicted values ​​of renewable energy output and load demand, as well as the associated power deviation confidence intervals, for each non-uniform scheduling period. The processing module is also used to, during the rolling execution of intraday energy dispatch, when the real-time power prediction deviation in the future non-uniform dispatch period exceeds the corresponding power deviation confidence interval, redistribute the power of the adjustment resources in the future non-uniform dispatch period with a preset benchmark dispatch strategy as the boundary condition, and generate a power adjustment instruction for global energy optimization dispatch. The execution module is used to send the power adjustment command to the execution mechanism of the integrated energy management platform for integrated energy optimization scheduling based on big data analysis.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the integrated energy optimization scheduling method based on big data analysis as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the comprehensive energy optimization scheduling method based on big data analysis as described in any one of claims 1 to 7.

Images