Multi-time-scale optimal scheduling method for regional integrated energy system and computing equipment

Abstract

The invention provides a regional integrated energy system multi-time scale optimization scheduling method and computing equipment, and is applied to the technical field of new energy prediction. The method comprises the following steps: acquiring data of each energy system; constructing an optimization model of the regional integrated energy system, wherein the optimization model sequentially performs day-ahead optimization scheduling, intra-day rolling optimization scheduling and real-time optimization scheduling on the regional integrated energy system; training and solving the optimization model of the regional integrated energy system through a training data set; and performing multi-time-scale optimization scheduling of the regional integrated energy system by using the trained model. By utilizing the method, the operation efficiency and reliability of the energy system can be improved.

Description

Technical Field

[0001] The present invention relates to the fields of power systems and energy management, and particularly to a method and a computing device for multi-time scale optimal scheduling of a regional integrated energy system. Background Art

[0002] Under the background of the continuously intensifying global energy crisis and the increasingly serious environmental pollution problem in today's world, the traditional energy system has been difficult to meet the diverse energy demands of modern society. Moreover, with the continuous growth of the installed capacity of renewable energy, its status in the energy field has been gradually improving and has become the focus of current research and attention. However, due to the influence of weather conditions, the power generation using renewable energy has significant intermittency and volatility. Thus, directly connecting a large-scale renewable energy to the power grid may lead to the mismatch between power grid supply and demand, and further bring huge challenges to the safe and stable operation of the power grid. To reduce this limitation of renewable energy and improve the utilization efficiency and consumption rate, the integrated energy system has come into people's view.

[0003] The integrated energy system can integrate different energy forms such as electricity, heat energy, cold energy, and natural gas, realize the complementarity and optimal allocation of energy, help improve energy utilization efficiency, reduce energy consumption, and reduce environmental pollution, thereby promoting the sustainable development of energy. At the same time, the new generation of information technologies, including big data, Internet of Things, cloud computing, artificial intelligence, etc., provide technical support for the realization of the integrated energy system. Because of these technologies, the monitoring, control, and management of the energy system become more intelligent, and thus the operation efficiency and reliability of the energy system are improved.

[0004] Therefore, a method for multi-time scale optimal scheduling of a regional integrated energy system is needed. Summary of the Invention

[0005] The present invention aims to provide a method and a computing device for multi-time scale optimal scheduling of a regional integrated energy system, which improve the operation efficiency and stability of the entire energy system and provide solid technical support and guarantee for the efficient operation of the regional integrated energy system.

[0006] According to one aspect of the present invention, there is provided a method for multi-time scale optimal scheduling of a regional integrated energy system, including:

[0007] Obtaining data of each energy system;

[0008] Constructing an optimization model of the regional integrated energy system, and the optimization model performs day-ahead optimal scheduling, intra-day rolling optimal scheduling, and real-time optimal scheduling on the regional integrated energy system in sequence;

[0009] Training and solving the optimization model of the regional integrated energy system through a training data set;

[0010] Use the trained model to perform multi-time scale optimal scheduling for the regional integrated energy system.

[0011] According to some embodiments, the data includes:

[0012] Equipment data of each energy region;

[0013] Historical load prediction data and actual data on both sides of the source and load;

[0014] Grid and gas network price signal data.

[0015] According to some embodiments, the regional integrated energy system includes:

[0016] A dispatching center and each single integrated energy system;

[0017] The single integrated energy systems are interconnected with each other through power lines and natural gas pipelines.

[0018] According to some embodiments, training and solving the optimization model of the regional integrated energy system through a training data set includes:

[0019] Integrate the load prediction data on both sides of the source and load with the actual data, and check and clean the quality of the price signal data;

[0020] Extract statistical features from the load prediction data on both sides of the source and load and the price signal data;

[0021] Perform standardization or normalization processing on the data;

[0022] Use the processed data to train and solve the optimization model.

[0023] According to some embodiments, the optimization model includes Q-Learning, K-Means, decision tree, and / or linear regression model.

[0024] According to some embodiments, using the processed data to train and solve the optimization model includes:

[0025] Train the Q–Learning model: Define the state space, action space, and reward function, and perform iterative training through time steps. Make decisions based on the load prediction data on both sides of the source and load and the price signal to achieve model training; and / or

[0026] Train the K-Means model: Assign the load prediction data on both sides of the source and load and the price signal data points to the cluster centers, and calculate the mean of the data points in each cluster until the cluster centers reach the convergence condition; and / or

[0027] Training a decision tree model: by calculating the information gain or Gini index for each feature, selecting the first feature as the split point; and / or

[0028] Training the linear regression model: using a least squares method or a gradient descent algorithm to adjust the coefficients of the optimization model of the regional integrated energy system.

[0029] According to some embodiments, performing day-ahead optimization scheduling on the regional integrated energy system includes:

[0030] The goal is to reduce the daily operating costs of the regional integrated energy system;

[0031] Optimize the entire area based on energy purchase constraints, equipment operation constraints, transmission power constraints, and energy balance constraints;

[0032] The time resolution of the day-ahead optimization scheduling is 1 to n hours, n is less than 2, and the control time domain length is 24 hours.

[0033] According to some embodiments, performing intraday rolling optimization scheduling on the regional integrated energy system includes:

[0034] The goal is to reduce the daily operating costs of a single integrated energy system in the region;

[0035] Energy purchase constraints, equipment operation constraints, and energy balance constraints are used as constraints;

[0036] The time resolution of the intraday optimization rolling optimization scheduling is 15 to m minutes, m is less than 30, and the control time domain length is 2 hours.

[0037] According to some embodiments, performing real-time optimization scheduling on the regional integrated energy system includes:

[0038] The target is the first cost of operating a single integrated energy system taking into account the penalty cost of equipment fluctuation;

[0039] Update energy load forecast data in real time and adjust equipment power;

[0040] The time resolution of the real-time optimization scheduling is 5 to p minutes, p is less than 10, and the control time domain length is 15 minutes.

[0041] According to another aspect of the present invention, there is provided a computing device comprising:

[0042] Processor; and

[0043] A memory stores a computer program, which, when executed by the processor, causes the processor to perform any of the aforementioned methods.

[0044] According to the exemplary embodiments of the present invention, an optimization model of the regional integrated energy system is constructed. By obtaining detailed historical and real-time data, the accuracy and reliability of the model can be improved, and all components of the system and their interactions can be comprehensively described, thereby achieving the overall optimal configuration. The day-ahead optimization scheduling, intraday rolling optimization scheduling, and real-time optimization scheduling of the regional integrated energy system are carried out, realizing seamless connection and collaborative optimization between different time scales, and improving the operation efficiency and stability of the entire energy system. By training and solving the optimization model of the regional integrated energy system with a training data set, the model parameters can be adjusted according to the characteristics and requirements of different regions, and the trained model is used for multi-time scale optimization scheduling of the regional integrated energy system, effectively improving the solving accuracy and speed of the model, and effectively promoting the efficient operation of the energy system.

[0045] According to the exemplary embodiments, the method for multi-time scale optimization scheduling of the regional integrated energy system can effectively integrate renewable energy and achieve comprehensive energy management and optimal utilization.

[0046] It should be understood that the above general description and the following detailed description are only exemplary and do not limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below.

[0048] Figure 1 The flowchart showing a method for multi-time scale optimization scheduling of a regional integrated energy system according to an exemplary embodiment is shown.

[0049] Figure 2 The schematic block diagram showing a regional integrated energy system according to an exemplary embodiment is shown.

[0050] Figure 3 The schematic flow diagram showing a three-level optimization scheduling model according to an exemplary embodiment is shown.

[0051] Figure 4 The block diagram showing a computing device according to an exemplary embodiment of the present invention is shown. DETAILED DESCRIPTION

[0052] The exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. Identical reference numerals in the figures denote identical or similar parts, and thus their repeated description will be omitted.

[0053] In addition, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present invention. However, those skilled in the art will realize that the technical solutions of the present invention may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the present invention.

[0054] The block diagrams shown in the drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0055] The flowcharts shown in the drawings are merely illustrative and do not necessarily include all the content and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps may be decomposed, while some operations / steps may be combined or partially combined, so the actual execution order may change according to the actual situation.

[0056] It should be understood that although terms such as first, second, and third may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Therefore, the first component discussed below may be referred to as the second component without departing from the teachings of the concept of the present invention. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0057] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of the relevant data need to comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entrances are provided for the user to select authorization or rejection.

[0058] Those skilled in the art can understand that the drawings are only schematic diagrams of exemplary embodiments, and the modules or processes in the drawings are not necessarily essential for implementing the present invention, so they cannot be used to limit the protection scope of the present invention.

[0059] The core goal of the integrated energy system is to integrate renewable energy and achieve comprehensive energy management and optimal utilization. However, due to the uncertainties of renewable energy and user loads, as well as the lack of a universal large-scale long-term energy storage technology, long-term scheduling plans are difficult to adapt to the actual energy usage situation, resulting in excessively high operating costs and serious phenomena of wind and solar curtailment. At the same time, due to the small scale of a single integrated energy system and limited dispatchable resources, it is difficult to achieve a higher degree of multi-energy coordination. In order to promote the optimal allocation of renewable energy on a broader spatial scale, further improve its consumption rate, and simultaneously improve the economic efficiency of operation, different integrated energy systems can be interconnected by means of inter-regional power lines, natural gas pipelines, and heat pipelines, and then a regional integrated energy system can be constructed.

[0060] The main disadvantages of the existing integrated energy system scheduling are as follows: insufficient ability to cope with uncertainties. Renewable energy and user loads are uncertain, and it is difficult to flexibly respond to such uncertainties with a single-time-scale optimization; relatively high economic costs. Since the optimal allocation and coordination of energy cannot be achieved, more or larger equipment and resources need to be invested to meet energy demands, resulting in increased construction and operation costs; there are limitations in the optimal dispatch decision-making. It is difficult to adapt to the complex and changing operating conditions in the integrated energy system when formulating a scheduling plan based on fixed rules; the scale and dispatchable resources are limited. Due to the limitations of the scale and dispatchable resources of a single integrated energy system, it is often difficult to achieve a higher level of energy optimal allocation and multi-energy coordination.

[0061] Therefore, it is necessary for multiple integrated energy systems to be able to effectively optimize and supplement each other, so as to achieve the efficient utilization and reasonable distribution of energy on a larger scale, and then improve the stability and reliability of the entire energy system.

[0062] The present invention provides a method for multi-time-scale optimal scheduling of a regional integrated energy system, which can perform the optimal scheduling of the regional integrated energy system more accurately and efficiently, so as to reduce the economic cost of the regional integrated energy system and improve the consumption rate of renewable energy.

[0063] The exemplary embodiments of the present invention will be described below with reference to the accompanying drawings.

[0064] Figure 1 The flowchart showing a method for multi-time-scale optimal scheduling of a regional integrated energy system according to an exemplary embodiment is shown.

[0065] According to the exemplary embodiment, referring to Figure 1 , the method steps of a multi-time-scale optimal scheduling of a regional integrated energy system are shown. Through this method, the prediction accuracy can be improved and the resource allocation can be optimized.

[0066] In S101, data of each energy system are obtained.

[0067] According to the exemplary embodiment, obtaining data of each energy system includes equipment data of each energy region, historical load forecasting data and actual data on both sides of the source and load, and grid and gas network price signal data. Among them, the equipment involves equipment for multiple types of energy such as electricity, gas, heat, and cold, and the equipment data includes equipment ramp power, equipment capacity, maximum equipment power, self-discharge power of energy storage equipment, and conversion efficiency of each equipment; the historical load forecasting data on both sides of the source and load includes load change conditions within different time scales.

[0068] Through data collection, solid data support can be provided for the operation of the integrated energy system, thereby achieving more accurate load forecasting and efficient energy scheduling.

[0069] In S103, an optimization model of the regional integrated energy system is constructed.

[0070] According to the exemplary embodiment, an optimization model of the regional integrated energy system is constructed. The regional integrated energy system includes a dispatching center and each single integrated energy system, covering a variety of integrated energy systems. The single integrated energy systems are interconnected through power lines and natural gas pipelines.

[0071] In S105, the optimization model performs day-ahead optimal dispatching, intra-day rolling optimal dispatching, and real-time optimal dispatching on the regional integrated energy system in sequence.

[0072] According to the exemplary embodiment, the optimization model performs day-ahead optimal dispatching, intra-day rolling optimization, and real-time optimal dispatching on the regional integrated energy system in sequence. To construct a three-level optimal dispatching model of the regional integrated energy system (R-IRES), the objectives and constraints of each stage need to be accurately set. For the specific process schematic diagram, see Figure 3 .

[0073] The day-ahead optimal dispatching aims to reduce the daily operating cost of the single integrated energy system within the region. Taking energy purchase constraints, equipment operation constraints, transmission power constraints, and energy balance constraints as conditions, a day-ahead dispatching plan is formulated, and machine learning is used to correct the results for optimizing the entire region.

[0074] The intra-day rolling optimal dispatching aims to reduce the daily operating cost of the single integrated energy system within the region. Taking energy purchase constraints, equipment operation constraints, and energy balance constraints as constraints, the intra-day optimization time resolution is 15 minutes, and the control time domain length is 2 hours. An intra-day rolling dispatching plan is formulated, and machine learning is used to correct the results.

[0075] The real-time optimal scheduling aims to minimize the operating cost of a single integrated energy system considering the device fluctuation penalty cost. The constraint conditions are the same as those of the intraday rolling optimal scheduling, with energy purchase constraints, device operation constraints, and energy balance constraints as the constraint conditions. The energy load, wind power, and photovoltaic prediction data are updated in real time. The real-time optimization time resolution is 5 minutes, and the control time domain length is 15 minutes. A real-time scheduling plan is formulated, and machine learning is used to correct the results and adjust the device power.

[0076] Through three-level continuous optimization of day-ahead optimal scheduling, intraday rolling optimal scheduling, and real-time optimal scheduling, a scheduling plan for the regional integrated energy system is gradually formed.

[0077] In S107, the optimization model of the regional integrated energy system is trained and solved using the training dataset.

[0078] According to the exemplary embodiment, the optimization model of the regional integrated energy system is trained and solved using the training dataset. First, in the data preprocessing stage, the load prediction data and actual data on both sides of the source and load are integrated, and the quality of the price signal data is inspected and cleaned. The integrity of the data is checked to ensure that there are no missing values or outliers. For possible missing data points, interpolation methods such as linear interpolation and spline interpolation can be used for filling. Outliers need to be identified and corrected or removed through statistics using methods such as JH analysis. For example, if it is found that the load data at a certain moment significantly deviates from the normal range and is judged to be abnormal, it can be reasonably corrected according to the trend of the surrounding data.

[0079] The quality of the price signal data is inspected and cleaned; the quality of the price signal data of the external gas network and power grid is also inspected and cleaned, and data in different formats or from different sources are uniformly processed to ensure the consistency of the time series and the accuracy of the data.

[0080] Statistical features are extracted from the load prediction data and price signal data on both sides of the source and load. For the load prediction data on both sides of the source and load, statistical features such as daily average load, weekly average load, and monthly average load can be calculated, as well as dynamic features such as the change rate of the load and the peak-to-valley difference. These features help the model better understand the load pattern and change trend. For the price signal data, in addition to the original price values, features such as price volatility and the correlation of prices in different time periods can be calculated. For example, analyze the fluctuation differences of the power grid price on weekdays and holidays and then input them into the model to improve the sensitivity of the model to the impact of price factors.

[0081] Normalize or standardize the data. Since the load forecasting data (source-load data) and price signal data on both sides of the source and load may have different dimensions and value ranges, in order to make the model training more stable and efficient, it is necessary to normalize or standardize the data. Different data processing methods are selected according to different models. For models such as linear regression, the standardization method is usually adopted to convert the data into a distribution with a mean of 0 and a standard deviation of 1. For example, for a certain load feature column, calculate the mean and standard deviation, and then subtract the mean from each data point and divide by the standard deviation; for clustering models such as K-Means, normalizing to the [0,1] interval may be more appropriate to better determine the relative distance between data points and the clustering center.

[0082] Use the processed data to train and solve the optimization model. The optimization model includes Q-Learning, K-Means, decision tree, and / or linear regression model.

[0083] According to the exemplary embodiment, use the processed data to train and solve the optimization model, and perform model selection and initialization after the data preprocessing stage.

[0084] Q-Learning is a reinforcement learning algorithm used to find the optimal strategy in a dynamic environment. In the regional integrated energy system, it is necessary to define the state space, action space, and reward function. The state space can be a certain combination of the load forecasting data and price signals on both sides of the source and load, such as the load level, price, and historical change trends of both at the current moment; the action space can include operations such as adjusting the power generation plan and load control strategy; the reward function is designed according to the system's performance indicators, such as reducing costs and improving energy utilization efficiency. In the initialization stage, all Q values are usually initialized to a small random value or 0.

[0085] Train the Q–Learning model. Through iterative training in time steps, make decisions based on the source-load data and price signals to achieve model training; for each time step, the agent (which may be the decision-making module of the energy management system in this case) observes the current state s t , selects an action a according to the current Q table t . After performing the action, the environment will feedback the next state s t+1 and the reward r t , and then update the values in the Q table according to the Q-Learning update formula ; where α is the learning rate and γ is the discount factor.

[0086] Through a large number of time-step iterations for training, the Q-table gradually converges, and the agent can learn the optimal policy, that is, make the best decisions based on the load prediction data and price signals on both sides of the source and load to achieve the optimization goal.

[0087] Train the K-Means model. Assign the load prediction data and price signal data points on both sides of the source and load to the cluster centers, calculate the mean of the data points in each cluster, and use the mean as the new cluster center. Repeat this process until the cluster centers no longer change significantly, that is, the convergence condition is reached. In each iteration, reassign the data points according to the new cluster centers, continuously optimize the cluster division, so that the data points within the same cluster have the highest similarity, and the data points between different clusters have the lowest similarity. For example, for the clustering of source and load data, make the load patterns within the same cluster similar and the load patterns between different clusters quite different, so that different energy management strategies can be formulated according to the characteristics of different clusters. K-Means is a clustering algorithm used to divide data points into K clusters. In the regional integrated energy system, the number of clusters K needs to be determined according to the characteristics of the data and business requirements. For example, if clustering different types of load patterns, the appropriate value of K can be determined first through empirical analysis or the elbow method, etc. At initialization, randomly initialize K cluster centers, which can be points in the feature space of source and load data or price signals.

[0088] Train the decision tree model. Starting from the root node, select the best feature according to the splitting criterion to divide the data set into subsets. For each subset, continue the splitting operation recursively to construct the branches and nodes of the decision tree. During the splitting process, by calculating metrics such as the information gain or Gini index of each feature, select the feature that can best reduce the uncertainty for splitting. Select the first feature as the splitting point. For example, when considering the source and load data and price signals, determine which feature has the greatest impact on the load prediction error or cost, and use this feature as the splitting point. Repeat this process continuously until the stopping condition is met to construct a complete decision tree model, which can make corresponding decisions according to the input source and load data and price signal features, such as load adjustment strategies or energy procurement plans.

[0089] The decision tree model constructs a tree structure based on the features of the data for decision-making. First, determine the splitting criterion of the decision tree, such as information gain, Gini index, etc.; for the load prediction data and price signal data on both sides of the source and load, select the most discriminative feature as the root node for splitting according to the selected splitting criterion, and then recursively split each child node until the stopping condition is met, such as the depth of the tree reaches the preset value, the number of samples in the node is too small, etc. At initialization, the decision tree is empty, waiting for the input of data and the construction during the training process.

[0090] The form of the model is:

[0091] y = β0 + β1x1 + β2x2 + … + β n x n + ∈,

[0092] where y is the prediction target, x i is the feature variable, β i are the coefficients to be trained, and ∈ is the error term.

[0093] Train the linear regression model and use the least squares method or gradient descent algorithm to adjust the coefficients of the optimization model of the regional integrated energy system. Use optimization algorithms such as the least squares method or gradient descent method to adjust the coefficients β i . For the optimal solution of the coefficients by the least squares method:

[0094] β = (X T X) -1 X T y, we get

[0095] where X is the feature matrix and y is the target vector.

[0096] For the gradient descent method, calculate the loss function through the following formula, such as the mean squared error:

[0097]

[0098] where m is the number of samples, y i is the true value, is the predicted value.

[0099] Regarding the gradient of the coefficient β i , then update the coefficient in the opposite direction of the gradient,

[0100] that is

[0101] where α is the learning rate.

[0102] Through multiple iterations, continuously reduce the value of the loss function so that the model can better fit the relationship between the source-load data and price signals and the target variable, thereby achieving accurate prediction or optimal decision-making.

[0103] In S109, use the trained model to perform multi-time scale optimal scheduling of the regional integrated energy system.

[0104] According to the exemplary embodiments, a trained model is used for multi-time scale optimal scheduling of the regional integrated energy system, and the trained model is evaluated using the training and test set. Taking four different models as examples in the present invention, for the Q-Learning model, indicators such as the average reward and the stability of the policy in the test environment can be evaluated. For example, observe whether the decisions made by the model can effectively reduce costs and maintain the stable operation of the system under different source-load conditions and price signals.

[0105] For the K-Means model, observe whether the clustering results of the load prediction data and price signals on both sides of the source-load are reasonable, whether different patterns or categories can be clearly divided, and evaluate the quality of the clustering by calculating indicators such as the silhouette coefficient of the clustering. The silhouette coefficient measures the similarity of each data point to other points within the cluster and the difference from other clusters. The higher the silhouette coefficient value, the better the clustering effect.

[0106] For the decision tree model, calculate classification indicators such as accuracy and recall. If the decision tree model is used for classification tasks, judge the energy supply status or regression indicators such as the mean square error based on the load prediction data and price signals on both sides of the source-load. If the decision tree model is used for prediction tasks, such as load prediction, evaluate the decision accuracy and prediction accuracy of the decision tree on the test set, and judge whether it can effectively make correct judgments or predictions according to the input data.

[0107] For the linear regression model, mainly evaluate indicators such as the mean square error and the mean absolute error on the test set, observe the fitting degree of the model to the source-load data and price signals, as well as the deviation between the predicted value and the actual value, and determine whether the model can accurately capture the relationships in the data and make reliable predictions or optimization decisions.

[0108] Select the Q-Learning, K-Means, decision tree, and / or linear regression models according to different application scenarios.

[0109] If the model performance does not meet the expectations, tuning can be performed. For the Q-Learning model, the learning rate α, the discount factor γ, and the definitions of the state space and action space can be adjusted. For example, increasing the learning rate may cause the model to converge faster, but it may also lead to instability; adjusting the discount factor can change the degree of emphasis on future rewards.

[0110] For the K-Means model, different initial clustering center selection methods can be tried, the number of clusters K can be adjusted, or other improved clustering algorithms (such as K-Medoids) can be adopted. By repeatedly experimenting with different parameters and algorithm variants, find the configuration that is most suitable for clustering the source-load data and price signals.

[0111] For a decision tree model, the depth of the tree, parameters of the splitting criterion (such as the threshold of information gain), pruning strategies, etc. can be adjusted. For example, increasing the depth of the tree may improve the fitting ability of the model on the training set, and pruning can reduce the risk of overfitting and improve the generalization ability of the model.

[0112] For a linear regression model, features can be added or removed, and regularization methods such as ridge regression and Lasso regression can be adopted to prevent overfitting. Parameters such as the learning rate and the number of iterations are adjusted to optimize the training process of the model, improve the performance of the model on the test set, enable it to better adapt to the changes in source load data and price signals, and make accurate predictions and optimization decisions.

[0113] Figure 2 A schematic block diagram of a regional integrated energy system according to an exemplary embodiment is shown.

[0114] According to the exemplary embodiment, the regional integrated energy system covers multiple integrated energy systems, as shown in Figure 2 and consists of a dispatching center and each individual integrated energy system. Different integrated energy systems are interconnected through power lines and natural gas pipelines.

[0115] Each individual integrated energy system has functional areas. As shown in Figure 2 , the dispatching center 001 delivers electric power, heat energy, natural gas resources, and dispatching signals to the first integrated energy system 011, the second integrated energy system 013, the third integrated energy system 015, and the fourth integrated energy system 017 respectively. The power grid and gas network among the individual integrated energy systems are interconnected; the individual integrated energy systems shown can be a residential area integrated energy system, an office area integrated energy system, a commercial area integrated energy system, and / or a medical area integrated energy system.

[0116] Each individual regional integrated energy system uses photovoltaic to convert solar energy into electric power, converts wind energy into electric power through a wind power generation system, provides electric power for electric loads, and stores excess electric power through electric energy storage; the regional integrated energy system uses a gas turbine and a heat exchanger to generate heat by burning gas, and is equipped with gas energy storage for storing excess gas; the cold load demand is met through an electric refrigeration machine and a gas refrigerant or an absorption refrigeration machine. Different energy conversion forms are selected according to the different demands of different functional areas.

[0117] In this embodiment, by integrating multiple energy resources such as photovoltaic, wind power, and gas, and through various technical means such as energy storage, refrigeration, and heating, the efficient utilization of energy and the balance between supply and demand are achieved. The regional integrated energy system can not only improve the energy utilization efficiency, but also reduce the dependence on traditional fossil fuels, contributing to sustainable development.

[0118] Figure 3A schematic flow chart of a three-level optimization scheduling model according to an example embodiment is shown.

[0119] According to an example embodiment, the three-level optimization is to sequentially perform day-ahead optimization scheduling, intraday rolling optimization scheduling, and real-time optimization scheduling on the regional integrated energy system, thereby gradually forming a scheduling plan for the regional integrated energy system.

[0120] Day-ahead optimization scheduling aims to reduce the daily operating costs of the regional integrated energy system. It uses the input day-ahead forecast data of photovoltaic and wind power output, day-ahead forecast data of electricity, gas, heat, and cooling loads, and time-of-use electricity prices, gas prices, and heat prices to formulate a day-ahead scheduling plan. The results are corrected using machine learning. The entire region is optimized based on energy purchase constraints, equipment operation constraints, transmission power constraints, and energy balance constraints. The day-ahead optimization control time domain length is 24 hours. In this embodiment, the time resolution is 1 hour.

[0121] The intraday rolling optimization scheduling aims to reduce the daily operating cost of a single integrated energy system in a region, and optimizes the single regional integrated energy system with energy purchase constraints, equipment operation constraints, and energy balance constraints as constraints.

[0122] The intraday optimization control time domain length is 2 hours. In this embodiment, the time step is set to 15 minutes (Δt=15min), that is, the time resolution is 15 minutes. The intraday rolling scheduling plan is formulated with a 2-hour cycle and a 15-minute scheduling period, and the results are corrected using machine learning;

[0123] Update the source and load forecast data within the control time domain, but only execute the plan for the first 15 minutes of the control time domain. Then, starting from the 16th minute, reconstruct the control time domain and repeat the above calculation steps in a rolling manner. When the time is greater than 24 hours, continue to optimize the scheduling in real time.

[0124] Real-time optimization scheduling aims to minimize the operating cost of a single integrated energy system taking into account the penalty cost of equipment fluctuations. It updates energy load, wind power, and photovoltaic forecast data in real time, and uses energy purchase constraints, equipment operation constraints, and energy balance constraints as constraints. In this embodiment, the real-time optimization control time domain length is 15 minutes, and the time resolution is 5 minutes. By formulating a real-time scheduling plan, machine learning is used to correct the results.

[0125] The machine learning-based method can innovatively construct an optimization model framework for the unified regional integrated energy system. Within the model framework, the scheduling objectives and constraint conditions corresponding to different time scales are organically integrated and combined. In this way, seamless connection and collaborative optimization between different time scales are achieved, breaking the gap between time scales in the traditional scheduling mode, greatly improving the operation efficiency and stability of the entire energy system, and providing solid technical support and guarantee for the efficient operation of the regional integrated energy system.

[0126] In this embodiment, aiming at the rich information resources such as the load prediction data and actual data of the historical source and load sides at different time scales, as well as the time-sharing price data of the power grid and gas grid, a multi-time scale optimization model of the integrated energy system is constructed. By using the machine learning method, efforts are made to improve the solving accuracy and speed of the model, and thus significantly improve the scheduling economy and the renewable energy consumption rate to a great extent.

[0127] The multi-time scale optimal scheduling method for the regional integrated energy system based on machine learning conducts day-ahead optimal scheduling, intra-day rolling optimal scheduling and real-time optimal scheduling for the regional integrated energy system. With the help of advanced machine learning algorithms, it deeply learns and accurately predicts key factors such as long-term energy demand trends and seasonal changes. By grasping the information at the macro level, it provides a forward-looking and guiding strategic basis for the formulation of medium-term and short-term scheduling plans. The present invention makes full use of the high timeliness characteristics of short-term and real-time data to flexibly and dynamically adjust and optimize the scheduling plan, enabling the scheduling plan to more effectively cope with the interference of various uncertain factors in the energy system, such as the fluctuation of renewable energy output power and the sudden change of user load, and ensuring the stable operation of the energy system.

[0128] A method for multi-time scale optimal scheduling of a regional integrated energy system proposed by the present invention fully explores the value of multi-source data. With the powerful ability of machine learning, it opens up a new path for the optimal scheduling of the integrated energy system, effectively promotes the efficient operation and sustainable development of the energy system, makes the scheduling process more reasonable at the economic level, and at the same time greatly improves the utilization efficiency and consumption level of renewable energy in the system.

[0129] Figure 4 A block diagram of a computing device according to an exemplary embodiment of the present invention is shown.

[0130] As Figure 4 shown, the computing device 30 includes a processor 12 and a memory 14. The computing device 30 may further include a bus 22, a network interface 16, and an I / O interface 18. The processor 12, the memory 14, the network interface 16, and the I / O interface 18 may communicate with each other through the bus 22.

[0131] The processor 12 may include one or more general-purpose CPUs (Central Processing Units), microprocessors, or application-specific integrated circuits, etc., for executing relevant program instructions.

[0132] The memory 14 may include a machine system-readable medium in the form of volatile memory, such as random access memory (RAM), read-only memory (ROM), and / or cache memory. The memory 14 is used to store one or more programs containing instructions and data. The processor 12 can read the instructions stored in the memory 14 to execute the method according to the embodiments of the present invention described above.

[0133] The computing device 30 can also communicate with one or more networks through the network interface 16. The network interface 16 can be a wireless network interface.

[0134] The bus 22 can include an address bus, a data bus, a control bus, etc. The bus 22 provides a path for exchanging information between components.

[0135] It should be noted that in the specific implementation process, the computing device 30 may also include other components necessary for normal operation. In addition, those skilled in the art can understand that the above devices may also only include the components necessary to implement the solutions of the embodiments of this specification, and do not necessarily include all the components shown in the figure.

[0136] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical discs, DVDs, CD-ROMs, microdrives, and magneto-optical discs, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nano-systems (including molecular memory ICs), network storage devices, cloud storage devices, or any type of medium or device suitable for storing instructions and / or data.

[0137] The embodiments of the present invention also provide a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps of any one of the methods described in the above method embodiments.

[0138] Those skilled in the art can clearly understand that the technical solutions of the present invention can be implemented by means of software and / or hardware. The "units" and "modules" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field-programmable gate array, an integrated circuit, etc.

[0139] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0140] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0141] In several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some service interfaces. The indirect couplings or communication connections of the devices or units can be in electrical or other forms.

[0142] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0143] In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0144] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present invention.

[0145] In the above embodiments, the descriptions of the various embodiments each have their own emphasis. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0146] The above has specifically shown and described exemplary embodiments of the present invention. It should be understood that the present invention is not limited to the detailed structures, settings, or implementation methods described herein; on the contrary, the present invention is intended to cover various modifications and equivalent settings included within the spirit and scope of the appended claims.

[0147] Those skilled in the art can clearly understand that the technical solutions of the present invention can be implemented by means of software and / or hardware. The "units" and "modules" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field-programmable gate array, an integrated circuit, etc.

[0148] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0149] In the above embodiments, the descriptions of the various embodiments each have their own emphasis. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0150] In several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some service interfaces. The indirect couplings or communication connections of devices or units can be in electrical or other forms.

[0151] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0152] In addition, in each embodiment of the present invention, each functional unit can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0153] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present invention.

[0154] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0155] The above specifically shows and describes the exemplary embodiments of the present invention. It should be understood that the present invention is not limited to the detailed structures, setting manners, or implementation methods described herein; on the contrary, the present invention is intended to cover various modifications and equivalent settings included within the spirit and scope of the appended claims.

Claims

1. A method for multi-time scale optimal scheduling of a regional integrated energy system, characterized in that Including: Obtaining data of each energy system; Constructing an optimization model for the regional integrated energy system, and the optimization model performs day-ahead optimal scheduling, intra-day rolling optimal scheduling, and real-time optimal scheduling on the regional integrated energy system in sequence; Training and solving the optimization model of the regional integrated energy system through a training data set; Performing multi-time scale optimal scheduling of the regional integrated energy system by using the trained model.

2. The method according to claim 1, characterized in that The data includes: Equipment data of each energy region; Historical load prediction data and actual data on both sides of the source and load; Grid and gas network price signal data.

3. The method according to claim 1, wherein the regional integrated energy system includes: A dispatching center and each single integrated energy system; The single integrated energy systems are interconnected with each other through power lines and natural gas pipelines.

4. The method according to claim 2, wherein Training and solving the optimization model of the regional integrated energy system through a training data set includes: Integrating the load prediction data on both sides of the source and load with the actual data, and checking and cleaning the quality of the price signal data; Extracting statistical features from the load prediction data on both sides of the source and load and the price signal data; Performing standardization or normalization processing on the data; Using the processed data to train and solve the optimization model.

5. The method according to claim 4, characterized in that, The optimization model includes Q-Learning, K-Means, decision tree, and / or linear regression model.

6. The method according to claim 5, wherein Using the processed data to train and solve the optimization model includes: Training for the Q–Learning model: Defining the state space, action space, and reward function, and performing iterative training through time steps, and making decisions based on the load prediction data on both sides of the source and load and the price signal to achieve model training; and / or Training for the K-Means model: Assigning the load prediction data on both sides of the source and load and the price signal data points to the cluster centers, calculating the mean of the data points in each cluster until the cluster centers reach the convergence condition; and / or Training for the decision tree model: Selecting the first feature as the splitting point by calculating the information gain or Gini index index of each feature; and / or Training for the linear regression model: Adjusting the coefficients of the optimization model of the regional integrated energy system by using the least squares method or the gradient descent algorithm Case number: 241214CI Performing day-ahead optimal scheduling on the regional integrated energy system includes:

7. The method according to claim 1, characterized in that Taking the reduction of the daily operation cost of the regional integrated energy system as the goal; Optimizing the entire region with energy purchase constraints, equipment operation constraints, transmission power constraints, and energy balance constraints as conditions; The time resolution of the day-ahead optimal scheduling is 1 to n hours, n is less than 2, and the control time domain length is 24 hours. Performing intra-day rolling optimal scheduling on the regional integrated energy system includes:

8. The method according to claim 1, wherein Taking the reduction of the daily operation cost of the single integrated energy system within the region as the goal; Taking energy purchase constraints, equipment operation constraints, and energy balance constraints as constraint conditions; The time resolution of the intra-day rolling optimal scheduling is 15 to m minutes, m is less than 30, and the control time domain length is 2 hours. Performing real-time optimal scheduling on the regional integrated energy system includes:

9. The method according to claim 1, wherein ​ Targeting the first cost of the operation of a single integrated energy system considering the equipment fluctuation penalty cost; Real-time update of energy load prediction data and adjustment of equipment power; The time resolution of the real-time optimal scheduling is 5 to p minutes, where p is less than 10, and the control time domain length is 15 minutes.

10. A computing device, characterized in that, Including: A processor; And A memory storing a computer program, which when executed by the processor causes the processor to execute the method according to any one of claims 1-9.

Images