Multi-objective real-time operation optimization scheduling method and system for microgrids

Abstract

This invention provides a multi-objective real-time operation optimization scheduling method and system for microgrids, relating to the field of power dispatching technology. The invention employs a composite prediction model combining TCN and GCN to reduce the impact of supply and demand uncertainties. The supply and demand prediction model integrates TCN and GCN, effectively handling periodic and trend components in the data, and using noise filtering technology to process random components, significantly improving prediction accuracy. This makes it more adaptable to the complexity of power demand and renewable energy output fluctuations, better reducing the impact of supply and demand uncertainties on microgrid optimal scheduling and improving the accuracy of microgrid optimal scheduling.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, specifically to a multi-objective real-time operation optimization dispatching method and system for microgrids. Background Technology

[0002] Microgrids, as small-scale power systems that organically combine distributed power sources (such as solar photovoltaic power generation and wind power generation), energy storage devices, and loads, have received widespread attention and application.

[0003] However, the output power of distributed generation in microgrids is intermittent, fluctuating, and uncertain, while load demand also varies randomly, resulting in low accuracy of microgrid operation optimization. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a multi-objective real-time operation optimization and scheduling method and system for microgrids, which solves the technical problem of low accuracy in microgrid operation optimization in existing technologies.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a multi-objective real-time operation optimization scheduling method for microgrids, comprising:

[0009] Raw load data, renewable energy output data, and meteorological data are acquired and preprocessed to obtain time series data.

[0010] The time series data is decomposed into sequences, and features are extracted and optimized based on the decomposed dataset to obtain periodic features, trend features, and random features that are strongly correlated with the target predictor variable.

[0011] Periodic features are processed by a pre-built temporal convolutional network model to obtain periodic prediction results; trend features are processed by a pre-built graph convolutional network model to obtain trend prediction results; and random features are processed using noise filtering techniques to obtain random prediction results.

[0012] Based on the periodic, trend, and stochastic prediction results, a microgrid operation optimization scheduling model is constructed and solved using a multi-objective particle swarm optimization algorithm. The microgrid is then optimized and scheduled based on the solution results.

[0013] Preferably, the step of performing sequence decomposition on the time series data includes:

[0014] Using seasonal decomposition time series technology, time series data is decomposed into three parts: trend component, seasonal component, and random component, as expressed below:

[0015] Y t =T t +S t +R t

[0016] Among them, Y t It is the observation value of time series data at time point t, where T is the time series data. t It is a trend component, S t It is a seasonal ingredient, R t It is a random component.

[0017] Preferably, the pre-built temporal convolutional network model includes: an input layer, multiple convolutional layers, and an output layer; wherein the input layer is used to receive periodic features, the multiple convolutional layers are used to capture temporal patterns in the periodic features, and the output layer is used to output periodic prediction results.

[0018] Preferably, the pre-built graph convolutional network model includes: an input layer, multiple graph convolutional layers, and an output layer; wherein the input layer is used to receive trend features, the multiple graph convolutional layers are used to capture the interdependent relationships between trend features, and the output layer is used to output trend prediction results.

[0019] Preferably, the microgrid multi-objective real-time operation optimization scheduling method further includes: using a near-end strategy optimization algorithm to perform real-time dynamic optimal energy management, and optimizing and updating the microgrid operation optimization scheduling model.

[0020] Preferably, the microgrid operation optimization scheduling model includes an objective function and constraints, wherein the objective function includes:

[0021] Cost minimization objective function:

[0022] Emissions minimization objective function:

[0023] The objective function for maximizing energy efficiency can be transformed into a minimization form:

[0024]

[0025] Where, N G It is the number of generators, C i P is the unit power generation cost of the i-th generator. i It is its output power, N s It refers to the number of energy storage devices, M. j Q is the operation and maintenance cost of the j-th energy storage device.j It refers to its charging and discharging capacity; transaction cost C includes the cost of buying and selling electricity with the main grid; E i N is the carbon dioxide emission per unit of electricity generated by the i-th generator; R It is the number of renewable energy generators, R k This is the energy efficiency of the k-th renewable energy generator;

[0026] Among them, P i =P per +P trend +P random P per ,P trend ,P random ∈(P t ∪P s ∪P r That is, the output power includes the periodic prediction result of the energy output power obtained based on the time convolutional network model, the corresponding trend prediction result obtained based on the graph convolutional network model, and the corresponding random prediction result after noise reduction; P t For periodic prediction results; P s This is a trend prediction result; P r This is a random prediction result;

[0027] Q j =Q per +Q trend +Q random Q per Q trend Q random ∈(P t ∪P s ∪P r That is, the charging and discharging quantities include the periodic prediction results of the energy storage system's charging and discharging quantities obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction.

[0028] E i =E per +E trend +E random E per E trend E random ∈(P t ∪P s ∪P r The carbon dioxide emissions per unit of electricity generated include the periodic prediction results of the carbon dioxide emissions of the generator obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding stochastic prediction results after noise reduction.

[0029] The constraints include:

[0030] Power balance constraint: P L (t)=P′ pv (t)+P′ W (t)+P grid (t)+P di (t)+P gas (t)+P ba (t)

[0031] Output constraints of diesel generators:

[0032] Output constraints of micro gas turbines:

[0033] Power constraints for power exchange between the main microgrid and the microgrid:

[0034] Energy storage device lifespan constraints:

[0035] in, These represent the upper and lower limits of the output power of the diesel generator and the micro gas turbine, respectively. di r gas These represent the upper and lower limits of the output power of the diesel generator and the ramping power of the micro gas turbine, respectively. These are the upper and lower limits of the power transmitted via the tie line. State of Charge (SOC) represents the upper and lower limits of the energy storage device's output power; its positive or negative value indicates the power input and output. min (t), SOC max (t) represents the upper and lower limits of the energy storage capacity at time t.

[0036] Secondly, the present invention provides a microgrid multi-objective real-time operation optimization scheduling system, comprising:

[0037] The data processing module is used to acquire raw load data, renewable energy output data, and meteorological data, and to preprocess them to obtain time series data.

[0038] The feature extraction and optimization module is used to decompose time series data and extract and optimize features based on the decomposed dataset to obtain periodic features, trend features and random features that are strongly correlated with the target predictor variable.

[0039] The prediction module is used to process periodic features through a pre-built temporal convolutional network model to obtain periodic prediction results; to process trend features through a pre-built graph convolutional network model to obtain trend prediction results; and to process random features using noise filtering techniques to obtain random prediction results.

[0040] The optimization scheduling module is used to construct an optimized scheduling model for microgrid operation based on periodic, trend, and stochastic prediction results, solve it using a multi-objective particle swarm optimization algorithm, and perform optimized scheduling of the microgrid based on the solution results.

[0041] Preferably, the microgrid multi-objective real-time operation optimization and scheduling system further includes:

[0042] The model adjustment module is used to perform real-time dynamic optimal energy management and optimize and update the microgrid operation optimization scheduling model by employing a near-end strategy optimization algorithm.

[0043] Thirdly, the present invention provides a computer-readable storage medium storing a computer program for a multi-objective real-time operation optimization scheduling method for microgrids, wherein the computer program causes a computer to execute the multi-objective real-time operation optimization scheduling method for microgrids as described above.

[0044] Fourthly, the present invention provides an electronic device, comprising:

[0045] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing a microgrid multi-objective real-time operation optimization scheduling method as described above.

[0046] (III) Beneficial Effects

[0047] This invention provides a method and system for multi-objective real-time operation optimization scheduling of microgrids. Compared with existing technologies, it has the following advantages:

[0048] This invention employs a composite forecasting model combining TCN and GCN to reduce the impact of supply and demand uncertainties. The supply and demand forecasting model integrates TCN and GCN, effectively handling periodic and trending components in the data, and using noise filtering technology to process random components, significantly improving forecast accuracy. This makes it more adaptable to the complexity of fluctuations in electricity demand and renewable energy output, better mitigating the impact of supply and demand uncertainties on microgrid optimal scheduling and improving the accuracy of microgrid optimal scheduling. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a block diagram of a multi-objective real-time operation optimization scheduling method for microgrids according to an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] This application provides a multi-objective real-time operation optimization scheduling method and system for microgrids, which solves the technical problem of low accuracy in microgrid operation optimization in the prior art, achieves accurate supply and demand forecasting, reduces the impact of supply and demand uncertainty on the optimization scheduling model, and improves the accuracy of microgrid operation optimization.

[0053] The technical solution in this application is to solve the above-mentioned technical problems, and the general idea is as follows:

[0054] The output power of distributed generation in microgrids is characterized by intermittency, volatility, and uncertainty, while load demand also varies randomly. This poses a significant challenge to the stable operation and optimal scheduling of microgrids, as specifically manifested in the following ways:

[0055] First, accurate supply and demand forecasting can better mitigate the impact of supply and demand uncertainties and improve the accuracy of microgrid optimal scheduling. Commonly used machine learning algorithms, such as Recurrent Neural Networks (RNNs), Support Vector Machines (SVMs), and Long Short-Term Memory Networks (LSTMs), have certain capabilities in capturing trends, seasonal variations, and periodic features of load data, thus outputting relatively accurate forecast results. However, these single forecasting methods also have their own limitations. These limitations become increasingly apparent when dealing with large-scale, high-dimensional, and structurally complex microgrid data. These algorithms typically require extremely cumbersome and time-consuming feature engineering processes to extract key information valuable for forecasting, and their generalization ability in handling nonlinear problems is insufficient, making it difficult to comprehensively capture deep-seated features and temporal dependencies within the data. Furthermore, their computational efficiency is relatively low, making it difficult to meet the stringent real-time requirements of microgrid forecasting tasks, and they are also prone to gradient vanishing or exploding problems, which undoubtedly greatly increases the complexity of model training and the uncertainty of the results. Therefore, to better address the complexity, intermittency, and volatility of diverse and heterogeneous data, different prediction models can be combined to complement each other's strengths and reduce the errors caused by a single prediction model. Furthermore, composite prediction can dynamically adjust model combinations or parameters according to different operating scenarios, enabling the prediction model to better adapt to various complex and changing operating environments and enhancing its adaptability.

[0056] Secondly, in the area of ​​multi-objective optimal scheduling of microgrids, the optimal scheduling of microgrids involves multiple objectives such as economic costs and environmental governance costs. These objectives often have complex nonlinear relationships and conflicts. While the traditional linear weighted summation method can transform a multi-objective problem into a single-objective problem for solution, the allocation of weights often relies on the decision-maker's subjective judgment, lacking objective basis and easily leading to instability and bias in the optimization results. Furthermore, information may be lost during normalization, further affecting the accuracy of the optimization results. On the other hand, while the Pareto optimal solution set method can provide a set containing multiple possible solutions, offering decision-makers a choice space, the solution set can become extremely large when the objective space is too complex or high-dimensional, making selection difficult. Moreover, the Pareto optimal solution set method cannot assign specific weights to each solution, meaning that some solutions more meaningful to the decision-maker may not be highlighted. Simultaneously, its computational complexity is high, especially when dealing with large-scale or high-dimensional problems, resulting in high solution and time costs.

[0057] Finally, over time and with the continuous changes in the microgrid's operating environment, such as fluctuations in renewable energy generation and changes in load demand, the accuracy of existing forecasting models may decline. Simultaneously, optimization methods rely on accurate forecasting models to formulate effective optimization strategies; inaccurate forecasting models will directly lead to the failure of optimization strategies, thereby weakening the optimization effect. Furthermore, from an economic perspective, failing to adjust models may also impair the microgrid's economics, as existing forecasting and optimization models may not accurately reflect changes in key factors such as electricity prices, energy costs, and equipment maintenance expenses. Finally, in the face of sudden disturbances or extreme situations, forecasting and optimization models that have not been adjusted in a timely manner may lack sufficient robustness and be unable to effectively cope with these challenges, thus jeopardizing the overall system performance. Therefore, regular model adjustments are crucial for ensuring the accuracy, economy, and robustness of microgrid management systems.

[0058] As can be seen from the above description, the existing microgrid operation optimization and scheduling mainly has the following shortcomings:

[0059] 1. Microgrid operation optimization and scheduling models have shortcomings in addressing supply and demand uncertainties. The randomness and volatility of renewable energy sources, as well as the uncertainty of load demand, increase the difficulty of microgrid operation optimization and scheduling, negatively impacting the accuracy of microgrid operation optimization.

[0060] 2. In the multi-objective optimization scheduling of microgrids, the target of renewable energy utilization rate is rarely considered, making it difficult to maximize the use of renewable energy, which will further lead to energy waste and affect the sustainable application of renewable energy in microgrids.

[0061] 3. Existing microgrid optimal dispatch strategies lack dynamic model updates, resulting in insufficient robustness in responding to sudden disturbances or extreme situations. Fluctuations in renewable energy generation and continuous changes in load demand lead to a decrease in the accuracy of microgrid optimal dispatch. Furthermore, microgrid optimal dispatch models cannot adjust in a timely manner to sudden disturbances or extreme situations.

[0062] To address the aforementioned issues, this invention proposes a multi-objective real-time operation optimization scheduling method and system for microgrids based on supply and demand forecasting. Firstly, to reduce the impact of supply and demand uncertainty on the optimization scheduling model, accurate supply and demand forecasting is necessary. This invention integrates the advantages of Temporal Convolutional Networks (TCNs) and Graph Convolutional Networks (GCNs) to construct a composite forecasting model covering both supply and demand sides. The combination of TCNs' deep mining capabilities for time-series data and GCNs' efficient extraction capabilities for spatial structure information significantly improves the accuracy of supply and demand forecasting, providing a solid foundation for the coordinated and optimized operation of microgrids' source-grid-load-storage systems.

[0063] Subsequently, this embodiment of the invention constructs a multi-objective operation optimization scheduling model, comprehensively considering three objectives: microgrid operating cost, emission level, and renewable energy utilization rate. This ensures that the microgrid maintains a highly efficient, stable, and low-cost state under various operating scenarios, while maximizing the utilization of renewable energy. Finally, to enable real-time monitoring and analysis of the system's operating status and to rapidly respond to complex and changing environmental conditions and market demand changes, this embodiment of the invention also adjusts the model. Model adjustments further improve the system's operating efficiency, resilience and rapid recovery capability in the face of emergencies, and its sustainable benefits to the ecological environment and society. This allows for a rapid response to complex and changing environments, enhancing the system's adaptability and robustness in such conditions.

[0064] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0065] This invention provides a multi-objective real-time operation optimization scheduling method for microgrids, such as... Figure 1 As shown, it includes:

[0066] S1. Acquire raw load data, renewable energy output data, and meteorological data, and preprocess them to obtain time series data;

[0067] S2. Perform sequence decomposition on the time series data, and extract and optimize features based on the decomposed dataset to obtain periodic features, trend features, and random features that are strongly correlated with the target predictor variable.

[0068] S3. Periodic features are processed using a pre-built temporal convolutional network model to obtain periodic prediction results; trend features are processed using a pre-built graph convolutional network model to obtain trend prediction results; and random features are processed using noise filtering techniques to obtain random prediction results.

[0069] S4. Based on the periodic, trend, and stochastic prediction results, construct a microgrid operation optimization scheduling model and solve it using a multi-objective particle swarm optimization algorithm. Perform microgrid optimization scheduling based on the solution results.

[0070] This invention employs a composite forecasting model combining TCN and GCN to reduce the impact of supply and demand uncertainties. The supply and demand forecasting model combines TCN and GCN, effectively handling periodic and trend components in the data, and using noise filtering technology to handle random components, significantly improving forecast accuracy. This makes it more adaptable to the complexity of fluctuations in electricity demand and renewable energy output, better reducing the impact of supply and demand uncertainties on microgrid optimal scheduling and improving the accuracy of microgrid optimal scheduling.

[0071] In one embodiment, S1, raw load data, renewable energy output data, and meteorological data are acquired and preprocessed to obtain time series data. Specifically, this includes:

[0072] Collect raw load data, renewable energy output data (such as solar and wind power), and meteorological data (such as temperature, wind speed, and humidity).

[0073] Data sources include, but are not limited to, distributed photovoltaic power generation systems, wind power generation systems, energy storage systems (such as batteries), load equipment (such as household electricity and commercial electricity), and grid connections.

[0074] The system regularly collects real-time data from photovoltaic panels, wind turbines, energy storage devices, and load equipment, including power generation, power consumption, and battery charging and discharging status, and obtains weather information such as temperature, wind speed, and humidity from meteorological stations.

[0075] The collected data is cleaned to remove outliers (such as data points outside the normal range) and duplicates (such as recording the same data multiple times at the same time point), resulting in time series data. The cleaning process ensures the accuracy and consistency of the data, providing a reliable data foundation for subsequent analysis.

[0076] In one embodiment, S2, the time series data is decomposed, and features are extracted and optimized based on the decomposed dataset to obtain periodic features, trend features, and random features that are strongly correlated with the target predictor variable. Specifically, this includes:

[0077] Identify and separate the periodic, trend, and random components in time series data. Use Seasonal Decomposition Time Series (STL) or other decomposition techniques to decompose the original data into three parts:

[0078] Periodic components: reflect the periodic changes in data over time.

[0079] Trend components: Reflect the long-term changing trend of data over time.

[0080] Randomness component: Represents the uncertainty in the data.

[0081] Time series data is decomposed into three parts: trend, seasonality, and residuals. The residuals are the random component.

[0082] Y t =T t +S t +R t

[0083] Among them, Yt It is the observation value (Y) of the original time series data at time point t. t This could be the power load, renewable energy generation, energy storage system status, or grid connection status at time point t. t It is a trend component, S t It is a seasonal ingredient, R t It is the residual (random component).

[0084] Seasonal time series STL decomposition steps:

[0085] The trend component T is estimated from time series data using the Loess smoothing algorithm. t and seasonal component S t ,

[0086] R t The residuals are calculated using the following formula:

[0087] R t =Y t -(T t +S t )

[0088] Feature engineering techniques are used to select and optimize the extracted features. Features strongly correlated with the target predictor variables are selected through methods such as correlation analysis. Focus is placed on features closely related to load forecasting and renewable energy production forecasting, such as historical load, historical renewable energy production, temperature, and wind speed. Optimization steps:

[0089] Feature selection: Feature selection algorithms (such as Recursive Feature Emission (RFE)) are used to determine the most important features. The Recursive Feature Emission (RFE) method is as follows:

[0090] The initial feature set F contains all features.

[0091] Select a model M (linear regression) and an evaluation index E (mean squared error).

[0092] Recursively remove one or more features and retrain the model M.

[0093] Select the feature subset that best performs the model M. This feature subset includes periodic features, trend features, and random features.

[0094] Further processing of the selected features, such as normalization and standardization, is performed to scale the features to the same scale, resulting in periodic features F that are strongly correlated with the target predictor variable. t Trend characteristics F s and randomness characteristic F r .

[0095] In one embodiment, S3, F is processed by a pre-built time-in-time convolutional network model. t After processing, the periodic prediction result P is obtained. t ; Using a pre-built graph convolutional network model to apply F s After processing, the trend prediction result P is obtained. s Using noise filtering technology for F r The process is performed to obtain the randomness prediction result P. r Specifically, this includes:

[0096] A pre-built Temporal Convolutional Network (TCN) model consists of an input layer, multiple convolutional layers, and an output layer. The input layer receives the input signal (F). t Multiple convolutional layers are used to capture temporal patterns in periodic features, and the output layer outputs the periodic prediction results. This model can effectively capture the temporal dependencies in time series data and is suitable for predicting periodically changing data.

[0097] Input layer receives F t , where F t The periodic features at time t that are strongly correlated with the target predictor variable are obtained through feature selection and optimization.

[0098] In the convolutional operation of layer l, causal convolution is used to ensure that the model only sees data prior to the current time point. The convolutional operation can be represented as:

[0099]

[0100] in, It is the output of the l-th layer. F is the weight of the convolutional kernel in the l-th layer. t (tk) represents the periodic characteristic of strong correlation of the target predictor variables at time tk (i.e., a historical value of the periodic characteristic of strong correlation of the target predictor variables at time t), b (l) is the bias term of the l-th layer, and K is the size of the convolution kernel.

[0101] To mitigate the vanishing gradient problem, TCN uses residual connections. Assume... The input of layer l is the residual connection, and the output can be expressed as:

[0102]

[0103] Among them, Y t (l) It is the final output of the l-th layer. These are the features input to this layer.

[0104] The output after residual concatenation is nonlinearly transformed using an activation function (such as ReLU):

[0105]

[0106] The final output layer is a stack of multiple convolutional layers, which outputs the prediction result P. t :

[0107]

[0108] Where L is the index of the last layer, and f is the activation function of the output layer (such as a linear activation function).

[0109] Pre-built graph convolutional network (GCN) models can capture the dependencies between nodes and are suitable for predicting trending data. The model includes an input layer, multiple graph convolutional layers, and an output layer.

[0110] Input layer receives F s , where F s These are trend features that are strongly correlated with the target predictor variable, obtained through feature selection and optimization.

[0111] Let A be the adjacency matrix of the graph, representing the connection relationships between nodes, and D be the degree matrix, representing the number of connections for each node.

[0112] In the graph convolution operation at layer l, the node representation update can be represented as:

[0113]

[0114] Among them, H (l) H is the node representation matrix of the l-th layer. (l-1) W is the node representation matrix of the (l-1)th layer. (l) A is the weight matrix of the l-th layer. ～ =A+I is the adjacency matrix plus a self-loop (I is the identity matrix), D ～ It is A ～ The degree matrix is ​​σ, where σ is the activation function (such as ReLU).

[0115] The final output layer will be a stack of multiple graph convolutional layers, outputting the prediction result P. s :

[0116] P s =f(H (L) )

[0117] Where L is the index of the last layer, and f is the activation function of the output layer (such as a linear activation function).

[0118] F r It is treated as a noise term in the model to reduce its impact on the prediction results.

[0119] Use noise filtering techniques (such as smoothing filters) to handle randomness features.

[0120] Use a Savitzky-Golay filter to smooth F r Its formula is:

[0121]

[0122] Among them, F r It is a random feature that is strongly correlated with the target predictor variable, obtained through feature selection and optimization. j These are the filter coefficients (usually calculated using the least squares method), and m is half the size of the filter window (determining the degree of smoothing).

[0123] In one embodiment, S4, based on periodic prediction results, trend prediction results, and stochastic prediction results, a microgrid operation optimization scheduling model is constructed and solved using a multi-objective particle swarm optimization algorithm. Microgrid optimization scheduling is then performed based on the solution results. Specifically, this includes:

[0124] Cost minimization objective function:

[0125] Where, N G It is the number of generators, C i P is the unit power generation cost of the i-th generator. i It is its output power, N s It refers to the number of energy storage devices, M. j Q is the operation and maintenance cost of the j-th energy storage device. j It refers to its charging and discharging capacity, and the transaction cost C includes the cost of buying and selling electricity with the main grid.

[0126] Among them, P i =P per +P trend +P random P per ,P trend ,P random ∈(P t ∪P s ∪P r That is, the output power includes the periodic prediction results of energy output power obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction.

[0127] Q j =Q per +Q trend +Q random Q per Q trend Qrandom ∈(P t ∪P s ∪P r That is, the charging and discharging quantities include the periodic prediction results of the energy storage system's charging and discharging quantities obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction.

[0128] Emissions minimization objective function:

[0129] Among them, E i It is the carbon dioxide emission per unit of electricity generated by the i-th generator.

[0130] illustrate:

[0131] E i =E per +E trend +E random E per E trend E random ∈(P t ∪P s ∪P r The carbon dioxide emissions per unit of electricity generated include the periodic prediction results of the carbon dioxide emissions of the generator obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding stochastic prediction results after noise reduction.

[0132] The objective function for maximizing energy efficiency (transformed into minimization form):

[0133] Where, N R It is the number of renewable energy generators, R k It is the energy efficiency of the k-th renewable energy generator.

[0134] Optimize scheduling constraints:

[0135] Power balance constraint: P L (t)=P′ pv (t)+P′ W (t)+P grid (t)+P di (t)+P gas (t)+P ba (t)

[0136] Output constraints of diesel generators:

[0137] Output constraints of micro gas turbines:

[0138] Power constraints for power exchange between the main microgrid and the microgrid:

[0139] Energy storage device lifespan constraints:

[0140] in, These represent the upper and lower limits of the output power of the diesel generator and the micro gas turbine, respectively. di r gas These represent the upper and lower limits of the output power of the diesel generator and the ramping power of the micro gas turbine, respectively. These are the upper and lower limits of the power transmitted via the tie line. State of Charge (SOC) represents the upper and lower limits of the energy storage device's output power; its positive or negative value indicates the power input and output. min (t), SOC max (t) represents the upper and lower limits of the energy storage capacity at time t, ensuring its service life.

[0141] In this embodiment of the invention, the MOPSO algorithm is used to solve the microgrid operation optimization scheduling model, specifically including:

[0142] In the MOPSO algorithm, updating the position and velocity of particles is a key step, and the fitness value of each particle needs to be calculated based on the objective function.

[0143] Position: For each particle, its position can be represented as a vector, where each element corresponds to a decision variable (such as the output power of a generator, the charging and discharging amount of an energy storage device, etc.).

[0144] Let the position of particle p be X p ,but

[0145] Velocity: The velocity of a particle is also a vector, representing the change of each decision variable in each iteration. Let the velocity of particle p be V. p ,but

[0146] Location update:

[0147] Speed ​​update: Speed ​​is updated according to the standard formula of the PSO algorithm, such as the influence of inertia weight, cognitive part and social part.

[0148] Update the individual optimal position and the global optimal position (Pareto front):

[0149] For each particle, compare its current position with its optimal position. If the current position is better, update the optimal position.

[0150] For the globally optimal position, the Pareto front concept is used for updating, that is, all non-dominated solutions are retained to form a new Pareto front.

[0151] It should be noted that, in the specific implementation process, in order to enable the microgrid operation optimization scheduling model to adapt quickly to new data distributions and environmental changes and maintain its prediction and optimization accuracy, the embodiments of the present invention also include: S5, using a near-end strategy optimization algorithm to perform real-time dynamic optimal energy management, and optimizing and updating the microgrid operation optimization scheduling model.

[0152] Specifically, it includes:

[0153] For model updates and optimization, the Proximal Policy Optimization (PPO) algorithm is employed for real-time dynamic optimal energy management (OEM). Leveraging its ability to learn from historical data, the PPO algorithm accurately captures the uncertainties in renewable energy generation and load consumption, continuously updating and optimizing the model to ensure accurate and efficient energy management. Incremental learning and transfer learning techniques are also used to enable the model to quickly adapt to new data distributions and environmental changes, maintaining its predictive and optimization accuracy. For performance evaluation, key indicators such as Expected Energy Deficiency (EENS) and Probability of Load Loss (LOLP) are used to comprehensively examine the microgrid's operational effectiveness. Furthermore, the model's predictive performance is regularly and meticulously evaluated across multiple dimensions, including prediction error and stability, to ensure continuous system optimization and improvement.

[0154] Its optimizations mainly include:

[0155] Hyperparameter tuning: Adjust the model's hyperparameters, such as learning rate, regularization parameter, and tree depth, through methods such as grid search, random search, or Bayesian optimization.

[0156] Feature selection: Select the features that have the greatest impact on the prediction results, and remove noise and irrelevant features.

[0157] Feature optimization: Create new features or transform existing features to better capture patterns in the data.

[0158] This invention also provides a microgrid multi-objective real-time operation optimization scheduling system, comprising:

[0159] The data processing module is used to acquire raw load data, renewable energy output data, and meteorological data, and to preprocess them to obtain time series data.

[0160] The feature extraction and optimization module is used to decompose time series data and extract and optimize features based on the decomposed dataset to obtain periodic features, trend features and random features that are strongly correlated with the target predictor variable.

[0161] The prediction module is used to process periodic features through a pre-built temporal convolutional network model to obtain periodic prediction results; to process trend features through a pre-built graph convolutional network model to obtain trend prediction results; and to process random features using noise filtering techniques to obtain random prediction results.

[0162] The optimization scheduling module is used to construct an optimized scheduling model for microgrid operation based on periodic, trend, and stochastic prediction results, solve it using a multi-objective particle swarm optimization algorithm, and perform optimized scheduling of the microgrid based on the solution results.

[0163] In the specific implementation process, the microgrid multi-objective real-time operation optimization scheduling system also includes a model adjustment module, which is used to perform real-time dynamic optimal energy management and optimize and update the microgrid operation optimization scheduling model by adopting a near-end strategy optimization algorithm.

[0164] It is understood that the microgrid multi-objective real-time operation optimization scheduling system provided in this embodiment of the invention corresponds to the above-mentioned microgrid multi-objective real-time operation optimization scheduling method. The explanation, examples, and beneficial effects of the relevant content can be referred to the corresponding content in the microgrid multi-objective real-time operation optimization scheduling method, and will not be repeated here.

[0165] This invention also provides a computer-readable storage medium storing a computer program for multi-objective real-time operation optimization scheduling of a microgrid, wherein the computer program causes a computer to execute the microgrid multi-objective real-time operation optimization scheduling method as described above.

[0166] This invention also provides an electronic device, including: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing the microgrid multi-objective real-time operation optimization scheduling method as described above.

[0167] In summary, compared with existing technologies, it has the following beneficial effects:

[0168] 1. The embodiments of the present invention use a composite prediction model of TCN and GCN for prediction, which reduces the impact of supply and demand uncertainty. The supply and demand prediction model combines TCN and GCN, effectively handles the periodic and trend components in the data, and handles the random components through noise filtering technology, which greatly improves the prediction accuracy, making it more adaptable to the complexity of power demand and renewable energy output fluctuations, better reducing the impact of supply and demand uncertainty on microgrid optimal scheduling, and improving the accuracy of microgrid optimal scheduling.

[0169] 2. This invention proposes a microgrid operation optimization scheduling model that minimizes overall costs, emissions, and renewable energy utilization. This model better ensures that microgrids maintain high efficiency, stability, and low cost under various operating scenarios while maximizing the utilization of renewable energy.

[0170] 3. This invention utilizes a near-end strategy optimization algorithm to continuously update and optimize the model, ensuring accurate and efficient energy management. When faced with sudden disturbances or extreme situations, the real-time adjusted prediction and optimization model can respond rapidly, exhibiting stronger robustness and ensuring that the overall system performance remains unaffected.

[0171] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0172] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-objective real-time operation optimization scheduling method for microgrids, characterized in that, include: Raw load data, renewable energy output data, and meteorological data are acquired and preprocessed to obtain time series data. The time series data is decomposed into sequences, and features are extracted and optimized based on the decomposed dataset to obtain periodic features, trend features, and random features that are strongly correlated with the target predictor variable. Periodic features are processed by a pre-built temporal convolutional network model to obtain periodic prediction results; trend features are processed by a pre-built graph convolutional network model to obtain trend prediction results. Noise filtering techniques are used to process the randomness characteristics to obtain randomness prediction results; Based on the periodic, trend, and stochastic prediction results, a microgrid operation optimization scheduling model is constructed and solved using a multi-objective particle swarm optimization algorithm. Microgrid optimization scheduling is then performed based on the solution results. The microgrid operation optimization scheduling model includes an objective function and constraints. The objective function includes: Cost minimization objective function: Emissions minimization objective function: The objective function for maximizing energy efficiency can be transformed into a minimization form: Where, N G It is the number of generators, C i P is the unit power generation cost of the i-th generator. i It is its output power, N s It refers to the number of energy storage devices, M. j Q is the operation and maintenance cost of the j-th energy storage device. j It refers to its charging and discharging capacity; transaction cost C includes the cost of buying and selling electricity with the main grid; E i N is the carbon dioxide emission per unit of electricity generated by the i-th generator; R It is the number of renewable energy generators, R k This is the energy efficiency of the k-th renewable energy generator; Among them, P i =P per +P trend +P random P per ,P trend ,P random ∈(P t ∪P s ∪P r That is, the output power includes the periodic prediction result of the energy output power obtained based on the time convolutional network model, the corresponding trend prediction result obtained based on the graph convolutional network model, and the corresponding random prediction result after noise reduction; P t For periodic prediction results; P s This is a trend prediction result; P r This is a random prediction result; Q j =Q per +Q trend +Q random Q per Q trend Q random ∈(P t ∪P s ∪P r That is, the charging and discharging quantities include the periodic prediction results of the energy storage system's charging and discharging quantities obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction. E i =E per +E trend +E random E per E trend E random ∈(P t ∪P s ∪P r That is, the carbon dioxide emissions per unit of electricity generated include the periodic prediction results of the carbon dioxide emissions of the generator obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction. The constraints include: Power balance constraint: P L (t)=P′ pv (t)+P′ W (t)+P grid (t)+P di (t)+P gas (t)+P ba (t) Output constraints of diesel generators: Output constraints of micro gas turbines: Power constraints for power exchange between the main microgrid and the microgrid: Energy storage device lifespan constraints: in, These represent the upper and lower limits of the output power of the diesel generator and the micro gas turbine, respectively. di r gas These represent the upper and lower limits of the output power of the diesel generator and the ramping power of the micro gas turbine, respectively. These are the upper and lower limits of power transmitted via the tie line. State of Charge (SOC) represents the upper and lower limits of the energy storage device's output power; the sign of the value indicates the power input and output. min (t), SOC max (t) represents the upper and lower limits of the energy storage capacity at time t.

2. The microgrid multi-objective real-time operation optimization scheduling method as described in claim 1, characterized in that, The process of performing sequence decomposition on time series data includes: Using seasonal decomposition time series technology, time series data is decomposed into three parts: trend component, seasonal component, and random component, as expressed below: Y t =T t +S t +R t Among them, Y t It is the observation value of time series data at time point t, where T is the time series data. t It is a trend component, S t It is a seasonal ingredient, R t It is a random component.

3. The microgrid multi-objective real-time operation optimization scheduling method as described in claim 1, characterized in that, The pre-built temporal convolutional network model includes an input layer, multiple convolutional layers, and an output layer; wherein the input layer is used to receive periodic features, the multiple convolutional layers are used to capture the temporal patterns in the periodic features, and the output layer is used to output periodic prediction results.

4. The microgrid multi-objective real-time operation optimization scheduling method as described in claim 1, characterized in that, The pre-built graph convolutional network model includes an input layer, multiple graph convolutional layers, and an output layer. The input layer is used to receive trend features, the multiple graph convolutional layers are used to capture the interdependencies between trend features, and the output layer is used to output trend prediction results.

5. The microgrid multi-objective real-time operation optimization scheduling method as described in claim 1, characterized in that, The microgrid multi-objective real-time operation optimization scheduling method further includes: using a near-end strategy optimization algorithm to perform real-time dynamic optimal energy management, and optimizing and updating the microgrid operation optimization scheduling model.

6. A multi-objective real-time operation optimization and scheduling system for microgrids, characterized in that, include: The data processing module is used to acquire raw load data, renewable energy output data, and meteorological data, and to preprocess them to obtain time series data. The feature extraction and optimization module is used to decompose time series data and extract and optimize features based on the decomposed dataset to obtain periodic features, trend features and random features that are strongly correlated with the target predictor variable. The prediction module is used to process periodic features through a pre-built temporal convolutional network model to obtain periodic prediction results; to process trend features through a pre-built graph convolutional network model to obtain trend prediction results; and to process random features using noise filtering techniques to obtain random prediction results. The optimization scheduling module is used to construct a microgrid operation optimization scheduling model based on periodic forecast results, trend forecast results, and stochastic forecast results, and solve it using a multi-objective particle swarm optimization algorithm. Based on the solution results, the microgrid is optimized and scheduled. The microgrid operation optimization scheduling model includes an objective function and constraints. The objective function includes: Cost minimization objective function: Emissions minimization objective function: The objective function for maximizing energy efficiency can be transformed into a minimization form: Where, N G It is the number of generators, C i P is the unit power generation cost of the i-th generator. i It is its output power, N s It refers to the number of energy storage devices, M. j Q is the operation and maintenance cost of the j-th energy storage device. j It refers to its charging and discharging capacity; transaction cost C includes the cost of buying and selling electricity with the main grid; E i N is the carbon dioxide emission per unit of electricity generated by the i-th generator; R It is the number of renewable energy generators, R k This is the energy efficiency of the k-th renewable energy generator; Among them, P i =P per +P trend +P random P per ,P trend ,P random ∈(P t ∪P s ∪P r That is, the output power includes the periodic prediction result of the energy output power obtained based on the time convolutional network model, the corresponding trend prediction result obtained based on the graph convolutional network model, and the corresponding random prediction result after noise reduction; P t For periodic prediction results; P s This is a trend prediction result; P r This is a random prediction result; Q j =Q per +Q trend +Q random Q per Q trend Q random ∈(P t ∪P s ∪P r That is, the charging and discharging quantities include the periodic prediction results of the energy storage system's charging and discharging quantities obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction. E i =E per +E trend +E random E per E trend E random ∈(P t ∪P s ∪P r That is, the carbon dioxide emissions per unit of electricity generated include the periodic prediction results of the carbon dioxide emissions of the generator obtained based on the time convolutional network model, the corresponding trend prediction results obtained based on the graph convolutional network model, and the corresponding random prediction results after noise reduction. The constraints include: Power balance constraint: P L (t)=P′ pv (t)+P′ W (t)+P grid (t)+P di (t)+P gas (t)+P ba (t) Output constraints of diesel generators: Output constraints of micro gas turbines: Power constraints for power exchange between the main microgrid and the microgrid: Energy storage device lifespan constraints: in, These represent the upper and lower limits of the output power of the diesel generator and the micro gas turbine, respectively. di r gas These represent the upper and lower limits of the output power of the diesel generator and the ramping power of the micro gas turbine, respectively. These are the upper and lower limits of power transmitted via the tie line. State of Charge (SOC) represents the upper and lower limits of the energy storage device's output power; its positive or negative value indicates the power input and output. min (t), SOC max (t) represents the upper and lower limits of the energy storage capacity at time t.

7. The microgrid multi-objective real-time operation optimization scheduling system as described in claim 6, characterized in that, The microgrid multi-objective real-time operation optimization and scheduling system also includes: The model adjustment module is used to perform real-time dynamic optimal energy management and optimize and update the microgrid operation optimization scheduling model by employing a near-end strategy optimization algorithm.

8. A computer-readable storage medium, characterized in that, It stores a computer program for a multi-objective real-time operation optimization scheduling method for microgrids, wherein the computer program causes a computer to execute the multi-objective real-time operation optimization scheduling method for microgrids as described in any one of claims 1 to 5.

9. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing the microgrid multi-objective real-time operation optimization scheduling method as described in any one of claims 1 to 5.

Images