Source-network-load-storage integrated multi-energy complementary scheduling method and system

Abstract

The invention discloses a source, network, load and storage integrated multi-energy complementary scheduling method. The method comprises the following steps: acquiring related data of each link of a source, a network, a load and storage; performing feature extraction on the related data in combination with an FNN network and a GRU-BiLSTM-Transform network to obtain a comprehensive feature vector; constructing a multi-energy complementary scheduling model based on the comprehensive feature vector; and solving the multi-energy complementary scheduling model through an improved genetic algorithm to obtain a corresponding multi-energy complementary scheduling scheme. The method not only can comprehensively capture complex characteristics and dynamic changes of the system, but also can deeply excavate complementary advantages among multiple energy sources, and realizes more intelligent, flexible and reliable scheduling decisions.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and more specifically to a multi-energy complementary dispatching method and system integrating source, grid, load and storage. Background Art

[0002] With the diversification of energy structure and the development of intelligent power grid, multi-energy complementary dispatching has become an important means to improve energy utilization efficiency and ensure the safe and stable operation of power grid. However, the traditional single energy dispatching method can no longer meet the complex needs of modern power grid.

[0003] When dealing with the problem of multi-energy complementary scheduling, existing technologies often have defects such as incomplete data feature extraction, inaccurate model construction, and low efficiency of solution algorithms. Traditional data processing methods often focus on feature extraction of a single dimension, ignoring the complex correlation and dynamic changes between data, resulting in the difficulty of constructing a scheduling model to fully reflect the actual situation of the system. At the same time, traditional solution algorithms are unable to cope with multi-objective optimization and complex constraints, and are prone to premature convergence or falling into local optimal solutions, making it difficult to ensure the optimality and reliability of scheduling solutions.

[0004] In addition, existing technologies lack in-depth exploration and utilization of the complementary characteristics between multiple energy sources. Different energy types have their own advantages and disadvantages in terms of supply stability, cost-effectiveness and environmental performance, but existing technologies often fail to fully consider these differences and complementarities, resulting in a lack of pertinence and flexibility in scheduling decisions. At the same time, with the rapid development of smart grid and distributed energy technologies, the grid structure and operating environment are becoming increasingly complex and changeable, which puts higher requirements on the real-time, intelligent and scalable performance of the scheduling system. However, existing technologies still have major deficiencies in this regard and are difficult to meet the actual needs and development trends of modern power grids.

[0005] Therefore, how to design a multi-energy complementary scheduling method with integrated source, grid, load and storage that can fully capture the complex characteristics and dynamic changes of the system and achieve more intelligent, flexible and reliable scheduling decisions is an urgent problem that technical personnel in this field need to solve. Summary of the invention

[0006] In view of this, the present invention provides a multi-energy complementary scheduling method with integrated source, grid, load and storage. By combining the FNN network and the GRU-BiLSTM-Transformer network for comprehensive feature extraction, an accurate multi-energy complementary scheduling model is constructed, and an improved genetic algorithm is used for efficient solution, which effectively overcomes the defects of the prior art.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a multi-energy complementary scheduling method integrating source, grid, load and storage, comprising the following steps:

[0009] S1. Obtain relevant data of each link of source, network, load and storage;

[0010] S2. Extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector;

[0011] S3. Construct a multi-energy complementary scheduling model based on the comprehensive feature vector;

[0012] S4. Solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

[0013] Furthermore, in S1, the relevant data includes:

[0014] The real-time power and power forecast data of different types of power generation units at the source end, the operating status information of the power grid at the network end, the power demand and demand forecast data at the load end, and the current power storage and maximum charge and discharge rate data at the energy storage end.

[0015] Furthermore, the S2 includes:

[0016] S21, construct static attribute data for input to the FNN network and time series data for input to the GRU-BiLSTM-Transformer network;

[0017] S22, using FNN network and GRU-BiLSTM-Transformer network to extract static attribute features and time series features respectively;

[0018] S23, fusing the static attribute features and the time series features through a fusion layer, and outputting a comprehensive feature vector.

[0019] Furthermore, the FNN network includes: three linear layers;

[0020] Among them, each linear layer is connected to the BN layer and is provided with a ReLU activation function, and at least one linear layer is provided with a Dropout layer.

[0021] Furthermore, the GRU-BiLSTM-Transformer network includes:

[0022] Gated recurrent unit layer: used to preprocess time series data;

[0023] Bidirectional long short-term memory network layer: used to capture bidirectional dependencies in preprocessed time series data;

[0024] Embedding layer: used to map data into a high-dimensional space;

[0025] Multi-head attention mechanism: used to parallelize the calculation of multiple attention heads to enhance the attention to information at different locations;

[0026] Normalization layer: used to ensure that the data in each batch has a similar distribution;

[0027] Fully connected layer: used to process the results of the multi-head attention mechanism and output deep features.

[0028] Furthermore, in S3, the objective function of the multi-energy complementary scheduling model is expressed as:

[0029]

[0030] Where T represents the number of time steps in the scheduling cycle, N g Represents the number of power generation units, N s represents the number of energy storage devices, C i (P g,i,t ) represents the operating cost function of the i-th power generation unit at time t, C j (P s,i,t ) represents the charge and discharge cost function of the jth energy storage device at time t, C loss (P s,i,t ) represents the network transmission loss cost function, P g,i,t represents the output power of the i-th power generation unit at time t, P s,i,t represents the charge and discharge power of the jth energy storage device at time t, P l,t Represents the transmission power loss of the power grid at time t.

[0031] Furthermore, in S3, the constraints of the multi-energy complementary scheduling model include:

[0032] Power balance constraints:

[0033]

[0034] Among them, D t represents the load demand at time t, P g,i,t The output power of the i-th power generation unit at time t, P d,j,t represents the discharge power of the jth energy storage device at time t, P c,j,t represents the charging power of the jth energy storage device at time t;

[0035] Power generation unit output power limit:

[0036] P g,i,min ≤P g,i,t ≤P g,i,max

[0037] Among them, P g,i,min Represents the minimum output power of the power generation unit, P g,i,max Indicates the maximum output power of the power generation unit;

[0038] Energy storage system limitations:

[0039] E j,min ≤E j,t ≤E j,max

[0040] 0≤P c,j,t ≤P c,j,max

[0041] 0≤P d,j,t ≤P d,j,max

[0042] Among them, E j,t represents the power storage state of the jth energy storage device at time t, E j,min and E j,max Respectively represent the minimum and maximum storage capacity of the jth energy storage device, P c,j,max and P d,j,max denote the maximum charging and discharging rates of the j-th energy storage device, respectively;

[0043] Grid security constraints:

[0044] V min ≤V≤V max

[0045] I min ≤I≤I max

[0046] Where V represents the voltage intensity at time t, V min and V max represent the preset minimum and maximum voltage strengths, I represents the current strength at time t, and I min and I max Indicates the preset minimum and maximum current strengths respectively.

[0047] Furthermore, the S4 includes:

[0048] S41, generating an initial population; wherein each individual in the population represents a possible scheduling solution that satisfies all constraints;

[0049] S42, calculate the total cost of each individual in the initial population and check the constraints, and impose penalties on individuals that do not meet the constraints;

[0050] S43, performing selection, crossover and mutation operations to generate a new population;

[0051] S44, calculating the fitness of the new population;

[0052] S45. Repeat S43 to S44 above until the preset conditions are met and the optimal multi-energy complementary scheduling plan is output.

[0053] Furthermore, the S45 also includes: optimizing the excellent individuals in the current population by using a local search algorithm, recalculating the fitness, and incorporating the optimized excellent individuals into the next generation population.

[0054] In a second aspect, the present invention provides a multi-energy complementary dispatching system integrating source, grid, load and storage, comprising:

[0055] Data acquisition and preprocessing module: used to obtain relevant data from source, network, load and storage;

[0056] Feature extraction and fusion module: used to extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector;

[0057] Scheduling model building module: used to build a multi-energy complementary scheduling model based on comprehensive feature vectors;

[0058] Optimization solution module: used to solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

[0059] It can be seen from the above technical solution that compared with the prior art, the technical solution of the present invention has the following advantages:

[0060] Beneficial effects:

[0061] 1. This method combines the FNN network and the GRU-BiLSTM-Transformer network, which can comprehensively and efficiently process static attribute data and time series data in multi-energy systems. Finally, a comprehensive feature vector is generated, which provides a rich information basis for the subsequent construction of the dispatch model. This multi-level and multi-dimensional data processing method significantly improves the model's adaptability and prediction accuracy to complex and changeable energy systems.

[0062] 2. An improved genetic algorithm is used to solve the multi-energy complementary scheduling model. In each generation, a new population is generated through selection, crossover and mutation operations, and a local search mechanism is introduced to further optimize the excellent individuals. This method can not only find the global optimal solution, but also effectively avoid falling into the local optimal solution, ensuring the economy, reliability and environmental friendliness of the scheduling plan.

[0063] 3. When constructing a multi-energy complementary scheduling model, the objective function comprehensively considers the operating cost of the power generation unit, the charging and discharging cost of the energy storage device, and the network transmission loss cost, so as to minimize the total operating cost. At the same time, the constraints cover power balance, power generation unit output power limit, energy storage system limit, and grid security constraint, ensuring the stable operation of the system and the satisfaction of various technical indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0065] Figure 1 A flow chart of a multi-energy complementary scheduling method with integrated source, grid, load and storage provided in an embodiment of the present invention;

[0066] Figure 2 A schematic diagram of a comprehensive feature vector acquisition process provided by an embodiment of the present invention;

[0067] Figure 3 A schematic diagram of a comprehensive feature vector acquisition process based on an FNN network and a GRU-BiLSTM-Transformer network structure provided in an embodiment of the present invention;

[0068] Figure 4 A schematic diagram of a process for obtaining a corresponding multi-energy complementary scheduling scheme based on an improved genetic algorithm provided in an embodiment of the present invention;

[0069] Figure 5 A framework diagram of a multi-energy complementary scheduling system integrating source, grid, load and storage provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] Embodiment 1;

[0072] like Figure 1 As shown, this embodiment provides a multi-energy complementary scheduling method integrating source, grid, load and storage, including the following steps:

[0073] S1. Obtain relevant data of each link of source, network, load and storage;

[0074] S2. Extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector;

[0075] S3. Construct a multi-energy complementary scheduling model based on the comprehensive feature vector;

[0076] S4. Solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

[0077] This method uses FNN and GRU-BiLSTM-Transformer networks to extract deep features, construct a comprehensive feature vector containing comprehensive information, and then accurately construct a multi-energy complementary scheduling model. At the same time, the improved genetic algorithm is used to efficiently solve the model, which not only significantly improves the accuracy and solution efficiency of the scheduling plan, but also shows strong flexibility and adaptability, and can flexibly respond to changing energy scheduling needs and realize efficient complementary utilization of multiple energy sources.

[0078] The above steps and related technical features are further described in detail below:

[0079] In this embodiment S1, the relevant data includes:

[0080] The real-time power and power forecast data of different types of power generation units at the source end, the operating status information of the power grid at the network end, the power demand and demand forecast data at the load end, and the current power storage and maximum charge and discharge rate data at the energy storage end.

[0081] Specifically, in the source data, the real-time power of different types of power generation units includes the real-time output power of various types of power generation units such as wind power generation, solar power generation, hydropower generation, thermal power generation, etc. The power forecast data is based on historical data and weather forecasts, and is the forecast result of the output power of various types of power generation units in the future.

[0082] In the network-side data, the operating status information of the power grid includes the real-time voltage, current, frequency and other parameters of the power grid, as well as the load conditions of the lines and transformers.

[0083] In the load-side data, with the rapid growth of the electric vehicle market, the electricity demand includes real-time monitoring of the user's electricity load, including different types of electricity demand such as industry, commerce, and residents. The electricity demand of charging piles has become an important part of the power grid load. Specifically, it is necessary to collect the real-time electricity load of each charging pile through the charging pile management system; it also includes combining historical charging records, weather forecasts (such as temperature, humidity, rainfall, and other factors that affect charging demand), holiday arrangements, and electric vehicle user behavior patterns, and using machine learning methods (such as time series analysis, random forests, etc.) to predict the electricity demand of charging piles in the future.

[0084] In the energy storage data, the current power storage is the current storage power of the energy storage device (such as battery energy storage system, pumped storage, etc.). The maximum charge and discharge rate data is the maximum charge and discharge rate of the energy storage device, as well as the current charge and discharge status.

[0085] By fully acquiring the above data, it not only covers the real-time operating status, but also includes the prediction of future trends, providing a solid foundation for optimizing scheduling.

[0086] In this embodiment S2, Figure 2 and Figure 3 As shown, the FNN network and the GRU-BiLSTM-Transformer network are combined to extract features of the relevant data to obtain a comprehensive feature vector; specifically, the following steps are included:

[0087] S21. Construct static attribute data for input into the FNN network and time series data for input into the GRU-BiLSTM-Transformer network; wherein, static attribute data generally refers to data that does not change frequently over time or changes only under specific conditions, such as the rated value of the current power storage at the energy storage end and the rated value of the maximum charge and discharge rate at the energy storage end; and time series data refers to data that changes over time, including the real-time power of different types of power generation units at the source end, power forecast data at the source end, operating status information of the power grid at the network end, power demand at the load end, demand forecast data at the load end, and current power storage at the energy storage end.

[0088] S22, using FNN network and GRU-BiLSTM-Transformer network to extract static attribute features and time series features respectively;

[0089] The FNN network includes: three linear layers; each linear layer is connected to the BN layer and is provided with a ReLU activation function, and at least one linear layer is provided with a Dropout layer.

[0090] In the FNN network, static attribute data is processed sequentially through three linear layers. Each linear layer is followed by a BatchNormalization (BN) layer to accelerate the training process and reduce internal covariate shifts. At the same time, the ReLU (Rectified Linear Unit) activation function is used to introduce nonlinearity to enhance the network's expressiveness. A Dropout layer is set after at least one linear layer to reduce overfitting. By randomly discarding the output of some neurons, the network is forced to learn more robust features.

[0091] The GRU-BiLSTM-Transformer network includes:

[0092] Gated recurrent unit layer: used to preprocess time series data;

[0093] Bidirectional long short-term memory network layer: used to capture bidirectional dependencies in preprocessed time series data;

[0094] Embedding layer: used to map data into a high-dimensional space;

[0095] Multi-head attention mechanism: used to parallelize the calculation of multiple attention heads to enhance the attention to information at different locations;

[0096] Normalization layer: used to ensure that the data in each batch has a similar distribution;

[0097] Fully connected layer: used to process the results of the multi-head attention mechanism and output deep features.

[0098] In the GRU-BiLSTM-Transformer network, the time series data is first processed preliminarily, and the flow of information is controlled through the gating mechanism to capture the short-term dependencies in the sequence. Based on the GRU layer, the BiLSTM layer processes the forward and reverse time series data at the same time, capturing the bidirectional dependencies in the data, so as to more comprehensively understand the contextual information of the sequence. The embedding layer maps the input of each time step in the time series data to a high-dimensional space, so as to better capture the nonlinear relationship between data in the subsequent layers. Then, by calculating multiple attention heads in parallel, the model can focus on different positions of the input sequence at the same time, enhancing the model's ability to capture key information. The normalization layer ensures that each batch of data maintains a similar distribution during the training process, accelerating the convergence process. Finally, the deep features of the multi-head attention mechanism output are processed by the fully connected layer.

[0099] S23, fusing the static attribute features and the time series features through a fusion layer, and outputting a comprehensive feature vector.

[0100] The static attribute features extracted by the FNN network and the time series features extracted by the GRU-BiLSTM-Transformer network are fused through the fusion layer. The fusion layer can adopt simple splicing, weighted summation or more complex attention mechanism. The fused feature vector is a comprehensive feature vector, which contains comprehensive information of each link of source, network, load and storage, providing strong support for the subsequent construction of multi-energy complementary scheduling model.

[0101] In this embodiment S3, a multi-energy complementary scheduling model is constructed based on the comprehensive feature vector; the goal of this model is to maximize economy, reliability and environmental friendliness by optimizing the scheduling plan.

[0102] The objective function of the multi-energy complementary scheduling model is expressed as:

[0103]

[0104] Where T represents the number of time steps in the scheduling cycle, N g Represents the number of power generation units, N s represents the number of energy storage devices, C i (P g,i,t ) represents the operating cost function of the i-th power generation unit at time t, C j (P s,i,t ) represents the charge and discharge cost function of the jth energy storage device at time t, C loss (P s,i,t ) represents the network transmission loss cost function, P g,i,t represents the output power of the i-th power generation unit at time t, P s,i,t represents the charge and discharge power of the jth energy storage device at time t, P l,t Represents the transmission power loss of the power grid at time t.

[0105] The constraints of the multi-energy complementary scheduling model include:

[0106] Power balance constraints:

[0107]

[0108] Among them, D t represents the load demand at time t, P g,i,t The output power of the i-th power generation unit at time t, P d,j,t represents the discharge power of the jth energy storage device at time t, P c,j,t represents the charging power of the jth energy storage device at time t;

[0109] Specifically, assuming that the total load demand that the system needs to meet at time t includes traditional user demand D user800MW and charging pile demand D ev 200MW, total load demand D t 1000MW;

[0110] At this time, the output power of the power generation unit is P of wind power plant A and P of wind power plant A. g,1,t 200MW, P of solar power station B g,2,t 300MW, coal-fired power plant C P g,3,t 250MW; the energy storage device is in a discharging state with a discharging power of 100MW, and the energy storage device E is in a charging state with a charging power of 50MW;

[0111] According to the power balance constraint, D t =200MW+300MW+250MW+100MW-50MW, that is, the actual load demand is 800MW. Therefore, it is necessary to adjust the output of the power generation unit or the working state of the energy storage device to meet the load demand. For example, the output power of the coal-fired power plant C can be increased or the energy storage device E can be stopped from charging and started to discharge.

[0112] Power generation unit output power limit:

[0113] P g,i,min ≤P g,i,t ≤P g,i,max

[0114] Among them, P g,i,min Represents the minimum output power of the power generation unit, P g,i,max Indicates the maximum output power of the power generation unit;

[0115] Energy storage system limitations:

[0116] E j,min ≤E j,t ≤E j,max

[0117] 0≤P c,j,t ≤P c,j,max

[0118] 0≤P d,j,t ≤P d,j,max

[0119] Among them, E j,t represents the power storage state of the jth energy storage device at time t, E j,min and E j,max Respectively represent the minimum and maximum storage capacity of the jth energy storage device, P c,j,max and P d,j,max denote the maximum charging and discharging rates of the j-th energy storage device, respectively;

[0120] Grid security constraints:

[0121] V min ≤V≤V max

[0122] I min ≤I≤I max

[0123] Where V represents the voltage intensity at time t, V min and V max represent the preset minimum and maximum voltage strengths, I represents the current strength at time t, and I min and I max Indicates the preset minimum and maximum current strengths respectively.

[0124] The model aims to minimize the total system cost by optimizing the coordinated operation between power generation units, energy storage devices and grid transmission. The objective function clearly states the optimization goal, which is to minimize the sum of the operating costs of all power generation units, the charging and discharging costs of energy storage devices, and the network transmission loss costs during the entire scheduling period (defined by the number of time steps T). These cost functions take into account the power output or charging and discharging state of each at a specific point in time.

[0125] In order to ensure the feasibility of the scheduling plan and the stability of system operation, the model also sets a series of constraints.

[0126] The first is the power balance constraint, which requires that the system's power supply at any point in time must strictly match the load demand, which is achieved by accurately calculating the output power of the power generation unit and the charging and discharging power of the energy storage device. Secondly, the output power limit of the power generation unit ensures that the power generation equipment operates within a reasonable range to avoid damage to the equipment caused by overload or underload. The energy storage system limit clearly defines the power storage state, storage capacity, and charging and discharging rate of the energy storage device to protect the energy storage device and optimize its use efficiency. Finally, the grid safety constraint ensures that the grid operates within a stable and safe parameter range by setting upper and lower limits on voltage and current intensity.

[0127] In this embodiment S4, Figure 4 As shown, by improving the genetic algorithm, the multi-energy complementary scheduling model is solved to obtain the corresponding multi-energy complementary scheduling scheme; specifically including:

[0128] S41. Generate an initial population; each individual in the population represents a possible scheduling scheme that satisfies all constraints; each individual can be represented as a vector, in which each element corresponds to scheduling decision variables such as the output power of the power generation unit and the charging and discharging power of the energy storage device within a time step.

[0129] S42. Calculate the total cost of each individual in the initial population and check the constraints, and impose penalties on individuals that do not meet the constraints; for each individual, calculate its total operating cost according to the objective function, including the operating cost of the power generation unit, the charging and discharging cost of the energy storage device, and the transmission loss cost of the power grid. Further, verify whether each individual meets all the constraints, including power balance constraints, power generation unit output power limits, energy storage system limits, and power grid security constraints. Penalize individuals that do not meet the constraints, such as increasing their total cost or reducing their fitness values, to ensure that these individuals are excluded or have a reduced probability of selection during the selection process.

[0130] S43, performing selection, crossover and mutation operations to generate a new population; wherein the selection operation is to select a certain number of excellent individuals from the current population as parents according to the fitness value (the inverse of the total cost or the fitness value after penalty adjustment). The crossover operation is to perform a crossover operation on the selected parent individuals to generate new offspring individuals. The mutation operation is to perform a mutation operation on the newly generated offspring individuals to introduce new gene combinations and increase population diversity.

[0131] S44, calculate the fitness of the new population; recalculate the total cost of each individual in the newly generated population, and check the constraints. Update the fitness value of each individual according to the new total cost and the satisfaction of the constraints.

[0132] S45, repeating the above S43 to S44 until the preset conditions are met and the optimal multi-energy complementary scheduling scheme is output, wherein the preset conditions include the maximum number of iterations or the fitness value converging to a certain threshold.

[0133] The S45 also includes: optimizing the excellent individuals in the current population through a local search algorithm, recalculating the fitness, and incorporating the optimized excellent individuals into the next generation population. The local search algorithm adopts a hill climbing method or a simulated annealing algorithm, wherein the hill climbing method starts from the current solution and gradually searches for a better solution in the neighborhood until no better solution can be found. The simulated annealing algorithm allows a certain degree of acceptance of inferior solutions, and controls the probability of accepting inferior solutions by gradually reducing the temperature parameter, thereby avoiding falling into a local optimum.

[0134] In this step, the multi-energy complementary scheduling model is solved by improving the genetic algorithm, generating an initial population that meets the constraints, calculating the total cost of individuals and imposing penalties to ensure that the constraints are met, and iteratively evolving the population through selection, crossover, and mutation operations. During this period, the local search algorithm is applied to optimize the outstanding individuals until the preset conditions are met and the optimal scheduling plan is output. The hill climbing method is combined with the simulated annealing algorithm to enhance the local search capability and improve the efficiency and quality of the solution.

[0135] The multi-energy complementary scheduling method with integrated source, grid, load and storage proposed in this embodiment, through comprehensive feature extraction and improved genetic algorithm optimization, accurately constructs and efficiently solves the multi-energy scheduling model, realizes the complementary utilization of multiple energy sources, and improves the system economy, reliability and environmental friendliness.

[0136] Embodiment 2;

[0137] like Figure 5 As shown, this embodiment provides a multi-energy complementary scheduling system integrating source, grid, load and storage, including:

[0138] Data acquisition and preprocessing module: used to obtain relevant data from source, network, load and storage;

[0139] Feature extraction and fusion module: used to extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector;

[0140] Scheduling model building module: used to build a multi-energy complementary scheduling model based on comprehensive feature vectors;

[0141] Optimization solution module: used to solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

[0142] This embodiment provides a multi-energy complementary scheduling system that integrates source, grid, load and storage, including four main modules: data acquisition and preprocessing module, used to obtain relevant data of each link of source, grid, load and storage; feature extraction and fusion module, combining FNN network and GRU-BiLSTM-Transformer network to perform deep feature extraction on data and generate comprehensive feature vector; scheduling model construction module, based on comprehensive feature vector, constructs multi-energy complementary scheduling model, optimizes economy, reliability and environmental friendliness, and meets multiple constraints; optimization solution module, through improved genetic algorithm, efficiently solves scheduling model, and obtains the optimal multi-energy complementary scheduling scheme. The system realizes efficient coordinated scheduling of multi-energy systems through advanced data processing and optimization technology.

[0143] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0144] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-energy complementary scheduling method integrating source, grid, load and storage, characterized in that: The following steps are involved: S1. Obtain relevant data of each link of source, network, load and storage; S2. Extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector; S3. Construct a multi-energy complementary scheduling model based on the comprehensive feature vector; S4. Solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

2. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: In S1, the relevant data include: The real-time power and power forecast data of different types of power generation units at the source end, the operating status information of the power grid at the network end, the power demand and demand forecast data at the load end, and the current power storage and maximum charge and discharge rate data at the energy storage end.

3. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: The S2 comprises: S21, construct static attribute data for input to the FNN network and time series data for input to the GRU-BiLSTM-Transformer network; S22, using FNN network and GRU-BiLSTM-Transformer network to extract static attribute features and time series features respectively; S23, fusing the static attribute features and the time series features through a fusion layer, and outputting a comprehensive feature vector.

4. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: The FNN network includes: three linear layers; Among them, each linear layer is connected to the BN layer and is provided with a ReLU activation function, and at least one linear layer is provided with a Dropout layer.

5. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: The GRU-BiLSTM-Transformer network includes: Gated recurrent unit layer: used to preprocess time series data; Bidirectional long short-term memory network layer: used to capture bidirectional dependencies in preprocessed time series data; Embedding layer: used to map data into a high-dimensional space; Multi-head attention mechanism: used to parallelize the calculation of multiple attention heads to enhance the attention to information at different locations; Normalization layer: used to ensure that the data in each batch has a similar distribution; Fully connected layer: used to process the results of the multi-head attention mechanism and output deep features.

6. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: In S3, the objective function of the multi-energy complementary scheduling model is expressed as: Where T represents the number of time steps in the scheduling cycle, N g Represents the number of power generation units, N s represents the number of energy storage devices, C i (P g，i，t ) represents the operating cost function of the i-th power generation unit at time t, C j (P s，i，t ) represents the charge and discharge cost function of the jth energy storage device at time t, C loss (P s，i，t ) represents the network transmission loss cost function, P g，i，t represents the output power of the i-th power generation unit at time t, P s，i，t represents the charge and discharge power of the jth energy storage device at time t, P l，t Represents the transmission power loss of the power grid at time t.

7. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: In S3, the constraints of the multi-energy complementary scheduling model include: Power balance constraints: Among them, D t represents the load demand at time t, P g,i,t The output power of the i-th power generation unit at time t, P d,j,t represents the discharge power of the jth energy storage device at time t, P c,j,t represents the charging power of the jth energy storage device at time t; Power generation unit output power limit: P g,i,min ≤P g,i,t ≤P g,i,max Among them, P g,i,min Represents the minimum output power of the power generation unit, P g,i,max Indicates the maximum output power of the power generation unit; Energy storage system limitations: AND j,min ≤E j,t ≤E j,max 0≤P c,j,t ≤P c,j,max 0≤P d,j,t ≤P d,j,max Among them, E j,t represents the power storage state of the jth energy storage device at time t, E j,min and E j,max Respectively represent the minimum and maximum storage capacity of the jth energy storage device, P c,j,max and P d,j,max denote the maximum charging and discharging rates of the j-th energy storage device, respectively; Grid security constraints: In min ≤V≤V max I min ≤I≤I max Where V represents the voltage intensity at time t, V min and V max represent the preset minimum and maximum voltage strengths, I represents the current strength at time t, and I min and I max Indicates the preset minimum and maximum current strengths respectively.

8. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 1 is characterized in that: The S4 comprises: S41, generating an initial population; wherein each individual in the population represents a possible scheduling solution that satisfies all constraints; S42, calculate the total cost of each individual in the initial population and check the constraints, and impose penalties on individuals that do not meet the constraints; S43, performing selection, crossover and mutation operations to generate a new population; S44, calculating the fitness of the new population; S45. Repeat S43 to S44 above until the preset conditions are met and the optimal multi-energy complementary scheduling plan is output.

9. The multi-energy complementary scheduling method of source-grid-load-storage integration according to claim 8 is characterized in that: The S45 also includes: optimizing the excellent individuals in the current population by using a local search algorithm, recalculating the fitness, and incorporating the optimized excellent individuals into the next generation population.

10. A multi-energy complementary dispatching system integrating source, grid, load and storage, characterized in that: include: Data acquisition and preprocessing module: used to obtain relevant data from source, network, load and storage; Feature extraction and fusion module: used to extract features of the relevant data by combining the FNN network and the GRU-BiLSTM-Transformer network to obtain a comprehensive feature vector; Scheduling model building module: used to build a multi-energy complementary scheduling model based on comprehensive feature vectors; Optimization solution module: used to solve the multi-energy complementary scheduling model by improving the genetic algorithm to obtain the corresponding multi-energy complementary scheduling plan.

Images