A coordinated control method for microgrids with multi-energy complementarity

Abstract

The present invention discloses a multi-energy complementary microgrid coordinated control method, which relates to the technical field of microgrid coordinated control. The method obtains the voltage, current, operating power, active power and circuit parameters of gas generators, diesel generators and photovoltaic power generation systems to construct a microgrid link topology diagram; within the gas generator and diesel generator microgrid, load fluctuations are automatically shared according to preset frequency-power and voltage-reactive Droop curves; when the bus frequency deviates from the nominal value, each power generation device increases or decreases the output at a preset ratio; the present invention calculates the shortest path based on transmission loss, current carrying capacity and voltage drop on a weighted topology diagram, and dynamically verifies and adjusts the first transmission path to achieve rapid switching and reconstruction in the event of a fault; the monitoring and path recalculation mechanism can effectively improve the safety and reliability of the microgrid and reduce the risk of power supply interruption due to line or equipment abnormalities.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid coordinated control, and in particular to a multi-energy complementary microgrid coordinated control method. Background Art

[0002] As the proportion of renewable energy continues to increase, photovoltaic power generation is widely used due to its clean and renewable characteristics. However, its output is easily affected by meteorological factors such as cloud cover, temperature, and humidity, resulting in large fluctuations in power generation. Traditional microgrid scheduling strategies mainly rely on gas-fired and diesel generators to support loads, making it difficult to fully utilize photovoltaic output. In addition, when photovoltaic output suddenly changes or load fluctuates drastically, frequent starting and stopping and large-scale adjustments of units increase fuel consumption and equipment wear, reducing system economy and reliability.

[0003] Existing power transmission path planning within power grids is generally based on static topology information, without fully considering dynamic factors such as real-time load distribution, line resistance, and voltage drop. The dynamic verification mechanism for transmission losses is also incomplete.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides a multi-energy complementary microgrid coordinated control method.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A multi-energy complementary microgrid coordinated control method, comprising:

[0007] S1. Obtain the voltage, current, operating power, active power, and circuit parameters of the gas generator, diesel generator, and photovoltaic power generation system, and construct a microgrid link topology diagram;

[0008] S2: Within the gas-fired and diesel-powered microgrids, load fluctuations are automatically shared according to preset frequency-power and voltage-reactive power curves. When the bus frequency deviates from the nominal value, each generator increases or decreases its output at a preset ratio.

[0009] S3. Obtain historical power generation and meteorological data sequences, use LSTM to build a photovoltaic power generation prediction model, and output a photovoltaic power generation curve. Generate a load demand curve based on historical electricity consumption data. During scheduling, gas and diesel generators are connected first, and the remaining load is borne by the photovoltaic power generation system.

[0010] S4. Calculate a first power transmission path based on the energy-saving strategy for the microgrid link topology diagram, convert the first power transmission path into a switch operation instruction, and send the instruction to the power distribution device.

[0011] The process of constructing the microgrid link topology diagram includes:

[0012] S11. Create a unique node for each gas generator, diesel generator, and photovoltaic power generation system. Treat the busbar or combiner box where the power generation equipment outputs and receives as an independent node. Treat important primary and secondary load branches as separate nodes.

[0013] S12. If there is a cable or bus connection between any two nodes, add an undirected edge to the graph and attach impedance, flow, line length, and voltage level attributes to each edge;

[0014] S13. When a node or undirected edge is detected to be abnormal, the corresponding node or undirected edge is immediately marked as invalid in the graph, and the corresponding edge is dynamically added or removed through the closing / opening signal of the switch node.

[0015] The preset frequency-power and voltage-reactive droop curves for automatically sharing load fluctuations include: determining the maximum allowable frequency deviation and voltage deviation amplitude based on the rated frequency and rated active power, rated voltage, and rated reactive power of the unit; then calculating the active power correction coefficient and reactive power correction coefficient according to the predetermined frequency-power and voltage-reactive droop ratios, respectively, and writing them into the local controller parameters; verifying whether the unit's active and reactive power responses to frequency and voltage conform to a preset linear relationship by applying small frequency deviations and voltage deviations, and fine-tuning the coefficients based on the test results to obtain a characteristic curve; finally, using the verified characteristic curve according to the time sequence to obtain an expected power amount, and subtracting the operating power from the expected power amount to obtain the active power correction amount;

[0016] The active power correction amount is converted into a fuel valve opening instruction for a gas engine or a fuel injection amount change instruction for a diesel engine. The controller monitors the correction effect in each control cycle and continuously iterates the Droop operation until the bus frequency and voltage return to the preset range.

[0017] Constructing a photovoltaic power generation prediction model to output photovoltaic power generation curves includes:

[0018] S21. Obtain a historical power generation sequence of the photovoltaic inverter, obtain a historical meteorological data sequence, and align the historical power generation sequence and the historical meteorological data sequence at a unified time interval to form a preliminary data set;

[0019] S22. Cut windows of length W+H from the aligned dataset in sequence, use the meteorological and power generation features of the first W steps as input X, and use the power generation power of the next H steps as label Y, to form many-to-one or many-to-many training samples;

[0020] S23. Construct a multi-layer LSTM network. The input layer inputs multi-dimensional meteorological and power generation features with a length of W steps. The hidden layer is configured with 1 to 2 layers of LSTM units, each containing M hidden states. The fully connected output layer maps the output of the last LSTM layer to the H-step photovoltaic power prediction value. The mean square error (MSE) is selected as the loss function, the Adam optimizer is used, and the initial learning rate is set to η.

[0021] S24. Calling a photovoltaic power generation prediction model based on LSTM, the model takes current and recent historical meteorological characteristics as input, the meteorological characteristics including cloud cover, cloud distribution, temperature, humidity and time label, and outputs the photovoltaic power generation curve.

[0022] An average demand value for the current time segment is generated based on historical electricity consumption data, and the gas generator and diesel generator are first connected to the grid and put into operation at their respective minimum stable operating powers. When the sum of the minimum stable outputs of the gas generator and the diesel generator is greater than or equal to the average demand value, the photovoltaic power generation system provides reactive power compensation or charges the energy storage system. When the sum of the minimum stable outputs of the gas generator and the diesel generator is less than the average demand value, the system calculates the difference between the sum of the minimum stable outputs and the average demand value as the residual load.

[0023] The available photovoltaic power generation power in the current time period is calculated based on the photovoltaic power generation curve, and the available photovoltaic power generation power is compared with the remaining load: if the available photovoltaic power generation power is greater than or equal to the remaining load, the remaining load is entirely allocated to the output of the photovoltaic power generation system, while the gas generator and the diesel generator are maintained at the minimum stable operating power, and the excess power of the photovoltaic power generation system is used for reactive power regulation or charging the energy storage system; if the available photovoltaic power generation power is less than the remaining load, the gas generator and the diesel generator are used to supplement the remaining load as needed based on their minimum stable output.

[0024] Calculating the first power transmission path based on the energy-saving strategy includes:

[0025] S41. The obtained microgrid link topology and edge attributes are used to establish a weighted directed graph for each source node (including gas generator nodes, diesel generator nodes, and photovoltaic power generation system nodes) and the equivalent load node, where the weight wij of each edge eij is calculated according to the formula; Where Rij is the resistance value of edge eij, Unom is the rated voltage, αij is the historical loss correction factor, and Pflow is the power to be transmitted;

[0026] S42. Invoke the Dijkstra shortest path algorithm for each source node in turn to find the path with the minimum cumulative loss to the equivalent load node. After the calculation is completed, compare the minimum cumulative loss values ​​corresponding to all source nodes and select the path with the minimum loss and its corresponding source node as the first candidate transmission path.

[0027] S43. Perform ampacity and voltage drop checks on each edge in the first alternative transmission path: Calculate the actual current Iij using the formula Iij = Pflow / Unom, and determine whether it exceeds the rated ampacity Iijmax of the edge. Simultaneously, calculate the cumulative voltage drop ∑Rij × Iij of the path, and determine whether it exceeds the set voltage drop threshold. If any of the above checks are not met, remove the corresponding edge from the weighted graph and repeat step S42.

[0028] S44: When the candidate first power transmission path passes all checks, the system confirms it as the final first power transmission path;

[0029] S45. Extract the circuit breaker identifiers corresponding to the path edges from the final first transmission path, and generate corresponding closing operation instructions according to the start-end sequence of the path; the system also generates opening operation instructions for all edges adjacent to the path nodes but not on the path.

[0030] The closing and opening operation instructions are encapsulated into a control message containing the target circuit breaker ID, operation type, execution timing and safety check code according to the predetermined protocol format, and sent to each distribution terminal through the communication channel; when it is detected that any circuit breaker fails to execute the instruction successfully or an abnormality occurs, the corresponding edge is marked as failed and the recalculation process starts from step S41.

[0031] The present invention provides a multi-energy complementary microgrid coordinated control method. Compared with the existing technology, it has the following advantages:

[0032] The present invention calculates the shortest path based on transmission loss, current carrying capacity, and voltage drop on a weighted topology graph, and dynamically verifies and adjusts the first transmission path to achieve rapid switching and reconstruction in the event of a fault. This monitoring and path recalculation mechanism can effectively improve the safety and reliability of the microgrid and reduce the risk of power outages caused by line or equipment anomalies.

[0033] The present invention estimates the available photovoltaic output by collecting historical power generation power sequences and historical meteorological data sequences and constructing an LSTM prediction model. It then generates a stable load demand based on historical electricity consumption curves and prioritizes the use of photovoltaic output to cover the remaining load, thereby significantly reducing the operating time and fuel consumption of gas and diesel engines and improving the overall energy efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0036] Example 1

[0037] See also Figure 1 , this application provides a multi-energy complementary microgrid coordinated control method, including:

[0038] S1. Obtain the voltage, current, operating power, active power, and circuit parameters of the gas generator, diesel generator, and photovoltaic power generation system, and construct a microgrid link topology diagram;

[0039] S2: Within the gas-fired and diesel-powered microgrids, load fluctuations are automatically shared according to preset frequency-power and voltage-reactive power curves. When the bus frequency deviates from the nominal value, each generator increases or decreases its output at a preset ratio.

[0040] S3. Obtain historical power generation and meteorological data sequences, use LSTM to build a photovoltaic power generation prediction model, and output a photovoltaic power generation curve. Generate a load demand curve based on historical electricity consumption data. During scheduling, gas and diesel generators are connected first, and the remaining load is borne by the photovoltaic power generation system.

[0041] S4. Calculate a first power transmission path based on the energy-saving strategy for the microgrid link topology diagram, convert the first power transmission path into a switch operation instruction, and send the instruction to the power distribution device.

[0042] The process of constructing a microgrid link topology diagram includes:

[0043] S11. Create a unique node for each gas generator, diesel generator, and photovoltaic power generation system. Treat the busbar or combiner box where the power generation equipment outputs and receives as an independent node. Treat important primary and secondary load branches as separate nodes.

[0044] S12. If there is a cable or bus connection between any two nodes, add an undirected edge to the graph and attach impedance, flow, line length, and voltage level attributes to each edge;

[0045] S13. When a node or undirected edge is detected to be abnormal, the corresponding node or undirected edge is immediately marked as invalid in the graph, and the corresponding edge is dynamically added or removed through the closing / opening signal of the switch node.

[0046] Specifically, for the gas generator, diesel generator 1, diesel generator 2, and each inverter in the photovoltaic power generation system, a corresponding power generation node (node ​​ID) is created according to the unique device number, and the following metadata is associated with each power generation node: physical location (such as cabinet number, equipment room coordinates), rated voltage level, rated active / reactive power, access point bus number, and device operating status indicator (used for real-time reading of voltage, current, frequency, and active / reactive power);

[0047] For the busbars or combiner boxes that the PV panels or generators output and receive, create independent busbar nodes. The node ID corresponds to the corresponding busbar number or combiner box number.

[0048] Initialize the following attributes for each bus node: bus voltage level (e.g., 400 V, 10.5 kV), real-time voltage measurement point pointer, real-time current accumulation pointer (used to count the parallel outputs of multiple connected devices). Important primary loads in the system (e.g., critical communications, command centers, and life support equipment) are represented as independent load nodes in the topology diagram and assigned node IDs.

[0049] The preset frequency-power and voltage-reactive droop curves automatically share load fluctuations. The following steps are used: Based on the unit's rated frequency, rated active power, rated voltage, and rated reactive power, the maximum allowable frequency and voltage deviations are determined. Active and reactive correction factors are then calculated based on the predetermined frequency-power and voltage-reactive droop ratios, respectively, and written into the local controller parameters. By applying small frequency and voltage deviations, the unit's active and reactive responses to frequency and voltage are verified to conform to the preset linear relationship. The coefficients are then fine-tuned based on the test results to create a characteristic curve. Finally, the verified characteristic curve is used to derive the predicted power based on the time sequence. The active correction factor is then subtracted from the operating power to obtain the predicted power.

[0050] Unit local droop adjustment: Based on the nameplate parameters of each unit, the rated frequency, rated active / reactive power, and rated voltage are read, the maximum frequency deviation and voltage deviation amplitude are selected, the initial frequency-active power and voltage-reactive power droop coefficients are calculated, and the coefficients are sent to the gas engine excitation system, diesel engine governor, or photovoltaic inverter reactive power control unit.

[0051] In the field test, a small frequency deviation or voltage deviation is applied to the unit, and the unit's active and reactive responses are measured to verify their linear relationship. The Droop coefficient is then fine-tuned based on the test results until the error meets the set accuracy.

[0052] During power-on operation, the unit measures the bus frequency and voltage in real time within every 10 ms to 100 ms control cycle, calculates the expected active / reactive output by table lookup or linear interpolation, converts the correction amount into the gas valve opening or diesel injection amount / photovoltaic inverter reference value, and quickly injects or absorbs reactive power to ensure that the bus frequency and voltage recover smoothly within the deviation range.

[0053] The active power correction amount is converted into a fuel valve opening instruction for a gas engine or a fuel injection amount change instruction for a diesel engine. The controller monitors the correction effect in each control cycle and continuously iterates the Droop operation until the bus frequency and voltage return to the preset range.

[0054] Constructing a photovoltaic power generation prediction model to output photovoltaic power generation curves includes:

[0055] S21. Obtain a historical power generation sequence of the photovoltaic inverter, obtain a historical meteorological data sequence, and align the historical power generation sequence and the historical meteorological data sequence at a unified time interval to form a preliminary data set;

[0056] S22. Cut windows of length W+H from the aligned dataset in sequence, use the meteorological and power generation features of the first W steps as input X, and use the power generation power of the next H steps as label Y, to form many-to-one or many-to-many training samples;

[0057] S23. Construct a multi-layer LSTM network. The input layer inputs multi-dimensional meteorological and power generation features with a length of W steps. The hidden layer is configured with 1 to 2 layers of LSTM units, each containing M hidden states. The fully connected output layer maps the output of the last LSTM layer to the H-step photovoltaic power prediction value. The mean square error (MSE) is selected as the loss function, the Adam optimizer is used, and the initial learning rate is set to η.

[0058] S24. Call the LSTM-based photovoltaic power generation prediction model. The model takes the current and recent historical meteorological characteristics as input, including cloud cover, cloud distribution, temperature, humidity and time tags, and outputs the photovoltaic power generation curve.

[0059] Multi-layer LSTM network structure: The first LSTM hidden layer contains M1 hidden units (such as 64 or 128), return_sequences=True;

[0060] The second LSTM hidden layer (optional): contains M2 hidden units (such as 32 or 64), return_sequences=False;

[0061] Fully connected output layer: For many-to-many models, connect to TimeDistributedDense to map each time step to a scalar prediction; for many-to-one models, connect to the Dense layer only at the last moment and output a photovoltaic power generation curve of length H.

[0062] Training process: Divide the samples generated by the sliding window into a training set (70%), a validation set (20%), and a test set (10%), maintaining chronological order; set the batch size to 32 or 64, and loop training for a number of epochs (e.g., 50-200); after each epoch, evaluate the MSE and MAE on the validation set. If the validation error does not decrease after p consecutive epochs (e.g., 10), trigger early stopping and save the current optimal model weights; perform a final evaluation on the test set, calculate indicators such as MSE, MAE, and MAPE, and generate a comparison chart of the photovoltaic power generation curve and the actual curve to verify the model's generalization ability under various weather conditions such as sunny days, cloudy days, and showers.

[0063] Beneficial effects: The present invention estimates the available photovoltaic output by collecting historical power generation power sequences and historical meteorological data sequences and constructing an LSTM prediction model; it generates a stable load demand based on the historical electricity consumption curve, and gives priority to using photovoltaic output to cover the remaining load, thereby significantly reducing the operating time and fuel consumption of gas engines and diesel engines, and improving the overall energy utilization efficiency of the system.

[0064] Example 2

[0065] Based on historical electricity consumption data, the average demand value for the current time segment is generated, and the gas generator and diesel generator are first connected to the grid and put into operation at their respective minimum stable operating power. When the sum of the minimum stable output of the gas generator and the diesel generator is greater than or equal to the average demand value, the photovoltaic power generation system provides reactive power compensation or charges the energy storage system. When the sum of the minimum stable output of the gas generator and the diesel generator is less than the average demand value, the system calculates the difference between the sum of the aforementioned minimum stable outputs and the average demand value as the residual load.

[0066] Generating the average demand value for the current time segment based on the historical electricity consumption data includes collecting the historical active power time series of each level of load (including primary, secondary, and tertiary loads) in the microgrid in real time through the power distribution monitoring system, with the sampling granularity consistent with the photovoltaic forecast (e.g., once every 5 minutes), and recording the corresponding timestamp;

[0067] Then the collected load power data is processed for missing values: data gaps caused by communication delays or failures are filled with forward filling or the mean of adjacent time periods;

[0068] The above historical load data is segmented by time granularity (for example, each 5-minute period of the upcoming day is processed separately), and the arithmetic mean of the load data corresponding to all historical dates in each period is calculated to obtain the average demand value of the current time segment.

[0069] The available photovoltaic power generation power in the current time period is calculated based on the photovoltaic power generation curve, and the available photovoltaic power generation power is compared with the remaining load: if the available photovoltaic power generation power is greater than or equal to the remaining load, the remaining load is allocated entirely to the photovoltaic power generation system output, while the gas generator and diesel generator maintain the minimum stable operating power. The excess power of the photovoltaic power generation system is used for reactive power regulation or charging the energy storage system; if the available photovoltaic power generation power is less than the remaining load, the gas generator and diesel generator are used to supplement the remaining load as needed based on their minimum stable output.

[0070] Photovoltaic system reactive power compensation or energy storage charging includes:

[0071] When the base load of the unit can fully meet the average load, the photovoltaic system does not need to bear active output after being connected to the grid, and all the available photovoltaic power can be used for reactive power compensation to maintain bus voltage stability; if there is an energy storage device and it is in a rechargeable state, the available photovoltaic power can also be used to charge the energy storage device first to reserve backup energy in advance; in this state, if the actual photovoltaic output is lower than the predicted value, only reactive power adjustment can be performed or energy storage charging can be suspended, and the unit continues to maintain the lowest base load operation without the need for active power output adjustment.

[0072] Calculating the first transmission path based on the energy-saving strategy includes:

[0073] S41. The obtained microgrid link topology and edge attributes are used to establish a weighted directed graph for each source node (including gas generator nodes, diesel generator nodes, and photovoltaic power generation system nodes) and the equivalent load node, where the weight wij of each edge eij is calculated according to the formula; Where Rij is the resistance value of edge eij, Unom is the rated voltage, αij is the historical loss correction factor, and Pflow is the power to be transmitted;

[0074] S42. Invoke the Dijkstra shortest path algorithm for each source node in turn to find the path with the minimum cumulative loss to the equivalent load node. After the calculation is completed, compare the minimum cumulative loss values ​​corresponding to all source nodes and select the path with the minimum loss and its corresponding source node as the first candidate transmission path.

[0075] S43. Perform ampacity and voltage drop checks on each edge in the first alternative transmission path: Calculate the actual current Iij using the formula Iij = Pflow / Unom, and determine whether it exceeds the rated ampacity Iijmax of the edge. Simultaneously, calculate the cumulative voltage drop ∑Rij × Iij of the path, and determine whether it exceeds the set voltage drop threshold. If any of the above checks are not met, remove the corresponding edge from the weighted graph and repeat step S42.

[0076] S44: When the candidate first power transmission path passes all checks, the system confirms it as the final first power transmission path;

[0077] S45. Extract the circuit breaker identifiers corresponding to the path edges from the final first transmission path, and generate corresponding closing operation instructions according to the start-end sequence of the path; the system also generates opening operation instructions for all edges adjacent to the path nodes but not on the path.

[0078] The specific implementation method of calling the shortest path algorithm to find the minimum cumulative loss path to the equivalent load node includes:

[0079] For each source node (including gas generator nodes, diesel generator nodes, and photovoltaic power generation system nodes), a corresponding weighted directed graph is obtained from step S41. The graph includes all power generation nodes, intermediate bus nodes, and equivalent load nodes in the microgrid, and each edge is assigned a weight calculated based on the line resistance R, rated voltage U_nom, historical loss correction coefficient α, and power to be transmitted P_flow. Then, for the current source node s, the system initializes the distance values ​​for all nodes in the local temporary data structure, sets the initial distance of the source node to zero, and sets the remaining nodes to infinity, and establishes a predecessor pointer prev for each node to record the optimal path information. At the same time, all nodes and their distance values ​​are inserted into the priority queue Q. Then, when the queue Q is not empty, the loop is executed: the node u with the smallest current distance is taken from Q, and if the node has not been visited, it is marked as visited; if u is an equivalent load node, it means that the shortest cumulative loss path from the source node s to the equivalent load node has been found, and the loop can be terminated early; otherwise, the system traverses the nodes starting from u For all edges with u as the starting point, for each edge (u,v) that is not marked as invalid and has a non-infinite weight, calculate the candidate distance alt = dist[u]+w(u,v). If alt is less than the currently stored dist[v], update dist[v] = alt and set prev[v] = u. At the same time, update the priority of node v in queue Q to alt. After the loop ends, if the distance to the equivalent load node is still infinite, mark the source node as unreachable and no longer participate in the candidate path comparison; otherwise, reconstruct the optimal path Path_s by backtracking the predecessor pointer prev from the equivalent load node to the source node s and record the cumulative loss value Cost_s of the path. After completing the above shortest path search for all source nodes, the system compares the minimum cumulative loss value Cost_s corresponding to each source node and selects the source node s_min corresponding to the minimum value and its path Path_{s_min} as the first transmission path.

[0080] The closing and opening operation instructions are encapsulated into a control message containing the target circuit breaker ID, operation type, execution timing and safety check code according to the predetermined protocol format, and sent to each distribution terminal through the communication channel; when it is detected that any circuit breaker fails to execute the instruction successfully or an abnormality occurs, the corresponding edge is marked as failed and the recalculation process starts from step S41.

[0081] Specifically, the circuit breaker number corresponding to each edge is extracted from the final path, and closing instructions are generated in the order of the path from the source node to the load node. Opening instructions are generated for other edges outside the path but adjacent to the path nodes. All closing / opening operations are encapsulated as control messages in the communication protocol format (including circuit breaker ID, operation type, execution timing and safety check code), and sent to each distribution terminal through the SCADA / IEC61850 channel. After execution, each terminal will provide real-time feedback on the circuit breaker status and voltage and current data. If it is detected that any circuit breaker fails to complete the operation as instructed or an abnormality occurs, the corresponding path edge will be marked as invalid and recalculated from the first step.

[0082] Beneficial effects: The present invention calculates the shortest path based on transmission loss, current carrying capacity and voltage drop on a weighted topology graph, and dynamically verifies and adjusts the first transmission path to achieve rapid switching and reconstruction in the event of a fault; this monitoring and path recalculation mechanism can effectively improve the safety and reliability of the microgrid and reduce the risk of power interruption due to line or equipment abnormalities.

[0083] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0084] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A multi-energy complementary microgrid coordinated control method, characterized in that: include: S1. Obtain the voltage, current, operating power, active power, and circuit parameters of the gas generator, diesel generator, and photovoltaic power generation system, and construct a microgrid link topology diagram; S2: Within the gas-fired and diesel-powered microgrids, load fluctuations are automatically shared according to preset frequency-power and voltage-reactive power curves. When the bus frequency deviates from the nominal value, each generator increases or decreases its output at a preset ratio. S3. Obtain historical power generation and meteorological data sequences, use LSTM to build a photovoltaic power generation prediction model, and output a photovoltaic power generation curve. Generate a load demand curve based on historical electricity consumption data. During scheduling, gas and diesel generators are connected first, and the remaining load is borne by the photovoltaic power generation system. S4, the microgrid link topology diagram based on the energy-saving strategy to calculate the first transmission path, the first transmission path is converted into a switch operation instruction, sent to the distribution device; Calculating the first power transmission path based on the energy-saving strategy includes: S41. Obtain the microgrid link topology and edge attributes, and establish a weighted directed graph for each source node and equivalent load node. Each source node includes a gas generator node, a diesel generator node, and a photovoltaic power generation system node, wherein the weight wij of each edge eij is calculated according to the formula; Where Rij is the resistance value of edge eij, Unom is the rated voltage, αij is the historical loss correction factor, and Pflow is the power to be transmitted; S42. Invoke the Dijkstra shortest path algorithm for each source node in turn to find the path with the minimum cumulative loss to the equivalent load node. After the calculation is completed, compare the minimum cumulative loss values ​​corresponding to all source nodes and select the path with the minimum loss and its corresponding source node as the first candidate transmission path. S43. Perform ampacity and voltage drop checks on each edge in the first alternative transmission path: Calculate the actual current Iij using the formula Iij = Pflow / Unom, and determine whether it exceeds the rated ampacity Iijmax of the edge. Simultaneously, calculate the cumulative voltage drop ∑Rij × Iij of the path, and determine whether it exceeds the set voltage drop threshold. If any of the above checks are not met, remove the corresponding edge from the weighted graph and repeat step S42. S44: When the candidate first power transmission path passes all checks, the system confirms it as the final first power transmission path; S45. Extract the circuit breaker identifiers corresponding to the path edges from the final first transmission path, and generate corresponding closing operation instructions according to the start-end sequence of the path; the system also generates opening operation instructions for all edges adjacent to the path nodes but not on the path.

2. A multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: The process of constructing the microgrid link topology diagram includes: S11. Create a unique node for each gas generator, diesel generator, and photovoltaic power generation system. Treat the busbar or combiner box where the power generation equipment outputs and receives as an independent node. Treat important primary and secondary load branches as separate nodes. S12. If there is a cable or bus connection between any two nodes, add an undirected edge to the graph and attach impedance, flow, line length, and voltage level attributes to each edge; S13. When a node or undirected edge is detected to be abnormal, the corresponding node or undirected edge is immediately marked as invalid in the graph, and the corresponding edge is dynamically added or removed through the closing / opening signal of the switch node.

3. The multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: The preset frequency-power and voltage-reactive droop curves for automatically sharing load fluctuations include: determining the maximum allowable frequency deviation and voltage deviation amplitude based on the rated frequency and rated active power, rated voltage, and rated reactive power of the unit; then calculating the active power correction coefficient and reactive power correction coefficient according to the predetermined frequency-power and voltage-reactive droop ratios, respectively, and writing them into the local controller parameters; verifying whether the unit's active and reactive power responses to frequency and voltage conform to a preset linear relationship by applying small frequency deviations and voltage deviations, and fine-tuning the coefficients based on the test results to obtain a characteristic curve; finally, using the verified characteristic curve according to the time sequence to obtain an expected power amount, and subtracting the operating power from the expected power amount to obtain the active power correction amount; The active power correction amount is converted into a fuel valve opening instruction for a gas engine or a fuel injection amount change instruction for a diesel engine. The controller monitors the correction effect in each control cycle and continuously iterates the Droop operation until the bus frequency and voltage return to the preset range.

4. The multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: Constructing a photovoltaic power generation prediction model to output photovoltaic power generation curves includes: S21. Obtain a historical power generation sequence of the photovoltaic inverter, obtain a historical meteorological data sequence, and align the historical power generation sequence and the historical meteorological data sequence at a unified time interval to form a preliminary data set; S22. Cut windows of length W+H from the aligned dataset in sequence, use the meteorological and power generation features of the first W steps as input X, and use the power generation power of the next H steps as label Y, to form many-to-one or many-to-many training samples; S23. Construct a multi-layer LSTM network. The input layer inputs multi-dimensional meteorological and power generation features with a length of W steps. The hidden layer is configured with 1 to 2 layers of LSTM units, each containing M hidden states. The fully connected output layer maps the output of the last LSTM layer to the H-step photovoltaic power prediction value. The mean square error (MSE) is selected as the loss function, the Adam optimizer is used, and the initial learning rate is set to η. S24. Calling a photovoltaic power generation prediction model based on LSTM, the model takes current and recent historical meteorological characteristics as input, the meteorological characteristics including cloud cover, cloud distribution, temperature, humidity and time label, and outputs the photovoltaic power generation curve.

5. The multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: Based on historical electricity consumption data, an average demand value for the current time segment is generated. The gas generator and diesel generator are first connected to the grid and put into operation at their respective minimum stable operating powers. When the sum of the minimum stable outputs of the gas generator and the diesel generator is greater than or equal to the average demand value, the photovoltaic power generation system provides reactive power compensation or charges the energy storage system. When the sum of the minimum stable outputs of the gas generator and the diesel generator is less than the average demand value, the system calculates the difference between the sum of the minimum stable outputs and the average demand value as the residual load.

6. The multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: The available photovoltaic power generation power in the current time period is calculated based on the photovoltaic power generation curve, and the available photovoltaic power generation power is compared with the remaining load: if the available photovoltaic power generation power is greater than or equal to the remaining load, the remaining load is entirely allocated to the output of the photovoltaic power generation system, while the gas generator and the diesel generator are maintained at the minimum stable operating power, and the excess power of the photovoltaic power generation system is used for reactive power regulation or charging the energy storage system; if the available photovoltaic power generation power is less than the remaining load, the gas generator and the diesel generator are used to supplement the remaining load as needed based on their minimum stable output.

7. The multi-energy complementary microgrid coordinated control method according to claim 1, characterized in that: The closing and opening operation instructions are encapsulated into a control message containing the target circuit breaker ID, operation type, execution timing and safety check code according to the predetermined protocol format, and sent to each distribution terminal through the communication channel; when it is detected that any circuit breaker fails to execute the instruction successfully or an abnormality occurs, the corresponding edge is marked as failed and the recalculation process starts from step S41.

Images