Microgrid control method and system

Abstract

The invention provides a micro-grid control method and system, and the method comprises the steps: detecting the operation state of each device in a micro-grid in real time, and obtaining the load demand data of each device; preprocessing the load demand data, and extracting feature data of the load demand data; acquiring historical operation state data of each device in the micro-grid, inputting the historical operation state data into a preset deep learning algorithm, and predicting future load demand and renewable energy output of the micro-grid to obtain prediction data; based on the prediction data and the feature data, determining a scheduling strategy corresponding to the power equipment by adopting a genetic algorithm; and generating a control instruction of the equipment according to the scheduling strategy, sending the control instruction to the corresponding power equipment, and monitoring the execution condition of the power equipment. Based on the method, the invention further provides a micro-grid control system. According to the invention, each device can be ensured to operate according to the optimal strategy, and the operation efficiency of the micro-grid is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of microgrid control, and in particular relates to a microgrid control method and system. Background Art

[0002] With the transformation of energy structure and the continuous development of smart grid technology, microgrid, as a small power system integrating distributed energy, energy storage devices, loads and control systems, has attracted more and more attention. Microgrid can operate independently or in parallel with the main grid, which improves the flexibility and reliability of the power system, especially in renewable energy access and energy conservation and emission reduction. However, the efficient and stable operation of microgrid depends on accurate equipment scheduling and control strategies.

[0003] Traditional microgrid dispatching methods mostly rely on manual experience and simplified models, which are difficult to respond to complex and changing operating conditions and load demands in real time, resulting in low energy efficiency and even affecting the stability and security of microgrids. In addition, the intermittent and uncertain nature of renewable energy also poses a huge challenge to the dispatching and control of microgrids. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a microgrid control method and system, which can ensure that each device operates according to the optimal strategy and improve the operation efficiency of the microgrid.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A microgrid control method comprises the following steps:

[0007] Real-time detection of the operating status of each device in the microgrid to obtain the load demand data of each device; after pre-processing the load demand data, the characteristic data of the load demand data is extracted;

[0008] Obtain the historical operating status data of each device in the microgrid, input the historical operating status data into the preset deep learning algorithm, and predict the future load demand and renewable energy output of the microgrid to obtain forecast data;

[0009] Based on the predicted data and characteristic data, a genetic algorithm is used to determine the dispatching strategy corresponding to the power equipment; a control instruction of the equipment is generated according to the dispatching strategy, and the control instruction is sent to the corresponding power equipment, and the execution of the power equipment is monitored.

[0010] Furthermore, the process of extracting characteristic data of the load demand data after preprocessing the load demand data includes:

[0011] Traverse the load demand data of each time period after preprocessing, use the maximum value function to determine the maximum load value in each cycle, and mark it as the load peak value;

[0012] Traverse the load demand data of each time period after preprocessing, use the minimum function to determine the minimum load value in each cycle, and mark it as the load valley value;

[0013] Define the time window for fluctuation calculation. In each time window, calculate the standard deviation of load demand data as the load fluctuation rate in the current window.

[0014] The load peak value, the load valley value and the load fluctuation rate are fused to obtain a fused data set; K data points are selected from the fused data set as initial cluster centers, and for each data point in the fused data set, the distance between each data point and each cluster center is calculated, and each data point is assigned to a cluster center corresponding to the distance to form K clusters;

[0015] For each cluster, the average value of all data points is calculated, and the average value is used as the cluster center again. The operation is iterated until the preset number of iterations is reached; for each cluster, the data point corresponding to the distance from the cluster center is used as the representative data point of the cluster; the representative data point of each cluster is used as the fused data point, where the representative data points include load peak value, load valley value and load fluctuation rate.

[0016] Furthermore, the process of calculating the distance between each data point and each cluster center includes:

[0017]

[0018] Among them, x i is the data point; c k is the cluster center; i represents the index of the data point; k represents the index of the cluster center; x i -c k is the data point x i With cluster center c k The difference vector between -1 is the inverse matrix of the covariance matrix Σ; W is the weighting matrix; σ is the scale parameter in the Gaussian kernel function; T represents the transpose.

[0019] Furthermore, the historical operating status data is input into a preset deep learning algorithm to predict the future load demand and renewable energy output of the microgrid to obtain the predicted data, including:

[0020] Divide the historical operation status data into training set, validation set and test set according to the preset ratio;

[0021] The long short-term memory network is trained using the training set, and the long short-term memory network after training is evaluated using the test set to obtain the evaluation result;

[0022] The LSTM network is tuned according to the evaluation results, and the trained LSTM network is used to predict the future load demand and renewable energy output of the microgrid to obtain prediction data.

[0023] Furthermore, based on the prediction data and the characteristic data, the process of using a genetic algorithm to determine a dispatching strategy corresponding to the power equipment includes:

[0024] integrating the forecast data and the characteristic data to form a comprehensive data set including time, load demand, renewable energy production and equipment characteristics;

[0025] According to the operation objectives and constraints of the microgrid, the objective function is determined; the expression of the objective function is:

[0026]

[0027] Where K is the index of the power equipment; R is the total number of equipment; t is the index of the time step; T is the total number of time steps; C gen,K represents the unit power generation cost of equipment K; P gen,K (t) represents the power generated by device K at time t; E gen,K (t) represents the efficiency of equipment K at time t;

[0028] The dispatching decisions of power equipment are encoded as genes in the genetic algorithm, and each gene represents a specific dispatching action or equipment state;

[0029] An initial population is randomly generated, in which each individual represents a possible scheduling strategy.

[0030] Furthermore, the control instructions include starting and stopping of the generator set, power output adjustment, charge and discharge control of the energy storage device, and switching of the load device.

[0031] Furthermore, the specific process of sending the control instruction to the corresponding power equipment and monitoring the execution status of the power equipment includes:

[0032] Dividing the control instruction into multiple independent data blocks according to preset rules;

[0033] Performing cluster analysis on the segmented data blocks to obtain clustered data block groups, and then calculating the weights of the data block groups;

[0034] The comprehensive score of the data block group is calculated according to the weight of each data block group, the size of the data block group and the transmission difficulty. According to the comprehensive score, the data block groups are sorted from high to low and sent to the corresponding power equipment in sequence.

[0035] Furthermore, the process of performing cluster analysis on the segmented data blocks to obtain clustered data block groups includes:

[0036] Calculate the Euclidean distance between each data block and the corresponding first cluster center, assign each data block to the corresponding first cluster center, and form multiple initial clusters;

[0037] The average value of all data blocks is calculated to obtain the second cluster center, and the data blocks are redistributed according to the second cluster center, and the position of the second cluster center is updated until the preset maximum number of iterations is reached to obtain the final clustering result and output the data block group.

[0038] Furthermore, the calculation formula of the comprehensive score is:

[0039]

[0040] Among them, CS j represents the comprehensive priority score of the jth data block group; α is the first weight coefficient, β is the second weight coefficient, γ is the third weight coefficient, δ is the fourth weight coefficient, ∈ is the fifth weight coefficient and ζ represents the sixth weight coefficient; K j K represents the criticality coefficient of the jth data block group; max represents the maximum criticality coefficient; T ow Indicates the current time; T en Indicates the generation time of the data block group; T dead Indicates the data block group deadline; W j Indicates type weight; s j Indicates the size of the data block group; S max Indicates the maximum data block group size; d j Indicates the transmission distance; D max Indicates the maximum transmission distance; B j represents the protocol complexity weight coefficient; E j Indicates the current bandwidth utilization; G indicates the maximum bandwidth utilization.

[0041] The present invention proposes a microgrid control system, comprising a preprocessing module, a prediction module and a control module;

[0042] The preprocessing module is used to detect the operating status of each device in the microgrid in real time and obtain the load demand data of each device; after preprocessing the load demand data, the characteristic data of the load demand data is extracted;

[0043] The prediction module is used to obtain the historical operating status data of each device in the microgrid, input the historical operating status data into a preset deep learning algorithm, and predict the future load demand and renewable energy output of the microgrid to obtain prediction data;

[0044] The control module is used to determine the dispatching strategy corresponding to the power equipment based on the prediction data and characteristic data using a genetic algorithm; generate control instructions for the equipment according to the dispatching strategy, send the control instructions to the corresponding power equipment, and monitor the execution of the power equipment.

[0045] The effects provided in the content of the invention are only the effects of the embodiments, not all the effects of the invention. One of the above technical solutions has the following advantages or beneficial effects:

[0046] The present invention proposes a microgrid control method and system, which includes the following steps: real-time detection of the operating status of each device in the microgrid, and acquisition of the load demand data of each device; extracting the characteristic data of the load demand data after preprocessing the load demand data; acquiring the historical operating status data of each device in the microgrid, and inputting the historical operating status data into a preset deep learning algorithm, and predicting the future load demand and renewable energy output of the microgrid to obtain prediction data; based on the prediction data and characteristic data, using a genetic algorithm to determine the dispatching strategy corresponding to the power equipment; generating control instructions for the equipment according to the dispatching strategy, and sending the control instructions to the corresponding power equipment, and monitoring the execution of the power equipment. Based on a microgrid control method, a microgrid control system is also proposed. By real-time monitoring and collecting the operating status and load demand data of each device in the microgrid, the present invention can accurately grasp the real-time situation of the power grid and provide accurate data support for subsequent dispatching and control.

[0047] The present invention pre-processes the collected data, can remove noise and outliers, and improve data quality. At the same time, it extracts feature data useful for power equipment control, which helps to formulate more accurate and efficient scheduling strategies.

[0048] The present invention uses deep learning algorithms to analyze historical data and can accurately predict the future load demand and renewable energy output of microgrids. This prediction capability enables the system to respond in advance and optimize resource allocation, thereby improving the stability and economy of microgrids.

[0049] The present invention uses a genetic algorithm to determine the dispatching strategy corresponding to the power equipment based on the prediction results and the characteristic data of the current equipment. This intelligent dispatching strategy can be dynamically adjusted according to the actual situation to ensure that the power equipment operates in the optimal state and improve energy utilization efficiency.

[0050] The present invention generates specific device-level control instructions according to the dispatching strategy, including the start and stop of the generator set, power output adjustment, charge and discharge control of the energy storage device, and switching of the load device, etc. This refined control method can ensure that each device operates according to the optimal strategy, further improving the operating efficiency of the microgrid.

[0051] The present invention sends control instructions to the corresponding power equipment through the communication network and monitors the execution of the power equipment in real time. This efficient information transmission and feedback mechanism can ensure the effective execution of the dispatching strategy, timely discover and deal with potential problems, and ensure the safe and stable operation of the microgrid. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A flow chart of a microgrid control method proposed in Embodiment 1 of the present invention;

[0053] Figure 2 This is a schematic diagram of a microgrid control system proposed in Example 2 of the present invention. DETAILED DESCRIPTION

[0054] In order to clearly illustrate the technical features of the present solution, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings. The disclosure below provides many different embodiments or examples for realizing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and settings of specific examples are described below. In addition, the present invention may repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. It should be noted that the components illustrated in the accompanying drawings are not necessarily drawn to scale. The present invention omits the description of known components and processing techniques and processes to avoid unnecessary limitations on the present invention.

[0055] Example 1

[0056] A microgrid control method proposed in Embodiment 1 of the present invention is used to solve the technical problems existing in the dispatching control of microgrids in the prior art. Figure 1 This is a flow chart of a microgrid control method proposed in Example 1 of the present invention.

[0057] In step S100, the operating status of each device in the microgrid is detected in real time to obtain the load demand data of each device.

[0058] In step S200, after preprocessing the load demand data, characteristic data of the load demand data is extracted.

[0059] Traverse the load data of each time period in the preprocessed data; use the maximum value function to find the maximum load value in each cycle and mark it as the load peak value;

[0060] Traverse the load data of each time period in the preprocessed data, use the minimum function to determine the minimum load value in each period, and mark it as the load valley value;

[0061] Define the time window for volatility calculation. In each time window, calculate the standard deviation of load data as the load volatility in the window.

[0062] The extracted load peak value, load valley value and load fluctuation rate are fused to obtain a fused data set;

[0063] Randomly select K data points from the fused data set as the initial cluster centers. For each data point in the fused data set, calculate the distance between it and each cluster center, and assign each data point to the cluster center corresponding to the distance to form K clusters.

[0064] For each data point in the fused dataset, Calculate the distance between it and each cluster center.

[0065] Among them, x i is a data point, which indicates the status data of a certain moment or a certain device in the microgrid; c k is the cluster center, which represents the center point in cluster k; x i -c k is the data point x i With cluster center c k The difference vector between them, that is, the difference between the data point and the cluster center in each feature dimension; Σ -1 is the inverse matrix of the covariance matrix Σ; W is a weighting matrix; σ is the scale parameter in the Gaussian kernel function; T represents transpose, which is an operation required for dimensional matching in matrix operations; i represents the index of the data point; k represents the index of the cluster center.

[0066] For each cluster, the average value of all data points in it is calculated, and the average value is used as the new cluster center. The operation is iterated until the preset number of iterations is reached; for each cluster, the data point corresponding to the distance from the cluster center is used as the representative data point of the cluster.

[0067] The representative data points of each cluster are used as fused data points, where the representative data points include load peak value, load valley value and load fluctuation rate.

[0068] In step S300, historical operating status data of each device in the microgrid is obtained, and the historical operating status data is input into a preset deep learning algorithm to predict the future load demand and renewable energy output of the microgrid to obtain prediction data;

[0069] Collect historical load demand data of the microgrid, including time series of electricity consumption and peak load; Collect historical data of renewable energy production, including solar power generation and wind power generation;

[0070] Preprocessing historical load demand data and historical data of renewable energy output to obtain a preprocessed data set; dividing the preprocessed data set into a training set, a validation set, and a test set;

[0071] The LSTM network is trained using the training set, and the LSTM parameters are adjusted by the back propagation algorithm and the gradient descent optimizer to minimize the error between the predicted value and the true value. During the training process, the trained LSTM network is evaluated using the test set to obtain the evaluation results.

[0072] The LSTM network is tuned according to the evaluation results, and the trained LSTM network is used to predict the future load demand and renewable energy output of the microgrid. By inputting future timestamps or related features, the trained LSTM network will output the corresponding predicted values.

[0073] In step S400, a dispatching strategy corresponding to the power equipment is determined using a genetic algorithm based on the prediction data and the characteristic data.

[0074] The prediction results of future load demand and renewable energy output obtained in step S300 are organized into a data set; characteristic data of current power equipment are collected, including equipment capacity, efficiency, operating status, maintenance requirements, etc.; the prediction results are integrated with the equipment characteristic data to form a comprehensive data set including time, load demand, renewable energy output and equipment characteristics;

[0075] Define an objective function based on the microgrid’s operational objectives and constraints;

[0076] The dispatching decision of power equipment is encoded into genes in the genetic algorithm, and each gene represents a specific dispatching action or equipment state;

[0077] Randomly generate an initial population, where each individual represents a possible scheduling strategy;

[0078] For each individual in the population (i.e., each scheduling strategy), its fitness is evaluated using the defined objective function; based on the fitness value, individuals with better performance in the population are selected as parents, and new offspring individuals are generated through crossover operations (such as single-point crossover, multi-point crossover, etc.) to inherit the excellent characteristics of the parents, and mutation operations are performed on offspring individuals (such as randomly changing the values ​​of certain genes) to increase the diversity and exploration ability of the population;

[0079] Repeat the fitness evaluation, selection, crossover and mutation operations until the termination conditions are met (such as reaching the preset number of iterations, fitness value convergence, etc.). During the iteration process, record and update the optimal solution (i.e., the individual with the highest fitness value), and decode the optimal solution (individual) obtained by the genetic algorithm into a specific power equipment scheduling strategy.

[0080] The expression of the objective function is:

[0081]

[0082] Where K is the index of the power equipment; R is the total number of equipment; t is the index of the time step; T is the total number of time steps; C gen,K represents the unit power generation cost of equipment K; P gen,K (t) represents the power generated by device K at time t; E gen,K (t) represents the efficiency of equipment K at time t.

[0083] The dispatching decisions of power equipment are encoded as genes in the genetic algorithm, each gene represents a specific dispatching action or equipment state; an initial population is randomly generated, in which each individual represents a possible dispatching strategy.

[0084] Generate new offspring individuals through crossover operations (such as single-point crossover, multi-point crossover, etc.). Randomly select two individuals from the current population as parents, and these two individuals will participate in the crossover operation to generate new offspring; randomly select a position in the gene sequence of the parent individual as the crossover point, which is an integer index between the start and end of the gene sequence; exchange the part of the first parent individual from the crossover point to the end of the gene sequence with the part of the second parent individual at the same position; similarly, exchange the part of the second parent individual from the crossover point to the end of the gene sequence with the part of the first parent individual at the same position, so that two new offspring individuals are generated, each of which contains a part of the genes from the first parent and a part of the genes from the second parent; check whether the generated offspring individuals meet the constraints of the problem (such as the validity and range of the genes, etc.). If the offspring individuals are invalid, valid offspring can be generated by reselecting parents or re-performing the crossover operation.

[0085] Similar to single-point crossover, two individuals are randomly selected from the current population as parents; multiple positions are randomly selected in the gene sequence of the parent individuals as crossover points, which divide the gene sequence into multiple fragments; for each pair of adjacent gene fragments between crossover points, it is randomly decided whether to keep the fragment of the first parent or the fragment of the second parent. Based on this decision, the corresponding fragment is copied from the selected parent to the new offspring individual. This process is repeated until all fragments are processed, thereby generating two new offspring individuals. Similar to single-point crossover, check whether the generated offspring individuals meet the constraints of the problem. If the offspring individuals are invalid, take corresponding measures to generate valid offspring.

[0086] In step S500, control instructions for the equipment are generated according to the scheduling strategy, including the start and stop of the generator set, power output adjustment, charge and discharge control of the energy storage equipment, and switching of the load equipment.

[0087] In step S600, the control instruction is sent to the corresponding power device, and the execution status of the power device is monitored.

[0088] Divide the control instructions into multiple independent data blocks according to preset rules; the preset rules adopt data length, function type, etc.;

[0089] Performing cluster analysis on the segmented data blocks to obtain clustered data block groups, and then calculating the weights of the data block groups;

[0090] Among them, the process of clustering the divided data blocks to obtain the clustered data block group includes: extracting features for clustering according to the data size, type and urgency of the data blocks; setting a cluster number K, where the K value represents the number of groups into which the data blocks are divided, randomly selecting K data blocks as initial cluster centers, and the cluster centers as the starting points of the K-means algorithm; for each data block, calculating the Euclidean distance between each data block and the corresponding first cluster center, assigning each data block to the corresponding first cluster center, and forming multiple initial clusters; calculating the average value of all data blocks to obtain the second cluster center, reallocating the data blocks according to the second cluster center, and updating the position of the second cluster center until the preset maximum number of iterations is reached to obtain the final clustering result and output the data block group.

[0091] The comprehensive score of the data block group is calculated according to the weight of each data block group, the size of the data block group and the transmission difficulty. According to the comprehensive score, the data block groups are sorted from high to low in priority, and the sorted data blocks are sent to the corresponding power equipment in sequence.

[0092] The formula for calculating the comprehensive score is:

[0093]

[0094] Among them, CS i represents the comprehensive priority score of the i-th data block group; α is the first weight coefficient, β is the second weight coefficient, γ is the third weight coefficient, δ is the fourth weight coefficient, ∈ is the fifth weight coefficient and ζ represents the sixth weight coefficient; K i represents the criticality coefficient of the i-th data block group; K max represents the maximum criticality coefficient; T ow Indicates the current time; T en Indicates the generation time of the data block group; T dead Indicates the data block group deadline; W i Indicates type weight; s i Indicates the size of the data block group; S max Indicates the maximum data block group size; d i Indicates the transmission distance; D max Indicates the maximum transmission distance; B i represents the protocol complexity weight coefficient; E i Indicates the current bandwidth utilization; G indicates the maximum bandwidth utilization.

[0095] A microgrid control method proposed in Example 1 of the present invention can accurately grasp the real-time situation of the power grid by real-time monitoring and collecting the operating status and load demand data of each device in the microgrid, and provide accurate data support for subsequent scheduling and control.

[0096] A microgrid control method proposed in Embodiment 1 of the present invention preprocesses the collected data to remove noise and abnormal values ​​and improve data quality. At the same time, feature data useful for power equipment control is extracted to help formulate a more accurate and efficient scheduling strategy.

[0097] A microgrid control method proposed in Embodiment 1 of the present invention uses a deep learning algorithm to analyze historical data, and can accurately predict the future load demand and renewable energy output of the microgrid. This prediction capability enables the system to respond in advance and optimize resource allocation, thereby improving the stability and economy of the microgrid.

[0098] A microgrid control method proposed in Example 1 of the present invention uses a genetic algorithm to determine the dispatching strategy corresponding to the power equipment based on the prediction results and the characteristic data of the current equipment. This intelligent dispatching strategy can be dynamically adjusted according to the actual situation to ensure that the power equipment operates in the optimal state and improve energy utilization efficiency.

[0099] A microgrid control method proposed in Example 1 of the present invention generates specific device-level control instructions according to the dispatching strategy, including the start and stop of the generator set, power output adjustment, charge and discharge control of the energy storage device, and switching of the load device, etc. This refined control method can ensure that each device operates according to the optimal strategy, further improving the operating efficiency of the microgrid.

[0100] A microgrid control method proposed in Example 1 of the present invention sends control instructions to corresponding power equipment through a communication network and monitors the execution of the power equipment in real time. This efficient information transmission and feedback mechanism can ensure the effective execution of the scheduling strategy, timely discover and deal with potential problems, and ensure the safe and stable operation of the microgrid.

[0101] Example 2

[0102] Based on a microgrid control method proposed in Embodiment 1 of the present invention, Embodiment 2 of the present invention further proposes a microgrid control system. Figure 2 A schematic diagram of a microgrid control system proposed in Embodiment 2 of the present invention, the system includes a preprocessing module, a prediction module and a control module;

[0103] The preprocessing module is used to detect the operating status of each device in the microgrid in real time and obtain the load demand data of each device; after preprocessing the load demand data, the characteristic data of the load demand data is extracted;

[0104] The prediction module is used to obtain the historical operating status data of each device in the microgrid, input the historical operating status data into the preset deep learning algorithm, and predict the future load demand and renewable energy output of the microgrid to obtain prediction data;

[0105] The control module is used to determine the dispatching strategy corresponding to the power equipment based on the predicted data and characteristic data by using a genetic algorithm; generate control instructions for the equipment according to the dispatching strategy, send the control instructions to the corresponding power equipment, and monitor the execution of the power equipment.

[0106] In the preprocessing module, after the load demand data is preprocessed, the process of extracting the characteristic data of the load demand data includes: traversing the load demand data of each time period after preprocessing, using the maximum function to determine the maximum load value in each cycle, and marking it as the load peak value; traversing the load demand data of each time period after preprocessing, using the minimum function to determine the minimum load value in each cycle, and marking it as the load valley value; defining a time window for volatility calculation, and in each time window, calculating the standard deviation of the load demand data as the load volatility in the current window; fusing the load peak value, load valley value and load volatility to obtain a fused data set; selecting K data points from the fused data set as initial clustering centers, for each data point in the fused data set, calculating the distance between each data point and each clustering center, and assigning each data point to the clustering center corresponding to the distance to form K clusters.

[0107] For each cluster, the average value of all data points is calculated, and the average value is used as the cluster center again. The operation is iterated until the preset number of iterations is reached; for each cluster, the data point corresponding to the distance from the cluster center is used as the representative data point of the cluster; the representative data point of each cluster is used as the fused data point, where the representative data points include load peak value, load valley value and load fluctuation rate.

[0108] The process of calculating the distance between each data point and each cluster center includes:

[0109]

[0110] Among them, x i is the data point; c k is the cluster center; i represents the index of the data point; k represents the index of the cluster center; x i -c k is the data point x i With cluster center c k The difference vector between -1 is the inverse matrix of the covariance matrix Σ; W is the weighting matrix; σ is the scale parameter in the Gaussian kernel function; T represents the transpose.

[0111] In the prediction module, the historical operating status data is input into the preset deep learning algorithm, and the process of predicting the future load demand and renewable energy output of the microgrid to obtain the prediction data includes: dividing the historical operating status data into a training set, a validation set, and a test set according to a preset ratio; using the training set to train the long short-term memory network, and using the test set to evaluate the trained long short-term memory network to obtain the evaluation result; tuning the long short-term memory network according to the evaluation result, and using the trained long short-term memory network to predict the future load demand and renewable energy output of the microgrid to obtain the prediction data.

[0112] In the control module, the process of using the genetic algorithm to determine the dispatch strategy corresponding to the power equipment based on the predicted data and the characteristic data includes: integrating the predicted data and the characteristic data to form a comprehensive data set including time, load demand, renewable energy output and equipment characteristics; determining the objective function according to the operation objectives and constraints of the microgrid; the expression of the objective function is:

[0113]

[0114] Where K is the index of the power equipment; R is the total number of equipment; t is the index of the time step; T is the total number of time steps; C gen,K represents the unit power generation cost of equipment K; P gen,K (t) represents the power generated by device K at time t; E gen,K (t) represents the efficiency of equipment K at time t.

[0115] The dispatching decisions of power equipment are encoded as genes in the genetic algorithm, each gene represents a specific dispatching action or equipment state; an initial population is randomly generated, in which each individual represents a possible dispatching strategy.

[0116] The control instructions include starting and stopping the generator set, adjusting the power output, charging and discharging control of the energy storage equipment, and switching on and off of the load equipment.

[0117] The specific process of sending the control instruction to the corresponding power equipment and monitoring the execution of the power equipment includes: dividing the control instruction into multiple independent data blocks according to preset rules; clustering the divided data blocks to obtain clustered data block groups, and then calculating the weight of the data block groups; calculating the comprehensive score of the data block groups according to the weight of each data block group, the size of the data block group and the transmission difficulty, and according to the comprehensive score, sorting the data block groups from high to low and sending them to the corresponding power equipment in turn.

[0118] The process of clustering the segmented data blocks to obtain the clustered data block group includes: calculating the Euclidean distance between each data block and the corresponding first cluster center, assigning each data block to the corresponding first cluster center to form multiple initial clusters; calculating the average value of all data blocks to obtain the second cluster center, reallocating the data blocks according to the second cluster center, and updating the position of the second cluster center until the preset maximum number of iterations is reached to obtain the final clustering result and output the data block group.

[0119] The formula for calculating the comprehensive score is:

[0120]

[0121] Among them, CSi represents the comprehensive priority score of the i-th data block group; α is the first weight coefficient, β is the second weight coefficient, γ is the third weight coefficient, δ is the fourth weight coefficient, ∈ is the fifth weight coefficient and ζ represents the sixth weight coefficient; K i represents the criticality coefficient of the i-th data block group; K max represents the maximum criticality coefficient; T ow Indicates the current time; T en Indicates the generation time of the data block group; T dead Indicates the data block group deadline; W i Indicates type weight; s i Indicates the size of the data block group; S max Indicates the maximum data block group size; d i Indicates the transmission distance; D max Indicates the maximum transmission distance; B i represents the protocol complexity weight coefficient; E i Indicates the current bandwidth utilization; G indicates the maximum bandwidth utilization.

[0122] A microgrid control system proposed in Example 2 of the present invention can accurately grasp the real-time situation of the power grid by real-time monitoring and collecting the operating status and load demand data of each device in the microgrid, and provide accurate data support for subsequent scheduling and control.

[0123] A microgrid control system proposed in Embodiment 2 of the present invention preprocesses the collected data to remove noise and abnormal values ​​and improve data quality. At the same time, it extracts characteristic data useful for power equipment control, which helps to formulate a more accurate and efficient dispatching strategy.

[0124] A microgrid control system proposed in Embodiment 2 of the present invention uses a deep learning algorithm to analyze historical data and can accurately predict the future load demand and renewable energy output of the microgrid. This prediction capability enables the system to respond in advance and optimize resource allocation, thereby improving the stability and economy of the microgrid.

[0125] A microgrid control system proposed in Example 2 of the present invention uses a genetic algorithm to determine the dispatching strategy corresponding to the power equipment based on the prediction results and the characteristic data of the current equipment. This intelligent dispatching strategy can be dynamically adjusted according to the actual situation to ensure that the power equipment operates in the optimal state and improve energy utilization efficiency.

[0126] A microgrid control system proposed in Example 2 of the present invention generates specific device-level control instructions according to the dispatching strategy, including the start and stop of the generator set, power output adjustment, charge and discharge control of the energy storage device, and switching of the load device, etc. This refined control method can ensure that each device operates according to the optimal strategy, further improving the operating efficiency of the microgrid.

[0127] A microgrid control system proposed in Example 2 of the present invention sends control instructions to corresponding power equipment through a communication network and monitors the execution of the power equipment in real time. This efficient information transmission and feedback mechanism can ensure the effective execution of the scheduling strategy, timely discover and deal with potential problems, and ensure the safe and stable operation of the microgrid.

[0128] The description of the relevant parts of a microgrid control system provided in Example 2 of the present application can be found in the detailed description of the corresponding parts of a microgrid control method provided in Example 1 of the present application, and will not be repeated here.

[0129] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment that includes a series of elements are inherent to the elements. In the absence of more restrictions, the elements limited by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or equipment that includes the elements. In addition, the above-mentioned technical solution provided in the embodiment of the present application is consistent with the corresponding technical solution in the prior art in principle, and the part is not described in detail, so as not to repeat too much.

[0130] Although the above describes the specific implementation of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. For those skilled in the art, other different forms of modifications or deformations can be made on the basis of the above description. It is not necessary and impossible to list all the implementation methods here. On the basis of the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative work are still within the scope of protection of the present invention.

Claims

1. A microgrid control method, characterized in that: The following steps are involved: Real-time detection of the operating status of each device in the microgrid to obtain the load demand data of each device; after pre-processing the load demand data, the characteristic data of the load demand data is extracted; Obtain the historical operating status data of each device in the microgrid, input the historical operating status data into the preset deep learning algorithm, and predict the future load demand and renewable energy output of the microgrid to obtain forecast data; Based on the predicted data and characteristic data, a genetic algorithm is used to determine the dispatching strategy corresponding to the power equipment; a control instruction of the equipment is generated according to the dispatching strategy, and the control instruction is sent to the corresponding power equipment, and the execution of the power equipment is monitored.

2. A microgrid control method according to claim 1, characterized in that: After the load demand data is preprocessed, the process of extracting characteristic data of the load demand data includes: Traverse the load demand data of each time period after preprocessing, use the maximum value function to determine the maximum load value in each cycle, and mark it as the load peak value; Traverse the load demand data of each time period after preprocessing, use the minimum function to determine the minimum load value in each cycle, and mark it as the load valley value; Define the time window for fluctuation calculation. In each time window, calculate the standard deviation of load demand data as the load fluctuation rate in the current window. The load peak value, the load valley value and the load fluctuation rate are fused to obtain a fused data set; K data points are selected from the fused data set as initial cluster centers, and for each data point in the fused data set, the distance between each data point and each cluster center is calculated, and each data point is assigned to a cluster center corresponding to the distance to form K clusters; For each cluster, the average value of all data points is calculated, and the average value is used as the cluster center again. The operation is iterated until the preset number of iterations is reached; for each cluster, the data point corresponding to the distance from the cluster center is used as the representative data point of the cluster; the representative data point of each cluster is used as the fused data point, where the representative data points include load peak value, load valley value and load fluctuation rate.

3. A microgrid control method according to claim 2, characterized in that: The process of calculating the distance between each data point and each cluster center includes: Among them, x i is the data point; c k is the cluster center; i represents the index of the data point; k represents the index of the cluster center; x i -c k is the data point x i With cluster center c k The difference vector between -1 is the inverse matrix of the covariance matrix Σ; W is the weighting matrix; σ is the scale parameter in the Gaussian kernel function; T represents the transpose.

4. A microgrid control method according to claim 1, characterized in that: The process of inputting historical operating status data into a preset deep learning algorithm to predict the future load demand and renewable energy output of the microgrid to obtain the predicted data includes: Divide the historical operation status data into training set, validation set and test set according to the preset ratio; The long short-term memory network is trained using the training set, and the long short-term memory network after training is evaluated using the test set to obtain the evaluation result; The LSTM network is tuned according to the evaluation results, and the trained LSTM network is used to predict the future load demand and renewable energy output of the microgrid to obtain prediction data.

5. A microgrid control method according to claim 1, characterized in that: Based on the prediction data and the characteristic data, the process of using the genetic algorithm to determine the dispatching strategy corresponding to the power equipment includes: integrating the forecast data and the characteristic data to form a comprehensive data set including time, load demand, renewable energy production and equipment characteristics; According to the operation objectives and constraints of the microgrid, the objective function is determined; the expression of the objective function is: Where K is the index of the power equipment; R is the total number of equipment; t is the index of the time step; T is the total number of time steps; C gen,K represents the unit power generation cost of equipment K; P gen,K (t) represents the power generated by device K at time t; E gen,K (t) represents the efficiency of equipment K at time t; The dispatching decisions of power equipment are encoded as genes in the genetic algorithm, and each gene represents a specific dispatching action or equipment state; An initial population is randomly generated, in which each individual represents a possible scheduling strategy.

6. A microgrid control method according to claim 1, characterized in that: The control instructions include starting and stopping the generator set, adjusting the power output, charging and discharging control of the energy storage device, and switching on and off of the load device.

7. A microgrid control method according to claim 1, characterized in that: The specific process of sending control instructions to corresponding power equipment and monitoring the execution of power equipment includes: Dividing the control instruction into multiple independent data blocks according to preset rules; Performing cluster analysis on the segmented data blocks to obtain clustered data block groups, and then calculating the weights of the data block groups; The comprehensive score of the data block group is calculated according to the weight of each data block group, the size of the data block group and the transmission difficulty. According to the comprehensive score, the data block groups are sorted from high to low and sent to the corresponding power equipment in sequence.

8. A microgrid control method according to claim 7, characterized in that: The process of clustering the segmented data blocks to obtain clustered data block groups includes: Calculate the Euclidean distance between each data block and the corresponding first cluster center, assign each data block to the corresponding first cluster center, and form multiple initial clusters; The average value of all data blocks is calculated to obtain the second cluster center, and the data blocks are redistributed according to the second cluster center, and the position of the second cluster center is updated until the preset maximum number of iterations is reached to obtain the final clustering result and output the data block group.

9. A microgrid control method according to claim 8, characterized in that: The calculation formula of the comprehensive score is: Among them, CS i represents the comprehensive priority score of the i-th data block group; α is the first weight coefficient, β is the second weight coefficient, γ is the third weight coefficient, δ is the fourth weight coefficient, ∈ is the fifth weight coefficient and ζ represents the sixth weight coefficient; K i represents the criticality coefficient of the i-th data block group; K max represents the maximum criticality coefficient; T ow Indicates the current time; T en Indicates the generation time of the data block group; T dead Indicates the data block group deadline; W i Indicates type weight; s i Indicates the size of the data block group; S max Indicates the maximum data block group size; d i Indicates the transmission distance; D max Indicates the maximum transmission distance; B i represents the protocol complexity weight coefficient; E i Indicates the current bandwidth utilization; G indicates the maximum bandwidth utilization.

10. A microgrid control system, characterized in that: It includes a preprocessing module, a prediction module and a control module; The preprocessing module is used to detect the operating status of each device in the microgrid in real time and obtain the load demand data of each device; after preprocessing the load demand data, the characteristic data of the load demand data is extracted; The prediction module is used to obtain the historical operating status data of each device in the microgrid, input the historical operating status data into a preset deep learning algorithm, and predict the future load demand and renewable energy output of the microgrid to obtain prediction data; The control module is used to determine the dispatching strategy corresponding to the power equipment based on the prediction data and characteristic data using a genetic algorithm; generate control instructions for the equipment according to the dispatching strategy, send the control instructions to the corresponding power equipment, and monitor the execution of the power equipment.

Images