Wind-solar off-grid micro-grid system and optimization control method

Abstract

The invention provides a wind and light off-grid micro-grid system and an optimization control method, and relates to the technical field of clean energy, the system comprises a photovoltaic power generation system, a wind power generation system, an optimization control system, an energy storage system and a load, the wind and light off-grid micro-grid system is not connected with a power grid, but uses a storage battery and the like as the energy storage system; the system is mainly composed of a photovoltaic power generation system, a wind power generation system and a power utilization system (load), and the photovoltaic power generation system and the wind power generation system comprise a photovoltaic array, a direct current combiner box, a direct current combiner cable, a photovoltaic controller, a storage battery pack, a photovoltaic off-grid inverter and the like. The system is mainly used for remote areas without power grids and with scattered population or is used for meeting some special requirements, application is very flexible, and therefore development of the system has important practical significance. By introducing the intelligent prediction algorithm and the multi-target optimization control strategy, the power supply stability and economy of the wind-solar off-grid micro-grid system are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of clean energy technology, and in particular to a wind-solar off-grid microgrid system and an optimization control method. Background Art

[0002] With the transformation of the global energy structure and the rapid development of renewable energy, wind and solar energy, as representatives of clean energy, have gradually become important components of off-grid microgrid systems. By combining wind power generation and photovoltaic power generation, wind-solar off-grid microgrid systems can effectively utilize natural resources and provide stable power supply to remote areas, islands, mountainous areas and other areas without grid coverage. However, wind and solar energy have significant intermittent and volatile characteristics, resulting in poor power supply stability of the system, and the service life and economy of energy storage devices are also facing challenges.

[0003] Traditional off-grid wind and solar microgrid systems mostly use simple energy storage devices (such as lead-acid batteries) to balance power generation and power demand, but there are the following problems: 1. Low energy management efficiency: Traditional systems lack accurate predictions of wind and solar power generation, resulting in unreasonable charging and discharging strategies for energy storage devices, which can easily lead to energy waste or insufficient power supply; 2. Insufficient economic efficiency: The cost of energy storage devices is high, and traditional systems fail to fully utilize the complementarity of wind and solar energy, resulting in poor overall economic efficiency of the system; 3. Insufficient prediction accuracy: Existing prediction methods are mostly based on a single model, which makes it difficult to accurately capture the multi-factor coupling characteristics of wind energy, solar energy and load demand, resulting in large deviations in prediction results.

[0004] Therefore, it is necessary to design a wind-solar off-grid microgrid system and an optimization control method. Summary of the invention

[0005] In order to overcome the deficiencies of the prior art, an object of the present invention is to provide a wind-solar off-grid microgrid system and an optimization control method.

[0006] To achieve the above object, the present invention provides the following solutions: The present invention provides a wind-solar off-grid microgrid system and an optimization control method, comprising: a photovoltaic power generation system, a wind power generation system, an optimization control system, an energy storage system and a load, wherein the photovoltaic power generation system and the wind power generation system are connected to the energy storage system and the load through the optimization control system; The optimization control system includes a data acquisition module, a photovoltaic power generation prediction module, a wind power generation prediction module, a load prediction module, and a scheduling optimization module. The data acquisition module is used to collect relevant data on photovoltaic power generation, wind power generation and load power consumption. The photovoltaic power generation prediction module is used to predict the photovoltaic power generation data at a certain time. The wind power generation prediction module is used to predict the wind power generation data at a certain time. The load prediction module is used to predict the load data at a certain time. The scheduling optimization module is used to perform scheduling optimization based on the predicted photovoltaic power generation data, wind power generation data and load data.

[0007] The present invention also provides an optimization control method for a wind-solar off-grid microgrid system, comprising: Step 1: The data acquisition module collects relevant data of photovoltaic power generation, wind power generation and load power consumption; Step 2: The photovoltaic power generation prediction module predicts photovoltaic power generation at a certain time based on the relevant data of photovoltaic power generation; Step 3: The wind power generation prediction module performs wind power generation prediction at the same time based on the relevant data of wind power generation; Step 4: The load prediction module performs load prediction at the same time based on the relevant data of load power consumption; Step 5: The scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time.

[0008] Preferably, in step 2, the photovoltaic power generation prediction module performs photovoltaic power generation prediction at a certain time based on relevant data of photovoltaic power generation, specifically: Step 201: Acquire relevant data of photovoltaic power generation; Step 202: Building a photovoltaic power generation prediction model based on the Adaboost ensemble learning algorithm and the CNN-LSTM network structure; Step 203: Based on the relevant data of photovoltaic power generation, photovoltaic power generation prediction at a certain time is performed through a photovoltaic power generation prediction model.

[0009] Preferably, in step 3, the wind power generation prediction module performs wind power generation prediction at the same time based on relevant data of wind power generation, specifically: Step 301: Acquire relevant data of wind power generation; Step 302: constructing a wind power generation prediction model based on the SSA-VMD-SPSA-LSTM network structure; Step 303: training a wind power generation prediction model based on relevant data of wind power generation; Step 304: Perform wind power generation simultaneously with photovoltaic power generation prediction based on the trained wind power generation prediction model.

[0010] Preferably, in step 4, the load prediction module performs load prediction at the same time based on relevant data of load power consumption, specifically: Step 401: Obtaining data related to load power consumption; Step 402: construct a load prediction model based on the SSA-CNN-LSTM-SES network structure; Step 403: training a load prediction model based on relevant data of load power consumption; Step 404: Perform load prediction simultaneously with photovoltaic power generation prediction based on the trained load prediction model.

[0011] Preferably, in step 5, the scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time, specifically: Based on the photovoltaic power generation data, wind power generation data and load data at the predicted time, determine whether there is excess output power. If so, charge the energy storage system; If the output power is insufficient, the energy storage system is controlled to discharge; In scheduling optimization, the objective function is established based on the energy storage aging cost and power generation maintenance cost: ; In the formula, is the energy storage aging cost, and the unit charge and discharge aging cost of energy storage is: ; The corresponding mathematical expression is: ; In the formula, is the replacement cost of energy storage, Le is the logarithmic function of the number of energy storage cycles and the depth of discharge, is the discharge depth of the energy storage; In summary, the aging cost expression of energy storage is: ; In the formula, is the unit aging cost of energy storage, It is the charging and discharging power of energy storage; The power generation maintenance cost refers to the maintenance and management cost generated during the operation of the power generation unit, and its expression is: ; In the formula, is the power generation maintenance cost coefficient, To generate power for photovoltaic power, Providing power for wind power generation; When scheduling the energy storage system in the system, some constraints need to be met, including: Power balance constraints: When the system is running, the wind power, photovoltaic, energy storage and load in the system must meet the active power balance, that is: ; In the formula, is the fixed load of the system; ; In the formula, is a rigid load; The load can be cut; For transferable loads; It is the charging and discharging power of energy storage; Contribute to photovoltaic power generation; Providing power for wind power generation; The power for interaction with the large power grid; Controllable load constraints: Cuttable load refers to cutting off some unimportant loads according to the power priority to maintain the normal operation of the system. The cuttable load constraints are: ; Transferable load refers to the load that can transfer its usage time when the wind and solar power output is normal and the energy storage can be charged and discharged normally, and has a specific operation cycle; Transferable load time constraints: ; is the time interval during which the load can be transferred; The time to start the transfer; To end the transfer time; Energy storage charging and discharging constraints: In order to ensure the safety and life of energy storage, the state of charge constraint and energy storage power constraint need to be met. The specific constraint conditions are shown in the following formula: ; ; ; In the formula, is the charging power; is the discharge power; is the largest; To ensure that it can be charged or discharged at any time, ; The output of energy storage is taken as the optimization object, and the roulette method is used to select individuals. The constraint conditions are modified in the form of penalty functions to modify the fitness values ​​of individuals, and then the best solutions are retained and the worst solutions are eliminated.

[0012] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: The present invention provides a wind-solar off-grid microgrid system and an optimization control method, the system includes a photovoltaic power generation system, a wind power generation system, an optimization control system, an energy storage system and a load; the method includes: a data acquisition module collects relevant data of photovoltaic power generation, wind power generation and load power consumption, a photovoltaic power generation prediction module performs photovoltaic power generation prediction at a certain time based on the relevant data of photovoltaic power generation, a wind power generation prediction module performs wind power generation prediction at the same time based on the relevant data of wind power generation, a load prediction module performs load prediction at the same time based on the relevant data of load power consumption, and a scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time. The present invention has the following advantages: 1. Improve power supply stability. Through intelligent prediction algorithms, it can accurately predict future wind and solar power generation and load demand. Based on the prediction results, the scheduling optimization module can adjust the charging and discharging strategy of the energy storage device in real time to ensure that the system maintains power supply stability and reduces power outages when power generation fluctuates and load demand changes; 2. Extend the life of energy storage devices. Through multi-objective optimization control strategies, optimize the number and depth of charge and discharge of energy storage devices, avoid over-charge and discharge, extend their service life, introduce energy storage aging cost models, comprehensively consider the number of cycles and discharge depth of energy storage devices, and further optimize the use strategy of energy storage; 3. Reduce system operating costs by optimizing the use of energy storage devices and the operation of backup power generation equipment, reduce system operating costs, improve economic efficiency, introduce a power generation maintenance cost model, comprehensively consider the maintenance costs of photovoltaic power generation and wind power generation, and further optimize the system's economic operation strategy; 4. Realize intelligent scheduling. The scheduling optimization module can realize intelligent scheduling based on the prediction results and real-time collected data to ensure that the system can meet the load demand while extending the life of the energy storage device and reducing the system operating cost. By introducing controllable load constraints and energy storage charging and discharging constraints, the system scheduling strategy is further optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0014] Figure 1 A flow chart of a method provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of the CNN-LSTM network structure; Figure 3 This is a schematic diagram of the CNN-LSTM network structure prediction framework based on the Adaboost ensemble learning method; Figure 4 Schematic diagram of the process of building the SSA-VMD-SPSA-LSTM network structure; Figure 5 This is a schematic diagram of the SSA-CNN-LSTM-SES network structure; Figure 6 A schematic diagram of the specific flow chart for optimizing control calculation; Figure 7 This is a structural diagram of a wind-solar off-grid microgrid system. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] The purpose of the present invention is to provide a wind-solar off-grid microgrid system and an optimization control method. By introducing an intelligent prediction algorithm and a multi-objective optimization control strategy, the power supply stability, energy utilization efficiency and economy of the wind-solar off-grid microgrid system are significantly improved, and it has broad application prospects. At the same time, the technical solution provided by the present invention can effectively solve the problems of low energy management efficiency, poor system stability and insufficient economy in traditional wind-solar off-grid microgrid systems, and provides strong technical support for the efficient utilization of renewable energy and the intelligent development of off-grid microgrid systems.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The present invention provides a wind-solar off-grid microgrid system, comprising: a photovoltaic power generation system, a wind power generation system, an optimization control system, an energy storage system and a load, wherein the photovoltaic power generation system and the wind power generation system are both connected to the energy storage system and the load through the optimization control system.

[0019] The off-grid wind and solar power generation system is not connected to the power grid, but uses batteries as energy storage systems. Its structure is as follows: Figure 7As shown, it is mainly composed of two parts: wind and solar power generation system (photovoltaic power generation system and wind power generation system recorded in the present invention) and power consumption system (load), wherein the power generation system includes photovoltaic array, DC junction box, DC junction cable, photovoltaic controller, battery pack and photovoltaic off-grid inverter, etc. Off-grid photovoltaic power generation system is generally small in scale, mainly used in remote areas without power grid and with dispersed population or to meet certain special needs, and its application is very flexible, so the development of this system has important practical significance.

[0020] The optimization control system includes a data acquisition module, a photovoltaic power generation prediction module, a wind power generation prediction module, a load prediction module, and a scheduling optimization module. The data acquisition module is used to collect relevant data on photovoltaic power generation, wind power generation and load power consumption. The photovoltaic power generation prediction module is used to predict the photovoltaic power generation data at a certain time. The wind power generation prediction module is used to predict the wind power generation data at a certain time. The load prediction module is used to predict the load data at a certain time. The scheduling optimization module is used to perform scheduling optimization based on the predicted photovoltaic power generation data, wind power generation data and load data.

[0021] Introduction to wind power generation system: The function of wind power generation system is to convert wind energy into electrical energy. The main components include wind turbines and generators. Wind turbines are mainly composed of wind rotors, towers, and wind-control devices. The function of wind rotors is to absorb energy from the wind. In actual use, the wind rotors need to be controlled to ensure that the wind rotors will not be damaged due to excessive wind speed. The blades on the wind rotors have a good aerodynamic shape and can generate aerodynamic force under the action of airflow to rotate the wind rotors, thereby converting energy. The tower is a device that supports the wind rotors from the ground. The function of the wind-control device is to ensure that when the wind direction changes, the rotating surface of the wind rotor is always aligned with the wind direction.

[0022] Wind power generation systems are mainly classified into the following categories: 1. Classification by generator: asynchronous, synchronous; 2. According to the direction of the fan main axis: horizontal axis, vertical axis; 3. Classification by speed: fixed speed, variable speed; 4. Power adjustment method according to the wind energy received by the blades: fixed pitch, variable pitch; Although wind turbines come in various types, their working principles are the same, which is to output wind energy in the form of electrical energy. The present invention does not limit their specific models and types.

[0023] The photovoltaic power generation system is introduced: it includes solar photovoltaic cells. Solar photovoltaic cells are metal semiconductor devices that use the photoelectric effect to convert energy. The so-called photoelectric effect is the phenomenon that metal semiconductors release electrons when exposed to light. Photovoltaic cells are generally composed of P-type (hole-type) semiconductors and N-type (electron-type) semiconductors. When two different types of semiconductors are connected, diffusion movement of electrons and holes occurs inside the semiconductor. When sunlight shines on this device, electron-hole pairs are generated because of the sunlight. They move successively to generate a photogenerated electric field. At this time, the semiconductor device is connected to an external circuit to generate current. This is the basic principle of solar photovoltaic cell power generation, which ultimately converts solar energy into electrical energy for output; At present, crystalline silicon is the most commonly used material for making solar cells, and silicon solar cells are also widely used. As for silicon cells, there are mainly amorphous silicon solar photovoltaic cells, polycrystalline silicon solar photovoltaic cells and monocrystalline silicon solar photovoltaic cells. A single cell has a sheet structure and can be manufactured into various shapes such as square and rectangular. Similarly, the present invention does not limit it.

[0024] Introduction to the energy storage system: Because the energy storage system has the characteristics of absorbing and releasing energy, it is often used in off-grid wind and solar microgrid systems to make up for the shortcomings of wind and solar output volatility and randomness, and to adjust the mismatch between output intensity and time and usage. When the wind is strong, in addition to providing energy to the load, the excess energy can also be stored through energy storage; when the wind is not strong enough, the energy in the energy storage can be released to provide to important loads until the wind and solar return to normal output, which improves the power supply reliability and economic benefits of the system to a certain extent.

[0025] The energy storage commonly used in the current wind-solar hybrid power generation system includes: lead-acid batteries, compressed air energy storage, flywheel energy storage, pumped storage, superconducting energy storage, etc.

[0026] Lead-acid batteries are often used in energy storage systems. The principle is that the chemical substances inside the battery are charged and discharged through reversible chemical reactions. Its positive electrode material is aluminum dioxide and its negative electrode is lead. The charging and discharging principle is shown in the formula: ; like Figure 1 As shown, the present invention also provides an optimization control method for a wind-solar off-grid microgrid system, comprising: Step 1: The data acquisition module collects relevant data of photovoltaic power generation, wind power generation and load power consumption; Step 2: The photovoltaic power generation prediction module predicts photovoltaic power generation at a certain time based on the relevant data of photovoltaic power generation; Step 3: The wind power generation prediction module performs wind power generation prediction at the same time based on the relevant data of wind power generation; Step 4: The load prediction module performs load prediction at the same time based on the relevant data of load power consumption; Step 5: The scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time.

[0027] In step 2, the photovoltaic power generation prediction module predicts photovoltaic power generation at a certain time based on the relevant data of photovoltaic power generation, specifically: Step 201: Acquire relevant data of photovoltaic power generation; Step 202: Building a photovoltaic power generation prediction model based on the Adaboost ensemble learning algorithm and the CNN-LSTM network structure; Step 203: Based on the relevant data of photovoltaic power generation, photovoltaic power generation prediction at a certain time is performed through a photovoltaic power generation prediction model.

[0028] In step 201, relevant data of photovoltaic power generation is obtained, specifically: Relevant data for photovoltaic power generation include weather, humidity, wind speed and radiation intensity information.

[0029] In step 202, a photovoltaic power generation prediction model is built based on the Adaboost ensemble learning algorithm and the CNN-LSTM network structure, specifically: First, the CNN-LSTM network structure is introduced: photovoltaic power generation is affected by the coupling of multiple factors such as weather characteristics, ambient temperature, wind speed, and solar radiation intensity. The model input data often contains the power generation and influencing factors of the time series, which are diversified and have large data volume. The CNN model can effectively solve the overfitting problem caused by too many parameters, but it cannot effectively extract feature information of different time scales, which is not conducive to capturing the influence of environmental factors of different time resolutions on power. The LSTM model can capture the time dependency through long and short-time neurons, and can mine the long-term dependence of multiple factors on power generation in the time series. According to the characteristics of photovoltaic power generation prediction and the influencing factors of power generation, the present invention adaptively designs CNN and LSTM models, and effectively extracts feature information by combining the advantages of CNN and LSTM models to ensure that the network can capture the influence of different time scales on power generation. During the forward reasoning process of the convolution kernel, there is a discrepancy between the weight value and the influencing factors of power generation, but the c of the feature extraction layer 11 With c 21The weight matching factors are the same. Therefore, the applicability of the convolution kernel is improved. According to the input sample, the adaptive convolution kernel is designed to perform forward reasoning on the multi-resolution input data to extract the law of the influence of the diversity of power generation factors on power. The lateral size of the adaptive convolution kernel is set to be consistent with the number of power generation factors in the sample data input. The feature layer {c 11 , c 12 , …, c nn The weights in} are consistent with the matching of the power impact factors, which can achieve efficient extraction of feature data and capture local correlation; On this basis, combined with the LSTM network model to learn the long-term dependence advantages of different time scales, the feature data extracted by CNN is output to the LSTM network model, and the gate operation, forgetting layer, learning layer, and output layer are coupled and linked. The specific model structure is as follows Figure 2 As shown in the figure; the CNN feature extraction module uses two sets of adaptive convolution kernels to capture local feature information. To improve the capture speed, if the number of power influencing factors is 4, the convolution kernel adopts a 3 × 4 structure to learn the low-level feature information of the initial layer structure. The second layer adopts the MaxPooling pooling structure to reduce the limitation of the invariance of the feature layer information mapping and reduce the overall calculation amount of the model. The CNN feature structure design adopts a feature extraction combination of two sets of convolution pooling and activation functions. Considering the idea of ​​extracting feature information from coarse to fine by the convolution kernel, these two sets of feature extraction modules adopt a pyramid structure. The kernel of the first set of convolution layers adopts a dimension of 48, and the kernel of the second set of convolution layers adopts a dimension of 32. This structure can further avoid overfitting and reduce the number of trainable parameters. The LSTM link uses 32 long and short-time neuron structures to capture the feature dependency in the time series. The Dropout layer is connected before the fully connected layer, and neurons are randomly selected at a ratio of 0.25 to further alleviate the overfitting problem. Finally, the network structure connects the fully connected layer built with 20 neurons to output the prediction information of photovoltaic power. Based on the Adaboost ensemble learning method, an ensemble learning framework for photovoltaic power generation prediction is constructed. The overall prediction process of the model is as follows: first, the data set is cut and divided, and the training set and test set of the network model are randomly divided in proportion, and the weight value of the initial weak learner is set; the improved CNN-LSTM neural network power generation prediction model is used to perform forward reasoning of power generation in a cross-validation manner, and the error rate e1 and weight coefficient α1 are calculated between the prediction result and the true power value; the weight value of the second round of weak learner training is reassigned according to the weight coefficient α1, and a weak learner is constructed using the CNN-LSTM neural network. The training set divided by the digital twin system is used for training to obtain a new round of error rate e2 and weight coefficient α2; the weight value of the weak learner is reassigned again until the weight value of the trained weak classifier with a small error rate is the largest, achieving the effect of "selecting the best from the best". The specific prediction framework is shown in the figure below. Figure 3 shown.

[0030] In step 3, the wind power generation prediction module performs wind power generation prediction at the same time based on relevant data of wind power generation, specifically: Step 301: Acquire relevant data of wind power generation; Step 302: constructing a wind power generation prediction model based on the SSA-VMD-SPSA-LSTM network structure; Step 303: training a wind power generation prediction model based on relevant data of wind power generation; Step 304: Perform wind power generation simultaneously with photovoltaic power generation prediction based on the trained wind power generation prediction model.

[0031] In step 301, relevant data of wind power generation is obtained, specifically: First, let’s introduce the fan characteristics: The fan power expression is: ; Where, P is the fan output power, in kW; is the air density, in kg / is the wind rotor power coefficient; A is the wind rotor swept area, unit is According to the formula, P and is proportional to the cube of The wind turbine power coefficient is also proportional to the wind power output. The wind turbine power coefficient is also called the wind energy utilization coefficient. Theoretical analysis shows that the wind turbine power coefficient is proportional to its output power. According to Betz theory, the maximum wind energy utilization coefficient of a wind turbine is 0.593, but this is only a theoretical value. At present, the wind energy utilization coefficient of wind turbines in wind farms is generally around 0.45. In summary, the remaining parameters are known, and the wind speed can be calculated to obtain the wind turbine power, so the present invention mainly predicts the wind speed.

[0032] In step 302, a wind power generation prediction model is constructed based on the SSA-VMD-SPSA-LSTM network structure, specifically: The SSA-VMD-SPSA-LSTM network structure includes the SSA denoising algorithm, VMD data decomposition method, SPSA intelligent optimization algorithm and LSTM network, which will not be introduced in detail; The present invention uses the average interpolation method to estimate the missing values ​​of the original wind speed data set. After ensuring that there is no missing data, SSA is used to remove the noise data to improve the data quality of the original sequence. Then, VMD is used to decompose the denoised data into multiple IMFs and a residual. In order to avoid the extension of the calculation time caused by multiple components, SE is introduced to calculate the time complexity of each IMF. According to similar SE values, these IMFs are reorganized and merged to obtain several new subsequences. After the above two-stage preprocessing, subsequences of different frequencies are predicted by the LSTM model optimized by SPSA. Finally, the predicted values ​​of all components are superimposed to obtain the final prediction result. The specific flow chart is shown as follows. Figure 4 shown.

[0033] In step 4, the load prediction module performs load prediction at the same time based on the relevant data of load power consumption, specifically: Step 401: Obtaining data related to load power consumption; Step 402: construct a load prediction model based on the SSA-CNN-LSTM-SES network structure; Step 403: training a load prediction model based on relevant data of load power consumption; Step 404: Perform load prediction simultaneously with photovoltaic power generation prediction based on the trained load prediction model.

[0034] In step 401, relevant data of load power consumption is obtained, including: date, weather data, and daily load data.

[0035] In step 402, a load prediction model is constructed based on the SSA-CNN-LSTM-SES network structure, specifically: First, we build a CNN-LSTM model. The network in the CNN LSTM model is divided into four layers. The first layer is the input layer, which inputs the processed data into the model. The second layer is the CNN layer, which uses the convolution layer to extract important information from the data, and reduces the dimension through the pooling layer to obtain the output data. The third layer inputs the output data into the LSTM layer for training. The fourth layer is the fully connected layer, which maps the feature space calculated by the previous layer (convolution, pooling, etc.) to the sample label space, and finally obtains the load output value. The single exponential smoothing method (SES) is suitable for short-term prediction and makes up for the long-term dependence of the CNN-LSTM model on data. At the same time, in order to determine the optimal parameters of the above model, the sparrow search algorithm (SSA) is introduced. The sparrow search algorithm not only has higher search accuracy and faster convergence rate, but also has better stability; The construction process of the SSA-CNN-LSTM-SES network structure is as follows: First, use SSA to optimize the hyperparameters of the CNN-LSTM model and the smoothing coefficient of the SES model respectively to obtain the optimized models SSA-CNN-LSTM and SSA-SES. Then, combine the two optimized models, assign corresponding weights to the outputs of the two models, and obtain the output of the final combined model. The combined model process is as follows: Figure 5 shown.

[0036] In step 5, the scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time, specifically: When the wind and solar output cannot meet the load, the energy storage is discharged and the charge of the energy storage is at the critical level of the energy storage. According to the power priority, part of the load is cut off, and only some important equipment is maintained in operation until the wind and solar output can meet the load demand. Then the power supply of the cut load is gradually restored. When the energy storage can operate normally, the use time of part of the load can be shifted from the peak power consumption period to the low power consumption period, or from the time period when the wind and solar output is insufficient to the time period when the wind and solar output is sufficient. Specifically: 1. When the wind and solar power output is sufficient, the output power is in excess, and the energy storage enters the charging state.

[0037] If the charging power is within the constraints of the energy storage, the energy storage will be charged; if the charging power exceeds the constraints of the energy storage, the energy storage charge is at the upper limit, which can stimulate the controllable load to absorb the energy, and transfer it from the time period when the wind and solar power output is insufficient to the time period when the output is sufficient, and from the high price time to the low price time.

[0038] When the amount of power regulated by the controllable load itself reaches the maximum value and there is still surplus power, energy is abandoned.

[0039] 2. When the wind and solar power output is insufficient, the output power is insufficient and the energy storage enters the discharge state.

[0040] If the discharge power can meet the power shortage, the energy storage will be discharged; if the discharge power exceeds the constraint conditions of the energy storage and the energy storage charge is at the lower critical limit, part of the load can be removed.

[0041] When the amount of power regulated by the controllable load itself reaches the maximum value and there is still a power shortage, the backup power supply is enabled.

[0042] In scheduling optimization, the objective function is established based on the energy storage aging cost and power generation maintenance cost: ; The aging cost of energy storage is generally that energy storage needs to be replaced when it drops to 80% of its factory capacity. The number of cycles before this is one of its most important performance indicators, so the unit charge and discharge aging cost of energy storage is: ; The corresponding mathematical expression is: ; In the formula, is the replacement cost of energy storage, Le is the logarithmic function of the number of energy storage cycles and the depth of discharge, is the discharge depth of the energy storage; In summary, the aging cost expression of energy storage is: ; In the formula, is the unit aging cost of energy storage, It is the charging and discharging power of energy storage; The power generation maintenance cost refers to the maintenance and management cost generated during the operation of the power generation unit, and its expression is: ; In the formula, is the power generation maintenance cost coefficient, To generate power for photovoltaic power, Providing power for wind power generation; When scheduling the energy storage system in the system, some constraints need to be met, including: 1. Power balance constraints When the system is running, the wind power, photovoltaic, energy storage and load in the system must meet the active power balance, that is: ; In the formula, is the fixed load of the system; ; In the formula, is a rigid load; The load can be cut; For transferable loads; It is the charging and discharging power of energy storage; Contribute to photovoltaic power generation; Providing power for wind power generation; The power for interaction with the large power grid; 2. Controllable load constraints Cuttable load refers to cutting off some unimportant loads according to the power priority to maintain the normal operation of the system. The cuttable load constraints are: ; Transferable load refers to the load that can transfer its usage time when the wind and solar power output is normal and the energy storage can be charged and discharged normally, and has a specific operation cycle; Transferable load time constraints: ; is the time interval during which the load can be transferred; The time to start the transfer; To end the transfer time; 3. Energy storage charging and discharging constraints In order to ensure the safety and life of energy storage, the state of charge constraint and energy storage power constraint need to be met. The specific constraint conditions are shown in the following formula: ; ; ; In the formula, is the charging power; is the discharge power; is the largest; To ensure that it can be charged or discharged at any time, ; The present invention takes the output of energy storage as the optimization object. This paper adopts the roulette method to select individuals, and modifies the fitness value of individuals in the form of a penalty function. Then, the optimal solution is retained and the inferior solution is eliminated. The specific flow chart is as follows: Figure 6 shown.

[0043] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0044] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A wind-solar off-grid microgrid system, characterized in that: include: Photovoltaic power generation system, wind power generation system, optimization control system, energy storage system and load, the photovoltaic power generation system and wind power generation system are connected to the energy storage system and load through the optimization control system; The optimization control system includes a data acquisition module, a photovoltaic power generation prediction module, a wind power generation prediction module, a load prediction module, and a scheduling optimization module. The data acquisition module is used to collect relevant data on photovoltaic power generation, wind power generation and load power consumption. The photovoltaic power generation prediction module is used to predict the photovoltaic power generation data at a certain time. The wind power generation prediction module is used to predict the wind power generation data at a certain time. The load prediction module is used to predict the load data at a certain time. The scheduling optimization module is used to perform scheduling optimization based on the predicted photovoltaic power generation data, wind power generation data and load data.

2. An optimization control method for a wind-solar off-grid microgrid system, characterized in that: include: Step 1: The data acquisition module collects relevant data of photovoltaic power generation, wind power generation and load power consumption; Step 2: The photovoltaic power generation prediction module predicts photovoltaic power generation at a certain time based on the relevant data of photovoltaic power generation; Step 3: The wind power generation prediction module performs wind power generation prediction at the same time based on the relevant data of wind power generation; Step 4: The load prediction module performs load prediction at the same time based on the relevant data of load power consumption; Step 5: The scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time.

3. The method according to claim 2, characterized in that In step 2, the photovoltaic power generation prediction module predicts photovoltaic power generation at a certain time based on the relevant data of photovoltaic power generation, specifically: Step 201: Acquire relevant data of photovoltaic power generation; Step 202: Building a photovoltaic power generation prediction model based on the Adaboost ensemble learning algorithm and the CNN-LSTM network structure; Step 203: Based on the relevant data of photovoltaic power generation, photovoltaic power generation prediction at a certain time is performed through a photovoltaic power generation prediction model.

4. The method according to claim 3, characterized in that: In step 3, the wind power generation prediction module performs wind power generation prediction at the same time based on relevant data of wind power generation, specifically: Step 301: Acquire relevant data of wind power generation; Step 302: constructing a wind power generation prediction model based on the SSA-VMD-SPSA-LSTM network structure; Step 303: training a wind power generation prediction model based on relevant data of wind power generation; Step 304: Perform wind power generation simultaneously with photovoltaic power generation prediction based on the trained wind power generation prediction model.

5. The method according to claim 4, characterized in that In step 4, the load prediction module performs load prediction at the same time based on the relevant data of load power consumption, specifically: Step 401: Obtaining data related to load power consumption; Step 402: construct a load prediction model based on the SSA-CNN-LSTM-SES network structure; Step 403: training a load prediction model based on relevant data of load power consumption; Step 404: Perform load prediction simultaneously with photovoltaic power generation prediction based on the trained load prediction model.

6. The method according to claim 5, characterized in that In step 5, the scheduling optimization module performs scheduling optimization based on the photovoltaic power generation data, wind power generation data and load data at that time, specifically: Based on the photovoltaic power generation data, wind power generation data and load data at the predicted time, determine whether there is excess output power. If so, charge the energy storage system; If the output power is insufficient, the energy storage system is controlled to discharge; In scheduling optimization, the objective function is established based on the energy storage aging cost and power generation maintenance cost: ； In the formula, is the energy storage aging cost, and the unit charge and discharge aging cost of energy storage is: ； The corresponding mathematical expression is: ； In the formula, is the replacement cost of energy storage, Le is the logarithmic function of the number of energy storage cycles and the depth of discharge, is the discharge depth of the energy storage; In summary, the aging cost expression of energy storage is: ； In the formula, is the unit aging cost of energy storage, It is the charging and discharging power of energy storage; The power generation maintenance cost refers to the maintenance and management cost generated during the operation of the power generation unit, and its expression is: ； In the formula, is the power generation maintenance cost coefficient, To generate power for photovoltaic power, Providing power for wind power generation; When scheduling the energy storage system in the system, some constraints need to be met, including: Power balance constraints: When the system is running, the wind power, photovoltaic, energy storage and load in the system must meet the active power balance, that is: ； In the formula, is the fixed load of the system; ； In the formula, is a rigid load; The load can be cut; For transferable loads; It is the charging and discharging power of energy storage; Contribute to photovoltaic power generation; Providing power for wind power generation; The power for interaction with the large power grid; Controllable load constraints: Cuttable load refers to cutting off some unimportant loads according to the power priority to maintain the normal operation of the system. The cuttable load constraints are: ； Transferable load refers to the load that can transfer its usage time when the wind and solar power output is normal and the energy storage can be charged and discharged normally, and has a specific operation cycle; Transferable load time constraints: ； is the time interval during which the load can be transferred; The time to start the transfer; To end the transfer time; Energy storage charging and discharging constraints: In order to ensure the safety and life of energy storage, the state of charge constraint and energy storage power constraint need to be met. The specific constraint conditions are shown in the following formula: ； ； ； In the formula, is the charging power; is the discharge power; is the largest; To ensure that it can be charged or discharged at any time, ; The output of energy storage is taken as the optimization object, and the roulette method is used to select individuals. The constraint conditions are modified in the form of penalty functions to modify the fitness values ​​of individuals, and then the best solutions are retained and the worst solutions are eliminated.

Images