Distributed energy self-absorption optimization method and control system based on optical storage direct-flexible system

Abstract

The invention relates to the technical field of energy scheduling, in particular to a distributed energy self-consumption optimization method and control system based on optical storage direct flexible, and the method comprises the steps: obtaining a historical operation data set, screening the historical operation data set, obtaining an effective operation data set, carrying out the training through the effective operation data set, and obtaining a target power prediction model set; and according to the target power prediction model group, obtaining predicted power generation power, judging whether the predicted power generation power is greater than real-time load power, if so, performing energy storage on energy storage equipment based on the predicted power generation power and the real-time load power, and if not, performing power calling on an external power grid and the energy storage equipment respectively. According to the method, the accuracy of photovoltaic power generation prediction and the safety of the optical storage direct flexible system can be improved, and the flexibility of energy scheduling in the energy self-absorption process is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy scheduling, in particular to a distributed energy self-consumption optimization method and control system based on light storage direct flexibility. BACKGROUND

[0002] With the large-scale access of distributed energy (such as photovoltaic power generation), how to realize efficient utilization and self-consumption of energy in the local area becomes the key. Energy self-consumption can effectively reduce the dependence on traditional power grids, improve energy utilization efficiency, reduce energy transmission losses, and has important significance for improving the stability and economy of energy control systems. At the same time, it also helps to cope with the intermittency and instability of new energy generation, promotes the large-scale application of renewable energy, and promotes the optimization and upgrading of energy structure, providing strong support for achieving the carbon peak and carbon neutral goals.

[0003] Traditional technologies usually use simple rules and strategies for energy distribution and scheduling to realize energy self-consumption. However, traditional technologies are difficult to accurately predict photovoltaic power generation, resulting in inaccurate energy scheduling. Moreover, traditional technologies lack flexibility and adaptability when facing complex environmental factors and dynamic energy demand, and cannot effectively regulate energy storage devices and flexible electricity devices. SUMMARY

[0004] The present application provides a distributed energy self-consumption optimization method and control system based on light storage direct flexibility, which mainly aims to improve the accuracy of photovoltaic power generation prediction and the safety of light storage direct flexibility system, and improve the flexibility of energy scheduling in the process of energy self-consumption.

[0005] To achieve the above purpose, the present application provides a distributed energy self-consumption optimization method based on light storage direct flexibility, which comprises:

[0006] Receiving a self-consumption optimization instruction, determining a to-be-optimized unit based on the self-consumption optimization instruction, wherein the to-be-optimized unit includes photovoltaic power generation equipment, energy storage equipment and flexible electricity equipment;

[0007] Obtaining a historical operation data set of the photovoltaic power generation equipment in the to-be-optimized unit, wherein the historical operation data set includes a plurality of historical operation data, and the historical operation data includes a historical power generation power and a historical environmental parameter group;

[0008] Performing environmental power correlation screening on the historical operation data set to obtain an effective operation data set;

[0009] Performing environmental climate labeling on the effective operation data set to obtain a training operation data set, and training a pre-obtained neural network model group using the training operation data set to obtain a target power prediction model group, wherein the training operation data in the training operation data set includes a target climate feature parameter group;

[0010] monitoring the environment of the unit to be optimized to obtain a real-time climate parameter group, predicting power generation of the photovoltaic power generation equipment according to the real-time climate parameter group and a target power prediction model group to obtain predicted power generation;

[0011] detecting real-time power of the flexible power consumption equipment to obtain real-time load power, and judging whether the predicted power generation is greater than the real-time load power;

[0012] if the predicted power generation is greater than the real-time load power, calculating real-time energy storage power based on the predicted power generation and the real-time load power, and storing energy in the energy storage equipment according to the real-time energy storage power;

[0013] if the predicted power generation is not greater than the real-time load power, obtaining real-time power grid load of a pre-constructed external power grid, and calling power of the external power grid and the energy storage equipment respectively based on the real-time power grid load.

[0014] Optionally, the historical operation data set is subjected to environment power correlation screening to obtain an effective operation data set, which comprises:

[0015] determining a photovoltaic operation environment of the photovoltaic power generation equipment, and setting a photovoltaic environment index group based on the photovoltaic operation environment;

[0016] extracting a historical power generation power set from the historical operation data set, and calculating a mean value of the historical power generation power set to obtain an average power generation power;

[0017] calculating correlation of the photovoltaic environment index group according to the historical power generation power set and the average power generation power to obtain an environment correlation value group, wherein the environment correlation value in the environment correlation value group corresponds to the photovoltaic environment index in the photovoltaic environment index group one by one;

[0018] screening the environment correlation value group based on a preset environment correlation threshold to obtain a high correlation value group, wherein the high correlation value group comprises a plurality of high correlation values, and the high correlation value is not less than the environment correlation threshold;

[0019] determining a correlation environment index group corresponding to the high correlation value group in the photovoltaic environment index group;

[0020] cleaning the historical operation data set based on the correlation environment index group to obtain an effective operation data set, wherein the effective operation data set comprises a plurality of effective operation data, and the effective operation data comprises an effective power generation power and an effective environment parameter group.

[0021] Optionally, the calculating correlation of the photovoltaic environment index group according to the historical power generation power set and the average power generation power to obtain an environment correlation value group comprises:

[0022] Extract the photovoltaic environmental indicators sequentially from the photovoltaic environmental indicator group;

[0023] The photovoltaic environmental parameter set is identified in the historical operation dataset based on the photovoltaic environmental indicators, and the mean of the photovoltaic environmental parameter set is calculated to obtain the average environmental parameters. Each photovoltaic environmental parameter in the photovoltaic environmental parameter set represents the same photovoltaic environmental indicator.

[0024] Based on average environmental parameters, average power generation, photovoltaic environmental parameter set, and historical power generation set, a power correlation assessment is performed on the photovoltaic environmental indicators to obtain environmental correlation values, which are expressed as follows:

[0025] ,

[0026] in, Indicates the environmental correlation value. The first one representing the historical concentration of power generation Historical power generation capacity, Indicates average power generation. Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, Represents average environmental parameters. The first one representing the historical concentration of power generation Historical power generation capacity, Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, This indicates the number of photovoltaic environmental parameters in the photovoltaic environmental parameter set or the number of historical power generation parameters in the historical power generation set.

[0027] The environmental correlation values ​​are summarized to obtain the environmental correlation value group.

[0028] Optionally, the step of labeling the effective running dataset with environmental climate data to obtain the training running dataset includes:

[0029] Multiple initial clustering distributions are set, including: distributions for severely cold regions, cold regions, hot summer and cold winter regions, or hot summer and warm winter regions.

[0030] Climate features are constructed for each of the multiple initial cluster distributions to obtain multiple sets of climate feature parameters, where one initial cluster distribution corresponds to one set of climate feature parameters.

[0031] Each initial cluster distribution among multiple initial cluster distributions is initialized to obtain multiple initial cluster parameter sets. The initial cluster parameter sets among the multiple initial cluster parameter sets correspond one-to-one with the initial cluster distributions among the multiple initial cluster distributions. The initial cluster distributions include: initial mean, initial covariance, and initial mixing weights.

[0032] Based on multiple initial clustering parameter sets and valid running datasets, multiple initial clustering distributions are updated to obtain multiple updated clustering distributions;

[0033] Parameters are calculated for each of the multiple updated clustering distributions to obtain multiple sets of updated clustering parameters;

[0034] The parameter change values ​​are calculated based on multiple updated clustering parameter sets and multiple initial clustering parameter sets;

[0035] Multiple updated clustering parameter sets and multiple updated clustering distributions are respectively used as multiple initial clustering parameter sets and multiple initial clustering distributions, and the step of updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets is returned until the parameter change value is not greater than the preset convergence change value.

[0036] When the parameter change value is not greater than the convergence change value, the multiple updated clustering distributions are recorded as multiple target clustering distributions, and climate labeling is performed based on the multiple target clustering distributions to obtain the training running dataset.

[0037] Optionally, updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets to obtain multiple updated clustering distributions includes:

[0038] Extract valid runtime data sequentially from the valid runtime dataset, and construct a valid runtime vector based on the valid runtime data;

[0039] Based on multiple initial clustering parameter sets, the posterior probability between the effective running vector and each initial clustering distribution in the multiple initial clustering distributions is calculated to obtain multiple posterior probabilities;

[0040] The multiple posterior probabilities corresponding to each valid running data are summarized to obtain a posterior probability set. The posterior probability set corresponds one-to-one with the valid running data in the valid running dataset, and the posterior probability set includes multiple posterior probabilities.

[0041] Based on the posterior probability set, the effective running dataset is distributed to the multiple initial clustering distributions to obtain multiple updated clustering distributions.

[0042] Optionally, the climate labeling based on multiple target clustering distributions to obtain the training dataset includes:

[0043] Target clustering distributions are extracted sequentially from multiple target clustering distributions, and target climate feature parameter groups corresponding to the target clustering distributions are determined from multiple climate feature parameter groups.

[0044] Identify the target running data group contained in the target cluster distribution;

[0045] Each target operational data point in the target operational data set is labeled using the target climate characteristic parameter set to obtain the training operational data set;

[0046] The training dataset is obtained by merging the training data sets corresponding to each of the multiple target clustering distributions.

[0047] Optionally, the step of predicting the power generation of photovoltaic power generation equipment based on real-time climate parameter sets and target power prediction model sets to obtain the predicted power generation includes:

[0048] The target power prediction model is extracted sequentially from the target power prediction model group, and the real-time climate parameter group is input into the target power prediction model to obtain the initial power generation.

[0049] Record the average loss value of the target power prediction model during the training step;

[0050] By summing up the average loss values ​​and the initial power generation, we obtain the average loss value group and the initial power generation group;

[0051] The model weights are allocated based on the average loss value group to obtain the model prediction weight reorganization, where the model prediction weights in the model prediction weight reorganization correspond one-to-one with the target power prediction models in the target power prediction model group.

[0052] The initial power generation group is weighted and averaged according to the model prediction weight reorganization to obtain the predicted power generation.

[0053] Optionally, the step of allocating model weights based on the average loss value set to obtain a reorganized model prediction weight includes:

[0054] The average loss values ​​are extracted sequentially from the average loss value group, and the model prediction weights are calculated based on the average loss values. The model prediction weights are expressed as follows:

[0055] ,

[0056] in, Indicates the model's predicted weights. This represents the average loss value. This indicates the number of target power prediction models in the target power prediction model group. Represents the first in the average loss value group Average loss value;

[0057] The model prediction weights are aggregated to obtain the model prediction weight reorganization.

[0058] Optionally, the step of obtaining the real-time grid load of the pre-constructed external grid, and based on the real-time grid load, performing power dispatching to the external grid and energy storage devices respectively, includes:

[0059] Calculate the real-time power deficit between the predicted power generation and the real-time load power;

[0060] Obtain the maximum load of the external power grid;

[0061] If the real-time grid load is less than the maximum dispatch load, the external grid dispatch power and energy storage dispatch power are calculated based on the maximum dispatch load, the real-time grid load, and the real-time power deficit.

[0062] Power is dispatched to the external power grid and energy storage devices respectively based on the power dispatched from the external power grid and the power dispatched from the energy storage device.

[0063] If the real-time grid load is not less than the maximum dispatch load, the maximum discharge power is calculated based on the preset maximum state of charge difference, the preset sampling interval, and the preset maximum energy storage capacity.

[0064] Determine whether the maximum discharge power is greater than the real-time deficit power;

[0065] If the maximum discharge power is greater than the real-time deficit power, then the energy storage device is discharged based on the real-time deficit power to complete the self-consumption optimization of distributed energy based on photovoltaic-storage-direct-flexible energy storage.

[0066] If the maximum discharge power is not greater than the real-time deficit power, then calculate the load regulation power between the maximum discharge power and the real-time deficit power.

[0067] The energy storage device is discharged according to the maximum discharge power, and the flexible power device is reduced in power based on the load adjustment power.

[0068] To achieve the above objectives, the present invention also provides a distributed energy self-consumption optimization control system based on photovoltaic-storage-direct-flexible architecture, comprising:

[0069] The historical data acquisition module is used to receive the consumption optimization instruction, determine the unit to be optimized based on the consumption optimization instruction, wherein the unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment, and acquire the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized, wherein the historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups;

[0070] The prediction model training module is used to filter the environmental power correlation of the historical running dataset to obtain the effective running dataset, label the effective running dataset with environmental climate to obtain the training running dataset, and use the training running dataset to train the pre-acquired neural network model group to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate feature parameter group.

[0071] The real-time data acquisition module is used to monitor the environment of the unit to be optimized, obtain the real-time climate parameter set, predict the power generation of the photovoltaic power generation equipment based on the real-time climate parameter set and the target power prediction model set, obtain the predicted power generation, detect the real-time power of the flexible power consumption equipment, obtain the real-time load power, and determine whether the predicted power generation is greater than the real-time load power.

[0072] The real-time power regulation module is used to calculate the real-time energy storage power based on the predicted power generation and the real-time load power, and to store energy in the energy storage device according to the real-time energy storage power. If the predicted power generation is not greater than the real-time load power, the module obtains the real-time grid load of the pre-built external grid and performs power dispatching to the external grid and the energy storage device respectively based on the real-time grid load.

[0073] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0074] Memory, storing at least one instruction; and

[0075] The processor executes the instructions stored in the memory to implement the above-described optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy.

[0076] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the aforementioned optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-flexible energy transmission.

[0077] To address the problems described in the background art, this invention first obtains the historical operating dataset of the photovoltaic power generation equipment in the unit to be optimized. This step collects historical operating data of the photovoltaic power generation equipment, providing a rich data foundation for subsequent data analysis and model training, and helping to gain a deeper understanding of the operating characteristics of the photovoltaic power generation equipment under different environmental conditions. Next, the historical operating dataset is filtered for environmental power correlation to obtain a valid operating dataset. This step effectively removes noise and irrelevant factors from the data by filtering out environmental parameters that are highly correlated with power generation, thereby improving data quality and making the model trained based on this data more accurate and reliable. This enhances the accuracy of subsequent power allocation, better reflects the real operating patterns of the photovoltaic power generation equipment, and improves the model's predictive accuracy. Then, the valid operating dataset is labeled with environmental climate data to obtain a training operating dataset. This step incorporates climate factors into the data labeling and model training process, enabling the trained power prediction model to... This invention fully considers the impact of different climate characteristics on photovoltaic power generation, improving the model's generalization ability and prediction accuracy in practical applications. Finally, it compares the predicted power generation with the real-time load power and performs different operations. If the predicted power generation is greater than the real-time load power, the energy storage device is used to store excess energy, achieving energy self-consumption, avoiding energy waste, and improving energy utilization efficiency. If the predicted power generation is not greater than the real-time load power, power is allocated to the external grid and the energy storage device based on the real-time grid load. This step effectively compensates for the gap between power generation and load power when power generation is insufficient by rationally allocating power from the external grid, the energy storage device, and adjusting the power of flexible power-consuming equipment. This ensures the normal operation of power-consuming equipment, avoids over-discharge of the energy storage device, extends its lifespan, and improves the safety of the photovoltaic-storage-DC-flexible system. Therefore, this invention can improve the accuracy of photovoltaic power generation prediction and the safety of the photovoltaic-storage-DC-flexible system, enhancing the flexibility of energy dispatch during energy self-consumption. Attached Figure Description

[0078] Figure 1 This is a flowchart illustrating an embodiment of the self-consumption optimization method for distributed energy based on photovoltaic-storage-direct-flexible energy transfer.

[0079] Figure 2 A functional block diagram of a distributed energy self-consumption optimization control system based on photovoltaic-storage-direct-flexible transmission and energy storage provided in an embodiment of the present invention;

[0080] Figure 3 This is a schematic diagram of the structure of an electronic device that implements the distributed energy self-consumption optimization method based on photovoltaic-storage-direct-flexible transmission, as provided in an embodiment of the present invention.

[0081] Explanation of reference numerals in the attached figures:

[0082] 10. Electronic device; 11. Processor; 12. Memory; 13. Bus.

[0083] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0084] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0085] This application provides a distributed energy self-consumption optimization method based on photovoltaic-storage-DC-flexible architecture. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the distributed energy self-consumption optimization method based on photovoltaic-storage-DC-flexible architecture can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0086] Reference Figure 1 The diagram shown is a flowchart illustrating a distributed energy self-consumption optimization method based on photovoltaic-storage-DC-flexible energy transfer, according to an embodiment of the present invention. In this embodiment, the distributed energy self-consumption optimization method based on photovoltaic-storage-DC-flexible energy transfer includes:

[0087] S1. Receive the power consumption optimization instruction and determine the unit to be optimized based on the power consumption optimization instruction. The unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment.

[0088] Understandably, the aforementioned absorption optimization command refers to a manually initiated command for a specific photovoltaic-storage-DC-flexible control system to absorb energy. This command is received by intelligent monitoring equipment, which is the control center that controls the power of the unit to be optimized. The unit to be optimized refers to the specific photovoltaic-storage-DC-flexible control system specified in the absorption optimization command. This unit includes: photovoltaic power generation equipment, energy storage equipment, and flexible power consumption equipment. Photovoltaic power generation equipment refers to equipment that converts solar energy into electrical energy, such as rooftop photovoltaic panels or photovoltaic power stations. Energy storage equipment refers to equipment that stores electrical energy and releases it when needed, such as lithium battery packs or supercapacitors. Flexible power consumption equipment refers to power terminals with dynamically adjustable power, such as electric vehicle charging piles. When the power output of the photovoltaic power generation equipment cannot meet the load power of the electric vehicle charging piles, the number of electric vehicle charging piles can be reduced, thereby reducing the load power of the flexible power consumption equipment and achieving self-absorption. In addition to photovoltaic power generation equipment, energy storage equipment, and flexible power consumption equipment, the aforementioned units to be optimized also include: a DC power supply control system. This DC power supply control system refers to a control system that directly transmits and distributes electrical energy in the form of DC power. Traditional AC power supply control systems may experience certain losses during energy conversion, while using DC power supply can improve energy transmission and utilization efficiency in some scenarios.

[0089] S2. Obtain the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized. The historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups.

[0090] It is clear that the historical operating data refers to the operating data of photovoltaic power generation equipment detected in previous periods. This operating data includes: historical power generation and historical environmental parameter sets. The historical power generation refers to the power generation of the photovoltaic power generation equipment detected in previous periods, and the historical environmental parameter sets refer to the parameters of natural environmental factors detected during the operation of the photovoltaic power generation equipment in previous periods. These natural environmental factors include, for example, ambient temperature, ambient humidity, solar radiation intensity, and solar incidence angle. It should be noted that the historical power generation and historical environmental parameter sets in the same historical operating data are collected at the same time.

[0091] Furthermore, the aforementioned historical operation dataset is recorded by IoT devices, and relevant operators can retrieve the historical operation dataset from the IoT devices.

[0092] S3. Perform environmental power correlation filtering on the historical operating dataset to obtain the effective operating dataset.

[0093] It should be explained that the effective running dataset refers to the historical running dataset after correlation filtering. Since not all historical environmental parameters in the historical environmental parameter group are highly correlated with the power generation of photovoltaic power generation equipment, historical environmental parameters with low correlation to power generation will affect the accuracy of subsequent network training. Therefore, it is necessary to remove historical environmental parameters with low correlation (i.e., environmental power correlation filtering). Low correlation means that the subsequent environmental correlation value is greater than the environmental correlation threshold.

[0094] In detail, the process of filtering historical operational datasets based on environmental power correlations to obtain valid operational datasets includes:

[0095] Determine the photovoltaic operating environment of the photovoltaic power generation equipment, and set up a photovoltaic environment index group based on the photovoltaic operating environment;

[0096] Historical power generation sets are extracted from historical operation datasets, and the average power generation is calculated by averaging the historical power generation sets.

[0097] The correlation calculation of the photovoltaic environmental index group is performed based on the historical power generation set and the average power generation to obtain the environmental correlation value group, wherein the environmental correlation value in the environmental correlation value group corresponds one-to-one with the photovoltaic environmental index in the photovoltaic environmental index group.

[0098] The environmental association value group is filtered based on the preset environmental association threshold to obtain the highly associated value group. The highly associated value group includes multiple highly associated values, and the highly associated values ​​are not less than the environmental association threshold.

[0099] Determine the associated environmental indicator group corresponding to the highly correlated value group in the photovoltaic environmental indicator group;

[0100] Data cleaning is performed on the historical operating dataset based on the associated environmental indicator group to obtain the effective operating dataset. The effective operating dataset includes multiple effective operating data, including: effective power generation and effective environmental parameter group.

[0101] It is clear that the photovoltaic operating environment refers to the environment in which the photovoltaic power generation equipment operates, such as an outdoor open-air site or building rooftop. The photovoltaic environment index group includes multiple photovoltaic environment indicators, which are indicators that reflect the impact on the photovoltaic power generation equipment within the photovoltaic operating environment. These photovoltaic environment indicators are set manually, such as ambient temperature, ambient humidity, and weather labels. The historical power generation set refers to the collection of all historical power generation data in the historical operating dataset. The weather label refers to a numerical label that represents the target cluster distribution. This weather label includes sunny, cloudy, and rainy labels. In addition, the weather label may also include thunderstorms, rain, snow, etc.

[0102] Furthermore, the average power generation refers to the average of all historical power generation in the historical power generation set. The environmental correlation value group includes multiple environmental correlation values, and the environmental correlation value refers to a numerical value that quantifies the degree of correlation between photovoltaic environmental indicators and photovoltaic power generation. The larger the environmental correlation value, the higher the degree of correlation between photovoltaic environmental indicators and photovoltaic power generation. The environmental correlation threshold refers to a human-set constant. When the environmental correlation value is greater than the environmental correlation threshold (i.e., a highly correlated value), it indicates that the photovoltaic environmental indicator corresponding to the environmental correlation value has a high correlation with photovoltaic power generation. The correlated environmental indicator group refers to the combination of multiple photovoltaic environmental indicators corresponding to multiple highly correlated values ​​in the highly correlated value group. The data cleaning refers to: sequentially extracting historical operating data from the historical operating dataset, identifying the historical environmental parameter group in the historical operating data, determining the correlated environmental parameter group in the historical environmental parameter group based on the correlated environmental indicator group, where the correlated environmental parameter group refers to the combination of multiple historical environmental parameters corresponding to multiple correlated environmental indicators in the correlated environmental indicator group, and recording the correlated environmental parameter group as the valid environmental parameter group. The above operation is performed on all historical operating data to obtain a valid operating dataset. The effective power generation refers to the historical power generation in the effective operating data, and the effective environmental parameter set refers to the historical environmental parameter set in the effective operating data after cleaning.

[0103] In detail, the step of performing correlation calculations on the photovoltaic environmental indicator group based on historical power generation sets and average power generation to obtain an environmental correlation value group includes:

[0104] Extract the photovoltaic environmental indicators sequentially from the photovoltaic environmental indicator group;

[0105] The photovoltaic environmental parameter set is identified in the historical operation dataset based on the photovoltaic environmental indicators, and the mean of the photovoltaic environmental parameter set is calculated to obtain the average environmental parameters. Each photovoltaic environmental parameter in the photovoltaic environmental parameter set represents the same photovoltaic environmental indicator.

[0106] Based on average environmental parameters, average power generation, photovoltaic environmental parameter set, and historical power generation set, a power correlation assessment is performed on the photovoltaic environmental indicators to obtain environmental correlation values, which are expressed as follows:

[0107] ,

[0108] in, Indicates the environmental correlation value. The first one representing the historical concentration of power generation Historical power generation capacity, Indicates average power generation. Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, Represents average environmental parameters. The first one representing the historical concentration of power generation Historical power generation capacity, Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, This indicates the number of photovoltaic environmental parameters in the photovoltaic environmental parameter set or the number of historical power generation parameters in the historical power generation set.

[0109] The environmental correlation values ​​are summarized to obtain the environmental correlation value group.

[0110] It is clear that the photovoltaic environmental parameter set refers to the collection of all historical environmental parameters corresponding to photovoltaic environmental indicators in the historical operation dataset. The average environmental parameter refers to the average value of all photovoltaic environmental parameters in the photovoltaic environmental parameter set.

[0111] S4. Environmental climate labeling is performed on the effective running dataset to obtain the training running dataset. The pre-acquired neural network model group is trained using the training running dataset to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate characteristic parameter group.

[0112] Understandably, the training dataset refers to the valid operating dataset after climate labeling. Since the power output of photovoltaic power generation equipment varies under different climatic conditions, it is necessary to climate label each valid operating data point in the valid operating dataset to make subsequent training steps more accurate. The neural network model group includes multiple neural network models, such as Convolutional Neural Networks (CNN) and Multilayer Perceptrons (MLP). The target power prediction model group includes multiple target power prediction models, which refer to the trained neural network models, and there is a one-to-one correspondence between the target power prediction models and the neural network models.

[0113] Furthermore, in practical applications, the target power prediction model group can be connected to a large language model to obtain an AI model. The large language model in this AI model can generate natural language based on the output of the target power prediction model group and display the generated results to relevant operators.

[0114] Specifically, the step of performing environmental and climate labeling on the effective running dataset to obtain the training running dataset includes:

[0115] Multiple initial clustering distributions are set, including: distributions for severely cold regions, cold regions, hot summer and cold winter regions, or hot summer and warm winter regions.

[0116] Climate features are constructed for each of the multiple initial cluster distributions to obtain multiple sets of climate feature parameters, where one initial cluster distribution corresponds to one set of climate feature parameters.

[0117] Each initial cluster distribution among multiple initial cluster distributions is initialized to obtain multiple initial cluster parameter sets. The initial cluster parameter sets among the multiple initial cluster parameter sets correspond one-to-one with the initial cluster distributions among the multiple initial cluster distributions. The initial cluster distributions include: initial mean, initial covariance, and initial mixing weights.

[0118] Based on multiple initial clustering parameter sets and valid running datasets, multiple initial clustering distributions are updated to obtain multiple updated clustering distributions;

[0119] Parameters are calculated for each of the multiple updated clustering distributions to obtain multiple sets of updated clustering parameters;

[0120] The parameter change values ​​are calculated based on multiple updated clustering parameter sets and multiple initial clustering parameter sets;

[0121] Multiple updated clustering parameter sets and multiple updated clustering distributions are respectively used as multiple initial clustering parameter sets and multiple initial clustering distributions, and the step of updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets is returned until the parameter change value is not greater than the preset convergence change value.

[0122] When the parameter change value is not greater than the convergence change value, the multiple updated clustering distributions are recorded as multiple target clustering distributions, and climate labeling is performed based on the multiple target clustering distributions to obtain the training running dataset.

[0123] It should be explained that the initial clustering distribution refers to the pre-defined climate category distribution in the Gaussian Mixture Model (GMM), where the distributions of severe cold regions, cold regions, hot summer and cold winter regions, or hot summer and warm winter regions correspond to the severe cold climate category, cold climate category, hot summer and cold winter climate category, and hot summer and warm winter climate category, respectively. The construction of multiple climate feature parameter sets for each of the multiple initial clustering distributions refers to the fact that different initial clustering distributions represent different climate categories with their own climate characteristics, such as solar radiation intensity. By constructing climate feature parameter sets, the data characteristics of the corresponding initial clustering distributions can be quantified, thereby improving the accuracy of subsequent neural network training. Furthermore, adding climate features to the neural network training data can further improve the accuracy of predicting the power generation of photovoltaic power generation equipment under different climate categories.

[0124] Furthermore, the initialization refers to setting the initial clustering parameter set for each initial cluster distribution, and the initialization method is either initialization using a random function or initialization using the K-means++ algorithm. The initial clustering parameter set includes: initial mean, initial covariance, and initial mixture weights. The initial mean refers to the centroid vector of the initial cluster distribution, the initial covariance refers to the data dispersion matrix of the initial cluster distribution, and the initial mixture weights refer to the proportion of that cluster distribution in the overall model. The detailed definitions and calculation processes of the above-mentioned mean, covariance, and mixture weights are all existing and common techniques, and will not be elaborated here.

[0125] Furthermore, the multiple updated clustering distributions refer to the multiple initial clustering distributions after updates. The multiple updated clustering parameter sets refer to the multiple initial clustering parameter sets corresponding to the multiple updated clustering distributions. The formula for calculating the updated clustering parameter sets is a general technique and will not be elaborated here. The parameter change value refers to a numerical value quantifying the similarity between the multiple updated clustering parameter sets and the multiple initial clustering parameter sets. The larger the parameter change value, the higher the similarity between the multiple updated clustering parameter sets and the multiple initial clustering parameter sets. The parameter change value is calculated as follows: an updated parameter vector is constructed based on the multiple updated clustering parameter sets, where the vector parameters in the updated parameter vector are composed of the parameters in the multiple updated clustering parameter sets. An initial parameter vector is constructed based on the multiple initial clustering parameter sets, where the vector parameters in the initial parameter vector are composed of the parameters in the multiple initial clustering parameter sets. Then, the vector cosine value between the updated parameter vector and the initial parameter vector is calculated, and this vector cosine value is used as the parameter change value. The convergence change value refers to a manually set constant about the parameter change value. When the parameter change value is not greater than this convergence change value, it indicates that the clustering has reached convergence. The multiple target clustering distributions refer to the multiple updated clustering distributions when the clustering reaches convergence.

[0126] In detail, the step of updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets to obtain multiple updated clustering distributions includes:

[0127] Extract valid runtime data sequentially from the valid runtime dataset, and construct a valid runtime vector based on the valid runtime data;

[0128] Based on multiple initial clustering parameter sets, the posterior probability between the effective running vector and each initial clustering distribution in the multiple initial clustering distributions is calculated to obtain multiple posterior probabilities;

[0129] The multiple posterior probabilities corresponding to each valid running data are summarized to obtain a posterior probability set. The posterior probability set corresponds one-to-one with the valid running data in the valid running dataset, and the posterior probability set includes multiple posterior probabilities.

[0130] Based on the posterior probability set, the effective running dataset is distributed to the multiple initial clustering distributions to obtain multiple updated clustering distributions.

[0131] It is clear that the effective operating vector refers to a vector composed of effective operating data. For example, if the effective operating data is: effective power generation A, effective environmental parameter set (B1, B1, B1), then the effective operating vector composed of this effective operating data is ( The posterior probability refers to the probability that valid running data belongs to a specific cluster (initial cluster distribution), and the formula for calculating the posterior probability is:

[0132] ,

[0133] in, Represents the posterior probability. This represents the initial mixing weights in the initial clustering parameter set corresponding to the posterior probability. This represents the Gaussian probability density function. This represents the effective running vector corresponding to the posterior probability. This represents the initial mean of the initial clustering parameter set corresponding to the posterior probability. This represents the initial covariance in the initial clustering parameter set corresponding to the posterior probability. This indicates the number of initial clustering parameter groups in a set of multiple initial clustering parameter groups. Represents the first clustering parameter in multiple initial clustering parameter sets. Initial mixing weights in an initial clustering parameter set, Represents the first clustering parameter in multiple initial clustering parameter sets. The initial mean of each initial cluster parameter set. Represents the first clustering parameter in multiple initial clustering parameter sets. The initial covariance in the initial clustering parameter set.

[0134] Furthermore, the step of allocating the effective running dataset to the multiple initial clustering distributions based on the posterior probability set refers to: sequentially extracting posterior probability sets from the posterior probability set, identifying the effective running data corresponding to the posterior probability set, identifying the maximum posterior probability in the posterior probability set, determining the initial clustering distribution corresponding to the maximum posterior probability, and then placing the effective running data corresponding to the posterior probability set into the initial clustering distribution corresponding to the maximum posterior probability. This process is repeated until all posterior probability sets in the posterior probability set have been extracted. At this point, all effective running data in the effective running dataset has been placed, and the multiple initial clustering distributions after all effective running data has been placed are recorded as multiple updated clustering distributions.

[0135] Specifically, the climate labeling based on multiple target clustering distributions yields a training dataset, including:

[0136] Target clustering distributions are extracted sequentially from multiple target clustering distributions, and target climate feature parameter groups corresponding to the target clustering distributions are determined from multiple climate feature parameter groups.

[0137] Identify the target running data group contained in the target cluster distribution;

[0138] Each target operational data point in the target operational data set is labeled using the target climate characteristic parameter set to obtain the training operational data set;

[0139] The training dataset is obtained by merging the training data sets corresponding to each of the multiple target clustering distributions.

[0140] Understandably, the target climate characteristic parameter set refers to the climate characteristic parameter set corresponding to the target clustering distribution. The labeling refers to placing the climate characteristic parameter set into each target operational data in the target operational data set.

[0141] S5. Perform environmental monitoring on the unit to be optimized to obtain a set of real-time climate parameters. Based on the set of real-time climate parameters and the target power prediction model set, predict the power generation of the photovoltaic power generation equipment to obtain the predicted power generation.

[0142] It is clear that the real-time climate parameter set refers to the environmental parameter set and climate characteristic parameter set of the photovoltaic power generation equipment of the unit to be optimized at the current moment. Among them, the environmental parameters in the environmental parameter set of the real-time climate parameter set correspond one-to-one with the associated environmental indicators in the associated environmental indicator set. The predicted power generation refers to the power generation of the photovoltaic power generation equipment after a certain period of time after prediction. The certain period of time is the time interval between two adjacent environmental monitoring sessions, that is, the subsequent sampling interval.

[0143] In detail, the step of predicting the power generation of photovoltaic power generation equipment based on real-time climate parameter sets and target power prediction model sets to obtain the predicted power generation includes:

[0144] The target power prediction model is extracted sequentially from the target power prediction model group, and the real-time climate parameter group is input into the target power prediction model to obtain the initial power generation.

[0145] Record the average loss value of the target power prediction model during the training step;

[0146] By summing up the average loss values ​​and the initial power generation, we obtain the average loss value group and the initial power generation group;

[0147] The model weights are allocated based on the average loss value group to obtain the model prediction weight reorganization, where the model prediction weights in the model prediction weight reorganization correspond one-to-one with the target power prediction models in the target power prediction model group.

[0148] The initial power generation group is weighted and averaged according to the model prediction weight reorganization to obtain the predicted power generation.

[0149] Understandably, the initial power generation refers to the output value of the target power prediction model. The average loss value refers to the average of various loss values ​​of the target power prediction model during training. The loss value is an indicator used to measure the difference between the model's prediction result and the actual target value during neural network model training. It reflects the model's current performance; the smaller the loss value, the closer the model's prediction result is to the real data, and the better the model's performance. For example, mean squared error (MSE) is one of the commonly used loss functions, and the loss value can be calculated using this function. The initial power generation group refers to a combination of multiple initial power generation values, and the average loss value group refers to a combination of multiple average loss values. The model prediction weighting group includes a combination of multiple model prediction weights, and the model prediction weight refers to the weight of the initial power generation of the target power prediction model in subsequent calculations of the predicted power generation. The weighted average is calculated using the following formula:

[0150] ,

[0151] in, Indicates the predicted power generation capacity. This indicates the number of target power prediction models in the target power prediction model group. In the model, the first prediction weight reorganization is represented by the first... Each model predicts weights. Represents the first in the initial power generation group Initial power generation.

[0152] In detail, the process of allocating model weights based on the average loss value set to obtain a reorganized model prediction weight includes:

[0153] The average loss values ​​are extracted sequentially from the average loss value group, and the model prediction weights are calculated based on the average loss values. The model prediction weights are expressed as follows:

[0154] ,

[0155] in, Indicates the model's predicted weights. This represents the average loss value. This indicates the number of target power prediction models in the target power prediction model group. Represents the first in the average loss value group Average loss value;

[0156] The model prediction weights are aggregated to obtain the model prediction weight reorganization.

[0157] S6. Perform real-time power detection on flexible electrical equipment to obtain real-time load power and determine whether the predicted power generation is greater than the real-time load power.

[0158] Understandably, the real-time load power refers to the power consumption of flexible electrical equipment at the current moment. Subsequent power regulation can be performed by comparing the predicted power generation with the real-time load power.

[0159] S7. If the predicted power generation is greater than the real-time load power, the real-time energy storage power is calculated based on the predicted power generation and the real-time load power, and the energy storage device is used to store energy according to the real-time energy storage power.

[0160] It is clear that when the predicted power generation is greater than the real-time load power, it means that in the subsequent period of time (subsequent sampling intervals), the power generation of the photovoltaic power generation equipment alone can meet the power demand of the flexible power consumption equipment. In addition to meeting the power demand, the photovoltaic power generation equipment can also store energy for the energy storage equipment, thereby completing the self-consumption of energy. The charging power for storing energy for the energy storage equipment is: predicted power generation minus real-time load power, that is, real-time energy storage power is the energy storage power for the energy storage equipment.

[0161] Furthermore, the photovoltaic power generation equipment charges the energy storage device according to the real-time energy storage power until the energy storage device reaches its maximum power. Once the energy storage device reaches its maximum power, the photovoltaic power generation equipment can store the energy to the external power grid according to the real-time energy storage power. The external power grid refers to the public power transmission network.

[0162] S8. If the predicted power generation is not greater than the real-time load power, then obtain the real-time grid load of the pre-built external grid, and based on the real-time grid load, make power calls to the external grid and energy storage devices respectively.

[0163] Understandably, when the predicted power generation is not greater than the real-time load power, it indicates that the power generation of photovoltaic power generation equipment alone cannot meet the power demand of flexible power consumption equipment within a certain period of time (subsequent sampling intervals). At this time, energy storage equipment needs to discharge or reduce the load power of flexible power consumption equipment. The maximum state of charge difference refers to the maximum change value of the state of charge (SOC) of the energy storage equipment set by the user. The state of charge refers to the percentage of the remaining electricity of the energy storage equipment relative to its maximum energy storage capacity. Since drawing electricity from the energy storage equipment will cause the state of charge of the energy storage equipment to change, if the change in the state of charge of the energy storage equipment exceeds the maximum state of charge difference, it will cause significant damage to the service life of the energy storage equipment. The maximum state of charge difference can be empirically summarized by relevant personnel through historical data, or it can be set according to the usage specifications of the energy storage equipment.

[0164] In detail, the step of obtaining the real-time grid load of the pre-constructed external grid, and based on the real-time grid load, performing power dispatching to the external grid and energy storage devices respectively, includes:

[0165] Calculate the real-time power deficit between the predicted power generation and the real-time load power;

[0166] Obtain the maximum load of the external power grid;

[0167] If the real-time grid load is less than the maximum dispatch load, the external grid dispatch power and energy storage dispatch power are calculated based on the maximum dispatch load, the real-time grid load, and the real-time power deficit.

[0168] Power is dispatched to the external power grid and energy storage devices respectively based on the power dispatched from the external power grid and the power dispatched from the energy storage device.

[0169] If the real-time grid load is not less than the maximum dispatch load, the maximum discharge power is calculated based on the preset maximum state of charge difference, the preset sampling interval, and the preset maximum energy storage capacity.

[0170] Determine whether the maximum discharge power is greater than the real-time deficit power;

[0171] If the maximum discharge power is greater than the real-time deficit power, then the energy storage device is discharged based on the real-time deficit power to complete the self-consumption optimization of distributed energy based on photovoltaic-storage-direct-flexible energy storage.

[0172] If the maximum discharge power is not greater than the real-time deficit power, then calculate the load regulation power between the maximum discharge power and the real-time deficit power.

[0173] The energy storage device is discharged according to the maximum discharge power, and the flexible power device is reduced in power based on the load adjustment power.

[0174] It is clear that the real-time power deficit refers to the power obtained by subtracting the predicted power generation from the real-time load power. The maximum load capacity refers to the maximum load power that the external power grid can withstand. When the real-time grid load is not less than this maximum load capacity, the external power grid will be unable to transmit power. This maximum load capacity can be obtained by summarizing historical power consumption data from the external power grid or set by relevant personnel. The external power grid power consumption refers to the power requested by the unit to be optimized from the external power grid, and the energy storage power consumption refers to the power released by the energy storage devices in the unit to be optimized. The external power grid power consumption and energy storage power consumption are calculated as follows: External power grid power consumption = Real-time power deficit. (Real-time grid load / maximum call load), energy storage call power = real-time deficit power - external grid call power.

[0175] It should be explained that the sampling interval refers to the time interval between two adjacent environmental monitoring sessions. The maximum energy storage capacity refers to the maximum amount of electricity that the energy storage device can store. The maximum discharge power refers to the maximum power that the energy storage device can utilize during discharge, wherein the maximum discharge power is expressed as: ,in, Indicates the maximum discharge power. Indicates the maximum energy storage capacity. Indicates the difference in maximum state of charge. Indicates the sampling interval.

[0176] Furthermore, when the predicted power generation is not greater than the real-time load power and it is impossible to draw power from the external grid (i.e., the real-time grid load is not less than the maximum draw load), the difference between the predicted power generation and the real-time load power (i.e., the real-time power deficit) can be compensated by discharging the energy storage device. However, the power of the energy storage device during discharge is constrained, i.e., it cannot exceed the maximum discharge power. When the real-time power deficit is not greater than the maximum discharge power, the real-time power deficit can be met solely by discharging the energy storage device (i.e., the energy storage device is discharged based on the real-time power deficit). When the real-time power deficit is greater than the maximum discharge power, it indicates that the real-time power deficit cannot be met solely by discharging the energy storage device. In this case, it is necessary to reduce the power consumption of flexible power consumption equipment to compensate for the difference between the real-time power deficit and the maximum discharge power. That is, the load regulation power refers to the difference between the real-time power deficit and the maximum discharge power. The power reduction operation of the flexible electrical equipment based on load regulation power refers to reducing the real-time load power of the flexible electrical equipment by the amount of reduction, which is equal to the load regulation power. The specific reduction operation is set manually. For example, if a flexible electrical equipment is an electric vehicle charging pile, the real-time load power of the electric vehicle charging pile can be reduced by shutting down some of the charging piles.

[0177] To address the problems described in the background art, this invention first obtains the historical operating dataset of the photovoltaic power generation equipment in the unit to be optimized. This step collects historical operating data of the photovoltaic power generation equipment, providing a rich data foundation for subsequent data analysis and model training, and helping to gain a deeper understanding of the operating characteristics of the photovoltaic power generation equipment under different environmental conditions. Next, the historical operating dataset is filtered for environmental power correlation to obtain a valid operating dataset. This step effectively removes noise and irrelevant factors from the data by filtering out environmental parameters that are highly correlated with power generation, thereby improving data quality and making the model trained based on this data more accurate and reliable. This enhances the accuracy of subsequent power allocation, better reflects the real operating patterns of the photovoltaic power generation equipment, and improves the model's predictive accuracy. Then, the valid operating dataset is labeled with environmental climate data to obtain a training operating dataset. This step incorporates climate factors into the data labeling and model training process, enabling the trained power prediction model to... This invention fully considers the impact of different climate characteristics on photovoltaic power generation, improving the model's generalization ability and prediction accuracy in practical applications. Finally, it compares the predicted power generation with the real-time load power and performs different operations. If the predicted power generation is greater than the real-time load power, the energy storage device is used to store excess energy, achieving energy self-consumption, avoiding energy waste, and improving energy utilization efficiency. If the predicted power generation is not greater than the real-time load power, power is allocated to the external grid and the energy storage device based on the real-time grid load. This step effectively compensates for the gap between power generation and load power when power generation is insufficient by rationally allocating power from the external grid, the energy storage device, and adjusting the power of flexible power-consuming equipment. This ensures the normal operation of power-consuming equipment, avoids over-discharge of the energy storage device, extends its lifespan, and improves the safety of the photovoltaic-storage-DC-flexible system. Therefore, this invention can improve the accuracy of photovoltaic power generation prediction and the safety of the photovoltaic-storage-DC-flexible system, enhancing the flexibility of energy dispatch during energy self-consumption.

[0178] like Figure 2 The diagram shown is a functional block diagram of a distributed energy self-consumption optimization control system based on photovoltaic-storage-direct-flexible energy transmission, provided in an embodiment of the present invention.

[0179] The distributed energy self-consumption optimization control system 100 based on photovoltaic-storage-DC-flexible power transmission and control, as described in this invention, can be installed in an electronic device. Depending on the functions implemented, the distributed energy self-consumption optimization control system 100 may include a historical data acquisition module 101, a prediction model training module 102, a real-time data acquisition module 103, and a real-time power adjustment module 104. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0180] The historical data acquisition module 101 is used to receive the consumption optimization instruction, determine the unit to be optimized based on the consumption optimization instruction, wherein the unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment, and acquire the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized, wherein the historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups.

[0181] The prediction model training module 102 is used to filter the historical running dataset for environmental power correlation to obtain a valid running dataset, to label the valid running dataset for environmental climate to obtain a training running dataset, and to train the pre-acquired neural network model group using the training running dataset to obtain a target power prediction model group. The training running data in the training running dataset includes a target climate feature parameter group.

[0182] The real-time data acquisition module 103 is used to monitor the environment of the unit to be optimized, obtain a real-time climate parameter set, predict the power generation of the photovoltaic power generation equipment based on the real-time climate parameter set and the target power prediction model set, obtain the predicted power generation, perform real-time power detection on the flexible power consumption equipment, obtain the real-time load power, and determine whether the predicted power generation is greater than the real-time load power.

[0183] The real-time power regulation module 104 is used to calculate the real-time energy storage power based on the predicted power generation and the real-time load power, and to store energy in the energy storage device according to the real-time energy storage power. If the predicted power generation is not greater than the real-time load power, the real-time grid load of the pre-constructed external grid is obtained, and power is called to the external grid and the energy storage device respectively based on the real-time grid load.

[0184] In detail, the modules in the distributed energy self-consumption optimization control system 100 based on photovoltaic-storage-direct-drive-flexible architecture described in this embodiment of the invention employ the same methods as described above during use. Figure 1 The method used is the same as the distributed energy self-consumption optimization method based on photovoltaic-storage-direct-flexible energy storage described in the article, and can produce the same technical effect, so it will not be repeated here.

[0185] like Figure 3 The diagram shown is a schematic representation of an electronic device that implements a distributed energy self-consumption optimization method based on photovoltaic-storage-direct-flexible energy transmission, according to an embodiment of the present invention.

[0186] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a program for a distributed energy self-consumption optimization method based on photovoltaic-storage-direct-flexible energy transmission.

[0187] The memory 11 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. Furthermore, the memory 11 includes both internal storage units and external storage devices of the electronic device 1. The memory 11 can be used not only to store application software and various types of data installed on the electronic device 1, such as the code of a distributed energy self-consumption optimization method program based on photovoltaic-storage-direct-flexible energy transmission, but also to temporarily store data that has been output or will be output.

[0188] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., a distributed energy self-consumption optimization method program based on photovoltaic-storage-direct-drive-flexible energy transmission), and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0189] The bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 can be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to realize the connection and communication between the memory 11 and at least one processor 10, etc.

[0190] Figure 3 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0191] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management and control system, thereby enabling functions such as charging management, discharging management, and power consumption management through the power management and control system. The power supply may also include one or more DC or AC power supplies, a recharge control system, a power fault detection circuit, a power converter or inverter, a power status indicator, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0192] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the electronic device 1 and other electronic devices.

[0193] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual user interface.

[0194] The program for the self-consumption optimization method of distributed energy based on photovoltaic-storage-direct-flexible energy storage, stored in the memory 11 of the electronic device 1, is a combination of multiple instructions. When run in the processor 10, it can achieve the following:

[0195] Receive the power consumption optimization instruction, and determine the unit to be optimized based on the power consumption optimization instruction. The unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment.

[0196] Obtain the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized. The historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups.

[0197] Environmental power correlation was performed on historical operational datasets to obtain valid operational datasets;

[0198] Environmental climate labeling is performed on the effective running dataset to obtain the training running dataset. The pre-acquired neural network model group is trained using the training running dataset to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate feature parameter group.

[0199] Environmental monitoring is performed on the unit to be optimized to obtain a set of real-time climate parameters. Based on the set of real-time climate parameters and the target power prediction model set, the power generation of the photovoltaic power generation equipment is predicted to obtain the predicted power generation.

[0200] Real-time power detection is performed on flexible electrical equipment to obtain real-time load power, and it is determined whether the predicted power generation is greater than the real-time load power.

[0201] If the predicted power generation is greater than the real-time load power, the real-time energy storage power is calculated based on the predicted power generation and the real-time load power, and the energy storage device is used to store energy according to the real-time energy storage power.

[0202] If the predicted power generation is not greater than the real-time load power, then the real-time grid load of the pre-built external grid is obtained, and power is dispatched to the external grid and energy storage devices respectively based on the real-time grid load.

[0203] Specifically, the processor 10's implementation method for the above instructions can be found in [reference needed]. Figures 1 to 3 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0204] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or control system capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0205] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0206] Receive the power consumption optimization instruction, and determine the unit to be optimized based on the power consumption optimization instruction. The unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment.

[0207] Obtain the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized. The historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups.

[0208] Environmental power correlation was performed on historical operational datasets to obtain valid operational datasets;

[0209] Environmental climate labeling is performed on the effective running dataset to obtain the training running dataset. The pre-acquired neural network model group is trained using the training running dataset to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate feature parameter group.

[0210] Environmental monitoring is performed on the unit to be optimized to obtain a set of real-time climate parameters. Based on the set of real-time climate parameters and the target power prediction model set, the power generation of the photovoltaic power generation equipment is predicted to obtain the predicted power generation.

[0211] Real-time power detection is performed on flexible electrical equipment to obtain real-time load power, and it is determined whether the predicted power generation is greater than the real-time load power.

[0212] If the predicted power generation is greater than the real-time load power, the real-time energy storage power is calculated based on the predicted power generation and the real-time load power, and the energy storage device is used to store energy according to the real-time energy storage power.

[0213] If the predicted power generation is not greater than the real-time load power, then the real-time grid load of the pre-built external grid is obtained, and power is dispatched to the external grid and energy storage devices respectively based on the real-time grid load.

[0214] In the several embodiments provided by this invention, it should be understood that the disclosed devices, control systems, and methods can be implemented in other ways. For example, the control system embodiments described above are merely illustrative, and actual implementations may have other classification methods.

[0215] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0216] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0217] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0218] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing the self-consumption of distributed energy resources based on photovoltaic-storage-direct-drive-flexible energy transmission, characterized in that, The method includes: Receive the power consumption optimization instruction, and determine the unit to be optimized based on the power consumption optimization instruction. The unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment. Obtain the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized. The historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups. Environmental power correlation was performed on historical operational datasets to obtain valid operational datasets; Environmental climate labeling is performed on the effective running dataset to obtain the training running dataset. The pre-acquired neural network model group is trained using the training running dataset to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate feature parameter group. Environmental monitoring is performed on the unit to be optimized to obtain a set of real-time climate parameters. Based on the set of real-time climate parameters and the target power prediction model set, the power generation of the photovoltaic power generation equipment is predicted to obtain the predicted power generation. Real-time power detection is performed on flexible electrical equipment to obtain real-time load power, and it is determined whether the predicted power generation is greater than the real-time load power. If the predicted power generation is greater than the real-time load power, the real-time energy storage power is calculated based on the predicted power generation and the real-time load power, and the energy storage device is used to store energy according to the real-time energy storage power. If the predicted power generation is not greater than the real-time load power, then the real-time grid load of the pre-built external grid is obtained, and power is dispatched to the external grid and energy storage devices respectively based on the real-time grid load.

2. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 1, characterized in that, The process of filtering historical operational datasets based on environmental power correlations to obtain valid operational datasets includes: Determine the photovoltaic operating environment of the photovoltaic power generation equipment, and set up a photovoltaic environment index group based on the photovoltaic operating environment; Historical power generation sets are extracted from historical operation datasets, and the average power generation is calculated by averaging the historical power generation sets. The correlation calculation of the photovoltaic environmental index group is performed based on the historical power generation set and the average power generation to obtain the environmental correlation value group, wherein the environmental correlation value in the environmental correlation value group corresponds one-to-one with the photovoltaic environmental index in the photovoltaic environmental index group. The environmental association value group is filtered based on the preset environmental association threshold to obtain the highly associated value group. The highly associated value group includes multiple highly associated values, and the highly associated values ​​are not less than the environmental association threshold. Determine the associated environmental indicator group corresponding to the highly correlated value group in the photovoltaic environmental indicator group; Data cleaning is performed on the historical operating dataset based on the associated environmental indicator group to obtain the effective operating dataset. The effective operating dataset includes multiple effective operating data, including: effective power generation and effective environmental parameter group.

3. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 2, characterized in that, The step of performing correlation calculations on the photovoltaic environmental indicator group based on historical power generation sets and average power generation to obtain an environmental correlation value group includes: Extract the photovoltaic environmental indicators sequentially from the photovoltaic environmental indicator group; The photovoltaic environmental parameter set is identified in the historical operation dataset based on the photovoltaic environmental indicators, and the mean of the photovoltaic environmental parameter set is calculated to obtain the average environmental parameters. Each photovoltaic environmental parameter in the photovoltaic environmental parameter set represents the same photovoltaic environmental indicator. Based on average environmental parameters, average power generation, photovoltaic environmental parameter set, and historical power generation set, a power correlation assessment is performed on the photovoltaic environmental indicators to obtain environmental correlation values, which are expressed as follows: ， in, Indicates the environmental correlation value. The first one representing the historical concentration of power generation Historical power generation capacity, Indicates average power generation. Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, Represents average environmental parameters. The first one representing the historical concentration of power generation Historical power generation capacity, Represents the first in the photovoltaic environmental parameter set One photovoltaic environmental parameter, This indicates the number of photovoltaic environmental parameters in the photovoltaic environmental parameter set or the number of historical power generation parameters in the historical power generation set. The environmental correlation values ​​are summarized to obtain the environmental correlation value group.

4. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 3, characterized in that, The environmental climate labeling of the effective running dataset to obtain the training running dataset includes: Multiple initial clustering distributions are set, including: distributions for severely cold regions, cold regions, hot summer and cold winter regions, or hot summer and warm winter regions. Climate features are constructed for each of the multiple initial cluster distributions to obtain multiple sets of climate feature parameters, where one initial cluster distribution corresponds to one set of climate feature parameters. Each initial cluster distribution among multiple initial cluster distributions is initialized to obtain multiple initial cluster parameter sets. The initial cluster parameter sets among the multiple initial cluster parameter sets correspond one-to-one with the initial cluster distributions among the multiple initial cluster distributions. The initial cluster distributions include: initial mean, initial covariance, and initial mixing weights. Based on multiple initial clustering parameter sets and valid running datasets, multiple initial clustering distributions are updated to obtain multiple updated clustering distributions; Parameters are calculated for each of the multiple updated clustering distributions to obtain multiple sets of updated clustering parameters; The parameter change values ​​are calculated based on multiple updated clustering parameter sets and multiple initial clustering parameter sets; Multiple updated clustering parameter sets and multiple updated clustering distributions are respectively used as multiple initial clustering parameter sets and multiple initial clustering distributions, and the step of updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets is returned until the parameter change value is not greater than the preset convergence change value. When the parameter change value is not greater than the convergence change value, the multiple updated clustering distributions are recorded as multiple target clustering distributions, and climate labeling is performed based on the multiple target clustering distributions to obtain the training running dataset.

5. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 4, characterized in that, The process involves updating multiple initial clustering distributions based on multiple initial clustering parameter sets and valid running datasets to obtain multiple updated clustering distributions, including: Extract valid runtime data sequentially from the valid runtime dataset, and construct a valid runtime vector based on the valid runtime data; Based on multiple initial clustering parameter sets, the posterior probability between the effective running vector and each initial clustering distribution in the multiple initial clustering distributions is calculated to obtain multiple posterior probabilities; The multiple posterior probabilities corresponding to each valid running data are summarized to obtain a posterior probability set. The posterior probability set corresponds one-to-one with the valid running data in the valid running dataset, and the posterior probability set includes multiple posterior probabilities. Based on the posterior probability set, the effective running dataset is distributed to the multiple initial clustering distributions to obtain multiple updated clustering distributions.

6. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 5, characterized in that, The climate labeling based on multiple target clustering distributions yields a training dataset, including: Target clustering distributions are extracted sequentially from multiple target clustering distributions, and target climate feature parameter groups corresponding to the target clustering distributions are determined from multiple climate feature parameter groups. Identify the target running data group contained in the target cluster distribution; Each target operational data point in the target operational data set is labeled using the target climate characteristic parameter set to obtain the training operational data set; The training dataset is obtained by merging the training data sets corresponding to each of the multiple target clustering distributions.

7. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 6, characterized in that, The process of predicting the power generation of photovoltaic power generation equipment based on real-time climate parameter sets and target power prediction model sets to obtain the predicted power generation includes: The target power prediction model is extracted sequentially from the target power prediction model group, and the real-time climate parameter group is input into the target power prediction model to obtain the initial power generation. Record the average loss value of the target power prediction model during the training step; By summing up the average loss values ​​and the initial power generation, we obtain the average loss value group and the initial power generation group; The model weights are allocated based on the average loss value group to obtain the model prediction weight reorganization, where the model prediction weights in the model prediction weight reorganization correspond one-to-one with the target power prediction models in the target power prediction model group. The initial power generation group is weighted and averaged according to the model prediction weight reorganization to obtain the predicted power generation.

8. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 7, characterized in that, The step of allocating model weights based on the average loss value set to obtain a reorganized model prediction weight includes: The average loss values ​​are extracted sequentially from the average loss value group, and the model prediction weights are calculated based on the average loss values. The model prediction weights are expressed as follows: ， in, Indicates the model's predicted weights. This represents the average loss value. This indicates the number of target power prediction models in the target power prediction model group. Represents the first in the average loss value group Average loss value; The model prediction weights are aggregated to obtain the model prediction weight reorganization.

9. The optimized method for self-consumption of distributed energy based on photovoltaic-storage-direct-drive-flexible energy transmission as described in claim 8, characterized in that, The process of obtaining the real-time grid load of the pre-constructed external grid, and based on the real-time grid load, performing power dispatching to the external grid and energy storage devices respectively, includes: Calculate the real-time power deficit between the predicted power generation and the real-time load power; Obtain the maximum load of the external power grid; If the real-time grid load is less than the maximum dispatch load, the external grid dispatch power and energy storage dispatch power are calculated based on the maximum dispatch load, the real-time grid load, and the real-time power deficit. Power is dispatched to the external power grid and energy storage devices respectively based on the power dispatched from the external power grid and the power dispatched from the energy storage device. If the real-time grid load is not less than the maximum dispatch load, the maximum discharge power is calculated based on the preset maximum state of charge difference, the preset sampling interval, and the preset maximum energy storage capacity. Determine whether the maximum discharge power is greater than the real-time deficit power; If the maximum discharge power is greater than the real-time deficit power, then the energy storage device is discharged based on the real-time deficit power to complete the self-consumption optimization of distributed energy based on photovoltaic-storage-direct-flexible energy storage. If the maximum discharge power is not greater than the real-time deficit power, then calculate the load regulation power between the maximum discharge power and the real-time deficit power. The energy storage device is discharged according to the maximum discharge power, and the flexible power device is reduced in power based on the load adjustment power.

10. A control system using the distributed energy self-consumption optimization method based on photovoltaic-storage-direct-flexible energy transmission as described in any one of claims 1 to 9, characterized in that, The control system includes: The historical data acquisition module is used to receive the consumption optimization instruction, determine the unit to be optimized based on the consumption optimization instruction, wherein the unit to be optimized includes: photovoltaic power generation equipment, energy storage equipment and flexible power consumption equipment, and acquire the historical operation dataset of the photovoltaic power generation equipment in the unit to be optimized, wherein the historical operation dataset includes multiple historical operation data, and the historical operation data includes: historical power generation and historical environmental parameter groups; The prediction model training module is used to filter the environmental power correlation of the historical running dataset to obtain the effective running dataset, label the effective running dataset with environmental climate to obtain the training running dataset, and use the training running dataset to train the pre-acquired neural network model group to obtain the target power prediction model group. The training running data in the training running dataset includes the target climate feature parameter group. The real-time data acquisition module is used to monitor the environment of the unit to be optimized, obtain the real-time climate parameter set, predict the power generation of the photovoltaic power generation equipment based on the real-time climate parameter set and the target power prediction model set, obtain the predicted power generation, detect the real-time power of the flexible power consumption equipment, obtain the real-time load power, and determine whether the predicted power generation is greater than the real-time load power. The real-time power regulation module is used to calculate the real-time energy storage power based on the predicted power generation and the real-time load power, and to store energy in the energy storage device according to the real-time energy storage power. If the predicted power generation is not greater than the real-time load power, the module obtains the real-time grid load of the pre-built external grid and performs power dispatching to the external grid and the energy storage device respectively based on the real-time grid load.

Images