A method for estimating distributed photovoltaic output power

Abstract

The present invention provides a distributed photovoltaic output power estimation method, which includes installing a load sensor and a photovoltaic output sensor at an access point, collecting raw data including load data and photovoltaic output data in real time; processing the collected raw data; and synchronizing the load data and photovoltaic output data in time; extracting key features from the cleaned raw data, and predicting photovoltaic output through a time series analysis model; inputting the real-time collected load data and photovoltaic output data into a trained time series analysis model; the time series analysis model performs real-time calculations based on the load data and photovoltaic output data to estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and the real-time load data, the intelligent coordination controller performs dynamic adjustments to match the photovoltaic output with the load power. The present invention realizes efficient estimation and optimization control of photovoltaic output.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed photovoltaic power generation, and in particular to a distributed photovoltaic output power estimation method. Background Art

[0002] Distributed photovoltaic power generation is a way of generating electricity using solar energy. It installs photovoltaic power generation systems in decentralized locations near users, such as rooftops, parking lots, farmlands, etc. In order to plan, operate and dispatch distributed photovoltaic systems, it is necessary to estimate the distributed photovoltaic output power, which is a process of predicting and estimating the electrical energy that may be generated by distributed photovoltaic power generation systems under specific times and conditions; the estimation is of great significance for grid dispatching, power market transactions and user electricity consumption plans. However, the existing distributed photovoltaic power generation systems have technical problems in maximizing self-generation and self-use, ensuring power supply quality, improving grid stability and improving economic benefits.

[0003] Prior art 1, application number: CN 202310503584.3 discloses a method for estimating distributed photovoltaic output power without monitoring, including: observing existing aggregates and low-voltage distributed power sources, constructing estimation points of low-voltage distributed power sources, and determining low-voltage distributed power source partitions or aggregates according to their geographical locations or feeders; based on the consistency of the correlation between the distributed power source electricity data and power data, a linear estimation model is used to estimate the parameters; based on the accumulated electricity error, it is determined whether to re-perform the aggregation solution and state estimation parameter calculation, and the state estimation strategy is rolled over. Although it can solve the problem of low-voltage unsupervised photovoltaic output estimation in the absence of real-time data, it reduces the complexity of calculation in the form of aggregates, improves the efficiency of estimation, and combines the output curve correction to ensure the accuracy of photovoltaic output power estimation. However, there is no real-time monitoring of the load data and photovoltaic output data of the access point, and there is a lack of relevant equipment and hardware, resulting in low accuracy of the estimation results.

[0004] Prior art 2, application number: CN 202311448616.0 discloses a distributed photovoltaic output prediction method, system and storage medium, the method includes: extracting the shielding characteristics of satellite cloud image irradiance at different times; preprocessing the feature data; using the processed shielding characteristics to establish and train the DC-GAN network model to predict short-term cloud shielding characteristics; combining the irradiance data of centralized photovoltaic power stations and the spatial correlation of distributed photovoltaic power stations and the predicted cloud shielding characteristics of distributed photovoltaic power stations to establish a database; combining the database to establish a distributed photovoltaic power station output prediction model with CNN-LSTM as the structural feature; using the trained prediction model to predict photovoltaic output power. Although it can improve the accuracy of the distributed photovoltaic prediction model. However, it is too dependent on sample data, resulting in a large amount of data to be processed, which affects the efficiency of the prediction to a certain extent.

[0005] Prior art three, application number: CN 202410303807.6 discloses a short-term photovoltaic output prediction method based on separable convolution and deep gate structure neural network, including: identifying cloud types according to a cloud classification model based on separable convolution; predicting photovoltaic output according to a photovoltaic output prediction model based on a deep gate structure neural network; introducing photovoltaic output coefficients for different types of clouds; correcting the predicted photovoltaic output according to the photovoltaic output coefficient to obtain corrected photovoltaic output prediction data. Although by integrating the cloud classification model based on DSCVGG and the photovoltaic output prediction model based on the DGN network, the influence coefficients of different cloud types on photovoltaic output are introduced, the influence of cloud movement on photovoltaic output can be corrected, and the short-term photovoltaic output fluctuations blocked by cloud movement are effectively corrected, so that the obtained distributed photovoltaic output power is more accurate, which can effectively ensure the stability of the distribution network, avoid the phenomenon of abandoned light in distributed photovoltaic power generation systems, and improve the absorption rate. However, the refined data of the model does not include the load data of the access point and the photovoltaic output data, which results in the inability to maximize the energy utilization and the failure to achieve the best match between the photovoltaic output and the load power.

[0006] At present, the existing technologies 1, 2 and 3 do not monitor the load data and photovoltaic output data of the access point in real time, resulting in low accuracy of the estimation result. Therefore, the present invention provides a distributed photovoltaic output power estimation method. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a distributed photovoltaic output power estimation method, comprising the following steps:

[0008] Install load sensors and photovoltaic output sensors at access points to collect raw data including load data and photovoltaic output data in real time; clean the collected raw data to remove noise and abnormal values; and synchronize the load data and photovoltaic output data in time;

[0009] Extract key features including peak and valley periods of load and seasonal changes of photovoltaic output from the cleaned raw data, and predict photovoltaic output through time series analysis model;

[0010] The real-time collected load data and photovoltaic output data are input into the trained time series analysis model; the time series analysis model performs real-time calculations based on the load data and photovoltaic output data to estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller performs dynamic adjustments to match the photovoltaic output with the load power.

[0011] Optionally, the process of installing a load sensor and a photovoltaic output sensor at an access point includes the following steps:

[0012] Obtain the locations of transformers, distribution cabinets, key load equipment, photovoltaic panel arrays and inverters of distributed photovoltaic systems; and simultaneously obtain the input and output ends of transformers, distribution cabinets, key load equipment, photovoltaic panel arrays and inverters;

[0013] Load sensors are installed on transformers, power distribution cabinets, and key load equipment, and photovoltaic output sensors are installed at the output end of the photovoltaic panel array and the input and output ends of the inverter;

[0014] The load sensors and photovoltaic output sensors are networked through the cellular network, the cellular network parameters are configured, and the collected raw data are aggregated to the central node of the cellular network.

[0015] Optional, where the transformer outlet is the key node for converting electricity from the high-voltage side to the low-voltage side, and the load sensor monitors the output power and current of the transformer in real time; load sensors are installed at the incoming and outgoing ends of the distribution cabinet respectively, the incoming end sensor monitors the power load input from the upper power grid, and the outgoing end sensor monitors the power load distributed to each branch; load sensors are installed at the power input end of the key load equipment; photovoltaic output sensors are installed at the output end of the photovoltaic panel array; photovoltaic output sensors are installed at the input end of the inverter to monitor the DC power at the input end of the inverter; photovoltaic output sensors are installed at the output end of the inverter to monitor the AC power output of the inverter.

[0016] Optionally, the process of predicting photovoltaic output through a time series analysis model includes the following steps:

[0017] Decompose the load data into time series, decomposing it into trend terms, seasonal terms and residual terms; identify the peak and trough periods of load through seasonal term analysis;

[0018] Use the seasonal decomposition model to decompose the photovoltaic output data, extract the seasonal terms, and extract the seasonal variation characteristics of photovoltaic output based on the seasonal terms; use the periodic analysis method to model and analyze the seasonal variation of photovoltaic output data; extract the seasonal characteristics of photovoltaic output through periodic analysis;

[0019] A time series analysis model is constructed by combining the autoregressive integrated moving average model and the long short-term memory network model. The autoregressive integrated moving average model is used to capture the short-term fluctuations of photovoltaic output, and the long short-term memory network model is used to capture the long-term prediction of photovoltaic output. The results of short-term fluctuations of photovoltaic output and long-term prediction of photovoltaic output are integrated, and the weighted average method is used to obtain the final prediction result of photovoltaic output.

[0020] Optionally, the process of building a time series analysis model includes the following steps:

[0021] Use the augmented Dickey-Fuller test to test the stability of the photovoltaic output data; if the data is not stable, perform differential processing, and the selection of the number of differentials d is determined by observing the p value of the augmented Dickey-Fuller test and the data graph after differential; draw the autocorrelation function and partial autocorrelation function graphs, observe the significance level in the graph, and select the appropriate autoregressive order p and moving average order q values; use the selected parameters (p, d, q) to fit the autoregressive integrated moving average model to the photovoltaic output data;

[0022] The trained autoregressive integrated moving average model is used to generate the photovoltaic output forecast results in the short term in the future; the photovoltaic output data is normalized to the range of [0, 1] or [-1, 1], and the time series data is divided into multiple subsequences. The length of each subsequence is consistent with the time period of the data and the input requirements of the long short-term memory network model;

[0023] Design the LSTM network structure, including the input layer, multiple hidden layers, and output layer; use grid search to optimize the hyperparameters of the LSTM network model; use the segmented subsequence data to train the LSTM network model, and use the Adam optimizer for gradient descent; use the trained LSTM network model to generate photovoltaic output forecast results in the long term in the future;

[0024] Using a dynamic time window, the weights are dynamically adjusted according to the distance of the prediction time point; the short-term prediction results of the autoregressive integrated moving average model and the long-term prediction results of the long short-term memory network model are weighted averaged to generate the final photovoltaic output prediction results.

[0025] Optionally, the process of generating the final photovoltaic output prediction result includes the following steps:

[0026] Design a time decay function to dynamically adjust the weight according to the distance of the prediction time point; the closer the time is, the higher the weight of the autoregressive integrated moving average model; the farther the time is, the higher the weight of the long short-term memory network model;

[0027] According to the dynamic weight calculation results, the prediction results of the autoregressive integrated moving average model and the long short-term memory network are weighted averaged;

[0028] The weighted average result is output as the final photovoltaic output prediction result; according to the feedback of actual photovoltaic output data, the attenuation coefficient and weight switching time point are dynamically adjusted.

[0029] Optionally, the process of matching the photovoltaic output with the load power includes the following steps:

[0030] The intelligent coordination controller inputs the pre-processed data into the trained time series analysis model, performs real-time calculations based on the current load data and photovoltaic output data, and estimates the photovoltaic output power at the current moment;

[0031] Based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller generates corresponding strategies and adjusts the operating parameters of the distributed photovoltaic system; specifically, it includes:

[0032] Inverter output power adjustment: Based on the estimated photovoltaic output power, the intelligent coordination controller adjusts the inverter output power to ensure that the power generation capacity of the distributed photovoltaic system matches the power demand;

[0033] Adjustment of energy storage system charging and discharging strategy: the intelligent coordination controller adjusts the energy storage system charging and discharging strategy according to the current load demand and photovoltaic output power;

[0034] The intelligent coordination controller monitors the changes in photovoltaic output power and load data in real time, and makes feedback adjustments based on the changes, adjusting the output power of the inverter and the discharge strategy of the energy storage system.

[0035] Optionally, the process of the intelligent coordination controller adjusting the output power of the inverter includes the following steps:

[0036] The intelligent coordination controller first merges the estimated PV output power with the real-time load data to form a comprehensive data set; analyzes the current power demand trend, identifies the peak and trough periods of load, and the fluctuation of PV output;

[0037] Based on the preset strategy model, the intelligent coordination controller generates a preliminary strategy plan, including a power increase strategy during peak load periods and a power reduction strategy during low load periods; the preliminary strategy plan is optimized through particle swarm optimization;

[0038] The intelligent coordination controller identifies the operating parameters of the inverter, including output voltage, frequency and power factor; maps the optimized strategy scheme to a specific parameter adjustment scheme based on the strategy model; and dynamically adjusts the parameters of the inverter based on real-time data;

[0039] The intelligent coordination controller monitors the changes in load demand and photovoltaic output power in real time to obtain the latest operating status; according to the monitored changes in load demand, the intelligent coordination controller responds and adjusts the output power of the inverter.

[0040] Optionally, the process of mapping the optimized strategy scheme to a specific parameter adjustment scheme includes the following steps:

[0041] The intelligent coordination controller breaks down the optimization strategy into multiple specific goals, including power boost, power reduction and frequency adjustment;

[0042] Each target corresponds to the adjustment requirements of one or more inverter parameters; according to the operating parameters of the inverter, the targets in the strategy plan are further refined into specific requirements for the operating parameters;

[0043] Extract load characteristics from specific requirements and generate corresponding parameter adjustment strategies; generate corresponding parameter adjustment strategies based on the characteristics of photovoltaic output; generate strategies for simultaneously adjusting output voltage and frequency when both load and photovoltaic output reach peak values ​​at the same time; generate parameter adjustment strategies based on the results of feature mapping.

[0044] Optionally, the process of further refining the objectives in the strategy plan into specific requirements for operating parameters includes the following steps:

[0045] Identify the specific goals in the optimization strategy, break down each goal into more detailed sub-goals, and ensure that each sub-goal corresponds to the specific inverter parameter adjustment requirements;

[0046] Based on the inverter's operating parameters, including output voltage, frequency, power factor, and current, identify the specific parameters that need to be adjusted for each target; associate each target with the corresponding operating parameters;

[0047] According to the operating parameters of the inverter, the objectives in the strategy plan are further refined into specific requirements for the operating parameters and the adjustment range of the inverter's output voltage and frequency.

[0048] The data acquisition and preprocessing of the present invention can obtain the original data of load and photovoltaic output in real time by installing load sensors and photovoltaic output sensors, ensuring the timeliness and accuracy of the data; removing noise and outliers, improving the purity of the data, making subsequent analysis more reliable; synchronizing the load data and photovoltaic output data in time, ensuring the consistency of the data in the time dimension, and providing a solid foundation for subsequent analysis and prediction. The significance achieved: high-quality original data is the basis for subsequent analysis and prediction, ensuring the accuracy and reliability of the estimation results; through real-time monitoring and data cleaning, abnormal situations can be discovered and handled in time, improving the stability and safety of the distributed photovoltaic system. The feature extraction and prediction model extracts key features such as the peak and valley periods of the load and the seasonal changes of the photovoltaic output from the cleaned data. The key features are crucial for the prediction of the photovoltaic output; through the time series analysis model, the photovoltaic output can be accurately predicted, and the model can capture the time dependence and trend in the data. The significance achieved: by extracting key features and using the time series analysis model, the prediction accuracy of the photovoltaic output can be significantly improved, providing reliable data support for subsequent energy management; accurate prediction results help optimize the distribution and utilization of energy, reduce energy waste, and improve energy utilization efficiency. Real-time calculation and dynamic adjustment, input the real-time collected load data and photovoltaic output data into the trained time series analysis model for real-time calculation, which can quickly estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller can make dynamic adjustments to better match the photovoltaic output with the load power. Significance achieved: Through real-time calculation and dynamic adjustment, the distributed photovoltaic system can quickly respond to load changes and ensure the real-time and accuracy of photovoltaic output; the dynamic adjustment function makes the photovoltaic output better match the load power, reduces energy waste, and improves the overall efficiency of the distributed photovoltaic system; the introduction of the intelligent coordination controller enables the distributed photovoltaic system to have the ability to adapt and be intelligent, and can be optimized and adjusted according to actual conditions, which improves the intelligence level of the distributed photovoltaic system.

[0049] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0050] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0052] Figure 1 This is a flow chart of a method for estimating distributed photovoltaic output power in Embodiment 1 of the present invention;

[0053] Figure 2 This is a process diagram of installing a load sensor and a photovoltaic output sensor at an access point in Embodiment 2 of the present invention;

[0054] Figure 3 This is a process diagram of predicting photovoltaic output by using a time series analysis model in Example 3 of the present invention;

[0055] Figure 4 A process diagram of constructing a time series analysis model in Example 4 of the present invention;

[0056] Figure 5 A process diagram for generating a final photovoltaic output prediction result in Embodiment 5 of the present invention;

[0057] Figure 6 This is a process diagram for matching photovoltaic output with load power in Embodiment 6 of the present invention;

[0058] Figure 7 This is a process diagram of the intelligent coordination controller adjusting the output power of the inverter in Embodiment 7 of the present invention;

[0059] Figure 8 A process diagram of mapping the optimized strategy scheme to a specific parameter adjustment scheme in Embodiment 8 of the present invention;

[0060] Fig. 9 This is a process diagram for further refining the objectives in the strategy scheme into specific requirements for operating parameters in Example 9 of the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0062] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0063] When the following description relates to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are only examples of devices and methods consistent with some aspects of the present application. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not have to be used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0064] Example 1: Figure 1 As shown, an embodiment of the present invention provides a distributed photovoltaic output power estimation method, comprising the following steps:

[0065] S100: Install load sensors and photovoltaic output sensors at access points to collect raw data including load data and photovoltaic output data in real time; clean the collected raw data to remove noise and abnormal values; and synchronize the load data and photovoltaic output data in time;

[0066] S200: Extract key features including peak and valley periods of load and seasonal changes of photovoltaic output from the cleaned raw data, and predict photovoltaic output through a time series analysis model;

[0067] S300: Input the real-time collected load data and photovoltaic output data into the trained time series analysis model; the time series analysis model performs real-time calculations based on the load data and photovoltaic output data to estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller performs dynamic adjustments to match the photovoltaic output with the load power.

[0068] The working principle and beneficial effects of the above technical solution are as follows: in this embodiment, first, a load sensor and a photovoltaic output sensor are installed at the access point to collect raw data including load data and photovoltaic output data in real time; the collected raw data are cleaned to remove noise and outliers; at the same time, the load data and photovoltaic output data are synchronized in time; secondly, key features including peak and valley periods of load, seasonal changes of photovoltaic output, etc. are extracted from the cleaned raw data, and photovoltaic output is predicted through a time series analysis model; finally, the load data and photovoltaic output data collected in real time are input into a trained time series analysis model; the time series analysis model performs real-time calculations based on the load data and photovoltaic output data to estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller performs dynamic adjustments to match the photovoltaic output with the load power. Step S100 of the above scheme is data acquisition and preprocessing. By installing load sensors and photovoltaic output sensors, the original data of load and photovoltaic output can be obtained in real time to ensure the timeliness and accuracy of the data; noise and outliers are removed to improve the purity of the data, making subsequent analysis more reliable; load data and photovoltaic output data are synchronized in time to ensure the consistency of data in the time dimension, providing a solid foundation for subsequent analysis and prediction. Significance achieved: High-quality original data is the basis for subsequent analysis and prediction, ensuring the accuracy and reliability of the estimation results; through real-time monitoring and data cleaning, abnormal situations can be discovered and handled in a timely manner, improving the stability and safety of distributed photovoltaic systems. Step S200 feature extraction and prediction model, extracting key features such as peak and valley periods of load and seasonal changes in photovoltaic output from the cleaned data, key features are crucial for the prediction of photovoltaic output; through the time series analysis model, photovoltaic output can be accurately predicted, and the model can capture the time dependence and trend in the data. Significance achieved: By extracting key features and using the time series analysis model, the prediction accuracy of photovoltaic output can be significantly improved, providing reliable data support for subsequent energy management; accurate prediction results help optimize energy distribution and utilization, reduce energy waste, and improve energy efficiency. Step S300 Real-time calculation and dynamic adjustment, input the real-time collected load data and photovoltaic output data into the trained time series analysis model, perform real-time calculation, and quickly estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller can perform dynamic adjustments to better match the photovoltaic output with the load power.Significance achieved: Through real-time calculation and dynamic adjustment, the distributed photovoltaic system can quickly respond to load changes and ensure the real-time and accuracy of photovoltaic output; the dynamic adjustment function enables photovoltaic output to better match load power, reduces energy waste, and improves the overall efficiency of the distributed photovoltaic system; the introduction of intelligent coordination controllers enables distributed photovoltaic systems to have adaptive and intelligent capabilities, and can be optimized and adjusted according to actual conditions, thereby improving the intelligence level of distributed photovoltaic systems.

[0069] In summary, this embodiment achieves efficient estimation and optimization control of photovoltaic output through three steps: data collection and preprocessing, feature extraction and prediction model, and real-time calculation and dynamic adjustment. Each step together improves the stability, prediction accuracy and intelligence level of the distributed photovoltaic system, and provides strong support for the operation and management of the distributed photovoltaic system.

[0070] Example 2: Figure 2 As shown, based on Example 1, the process of installing a load sensor and a photovoltaic output sensor at an access point provided by an embodiment of the present invention includes the following steps:

[0071] S101: Obtain the locations of the transformer, distribution cabinet, key load equipment, photovoltaic panel array and inverter of the distributed photovoltaic system; and simultaneously obtain the input and output ends of the transformer, distribution cabinet, key load equipment, photovoltaic panel array and inverter;

[0072] S102: Load sensors are installed on transformers, power distribution cabinets, and key load equipment, and photovoltaic output sensors are installed at the output end of the photovoltaic panel array and the input and output ends of the inverter;

[0073] Among them, the transformer outlet is the key node for converting electricity from the high-voltage side to the low-voltage side. The load sensor monitors the output power and current of the transformer in real time. Load sensors are installed at the incoming and outgoing ends of the distribution cabinet. The incoming end sensor monitors the power load input from the upper power grid, and the outgoing end sensor monitors the power load distributed to each branch. Load sensors are installed at the power input end of key load equipment (such as large motors, air-conditioning units, etc.); photovoltaic output sensors are installed at the output end of the photovoltaic panel array; photovoltaic output sensors are installed at the input end of the inverter to monitor the DC power at the input end of the inverter; photovoltaic output sensors are installed at the output end of the inverter to monitor the AC power output of the inverter.

[0074] S103: Networking the load sensor and the photovoltaic output sensor through a cellular network, configuring cellular network parameters, and aggregating the collected raw data to a central node of the cellular network.

[0075] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first obtains the positions of the transformer, distribution cabinet, key load equipment, photovoltaic panel array and inverter of the distributed photovoltaic system; at the same time, obtains the input and output ends of the transformer, distribution cabinet, key load equipment, photovoltaic panel array and inverter; secondly, load sensors are arranged on the transformer, distribution cabinet and key load equipment, and photovoltaic output sensors are installed at the output end of the photovoltaic panel array, the input end and the output end of the inverter; among them, the transformer outlet is the key node for converting electricity from the high voltage side to the low voltage side, and the load sensor monitors the output power and current of the transformer in real time; and the load sensors are installed at the incoming and outgoing ends of the distribution cabinet respectively. Load sensors: Incoming sensors monitor the power load input from the upper power grid, and outgoing sensors monitor the power load distributed to each branch. Load sensors are installed at the power input of key load equipment (such as large motors, air-conditioning units, etc.). Photovoltaic output sensors are installed at the output of the photovoltaic panel array. Photovoltaic output sensors are installed at the input of the inverter to monitor the DC power at the input of the inverter. Photovoltaic output sensors are installed at the output of the inverter to monitor the AC power output of the inverter. Finally, the load sensors and photovoltaic output sensors are networked through the cellular network, the cellular network parameters are configured, and the collected raw data are aggregated to the central node of the cellular network. The above scheme installs load sensors at the input and output ends of transformers, distribution cabinets, and key load equipment to monitor the power and current of power loads in real time; installs photovoltaic output sensors at the output end of the photovoltaic panel array and the input and output ends of the inverter to monitor the DC and AC power of the photovoltaic system; networks all sensors through the cellular network, configures network parameters, and ensures that data can be stably transmitted to the central node of the cellular network; the central node of the cellular network receives and summarizes the raw data collected by all sensors; processes and analyzes the summarized data to generate reports, charts or real-time monitoring screens to provide decision support. The load sensor of this embodiment can monitor the power load of the transformer, distribution cabinet and key load equipment in real time, and provide accurate power and current data; by monitoring the load at the incoming and outgoing ends of the distribution cabinet, the power distribution can be optimized to ensure the load balance of each branch; the load sensor of the key load equipment can monitor the operating status of the equipment in real time, detect abnormal conditions in time, and prevent equipment failure; by monitoring the output power of the photovoltaic panel array and the input and output power of the inverter, the overall performance of the photovoltaic system can be evaluated; the input and output power monitoring of the inverter helps to analyze the energy conversion efficiency and optimize the working state of the inverter; by monitoring the key nodes of the photovoltaic system, faults in the system can be discovered and diagnosed in time, and the reliability and stability of the system can be improved; the cellular network provides a stable communication link to ensure that the sensor data can be transmitted to the central node in real time and accurately; through the cellular network, remote monitoring and management can be achieved, reducing the workload of on-site maintenance and improving operation and maintenance efficiency; cellular networks usually have high security and can protect privacy and security during data transmission.Significance achieved: By real-time monitoring and optimizing power distribution, energy utilization efficiency can be improved and energy waste can be reduced; real-time monitoring of the operating status of equipment and the performance of the photovoltaic system can promptly detect and handle abnormal situations and improve reliability and stability; remote monitoring and management reduce the workload of on-site maintenance and reduce operation and maintenance costs; through the analysis of collected data, intelligent decision-making support can be provided for energy management, and energy configuration and use strategies can be optimized; by monitoring and optimizing the performance of the photovoltaic system, the use of renewable energy can be promoted and the development of green energy can be promoted.

[0076] In this embodiment, the key equipment location and port information of the distributed photovoltaic system are obtained:

[0077] in: A matrix representing the locations of all key equipment, Indicates i The position vector of the transformer, Indicates i The position vector of the distribution cabinet, Indicates i The position vector of each key load device, Indicates i The position vector of the photovoltaic array, Indicates i The position vector of each inverter, The weight matrix representing the transformer position, The weight matrix representing the location of the distribution cabinet, The weight matrix representing the location of critical load equipment, The weight matrix representing the location of the photovoltaic panel array, The weight matrix representing the inverter locations, represents the normalization factor of the position matrix;

[0078] Laying out load sensors and photovoltaic output sensors:

[0079] in: represents the layout matrix of all sensors, Indicates j The load sensor layout vector on each transformer is: Indicates j Load sensor layout vector on the distribution cabinet, - Indicates j Load sensor layout vectors on key load devices, Indicates j The photovoltaic output sensor layout vector at the output end of the photovoltaic panel array, Indicatesj The photovoltaic output sensor layout vectors at the input and output ends of each inverter, represents the weight matrix of transformer sensor placement, represents the weight matrix of the distribution cabinet sensor layout, represents the weight matrix of the key load equipment sensor layout, represents the weight matrix of the photovoltaic panel array sensor layout, represents the weight matrix of inverter sensor layout, represents the normalization factor of the sensor placement matrix;

[0080] Data aggregation via cellular networks:

[0081] in: represents the configuration matrix of the cellular network, Indicates k A cellular network parameter vector, Indicates k data transmission vectors, Indicates k A vector of cellular network center nodes, represents the weight matrix of the cellular network parameters, represents the weight matrix of data transmission, represents the weight matrix of the central node, represents the normalization factor of the network configuration matrix;

[0082] Through the equation, the deployment and data aggregation of load sensors and photovoltaic output sensors in distributed photovoltaic systems can be realized, ensuring the intelligent management and efficient operation of the system.

[0083] In summary, this embodiment realizes comprehensive monitoring and optimization of distributed photovoltaic systems through networking of load sensors and photovoltaic output sensors, improves energy utilization efficiency and system reliability, reduces operation and maintenance costs, supports intelligent decision-making, and promotes the development of renewable energy.

[0084] Example 3: Figure 3 As shown, based on Example 1, the process of predicting photovoltaic output by using a time series analysis model provided by the embodiment of the present invention includes the following steps:

[0085] S201: Decompose the load data into a time series, a trend term, a seasonal term and a residual term; identify the peak and valley periods of the load through the analysis of the seasonal term;

[0086] S202: Decompose the photovoltaic output data using a seasonal decomposition model, extract seasonal items, and extract seasonal variation characteristics of photovoltaic output based on the seasonal items; use a periodic analysis method to model and analyze the seasonal variation of photovoltaic output data; extract seasonal characteristics of photovoltaic output through periodic analysis;

[0087] S203: A time series analysis model is constructed by combining the autoregressive integrated moving average model and the long short-term memory network model. The autoregressive integrated moving average model is used to capture the short-term fluctuation of photovoltaic output, and the long short-term memory network model is used to capture the long-term prediction of photovoltaic output. The results of short-term fluctuation of photovoltaic output and long-term prediction of photovoltaic output are integrated, and the weighted average method is used to obtain the final prediction result of photovoltaic output.

[0088] The working principle and beneficial effects of the above technical solution are as follows: the present embodiment firstly performs a time series decomposition on the load data, decomposing it into trend terms, seasonal terms and residual terms; identifies the peak and valley periods of the load through the analysis of seasonal terms; secondly, decomposes the photovoltaic output data using a seasonal decomposition model, extracts the seasonal terms, and extracts the seasonal variation characteristics of the photovoltaic output based on the seasonal terms; models and analyzes the seasonal variation of the photovoltaic output data using a periodic analysis method; extracts the seasonal characteristics of the photovoltaic output through periodic analysis; finally, combines an autoregressive integral moving average model and a long short-term memory network model to construct a time series analysis model, wherein the autoregressive integral moving average model is used to capture short-term fluctuations in photovoltaic output, and the long short-term memory network model is used to capture long-term predictions of photovoltaic output; integrates the results of short-term fluctuations in photovoltaic output and long-term predictions of photovoltaic output, and adopts a weighted average method to obtain the final prediction result of photovoltaic output. Step S201 of the above scheme decomposes the load data in time series, identifies the long-term trend of the load data, and helps understand the basic growth or decline pattern of the load; identifies the seasonal changes in the load data, helps determine the peak and trough periods of the load; identifies random fluctuations in the load data, and helps evaluate the stability of the data and the accuracy of the prediction. Significance achieved: By identifying the peak and trough periods, the scheduling and resource allocation of the distributed photovoltaic system can be optimized, and energy waste and cost can be reduced; understanding the seasonal changes in the load can help predict and respond to possible power shortages or surpluses. Step S202 uses a seasonal decomposition model to decompose the photovoltaic output data, identify the seasonal change characteristics of the photovoltaic output, and help understand the periodic fluctuations of the photovoltaic output; through periodic analysis, the seasonal characteristics of the photovoltaic output are extracted to help establish a more accurate prediction model. Significance achieved: Through seasonal decomposition and periodic analysis, the changes in photovoltaic output can be more accurately predicted, and the reliability and accuracy of the prediction can be improved; understanding the seasonal characteristics of photovoltaic output can help optimize the resource allocation of photovoltaic power generation systems and improve energy utilization efficiency. Step S203 combines the autoregressive integrated moving average model and the long short-term memory network model. The autoregressive integrated moving average model (ARIMA) is used to capture the short-term fluctuations of photovoltaic output and provide the accuracy of short-term prediction; the long short-term memory network model (LSTM) is used to capture the long-term trend of photovoltaic output and provide the accuracy of long-term prediction; through the weighted average method, the short-term and long-term prediction results are integrated to obtain the final photovoltaic output prediction result. Significance achieved: By combining the ARIMA and LSTM models, the short-term fluctuations and long-term trends of photovoltaic output can be captured at the same time, improving the comprehensiveness and accuracy of the prediction; accurate prediction results are helpful for the optimal scheduling, resource allocation and risk management of distributed photovoltaic systems, and improving the stability and economy of distributed photovoltaic systems.

[0089] In summary, this embodiment can comprehensively analyze and predict the changes in photovoltaic output, optimize the scheduling and resource allocation of distributed photovoltaic systems, and improve energy utilization efficiency and system stability. It not only helps to cope with changes in the energy market, but also provides technical support for the development of sustainable energy.

[0090] Example 4: Figure 4 As shown, based on Example 3, the process of constructing a time series analysis model provided by the embodiment of the present invention includes the following steps:

[0091] S2031: Use the augmented Dickey-Fuller test to test the stability of the photovoltaic output data; if the data is not stable, perform differential processing, and the selection of the number of differentials d is determined by observing the p value of the augmented Dickey-Fuller test and the data graph after differential; draw the autocorrelation function and partial autocorrelation function graphs, observe the significance level in the graph, and select the appropriate autoregressive order p and moving average order q values; use the selected parameters (p, d, q) to fit the autoregressive integrated moving average model to the photovoltaic output data;

[0092] S2032: Generate the photovoltaic output forecast results in the short term in the future by using the trained autoregressive integrated moving average model; normalize the photovoltaic output data to the range of [0, 1] or [-1, 1], and divide the time series data into multiple subsequences, and the length of each subsequence is consistent with the time period of the data and the input requirements of the long short-term memory network model;

[0093] S2033: Design the LSTM network structure, including the input layer, multiple hidden layers and output layer; use grid search to optimize the hyperparameters of the LSTM network model; use the segmented subsequence data to train the LSTM network model, and use the Adam optimizer for gradient descent; use the trained LSTM network model to generate the photovoltaic output forecast results in the long term in the future;

[0094] S2034: Use a dynamic time window to dynamically adjust the weight according to the distance of the prediction time point; perform weighted averaging on the short-term prediction results of the autoregressive integrated moving average model and the long-term prediction results of the long short-term memory network model to generate the final photovoltaic output prediction result.

[0095] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the augmented Dickey-Fuller test is first used to test the stability of photovoltaic output data; if the data is not stable, differential processing is performed, and the selection of the differential number d is determined by observing the p value of the augmented Dickey-Fuller test and the data graph after differential; the autocorrelation function and partial autocorrelation function graphs are plotted, the significance level in the graph is observed, and the appropriate autoregressive order p and moving average order q values ​​are selected; the selected parameters (p, d, q) are used to fit the autoregressive integrated moving average model to the photovoltaic output data; secondly, the trained autoregressive integrated moving average model is used to generate the photovoltaic output prediction results in the short term in the future; the photovoltaic output data is normalized to [0, 1] or [-1, 1], the time series data is divided into multiple subsequences, and the length of each subsequence is consistent with the time period of the data and the input requirements of the long short-term memory network model; then the long short-term memory network structure is designed, including the input layer, multiple hidden layers and the output layer; the hyperparameters of the long short-term memory network model are optimized using grid search; the long short-term memory network model is trained using the segmented subsequence data, and the Adam optimizer is used for gradient descent; the trained long short-term memory network model is used to generate the photovoltaic output prediction results in the long term in the future; finally, the dynamic time window is used to dynamically adjust the weight according to the distance of the prediction time point; the short-term prediction results of the autoregressive integrated moving average model and the long-term prediction results of the long short-term memory network model are weighted averaged to generate the final photovoltaic output prediction results. Step S2031 of the above scheme is used to test the stability of time series data; if the data is not stable, differential processing is performed to make the data stable; through differential, the trend and seasonality in the time series are eliminated to make it more suitable for the modeling of the autoregressive integrated moving average model; the autocorrelation function and partial autocorrelation function graph help determine the autoregressive order p and the moving average order q of the autoregressive integrated moving average model; using the selected parameters (p, d, q), the autoregressive integrated moving average model is fitted to the photovoltaic output data to generate short-term prediction results. Significance: Ensure that the time series data is statistically consistent and avoid the model being affected by non-stationarity; through the autocorrelation function and partial autocorrelation function graph, the parameters of the autoregressive integrated moving average model are scientifically selected to improve the prediction accuracy of the model; the autoregressive integrated moving average model is used to generate photovoltaic output prediction results in the short term, providing a basis for subsequent long-term predictions. Step S2032: Data preprocessing and short-term prediction: normalize the PV output data to the range of [0, 1] or [-1, 1] to facilitate model processing and improve the convergence speed of the model; divide the time series data into multiple subsequences, and the length of each subsequence is consistent with the time period of the data and the input requirements of the long short-term memory network model; use the trained autoregressive integrated moving average model to generate the PV output prediction results in the short term in the future.Significance: Normalization processing makes data of different dimensions comparable and improves the generalization ability of the model; prepares for the input of the long short-term memory network model, ensures that the length of each subsequence meets the input requirements of the long short-term memory network model, and improves the training efficiency of the model; provides short-term photovoltaic output prediction results, and provides reference for subsequent long-term prediction and comprehensive prediction. Step S2033 Long short-term memory network model design and training, including input layer, multiple hidden layers and output layer, to ensure that the model can capture long-term dependencies in time series; use grid search to optimize the hyperparameters of the long short-term memory network model to improve the prediction performance of the model; use Adam optimizer to perform gradient descent, train the long short-term memory network model, and generate photovoltaic output prediction results in the future long term. Significance: The long short-term memory network model can capture long-term dependencies in time series and improve the accuracy of long-term prediction; optimize hyperparameters through grid search to ensure that the model performs stably and efficiently on different data sets; generate photovoltaic output prediction results in the future long term, and provide important reference for the planning and scheduling of distributed photovoltaic systems. Step S2034 Dynamic weighting and comprehensive prediction: dynamically adjust the weight according to the distance of the prediction time point, so that the prediction result is more flexible and accurate; weight the short-term prediction results of the autoregressive integral sliding average model and the long-term prediction results of the long short-term memory network model to generate the final photovoltaic output prediction result. Significance: Dynamically adjust the weight according to the distance of the prediction time point to make the prediction result more in line with the actual situation; combine the short-term and long-term prediction results to generate the final comprehensive prediction result, improve the comprehensiveness and accuracy of the prediction; provide reliable prediction data for the operation and scheduling of distributed photovoltaic systems, and support the scientificity and effectiveness of decision-making.

[0096] In summary, the time series analysis model constructed in this embodiment can effectively deal with the stationarity problem of photovoltaic output data, select appropriate model parameters, generate short-term and long-term prediction results, and finally provide accurate photovoltaic output prediction results through dynamic weighted comprehensive prediction. It not only improves the accuracy of the prediction, but also provides important decision-making support for the operation and scheduling of distributed photovoltaic systems. Through the above-mentioned hierarchical construction process, combined with the autoregressive integral moving average model and the long short-term memory network model, the short-term fluctuations and long-term trends of photovoltaic output can be fully captured, and the prediction results can be fused by the weighted average method to improve the accuracy and reliability of photovoltaic output prediction. It not only optimizes the prediction performance of the model, but also provides strong technical support for the scheduling and resource allocation of distributed photovoltaic systems. By refining the steps of data preprocessing, model training and result fusion, the robustness and prediction accuracy of the model are further improved, ensuring the comprehensiveness and accuracy of photovoltaic output prediction.

[0097] Example 5: Figure 5As shown, based on Example 4, the process of generating the final photovoltaic output prediction result provided by the embodiment of the present invention includes the following steps:

[0098] S20341: Design a time decay function to dynamically adjust the weight according to the distance of the prediction time point; the closer the time is, the higher the weight of the autoregressive integrated moving average model; the farther the time is, the higher the weight of the long short-term memory network model;

[0099] formula:

[0100] In the formula, It's time point t The weight of k is the attenuation coefficient, which controls the speed of change of the weight. is the time point of weight switching;

[0101] S20342: Perform weighted averaging of the prediction results of the autoregressive integrated moving average model and the long short-term memory network according to the dynamic weight calculation results;

[0102] S20343: Output the weighted average result as the final photovoltaic output prediction result; dynamically adjust the attenuation coefficient and weight switching time point according to the feedback of the actual photovoltaic output data.

[0103] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, a time decay function is first designed to dynamically adjust the weight according to the distance of the prediction time point; the closer the time is, the higher the weight of the autoregressive integrated moving average model is; the farther the time is, the higher the weight of the long short-term memory network model is; secondly, according to the dynamic weight calculation result, the prediction results of the autoregressive integrated moving average model and the long short-term memory network are weighted averaged; finally, the weighted average result is output as the final photovoltaic output prediction result; according to the feedback of the actual photovoltaic output data, the decay coefficient and the weight switching time point are dynamically adjusted. Step S20341 of the above scheme designs a time decay function, and by designing a time decay function, the weight can be dynamically adjusted according to the distance of the prediction time point. The closer the time is, the higher the weight of the autoregressive integrated moving average model (ARIMA); the farther the time is, the higher the weight of the long short-term memory network model (LSTM); this dynamic adjustment mechanism enables the model to better capture the recent and long-term data characteristics; the introduction of the decay coefficient and the weight switching time point makes the weight distribution of the model on different time scales more flexible and can adapt to different application scenarios and data characteristics. Significance achieved: By dynamically adjusting the weights, the model can better balance the recent and long-term data information, thereby improving the accuracy of photovoltaic output forecasting; the dynamic weight adjustment mechanism enables the model to better adapt to different data distributions and time series characteristics, and enhances the robustness and adaptability of the model. Step S20342 performs weighted averaging according to the dynamic weight calculation results. By weighted averaging the prediction results of ARIMA and LSTM, the advantages of the two models can be comprehensively utilized to make up for the shortcomings of a single model. The ARIMA model is good at capturing linear trends, while the LSTM model is good at capturing nonlinear relationships. The weighted average can combine the advantages of the two models to improve the comprehensiveness of the forecast; the weighted average can smooth the forecast results of a single model, reduce the volatility of the forecast results, and make the final photovoltaic output forecast more stable and reliable. Significance achieved: Through weighted averaging, the forecast results of the model are smoother and more stable, reducing the forecast deviation and volatility that may occur in a single model, and improving the reliability of the forecast; more accurate forecast results help optimize the resource allocation of the photovoltaic system, improve energy efficiency, and reduce operating costs. Step S20343 outputs the final prediction result and dynamically adjusts the parameters. According to the feedback of the actual photovoltaic output data, the attenuation coefficient and the weight switching time point are dynamically adjusted, so that the model can adapt to new data features and environmental changes in real time; the dynamic parameter adjustment mechanism enables the model to continuously optimize its own performance, improve the prediction accuracy and stability, and enhance the model's adaptive ability. Significance achieved: The dynamic parameter adjustment mechanism enables the model to adapt to new data features and environmental changes in real time, improves the model's adaptive ability and robustness; by continuously adjusting the parameters, the model can continuously optimize the prediction effect, improve the accuracy and reliability of photovoltaic output prediction, and provide more reliable support for practical applications.

[0104] In summary, the photovoltaic output prediction model of this embodiment can not only dynamically adjust weights and comprehensively utilize the advantages of multiple models, but also provide real-time feedback and adjust parameters to continuously optimize the prediction effect. It not only improves the prediction accuracy and stability of the model, but also enhances the model's adaptability and robustness, providing strong support for the optimized operation and resource allocation of photovoltaic systems. Through an innovative dynamic weighting strategy, this embodiment can not only effectively combine the advantages of the autoregressive integral moving average model and the long short-term memory network model, but also dynamically adjust the weights according to the distance of the prediction time point, thereby improving the accuracy and stability of the prediction; compared with conventional static weighting methods, this strategy is more flexible and adaptable, and can better cope with the complexity and uncertainty of photovoltaic output data.

[0105] Example 6: Figure 6 As shown, based on Example 1, the process of matching photovoltaic output with load power provided by the embodiment of the present invention includes the following steps:

[0106] S301: The intelligent coordination controller inputs the pre-processed data into the trained time series analysis model, performs real-time calculations based on the current load data and photovoltaic output data, and estimates the photovoltaic output power at the current moment;

[0107] S302: Based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller generates corresponding strategies and adjusts the operating parameters of the distributed photovoltaic system; specifically, it includes:

[0108] Inverter output power adjustment: Based on the estimated photovoltaic output power, the intelligent coordination controller adjusts the inverter output power to ensure that the power generation capacity of the distributed photovoltaic system matches the power demand; for example, during peak load periods, the inverter can increase its output power to meet higher power demand; during low load periods, the inverter can reduce its output power to avoid excess power;

[0109] The intelligent coordination controller adjusts the charging and discharging strategy of the energy storage system according to the current load demand and photovoltaic output power. For example, when the photovoltaic output is excessive, the intelligent coordination controller can instruct the energy storage system to charge and store excess electricity; when the photovoltaic output is insufficient, the intelligent coordination controller can instruct the energy storage system to discharge and supplement the power supply.

[0110] S303: The intelligent coordination controller monitors the changes in photovoltaic output power and load data in real time, and makes feedback adjustments based on the changes to adjust the output power of the inverter and the discharge strategy of the energy storage system.

[0111] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the intelligent coordination controller first inputs the pre-processed data into the trained time series analysis model, performs real-time calculations based on the current load data and photovoltaic output data, and estimates the photovoltaic output power at the current moment; secondly, based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller generates corresponding strategies and adjusts the operating parameters of the distributed photovoltaic system; specifically, including: inverter output power adjustment, according to the estimated photovoltaic output power, the intelligent coordination controller adjusts the output power of the inverter to ensure that the power generation capacity of the distributed photovoltaic system matches the power demand; for example, during peak load periods, the inverter can Increase output power to meet higher power demand; during low load periods, the inverter can reduce output power to avoid excess power; the energy storage system charging and discharging strategy is adjusted, and the intelligent coordination controller adjusts the energy storage system charging and discharging strategy according to the current load demand and photovoltaic output power; for example, when the photovoltaic output is in excess, the intelligent coordination controller can instruct the energy storage system to charge and store excess electricity; when the photovoltaic output is insufficient, the intelligent coordination controller can instruct the energy storage system to discharge and supplement the power supply; finally, the intelligent coordination controller monitors the changes in photovoltaic output power and load data in real time, and makes feedback adjustments based on the changes to adjust the inverter's output power and the energy storage system's discharge strategy. Step S301 of the above scheme is model calculation and power estimation. Through real-time calculation, the intelligent coordination controller can quickly estimate the photovoltaic output power at the current moment to ensure the timeliness and accuracy of the data; using the trained time series analysis model, the photovoltaic output power can be estimated more accurately to reduce errors. Significance: Accurate power estimation is the basis for subsequent strategy generation and parameter adjustment, and provides reliable data support for the intelligent coordination controller; based on accurate power estimation, the intelligent coordination controller can make more optimized decisions and improve the operating efficiency of the photovoltaic system. Step S302 strategy generation and parameter adjustment, the intelligent coordination controller can dynamically adjust the parameters of the inverter and energy storage system according to real-time data to ensure that the operation of the photovoltaic system matches the power demand; by generating corresponding strategies, the intelligent coordination controller can adapt to different power demands and photovoltaic output conditions to improve the flexibility and adaptability of the system. Significance: By adjusting the inverter output power and the energy storage system charging and discharging strategy, the intelligent coordination controller can achieve a dynamic balance between power supply and demand and avoid excess or insufficient power; dynamic adjustment of parameters helps to improve the stability of the power system and reduce system failures caused by power fluctuations. Step S303 real-time monitoring and feedback adjustment, the intelligent coordination controller can monitor the changes in photovoltaic output power and load data in real time, and quickly make feedback adjustments to ensure the rapid response capability of the system; through real-time feedback adjustment, the intelligent coordination controller can continuously optimize the parameters of the inverter and energy storage system to improve the operating efficiency of the system.Significance: Real-time monitoring and feedback adjustment ensure that the system can respond quickly to changes in power demand and photovoltaic output, ensuring the real-time nature of power supply; through continuous feedback adjustment, the intelligent coordination controller can continuously optimize the system's operating parameters and improve the overall performance and efficiency of the photovoltaic system.

[0112] In summary, this embodiment achieves the matching of photovoltaic output and load power. It provides accurate data support, realizes dynamic adjustment and adaptability, and ensures real-time performance and continuous optimization. It not only improves the stability and reliability of the power system, but also optimizes the operating efficiency of the photovoltaic power generation system and achieves a dynamic balance between power supply and demand.

[0113] Example 7: Figure 7 As shown, based on Example 6, the process of adjusting the output power of the inverter by the intelligent coordination controller provided in the embodiment of the present invention includes the following steps:

[0114] S3021: The intelligent coordination controller first integrates the estimated PV output power with the real-time load data to form a comprehensive data set; analyzes the current power demand trend, identifies the peak and trough periods of load, and the fluctuation of PV output;

[0115] S3022: Based on the preset strategy model, the intelligent coordination controller generates a preliminary strategy plan, including a power increase strategy during the peak load period and a power reduction strategy during the low load period, etc.; the preliminary strategy plan is optimized by particle swarm optimization;

[0116] Among them, the preset strategy models include: load forecasting model, which is used to predict the load demand in the future and help identify the peak and valley periods of load; photovoltaic output forecasting model, which is used to predict the output power of photovoltaic system and help identify the fluctuation of photovoltaic output; power balance model, which is used to ensure the balance between photovoltaic output and load demand to avoid excess or insufficient power; optimization target model, which sets the optimization goals of the system, such as maximizing efficiency, minimizing cost, improving stability, etc.; particle swarm optimization PSO finds the optimal solution by simulating the group behavior of bird flocks or fish schools. In the intelligent coordination controller, particle swarm optimization is used to optimize the preliminary strategy plan;

[0117] S3023: The intelligent coordination controller identifies the operating parameters of the inverter, including output voltage, frequency, and power factor, etc.; maps the optimized strategy scheme to a specific parameter adjustment scheme according to the strategy model; and dynamically adjusts the parameters of the inverter according to real-time data;

[0118] S3024: The intelligent coordination controller monitors the changes in load demand and photovoltaic output power in real time to obtain the latest operating status; according to the monitored changes in load demand, the intelligent coordination controller responds and adjusts the output power of the inverter.

[0119] The working principle and beneficial effects of the above technical solution are as follows: in this embodiment, the intelligent coordination controller first integrates the estimated photovoltaic output power with the real-time load data to form a comprehensive data set; analyzes the current power demand trend, identifies the peak and trough periods of load, and the fluctuation of photovoltaic output; secondly, based on the preset strategy model, the intelligent coordination controller generates a preliminary strategy plan, including a power increase strategy during the peak load period and a power reduction strategy during the trough period of load; optimizes the preliminary strategy plan through particle swarm optimization; then the intelligent coordination controller identifies the operating parameters of the inverter, including output voltage, frequency and power factor; according to the strategy model, maps the optimized strategy plan to a specific parameter adjustment plan; dynamically adjusts the parameters of the inverter according to real-time data; finally, the intelligent coordination controller monitors the changes in load demand and photovoltaic output power in real time to obtain the latest operating status; according to the monitored changes in load demand, the intelligent coordination controller responds and adjusts the output power of the inverter. Step S3021 of the above scheme is data fusion and demand analysis. By integrating the estimated photovoltaic output power with the real-time load data, a comprehensive data set is formed to ensure that the basic data for strategy generation is comprehensive and real-time; analyze the power demand trend, identify the peak and valley periods of load, and the fluctuation of photovoltaic output, and help the controller understand the dynamic characteristics of the distributed photovoltaic system. Significance: Ensure that the basic data for strategy generation is accurate and reliable, and provide a solid foundation for subsequent strategy generation and parameter adjustment; through demand analysis, the controller can better understand the operating status of the distributed photovoltaic system, and provide a basis for subsequent strategy generation and dynamic adjustment. Step S3022 Strategy generation and optimization: Based on the preset strategy model, a preliminary strategy plan is generated, including a power increase strategy during the peak load period and a power reduction strategy during the valley load period; the preliminary strategy plan is optimized through the particle swarm optimization algorithm to ensure that the strategy maximizes the efficiency and stability of the system while meeting the power demand. Significance: Through the strategy model, the controller can generate a variety of strategy plans to deal with different scenarios, improve the flexibility and adaptability of the system; through the particle swarm optimization algorithm, the strategy plan can maximize the efficiency and stability of the system while meeting the power demand, and improve the overall performance of the system. Step S3023 parameter identification and adjustment, identify the operating parameters of the inverter, including output voltage, frequency, power factor, etc., to ensure the accuracy and pertinence of parameter adjustment; map the optimized strategy scheme to the specific parameter adjustment scheme according to the strategy model to ensure that the strategy can be accurately implemented; dynamically adjust the parameters of the inverter according to real-time data to ensure the stability and efficiency of the system. Significance: Through parameter identification, ensure the accuracy and pertinence of parameter adjustment, avoid system instability caused by improper parameter adjustment; through parameter mapping, ensure that the generated strategy can be accurately implemented and improve the execution effect of the strategy; through dynamic adjustment, ensure the stability and efficiency of the system, avoid fluctuations and overloads in system operation.Step S3024 real-time monitoring and dynamic adjustment, real-time monitoring of changes in load demand and photovoltaic output power, obtaining the latest operating status, ensuring that the controller can respond to system changes in a timely manner; according to the monitored load demand changes, the intelligent coordination controller responds quickly and adjusts the output power of the inverter to ensure the balance between photovoltaic output and load demand. Significance: Through real-time monitoring, the controller can quickly respond to system changes, improve the system's response speed and dynamic adjustment capabilities; through dynamic adjustment, ensure the balance between photovoltaic output and load demand, avoid excess or insufficient power, and improve the operating efficiency and stability of distributed photovoltaic systems.

[0120] In summary, the process of adjusting the inverter output power by the intelligent coordination controller of this embodiment ensures the precise matching of photovoltaic output and load power through steps such as data fusion and demand analysis, strategy generation and optimization, parameter identification and adjustment, real-time monitoring and dynamic adjustment. Each step together improves the operating efficiency, stability and response speed of the system, ensuring the safe, stable and efficient operation of the distributed photovoltaic system. The process of adjusting the inverter output power by the intelligent coordination controller of this embodiment is hierarchical and systematic. From strategy generation to parameter adjustment, and then to dynamic adjustment and optimization of the inverter output power, each step is closely linked to ensure the precise matching of photovoltaic output and load power. Through dynamic adjustment and optimization algorithms, the intelligent coordination controller can achieve precise control of the inverter output power and improve the operating efficiency and stability of the distributed photovoltaic system.

[0121] Example 8: Figure 8 As shown, based on Example 7, the process of mapping the optimized strategy scheme to a specific parameter adjustment scheme provided in the embodiment of the present invention includes the following steps:

[0122] S30231: The intelligent coordination controller decomposes the optimization strategy into multiple specific goals, including power increase, power reduction and frequency adjustment;

[0123] S30232: Each target corresponds to one or more inverter parameter adjustment requirements; according to the inverter operating parameters, the targets in the strategy scheme are further refined into specific requirements for the operating parameters;

[0124] S30233: Extract load characteristics from specific requirements and generate corresponding parameter adjustment strategies; generate corresponding parameter adjustment strategies based on the characteristics of photovoltaic output; generate a strategy for simultaneously adjusting output voltage and frequency when the load and photovoltaic output reach peak values ​​at the same time; generate parameter adjustment strategies based on the results of feature mapping.

[0125] The working principle and beneficial effects of the above technical solution are as follows: in this embodiment, first, the intelligent coordination controller decomposes the optimized strategy scheme into multiple specific goals, including power boost, power reduction and frequency adjustment, etc.; secondly, each goal corresponds to the adjustment requirements of one or more inverter parameters; according to the operating parameters of the inverter, the goals in the strategy scheme are further refined into specific requirements for the operating parameters; finally, the load characteristics are extracted from the specific requirements to generate the corresponding parameter adjustment strategy; according to the characteristics of the photovoltaic output, the corresponding parameter adjustment strategy is generated; when the load and photovoltaic output reach the peak at the same time, a strategy for adjusting the output voltage and frequency at the same time is generated; according to the result of the feature mapping, a parameter adjustment strategy is generated. Step S30231 of the above scheme: The intelligent coordination controller decomposes the optimized strategy scheme into multiple specific goals, and decomposes the complex optimization strategy into multiple specific goals, such as power boost, power reduction and frequency adjustment, so that the strategy is clearer and more operational; by decomposing the goals, the distributed photovoltaic system can manage multiple goals at the same time to ensure that each goal can be effectively executed. Significance: By refining the goals, the distributed photovoltaic system can execute the strategy more accurately and avoid ambiguity and uncertainty in the strategy execution process; multi-goal management enables the distributed photovoltaic system to flexibly respond to different operating scenarios and improve the adaptability and robustness of the distributed photovoltaic system. Step S30232 Each goal corresponds to the adjustment requirements of one or more inverter parameters. According to the operating parameters of the inverter, the goals in the strategy plan are further refined into specific requirements for the operating parameters to ensure that each goal has a clear parameter adjustment plan; by analyzing the operating parameters of the inverter, the distributed photovoltaic system can identify the correlation between different parameters, thereby generating a more reasonable parameter adjustment strategy. Significance: By refining the parameter adjustment requirements, the distributed photovoltaic system can adjust the operating parameters of the inverter more accurately to ensure the effective execution of the strategy; a reasonable parameter adjustment strategy can optimize the operating efficiency of the inverter and improve the overall performance of the distributed photovoltaic system. Step S30233 Extract load characteristics from specific requirements and generate corresponding parameter adjustment strategies. By analyzing the load characteristics, the distributed photovoltaic system can identify specific load patterns, such as peak periods and valley periods, and thus generate corresponding parameter adjustment strategies; based on the load characteristics, the distributed photovoltaic system can dynamically generate parameter adjustment strategies to ensure the best power balance under different load conditions. Significance: By dynamically adjusting parameters, the distributed photovoltaic system can maintain stable operation during both peak and valley periods; a reasonable parameter adjustment strategy can optimize energy utilization efficiency and reduce energy waste. Generate corresponding parameter adjustment strategies based on the characteristics of photovoltaic output.By analyzing the characteristics of photovoltaic output, the distributed photovoltaic system can identify specific modes of photovoltaic output, such as peak value, fluctuation, etc., and thus generate corresponding parameter adjustment strategies; according to the characteristics of photovoltaic output, the distributed photovoltaic system can dynamically generate parameter adjustment strategies to ensure the best power balance under different photovoltaic output conditions. Significance: By dynamically adjusting parameters, the distributed photovoltaic system can maintain the stable operation of the distributed photovoltaic system during both the peak and valley periods of photovoltaic output; reasonable parameter adjustment strategies can optimize the utilization efficiency of photovoltaic energy and improve the overall performance of photovoltaic distributed photovoltaic systems. When the load and photovoltaic output reach peak values ​​at the same time, a strategy for adjusting the output voltage and frequency at the same time is generated. When the load and photovoltaic output reach peak values ​​at the same time, the distributed photovoltaic system can comprehensively consider multiple characteristics and generate a coordinated parameter adjustment strategy; through real-time monitoring and feedback mechanisms, the distributed photovoltaic system can optimize the generated strategy to ensure the effectiveness and adaptability of the strategy. Significance: By collaboratively adjusting the output voltage and frequency, the distributed photovoltaic system can quickly respond to changes in load and photovoltaic output, improving the response speed of the distributed photovoltaic system; reasonable collaborative adjustment strategies can optimize power balance and ensure the stable operation of the distributed photovoltaic system under extreme conditions.

[0126] In summary, the intelligent coordinated control of the distributed photovoltaic system in this embodiment can map the optimized strategy scheme to a specific parameter adjustment scheme to achieve accurate parameter adjustment and strategy execution. Each step is reflected in improving the accuracy, stability, flexibility and energy efficiency of the distributed photovoltaic system. Ultimately, this logic- and feature-based parameter adjustment strategy can significantly improve the performance of the intelligent coordinated control of the distributed photovoltaic system and ensure the stable and efficient operation of the distributed photovoltaic system.

[0127] Example 9: Fig. 9 As shown, based on Example 7, the process of further refining the objectives in the strategy scheme into specific requirements for operating parameters provided by the embodiment of the present invention includes the following steps:

[0128] S30241: Identify the specific goals in the optimization strategy, break down each goal into more detailed sub-goals, and ensure that each sub-goal corresponds to the specific inverter parameter adjustment requirements;

[0129] S30242: Identify specific parameters that need to be adjusted for each target based on the operating parameters of the inverter, including output voltage, frequency, power factor, and current; and associate each target with the corresponding operating parameters;

[0130] S30243: Based on the operating parameters of the inverter, the objectives in the strategy plan are further refined into specific requirements for the operating parameters, including the adjustment range of the inverter's output voltage and frequency.

[0131] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first identifies the specific goals in the optimized strategy solution, decomposes each goal into more detailed sub-goals, and ensures that each sub-goal corresponds to a specific inverter parameter adjustment requirement; secondly, according to the operating parameters of the inverter, including output voltage, frequency, power factor and current, etc., identify the specific parameters that need to be adjusted for each goal; associate each goal with the corresponding operating parameters; finally, according to the operating parameters of the inverter, further refine the goals in the strategy solution into specific requirements for the operating parameters, to the adjustment range of the output voltage and frequency of the inverter. Step S30241 of the above solution identifies the specific goals in the optimized strategy solution, decomposes each goal into more detailed sub-goals, and ensures that each sub-goal corresponds to a specific inverter parameter adjustment requirement; by decomposing the goals in the strategy solution into more detailed sub-goals, ensure that each sub-goal can correspond to a specific inverter parameter adjustment requirement; ensure that each sub-goal accurately corresponds to the specific parameter adjustment requirement of the inverter to avoid ambiguity and uncertainty. Significance achieved: By refining the goals, the implementation of the strategy is ensured to be more accurate, and errors and deviations are reduced; accurate goal decomposition helps the system remain stable under different operating conditions and avoids system fluctuations caused by goal ambiguity. Step S30242 identifies the specific parameters that need to be adjusted for each goal based on the operating parameters of the inverter, including output voltage, frequency, power factor, and current; associates each goal with the corresponding operating parameters; identifies the specific parameters that need to be adjusted for each goal based on the operating parameters of the inverter to ensure the targeted parameter adjustment; associates each goal with the corresponding operating parameters to ensure that each goal can correspond to specific parameter adjustment requirements. Significance achieved: By accurately identifying and associating operating parameters, the parameter adjustment strategy is optimized and the operating efficiency of the system is improved; accurate parameter association helps the system respond quickly to different target requirements and improves the dynamic response capability of the system. Step S30243: According to the operating parameters of the inverter, the objectives in the strategy scheme are further refined into specific requirements for the operating parameters, and the adjustment range of the output voltage and frequency of the inverter; According to the operating parameters of the inverter, the objectives in the strategy scheme are further refined into specific requirements for the operating parameters to ensure the operability and accuracy of the strategy; The adjustment range of the output voltage and frequency of the inverter is clarified to ensure the rationality and effectiveness of the parameter adjustment. Significance achieved: By refining the target requirements, the implementation of the strategy is ensured to be operational and the uncertainty in the implementation of the strategy is avoided; the precise parameter adjustment range helps to improve the performance of the system and ensure that the system remains efficient and stable under different operating conditions.

[0132] In summary, this embodiment further refines the goals in the strategy scheme into specific requirements for operating parameters, which can significantly improve the performance and stability of the intelligent coordinated control system. Each step ensures the accurate implementation of the strategy and the stable operation of the system. Ultimately, this hierarchical refinement strategy can significantly improve the performance of the intelligent coordinated control system and ensure the stable and efficient operation of the power system.

[0133] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of equivalent technologies of the present invention, the present invention is also intended to include these modifications and variations.

Claims

1. A distributed photovoltaic power output estimation method, characterized in that: The following steps are involved: Install load sensors and photovoltaic output sensors at access points to collect raw data including load data and photovoltaic output data in real time; clean the collected raw data to remove noise and abnormal values; and synchronize the load data and photovoltaic output data in time; Extract key features including peak and valley periods of load and seasonal changes of photovoltaic output from the cleaned raw data, and predict photovoltaic output through time series analysis model; The real-time collected load data and photovoltaic output data are input into the trained time series analysis model; the time series analysis model performs real-time calculations based on the load data and photovoltaic output data to estimate the photovoltaic output power at the current moment; based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller performs dynamic adjustments to match the photovoltaic output with the load power; The process of predicting photovoltaic output through the time series analysis model includes the following steps: Decompose the load data into time series, decomposing it into trend terms, seasonal terms and residual terms; identify the peak and trough periods of load through seasonal term analysis; Use the seasonal decomposition model to decompose the photovoltaic output data, extract the seasonal terms, and extract the seasonal variation characteristics of photovoltaic output based on the seasonal terms; The seasonal variation of photovoltaic output data is modeled and analyzed using the periodic analysis method; the seasonal characteristics of photovoltaic output are extracted through periodic analysis; A time series analysis model is constructed by combining the autoregressive integrated moving average model and the long short-term memory network model. The autoregressive integrated moving average model is used to capture the short-term fluctuations of photovoltaic output, and the long short-term memory network model is used to capture the long-term prediction of photovoltaic output. The results of short-term fluctuations of photovoltaic output and long-term prediction of photovoltaic output are integrated, and the weighted average method is used to obtain the final prediction result of photovoltaic output.

2. The distributed photovoltaic output power estimation method according to claim 1, characterized in that: The process of installing load sensors and PV output sensors at access points includes the following steps: Obtain the locations of transformers, distribution cabinets, key load equipment, photovoltaic panel arrays and inverters of distributed photovoltaic systems; and simultaneously obtain the input and output ends of transformers, distribution cabinets, key load equipment, photovoltaic panel arrays and inverters; Load sensors are installed on transformers, power distribution cabinets, and key load equipment, and photovoltaic output sensors are installed at the output end of the photovoltaic panel array and the input and output ends of the inverter; The load sensors and photovoltaic output sensors are networked through the cellular network, the cellular network parameters are configured, and the collected raw data are aggregated to the central node of the cellular network.

3. The distributed photovoltaic output power estimation method according to claim 2, characterized in that: in, The transformer outlet is the key node for converting electricity from the high-voltage side to the low-voltage side. The load sensor monitors the output power and current of the transformer in real time. Load sensors are installed at the incoming and outgoing ends of the distribution cabinet. The incoming sensor monitors the power load input from the upper power grid, and the outgoing sensor monitors the power load distributed to each branch. Load sensors are installed at the power input end of key load equipment. Photovoltaic output sensors are installed at the output end of the photovoltaic panel array. Photovoltaic output sensors are installed at the input end of the inverter to monitor the DC power at the input end of the inverter. Photovoltaic output sensors are installed at the output end of the inverter to monitor the AC power output of the inverter.

4. The distributed photovoltaic output power estimation method according to claim 1, characterized in that: The process of building a time series analysis model includes the following steps: Use the augmented Dickey-Fuller test to test the stability of the photovoltaic output data; if the data is not stable, perform differential processing, and the selection of the number of differentials d is determined by observing the p value of the augmented Dickey-Fuller test and the data graph after differential; draw the autocorrelation function and partial autocorrelation function graphs, observe the significance level in the graph, and select the appropriate autoregressive order p and moving average order q values; use the selected parameters (p, d, q) to fit the autoregressive integrated moving average model to the photovoltaic output data; The trained autoregressive integrated moving average model is used to generate the photovoltaic output forecast results in the short term in the future; the photovoltaic output data is normalized to the range of [0, 1] or [-1, 1], and the time series data is divided into multiple subsequences. The length of each subsequence is consistent with the time period of the data and the input requirements of the long short-term memory network model; Design the LSTM network structure, including the input layer, multiple hidden layers, and output layer; use grid search to optimize the hyperparameters of the LSTM network model; use the segmented subsequence data to train the LSTM network model, and use the Adam optimizer for gradient descent; use the trained LSTM network model to generate photovoltaic output forecast results in the long term in the future; Using a dynamic time window, the weights are dynamically adjusted according to the distance of the prediction time point; the short-term prediction results of the autoregressive integrated moving average model and the long-term prediction results of the long short-term memory network model are weighted averaged to generate the final photovoltaic output prediction results.

5. The distributed photovoltaic output power estimation method according to claim 4, characterized in that: The process of generating the final PV output forecast results includes the following steps: Design a time decay function to dynamically adjust the weight according to the distance of the prediction time point; the closer the time is, the higher the weight of the autoregressive integrated moving average model; the farther the time is, the higher the weight of the long short-term memory network model; According to the dynamic weight calculation results, the prediction results of the autoregressive integrated moving average model and the long short-term memory network are weighted averaged; The weighted average result is output as the final photovoltaic output prediction result; according to the feedback of actual photovoltaic output data, the attenuation coefficient and weight switching time point are dynamically adjusted.

6. The distributed photovoltaic output power estimation method according to claim 1, characterized in that: The process of matching the photovoltaic output with the load power includes the following steps: The intelligent coordination controller inputs the pre-processed data into the trained time series analysis model, performs real-time calculations based on the current load data and photovoltaic output data, and estimates the photovoltaic output power at the current moment; Based on the estimated photovoltaic output power and real-time load data, the intelligent coordination controller generates corresponding strategies and adjusts the operating parameters of the distributed photovoltaic system; specifically, it includes: Inverter output power adjustment: Based on the estimated photovoltaic output power, the intelligent coordination controller adjusts the inverter output power to ensure that the power generation capacity of the distributed photovoltaic system matches the power demand; Adjustment of energy storage system charging and discharging strategy: the intelligent coordination controller adjusts the energy storage system charging and discharging strategy according to the current load demand and photovoltaic output power; The intelligent coordination controller monitors the changes in photovoltaic output power and load data in real time, and makes feedback adjustments based on the changes, adjusting the output power of the inverter and the discharge strategy of the energy storage system.

7. The distributed photovoltaic output power estimation method according to claim 6, characterized in that: The process of the intelligent coordination controller adjusting the output power of the inverter includes the following steps: The intelligent coordination controller first merges the estimated PV output power with the real-time load data to form a comprehensive data set; analyzes the current power demand trend, identifies the peak and trough periods of load, and the fluctuation of PV output; Based on the preset strategy model, the intelligent coordination controller generates a preliminary strategy plan, including a power increase strategy during peak load period and a power reduction strategy during low load period; the preliminary strategy plan is optimized through particle swarm optimization; The intelligent coordination controller identifies the operating parameters of the inverter, including output voltage, frequency and power factor; maps the optimized strategy scheme to a specific parameter adjustment scheme based on the strategy model; and dynamically adjusts the parameters of the inverter based on real-time data; The intelligent coordination controller monitors the changes in load demand and photovoltaic output power in real time to obtain the latest operating status; according to the monitored changes in load demand, the intelligent coordination controller responds and adjusts the output power of the inverter.

8. The distributed photovoltaic output power estimation method according to claim 7, characterized in that: The process of mapping the optimized strategy plan to a specific parameter adjustment plan includes the following steps: The intelligent coordination controller breaks down the optimization strategy into multiple specific goals, including power boost, power reduction and frequency adjustment; Each target corresponds to the adjustment requirements of one or more inverter parameters; according to the operating parameters of the inverter, the targets in the strategy plan are further refined into specific requirements for the operating parameters; Extract load characteristics from specific requirements and generate corresponding parameter adjustment strategies; generate corresponding parameter adjustment strategies based on the characteristics of photovoltaic output; generate strategies for simultaneously adjusting output voltage and frequency when both load and photovoltaic output reach peak values ​​at the same time; generate parameter adjustment strategies based on the results of feature mapping.

9. The distributed photovoltaic output power estimation method according to claim 8, characterized in that: The process of further refining the goals in the strategy plan into specific requirements for operating parameters includes the following steps: Identify the specific goals in the optimization strategy, break down each goal into more detailed sub-goals, and ensure that each sub-goal corresponds to the specific inverter parameter adjustment requirements; Based on the inverter's operating parameters, including output voltage, frequency, power factor, and current, identify the specific parameters that need to be adjusted for each target; associate each target with the corresponding operating parameters; According to the operating parameters of the inverter, the objectives in the strategy plan are further refined into specific requirements for the operating parameters and the adjustment range of the inverter's output voltage and frequency.

Images