CN-121984239-A - Medium-voltage distributed light Fu Qun group control method, device, equipment and medium

CN121984239ACN 121984239 ACN121984239 ACN 121984239ACN-121984239-A

Abstract

The invention belongs to the technical field of distributed light Fu Qun group control, and particularly relates to a medium-voltage distributed light Fu Qun group control method, a medium-voltage distributed light Fu Qun group control device, medium-voltage distributed light Fu Qun group control equipment and medium. The method comprises the steps of constructing an ultra-short-term irradiation prediction model, predicting irradiation intensity change in a future preset period by combining photovoltaic array layout information, constructing a shadow recognition module, recognizing a shadow shielding area and shielding degree of a photovoltaic array in real time, calculating power loss caused by shadow, constructing a self-adaptive power adjustment algorithm based on an irradiation prediction value, the shadow power loss and a photovoltaic system output voltage acquired in real time, calculating a smoothed power adjustment instruction, and issuing the adjustment instruction to each photovoltaic substation by a group control center to realize cooperative control of a medium-voltage distributed photovoltaic cluster. According to the invention, through a prospective prediction and accurate identification technology, a power adjustment strategy is optimized, the control smoothness is improved, the service life of equipment is prolonged, and the stable operation of a power grid is ensured.

Inventors

XU YINGQIANG
TAN CHAO
FENG HAO
HUANG LING
CHEN DAN
ZHANG WENJUN
LENG LI
HAN DIE
Jiao Muhan
WANG XICHUN

Assignees

国网北京市电力公司

Dates

Publication Date: 20260505
Application Date: 20260205

Claims (10)

1. The group control method for the medium-voltage distributed light Fu Qun is characterized by comprising the following steps of: collecting real-time operation data, environment data and image data of each array of the photovoltaic cluster, and filtering and preprocessing the data by adopting a moving average method to obtain purified irradiation intensity, output power, output voltage and environment parameters; based on the preprocessed irradiation intensity data, predicting the future by adopting an LSTM neural network model Predicted values of irradiance intensity for each array over time Wherein ; Building a shadow recognition module based on machine vision, acquiring aerial images or ground monitoring images of a photovoltaic array in real time, and recognizing shadow shielding areas through an image segmentation algorithm Photoelectric conversion efficiency combined with photovoltaic array Real-time value of irradiation intensity Calculating the amount of power loss due to shadows ; Based on the irradiation intensity predicted value Amount of power loss Real-time output voltage deviation Constructing an adaptive power adjustment algorithm, and calculating the smoothed power adjustment quantity of each photovoltaic substation ; The group control center adjusts the power Converting into control instructions, transmitting the control instructions to each photovoltaic substation, performing power adjustment operation by the substation, collecting the adjusted output voltage and power data in real time, and feeding the data back to a group control center to form closed-loop control.
2. The medium voltage distributed optical Fu Qun group control method according to claim 1, wherein the training process of the LSTM ultra short term irradiation prediction model includes: normalizing the historical irradiation intensity data according to the formula Wherein the method comprises the steps of As the raw irradiation intensity data, a light source is provided, Respectively a maximum value and a minimum value in the historical data; setting the dimension of an input layer of an LSTM network as 3, and setting input parameters as Wherein Is that The ambient temperature at the moment, the number of hidden layers is 2-3, the number of neurons is 64-128, the dimension of the output layer is 1, and the output is Adopting an Adam optimizer, taking a mean square error MSE as a loss function, carrying out iterative training until a model converges, and carrying out inverse normalization on the output normalized predicted value to obtain an irradiation intensity predicted value 。
3. The medium voltage distributed optical Fu Qun group control method as claimed in claim 1, wherein the amount of power loss is The calculation formula of (2) is as follows: wherein, the light transmittance of the shadow area is 0-0.3, and when the shadow is not present =1, When fully occluded =0。
4. The medium voltage distributed optical Fu Qun group control method according to claim 1, wherein the expression of the adaptive power adjustment algorithm is: Wherein Is a coefficient of proportionality and is used for the control of the power supply, As an integral coefficient of the power supply, For the irradiation to predict the compensation coefficient, Compensating the coefficient for shading losses.
5. The method of medium voltage distributed optical Fu Qun group control as defined in claim 4, wherein, The power adjustment amount is provided with an adjustment amplitude limit threshold When (when) In the time-course of which the first and second contact surfaces, The value is 5% -10% of rated power of each photovoltaic substation.
6. The method of claim 1, wherein the feedback adjustment period of the closed-loop control is 1-5s, when the feedback output voltage deviation is the same When the feedback adjustment period is shortened to 0.5-1s, The value of the voltage deviation threshold value is +/-2% of the rated voltage of the medium-voltage power grid.
7. The medium voltage distributed optical Fu Qun group control method according to claim 1, wherein the image segmentation algorithm is a U-Net semantic segmentation algorithm, and the accurate identification of the shadow area is realized by marking a shadow area sample image of the photovoltaic array and training a U-Net model.
8. A medium voltage distributed optical Fu Qun group control device, comprising: the data acquisition module is used for acquiring real-time operation data, environment data and image data of each array of the photovoltaic cluster, and filtering and preprocessing the data by adopting a moving average method to obtain purified irradiation intensity, output power, output voltage and environment parameters; irradiance prediction module for predicting future by LSTM neural network model based on the preprocessed irradiation intensity data Predicted values of irradiance intensity for each array over time Wherein ; The power loss calculation module is used for building a shadow recognition module based on machine vision, acquiring aerial images or ground monitoring images of the photovoltaic array in real time, and recognizing shadow shielding areas through an image segmentation algorithm Photoelectric conversion efficiency combined with photovoltaic array Real-time value of irradiation intensity Calculating the amount of power loss due to shadows ; An adjustment module for predicting value based on the irradiation intensity Amount of power loss Real-time output voltage deviation Constructing an adaptive power adjustment algorithm, and calculating the smoothed power adjustment quantity of each photovoltaic substation ; An instruction module for the group control center to adjust the power Converting into control instructions, transmitting the control instructions to each photovoltaic substation, performing power adjustment operation by the substation, collecting the adjusted output voltage and power data in real time, and feeding the data back to a group control center to form closed-loop control.
9. An electronic device comprising a processor and a memory, the processor configured to execute a computer program stored in the memory to implement the medium voltage distributed optical Fu Qun group control method according to any one of claims 1 to 7.
10. A computer readable storage medium storing at least one instruction that when executed by a processor implements the medium voltage distributed optical Fu Qun group control method of any one of claims 1 to 7.

Description

Medium-voltage distributed light Fu Qun group control method, device, equipment and medium Technical Field The invention belongs to the technical field of distributed light Fu Qun group control, and particularly relates to a medium-voltage distributed light Fu Qun group control method, a medium-voltage distributed light Fu Qun group control device, medium-voltage distributed light Fu Qun group control equipment and medium. Background The distributed photovoltaic is a photovoltaic power generation facility which is built nearby a user site, operates in a mode of self-power-consumption and surplus electric quantity surfing on the user side, and is characterized by balance adjustment in a power distribution system. For distributed light Fu Qun group control, a prior art scheme with patent publication number CN118523483a is currently commonly used, which includes calculating a corresponding power adjustment amount in a distributed photovoltaic cluster by sampling output voltages of each distributed photovoltaic system in real time, thereby recognizing a corresponding offset voltage. In addition, to improve the group control function for distributed light Fu Qun, the output voltage of each distributed photovoltaic system sampled in real time is often used to estimate the abnormal operation of the distributed photovoltaic system. The control strategy in the prior art has the defect of adaptability to illumination mutation and shadow shielding, and the specific problem is that the existing method does not explicitly consider illumination rapid change, such as power dip and surge caused by cloud layer movement or local shadow shielding, which can cause frequent and large-scale adjustment of control instructions and aggravate equipment abrasion and power grid fluctuation. The present invention has been made in view of this. Disclosure of Invention The present invention is directed to a method, apparatus, device and medium for group control of medium voltage distributed light Fu Qun to solve or improve at least one of the problems in the prior art. In order to achieve the above purpose, the present invention adopts the following technical scheme: the invention provides a group control method for medium-voltage distributed light Fu Qun, which comprises the following steps: Step S1, collecting real-time operation data, environment data and image data of each array of a photovoltaic cluster, and carrying out filtering pretreatment on the data by adopting a moving average method to obtain purified irradiation intensity, output power, output voltage and environment parameters; S2, predicting the future by adopting an LSTM neural network model based on the preprocessed irradiation intensity data Predicted values of irradiance intensity for each array over timeWherein; S3, constructing a shadow recognition module based on machine vision, acquiring an aerial image or a ground monitoring image of the photovoltaic array in real time, and recognizing a shadow shielding region through an image segmentation algorithmPhotoelectric conversion efficiency combined with photovoltaic arrayReal-time value of irradiation intensityCalculating the amount of power loss due to shadows; Step S4, based on the irradiation intensity predicted valueAmount of power lossReal-time output voltage deviationConstructing an adaptive power adjustment algorithm, and calculating the smoothed power adjustment quantity of each photovoltaic substation; S5, the group control center adjusts the power adjustment quantityConverting into control instructions, transmitting the control instructions to each photovoltaic substation, performing power adjustment operation by the substation, collecting the adjusted output voltage and power data in real time, and feeding the data back to a group control center to form closed-loop control. By fusing various real-time data sources and adopting a moving average method for filtering, environmental noise and measurement interference are filtered, and a high-quality data basis is provided for subsequent control decisions. By introducing an LSTM neural network to predict future short-term irradiation intensity, the system can predictively sense the illumination change trend, and the power adjustment strategy is formulated in advance to reduce the severe fluctuation of the power instruction caused by illumination mutation. The shadow shielding area is identified by combining a machine vision technology, and the power loss is quantized, so that the control system can accurately distinguish global illumination change from local shielding, and a differentiated compensation strategy is implemented. Finally, the smoothed power adjustment quantity is calculated through a self-adaptive algorithm and closed-loop control is formed, so that coordination and stable regulation and control of the distributed photovoltaic clusters are realized, equipment abrasion and power fluctuation of a power grid are effectively relieved, and the adaptabil