CN-116562355-B - Photovoltaic power station dust coverage determination method and device based on neural network

CN116562355BCN 116562355 BCN116562355 BCN 116562355BCN-116562355-B

Abstract

The application relates to a method and a device for determining dust coverage of a photovoltaic power station based on a neural network. The method comprises the steps of obtaining working condition data of a photovoltaic module at N moments on an M th day after dust cleaning of a new round of photovoltaic power station is completed, sending the working condition data of the photovoltaic module at the N moments to a second server, obtaining real-time output power predicted values of the power station at the N moments sent by the second server, obtaining real-time output power measured values of the power station at the N moments on the M th day, integrating the real-time output power predicted values of the power station at the N moments on the M th day to obtain a first value, integrating the real-time output power measured values of the power station at the N moments on the M th day to obtain a second value, and determining dust coverage of the photovoltaic power station on the M th day based on the first value and the second value. The application improves the accuracy of dust coverage calculation.

Inventors

ZHANG DU
HAN BIN
WANG ZHONGJIE
ZHAO YONG
XIE XIAOJUN
GAO PINGLIANG
WANG ZHAO

Assignees

西安热工研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20230330

Claims (8)

1. The dust coverage determining method of the photovoltaic power station based on the neural network is characterized by being applied to a first server, and comprises the following steps: acquiring working condition data of a photovoltaic module at N times on an M th day after dust cleaning of a new round of photovoltaic power station is completed, wherein the working condition data of the photovoltaic module comprises working temperature data of the photovoltaic module and irradiance data received by the photovoltaic module, M is an integer greater than 0, and N is an integer greater than 0; The N times of photovoltaic module working condition data are used for triggering the second server to determine the N times of power station real-time output power predicted values through a pre-trained power predicted model, wherein the power predicted model is a neural network model constructed according to N times of photovoltaic module working condition sample data in a preset time period; acquiring power station real-time output power predicted values of the N moments transmitted by the second server; acquiring real-time output power actual measurement values of the power station at the M th day and the N th time; Integrating the predicted values of the real-time output power of the power station at the N times on the M th day to obtain a first numerical value, and integrating the actual measured values of the real-time output power of the power station at the N times on the M th day to obtain a second numerical value; determining a dust coverage on day M of the photovoltaic power plant based on the first and second values; Before the working condition data of the photovoltaic module at the M th day after the dust cleaning of the new photovoltaic power station is completed is obtained, the method further comprises the steps of responding to the completion of the dust cleaning of the new photovoltaic power station, obtaining working condition sample data of the photovoltaic module at the N times in a preset time period according to preset frequency, wherein the working condition sample data of the photovoltaic module comprises working temperature sample data of the photovoltaic module, irradiance sample data received by the photovoltaic module and a real-time output power actual measurement value sample of the power station, sending the working condition sample data of the photovoltaic module at the N times in the preset time period to the second server, and triggering the second server to train the power prediction model based on the working condition sample data of the photovoltaic module at the N times in the preset time period, wherein the working condition sample data of the photovoltaic module at the N times in the preset time period is obtained according to the preset frequency; Before the working condition data of the photovoltaic module at the N times of the M th day after the dust cleaning of the new round of photovoltaic power station is completed is obtained, the method further comprises the steps of obtaining weather data of the photovoltaic power station, determining whether the M th day is a rainy day or not based on the weather data, determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station in response to the determination that the M th day is a rainy day, determining the M+1th day as a new M th day in response to the determination that the M+1th day is a rainy day, re-executing the steps of determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station in response to the determination that the M+1th day is a sunny day, determining the M+1th day as a new M th day, stopping executing the steps of determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station.
2. The method of claim 1, wherein the determining dust coverage on day M of the photovoltaic power plant based on the first value and the second value comprises: subtracting the first value from the second value to obtain a first difference; Dividing the first difference by the first numerical value to obtain dust coverage variation of the M day of the photovoltaic power station relative to the preset time period; Acquiring historical dust coverage, wherein the historical dust coverage is the dust coverage of the preset time period; And obtaining the dust coverage of the photovoltaic power station on the M day based on the historical dust coverage and the dust coverage variation.
3. The method of claim 1, wherein the N times within the predetermined time period are in one-to-one correspondence with the N times on the M day, and wherein the N times on the m+1th day are in one-to-one correspondence with the N times on the M day.
4. The dust coverage determining method of the photovoltaic power station based on the neural network is characterized by being applied to a second server, and comprises the following steps: receiving working condition data of a photovoltaic module at N times on an M th day sent by the first server, wherein the M th day is the M th day after dust cleaning of a new round of photovoltaic power station is completed, M is an integer greater than 0, and N is an integer greater than 0; The working condition data of the N moments on the M th day are input into a pre-trained power prediction model, and real-time power station output power predicted values of the N moments on the M th day output by the pre-trained power prediction model are obtained; The power station real-time output power predicted values at the N moments on the M th day are used for triggering the first server to determine dust coverage of the photovoltaic power station on the M th day based on the power station real-time output power predicted values at the N moments on the M th day and the power station real-time output power actual measurement values at the N moments on the M th day; Before receiving the photovoltaic module working condition data of the N times on the M th day sent by the first server, the method further comprises the following steps: Training a neural network model to be trained based on photovoltaic module working condition sample data sent by the first server to obtain a first power prediction model, wherein the photovoltaic module working condition sample data are photovoltaic module working condition sample data at N moments in a preset time period after a new round of dust cleaning of a photovoltaic power station is completed, and the photovoltaic module working condition sample data comprise working temperature sample data of the photovoltaic module, irradiance sample data received by the photovoltaic module and a real-time output power actual measurement value sample of the power station; Determining the first power prediction model as the pre-trained power prediction model.
5. A photovoltaic power plant dust coverage calculation device based on a neural network, characterized in that it is applied to a first server, said device comprising: The system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring working condition data of a photovoltaic module at N times on an M th day after dust cleaning of a new round of photovoltaic power station is completed, the working condition data of the photovoltaic module comprise working temperature data of the photovoltaic module and irradiance data received by the photovoltaic module, M is an integer greater than 0, and N is an integer greater than 0; The system comprises a first sending module, a second server, a power prediction model, a first power generation module, a second power generation module and a power generation module, wherein the first sending module is used for sending the working condition data of the photovoltaic module at the N moments to the second server; the second acquisition module is used for acquiring the power station real-time output power predicted values of the N moments sent by the second server; The third acquisition module is used for acquiring real-time output power actual measurement values of the power station at N times on the M th day; The integration module is used for integrating the predicted values of the real-time output power of the power station at the N times on the M th day to obtain a first numerical value, and integrating the actual measured values of the real-time output power of the power station at the N times on the M th day to obtain a second numerical value; A first determining module for determining a dust coverage of an mth day of the photovoltaic power plant based on the first value and the second value; Before the working condition data of the photovoltaic module at the M th day after the dust cleaning of the new photovoltaic power station is completed is obtained, the method further comprises the steps of responding to the completion of the dust cleaning of the new photovoltaic power station, obtaining working condition sample data of the photovoltaic module at the N times in a preset time period according to preset frequency, wherein the working condition sample data of the photovoltaic module comprises working temperature sample data of the photovoltaic module, irradiance sample data received by the photovoltaic module and a real-time output power actual measurement value sample of the power station, sending the working condition sample data of the photovoltaic module at the N times in the preset time period to the second server, and triggering the second server to train the power prediction model based on the working condition sample data of the photovoltaic module at the N times in the preset time period, wherein the working condition sample data of the photovoltaic module at the N times in the preset time period is obtained according to the preset frequency; Before the working condition data of the photovoltaic module at the N times of the M th day after the dust cleaning of the new round of photovoltaic power station is completed is obtained, the method further comprises the steps of obtaining weather data of the photovoltaic power station, determining whether the M th day is a rainy day or not based on the weather data, determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station in response to the determination that the M th day is a rainy day, determining the M+1th day as a new M th day in response to the determination that the M+1th day is a rainy day, re-executing the steps of determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station in response to the determination that the M+1th day is a sunny day, determining the M+1th day as a new M th day, stopping executing the steps of determining whether the M+1th day is a rainy day or not based on the weather data of the photovoltaic power station.
6. Photovoltaic power plant dust coverage calculation device based on neural network, characterized in that it is applied to the second server, said device comprises: The system comprises a first receiving module, a first server and a second server, wherein the first receiving module is used for receiving working condition data of a photovoltaic module at N moments on an M th day, the M th day is the M th day after dust cleaning of a new round of photovoltaic power station is completed, the M is an integer greater than 0, and the N is an integer greater than 0; the input module is used for inputting the working condition data of the N moments of the M th day into a pre-trained power prediction model; the acquisition module is used for acquiring power station real-time output power predicted values of N moments on the M th day output by the pre-trained power prediction model; the system comprises a first server, a sending module, a second server, a third server, a fourth server, a fifth server, a sixth server, a seventh server and a seventh server, wherein the sending module is used for sending the power station real-time output power predicted values of the N moments of the M th day to the first server; Before receiving the photovoltaic module working condition data of the N times on the M th day sent by the first server, the method further comprises the following steps: Training a neural network model to be trained based on photovoltaic module working condition sample data sent by the first server to obtain a first power prediction model, wherein the photovoltaic module working condition sample data are photovoltaic module working condition sample data at N moments in a preset time period after a new round of dust cleaning of a photovoltaic power station is completed, and the photovoltaic module working condition sample data comprise working temperature sample data of the photovoltaic module, irradiance sample data received by the photovoltaic module and a real-time output power actual measurement value sample of the power station; Determining the first power prediction model as the pre-trained power prediction model.
7. A storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform the neural network-based photovoltaic power plant dust coverage determination method of any one of claims 1 to 3, or to perform the neural network-based photovoltaic power plant dust coverage determination method of claim 4.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the neural network-based photovoltaic power plant dust coverage determination method of any one of claims 1 to 3 or performs the neural network-based photovoltaic power plant dust coverage determination method of claim 4 when executing the computer program.

Description

Photovoltaic power station dust coverage determination method and device based on neural network Technical Field The application relates to the technical field of photovoltaic power stations, in particular to a method and a device for determining dust coverage of a photovoltaic power station based on a neural network. Background In the related art, a photovoltaic power station can provide a group of template strings at any time. The template group string is cleaned by a cleaning robot or manually every day. And carrying out transverse comparison analysis on the generated energy or the real-time current integral value of the template string in a certain period and the generated energy or the real-time current integral value of the string in the same period influenced by dust coverage, calculating a difference value between the generated energy or the real-time current integral value and the generated energy or the real-time current integral value, and calculating the generated energy loss of the string due to dust coverage according to the difference value, thereby calculating the generated energy loss of the whole photovoltaic power station due to dust coverage. Because the sample plate strings need to be cleaned every day, the cleaning cost is relatively high by installing the automatic cleaning robot, and each cleaning robot can only be responsible for a very small amount of strings and is mainly limited by the arrangement and distribution conditions of the photovoltaic supports, the number of samples which can be provided by the photovoltaic power station is relatively small. Because of the inconsistency of the working states of the strings of the power station, a small number of template photovoltaic strings cannot describe the reference state that all strings of the power station are not covered by dust, so that the calculation of the generated energy loss of the strings due to dust coverage is inaccurate, and the timeliness of cleaning the strings is further affected. Disclosure of Invention Therefore, the application provides a method and a device for determining dust coverage of a photovoltaic power station based on a neural network. The technical scheme of the application is as follows: According to a first aspect of an embodiment of the present application, there is provided a method for determining dust coverage of a photovoltaic power station based on a neural network, applied to a first server, the method including: Acquiring working condition data of a photovoltaic module at N times on an M th day after dust cleaning of a new round of photovoltaic power station is completed, wherein the working condition data of the photovoltaic module comprises working temperature data of the photovoltaic module and irradiance data received by the photovoltaic module, M is an integer greater than 0, and N is an integer greater than 0; The N times of photovoltaic module working condition data are used for triggering the second server to determine the N times of power station real-time output power predicted values through a pre-trained power predicted model, wherein the power predicted model is a neural network model constructed according to N times of photovoltaic module working condition sample data in a preset time period; acquiring power station real-time output power predicted values of the N moments transmitted by the second server; acquiring real-time output power actual measurement values of the power station at the M th day and the N th time; Integrating the predicted values of the real-time output power of the power station at the N times on the M th day to obtain a first numerical value, and integrating the actual measured values of the real-time output power of the power station at the N times on the M th day to obtain a second numerical value; and determining the dust coverage of the photovoltaic power station on the M day based on the first numerical value and the second numerical value. According to one embodiment of the application, the determining the dust coverage on the mth day of the photovoltaic power plant based on the first and second values comprises: subtracting the first value from the second value to obtain a first difference; Dividing the first difference by the first numerical value to obtain dust coverage variation of the M day of the photovoltaic power station relative to the preset time period; Acquiring historical dust coverage, wherein the historical dust coverage is the dust coverage of the preset time period; And obtaining the dust coverage of the photovoltaic power station on the M day based on the historical dust coverage and the dust coverage variation. According to an embodiment of the present application, before the obtaining the working temperature data and irradiance data of the photovoltaic module at the nth time on the M th day after the dust cleaning of the new photovoltaic power station is completed, the method further includes: Responding to completion of dust cl