CN-121980397-A - Low-cost photovoltaic low-efficiency string identification analysis and inspection cleaning operation and maintenance method and device

CN121980397ACN 121980397 ACN121980397 ACN 121980397ACN-121980397-A

Abstract

A low-cost photovoltaic low-efficiency string identification analysis and inspection cleaning operation and maintenance method and device mainly comprise the steps of selecting a marker post string, analyzing the generating capacity of the marker post string, identifying abnormality of a non-marker post string, performing inspection cleaning linkage, optimizing a cleaning plan and the like. The method and the device realize accurate identification of short-term abnormality, long-term attenuation and electricity-limiting photovoltaic group strings by calculating daily relative power generation capacity and combining transverse sequencing and longitudinal slope analysis, carry out inspection on the group-series mobile unmanned aerial vehicle with short-term abnormality and individual component defects, avoid invalid operation and reduce operation and maintenance cost, optimize a cleaning plan by combining future power loss caused by dust coverage and weather prediction, dynamically adjust cleaning time or cancel cleaning, further reduce operation and maintenance cost and improve the overall power generation efficiency of the power station.

Inventors

SHEN ZHONGMING
HU KUN
WU XUEJIE
DENG XINYU
LI PEIHAN

Assignees

中电建新能源集团股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251218

Claims (7)

1. The low-cost photovoltaic low-efficiency string identification analysis and inspection cleaning operation and maintenance method is characterized by comprising the following steps of: s1, selecting a marker post group string, which is used for representing normal running states of different areas of the whole field; s2, carrying out generating capacity analysis on the selected marker post group string, wherein the generating capacity analysis comprises anomaly analysis, reference value calibration and relative generating capacity calculation; S3, identifying a non-standard pole group string with abnormal generated energy based on a set threshold value, and linking the unmanned aerial vehicle for inspection or group string cleaning according to the identification result; s4, optimizing a cleaning plan.
2. The method according to claim 1, wherein in the step S1, the specific method for selecting the string of the target group includes: Dividing three differentiation areas in east, west, south, middle and north according to the plane layout of the power station, and ensuring that the number of inverters covered by each area and the distribution density of strings are basically consistent; At least 1 inverter group string which has no history fault record, has the same assembly installation year and is far away from a shielding source is randomly selected from each area to be used as a target group string of the area.
3. The method according to claim 1 or 2, wherein in the step S1, the specific method for performing cleaning assurance on the string of targets comprises: The cleaning method comprises the steps of respectively providing low-cost cleaning robots adapting to photovoltaic module formats for each marker post group string, and setting high-frequency cleaning frequency, wherein the cleaning time is fixed to be in the early morning period of no-illumination and no-power generation operation; Meanwhile, a dust sensor is additionally arranged at the tail end of the cleaning robot, the coverage rate of dust on the surface of the component is automatically detected after each cleaning, the coverage rate is required to be less than or equal to a set threshold value, no stubborn dirt remains, the standard cleaning running state of the marker post group string is ensured all the time, and the generated energy is used as a reliable reference value for comparing the efficiency of the whole field group string.
4. The method according to claim 2, wherein in the step S2, the specific step of analyzing the power generation amount includes: the method comprises the steps of acquiring actual power generation amount Pi of a marker post group string and actual power generation amount Pi of other groups strings in the whole field every day through a power station SCADA system, wherein i=1, 2,..n, n is the total number of the groups strings in the whole field-the number of the marker post group strings, and acquiring weather data and maintenance record data in the same day; Judging whether the marker post group string is abnormal or not through cleaning state validity verification, generating capacity baseline deviation verification and multi-marker post cross verification, wherein if 2 or more marker post group strings in the region are abnormal, the method preferentially checks common interference factors by retrieving power station power grid monitoring data, SCADA system communication logs and marker post group string region images; The standard value calibration of the standard pole group string comprises the steps of taking an average value of the current day actual power generation of each standard pole group string as a standard value P0 if each standard pole group string is abnormal, taking the average value of the current day actual power generation of the rest normal standard pole group strings as the standard value P0 if only 1 area of standard pole group strings are abnormal, taking the calculated historical average power generation as a temporary standard value P0_temp if a historical standard value substitution scheme is started, and recalibrating the standard value after the standard pole group strings are recovered to be normal; calculating the relative power generation amount of each non-target rod group string: r_i=pi/P0, and if the temporary reference value p0_temp is used, r_i=pi/p0_temp.
5. The method according to claim 4, wherein the step S3 specifically includes: (1) Transverse contrast Sorting R_i of all non-standard pole group strings in the same day from large to small, screening out group strings with R_i < a set threshold value, and marking the group strings as 'group strings with lower generated energy'; When the number of the screened strings is small, marking the strings as 'single-day low-efficiency candidate strings'; (2) Longitudinal contrast Continuously collecting R_i data of 30 days for each non-standard pole group string, and constructing a daily relative power generation amount-time change curve; removing abnormal values in the curve by adopting a 3 sigma principle, namely calculating the average value mu and the standard deviation sigma of R_i for 30 days, and removing the numerical value of < mu-3 sigma or > mu+3 sigma; for the curve with abnormal values removed, calculating the descending slope k through linear regression, wherein k=delta R/delta t, delta R is the difference value of R_i in two adjacent days, delta t is the time interval, and the unit is day; sorting all group strings according to the absolute value of k from large to small, screening out group strings with |k| > set threshold, and marking the group strings as 'long-term attenuation candidate group strings'; (3) Electricity limiting identification From the group strings with lower generated energy, the group strings with reduced generated energy caused by the power grid electricity limiting are screened out, the abnormal candidate group string identities are eliminated, and the waste of invalid operation and maintenance cost is avoided: if a certain group of strings R_i is less than 0.9 and meets the requirements of 'power grid electricity limiting instruction exists on the same day' and 'no overhaul record exists in the group of strings', judging that the group of strings are 'electricity limiting group strings'; (4) Abnormal string type judgment If a certain group of strings are only marked as 'single-day low-efficiency candidate group strings' and are not limited in group strings, judging that the strings are 'short-term abnormal group strings', and supposing that the strings are temporary faults of components, and carrying out inspection on the unmanned aerial vehicle in a linkage way; If a group of strings are marked as a single-day low-efficiency candidate group string and a long-term attenuation candidate group string at the same time and are not limited by the group strings, further judging that only the group strings are abnormal if other group strings under the inverter are normal, judging that the group strings are individual component defect group strings, carrying out inspection on the linked unmanned aerial vehicle, and judging that the group strings are covered by regional dust if the group strings are abnormal, and carrying out a linked group string cleaning plan.
6. The method according to claim 5, wherein the step S4 specifically includes: (1) Electric quantity loss calculation For a zone dust cover group string, calculating theoretical electric quantity loss of 3 days in the future: Wherein R_imax takes the maximum value of R_i in 3 months of history, R_current is the average relative power generation amount R_i of the current regional group string, k is the kth day in 3 days in the future, and P0 mk is the power generation amount of the kth day of the mth marker post group string calculated according to the irradiance of weather forecast; (2) Setting a threshold value Setting an electric quantity loss threshold delta E0, and triggering a cleaning plan if delta E is more than or equal to delta E0; (3) Cleaning determination And (3) invoking weather forecast data to acquire weather data of 3 days in the future: a. If sand storm exists in the future 3 days, the cleaning is delayed until 1 day after the sand storm is ended; b. if heavy rain exists in the future 3 days, canceling the cleaning plan; c. if the weather is not particularly affected, the cleaning robot is immediately started to clean the regional dust covering group string.
7. A low cost photovoltaic inefficiency cluster recognition analysis and patrol cleaning operation device applying the method of any of claims 1-6, comprising: The marker post group string configuration module is used for selecting a marker post group string and carrying out cleaning guarantee on the marker post group string; The generating capacity analysis module is used for carrying out generating capacity analysis on the selected marker post group string and comprises anomaly analysis, reference value calibration and relative generating capacity calculation; The abnormal recognition and linkage module is used for recognizing a non-standard pole group string with abnormal generated energy based on a set threshold value and linking the unmanned aerial vehicle for inspection or group string cleaning according to a recognition result; and the cleaning plan optimization module is used for optimizing the cleaning plan.

Description

Low-cost photovoltaic low-efficiency string identification analysis and inspection cleaning operation and maintenance method and device Technical Field The invention relates to the technical field of photovoltaic operation and maintenance, in particular to a low-cost photovoltaic low-efficiency string identification analysis, inspection and cleaning operation and maintenance method and system. Background Currently, photovoltaic power stations are developing in a large-scale and intensive direction, and the power generation efficiency of a power station string directly determines the overall capacity of the power station. In order to improve the power generation efficiency of the photovoltaic string, the existing large-scale photovoltaic power station has widely adopted unmanned aerial vehicle to match with visible light, thermal infrared sensor and combine AI algorithm to identify the assembly defects such as assembly crack, hot spot, etc., the assembly surface pollution scheduling problem that dust, bird's droppings, fallen leaves, etc. caused, adopt semi-automatic cleaning vehicle and full-automatic cleaning robot to replace the manual work to wash the assembly to the assembly surface pollution. However, the unmanned aerial vehicle inspection and component cleaning frequency still faces the problems of standardized deficiency and high operation and maintenance cost, and along with the continuous upgrading of operation and maintenance requirements, how to accurately identify the low-efficiency group strings, reduce the operation and maintenance cost and avoid invalid inspection and cleaning operation becomes a core requirement in the operation and maintenance field of a large-scale photovoltaic power station, so that the transformation of the photovoltaic operation and maintenance technology from 'fixed period operation and maintenance' to 'data driving accurate operation and maintenance' is promoted. Disclosure of Invention The invention provides a low-cost photovoltaic low-efficiency string identification analysis, inspection and cleaning operation and maintenance method and system for a large photovoltaic power station, which are used for solving the problems of imperfect photovoltaic string identification analysis, lack of scientificity in inspection plans and cleaning plans, misjudgment of electricity limiting components and the like in the prior art. The low-cost photovoltaic low-efficiency group string identification analysis and inspection cleaning operation and maintenance method mainly comprises the following steps: s1, selecting a marker post group string, which is used for representing normal running states of different areas of the whole field; s2, carrying out generating capacity analysis on the selected marker post group string, wherein the generating capacity analysis comprises anomaly analysis, reference value calibration and relative generating capacity calculation; S3, identifying a non-standard pole group string with abnormal generated energy based on a set threshold value, and linking the unmanned aerial vehicle for inspection or group string cleaning according to the identification result; s4, optimizing a cleaning plan. Further, in the step S1, the specific method for selecting the marker post string includes: Dividing three differentiation areas in east, west, south, middle and north according to the plane layout of the power station, and ensuring that the number of inverters covered by each area and the distribution density of strings are basically consistent; At least 1 inverter group string which has no history fault record, has the same assembly installation year and is far away from a shielding source is randomly selected from each area to be used as a target group string of the area. Further, in the step S1, the specific method for cleaning and guaranteeing the string of the target rod group includes: The cleaning method comprises the steps of respectively providing low-cost cleaning robots adapting to photovoltaic module formats for each marker post group string, and setting high-frequency cleaning frequency, wherein the cleaning time is fixed to be in the early morning period of no-illumination and no-power generation operation; Meanwhile, a dust sensor is additionally arranged at the tail end of the cleaning robot, the coverage rate of dust on the surface of the component is automatically detected after each cleaning, the coverage rate is required to be less than or equal to a set threshold value, no stubborn dirt remains, the standard cleaning running state of the marker post group string is ensured all the time, and the generated energy is used as a reliable reference value for comparing the efficiency of the whole field group string. Further, in the step S2, the specific steps of the power generation amount analysis include: the method comprises the steps of acquiring actual power generation amount Pi of a marker post group string and actual power generation amount Pi of