Distributed photovoltaic cleaning demand determination method based on sand-dust-rainfall event sequence

Abstract

The present application relates to a kind of distributed photovoltaic cleaning demand determination method based on sand-dust-rainfall event sequence, the method is synchronous acquisition photovoltaic module to be monitored output power and meteorological parameter, automatically identify sand-dust and rainfall event, calculate sand-dust aftereffect recovery coefficient R, by with preset recovery threshold comparison, determine whether natural recovery is sufficient, only in recovery insufficient time trigger cleaning instruction.The present application will meteorological event sequence as natural experiment, without additional hardware, event-driven type is realized on-demand cleaning, significantly reduce operation and maintenance cost and water consumption, suitable for the cleaning decision of distributed photovoltaic power station in northwest arid region.

Description

Technical Field

[0001] This invention relates to the field of operation and maintenance technology for distributed photovoltaic systems, and in particular to a method for determining the clean energy needs of distributed photovoltaic systems based on dust-precipitation event sequences. Background Technology

[0002] Distributed photovoltaic power stations in the arid Northwest region have long faced the operational and maintenance decision-making challenge of "dust accumulation → decreased power generation efficiency → when to clean up." Existing technologies for determining cleaning needs mainly rely on two types of solutions: 1. Direct monitoring based on dust sensors: Optical dust sensors are installed on the surface or frame of the components, and a cleaning alarm is triggered when the dust accumulation exceeds a threshold. However, the sensors themselves are susceptible to failure due to dust cover, and can only reflect localized point dust accumulation, which cannot represent the cleanliness status of the entire array; for rooftop power stations that are already in operation and do not have reserved installation positions for sensors, the retrofit cost is high and the implementation is difficult.

[0003] 2. Preventative cleaning based on fixed time cycles: Cleaning is done weekly or monthly, without considering the actual degree of dust accumulation. This approach leads to delayed cleaning and significant power generation losses during seasons with frequent sandstorms, while causing over-cleaning and water waste during seasons with abundant rainfall. In the arid Northwest region, water resources are scarce, making fixed-cycle cleaning both economically and environmentally unfriendly.

[0004] CN121308663A proposes to calculate the power generation loss caused by dust accumulation by comparing the power generation of test components with that of reference components, and to determine the optimal cleaning cycle to maximize overall benefits. However, this method has the following limitations: First, it must rely on reference components (a clean control group), which is difficult to implement for distributed rooftop power stations that are already in operation and do not have reserved reference components; Second, it is based on an economic model and does not take into account the objective law that natural precipitation often follows dust storms in Northwest China, and still triggers cleaning according to the economic model during the rainy season, resulting in water waste; Third, it does not establish an automatic identification mechanism for dust storms and cannot distinguish between the two states of "natural recovery after dust storms" and "natural solidification of dust accumulation".

[0005] The common drawback of the aforementioned existing technologies is that they all fail to utilize the unique "dust storm-precipitation" meteorological event sequence specific to Northwest China. In fact, if a dust storm is followed by effective precipitation, dust accumulated on the component surface can be naturally washed away without human intervention; however, if there is a prolonged period without precipitation after a dust storm, the accumulated dust gradually solidifies naturally, necessitating manual cleaning. Existing technologies lack a systematic method for using meteorological event sequences as a "natural experiment" to determine cleaning needs, leading to cleaning decisions that either overly rely on hardware sensors or ignore natural recovery patterns, failing to meet the low-cost, high-efficiency operation and maintenance requirements of distributed photovoltaic systems in the arid Northwest region. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a low-cost and efficient method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences.

[0007] To address the aforementioned issues, the present invention provides a method for determining the cleaning needs of distributed photovoltaic systems based on a dust-precipitation event sequence. This method involves simultaneously collecting the output power and meteorological parameters of the photovoltaic modules to be monitored, automatically identifying dust and precipitation events, calculating the efficiency recovery coefficient R after the dust event, comparing it with a preset recovery threshold, determining whether the natural recovery is sufficient, and triggering a cleaning command only when the recovery is insufficient.

[0008] The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences, as described above, includes the following steps: (1) Deploy meteorological observation modules and power acquisition modules at the photovoltaic module array to be monitored to simultaneously acquire data on module DC output power, horizontal solar irradiance, environmental visibility, precipitation and wind speed; (2) Establish a sandstorm event recognition algorithm: Based on the combined characteristics of the sudden drop in the horizontal solar irradiance exceeding the preset irradiance threshold in a short period of time, accompanied by a decrease in environmental visibility and an increase in wind speed, the start and end times of the sandstorm event are automatically marked. (3) Establish a precipitation event identification algorithm: Based on the continuous period of precipitation exceeding the preset precipitation threshold, the start and end times of the precipitation event are automatically marked; (4) After the dust storm-precipitation event sequence ends, the normalized output power of the photovoltaic module during a preset sunny period before the dust storm event is extracted as a benchmark value. or pre The lowest normalized output power during the dust storm event was extracted as the dust storm attenuation value. or dust The normalized output power of the first preset sunny day period after the sequence ends is extracted as the recovery value. or post The normalized output power η = P / G, where P is the DC output power of the component and G is the horizontal solar irradiance. (5) Calculate the efficiency recovery coefficient R after dust storm: R = ( or post -or dust ) / ( or pre -or dust ); (6) Compare the recovery coefficient R with the preset recovery threshold and output a graded cleaning instruction: If R is greater than or equal to the first preset recovery threshold, it is determined that natural recovery is sufficient and no manual cleaning intervention is required; If R is less than or equal to the second preset recovery threshold and the duration after the sand-dust event ends exceeds the preset time threshold, it is determined that natural recovery is insufficient, and a cleaning instruction is triggered; If the second preset recovery threshold < R < the first preset recovery threshold, it is determined as a continuous attention state.

[0009] In step (1), the meteorological observation module includes a solar radiometer, a visibility meter, a rain gauge and a wind speed sensor.

[0010] In step (1), the power acquisition module is a DC power collector with a communication protocol, and its measuring range covers 0~120% of the rated power of the module.

[0011] In step (2), the preset irradiance threshold means that the horizontal solar irradiance drops by more than 30%~70% within a preset time window.

[0012] In the determination of the sand-dust event in step (2), it is also required that the ambient visibility is lower than the preset visibility threshold and the wind speed is higher than the preset wind speed threshold.

[0013] In step (4), the selection conditions for the preset sunny period are: the sky cloud cover rate does not exceed the preset cloud amount threshold, the horizontal solar irradiance is not lower than the preset sunny irradiance threshold, and the ambient wind speed is not higher than the preset sunny wind speed threshold.

[0014] In step (6), the first preset recovery threshold is 0.8~0.9, the second preset recovery threshold is 0.4~0.6, and the preset time threshold is 24~72 hours; the cleaning instruction includes at least one of manual cleaning, high-pressure water gun flushing or robot cleaning.

[0015] In step (6), when the continuous attention state is determined, a short-term power prediction correction step is further included: based on the sand-dust attenuation value or dust and the recovery value or post a sand-dust attenuation empirical model is established to perform attenuation correction prediction on the output power of the module within the next 24~72 hours.

[0016] Step (6) further includes a multi-module array partition determination step: when there are multiple photovoltaic module arrays in the monitoring area, the recovery coefficient R is calculated for each array respectively, and a sub-array cleaning demand map is output to realize differentiated operation and maintenance scheduling.

[0017] Compared with the prior art, the present invention has the following advantages: 1. Event-driven determination of cleaning demand is realized.

[0018] Existing technologies rely on fixed cycles or dust sensors, which cannot distinguish between "natural recovery after a sandstorm" and "dust accumulation and compaction." This invention uses the sandstorm-precipitation event sequence as a natural experiment, quantifies the degree of natural recovery through the recovery coefficient R, and avoids ineffective cleaning.

[0019] 2. Significantly reduces operation and maintenance costs and water consumption.

[0020] Water resources are scarce in the arid Northwest region, and traditional fixed-cycle cleaning results in significant waste during periods of abundant rainfall. This invention triggers cleaning only when natural recovery is insufficient (R is below a threshold) and there is a prolonged lack of rainfall, thus saving cleaning water.

[0021] 3. No additional hardware investment is required.

[0022] This invention utilizes existing weather stations and power acquisition data from power plants, eliminating the need for additional equipment such as dust sensors, making it particularly suitable for low-cost retrofitting of already operational power plants.

[0023] 4. It has the ability to predict and correct short-term power.

[0024] Following a sandstorm event, this invention can establish an empirical model of attenuation based on the recovery coefficient to correct and predict the power generation capacity for the next 1 to 3 days, thereby improving the accuracy of power prediction. Attached Figure Description

[0025] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0026] Figure 1 This is the overall flowchart of the present invention.

[0027] Figure 2 The flowchart for calculating the efficiency recovery coefficient R and determining cleaning requirements after a sandstorm is provided in the embodiments of the present invention.

[0028] Figure 3 Example 1 of the present invention provides a sandstorm-precipitation-recovery event sequence and recovery assessment diagram. Wherein: (a) is the daily change of normalized output power from March 10 to 16, 2026; (b) is a scatter plot of normalized power recovery assessment on March 16 (the first period after the sandstorm that meets the clear weather conditions).

[0029] Figure 4 These are comparative photographs of the surface state of the components provided in the embodiments of the present invention. Among them: (a) is the baseline state before the sandstorm; (b) is during severe sandstorm weather; (c) is the dust accumulation state on the surface of the components after the sandstorm event; and (d) is the clean state of the component surface after the precipitation event.

[0030] Figure 5 This is a conceptual diagram illustrating the derivation of Case 3, provided as an embodiment of the present invention. Detailed Implementation

[0031] The method for determining the cleaning demand of distributed photovoltaics based on the dust-precipitation event sequence refers to the method of simultaneously collecting the output power and meteorological parameters of the photovoltaic modules to be monitored, automatically identifying dust and precipitation events, calculating the efficiency recovery coefficient R after the dust event, and determining whether the natural recovery is sufficient by comparing it with the preset recovery threshold. The cleaning command is triggered only when the recovery is insufficient.

[0032] The specific process is as follows: Figure 1~2 As shown, it includes the following steps: (1) Deploy meteorological observation modules and power acquisition modules at the photovoltaic module array to be monitored to simultaneously acquire data on module DC output power, horizontal solar irradiance, environmental visibility, precipitation and wind speed.

[0033] The meteorological observation module includes a solar radiation meter (measuring horizontal solar irradiance G), a visibility meter (measuring visibility VIS), a rain gauge (measuring precipitation Pr), and a wind speed sensor (measuring ambient wind speed u).

[0034] The power acquisition module is a DC power acquisition device with a communication protocol, covering a range of 0~120% of the component's rated power, with an accuracy of ±0.5%, a sampling frequency of 1 Hz, and is connected to the data logger via an RS485 bus, with a storage interval of 5 minutes.

[0035] Environmental visibility data can also be obtained through the following methods: real-time collection using visibility meters deployed at the power plant site; access to publicly available visibility data released by the local meteorological bureau or environmental monitoring station (such as through meteorological data APIs or local weather forecasts); or simple estimation using grayscale threshold algorithms based on on-site camera images. For already operational distributed rooftop power plants, accessing publicly available meteorological bureau data is recommended, eliminating the need to purchase dedicated visibility meters and reducing deployment costs.

[0036] (2) Establish a sandstorm event recognition algorithm: Based on the combined characteristics of a sudden drop in horizontal solar irradiance exceeding a preset irradiance threshold within a short period, accompanied by decreased visibility and increased wind speed, the start and end times of a dust storm event are automatically marked. The preset irradiance threshold is defined as a drop in horizontal solar irradiance exceeding 30% to 70% within a preset time window. A dust storm event is also determined if visibility is below a preset visibility threshold and wind speed is above a preset wind speed threshold. The specific process is as follows: The data processing terminal monitors the horizontal solar irradiance G using a sliding window. When G drops by a percentage exceeding a preset irradiance threshold (e.g., 50%) within a preset time window (e.g., 30 minutes), and simultaneously the visibility VIS < preset visibility threshold (e.g., 1 km) and wind speed u > preset wind speed threshold (e.g., 2.5 m / s), the start time of the dust storm event is automatically marked. When G returns to the pre-dust level and VIS recovers, the end time of the dust storm event is marked.

[0037] (3) Establish a precipitation event identification algorithm: The start and end times of precipitation events are automatically marked based on consecutive periods when precipitation exceeds a preset precipitation threshold.

[0038] Precipitation events can overlap with dust storm events (dust storms accompanied by precipitation) or occur independently.

[0039] (4) After the dust storm-precipitation event sequence ends, extract the normalized output power of the photovoltaic modules during the preset clear weather period before the dust storm event as the baseline value. or pre The lowest normalized output power during the dust storm event was extracted as the dust storm attenuation value. or dust The normalized output power of the first preset sunny day period after the sequence ends is extracted as the recovery value. or post If precipitation occurs during or after a dust storm, then or post The normalized output power is the power output during the first preset sunny period after the precipitation event ends, reflecting the true recovery level of the component surface after natural erosion by precipitation.

[0040] Normalized output power η = P / G, where P is the DC output power of the module and G is the horizontal solar irradiance.

[0041] The selection criteria for the preset sunny period are: the cloud cover does not exceed the preset cloud cover threshold (cloud cover ≤ preset cloud cover threshold), the solar irradiance on the horizontal plane is not lower than the preset sunny irradiance threshold (G ≥ preset sunny irradiance threshold), and the ambient wind speed is not higher than the preset sunny wind speed threshold (u ≤ preset sunny wind speed threshold).

[0042] (5) Calculate the efficiency recovery coefficient R after dust storm: R = ( or post -or dust ) / ( or pre -or dust ).

[0043] (6) Compare the recovery coefficient R with a preset recovery threshold, and output a graded cleaning instruction: If R ≥ the first preset recovery threshold, it is determined that natural recovery is sufficient, and the dust accumulated on the surface of the module has been effectively removed by precipitation or strong wind, so no manual cleaning intervention is required. The first preset recovery threshold is 0.8~0.9, preferably 0.85.

[0044] If R ≤ the second preset recovery threshold and the duration after the end of the dust event (including the accompanying or subsequent precipitation recovery period) exceeds the preset time threshold, it is determined that natural recovery is insufficient, and a cleaning instruction is triggered. The second preset recovery threshold is 0.4~0.6, preferably 0.50; the preset time threshold is 24~72 hours, preferably 48 hours; the cleaning instruction includes at least one of manual cleaning, high-pressure water gun flushing or robot cleaning.

[0045] When the cleaning instruction is triggered, the data processing terminal pushes it to the operation and maintenance terminal through the Internet of Things module: ① Routine cleaning: 0.40 ≤ R ≤ 0.60, it is recommended to arrange manual cleaning within 3 days; ② Emergency cleaning: R < 0.40 and lasts for more than 8 hours, it is recommended to flush with a high-pressure water gun within 24 hours.

[0046] If the second preset recovery threshold < R < the first preset recovery threshold, it is determined as a continuous attention state, the system continues to monitor, and no cleaning is triggered for the time being.

[0047] When the state is determined as the continuous attention state, the method further includes a short-term power prediction correction step: based on the dust attenuation value or dust and the recovery value or post a dust attenuation empirical model is established to perform attenuation correction prediction on the output power of the module within the next 24 to 72 hours.

[0048] The present invention further includes a multi-module array partition determination step: when there are a plurality of photovoltaic module arrays in the monitoring area, the recovery coefficient R is calculated for each array respectively, and a cleaning demand map for each array is output, so as to realize differentiated operation and maintenance scheduling.

[0049] Embodiment A rooftop distributed photovoltaic power station in the central urban area of Lanzhou City is selected as the experimental site (35.87°N, 103.80°E), with an installed capacity of 2.7 kW, and the module type is CH-450WP bifacial monocrystalline silicon module. A Campbell CR1000X data logger is deployed beside the power station to synchronously collect the DC output power of the modules, horizontal plane solar irradiance, visibility, precipitation and wind speed, with a sampling interval of 5 minutes.

[0050] Reference calibration: March 10-11, 2026 was selected as the reference clear day, under preset clear day conditions (horizontal solar irradiance G ≥ 600 W / m²). 2 Under ambient wind speed u ≤ 1.5 m / s, the normalized output power reference value of the components was statistically obtained. or pre =2.725 m 2 (i.e. W·m) 2 The normalized power standard deviation σ = 0.1934, and the reference range [μ–1.5σ, μ+1.5σ] = [2.4349, 3.0151]. Inspection revealed no significant dust accumulation on the component surface during the reference period, and the power-irradiance linear relationship remained stable, confirming the validity of the reference.

[0051] Case 1: Sandstorm followed by effective precipitation, resulting in rapid and full recovery. Dust event identification: A severe dust storm occurred on March 12-13, 2026. Horizontal solar irradiance G increased from 800 W / m². 2 The temperature dropped sharply to 200 W / m 2 The following conditions were observed: a decrease in power exceeding 75%; ambient visibility <500 m; peak ambient wind speed reaching 3.2 m / s. The system automatically flagged this period as a sandstorm event. The lowest normalized output power of the components was recorded during the sandstorm event. or dust =0.0125 m 2 .

[0052] Recovery Assessment and Cleanliness Determination: A precipitation event occurred on March 14-15. The first period meeting the preset clear weather conditions after the dust storm ended was from 10:00 to 16:00 on March 16. During this period, there were 17 valid samples, and the normalized output power recovery value... or post = 2.398 m 2 Calculate the efficiency recovery coefficient after dust storm: R = ( or post -or dust ) / ( or pre -or dust = (2.398 - 0.0125) / (2.725 - 0.0125) ≈ 0.880 Since R=0.880 > the first preset recovery threshold of 0.85, the system determines that natural recovery is sufficient, and the dust on the component surface has been effectively washed away by rainfall, requiring no manual cleaning intervention. Figure 2~4 As shown.

[0053] Case 2: No precipitation after sand-dust event, gradual recovery after continuous monitoring A sand-dust weather event occurred on March 21, 2026, the minimum value of the normalized output power of the module or dust = 0.0061 m 2 . No effective precipitation occurred within 48 h after this sand-dust event, and the sand-dust persisted on March 22 (irradiance < 340 W / m 2 ), so the system could not obtain valid sunny day samples. On March 23, there were 15 valid samples in the first time period (10:00~16:00) that met the preset sunny day conditions, and the recovery value of the normalized output power or post = 2.027 m 2 . Calculate the recovery coefficient: R = ( or post -or dust ) / ( or pre -or dust ) = (2.027 - 0.0061) / (2.725 - 0.0061) ≈0.743 Since 0.50<R=0.743<0.85, the system determines a continuous monitoring state, does not trigger cleaning temporarily, and continues to monitor the power change in the subsequent time period. On March 24, the sunny day conditions are met again, the recovery value of the normalized output power or post = 2.540 m 2 , the calculated R≈ 0.932>0.85, the system updates the determination to sufficient natural recovery. This process shows that for short-term floating dust type sand-dust events, even without precipitation, 2~3 days of natural sedimentation and wind erosion can also achieve the recovery of module performance, but the recovery speed is slower than the precipitation-assisted recovery (Case 1).

[0054] Case 3: Long-term no precipitation after sand-dust event, dust accumulation caking, triggering cleaning (typical deduction) As shown Figure 5 , the complete event chain is demonstrated, which includes long-term no precipitation after sand-dust event leading to dust accumulation caking, triggering a cleaning instruction and verifying the cleaning effectiveness.

[0055] For comparison, based on the arid climate characteristics of northwestern China in the late spring and early summer, assume that no effective precipitation occurs within 72 h after the sand-dust event on May 18~19, 2026, and the recovery value of the normalized output power in the first time period (May 21) that meets the sunny day conditions or post =1.35 m 2 . Calculate the recovery coefficient: R = ( or post -or dust ) / ( or pre -or dust ) ≈ 0.493 Since R=0.493 < the second preset recovery threshold of 0.50, and the dust storm lasted for more than 72 hours after its end, the system determined that natural recovery was insufficient and the accumulated dust had solidified naturally, triggering a cleaning command. Maintenance personnel used a high-pressure water gun to clean the area on May 22nd. The normalized output power recovered to normal the following day (May 23rd). or clean =2.61 m 2 It falls within the reference range [2.4349, 3.0151], and is close to the reference value. or pre This verifies that the cleaning is effective.

[0056] The technical solution provided by this invention has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make several improvements and modifications to this invention without departing from its principles. For example, the recovery coefficient R can be combined with a machine learning algorithm to achieve adaptive threshold adjustment, or the method can be extended to determine the cleaning requirements of sand and dust deposition on wind turbine blades. These improvements and modifications also fall within the protection scope of the claims of this invention.

Claims

1. A method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences, characterized in that: This method involves simultaneously collecting the output power of the photovoltaic modules to be monitored and meteorological parameters, automatically identifying sandstorm and precipitation events, calculating the efficiency recovery coefficient R after the sandstorm, and determining whether the natural recovery is sufficient by comparing it with a preset recovery threshold. A cleaning command is only triggered when the recovery is insufficient.

2. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 1, comprising the following steps: (1) Deploy meteorological observation modules and power acquisition modules at the photovoltaic module array to be monitored to simultaneously acquire data on module DC output power, horizontal solar irradiance, environmental visibility, precipitation and wind speed; (2) Establish a sandstorm event recognition algorithm: Based on the combined characteristics of the sudden drop in the horizontal solar irradiance exceeding the preset irradiance threshold in a short period of time, accompanied by a decrease in environmental visibility and an increase in wind speed, the start and end times of the sandstorm event are automatically marked. (3) Establish a precipitation event identification algorithm: The start and end times of precipitation events are automatically marked based on the continuous period during which the precipitation exceeds a preset precipitation threshold. (4) After the dust storm-precipitation event sequence ends, extract the normalized output power of the photovoltaic module during a preset sunny period before the dust storm event as a benchmark value. η pre The lowest normalized output power during the dust storm event was extracted as the dust storm attenuation value. η dust The normalized output power of the first preset sunny day period after the end of the sequence is extracted as the recovery value. η post The normalized output power η = P / G, where P is the DC output power of the component and G is the horizontal solar irradiance. (5) Calculate the efficiency recovery coefficient R after dust storm: R = ( η post - η dust ) / ( η pre - η dust )； (6) Compare the recovery coefficient R with the preset recovery threshold and output a graded cleaning instruction: If R ≥ the first preset recovery threshold, then natural recovery is deemed sufficient and no manual cleaning intervention is required; If R ≤ the second preset recovery threshold and the duration after the sandstorm event ends exceeds the preset time threshold, then natural recovery is deemed insufficient, and a cleaning command is triggered. If the second preset recovery threshold < R < the first preset recovery threshold, then it is determined to be in a state of continuous monitoring.

3. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: The meteorological observation module in step (1) includes a solar radiation meter, a visibility meter, a rain gauge, and a wind speed sensor.

4. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: In step (1), the power acquisition module is a DC power acquisition device with a communication protocol, and its range covers 0 to 120% of the rated power of the component.

5. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: In step (2), the preset irradiance threshold is defined as the horizontal solar irradiance decreasing by more than 30% to 70% within a preset time window.

6. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: In step (2), the determination of a sandstorm event also requires that the environmental visibility be lower than a preset visibility threshold and the wind speed be higher than a preset wind speed threshold.

7. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: The selection criteria for the preset sunny time period in step (4) are: the cloud cover rate does not exceed the preset cloud cover threshold, the solar irradiance on the horizontal plane is not lower than the preset sunny irradiance threshold, and the ambient wind speed is not higher than the preset sunny wind speed threshold.

8. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: In step (6), the first preset recovery threshold is 0.8~0.9, the second preset recovery threshold is 0.4~0.6, and the preset time threshold is 24~72 hours; the cleaning instruction includes at least one of manual cleaning, high-pressure water gun rinsing, or robot cleaning.

9. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: When step (6) is determined to be in a state of continuous concern, a short-term power prediction correction step is also included: based on the dust attenuation value. η dust With recovery value η post An empirical model for dust attenuation was established to predict and correct the attenuation of component output power over the next 24 to 72 hours.

10. The method for determining the clean energy demand of distributed photovoltaic systems based on dust storm-precipitation event sequences as described in claim 2, characterized in that: Step 6 also includes a multi-module array partitioning determination step: when there are multiple photovoltaic module arrays in the monitoring area, the recovery coefficient R is calculated for each array, and a sub-array cleaning demand map is output to achieve differentiated operation and maintenance scheduling.

Images