Distributed photovoltaic output scheduling control method and system based on big data

Abstract

The invention relates to the technical field of photovoltaic output control, and discloses a distributed photovoltaic output scheduling control method and system based on big data, and the method comprises the steps: predicting the moving path of sand and dust according to meteorological remote sensing data and ground monitoring data, and carrying out the analysis of solar radiation attenuation, and obtaining the moving path of sand and dust and the radiation attenuation data; obtaining a photovoltaic output curve according to the distributed photovoltaic parameters, the sand and dust moving path and the radiation attenuation data; obtaining power load data according to the mining area load real-time data and the photovoltaic output curve; according to the mining area load prediction data and the photovoltaic output curve, the output load matching degree is calculated, and whether photovoltaic output scheduling control is triggered or not is judged; and generating an output scheduling strategy according to the power grid power deviation and the power grid frequency change curve. Based on big data and multi-source data fusion, through dynamic matching of photovoltaic power generation and loads, efficient management of a photovoltaic power generation system and stable operation of a power grid in sand and dust weather are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic output control, and particularly to a distributed photovoltaic output scheduling control method and system based on big data. Background Art

[0002] When building a photovoltaic power station in a gobi area, sandstorm weather will inevitably affect the power generation efficiency. During a sandstorm, the concentration of suspended substances in the atmosphere increases significantly. These suspended substances have selective attenuation characteristics for the solar radiation spectrum, and radiation in different bands is absorbed and scattered to different degrees, resulting in a significant reduction in the effective radiation received by the photovoltaic modules. This attenuation characteristic is closely related to the particle size, composition, and concentration of sand dust particles. However, existing models are difficult to accurately quantify this complex attenuation process. Especially during the movement of sand dust clusters, their concentration and distribution will change dynamically, further increasing the prediction difficulty.

[0003] In addition, when mining enterprises suspend production due to sandstorm weather, energy-intensive loads will suddenly drop. This sudden change in load will cause significant disturbances to the power grid frequency. Due to the intermittent and volatile nature of photovoltaic power generation itself, the reduction in power generation output and the sudden drop in load occur simultaneously during sandstorm weather, leading to further imbalance of the power grid frequency. Moreover, the movement trajectory of sand dust clusters is affected by various meteorological factors, and its prediction accuracy is limited. Therefore, how to achieve dynamic matching between photovoltaic power generation and load under special climate conditions has become a key technical challenge in the operation of gobi photovoltaic power stations. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a distributed photovoltaic output scheduling control method and system based on big data, which can solve the problems of inaccurate prediction of photovoltaic output, sudden drop in load, and power grid frequency fluctuation during sandstorm weather, achieve dynamic matching between photovoltaic power generation and load, and ensure the stable operation of the power grid.

[0005] In a first aspect, the present invention provides a distributed photovoltaic output scheduling control method based on big data, the method comprising: Predict the movement path of sand dust according to meteorological remote sensing data and ground monitoring data, and analyze the attenuation of solar radiation to obtain the sand dust movement path and radiation attenuation data; Predict the change in photovoltaic output according to the photovoltaic parameters of distributed photovoltaic, the sand dust movement path, and the radiation attenuation data to obtain a photovoltaic output curve; Analyze the disturbance of the sudden drop in load to the power grid frequency according to the real-time data of the mining area load and the photovoltaic output curve to obtain power load data, where the power load data includes mining area load prediction data, power grid power deviation, and power grid frequency change curve; Calculate the matching degree of power output and load according to the predicted data of the mining area load and the photovoltaic power output curve, and determine whether to trigger the scheduling control of photovoltaic power output according to the matching degree of power output and load; In response to triggering the scheduling control of photovoltaic power output, generate a power output scheduling strategy according to the power grid power deviation and the power grid frequency change curve, and perform scheduling control on the power output of distributed photovoltaic.

[0006] Further, the steps of predicting the movement path of sand and dust and analyzing the attenuation of solar radiation according to meteorological remote sensing data and ground monitoring data to obtain the movement path of sand and dust and radiation attenuation data include: Input the meteorological remote sensing data into a preset boundary extraction model to obtain sand and dust boundary feature data, and the boundary extraction model is constructed based on a convolutional neural network; Calculate the movement path of sand and dust by using the particle swarm optimization algorithm according to the wind direction and speed data, environmental temperature data, sand and dust concentration monitoring data and the sand and dust boundary feature data.

[0007] Further, the steps of predicting the movement path of sand and dust and analyzing the attenuation of solar radiation according to meteorological remote sensing data and ground monitoring data to obtain the movement path of sand and dust and radiation attenuation data further include: Calculate the median particle size of sand and dust according to the solar radiation monitoring data, and obtain the particle size distribution parameter according to the median particle size; Determine the extinction coefficient of sand and dust according to the particle size distribution parameter, and perform spatial distribution calculation on the sand and dust diffusion range by using Kriging interpolation method according to the sand and dust concentration monitoring data and the movement path of sand and dust to obtain sand and dust distribution data; Calculate the optical thickness according to the solar radiation wavelength, the sand and dust distribution data and the extinction coefficient; Input the optical thickness, the solar radiation wavelength and the particle size distribution parameter into a preset radiation attenuation model to obtain a radiation attenuation coefficient, and the radiation attenuation model is constructed based on a support vector regression model.

[0008] Further, the steps of predicting the change of photovoltaic power output according to the photovoltaic parameters of distributed photovoltaic, the movement path of sand and dust and the radiation attenuation data to obtain a photovoltaic power output curve include: Calculate the sand and dust radiation transmittance according to the optical thickness and the solar zenith angle; Correct the original intensity of solar radiation according to the radiation attenuation coefficient to obtain the solar radiation intensity; Correct the standard conversion efficiency according to the difference between the photovoltaic module temperature and the standard module temperature to obtain the photovoltaic module conversion efficiency; Obtain the photovoltaic output curve based on the solar radiation intensity, the conversion efficiency of the photovoltaic module, the area of the photovoltaic module, and the sand and dust radiation transmittance.

[0009] Further, the step of analyzing the disturbance of the sudden load drop to the grid frequency according to the real-time data of the mining area load and the photovoltaic output curve to obtain the power load data includes: Input the real-time data of the mining area load and the photovoltaic output curve into a preset load prediction model to obtain the sudden load drop amplitude, and the load prediction model is constructed based on a long short-term memory neural network; Calculate the predicted data of the mining area load according to the real-time data of the mining area load and the sudden load drop amplitude; Calculate the grid power deviation according to the predicted data of the mining area load and the photovoltaic output curve; Input the grid power deviation into the frequency dynamic equation to obtain the grid frequency change curve.

[0010] Further, the step of calculating the output-load matching degree according to the predicted data of the mining area load and the photovoltaic output curve, and judging whether to trigger the photovoltaic output dispatch control according to the output-load matching degree includes: Calculate the covariance between the predicted data of the mining area load and the photovoltaic output curve, and use the covariance as the output-load matching degree; Judge whether the output-load matching degree is less than the matching degree threshold. If so, trigger the photovoltaic output dispatch control.

[0011] Further, the step of generating an output dispatch strategy according to the grid power deviation and the grid frequency change curve to dispatch and control the output of distributed photovoltaics includes: Calculate the grid frequency deviation according to the grid frequency change curve, and use proportional-integral control according to the grid frequency deviation to calculate the initial power adjustment amount; Calculate the power deviation compensation according to the grid power deviation; Correct the initial power adjustment amount according to the power deviation compensation to obtain the power adjustment amount; Generate an output dispatch strategy for distributed photovoltaics through multi-machine coordination of distributed photovoltaics according to the power adjustment amount, and the multi-machine coordination includes equal-share coordination and priority coordination.

[0012] Further, the following formula is used to represent the initial power adjustment amount: In the formula, represents the initial power adjustment amount, represents the proportional gain coefficient, represents the integral gain coefficient, represents the grid frequency deviation, and t represents time; The power deviation compensation is represented by the following formula: In the formula, represents the power deviation compensation, represents the grid power deviation, represents the photovoltaic regulation efficiency coefficient; The power adjustment amount is represented by the following formula: In the formula, represents the power adjustment amount, represents the maximum adjustable capacity of the photovoltaic output.

[0013] Further, the step of generating the output scheduling strategy of the distributed photovoltaic through the multi-machine coordination of the distributed photovoltaic according to the power adjustment amount includes: When equal sharing coordination is adopted, the power adjustment amount is allocated to each distributed photovoltaic according to the capacity ratio of the distributed photovoltaic; When priority coordination is adopted, the priority of the distributed photovoltaic is determined according to the distance between the distributed photovoltaic and the mining area load center or the photovoltaic response speed; According to the priority, the distributed photovoltaic to be regulated is selected from the distributed photovoltaic, and the power adjustment amount is allocated to the distributed photovoltaic to be regulated.

[0014] In a second aspect, the present invention provides a distributed photovoltaic output scheduling control system based on big data, and the system includes: A data analysis module, configured to predict the movement path of the sand dust according to the meteorological remote sensing data and the ground monitoring data, and analyze the attenuation of the solar radiation to obtain the sand dust movement path and the radiation attenuation data; An output prediction module, configured to predict the change of the photovoltaic output according to the photovoltaic parameters of the distributed photovoltaic, the sand dust movement path and the radiation attenuation data, and obtain a photovoltaic output curve; A disturbance analysis module, configured to analyze the disturbance of the load sudden drop to the grid frequency according to the real-time data of the mining area load and the photovoltaic output curve, and obtain power load data, where the power load data includes mining area load prediction data, grid power deviation and grid frequency change curve; A scheduling trigger module, configured to calculate the output load matching degree according to the mining area load prediction data and the photovoltaic output curve, and determine whether to trigger the photovoltaic output scheduling control according to the output load matching degree; A scheduling control module, configured to respond to triggering of photovoltaic output scheduling control, generate an output scheduling strategy according to the grid power deviation and the grid frequency change curve, and perform scheduling control on the output of distributed photovoltaic.

[0015] The present invention provides a method and system for distributed photovoltaic output scheduling control based on big data. The present invention predicts the sand dust path through multi-data fusion, models the solar radiation attenuation by combining the physical optics model and the neural network model, effectively improves the accuracy of photovoltaic output prediction, analyzes the grid frequency disturbance caused by sudden load drop, improves the accuracy of mine area load and grid frequency prediction, and schedules the photovoltaic output according to the matching degree between the mine area load and the photovoltaic output, realizing the dynamic matching between photovoltaic power generation and load. The present invention is based on big data and multi-source data fusion, and realizes the efficient management of the photovoltaic power generation system and the stable operation of the power grid under sand dust weather through distributed photovoltaic output control and dynamic scheduling optimization. Description of the Drawings

[0016] Figure 1 is a schematic flow chart of the distributed photovoltaic output scheduling control method in an embodiment of the present invention; Figure 2 is a schematic structural diagram of the distributed photovoltaic output scheduling control system in an embodiment of the present invention. Detailed Embodiments

[0017] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Please refer to Figure 1 , a method for distributed photovoltaic output scheduling control based on big data proposed in the first embodiment of the present invention, which includes steps S10 to S50: Step S10, predicting the moving path of sand dust according to meteorological remote sensing data and ground monitoring data, and analyzing the solar radiation attenuation to obtain the sand dust moving path and radiation attenuation data; Step S20, predicting the change of photovoltaic output according to the photovoltaic parameters of distributed photovoltaic, the sand dust moving path and the radiation attenuation data to obtain a photovoltaic output curve; Step S30: Analyze the disturbance of the sudden load drop on the grid frequency based on the real-time data of the mining area load and the photovoltaic output curve to obtain power load data, where the power load data includes mining area load prediction data, grid power deviation, and grid frequency change curve; Step S40: Calculate the matching degree of output and load according to the mining area load prediction data and the photovoltaic output curve, and determine whether to trigger the photovoltaic output scheduling control according to the matching degree of output and load; Step S50: In response to triggering the photovoltaic output scheduling control, generate an output scheduling strategy according to the grid power deviation and the grid frequency change curve, and perform scheduling control on the output of distributed photovoltaics.

[0019] The present invention provides a photovoltaic output scheduling control method for the influence of sand and dust weather on the power generation efficiency of distributed photovoltaic power stations in the Gobi region. Since photovoltaics is directly related to the solar radiation intensity, the relationship between sand and dust weather and solar radiation is analyzed first. In this embodiment, the analysis includes two parts, namely, the prediction of the sand and dust movement path and the attenuation effect of sand and dust on solar radiation. Among them, the steps for predicting the sand and dust movement path include: Input meteorological remote sensing data into a preset boundary extraction model to obtain sand and dust boundary feature data, where the boundary extraction model is constructed based on a convolutional neural network; Calculate the sand and dust movement path using the particle swarm optimization algorithm according to the wind direction and speed data, ambient temperature data, sand and dust concentration monitoring data, and the sand and dust boundary feature data.

[0020] In this embodiment, first, meteorological remote sensing images are obtained from meteorological satellites, and the contrast of the remote sensing images is enhanced by the histogram equalization method. A boundary extraction model is pre-constructed using a convolutional neural network, and the equalized meteorological remote sensing data is input into the boundary extraction model to obtain a binary sand and dust mask. Sand and dust boundary feature data is obtained based on the binary sand and dust mask, where the sand and dust boundary feature data includes the area of the sand and dust cluster and the length of the boundary line, etc. Then, based on the sand and dust boundary feature data, combined with the wind direction and wind speed data and environmental temperature data collected by the meteorological detector, the particle swarm optimization algorithm is used to calculate the moving speed vector of the sand and dust, so as to obtain the sand and dust movement path. Among them, each particle represents a sand and dust microcluster, and the attributes include position, speed, and sand and dust concentration. Taking the minimization of the Euclidean distance between the predicted path and the measured data as the optimization goal, the particle swarm optimization algorithm PSO is used to calculate the moving speed vector of the sand and dust. When initializing PSO, the initial concentration weight of the particle swarm is adjusted according to the environmental temperature data. For example, the particle density in the high-temperature area is increased by 20%, the ground wind speed is used as the initial value of the particle horizontal speed, and the position is initialized based on the binary sand and dust mask. During the PSO speed calculation process, the wind direction data constrains the moving direction of the particle swarm. In addition, the vertical wind profile data can also be used to constrain the vertical speed component of the particle swarm. The position and speed of the particles are iteratively optimized, so as to output the optimal path set of the particle swarm. The specific calculation steps can refer to the iterative calculation steps of the conventional particle swarm algorithm, which will not be repeated here.

[0021] On the basis of obtaining the sand and dust movement path, the analysis steps for the attenuation of solar radiation by sand and dust include: According to the solar radiation monitoring data, calculate the median particle size of the sand and dust, and obtain the particle size distribution parameter according to the median particle size; According to the particle size distribution parameter, determine the extinction coefficient of the sand and dust, and use the Kriging interpolation method to perform spatial distribution calculation on the sand and dust diffusion range according to the sand and dust concentration monitoring data and the sand and dust movement path to obtain the sand and dust distribution data; According to the solar radiation wavelength, the sand and dust distribution data, and the extinction coefficient, calculate the optical thickness; Input the optical thickness, the solar radiation wavelength, and the particle size distribution parameter into a preset radiation attenuation model to obtain a radiation attenuation coefficient, and the radiation attenuation model is constructed based on a support vector regression model.

[0022] In this embodiment, for the current sand and dust weather, a sun photometer is used to directly measure the attenuation of solar radiation at different wavelengths through a multi-channel filter, and the optical thickness of aerosols corresponding to different bands can be directly obtained. The optical thicknesses at two wavelengths are selected, and then the Ångström exponent α is calculated. Preferably, since the visible light at 500 nm is sensitive to the scattering characteristics of sand and dust, and the near-infrared light at 1020 nm has strong penetrability and can represent the extinction dominated by coarse particles, therefore, the optical thicknesses at these two wavelengths are used to calculate the exponent α: In the formula, and represent the optical thicknesses corresponding to the wavelengths of 500 nm and 1020 nm, respectively.

[0023] An empirical formula based on the Ångström exponent and the median diameter is established in advance: In the formula, a, b, and c are coefficients respectively, α represents the Ångström exponent, and r med represents the median diameter.

[0024] The calculated exponent α is substituted into the empirical formula to estimate the median diameter r med . Since the sand and dust particle size follows a lognormal distribution, therefore, the median diameter is input into the lognormal distribution function n(r) to obtain the particle size distribution parameter r: In the formula, represents the geometric standard deviation, and this value is a preset fixed value.

[0025] According to the sand and dust movement path and the sand and dust concentration monitoring data, the Kriging interpolation method is used to calculate the spatial distribution of the diffusion range of the sand and dust group to obtain the sand and dust distribution data, that is, the concentration distribution data. Further, in Kriging interpolation, the wind direction and wind speed can also be used as covariates to correct the spatial distribution of the concentration field.

[0026] Align the coordinates of the sand and dust distribution data and the sand and dust movement path to ensure that the concentration value of each grid point corresponds to its movement direction. Then, according to the sand and dust concentration of each grid point, the optical thickness is calculated, and its formula is expressed as: In the formula, represents the optical thickness, represents the sand and dust concentration of the i-th grid point, represents the vertical thickness of the sand and dust layer at the i-th grid point, and this value can be estimated from the wind profile data in the meteorological prediction data, represents the extinction coefficient at the i-th grid point at the radiation wavelength λ.

[0027] Since the extinction coefficient is related to the particle size and the radiation wavelength, therefore, based on the particle size and the corresponding wavelength obtained from the above steps, the extinction coefficient can be calculated through the Mie theory, that is, the Mie scattering theory. , thereby obtaining the optical thickness.

[0028] Based on the optical principle, it is known that the optical thickness, the radiation wavelength, and the particle size distribution have a non-linear relationship with the attenuation of solar radiation. Therefore, in this embodiment, the vector regression method (SVR) is used to perform regression analysis on relevant historical data, thereby constructing a radiation attenuation model. Then, the optical thickness, the radiation wavelength, and the particle size distribution calculated above are input into the radiation attenuation model, and the radiation attenuation coefficient corresponding to different radiation wavelengths is output.

[0029] After obtaining the movement path of the sand dust and the data of the attenuation of solar radiation by the sand dust, based on these data and combined with the relevant parameters of the photovoltaic, the change of the photovoltaic output is analyzed. The specific steps include: Calculate the sand dust radiation transmittance according to the optical thickness and the solar zenith angle; Correct the original intensity of the solar radiation according to the radiation attenuation coefficient to obtain the solar radiation intensity; Correct the standard conversion efficiency according to the difference between the photovoltaic module temperature and the standard module temperature to obtain the photovoltaic module conversion efficiency; Obtain the photovoltaic output curve according to the solar radiation intensity, the photovoltaic module conversion efficiency, the photovoltaic module area, and the sand dust radiation transmittance.

[0030] In this embodiment, in the conventional formula of the photovoltaic output, the photovoltaic output is equal to the product of the solar radiation intensity, the photovoltaic module area, and the photovoltaic module conversion efficiency, that is: In the formula, G represents the solar radiation intensity, η represents the photovoltaic module conversion efficiency, and A represents the photovoltaic module area.

[0031] Since the attenuation of solar radiation caused by sand dust will affect the accuracy of the parameters related to solar radiation in the photovoltaic output formula, in order to improve the calculation accuracy of the photovoltaic output, it is also necessary to correct the photovoltaic output formula according to the radiation attenuation data calculated above.

[0032] First, according to the optical thickness and the solar zenith angle, based on the optical principle, calculate the radiation transmittance T: In the formula, represents the optical thickness, θ represents the solar zenith angle, and this angle can be calculated according to the geographical location and time.

[0033] Then, according to the difference between the surface temperature of the photovoltaic module and the standard module temperature under standard test conditions, the conversion efficiency of the photovoltaic module under standard test conditions is corrected to obtain the conversion efficiency of the photovoltaic module: In the formula, represents the conversion efficiency under standard test conditions, β represents the temperature coefficient, S PV represents the temperature of the photovoltaic module, S STC represents the standard module temperature.

[0034] In addition, dust will also cause attenuation of the solar radiation intensity. Therefore, it is also necessary to correct the original intensity of solar radiation according to the radiation attenuation coefficient. Specifically, the solar radiation is decomposed into multiple bands by a spectrometer to obtain the original radiation intensity of each band, and then the original intensity of solar radiation is corrected according to the attenuation coefficient of each band output by the attenuation model to obtain the attenuated radiation intensity: In the formula, represents the radiation attenuation coefficient output by the attenuation model, G(λ) represents the original intensity of solar radiation at wavelength λ, G eff (λ) represents the attenuated solar radiation intensity.

[0035] Since the photovoltaic module has different response efficiencies to different wavelengths, it is also necessary to weight and sum the effective radiation intensities of each band to obtain the total effective radiation G eff : In the formula, S(λ) represents the spectral response rate of the photovoltaic module.

[0036] Then, substitute the above correction values into the original photovoltaic output formula to obtain the corrected photovoltaic output P PV : According to the corrected photovoltaic output above, a photovoltaic output curve can be obtained.

[0037] In another preferred embodiment, the present invention also provides a method for correcting the original photovoltaic output, and the specific steps include: Input the solar radiation intensity, the temperature of the photovoltaic module, the optical thickness, and the solar zenith angle into a preset output correction model to obtain a dynamic correction factor, and the output correction model is constructed based on a recurrent neural network; Correct the original photovoltaic output according to the dynamic correction factor to obtain a photovoltaic output curve.

[0038] In this embodiment, since the theoretical formula of PV output assumes ideal conditions (steady-state irradiance, no environmental mutations), but there are actual dynamic interferences such as the dust diffusion rate and instantaneous cloud occlusion, a recurrent neural network model is used to model the complex relationship between solar radiation intensity, PV module temperature, optical thickness, solar zenith angle and PV output, so as to dynamically correct the solar radiation attenuation to make up for the static assumption defect of the theoretical model. Specifically, the time-series data of the actually measured solar radiation intensity, PV module temperature, optical thickness and solar zenith angle are input into the recurrent neural network model to predict the dynamic correction factor, and then the dynamic correction factor is multiplied by the PV output to achieve the correction of the PV output, so as to obtain the PV output curve. Here, the PV output can be the output calculated by the original formula or the corrected PV output in the above embodiment, that is, the PV output is corrected twice to further improve the accuracy of PV output prediction.

[0039] After predicting the PV output curve, according to the change of PV output and the real-time load of mining area enterprises, analyze the disturbance to the power grid frequency after the sudden drop of the mining area load caused by dust. The specific steps include: Input the real-time data of the mining area load and the PV output curve into a preset load prediction model to obtain the amplitude of the sudden load drop. The load prediction model is constructed based on a long short-term memory neural network; Calculate the predicted data of the mining area load according to the real-time data of the mining area load and the amplitude of the sudden load drop; Calculate the power grid power deviation according to the predicted data of the mining area load and the PV output curve; Input the power grid power deviation into the frequency dynamic equation to obtain the power grid frequency change curve.

[0040] In this embodiment, since the change of PV output is closely related to the weather, and the load of the mining area will be affected by the weather. For example, dust causes the mining area to stop production or partially stop production, resulting in a sudden drop in load. Therefore, a long short-term memory neural network model is used to model the relationship between the mining area load and the change of PV output. The real-time data of the mining area load and the predicted PV output curve are input into a preset load prediction model to output the amplitude of the sudden load drop and the corresponding mutation point timestamp, etc. Then, according to the difference between the real-time data of the mining area load and the sudden load drop, the predicted value of the mining area load can be obtained.

[0041] After obtaining the predicted data of the mining area load, subtract the PV output from the predicted value of the mining area load to get the power grid power deviation △P. If △P>0, it means a power deficit; otherwise, it means a power surplus. Then input the power grid power deviation into the frequency dynamic equation to obtain the frequency change rate of the power grid: Wherein, f represents frequency, t represents time, and H represents the system inertia constant.

[0042] Based on the rate of change of frequency, the frequency change curve of the power grid is obtained.

[0043] After obtaining power load data such as the mining area load prediction data, the power grid power deviation, and the power grid frequency change curve, based on these data, the photovoltaic output regulation determination is carried out. The specific determination steps include: Calculate the covariance between the mining area load prediction data and the photovoltaic output curve, and use the covariance as the output load matching degree. Judge whether the output load matching degree is less than the matching degree threshold. If so, trigger the photovoltaic output scheduling control.

[0044] In this embodiment, according to the correlation between the mining area load prediction data and the photovoltaic output curve, it is judged whether it is necessary to regulate the photovoltaic output. The correlation between the two is characterized by covariance. First, the data alignment and normalization processing are carried out on the mining area load prediction data and the photovoltaic output curve, and then the output load matching degree between the two is calculated through the covariance formula. And judge the size relationship between the output load matching degree and the preset matching degree threshold. If the output load matching degree is greater than or equal to the matching degree threshold, it is considered that the matching degree between the mining area load and the photovoltaic output is high, and there is no need to regulate the photovoltaic. Otherwise, it is considered that the matching degree between the two is low, and photovoltaic output scheduling is required.

[0045] When it is necessary to schedule the photovoltaic output, in this embodiment, through the power grid power deviation and the power grid frequency change curve, an output scheduling strategy is generated. The specific steps include: According to the power grid frequency change curve, calculate the power grid frequency deviation. According to the power grid frequency deviation, adopt proportional-integral control to calculate the initial power adjustment amount. According to the power grid power deviation, calculate the power deviation compensation. According to the power deviation compensation, correct the initial power adjustment amount to obtain the power adjustment amount. According to the power adjustment amount, through the multi-machine coordination of distributed photovoltaics, generate the output scheduling strategy of distributed photovoltaics. The multi-machine coordination includes equal distribution coordination and priority coordination.

[0046] In this embodiment, first, according to the difference between the power grid frequency change curve and the rated frequency value, the power grid frequency deviation Δf is obtained. Then, according to the power grid frequency deviation Δf, the power adjustment amount is calculated. Specifically, first, through proportional-integral control, that is, PI control, the initial power adjustment amount is calculated: Wherein, represents the initial power adjustment amount, represents the proportional gain coefficient, which is used to quickly respond to frequency changes, represents the integral gain coefficient, which is used to eliminate the steady-state error, represents the grid frequency deviation, and t represents time.

[0047] Then, based on the grid power deviation △P, the power deviation compensation is calculated: In the formula, represents the power deviation compensation, represents the grid power deviation, represents the photovoltaic regulation efficiency coefficient.

[0048] Finally, the initial power adjustment amount is corrected through the power deviation compensation, and with the maximum adjustable capacity of the photovoltaic output as the constraint condition, the final power adjustment amount is determined: In the formula, represents the power adjustment amount, represents the maximum adjustable capacity of the photovoltaic output.

[0049] Finally, according to the power adjustment amount, the output of the distributed photovoltaic is regulated. In this embodiment, a multi-machine coordination strategy is adopted to regulate the output of the distributed photovoltaic. Among them, the multi-machine coordination includes equal division coordination and priority coordination. Specifically, when equal division coordination is adopted, according to the capacity ratio of the distributed photovoltaic, the power adjustment amount is allocated to each photovoltaic to achieve output regulation; when priority coordination is adopted, according to the distance between the distributed photovoltaic and the mining area load center or the photovoltaic response speed, the priority of the distributed photovoltaic is determined. For example, the closer the distance between the photovoltaic and the load center or the higher the photovoltaic response speed, the higher the priority. Then, according to the priority of each photovoltaic, the photovoltaic to be regulated is selected from the distributed photovoltaic. For example, the photovoltaic with a priority greater than the priority threshold is selected as the photovoltaic to be regulated. Finally, the power adjustment amount is allocated to the photovoltaic to be regulated according to the capacity ratio of the photovoltaic to be regulated. It should be noted that after determining the power adjustment amount of the photovoltaic, the output scheduling control of the distributed photovoltaic can be realized through various regulation strategies. The output scheduling strategy in this embodiment is only a preference rather than a limitation.

[0050] A distributed photovoltaic power output scheduling control method based on big data provided by this embodiment. Through multi-data fusion, the present invention predicts the sand dust path and models the solar radiation attenuation by combining the physical optics model and the neural network model, effectively improving the accuracy of photovoltaic power output prediction. By analyzing the grid frequency disturbance caused by sudden load drop, the accuracy of mining area load and grid frequency prediction is improved, and the photovoltaic power output is scheduled according to the matching degree between the mining area load and the photovoltaic power output, realizing the dynamic matching of photovoltaic power generation and load. Based on big data and multi-source data fusion, through distributed photovoltaic power output control and dynamic scheduling optimization, the present invention realizes the efficient management of the photovoltaic power generation system and the stable operation of the power grid under sand dust weather.

[0051] Please refer to Figure 2 , based on the same inventive concept, a distributed photovoltaic power output scheduling control system proposed in the second embodiment of the present invention includes: A data analysis module 10, configured to predict the moving path of sand dust according to meteorological remote sensing data and ground monitoring data, and analyze the solar radiation attenuation, to obtain the sand dust moving path and radiation attenuation data; A power output prediction module 20, configured to predict the change of photovoltaic power output according to the photovoltaic parameters of the distributed photovoltaic, the sand dust moving path and the radiation attenuation data, to obtain a photovoltaic power output curve; A disturbance analysis module 30, configured to analyze the disturbance of the grid frequency caused by sudden load drop according to the real-time data of the mining area load and the photovoltaic power output curve, to obtain power load data, where the power load data includes mining area load prediction data, grid power deviation and grid frequency change curve; A scheduling trigger module 40, configured to calculate the matching degree of power output and load according to the mining area load prediction data and the photovoltaic power output curve, and determine whether to trigger the photovoltaic power output scheduling control according to the matching degree of power output and load; A scheduling control module 50, configured to respond to the triggered photovoltaic power output scheduling control, generate a power output scheduling strategy according to the grid power deviation and the grid frequency change curve, and schedule and control the power output of the distributed photovoltaic.

[0052] The technical features and technical effects of the distributed photovoltaic power output scheduling control system based on big data proposed in the embodiments of the present invention are the same as those of the method proposed in the embodiments of the present invention, and will not be elaborated here. Each module in the above-mentioned distributed photovoltaic power output scheduling control system based on big data can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in the processor in the computer device in the form of hardware or be independent of it, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0053] In summary, a distributed photovoltaic output scheduling control method and system proposed in an embodiment of the present invention. The method predicts the movement path of sand and dust by using meteorological remote sensing data and ground monitoring data, analyzes the attenuation of solar radiation, and obtains the movement path of sand and dust and radiation attenuation data; predicts the change of photovoltaic output according to the photovoltaic parameters of distributed photovoltaic, the movement path of sand and dust and the radiation attenuation data, and obtains a photovoltaic output curve; analyzes the disturbance of grid frequency caused by sudden load drop according to the real-time data of mine load and the photovoltaic output curve, and obtains power load data, where the power load data includes mine load prediction data, grid power deviation and grid frequency change curve; calculates the matching degree of output and load according to the mine load prediction data and the photovoltaic output curve, and determines whether to trigger the photovoltaic output scheduling control according to the matching degree of output and load; in response to triggering the photovoltaic output scheduling control, generates an output scheduling strategy according to the grid power deviation and the grid frequency change curve, and schedules and controls the output of distributed photovoltaic. The present invention predicts the sand and dust path through multi-data fusion, models the solar radiation attenuation by combining the physical optics model and the neural network model, effectively improves the accuracy of photovoltaic output prediction, analyzes the grid frequency disturbance caused by sudden load drop, improves the accuracy of mine load and grid frequency prediction, and schedules the photovoltaic output according to the matching degree between the mine load and the photovoltaic output, realizing the dynamic matching of photovoltaic power generation and load. The present invention is based on big data and multi-source data fusion, and realizes the efficient management of the photovoltaic power generation system and the stable operation of the power grid under sand and dust weather through distributed photovoltaic output control and dynamic scheduling optimization.

[0054] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0055] The above embodiments only represent several preferred embodiments of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the technical principle of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the protection scope of the claims.

Claims

1. A distributed photovoltaic output dispatching control method based on big data, characterized in that: include: Based on meteorological remote sensing data and ground monitoring data, the movement path of dust is predicted, and the attenuation of solar radiation is analyzed to obtain the movement path and radiation attenuation data of dust; According to the photovoltaic parameters of the distributed photovoltaic system, the dust movement path and the radiation attenuation data, the photovoltaic output change is predicted to obtain a photovoltaic output curve; According to the real-time data of the mining area load and the photovoltaic output curve, the disturbance of the load drop on the grid frequency is analyzed to obtain power load data, which includes the mining area load prediction data, the grid power deviation and the grid frequency change curve; Calculating the output load matching degree according to the mine area load forecast data and the photovoltaic output curve, and judging whether to trigger photovoltaic output dispatching control according to the output load matching degree; In response to triggering photovoltaic output dispatch control, an output dispatch strategy is generated according to the grid power deviation and the grid frequency change curve, and the output of distributed photovoltaic is dispatched and controlled.

2. The distributed photovoltaic output dispatching control method based on big data according to claim 1 is characterized in that: The step of predicting the movement path of dust according to meteorological remote sensing data and ground monitoring data, and analyzing solar radiation attenuation to obtain the movement path of dust and radiation attenuation data includes: The meteorological remote sensing data is input into a preset boundary extraction model to obtain dust boundary feature data, wherein the boundary extraction model is constructed based on a convolutional neural network; The dust movement path is calculated using a particle swarm optimization algorithm based on wind direction and speed data, ambient temperature data, dust concentration monitoring data, and the dust boundary feature data.

3. The distributed photovoltaic output dispatching control method based on big data according to claim 2 is characterized in that: The step of predicting the movement path of dust according to meteorological remote sensing data and ground monitoring data, and analyzing solar radiation attenuation to obtain the movement path of dust and radiation attenuation data also includes: Calculate the median particle size of the dust according to the solar radiation monitoring data, and obtain the particle size distribution parameter according to the median particle size; Determine the extinction coefficient of the dust according to the particle size distribution parameters, and calculate the spatial distribution of the dust diffusion range using the Kriging interpolation method according to the dust concentration monitoring data and the dust movement path to obtain dust distribution data; Calculating the optical thickness according to the solar radiation wavelength, the dust distribution data and the extinction coefficient; The optical thickness, the solar radiation wavelength and the particle size distribution parameters are input into a preset radiation attenuation model to obtain a radiation attenuation coefficient, and the radiation attenuation model is constructed based on a support vector regression model.

4. The distributed photovoltaic output dispatching control method based on big data according to claim 3 is characterized in that: The step of predicting the photovoltaic output change according to the photovoltaic parameters of the distributed photovoltaic, the dust movement path and the radiation attenuation data to obtain the photovoltaic output curve includes: Calculating the dust radiation transmittance according to the optical thickness and the solar zenith angle; Correcting the original solar radiation intensity according to the radiation attenuation coefficient to obtain the solar radiation intensity; According to the difference between the temperature of the photovoltaic module and the temperature of the standard module, the standard conversion efficiency is corrected to obtain the photovoltaic module conversion efficiency; A photovoltaic output curve is obtained according to the solar radiation intensity, the photovoltaic module conversion efficiency, the photovoltaic module area and the dust radiation transmittance.

5. The distributed photovoltaic output dispatching control method based on big data according to claim 1 is characterized in that: The step of analyzing the disturbance of the sudden drop in load on the grid frequency according to the real-time data of the mining area load and the photovoltaic output curve to obtain the power load data includes: Inputting the real-time data of the mine load and the photovoltaic output curve into a preset load prediction model to obtain the load drop amplitude, wherein the load prediction model is constructed based on a long short-term memory neural network; Calculate the mine area load forecast data based on the mine area load real-time data and the load sudden drop amplitude; Calculating the power deviation of the power grid according to the mine area load forecast data and the photovoltaic output curve; The power grid power deviation is input into the frequency dynamic equation to obtain a power grid frequency variation curve.

6. The distributed photovoltaic output dispatching control method based on big data according to claim 1 is characterized in that: The step of calculating the output load matching degree according to the mine area load prediction data and the photovoltaic output curve, and judging whether to trigger photovoltaic output dispatching control according to the output load matching degree includes: Calculating the covariance between the mine area load forecast data and the photovoltaic output curve, and using the covariance as the output load matching degree; It is determined whether the output load matching degree is less than a matching degree threshold, and if so, photovoltaic output dispatch control is triggered.

7. The distributed photovoltaic output dispatching control method based on big data according to claim 1 is characterized in that: The step of generating an output dispatching strategy according to the power grid power deviation and the power grid frequency change curve and dispatching and controlling the output of distributed photovoltaic power generation includes: Calculating the grid frequency deviation according to the grid frequency variation curve, and calculating the initial power adjustment amount by using proportional integral control according to the grid frequency deviation; Calculating power deviation compensation according to the power grid power deviation; Correcting the initial power adjustment amount according to the power deviation compensation to obtain a power adjustment amount; According to the power adjustment amount, an output dispatching strategy of distributed photovoltaic is generated through multi-machine coordination of distributed photovoltaic, wherein the multi-machine coordination includes equal distribution coordination and priority coordination.

8. The distributed photovoltaic output dispatching control method based on big data according to claim 7 is characterized in that: The initial power adjustment amount is expressed by the following formula: In the formula, Indicates the initial power adjustment amount, represents the proportional gain coefficient, represents the integral gain coefficient, represents the grid frequency deviation, and t represents the time; The power deviation compensation is expressed by the following formula: In the formula, Indicates power deviation compensation, Indicates the grid power deviation, Represents the photovoltaic regulation efficiency coefficient; The power adjustment amount is expressed by the following formula: In the formula, Indicates the power adjustment amount, Indicates the maximum adjustable capacity of photovoltaic output.

9. The distributed photovoltaic output dispatching control method based on big data according to claim 7 is characterized in that: The step of generating a distributed photovoltaic output dispatching strategy according to the power adjustment amount through multi-machine coordination of distributed photovoltaics includes: When equal distribution coordination is adopted, the power adjustment amount is distributed to each distributed photovoltaic according to the capacity ratio of the distributed photovoltaic; When priority coordination is adopted, the priority of distributed photovoltaic is determined according to the distance between distributed photovoltaic and the load center of the mining area or the photovoltaic response speed; According to the priority, a photovoltaic to be regulated is selected from the distributed photovoltaics, and the power adjustment amount is allocated to the photovoltaic to be regulated.

10. A distributed photovoltaic output dispatching and control system based on big data, characterized in that: include: The data analysis module is used to predict the movement path of sand and dust according to meteorological remote sensing data and ground monitoring data, and analyze the attenuation of solar radiation to obtain the movement path of sand and dust and radiation attenuation data; An output prediction module is used to predict the photovoltaic output change according to the photovoltaic parameters of the distributed photovoltaic, the dust movement path and the radiation attenuation data, and obtain a photovoltaic output curve; A disturbance analysis module is used to analyze the disturbance of the load drop on the grid frequency according to the real-time data of the mining area load and the photovoltaic output curve, and obtain power load data, wherein the power load data includes the mining area load prediction data, the grid power deviation and the grid frequency change curve; A scheduling trigger module, used to calculate the output load matching degree according to the mine area load forecast data and the photovoltaic output curve, and determine whether to trigger photovoltaic output scheduling control according to the output load matching degree; The dispatching control module is used to generate an output dispatching strategy in response to triggering photovoltaic output dispatching control according to the power grid power deviation and the power grid frequency change curve, and to dispatch and control the output of distributed photovoltaics.

Images