A photovoltaic panel sun tracking control method and system

Abstract

The application provides a photovoltaic panel sun tracking control method and system, the method comprising: based on the geographical position and time information of a photovoltaic power station, calculating the reference theoretical position of the sun through an astronomical algorithm to generate a sun reference position sequence; acquiring local environment sensing data and external weather forecast data of the photovoltaic power station, predicting the geometric shading relationship of the cloud layer to the sun in the future period based on the data, and calculating the prediction curve of the solar irradiance reaching the photovoltaic panel position; constructing a feedforward optimization model and solving, outputting the optimized feedforward instruction of the photovoltaic panel posture in the future period; collecting the rapid light sensing deviation signal and the absolute attitude feedback signal of the photovoltaic panel in real time, and weighting and fusing with the feedforward control instruction to generate a motor driving instruction to drive the photovoltaic panel to move to track the sun; the method can improve the tracking accuracy of the photovoltaic panel to the sun, effectively improve the power generation efficiency, and also reduce the mechanical loss and movement risk, and guarantee the stable and reliable operation of the system.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic panel solar tracking control technology, and in particular to a photovoltaic panel solar tracking control method and system. Background Technology

[0002] Landslides, collapses, and debris flows are types of natural disasters that seriously threaten people's lives and property and the operational safety of major infrastructure. In order to achieve all-weather, automated monitoring and early warning of potential geological disaster sites, slope monitoring Internet of Things (IoT) systems consisting of GNSS displacement monitors, deep displacement gauges, pore water pressure gauges, rain gauges, and high-definition video cameras have been widely deployed in engineering practice. These monitoring devices are usually deployed in remote mountainous areas, mining slopes, and high slopes of highways and railways in unattended outdoor environments where there is a lack of mains power supply. Their long-term stable operation depends entirely on independent photovoltaic power supply systems. Photovoltaic power supply systems are often equipped with solar tracking devices to track the sun in real time to generate electricity.

[0003] Existing solar tracking methods for photovoltaic panels, whether astronomical algorithms, photoelectric tracking, or visual calibration, can only make post-hoc corrections to the current solar position deviation or environmental changes. They lack the ability to predict and proactively plan for future events such as cloud cover, sudden changes in irradiance, and strong winds. This results in the system always being in a state of lagging adjustment, unable to achieve global optimization between power generation revenue, equipment lifespan, and operational safety, thus reducing the system stability of photovoltaic power plants. Summary of the Invention

[0004] In view of this, the present invention proposes a solar tracking control method and system for photovoltaic panels, which can improve the tracking accuracy of photovoltaic panels to the sun, effectively improve power generation efficiency, and comprehensively consider various cost factors, reduce mechanical losses and motion risks, and ensure stable and reliable operation of the system.

[0005] The technical solution of the present invention is implemented as follows: In a first aspect, the present invention provides a photovoltaic panel solar tracking control method, comprising the following steps: S1, based on the geographical location and time information of the photovoltaic power station, calculates the reference theoretical position of the sun through astronomical algorithms, and calibrates the reference theoretical position by combining it with measured data to generate a solar reference position sequence; S2, acquire local environmental perception data and external weather forecast data of photovoltaic power station, predict the geometric shading relationship of cloud layer on the sun in the future period based on the data, and calculate the solar irradiance prediction curve reaching the photovoltaic panel location; S3 comprehensively predicts power generation revenue, mechanical loss cost, and motion risk cost, constructs a feedforward optimization model, takes the theoretical solar position sequence, solar irradiance prediction curve, and physical constraints of photovoltaic panel motion as input, solves the feedforward optimization model, and outputs the optimized feedforward command of photovoltaic panel attitude in the future period. S4 collects the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and performs weighted fusion with the feedforward control command to generate motor drive command, which drives the photovoltaic panel to track the sun.

[0006] Based on the above technical solutions, preferably, step S1, which involves calculating the reference theoretical position of the sun using astronomical algorithms based on the geographical location and time information of the photovoltaic power station, and calibrating the reference theoretical position using measured data to generate a solar reference position sequence, includes the following sub-steps: S11: Obtain the geographical longitude and latitude of the photovoltaic power station, and obtain the current Coordinated Universal Time (UTC). Based on the geographical longitude and current time zone information, convert the UTC to the local mean solar time. S12, determine the annual day corresponding to Coordinated Universal Time, calculate the ecliptic longitude of the sun based on the annual day using the celestial motion model, and calculate the solar declination angle at the current moment; S13, correct the local mean solar time for mean time difference and longitude time difference to obtain the local true solar time. Using noon as the reference, calculate the solar hour angle at the current time based on the local true solar time. S14 calculates the theoretical azimuth and theoretical altitude angles of the sun using spherical trigonometric formulas based on geographical latitude, solar declination angle, and solar hour angle. S15, preset light intensity threshold. If the current light intensity of the photovoltaic power station is greater than the light intensity threshold, then collect the solar image at the current moment, process the solar image at the current moment, and identify the actual center position of the sun. S16. Compare the angle corresponding to the actual center position of the graphic with the theoretical azimuth and theoretical altitude angle of the sun to obtain the azimuth calibration deviation and altitude calibration deviation. S17. Based on the azimuth calibration deviation and elevation calibration deviation, the theoretical elevation angle and theoretical azimuth angle are compensated and corrected to generate the solar reference position and obtain the solar reference position sequence.

[0007] Based on the above technical solutions, preferably, step S2, which involves acquiring local environmental sensing data and external weather forecast data of the photovoltaic power station, predicting the geometric shading relationship of clouds on the sun in the future based on the data, and calculating the solar irradiance prediction curve reaching the photovoltaic panel location, includes the following sub-steps: S21, acquire local environmental perception data and external weather forecast data of photovoltaic power station. The local environmental perception data includes time-series sky images and real-time wind speed and direction. The external weather forecast data includes cloud cover spatial distribution forecast and solar irradiance grid forecast information for the future target period. S22, process the time-series sky image, extract the cloud region in the image, and use optical flow method to analyze the pixel displacement of the cloud region between consecutive frames to calculate and generate a two-dimensional motion vector field; S23 maps the two-dimensional motion vector field to the sky spherical coordinate system to predict the movement trajectory of the cloud layer within a preset time window in the future. S24, compare the movement trajectory of the clouds with the theoretical position sequence of the sun in time and space to determine the geometric occlusion relationship of the clouds on the sun, wherein the geometric occlusion relationship is the occlusion state and the degree of occlusion; S25, based on geometric shading relationships and combined with clear-sky atmospheric radiation models and external meteorological forecast data, calculates and corrects the direct and diffuse solar irradiance components reaching the photovoltaic panel location, and outputs the predicted curves of normal direct irradiance and total irradiance of the inclined surface reaching the photovoltaic panel within the future time window.

[0008] Based on the above technical solutions, preferably, step S22, which involves processing the temporal sky image, extracting the cloud region from the image, and using optical flow to analyze the pixel displacement of the cloud region between consecutive frames to calculate and generate a two-dimensional motion vector field, includes the following sub-steps: S221, Denoising and contrast enhancement are performed on the time-series sky image to obtain a standard sky image; S222: Based on the brightness value and color features of pixels, a segmentation threshold is set for the standard sky image to distinguish the image pixels into cloud pixel regions and non-cloud pixel regions, and a cloud binary mask is generated. S223: Based on two consecutive frames of temporal sky images and a binary cloud mask, the optical flow algorithm is used to calculate the displacement of pixels in the cloud pixel region to obtain the initial motion vector field. S224. Outlier removal and spatial smoothing are performed on the initial motion vector field to obtain a filtered motion vector field. Vector clustering is performed on the filtered motion vector field to extract the dominant motion vector that represents the main motion trend of the cloud. Combined with real-time wind speed and direction data, physical constraint verification and correction are performed on the dominant motion vector to generate the final two-dimensional motion vector field.

[0009] Based on the above technical solution, preferably, step S24, which involves spatiotemporally comparing the cloud movement trajectory with the theoretical position sequence of the sun to determine the geometric occlusion relationship of the cloud on the sun, wherein the geometric occlusion relationship is the occlusion state and the degree of occlusion, includes the following sub-steps: S241 aligns the predicted cloud movement trajectory within the future time window with the solar reference position sequence within the same future time window on the time axis; S242, For each aligned time point, calculate the angular distance between the predicted position of the cloud feature point and the corresponding theoretical position of the sun in the sky spherical coordinate system using the spherical trigonometry formula; S243, preset the occlusion radius threshold, compare the angular distance with the occlusion radius threshold, if the angular distance is greater than the occlusion radius threshold, it is determined to be an unoccluded state, if the angular distance is less than or equal to the occlusion radius threshold, it is determined to be an occlusion state, and calculate the occlusion degree coefficient based on the angular distance and the occlusion radius threshold; S245 integrates the comparison results of each time point within the future time window and outputs a time series including occlusion status and occlusion degree coefficient as a geometric occlusion relationship.

[0010] Based on the above technical solution, preferably, step S25 includes the following sub-steps: S251, based on the solar theoretical position sequence, uses the clear-sky atmospheric radiation transfer model to calculate the theoretical normal direct irradiance and theoretical total irradiance of the horizontal plane reaching the location of the photovoltaic power station within a future time window under cloudless conditions. S252. Based on the geometric shading relationship, determine the corresponding direct irradiance attenuation factor, and multiply the theoretical normal direct irradiance by the direct irradiance attenuation factor to obtain the predicted normal direct irradiance. S253, based on cloud cover information in external meteorological forecast data, uses an anisotropic scattering sky model to correct the scattering component in the theoretical total irradiance of the horizontal plane, and obtains the predicted scattering irradiance. S254, based on the attitude angle of the photovoltaic panel within a future time window, combines the predicted normal direct irradiance and the predicted scattered irradiance into the predicted total irradiance of the inclined surface reaching the photovoltaic panel through inclined surface projection transformation, and outputs the predicted curve of normal direct irradiance and the predicted curve of total irradiance of the inclined surface.

[0011] Based on the above technical solutions, preferably, step S3 includes the following sub-steps: S31. Based on the predicted change curve of total irradiance on the inclined plane, calculate the total predicted power generation in the future prediction period as the power generation revenue item. Based on the change amount and rate of change of the photovoltaic panel attitude angle, calculate the mechanical energy loss of the photovoltaic panel as the mechanical loss item. Based on the predicted wind speed and real-time attitude angle, assess the wind load risk as the motion risk item. S32, the power generation revenue, mechanical loss and motion risk are weighted and combined, and the photovoltaic panel azimuth and elevation angles at each time in the future period are used as decision variables to construct a feedforward optimization model; S33, with the solar reference position sequence, the predicted curve of total irradiance on the inclined plane, and the attitude angle range, maximum angular velocity and maximum angular acceleration of the photovoltaic panel as constraints, solve the optimization function; combine the solved future azimuth angle optimization sequence and elevation angle optimization sequence of the photovoltaic panel as optimization feedforward instructions; S34. Based on the signal-to-noise ratio of local sensing data, historical prediction errors, and weather change rate, a weighted calculation is performed to obtain the comprehensive confidence level of the optimized feedforward command.

[0012] Based on the above technical solutions, preferably, step S4 includes the following sub-steps: S41 uses a four-quadrant photoelectric sensor to collect two-axis angle deviation signals in real time, a dual-axis tilt sensor to collect absolute attitude angle feedback in real time, and a vision sensor to collect solar visual position data in real time. S42, calculate the first confidence level of the four-quadrant photoelectric sensor based on signal strength and stability, calculate the second confidence level of the dual-axis tilt sensor based on self-test status and data refresh rate, and calculate the third confidence level of the visual sensor based on image clarity and solar feature recognition success rate. S43, based on the second and third confidence levels, the attitude angle feedback and the solar visual position data are weighted and fused to obtain the fused absolute attitude angle; S44, the optimized feedforward command is corrected in real time according to the two-axis angle deviation signal to obtain the corrected feedforward target. Based on the comprehensive confidence of the optimized feedforward command and the confidence of the fused absolute attitude angle, the corrected feedforward target and the fused absolute attitude angle are weighted and fused to generate the final target attitude angle. S45 calculates the tracking error between the final target attitude angle and the current fused absolute attitude angle, and uses a PID control algorithm based on the tracking error to calculate the motor drive command to drive the photovoltaic panel to track the sun.

[0013] Secondly, the present invention provides a photovoltaic panel solar tracking control system, implemented using a photovoltaic panel solar tracking control method, comprising: The reference calibration module is used to calculate the reference theoretical position of the sun based on the geographical location and time information of the photovoltaic power station through astronomical algorithms, and to calibrate the reference theoretical position by combining it with measured data to generate a solar reference position sequence. The environmental perception and prediction module is used to acquire local environmental perception data and external weather forecast data of the photovoltaic power station, predict the geometric shading relationship of clouds on the sun in the future period based on the data, and calculate the solar irradiance prediction curve reaching the photovoltaic panel location. The feedforward optimization decision module is used to comprehensively predict power generation revenue, mechanical loss cost and motion risk cost, construct a feedforward optimization model, take the solar theoretical position sequence, solar irradiance prediction curve and photovoltaic panel motion physical constraints as input, solve the feedforward optimization model, and output the optimized feedforward instruction of photovoltaic panel attitude in the future period. The fusion control module is used to collect the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and to perform weighted fusion with the feedforward control command to generate execution control command to drive the photovoltaic panel to track the movement of the sun.

[0014] Thirdly, the present invention provides a computer-readable storage medium storing a photovoltaic panel solar tracking control method and system program, wherein the photovoltaic panel solar tracking control method program is executed to implement the photovoltaic panel solar tracking control method.

[0015] The photovoltaic panel solar tracking control method and system of the present invention have the following advantages over the prior art: (1) By combining astronomical algorithms with measured data, the solar reference position sequence is accurately generated. The solar irradiance is predicted using environmental perception and meteorological forecast data. The feedforward optimization model is constructed to output optimized feedforward instructions. At the same time, the real-time collected signals are integrated to generate motor drive instructions. This can improve the tracking accuracy of the photovoltaic panel to the sun, effectively improve the power generation efficiency, and also take into account various cost factors, reduce mechanical losses and motion risks, and ensure the stable and reliable operation of the system. (2) By acquiring the geographical and temporal information of the photovoltaic power station, the theoretical position of the sun is calculated. Then, the measured light intensity is used to trigger image acquisition and processing, identify the actual position of the sun for comparison and calibration, and finally generate a solar reference position sequence. This effectively improves the accuracy of the sun position calculation and provides a reliable data foundation for the subsequent precise tracking of the sun by photovoltaic panels and improvement of power generation efficiency. (3) By integrating and processing multi-source data, cloud motion characteristics are extracted and trajectories are predicted. Then, the geometric shading relationship is determined by combining the theoretical position of the sun. Finally, the solar irradiance prediction curve is calculated by integrating multiple models and correction methods. This can more accurately predict the impact of cloud shading, effectively improve the accuracy of solar irradiance prediction, provide a reliable basis for photovoltaic panel tracking control, and thus improve the efficiency and stability of photovoltaic power generation. (4) Taking into account power generation revenue, mechanical losses and motion risks, a feedforward optimization model is constructed by calculating and weighting various factors. The photovoltaic panel attitude optimization command is solved by combining multiple constraints. The comprehensive confidence level is also calculated based on multiple factors. This can scientifically balance the relationship between power generation, losses and risks, improve the rationality and accuracy of photovoltaic panel tracking control, effectively improve power generation efficiency, and ensure the stable and reliable operation of the system. (5) Multi-dimensional data are collected in real time by multiple sensors, and the confidence of each sensor is calculated separately to perform data weighted fusion to obtain a more accurate fused absolute attitude angle. At the same time, the feedforward command is corrected and fused in combination with the comprehensive confidence to generate the final target attitude angle. The motor drive command is calculated using the PID algorithm, which significantly improves the accuracy and stability of solar tracking by the photovoltaic panel and effectively improves the photovoltaic power generation efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the photovoltaic panel solar tracking control method of the present invention; Figure 2 This is a block diagram of the photovoltaic panel solar tracking control system of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, in a first aspect, a photovoltaic panel solar tracking control method of the present invention includes the following steps: S1, based on the geographical location and time information of the photovoltaic power station, calculates the reference theoretical position of the sun through astronomical algorithms, and calibrates the reference theoretical position by combining it with measured data to generate a solar reference position sequence; It should be noted that step S1 is used to establish a high-precision spatial reference for the entire tracking system. This step calculates the theoretical apparent position of the sun in the sky using astronomical algorithms and uses visual sensors to perform on-site calibration of systematic deviations caused by mechanical installation errors, structural deformation, etc., generating a series of continuous and accurate solar reference position sequences, providing a reliable absolute reference for subsequent feedforward prediction and feedback control.

[0020] Step S1 includes the following sub-steps: S11: Obtain the geographical longitude and latitude of the photovoltaic power station, and obtain the current Coordinated Universal Time (UTC). Based on the geographical longitude and current time zone information, convert the UTC to the local mean solar time. The system obtains the geographical longitude and latitude of the photovoltaic power station location through a GNSS module or pre-set power station archive information. Simultaneously, the system synchronously acquires the current Coordinated Universal Time (UTC) via a built-in real-time clock chip or network time protocol, with a recording accuracy of at least the second level. Based on the time zone information of the photovoltaic power station's geographical location, the UTC is converted to local mean solar time. The conversion expression is: LMT = UTC + Ts + λ × 4 min / °; In the formula, LMT is the local mean solar time, UTC is the Coordinated Universal Time, Ts is the time zone offset in hours, λ is the geographic longitude, and λ×4min / ° is used to correct the prime meridian time to the time corresponding to the local longitude.

[0021] S12, determine the annual day corresponding to Coordinated Universal Time, calculate the ecliptic longitude of the sun based on the annual day using the celestial motion model, and calculate the solar declination angle at the current moment; It should be noted that, based on the accumulated days of the year, the standard celestial motion model of the Wang Guolin formula is used to calculate the ecliptic longitude of the sun, and then the solar declination angle at the current moment is calculated. As a preferred embodiment, the solar declination angle can be calculated using the following simplified formula: d =0.006918-0.399912×cos(Γ)+0.070257×sin(Γ)-0.006758×cos(2Γ)+0.000907×sin(2Γ)-0.002697×cos(3Γ)+0.001480×sin(3Γ); In the formula, d Γ is the solar declination angle, Γ is the annual angle, Γ=2π×(DOY-1) / 365, and DOY is the yearly accumulated day corresponding to Coordinated Universal Time.

[0022] S13, correct the local mean solar time for mean time difference and longitude time difference to obtain the local true solar time. Using noon as the reference, calculate the solar hour angle at the current time based on the local true solar time. It should be noted that local mean solar time is corrected for mean time difference and longitude time difference to eliminate the difference between true solar time and mean solar time caused by Earth's orbital eccentricity and obliquity of the ecliptic. The expression for calculating the mean time difference is as follows: EOT=229.18×(0.000075+0.001868×cos(Γ)-0.032077×sin(Γ)-0.014615×cos(2Γ)-0.04089×sin(2Γ)); In the formula, EOT is the mean time difference.

[0023] Adding the mean solar time to the local mean solar time gives the local true solar time. Using noon as the reference, the current time is converted into solar hour angles, expressed as follows: ω=(L h +L m / 60+L s / 3600-12)×15°; In the formula, ω is the solar hour angle, and L h L is the local true solar time in hours. m L represents minutes of local true solar time. s The number of seconds in local true solar time; S14 calculates the theoretical azimuth and theoretical altitude angles of the sun using spherical trigonometric formulas based on geographical latitude, solar declination angle, and solar hour angle. The theoretical elevation angle expression in this embodiment is: sinα=sinφ×sinδ+cosφ×cosδ×cosω; α = arcsin(sinα); In the formula, α is the theoretical altitude angle, φ is the geographical latitude, δ is the solar declination angle, and ω is the solar hour angle; The theoretical azimuth expression in this embodiment is: sinγ = cosδ × sinω / cosα; cosγ=(sinα×sinφ-sinδ) / (cosα×cosφ); γ = atan2(sinγ, cosγ); In the formula, γ is the theoretical azimuth angle; Due to atmospheric refraction, the apparent position of the sun will be slightly higher than its theoretical position. To improve the consistency between theoretical calculations and actual observations, the theoretical elevation angle needs to be corrected for atmospheric refraction, as expressed by: α r =α+R; R=1.02 / tan(α+10.3 / (α+5.11)) (α≥10°) R=0.1594+0.0196α+0.00002α² (α<10°) In the formula, α r R is the corrected theoretical altitude angle of the sun, and R is the atmospheric refraction correction.

[0024] S15, preset light intensity threshold. If the current light intensity of the photovoltaic power station is greater than the light intensity threshold, then collect the solar image at the current moment, process the solar image at the current moment, and identify the actual center position of the sun. It should be noted that this step is used to eliminate systematic deviations caused by mechanical installation errors, long-term structural deformation, etc. The system presets a light intensity threshold to ensure that calibration is performed under clear conditions where the sun's glare is clearly visible. When the irradiance sensor installed at the photovoltaic power station detects that the total irradiance on the current horizontal plane is greater than the preset threshold, the system automatically triggers the calibration process. At this time, the vision sensor takes a picture of the sky and acquires a solar image containing the solar disk. The acquired solar image is first preprocessed, including grayscale conversion, filtering and noise reduction, and contrast stretching to enhance the solar edge features. Subsequently, an edge detection image processing algorithm is used to identify the precise position of the sun in the image and calculate its actual image center coordinates.

[0025] S16. Compare the angle corresponding to the actual center position of the graphic with the theoretical azimuth and theoretical altitude angle of the sun to obtain the azimuth calibration deviation and altitude calibration deviation. It should be noted that, by using the calibration parameters of the visual sensor, the coordinates of the sun's image center are mapped to the sky spherical coordinate system defined by azimuth and altitude angles, thus obtaining the visually measured azimuth and altitude angles. The visually measured azimuth and altitude angles are then compared with the corrected theoretical azimuth and altitude angles of the sun, respectively, and the calibration deviations are calculated to obtain the azimuth calibration deviation and altitude calibration deviation. These two sets of deviation values ​​comprehensively reflect the fixed installation error, zero-position offset, and possible non-orthogonal errors of the system on the azimuth and altitude axes.

[0026] S17. Based on the azimuth calibration deviation and elevation calibration deviation, the theoretical elevation angle and theoretical azimuth angle are compensated and corrected to generate the solar reference position and obtain the solar reference position sequence.

[0027] It should be noted that the azimuth angle calibration deviation and elevation angle calibration deviation are used as fixed calibration parameters for this photovoltaic power station. The system is repeatedly executed at a fixed time step to generate a series of continuous and smooth solar reference position data points corresponding to the time axis. By performing linear interpolation on these discrete data points, the solar reference position sequence is finally formed.

[0028] In addition, the system can be configured with a periodic calibration strategy or, when long-term tracking deviations are detected, steps S15 and S16 can be re-executed to update and iterate the calibration parameters in order to compensate for system parameter drift caused by factors such as structural aging and maintenance disturbances, and to ensure the long-term accuracy and stability of the reference position sequence.

[0029] It should be noted that by utilizing the latitude, longitude, and time information of the power station and calculating the theoretical apparent position of the sun based on a celestial motion model, the physical accuracy and all-weather availability of the reference were ensured. Furthermore, under clear conditions, the visual sensor was triggered to measure the sun's position, decoupling and compensating for systematic deviations such as mechanical installation errors, structural deformation, and zero-position offset. The calibration parameters were then solidified for real-time correction of the theoretical position. Through a periodic calibration mechanism and interpolation processing, a continuous, smooth, and long-term stable sequence of solar reference positions was finally generated. This solved the problem of accumulated errors caused by the lack of calibration in pure astronomical algorithms and the defect of pure sensor tracking being prone to failure on cloudy days. It provided accurate data for subsequent feedforward prediction and feedback control, improving the accuracy baseline, environmental adaptability, and long-term operational reliability of the entire tracking and control system.

[0030] S2 acquires local environmental perception data and external weather forecast data of the photovoltaic power station, predicts the geometric shading relationship of clouds on the sun in the future period based on the data, and calculates the solar irradiance prediction curve reaching the photovoltaic panel location.

[0031] It should be noted that step S2 is used to build the forward-looking predictive capability of the entire tracking system. This step achieves short-term prediction of cloud movement by integrating local visual perception data with external weather forecast information, and on this basis, establishes a geometric shading relationship model of the clouds on the sun, and finally quantifies the shading relationship into a predicted irradiance curve reaching the tilted surface of the photovoltaic panel.

[0032] Step S2 includes the following sub-steps: S21, acquire local environmental perception data and external weather forecast data of photovoltaic power station. The local environmental perception data includes time-series sky images and real-time wind speed and direction. The external weather forecast data includes cloud cover spatial distribution forecast and solar irradiance grid forecast information for the future target period. It should be noted that the temporal sky images were acquired by an all-sky imager deployed at the photovoltaic power station site. This device is equipped with a fisheye lens and continuously captures 180° field-of-view sky images at a fixed frequency, outputting RGB color image sequences. Real-time wind speed and direction were acquired by an ultrasonic meteorological sensor installed on the top of the tracking bracket. External meteorological forecast data were obtained through the numerical weather prediction API interface of the meteorological agency. The data includes cloud cover spatial distribution forecasts and solar irradiance grid forecasts for the future target time period. The cloud cover spatial distribution forecasts for the future target time period provide parameters such as total cloud cover, low cloud cover, and cloud base height in a gridded format for every 15 minutes within the next 0 to 4 hours. The solar irradiance grid forecasts provide forecast values ​​for total horizontal irradiance, direct normal irradiance, and diffuse irradiance at the same spatiotemporal resolution.

[0033] S22, process the time-series sky image, extract the cloud region in the image, and use optical flow method to analyze the pixel displacement of the cloud region between consecutive frames to calculate and generate a two-dimensional motion vector field; Step S22 includes the following sub-steps: S221, Denoising and contrast enhancement are performed on the time-series sky image to obtain a standard sky image; It should be noted that median filtering is used to denoise the time-series sky image with a kernel size of 5×5 to remove sensor noise during image acquisition. Contrast-limited adaptive histogram equalization is then used to stretch the local contrast of the image to a preset range, making the grayscale difference between the thin cloud area and the clear sky background more significant. Next, based on the sun position and camera exposure parameters, non-uniformity correction is performed on the image to compensate for the light reduction effect at the edge of the fisheye lens. The above processing yields a standard sky image, which serves as the input for subsequent segmentation and motion analysis.

[0034] S222: Based on the brightness value and color features of pixels, a segmentation threshold is set for the standard sky image to distinguish the image pixels into cloud pixel regions and non-cloud pixel regions, and a cloud binary mask is generated. It should be noted that a multi-feature threshold segmentation method is used to separate cloud pixels from clear sky pixels in a standard sky image. Clouds typically have higher grayscale values. The system adaptively calculates the brightness threshold using the Otsu method based on the image's global brightness histogram. In the RGB space, clouds usually exhibit similar R / G / B channel values, while the blue channel is significantly dominant in clear sky regions. The system calculates the normalized red-blue difference index as a color feature and sets a corresponding threshold. For each pixel in the image, if its brightness value is greater than a certain threshold and its color feature meets the cloud determination rules, it is marked as a cloud pixel; otherwise, it is marked as a non-cloud pixel. This generates a binary cloud mask with the same size as the original image. To eliminate isolated noise points, a morphological opening operation is performed on the binary mask with a kernel size of 3×3 to make the cloud region boundaries smooth and continuous.

[0035] S223: Based on two consecutive frames of temporal sky images and a binary cloud mask, the optical flow algorithm is used to calculate the displacement of pixels in the cloud pixel region to obtain the initial motion vector field. It should be noted that, based on two consecutive standard sky images and the corresponding binary cloud mask, the Farneback dense optical flow algorithm is used to calculate the motion vector of the cloud pixel region marked as 1 in the mask. The displacement vector of each cloud pixel in the image plane is calculated, and the set of all vectors constitutes the initial motion vector field.

[0036] S224. Outlier removal and spatial smoothing are performed on the initial motion vector field to obtain a filtered motion vector field. Vector clustering is performed on the filtered motion vector field to extract the dominant motion vector that represents the main motion trend of the cloud. Combined with real-time wind speed and direction data, physical constraint verification and correction are performed on the dominant motion vector to generate the final two-dimensional motion vector field.

[0037] It should be noted that the initial motion vector field contains anomalous vectors caused by sudden changes in local illumination, cloud deformation, or mismatches. These vectors need to be post-processed to obtain a stable and physically reasonable motion representation. Then, outlier removal and spatial smoothing are performed on the initial motion vector field to obtain the filtered motion vector field. First, the median of the vectors within the local window is calculated. If the Euclidean distance between the current vector and the median of its neighborhood exceeds a preset threshold, it is identified as an abnormal vector and removed. Its position is filled by the median of its neighborhood. The removed vector field is then Gaussian smoothed to obtain the filtered motion vector field. K-Means clustering analysis is performed on the filtered vector field to group vectors with similar directions and magnitudes into the same motion cluster. The cluster with the most pixels is selected, and the arithmetic mean of all its vectors is calculated to obtain the dominant motion vector that represents the main motion trend of the cloud. By using camera calibration parameters, the wind speed vector is converted into a projection vector under the image plane; the angular deviation between the dominant motion vector and the wind speed projection vector is calculated. If the angular deviation is greater than a preset deviation threshold, it indicates a contradiction between visual motion and atmospheric advection. The wind speed vector and the dominant motion vector are then weighted and fused, with the expression: u f =α· u d +(1-α)· u w ; v f =α· v d +(1-α)· v w ; In the formula, α is the visual weight coefficient, ( u d , v d ) is the dominant motion vector, ( v w , u w ) is the wind speed projection vector, ( u f , v f ) represents the final two-dimensional motion vector field.

[0038] S23 maps the two-dimensional motion vector field to the sky spherical coordinate system to predict the movement trajectory of the cloud layer within a preset time window in the future. It should be noted that the two-dimensional motion vector field is mapped to the sky spherical coordinate system and converted into cloud angular velocity. The prediction time window and time step are set. The position of the cloud feature point identified at the current moment in the sky spherical coordinate system is used as the initial state. The predicted cloud position at each future prediction time is recursively calculated using a linear motion extrapolation model. Cubic spline interpolation is performed on the predicted position sequence to generate a continuous and smooth cloud movement trajectory curve. This curve describes the movement path of the cloud feature point on the sky spherical surface within the prediction time window.

[0039] S24, compare the movement trajectory of the clouds with the theoretical position sequence of the sun in time and space to determine the geometric occlusion relationship of the clouds on the sun, wherein the geometric occlusion relationship is the occlusion state and the degree of occlusion; Step S24 includes the following sub-steps: S241 aligns the predicted cloud movement trajectory within the future time window with the solar reference position sequence within the same future time window on the time axis; S242, for each aligned time point, calculate the angular distance between the predicted position of the cloud feature point and the corresponding theoretical position of the sun in the sky spherical coordinate system using the spherical trigonometry formula; the expression is: d k =arccos[sinθ c (t k )·sinα sun (t k )+cosθ c (t k )·cosα sun (t k )·cos(φ c (t k )-γ sun (t k ))]; In the formula, d k Let θ be the angular distance. c (t k φ represents the predicted elevation angle of the cloud layer. c (t k ) represents the predicted azimuth angle of the cloud layer, γ sun (t k α is the theoretical azimuth angle of the sun. sun (t k ( ) represents the theoretical solar altitude angle; S243, preset the occlusion radius threshold, compare the angular distance with the occlusion radius threshold, if the angular distance is greater than the occlusion radius threshold, it is determined to be an unoccluded state, if the angular distance is less than or equal to the occlusion radius threshold, it is determined to be an occlusion state, and calculate the occlusion degree coefficient based on the angular distance and the occlusion radius threshold; It should be noted that when the angular distance is greater than the occlusion radius threshold, the system is considered to be in an unoccluded state, and the occlusion degree coefficient is 0. When the angular distance is less than or equal to the occlusion radius threshold, the system is considered to be in an occlusion state, and the occlusion degree coefficient is calculated based on the angular distance and the occlusion radius threshold. The expression is as follows: m k =1-d k / Rc; In the formula, Rc is the occlusion radius threshold. m k This is the occlusion degree coefficient. m k The closer the value is to 1, the more severe the cloud cover. m k The closer the value is to 0, the more likely it is to be located only at the edge of the occlusion.

[0040] S245 integrates the comparison results of each time point within the future time window and outputs a time series including occlusion status and occlusion degree coefficient as a geometric occlusion relationship.

[0041] S25, based on geometric shading relationships and combined with clear-sky atmospheric radiation models and external meteorological forecast data, calculates and corrects the direct and diffuse solar irradiance components reaching the photovoltaic panel location, and outputs the predicted curves of normal direct irradiance and total irradiance of the inclined surface reaching the photovoltaic panel within the future time window.

[0042] Step S25 includes the following sub-steps: S251, based on the solar theoretical position sequence, uses the clear-sky atmospheric radiation transfer model to calculate the theoretical normal direct irradiance and theoretical total irradiance of the horizontal plane reaching the location of the photovoltaic power station within a future time window under cloudless conditions. S252. Based on the geometric shading relationship, determine the corresponding direct irradiance attenuation factor, and multiply the theoretical normal direct irradiance by the direct irradiance attenuation factor to obtain the predicted normal direct irradiance. It should be noted that if the shading state is unobstructed, the direct sunlight attenuation factor is 1; if the shading state is completely obstructed, the direct sunlight attenuation factor is 0; if the shading state is partially obstructed, the direct sunlight attenuation factor is 1- m k .

[0043] S253, based on cloud cover information in external meteorological forecast data, uses an anisotropic scattering sky model to correct the scattering component in the theoretical total irradiance of the horizontal plane, and obtains the predicted scattering irradiance. It should be noted that the influence of clouds on diffuse irradiance is more complex, and the system uses an anisotropic scattering sky model for correction. As a preferred embodiment, the Hay scattering model is used, which divides sky scattering into three parts: isotropic background scattering, solar periphery bright ring scattering, and horizontal band scattering. The cloud cover information and solar position and photovoltaic panel attitude are input from external weather forecast data, and the scattering model outputs the corrected predicted diffuse irradiance.

[0044] S254, based on the attitude angle of the photovoltaic panel within a future time window, combines the predicted normal direct irradiance and the predicted scattered irradiance into the predicted total irradiance of the inclined surface reaching the photovoltaic panel through inclined surface projection transformation, and outputs the predicted curve of normal direct irradiance and the predicted curve of total irradiance of the inclined surface.

[0045] It should be noted that, firstly, the cosine of the angle between the solar incidence angle and the normal to the photovoltaic panel is calculated, and the expression is: cosθ i =sinα sun ·cosβ+cosα sun ·sinβ·cos(γ sun -γ panel ); In the formula, cosθ i Let α be the cosine of the incident angle. sun β is the theoretical solar elevation angle, β is the tilt angle of the photovoltaic panel, and γ is the solar theoretical elevation angle. panel Let θ be the azimuth angle of the photovoltaic panel, and cosβ be the tilt cosine. In the initial calculation, the photovoltaic panel attitude angle adopts the current real-time attitude or the astronomical tracking attitude. In the iterative optimization process of step S3, the optimized attitude sequence is adopted. The predicted total irradiance reaching the tilted surface of the photovoltaic panel is expressed as: GTI p =DNI p ·cosθ i +DHI p ·R d +ρ·GHI p ·(1-cosβ) / 2; In the formula, GTI p To predict the total irradiance of the slope, DNI p To predict normal direct irradiance, DHI p To predict scattered irradiance, R d GHI is the ratio of scattered irradiance, where ρ is the ground reflectance. p To predict the total irradiance on the horizontal surface.

[0046] The expression for the ratio of scattered irradiance is: ; The system integrates the normal direct irradiance prediction curve and the inclined plane total irradiance prediction curve according to the time series. The normal direct irradiance prediction curve describes the direct radiation power received per unit area perpendicular to the solar incidence direction in the future period; the inclined plane total irradiance prediction curve describes the total radiation power received on the actual installation plane of the photovoltaic panel in the future period.

[0047] It should be noted that, using time-series images from an all-sky imager, a two-dimensional motion vector field characterizing the main motion trend of clouds is generated through optical flow analysis, vector clustering, and physical constraint verification of wind speed data. This motion vector is then mapped to a celestial coordinate system, and continuous, smooth cloud movement trajectory curves are generated using linear extrapolation and interpolation. By calculating the spherical angular distance between the predicted cloud position and the solar reference position, and introducing an occlusion radius threshold, the cloud's occlusion effect on the sun is quantified into a geometric occlusion relationship time series containing discrete occlusion states and continuous occlusion degree coefficients. Finally, based on this quantified occlusion relationship, attenuation corrections are applied to direct irradiance, and an anisotropic model correction is applied to diffuse irradiance. These are then synthesized through oblique projection transformation to form predicted curves for normal direct irradiance and total oblique irradiance, improving the spatiotemporal resolution and physical rationality of short-term irradiance prediction and providing data support for subsequent feedforward optimization decisions.

[0048] S3 comprehensively predicts power generation revenue, mechanical loss costs, and motion risk costs, constructs a feedforward optimization model, takes the theoretical solar position sequence, solar irradiance prediction curve, and physical constraints of photovoltaic panel motion as inputs, solves the feedforward optimization model, and outputs optimized feedforward instructions for the photovoltaic panel attitude in the future time period.

[0049] In this embodiment, a multi-objective feedforward optimization model is constructed based on the solar reference position sequence as a spatial reference and the total irradiance prediction curve of the inclined plane as the energy prediction basis. Taking into account power generation revenue, mechanical losses and operational risks, a multi-objective feedforward optimization model is constructed. By solving this model, the optimal motion trajectory of the photovoltaic panel in the future period can be planned in advance. While pursuing the maximization of power generation, the lifespan and safety of the equipment are taken into account, and an optimized feedforward instruction with both economy and robustness is generated.

[0050] Step S3 includes the following sub-steps: S31. Based on the predicted change curve of total irradiance on the inclined plane, calculate the total predicted power generation in the future prediction period as the power generation revenue item. Based on the change amount and rate of change of the photovoltaic panel attitude angle, calculate the mechanical energy loss of the photovoltaic panel as the mechanical loss item. Based on the predicted wind speed and real-time attitude angle, assess the wind load risk as the motion risk item. It should be noted that, based on the predicted total irradiance of the inclined plane and combined with the physical parameters of the photovoltaic array, the instantaneous power generation at each discrete moment in the future prediction time domain is calculated and accumulated to obtain the total predicted power generation revenue. The formula for calculating instantaneous power generation is: P (t k )= or · A GTI p (t k )· f temp ( T cell ); In the formula, or GTI represents the photoelectric conversion efficiency of the photovoltaic module under standard test conditions, where A is the total effective power generation area of ​​the photovoltaic array. p (t k ) is the t-th k Predicted total irradiance reaching the tilted surface of the photovoltaic panel at any given time. f temp ( T cell () represents the temperature correction factor. Considering the impact of battery temperature on efficiency, a linear model can be used: ; In the formula, κ is the power temperature coefficient, and T ref For reference temperature, T cell It can be estimated from ambient temperature and irradiance.

[0051] The power generation revenue term is defined as the total power generation within the predicted time domain, and its calculation expression is: ; In the formula, Δ t pred To predict the time step; Frequent movement of the tracking bracket accelerates the wear of the reducer, motor and transmission mechanism, while consuming electrical energy. The mechanical loss is modeled as an energy cost function related to the change in attitude angle and the rate of change.

[0052] The expression for the change in attitude angle is: ; In the formula, α ( k ), β ( k ) is the first t k The azimuth and elevation angles of the photovoltaic panels at any given time are the decision variables. The expression for mechanical wear cost is: ; In the formula, c α ,c β c represents the energy consumption and wear cost coefficients corresponding to unit angular changes in the azimuth and altitude axes; ω This is the coefficient of the squared term of angular velocity, used to penalize strenuous exercise; Strong winds are the most significant natural hazard source for photovoltaic tracking systems. Based on real-time wind speed forecasts and the current orientation of the photovoltaic panels, wind load risks are assessed. The formula for calculating instantaneous wind load risk is: ; In the formula, v wind (t k ) is the t-th k Predicted wind speed at any time, γ wind To predict wind direction, β ( k ) is the decision variable for the tilt angle of the photovoltaic panel, α( k ) represents the azimuth angle of the photovoltaic panel as the decision variable. c wind Risk conversion factor; The formula for calculating the sports risk item is: ; S32, the power generation revenue, mechanical loss and motion risk are weighted and combined, and the photovoltaic panel azimuth and elevation angles at each time in the future period are used as decision variables to construct a feedforward optimization model; The feedforward optimization model is: ; In the formula, w 1, w 2, w 3 represents the weighting coefficient, which can be dynamically adjusted according to the power plant's operation strategy, increasing when maximizing power generation. w 1. Increased equipment aging period w 2. Increased during windy seasons w 3.

[0053] S33, with the solar reference position sequence, the predicted curve of total irradiance on the inclined plane, and the attitude angle range, maximum angular velocity and maximum angular acceleration of the photovoltaic panel as constraints, solve the optimization function; combine the solved future azimuth angle optimization sequence and elevation angle optimization sequence of the photovoltaic panel as optimization feedforward instructions; It should be noted that the SQP solver under the MPC framework is used to obtain the GTI prediction curve, wind speed prediction, and solar reference position sequence for the next N steps at time t0; with the current photovoltaic panel attitudes α(0) and β(0) as the initial state, the SQP solver is called to calculate the optimal sequence. The attitude command (α*(1),β*(1)) at the first moment in the sequence is output to the execution layer as the initial feedforward command. In the next control cycle, all prediction data are updated and the above process is repeated to achieve rolling optimization. The photovoltaic panel azimuth optimization sequence and elevation angle optimization sequence obtained by solving are merged to form the initial feedforward command.

[0054] S34. Based on the signal-to-noise ratio of local sensing data, historical prediction errors, and weather change rate, a weighted calculation is performed to obtain the comprehensive confidence level of the optimized feedforward command.

[0055] It should be noted that the formula for calculating the overall confidence level is: C total = l d · C data + l m · C model + l w · C weather ; In the formula, C data For data quality factors, C model This is the model accuracy factor. C weather Weather stability factors l d The weights of the data quality factor. l m The weights of the model accuracy factor. l w The weights of the weather stability factor; The expression for the data quality factor is: ; In the formula, SNR is the signal-to-noise ratio of the sky image, and SNR0 is the reference signal-to-noise ratio. k snr The slope coefficient of the Sigmoid function. I wind This is a valid indicator for wind speed data; a value of 1 indicates valid wind speed sensor data, while a value of 0 indicates invalid data. The expression for the model accuracy factor is: ; In the formula, MAE is the mean absolute error. M The length of the sliding window. d pred,i For the first iSecondary predicted angular distance, d meas,i For the first i Second measured angular distance; The expression that maps MAE to confidence level is: ; In the formula, MAE max This is the upper limit threshold for the error. The expression for the weather stability factor is: ; In the formula, s v For wind speed standard deviation, The average wind speed, The standard deviation of wind direction. k v This is the wind speed sensitivity coefficient. This is the wind direction sensitivity coefficient.

[0056] In this embodiment, the solar reference position sequence is used as the spatial reference, and the total irradiance prediction curve of the inclined plane is used as the energy prediction basis. Taking into account the physical constraints of photovoltaic panel motion, a rolling time-domain optimization solver under the model prediction control framework is adopted to plan the optimal attitude sequence of photovoltaic panels in the future period in advance, and generate optimized feedforward instructions that are both economical and safe. At the same time, a comprehensive confidence evaluation mechanism based on data quality, model accuracy and weather stability is introduced to quantify the reliability of feedforward prediction, provide dynamic weight basis for subsequent multi-source information fusion, effectively improve power generation while significantly reducing mechanical wear and wind load risks, and maximize the comprehensive benefits of the photovoltaic power station throughout its entire life cycle.

[0057] S4 collects the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and performs weighted fusion with the feedforward control command to generate motor drive command, which drives the photovoltaic panel to track the sun.

[0058] It should be noted that by fusing multi-source sensor feedback signals and feedforward optimization commands in real time, motor drive commands are generated in dynamic and uncertain environments to achieve precise closed-loop control of the photovoltaic panel's attitude. Through a confidence-driven adaptive weighted fusion strategy, the system can maintain stable and reliable tracking performance under various weather conditions and sensor anomalies.

[0059] Step S4 includes the following sub-steps: S41 uses a four-quadrant photoelectric sensor to collect two-axis angle deviation signals in real time, a dual-axis tilt sensor to collect absolute attitude angle feedback in real time, and a vision sensor to collect solar visual position data in real time. S42, calculate the first confidence level of the four-quadrant photoelectric sensor based on signal strength and stability, calculate the second confidence level of the dual-axis tilt sensor based on self-test status and data refresh rate, and calculate the third confidence level of the visual sensor based on image clarity and solar feature recognition success rate. It should be noted that the expression for calculating the first confidence level is: C quad = f sig · f stab ; In the formula, f sig Signal strength factor f stab This is the signal stability factor; The expression for calculating the signal strength factor is: ; In the formula, I sum This represents the total photocurrent. I min For the preset threshold, when I sum ≤ I min hour, f sig =0, when I sum > I min hour, f stab =1.

[0060] The expression for calculating the signal stability factor is: ; In the formula, σ Δ σ0 is the standard deviation of the deviation signal within a short time window, and σ0 is the preset allowable standard deviation of fluctuation. The expression for calculating the second confidence level of a dual-axis tilt sensor is as follows: ; In the formula, S self This is the sensor's self-test status flag; 1 indicates normal operation, and 0 indicates a fault. Δ t last The time interval since the last successful reception of valid data. t incl The time decay constant; The expression for calculating the third confidence level of a visual sensor is: ; In the formula, Q img Image quality factor R succ To improve the success rate of identification.

[0061] S43, based on the second and third confidence levels, the attitude angle feedback and the solar visual position data are weighted and fused to obtain the fused absolute attitude angle; Normalizing the second and third confidence levels yields the weighted fusion weights, expressed as follows: ; In the formula, w incl The weighted fusion weights for the second confidence level. w vision ε is the weighted fusion weight for the third confidence level, and ε is the error term; Based on the second and third confidence levels, the attitude angle feedback and the solar visual position data are weighted and fused to obtain the fused absolute attitude angle, expressed as: α fused = w incl · α actual + w vision · α vision ; β fused = w incl ·β actual + w vision ·β vision ; In the formula, α actual This is the actual azimuth angle. α vision For visual measurement of azimuth, β actual β is the actual elevation angle. vision For visual measurement of elevation angle, ( α fused ,β fused (This refers to the fusion of absolute attitude angles;) Based on the second and third confidence levels, the confidence level of the fused absolute attitude angle is calculated, and the expression is as follows: ; S44, the optimized feedforward command is corrected in real time according to the two-axis angle deviation signal to obtain the corrected feedforward target. Based on the comprehensive confidence of the optimized feedforward command and the confidence of the fused absolute attitude angle, the corrected feedforward target and the fused absolute attitude angle are weighted and fused to generate the final target attitude angle. It should be noted that the fast deviation angle in angular space is converted from the two-axis angular deviation signal through calibration coefficients, and the calculation expression is as follows: Δ α quad = k α ·Δ x ; Δβ quad = k β ·Δ y ; In the formula, (Δ α quad ,Δβ quad () represents the rapid deviation angle. k α , k β This is the proportionality coefficient; The first confidence level and the fast bias angle are used to correct the optimized feedforward instruction in real time, resulting in the corrected feedforward objective, expressed as: α corr = α ff +Δ α quad · C quad ; β corr = β ff +Δ β quad · C quad ; In the formula, ( α ff , β ff To optimize feedforward instructions; The corrected feedforward target and the fused absolute attitude are then subjected to a second weighted fusion. The overall confidence of the feedforward command and the confidence of the fused absolute attitude are normalized to obtain the weights of the overall confidence of the feedforward command and the fused absolute attitude, expressed as follows: ; In the formula, w ff The confidence weights for feedforward instructions are calculated.w fb To incorporate absolute attitude confidence weights; Based on the comprehensive confidence weight of the feedforward command and the fused absolute attitude confidence weight, the corrected feedforward target and the fused absolute attitude angle are weighted and fused to generate the final target attitude angle, expressed as: α target = w incl · α corr + w fb · α fused ; β target = w ff · β corr + w fb ·β fused ; In the formula, ( α target , β target () represents the final target attitude angle.

[0062] S45 calculates the tracking error between the final target attitude angle and the current fused absolute attitude angle, and uses a PID control algorithm based on the tracking error to calculate the motor drive command to drive the photovoltaic panel to track the sun.

[0063] In this embodiment, the instantaneous deviation signals of the four-quadrant photoelectric sensor, the absolute attitude of the dual-axis tilt sensor, and the high-precision solar position of the visual sensor are acquired in parallel. Based on features such as signal strength, data freshness, and image quality, the dynamic confidence of each sensor is calculated in real time. Then, a confidence normalization weighting strategy is adopted to fuse the tilt angle and visual data into a highly reliable fused absolute attitude angle. At the same time, the four-quadrant deviation signal is used to correct the feedforward command in real time under the confidence constraint. The combined confidence of the feedforward command and the confidence of the fused absolute attitude are used as dynamic weights for secondary fusion to generate the final target attitude angle that takes into account both prediction foresight and feedback correction capability. This is converted into smooth and continuous motor drive commands by a PID controller. This solves the problems of single sensor being susceptible to environmental interference, difficulty in quantifying feedforward prediction uncertainty, and poor adaptability of traditional fixed-weight fusion. It realizes seamless and smooth switching and robust tracking of the system under all operating conditions such as sunny / rainy, sensor normal / faulty, and reliable / failed prediction, improving the control accuracy, environmental adaptability, and fault tolerance of the photovoltaic tracking system.

[0064] like Figure 2As shown, in a second aspect, the present invention provides a photovoltaic panel solar tracking control system, implemented using a photovoltaic panel solar tracking control method, comprising: The reference calibration module is used to calculate the reference theoretical position of the sun based on the geographical location and time information of the photovoltaic power station through astronomical algorithms, and to calibrate the reference theoretical position by combining it with measured data to generate a solar reference position sequence. The environmental perception and prediction module is used to acquire local environmental perception data and external weather forecast data of the photovoltaic power station, predict the geometric shading relationship of clouds on the sun in the future period based on the data, and calculate the solar irradiance prediction curve reaching the photovoltaic panel location. The feedforward optimization decision module is used to comprehensively predict power generation revenue, mechanical loss cost and motion risk cost, construct a feedforward optimization model, take the solar theoretical position sequence, solar irradiance prediction curve and photovoltaic panel motion physical constraints as input, solve the feedforward optimization model, and output the optimized feedforward instruction of photovoltaic panel attitude in the future period. The fusion control module is used to collect the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and to perform weighted fusion with the feedforward control command to generate execution control command to drive the photovoltaic panel to track the movement of the sun.

[0065] It should be noted that this system corresponds to the aforementioned photovoltaic panel solar tracking control method and system. All implementation methods in the above method embodiments are applicable to the embodiments of this system and can achieve the same technical effect.

[0066] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0067] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the system and modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0068] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0069] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0070] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0071] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0072] Furthermore, it should be noted that in the system and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of the present invention. Moreover, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order; some steps can be executed in parallel or independently of each other. Those skilled in the art will understand that all or any step or component of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or a combination thereof. This is something that those skilled in the art can achieve by using their basic programming skills after reading the description of the present invention.

[0073] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing system. The computing system can be a known general-purpose system. Therefore, the object of the present invention can also be achieved simply by providing a program product containing program code implementing the method or apparatus. That is, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any known storage medium or any storage medium developed in the future. It should also be noted that in the apparatus and method of the present invention, it is obvious that the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent to the present invention. Furthermore, the steps performing the above series of processes can naturally be performed in the order described, but are not necessarily required to be performed in chronological order. Some steps can be performed in parallel or independently of each other.

[0074] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A solar tracking control method for photovoltaic panels, characterized in that, Includes the following steps: S1, based on the geographical location and time information of the photovoltaic power station, calculates the reference theoretical position of the sun through astronomical algorithms, and calibrates the reference theoretical position by combining it with measured data to generate a solar reference position sequence; S2, acquire local environmental perception data and external weather forecast data of photovoltaic power station, predict the geometric shading relationship of cloud layer on the sun in the future period based on the data, and calculate the solar irradiance prediction curve reaching the photovoltaic panel location; S3 comprehensively predicts power generation revenue, mechanical loss cost, and motion risk cost, constructs a feedforward optimization model, takes the theoretical solar position sequence, solar irradiance prediction curve, and physical constraints of photovoltaic panel motion as input, solves the feedforward optimization model, and outputs the optimized feedforward command of photovoltaic panel attitude in the future period. S4 collects the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and performs weighted fusion with the feedforward control command to generate motor drive command, which drives the photovoltaic panel to track the sun.

2. The photovoltaic panel solar tracking control method as described in claim 1, characterized in that, Step S1, which involves calculating the theoretical reference position of the sun based on the geographical location and time information of the photovoltaic power station using astronomical algorithms and calibrating the theoretical reference position using measured data to generate a solar reference position sequence, includes the following sub-steps: S11: Obtain the geographical longitude and latitude of the photovoltaic power station, and obtain the current Coordinated Universal Time (UTC). Based on the geographical longitude and current time zone information, convert the UTC to the local mean solar time. S12, determine the annual day corresponding to Coordinated Universal Time, calculate the ecliptic longitude of the sun based on the annual day using the celestial motion model, and calculate the solar declination angle at the current moment; S13, correct the local mean solar time for mean time difference and longitude time difference to obtain the local true solar time. Using noon as the reference, calculate the solar hour angle at the current time based on the local true solar time. S14 calculates the theoretical azimuth and theoretical altitude angles of the sun using spherical trigonometric formulas based on geographical latitude, solar declination angle, and solar hour angle. S15, preset light intensity threshold. If the current light intensity of the photovoltaic power station is greater than the light intensity threshold, then collect the solar image at the current moment, process the solar image at the current moment, and identify the actual center position of the sun. S16. Compare the angle corresponding to the actual center position of the graphic with the theoretical azimuth and theoretical altitude angle of the sun to obtain the azimuth calibration deviation and altitude calibration deviation. S17. Based on the azimuth calibration deviation and elevation calibration deviation, the theoretical elevation angle and theoretical azimuth angle are compensated and corrected to generate the solar reference position and obtain the solar reference position sequence.

3. The photovoltaic panel solar tracking control method as described in claim 2, characterized in that, Step S2, which involves acquiring local environmental sensing data and external weather forecast data for the photovoltaic power station, predicting the geometric shading relationship of clouds on the sun in the future based on the data, and calculating the solar irradiance prediction curve at the location of the photovoltaic panels, includes the following sub-steps: S21, acquire local environmental perception data and external weather forecast data of photovoltaic power station. The local environmental perception data includes time-series sky images and real-time wind speed and direction. The external weather forecast data includes cloud cover spatial distribution forecast and solar irradiance grid forecast information for the future target period. S22, process the time-series sky image, extract the cloud region in the image, and use optical flow method to analyze the pixel displacement of the cloud region between consecutive frames to calculate and generate a two-dimensional motion vector field; S23 maps the two-dimensional motion vector field to the sky spherical coordinate system to predict the movement trajectory of the cloud layer within a preset time window in the future. S24, compare the movement trajectory of the clouds with the theoretical position sequence of the sun in time and space to determine the geometric occlusion relationship of the clouds on the sun, wherein the geometric occlusion relationship is the occlusion state and the degree of occlusion; S25, based on geometric shading relationships and combined with clear-sky atmospheric radiation models and external meteorological forecast data, calculates and corrects the direct and diffuse solar irradiance components reaching the photovoltaic panel location, and outputs the predicted curves of normal direct irradiance and total irradiance of the inclined surface reaching the photovoltaic panel within the future time window.

4. The photovoltaic panel solar tracking control method as described in claim 3, characterized in that, Step S22, which involves processing the temporal sky image, extracting the cloud region from the image, and using optical flow to analyze the pixel displacement of the cloud region between consecutive frames to calculate and generate a two-dimensional motion vector field, includes the following sub-steps: S221, Denoising and contrast enhancement are performed on the time-series sky image to obtain a standard sky image; S222: Based on the brightness value and color features of pixels, a segmentation threshold is set for the standard sky image to distinguish the image pixels into cloud pixel regions and non-cloud pixel regions, and a cloud binary mask is generated. S223: Based on two consecutive frames of temporal sky images and a binary cloud mask, the optical flow algorithm is used to calculate the displacement of pixels in the cloud pixel region to obtain the initial motion vector field. S224, outlier removal and spatial smoothing are performed on the initial motion vector field to obtain the filtered motion vector field; Vector clustering is performed on the filtered motion vector field to extract the dominant motion vector that represents the main motion trend of the cloud layer; By combining real-time wind speed and direction data, physical constraints are verified and corrected on the dominant motion vector to generate the final two-dimensional motion vector field.

5. The photovoltaic panel solar tracking control method as described in claim 4, characterized in that, Step S24, which involves comparing the movement trajectory of the clouds with the theoretical position sequence of the sun in a spatiotemporal manner to determine the geometric occlusion relationship of the clouds on the sun, wherein the geometric occlusion relationship is the occlusion state and the degree of occlusion, includes the following sub-steps: S241 aligns the predicted cloud movement trajectory within the future time window with the solar reference position sequence within the same future time window on the time axis; S242, For each aligned time point, calculate the angular distance between the predicted position of the cloud feature point and the corresponding theoretical position of the sun in the sky spherical coordinate system using the spherical trigonometry formula; S243, preset the occlusion radius threshold, compare the angular distance with the occlusion radius threshold, if the angular distance is greater than the occlusion radius threshold, it is determined to be an unoccluded state, if the angular distance is less than or equal to the occlusion radius threshold, it is determined to be an occlusion state, and calculate the occlusion degree coefficient based on the angular distance and the occlusion radius threshold; S245 integrates the comparison results of each time point within the future time window and outputs a time series including occlusion status and occlusion degree coefficient as a geometric occlusion relationship.

6. The photovoltaic panel solar tracking control method as described in claim 5, characterized in that, Step S25 includes the following sub-steps: S251, based on the solar theoretical position sequence, uses the clear-sky atmospheric radiation transfer model to calculate the theoretical normal direct irradiance and theoretical total irradiance of the horizontal plane reaching the location of the photovoltaic power station within a future time window under cloudless conditions. S252. Based on the geometric shading relationship, determine the corresponding direct irradiance attenuation factor, and multiply the theoretical normal direct irradiance by the direct irradiance attenuation factor to obtain the predicted normal direct irradiance. S253, based on cloud cover information in external meteorological forecast data, uses an anisotropic scattering sky model to correct the scattering component in the theoretical total irradiance of the horizontal plane, and obtains the predicted scattering irradiance. S254, based on the attitude angle of the photovoltaic panel within a future time window, combines the predicted normal direct irradiance and the predicted scattered irradiance into the predicted total irradiance of the inclined surface reaching the photovoltaic panel through inclined surface projection transformation, and outputs the predicted curve of normal direct irradiance and the predicted curve of total irradiance of the inclined surface.

7. The photovoltaic panel solar tracking control method as described in claim 6, characterized in that, Step S3 includes the following sub-steps: S31. Based on the predicted change curve of total irradiance on the inclined plane, calculate the total predicted power generation in the future prediction period as the power generation revenue item. Based on the change amount and rate of change of the photovoltaic panel attitude angle, calculate the mechanical energy loss of the photovoltaic panel as the mechanical loss item. Based on the predicted wind speed and real-time attitude angle, assess the wind load risk as the motion risk item. S32, the power generation revenue, mechanical loss and motion risk are weighted and combined, and the photovoltaic panel azimuth and elevation angles at each time in the future period are used as decision variables to construct a feedforward optimization model; S33, with the solar reference position sequence, the predicted curve of total irradiance on the inclined plane, and the attitude angle range, maximum angular velocity and maximum angular acceleration of the photovoltaic panel as constraints, solve the optimization function; combine the solved future azimuth angle optimization sequence and elevation angle optimization sequence of the photovoltaic panel as optimization feedforward instructions; S34. Based on the signal-to-noise ratio of local sensing data, historical prediction errors, and weather change rate, a weighted calculation is performed to obtain the comprehensive confidence level of the optimized feedforward command.

8. The photovoltaic panel solar tracking control method as described in claim 7, characterized in that, Step S4 includes the following sub-steps: S41 uses a four-quadrant photoelectric sensor to collect two-axis angle deviation signals in real time, a dual-axis tilt sensor to collect absolute attitude angle feedback in real time, and a vision sensor to collect solar visual position data in real time. S42, calculate the first confidence level of the four-quadrant photoelectric sensor based on signal strength and stability, calculate the second confidence level of the dual-axis tilt sensor based on self-test status and data refresh rate, and calculate the third confidence level of the visual sensor based on image clarity and solar feature recognition success rate. S43, based on the second and third confidence levels, the attitude angle feedback and the solar visual position data are weighted and fused to obtain the fused absolute attitude angle; S44, the optimized feedforward command is corrected in real time according to the two-axis angle deviation signal to obtain the corrected feedforward target. Based on the comprehensive confidence of the optimized feedforward command and the confidence of the fused absolute attitude angle, the corrected feedforward target and the fused absolute attitude angle are weighted and fused to generate the final target attitude angle. S45 calculates the tracking error between the final target attitude angle and the current fused absolute attitude angle, and uses a PID control algorithm based on the tracking error to calculate the motor drive command to drive the photovoltaic panel to track the sun.

9. A photovoltaic panel solar tracking control system, implemented using the photovoltaic panel solar tracking control method as described in any one of claims 1-8, characterized in that, include: The reference calibration module is used to calculate the reference theoretical position of the sun based on the geographical location and time information of the photovoltaic power station through astronomical algorithms, and to calibrate the reference theoretical position by combining it with measured data to generate a solar reference position sequence. The environmental perception and prediction module is used to acquire local environmental perception data and external weather forecast data of the photovoltaic power station, predict the geometric shading relationship of clouds on the sun in the future period based on the data, and calculate the solar irradiance prediction curve reaching the photovoltaic panel location. The feedforward optimization decision module is used to comprehensively predict power generation revenue, mechanical loss cost and motion risk cost, construct a feedforward optimization model, take the solar theoretical position sequence, solar irradiance prediction curve and photovoltaic panel motion physical constraints as input, solve the feedforward optimization model, and output the optimized feedforward instruction of photovoltaic panel attitude in the future period. The fusion control module is used to collect the rapid light-sensing deviation signal and absolute attitude feedback signal of the photovoltaic panel in real time, and to perform weighted fusion with the feedforward control command to generate execution control command to drive the photovoltaic panel to track the movement of the sun.

10. A computer-readable storage medium, characterized in that, The storage medium stores a photovoltaic panel solar tracking control method and system program, which, when executed, implements the photovoltaic panel solar tracking control method as described in any one of claims 1-8.

Images