Photovoltaic array tracking method and system based on omnidirectional light intensity field sampling

Abstract

The invention provides a photovoltaic array tracking method based on omnidirectional light intensity field sampling, and the method comprises the steps: carrying out the time-space synchronous sampling of an omnidirectional light intensity detection network of a photovoltaic array, and obtaining a light intensity field original data set and a detection unit state set; removing invalid data through multi-scale space-time adaptive filtering to obtain an effective light intensity field data set, extracting multi-dimensional features of a time domain and a space domain of the effective light intensity field data set, resolving an initial optimal energy direction vector through space light intensity attenuation correction and three-dimensional curved surface fitting, and fusing environment reflection and atmospheric scattering factor correction to obtain an effective sun vector; and finally, combining the installation physical parameters of each photovoltaic panel and motion constraint to independently solve an optimal angle instruction, and driving a distributed execution mechanism to complete tracking. According to the method, the illumination perception and angle calculation precision is improved, the method is adaptive to a complex illumination environment, personalized independent tracking of the photovoltaic panel is realized, local optimum is avoided, and the global power generation efficiency, the system fault tolerance and the robustness are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent measurement and control technology for photovoltaic power generation, and in particular to a photovoltaic array tracking method and system based on omnidirectional light intensity field sampling. Background Technology

[0002] Photovoltaic array tracking technology is one of the core technologies of photovoltaic power generation systems. By adjusting the light-receiving angle of the photovoltaic panels, it ensures that the panels are always aligned with the direction of sunlight as much as possible, directly determining the light capture efficiency and overall power generation efficiency of the photovoltaic array. With the large-scale and refined development of the photovoltaic industry, traditional fixed-installation photovoltaic systems can no longer meet the demands for high-efficiency power generation.

[0003] Currently, while existing photovoltaic array tracking technology has upgraded from programmable control to sensor-based feedback, it still suffers from several technical shortcomings in complex outdoor lighting environments, making it difficult to achieve maximum global energy tracking of the photovoltaic array. Firstly, the sampling process lacks a systematic spatiotemporal synchronization mechanism; the sampling frequency is often fixed and cannot be adaptively adjusted according to the rate of change in illumination. Furthermore, the validity of the sampled data is not determined by considering operational data such as the working status and cleanliness of the detection unit, easily introducing invalid raw data. Secondly, the data processing and feature extraction stages are relatively crude, often employing simple filtering using a single time or spatial window without multi-scale spatiotemporal adaptive filtering. They also neglect the multi-dimensional extraction of the temporal dynamic characteristics and spatial gradient characteristics of the light intensity field, and fail to consider the attenuation of light intensity caused by factors such as atmospheric extinction and spatial shading. The problems include: firstly, the low accuracy of the optimal energy direction vector calculation; secondly, the imperfect design of the vector correction and angle calculation process, which simply integrates the theoretical solar vector with the measured light intensity direction without introducing dynamic correction factors for environmental reflection and atmospheric scattering, making it susceptible to being misled by environmental factors such as broken clouds and strong reflections; and thirdly, the photovoltaic panel angle calculation mostly adopts a unified synchronous adjustment method without combining the installation physical parameters and motion constraints of each photovoltaic panel for personalized independent calculation, which cannot adapt to the local illumination differences of the photovoltaic array, ultimately resulting in insufficient tracking accuracy and difficulty in fully improving the overall light energy capture efficiency of the photovoltaic array. Summary of the Invention

[0004] The purpose of this invention is to provide a photovoltaic array tracking method and system based on omnidirectional light intensity field sampling. Through multi-scale data processing, extraction of multi-dimensional features of light intensity field, calculation and correction of effective solar vector, the light energy capture efficiency and global power generation efficiency of photovoltaic array are improved.

[0005] To achieve the above objectives, the present invention provides the following solution: A photovoltaic array tracking method based on omnidirectional light intensity field sampling includes the following steps: Spatiotemporal synchronous sampling is performed on the omnidirectional light intensity detection network deployed in the photovoltaic array, and the original light intensity data, spatial position coordinate data and operating status data of each detection unit are collected to obtain the original light intensity field dataset and the detection unit status set. The original light intensity field dataset is processed by multi-scale spatiotemporal adaptive filtering using the state set of the detection unit, and invalid data is removed according to the operating state of the detection unit to obtain an effective light intensity field dataset. The temporal dynamic features and spatial gradient features of the effective light intensity field dataset are extracted to obtain a multidimensional feature set of the light intensity field; the multidimensional feature set of the light intensity field includes: spatiotemporal variation features of light intensity and spatial distribution features; Spatial light intensity attenuation correction and three-dimensional spatial surface fitting are performed on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector; The initial optimal energy direction vector is fused and corrected by using environmental reflection correction factor and atmospheric scattering coefficient to obtain effective solar vector; Based on the effective solar vector, the optimal azimuth and elevation commands are obtained by independently solving the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. The distributed actuators of the photovoltaic panel are driven to perform tracking actions based on the optimal azimuth and elevation angle commands.

[0006] Optionally, the omnidirectional light intensity detection network includes: a main detection unit and at least four auxiliary detection units, wherein both the main and auxiliary detection units are standard polyhedral structures with light intensity sensors; the sampling frequency of the spatiotemporal synchronous sampling is adaptively adjusted according to the rate of change of illumination, and the adjustment formula for the sampling frequency is: ;in, For real-time sampling frequency, Based on the sampling frequency, This is the frequency adjustment factor. This represents the light intensity value at the current sampling time. This represents the light intensity value at the previous sampling time. This represents the relative rate of change of light intensity.

[0007] Optionally, the original dataset of the light intensity field includes: light intensity value at the sampling time, sampling time, and spatial coordinates of the detection unit; the detection unit status set includes: working status identifier, sensor cleanliness, and power supply status.

[0008] Optionally, the original light intensity field dataset is subjected to multi-scale spatiotemporal adaptive filtering using the detector unit state set, and invalid data is removed based on the detector unit's operating state to obtain an effective light intensity field dataset, including: When the working status is 0, or the sensor cleanliness is less than the preset cleanliness threshold, or the power supply status is less than the preset power supply threshold, the original light intensity field data collected by the corresponding detection unit is removed, and the missing values ​​are filled in according to the original light intensity field data collected by the adjacent and normal detection unit to obtain the initial dataset. The time window length is adjusted based on the rate of change of light intensity over time in the original light intensity field dataset; the formula for calculating the time window length is: ;in, The length of the time window. Adjust the coefficient for the time window. The rate of change of light intensity over time. The base length of the time window; The spatial window radius is adjusted based on the spatial distribution density of the state set of the detection unit; the formula for calculating the spatial window radius is: ;in, To adapt the spatial window radius, This is the adjustment coefficient for the spatial window. The average spatial distance between detection units. The basic radius of the space window; Gaussian weighted filtering is applied to the initial dataset based on the time window length and spatial window radius to obtain an effective light intensity field dataset.

[0009] Optionally, the spatiotemporal variation characteristics of light intensity include: first-order difference of light intensity time series, second-order difference of light intensity time series, and moving average variance of light intensity; the spatial distribution characteristics include: spatial gradient magnitude of light intensity, spatial gradient direction of light intensity, and spatial gradient magnitude of light intensity.

[0010] Optionally, spatial intensity attenuation correction and three-dimensional spatial surface fitting are performed on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector, including: The spatiotemporal variation characteristics and spatial distribution characteristics of light intensity are weighted and assigned according to the spatial gradient contribution and temporal stability to obtain a weighted light intensity dataset. The atmospheric extinction coefficient, spatial shading coefficient, and light propagation distance are used as correction factors to perform point-by-point attenuation correction on the weighted light intensity dataset, resulting in the corrected light intensity dataset; the expression for the corrected light intensity dataset is: ;in, Let be the corrected light intensity value of the j-th detection unit at time i. Let be the weighted light intensity value of the j-th detection unit at time i. Atmospheric extinction coefficient, Let be the straight-line spatial distance from the j-th detection unit to the light incident reference point. Let be the spatial occlusion coefficient of the j-th detection unit; Using spatial location coordinate data as the independent variable and the corrected light intensity value as the dependent variable, a three-dimensional spatial light intensity distribution surface model is constructed using cubic spline interpolation. The light intensity distribution surface is obtained by fitting and solving the three-dimensional spatial light intensity distribution surface model using the least squares method. Find the extreme points of the light intensity distribution surface and normalize the surface normal vectors of the extreme points to obtain the initial optimal energy direction vector.

[0011] Optionally, the initial optimal energy direction vector is fused and corrected using an environmental reflection correction factor and an atmospheric scattering coefficient to obtain an effective solar vector, including: The theoretical solar vector is calculated based on the geographical location and UTC time of the photovoltaic array, combined with astronomical algorithms. The environmental reflection correction factor is determined based on the reflected light intensity value and the total light intensity value collected by the detection unit; The atmospheric scattering coefficient is determined by combining local real-time meteorological data and the temporal rate of change of light intensity. The theoretical solar vector is fused and corrected with the initial optimal energy direction vector by using the environmental reflection correction factor and atmospheric scattering coefficient to obtain the effective solar vector.

[0012] Optionally, based on the effective solar vector, the optimal azimuth and elevation commands are obtained independently for each photovoltaic panel in the photovoltaic array by using the installation physical parameters and motion constraints of each photovoltaic panel, including: The coordinates of the photovoltaic panel mounting base and the initial value of the photovoltaic panel normal vector are used as the installation physical parameters, and the range of values ​​for the photovoltaic panel azimuth angle, the range of values ​​for the elevation angle, and the limit of the single-step rotation rate of the photovoltaic panel are used as motion constraints. The optimal objective function is constructed by taking the maximization of the dot product between the photovoltaic panel normal vector and the effective solar vector as the core objective. The optimal objective function is iteratively solved based on the motion constraints to obtain the optimal azimuth and pitch commands.

[0013] A photovoltaic array tracking system based on omnidirectional light intensity field sampling includes: The data acquisition module is used to perform spatiotemporal synchronous sampling of the omnidirectional light intensity detection network deployed in the photovoltaic array, and to collect the original light intensity data, spatial location coordinate data and operating status data of each detection unit, so as to obtain the original light intensity field dataset and the detection unit status set. The data filtering module is used to perform multi-scale spatiotemporal adaptive filtering on the original light intensity field dataset through the state set of the detection unit, and to remove invalid data according to the operating status of the detection unit to obtain an effective light intensity field dataset. The feature extraction module is used to extract time-domain dynamic features and spatial-domain gradient features from the effective light intensity field dataset to obtain a multi-dimensional feature set of the light intensity field. The multi-dimensional feature set of the light intensity field includes: spatiotemporal variation features of light intensity and spatial distribution features. The energy calculation module is used to perform spatial light intensity attenuation correction and three-dimensional spatial surface fitting on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector. The feature fusion module is used to perform vector fusion correction on the initial optimal energy direction vector using the environmental reflection correction factor and atmospheric scattering coefficient to obtain an effective solar vector. The instruction generation module is used to independently calculate the optimal azimuth and optimal pitch instructions for each photovoltaic panel based on the effective solar vector and the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. The instruction execution module is used to drive the distributed actuators of the photovoltaic panel to perform tracking actions based on the optimal azimuth angle instruction and the optimal pitch angle instruction.

[0014] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The photovoltaic array tracking method based on omnidirectional light intensity field sampling provided by the present invention includes: performing spatiotemporal synchronous sampling on an omnidirectional light intensity detection network deployed in the photovoltaic array, and collecting the original light intensity data, spatial position coordinate data, and operating status data of each detection unit to obtain an original light intensity field dataset and a detection unit status set; performing multi-scale spatiotemporal adaptive filtering on the original light intensity field dataset through the detection unit status set, and removing invalid data according to the operating status of the detection units to obtain an effective light intensity field dataset; extracting time-domain dynamic features and spatial-domain gradient features from the effective light intensity field dataset to obtain... A multi-dimensional feature set of light intensity field is constructed, including spatiotemporal variation characteristics and spatial distribution characteristics of light intensity. Spatial light intensity attenuation correction and three-dimensional spatial surface fitting are applied to the multi-dimensional feature set to obtain an initial optimal energy direction vector. The initial optimal energy direction vector is then fused and corrected using an environmental reflection correction factor and atmospheric scattering coefficient to obtain an effective solar vector. Based on the effective solar vector, the optimal azimuth and elevation angle commands are independently calculated for each photovoltaic panel in the photovoltaic array, taking into account the installation physical parameters and motion constraints. The distributed actuators of the photovoltaic panels are then driven to perform tracking actions according to these commands. This method deploys an omnidirectional light intensity detection network in the photovoltaic array for spatiotemporal synchronous sampling, multi-scale data processing, and multi-dimensional feature extraction of the light intensity field. After calculating and correcting the effective solar vector, the optimal tracking angle is independently calculated for each photovoltaic panel and driven to execute the tracking. This achieves precise and intelligent tracking of the photovoltaic array under complex lighting conditions, improving the light capture efficiency and overall power generation efficiency of the photovoltaic array. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments 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.

[0016] Figure 1 This is a flowchart of the photovoltaic array tracking method based on omnidirectional light intensity field sampling of the present invention; Figure 2 This is a schematic diagram of the photovoltaic array tracking system based on omnidirectional light intensity field sampling according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, this embodiment of the invention provides a photovoltaic array tracking method based on omnidirectional light intensity field sampling, including the following steps: Step 100: Perform spatiotemporal synchronous sampling on the omnidirectional light intensity detection network deployed in the photovoltaic array, and collect the original light intensity data, spatial position coordinate data and operating status data of each detection unit to obtain the original light intensity field dataset and the detection unit status set; Step 200: Perform multi-scale spatiotemporal adaptive filtering on the original light intensity field dataset using the state set of the detection unit, and remove invalid data according to the operating state of the detection unit to obtain an effective light intensity field dataset; Step 300: Extract time-domain dynamic features and spatial-domain gradient features from the effective light intensity field dataset to obtain a multi-dimensional feature set of the light intensity field; the multi-dimensional feature set of the light intensity field includes: spatiotemporal variation features and spatial distribution features of light intensity; Step 400: Perform spatial light intensity attenuation correction and three-dimensional spatial surface fitting on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector; Step 500: The initial optimal energy direction vector is fused and corrected using the environmental reflection correction factor and atmospheric scattering coefficient to obtain the effective solar vector; Step 600: Based on the effective solar vector, the optimal azimuth and elevation commands are obtained by independently calculating the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. Step 700: Drive the distributed actuators of the photovoltaic panel to perform tracking actions according to the optimal azimuth angle command and the optimal pitch angle command.

[0020] In the specific implementation process, the omnidirectional light intensity detection network in step 100 consists of one main detection unit and at least four auxiliary detection units, all of which adopt a standard polyhedral structure with light intensity sensors (such as a regular hexahedron or octahedron, selected according to the light detection requirements of the photovoltaic array). Each facet is equipped with a high-sensitivity light intensity sensor. The main detection unit is deployed in the central area of ​​the photovoltaic array, and the auxiliary detection units are evenly distributed in different sub-areas of the photovoltaic array. All detection units receive satellite timing signals through a GPS module to obtain a unified UTC time as the sampling reference time, and perform spatial coordinate calibration on each detection unit to obtain its three-dimensional spatial coordinates (X, Y, Z) in the photovoltaic array coordinate system. These coordinates are the static basic coordinates. The sampling frequency is adaptively adjusted in real time according to the rate of change of light intensity to ensure that the sampling frequency is increased when the light intensity changes drastically and the basic sampling frequency is maintained when the light intensity is stable, thereby reducing invalid data collection. The formula for adjusting the sampling frequency is: ; in, For real-time sampling frequency, Based on the sampling frequency, This is the frequency adjustment factor. This represents the light intensity value at the current sampling time. This represents the light intensity value at the previous sampling time. This represents the relative rate of change of light intensity. Next, the omnidirectional light intensity value at the current moment is collected by the omnidirectional light intensity sensor onboard the detection unit. Each sensor collects the incident light intensity value in the direction corresponding to the surface. The main detection unit summarizes the collected values ​​from each sensor to obtain the comprehensive omnidirectional light intensity value of a single detection unit as the raw light intensity data. Simultaneously, each detection unit performs its own status monitoring, collecting three types of operational status data in real time: working status indicator, sensor cleanliness, and power supply status. The working status indicator is a binary value, with 1 indicating normal operation and 0 indicating a malfunction. Sensor cleanliness is a percentage value, obtained by calculating the attenuation rate of the light intensity sensor's receiving efficiency; lower cleanliness indicates more severe dust and dirt accumulation on the sensor surface, resulting in greater errors in the collected light intensity data. The power supply status is characterized by the real-time power supply voltage and current values ​​of the detection unit. If the real-time power supply voltage or current value exceeds a preset threshold, it is determined to be a power supply abnormality. Finally, the three types of data—light intensity value at sampling time, sampling time, and spatial coordinates of the detection unit—are used as row indexes with the unique number of the detection unit and column indexes with the sampling timestamp to construct the original dataset of the light intensity field. Similarly, the three types of data—working status identifier, sensor cleanliness, and power supply status—are used as primary indexes with the unique number of the detection unit and secondary indexes with the sampling timestamp to construct the detection unit status set.

[0021] In the specific implementation process, step 200 uses the working status identifier, sensor cleanliness, and power supply status of the detection unit status set as the core judgment indicators, and sets a single indicator veto system. That is, if the original light intensity data of the detection unit meets any of the following conditions, it is judged as invalid data and directly removed from the original light intensity field dataset: 1. A working status indicator of 0 indicates that the detection unit has a hardware fault or software malfunction, and the collected light intensity data has no practical reference value. 2. Sensor cleanliness is less than the preset cleanliness threshold (the threshold is set according to the environmental characteristics such as dust and precipitation in the area where the photovoltaic array is located; in this embodiment, it is 60%~80%): This means that the dust and dirt on the sensor surface will cause the light intensity receiving efficiency to decrease, and the collected value will deviate too much from the actual light intensity. 3. Power supply status less than the preset power supply threshold, i.e., the power supply voltage is lower than 85% of the rated operating voltage or the power supply current is lower than 90% of the rated operating current: This indicates that abnormal power supply will cause a decrease in the sampling accuracy of the light intensity sensor, resulting in random noise in the data. Then, the light intensity data with missing spatial location or sampling time in the original light intensity field dataset is filled in using the spatial neighborhood weighted average method. That is, taking the spatial coordinates of the detection unit corresponding to the missing value as the center, the detection units in the photovoltaic array detection network with a spatial Euclidean distance ≤ a preset distance threshold (1.5 times the average deployment spacing of the detection units in this embodiment) and in normal working condition are retrieved as reference units. The weighted weight is calculated according to the spatial Euclidean distance between the reference unit and the missing unit, thereby calculating the filled light intensity value of the missing unit and obtaining the initial dataset.

[0022] Next, based on the light intensity time series of a single detector unit in the initial dataset, the light intensity time change rate is calculated using the first-order forward difference method. The time window length is adjusted according to the rate of change of light intensity over time, and the adjustment calculation formula is as follows: ; in, The adaptive time window length represents the time range covered by the sliding window of the time filter; The time window adjustment coefficient is an empirical value, calibrated based on the light fluctuation characteristics of the area where the photovoltaic array is located, and is used to control the influence of the light intensity time change rate on the length of the time window. It is the absolute value of the rate of change of light intensity over time; is the base length of the time window, and is the minimum length of the time window.

[0023] Simultaneously, the spatial window radius is adaptively adjusted based on the spatial distribution density of the detector unit state set, enabling dynamic matching between the spatial window radius and the spatial characteristics of the detector unit. The adjustment formula is as follows: ; in, The adaptive spatial window radius represents the radius range of the circular window for spatial filtering, and the detection unit within the window is the spatial neighborhood reference unit; This is the spatial window adjustment coefficient, an empirical value calibrated based on the deployment density of the detection network, used to control the influence of the average spatial distance on the spatial window radius; The average spatial distance between detection units; Let be the basic radius of the spatial window, and be the minimum radius of the spatial window. Finally, a spatiotemporal joint Gaussian weighted filter is applied to the initial dataset. Through dual filtering in both the time and spatial dimensions, random noise and environmental interference noise in the light intensity data are removed, ultimately generating an effective light intensity field dataset.

[0024] It should be noted that step 200 achieves accurate identification and removal of invalid light intensity data based on three elements: working status identifier, sensor cleanliness, and power supply status in the detection unit status set. It also completes the basic data optimization function of filling in missing values ​​by sampling data from adjacent normal detection units. Furthermore, it can adaptively adjust the time window length based on the light intensity time change rate and the spatial window radius based on the spatial distribution density of the detection units, breaking through the limitations of traditional fixed window filtering. At the same time, it performs Gaussian weighted filtering on the completed initial dataset by combining the adaptively adjusted time window and spatial window. This not only removes invalid information and fills in data gaps from the data source, but also effectively filters out random noise and environmental interference noise in the light intensity data, taking into account the real-time performance and smoothness of light intensity data processing. It also completely preserves the time-domain dynamic change characteristics and spatial gradient distribution characteristics of the light intensity field. The final generated effective light intensity field dataset can realistically and comprehensively represent the light intensity field distribution characteristics of the entire photovoltaic array, improving the fault tolerance and overall robustness of the data processing link of the entire photovoltaic array tracking system.

[0025] In the specific implementation process, step 300 first reconstructs the effective light intensity field dataset in the time domain according to the unique number of the detection unit, generating a light intensity time series of continuous sampling times for each detection unit. Then, the first-order difference, second-order difference, and moving average variance of the light intensity time series are calculated sequentially to construct the spatiotemporal variation characteristics of light intensity. Subsequently, the effective light intensity field dataset is reconstructed in the spatial domain according to the sampling timestamp, accurately binding the effective light intensity value of each detection unit with its real-time three-dimensional spatial coordinates to construct a light intensity spatial distribution grid for the entire photovoltaic array. Then, gradient calculation is performed on the light intensity spatial distribution grid based on the three-dimensional spatial gradient operator to obtain the light intensity spatial distribution characteristics, including the light intensity spatial gradient magnitude, light intensity spatial gradient direction, and light intensity spatial gradient magnitude. Finally, the spatiotemporal variation characteristics and spatial distribution characteristics of light intensity of all detection units at each sampling time are structurally integrated. A feature association index is constructed according to the four dimensions of detection unit number-sampling timestamp-spatiotemporal variation feature group-spatial distribution feature group. The first-order difference, second-order difference, and moving average variance of light intensity are used as the core dimensions of the spatiotemporal variation feature group, and the spatial gradient magnitude and spatial gradient direction of light intensity are used as the core dimensions of the spatial distribution feature group. This achieves accurate bidirectional association between feature data and the original data of the effective light intensity field dataset, and finally forms a multidimensional feature set of light intensity field with complete structure and clear dimensions.

[0026] In the specific implementation process, step 400 first completes the weighted assignment based on the spatial gradient contribution and temporal stability. The spatial gradient contribution is determined by the proportion of the spatial gradient magnitude of light intensity in the global light intensity field, and the temporal stability is determined by the reciprocal of the moving average variance of light intensity. This weighted fusion calculation is then performed on various dimensions of the spatiotemporal variation and spatial distribution characteristics of light intensity to obtain the weighted light intensity value of each detection unit at each sampling time, thus constructing a weighted light intensity dataset. Then, three types of attenuation correction factors are obtained: the atmospheric extinction coefficient β, calibrated by combining real-time meteorological data (visibility, air humidity, aerosol concentration) and light propagation characteristics of the photovoltaic array location; and the spatial shading coefficient K. j By matching the 3D spatial model of the photovoltaic array with the actual spatial location of the detection unit, the obstructions around the detection unit (photovoltaic panel supports, buildings, vegetation, etc.) are identified, and the proportion of the obstructed area to the light-receiving area of ​​the detection unit is calculated to obtain the light propagation distance L. j The linear distance from the j-th detector unit to the light incident reference point (the three-dimensional spatial coordinates of the main detector unit at the center of the photovoltaic array) is calculated using the spatial distance formula between two points. A point-by-point attenuation correction is then applied to the weighted light intensity dataset to obtain the corrected light intensity dataset. The correction formula is: ; in, Let be the corrected light intensity value of the j-th detection unit at time i. Let be the weighted light intensity value of the j-th detection unit at time i. Atmospheric extinction coefficient, Let be the straight-line spatial distance from the j-th detection unit to the light incident reference point. Let be the spatial occlusion coefficient of the j-th detector unit. Then, using the three-dimensional spatial coordinates (X, Y, Z) of the detector unit as the independent variable and the corrected light intensity as the dependent variable, cubic spline interpolation is used to interpolate and complete the discrete detector unit coordinates and corresponding light intensity values, filling in the light intensity values ​​at locations where no detector units are deployed within the entire photovoltaic array, thus constructing a continuous three-dimensional spatial light intensity distribution surface model that can completely characterize the spatial distribution of light intensity in the photovoltaic array. Next, the surface model is fitted and solved using the least squares method. With the objective of minimizing the sum of squared residuals between the fitted light intensity values ​​and the actual corrected light intensity values, the surface model parameters are iteratively optimized to obtain the light intensity distribution surface with the highest accuracy and the highest matching degree to the actual light intensity field. Finally, the extreme point of this light intensity distribution surface is solved. This extreme point is the spatial location with the largest light intensity value within the entire photovoltaic array. The surface normal vector at this extreme point is then calculated, and the normal vector is normalized by dividing each component of the normal vector by its magnitude, resulting in the normalized initial optimal energy direction vector.

[0027] It should be noted that step 400, by solving the optimal energy direction vector, achieves scientific weighted fusion of multi-dimensional features of the light intensity field and accurate spatial light intensity attenuation correction. This effectively compensates for the light intensity data deviation caused by environmental and physical factors such as atmospheric extinction and spatial shading, and significantly improves the accuracy of the light intensity field data in representing actual illumination conditions. At the same time, the discrete light intensity data of the detection unit is transformed into a continuous three-dimensional spatial light intensity distribution surface model through cubic spline interpolation. Combined with the least squares fitting optimization, the light intensity field is accurately modeled from discrete points to a continuous surface, which can fully reflect the spatial distribution law of light intensity across the entire photovoltaic array. Furthermore, by solving for the extreme points of the light intensity distribution surface and normalizing its normal vector, the optimal incident position and direction of the photovoltaic array were accurately located. At the same time, the initial optimal energy direction vector obtained by the solution has the characteristics of high precision and high matching, which truly reflects the optimal energy incident direction of the light. This provides a high-quality and high-reliability basic vector for subsequent effective solar vector fusion correction combined with environmental reflection and atmospheric scattering factors, and greatly improves the robustness and adaptability of the energy direction vector solution.

[0028] In the specific implementation process, step 500 first performs accurate calculation of the theoretical solar vector. Using the actual geographical location parameters of the photovoltaic array, including geographical latitude, geographical longitude, altitude and real-time UTC time as input parameters, the solar astronomical parameters are calculated step by step using classical astronomical algorithms. First, the solar declination angle is calculated using the Brest formula. Then, the UTC time is converted to local true solar time and the solar hour angle is calculated. Subsequently, the solar altitude angle and solar azimuth angle are calculated. Finally, a three-dimensional spatial coordinate system is established with the plane where the photovoltaic array is located as the XY plane and the plane perpendicular to the photovoltaic array upward as the Z axis. The solar altitude angle and solar azimuth angle are converted into a three-dimensional theoretical solar vector and the vector is normalized to obtain the unit theoretical solar vector.

[0029] Next, the environmental reflection correction factor was quantitatively determined. Using polyhedral light intensity sensors in each detection unit of the omnidirectional light intensity detection network, the direct solar light intensity and environmental reflected light intensity of the photovoltaic array area were collected, and the total light intensity was statistically obtained. The environmental reflection correction factor was calculated by the ratio of direct light intensity to total light intensity. Subsequently, the atmospheric scattering coefficient was comprehensively determined. An atmospheric scattering coefficient calculation model was constructed by combining local real-time meteorological data and the light intensity temporal change rate extracted in step 300. First, real-time meteorological data of the photovoltaic array location was obtained, then the calculated light intensity temporal change rate was retrieved, and the atmospheric scattering coefficient was calculated using a multi-factor weighted formula. The calculation formula is as follows: ;in, Atmospheric scattering coefficient, , , The weighting coefficients for visibility, relative humidity, and the rate of change of light intensity over time are 0.4, 0.3, and 0.3, respectively, in this embodiment. For visibility, The relative humidity is taken as the atmospheric humidity. Finally, the initial optimal energy direction vector is normalized to obtain a unit initial optimal energy direction vector. Then, a fusion weight for the two-way quantities is constructed based on the environmental reflection correction factor and the atmospheric scattering coefficient, where the fusion weight is the unit theoretical solar vector. Fusion weights of unit initial optimal energy direction vector Then, the fusion vector is calculated using the weighted fusion formula and normalized to obtain the unit effective solar vector.

[0030] It should be noted that step 500 not only uses astronomical algorithms combined with the geographical location of the photovoltaic array and UTC time to calculate a theoretical solar vector with strong stability, thus compensating for the problem that the measured initial optimal energy direction vector is easily affected by external factors such as environmental reflection and atmospheric scattering, but also dynamically determines the environmental reflection correction factor and atmospheric scattering coefficient based on the actual sampling data of the detection unit and real-time environmental data, realizing the adaptive adjustment of the fusion weights and breaking the limitations of traditional fixed weight fusion. At the same time, the weighted fusion completes the accurate correction of the initial optimal energy direction vector, effectively eliminating the vector calculation errors caused by environmental reflection and atmospheric scattering. The final effective solar vector has both the stability of astronomical theory and the real-time performance of the measured light intensity field. It can accurately and realistically represent the actual incident direction of sunlight under complex lighting conditions, providing a high-precision and high-reliability directional basis for the subsequent independent calculation of the optimal azimuth and elevation angles of the photovoltaic panel. This greatly improves the adaptability, anti-interference and overall robustness of the entire photovoltaic array tracking method under complex lighting conditions such as haze, broken clouds and strong reflection.

[0031] In the specific implementation process, step 600 first involves the precise calibration of the physical parameters for each photovoltaic panel and the personalized definition of motion constraints. This involves extracting unique physical parameters for each photovoltaic panel, specifically including the three-dimensional coordinates of the photovoltaic panel mounting base (the static three-dimensional spatial coordinates of the fixed mounting point of the photovoltaic panel in the unified coordinate system of the photovoltaic array, determined by calibration data during the construction and deployment of the photovoltaic array); and the initial value of the photovoltaic panel normal vector. This parameter is the unit vector of the normal direction of the photovoltaic panel when no angle adjustment has been made, determined by the initial installation tilt angle, orientation, and other physical characteristics of the photovoltaic panel. Simultaneously, personalized motion constraints are defined for each photovoltaic panel, matching its own mechanical structure and installation space. These include the azimuth angle range, determined by the installation space around the photovoltaic panel and the mechanical limits of the support rotation; the pitch angle range, determined by the structural strength of the photovoltaic panel and the mechanical limits of the support pitch; and the single-step rotation rate limit of the photovoltaic panel, i.e., the maximum single-step rotation angle of the azimuth and pitch angles. All parameters are recorded in independent files based on the unique identifier of the photovoltaic panel.

[0032] Next, we construct the optimal solution objective function with maximizing light energy capture efficiency as the core, first by retrieving the unit effective solar vector. Next, a spatial rotational transformation relationship is established between the real-time normal vector of the photovoltaic panel and the azimuth and pitch angles, where the azimuth angle is the clockwise rotation angle of the photovoltaic panel around the Z-axis, and the pitch angle is the upward rotation angle of the photovoltaic panel around its own horizontal axis (X-axis). Then, with maximizing the dot product of the real-time normal vector of the photovoltaic panel and the unit effective solar vector as the core optimization objective, and combined with the defined motion constraints, a constrained optimal solution objective function is constructed, expressed as: ; ; ; ; ; in, It is the azimuth angle. The pitch angle, This represents the maximum rotation angle in a single step, representing the azimuth angle. The maximum rotation angle in a single step is the pitch angle. , , and These are the current real-time azimuth angle and the current real-time pitch angle of the photovoltaic panel, respectively.

[0033] Then, the gradient ascent method is used to iteratively solve the constrained optimal objective function. Specifically: First, the iteration parameters are initialized, using the current azimuth and elevation angles of the photovoltaic panel as the initial values, setting the iteration accuracy threshold (0.01°) and the maximum number of iterations; Second, the objective function value and gradient at the current iteration value are calculated, with the gradient direction being the direction of the fastest increase in the objective function value, i.e., the optimal adjustment direction of the photovoltaic panel angle; Third, the azimuth and elevation angles are updated along the gradient ascent direction to obtain candidate solutions; Fourth, the candidate solutions are constrained and verified, first verifying the range of angle values, if... Then take ,like Then take The elevation angle is verified according to the same rules; the fifth step is to determine convergence. When the iteration converges, the obtained objective function value is the optimal azimuth and optimal elevation angle of the photovoltaic panel. Finally, the optimal angle command is generated in a structured manner. The optimal azimuth and optimal elevation angles calculated independently for each photovoltaic panel are standardized and structured to generate a unique angle command package for each photovoltaic panel. The command package contains executable information such as the unique identifier of the photovoltaic panel, the optimal azimuth command value, the optimal elevation command value, the angle adjustment step size, the adjustment rate, and the target angle threshold.

[0034] It should be noted that step 600, through the core transformation from the solar direction vector to the actual execution command, achieves personalized and independent calculation for each photovoltaic panel, breaking the traditional model of uniformly and synchronously adjusting the angle of photovoltaic arrays. Furthermore, by fully combining the unique installation physical parameters and motion constraints of each photovoltaic panel, the calculation results are made more closely aligned with the actual mechanical characteristics and installation scenario of the photovoltaic panel, thus preventing mechanical failures such as excessive rotation or excessive adjustment from the outset. Simultaneously, by constructing an optimization objective function centered on maximizing the dot product of the photovoltaic panel normal vector and the effective solar vector, the optimal light-gathering angle of the photovoltaic panel is theoretically guaranteed. This approach maximizes light capture efficiency and employs a gradient ascent method for constrained iterative solutions, ensuring both accuracy and real-time performance while adapting to dynamic changes in outdoor lighting. The resulting structured angle command package provides precise and executable operational guidelines for the distributed actuators, enabling effective conversion from solar direction vectors to actual tracking commands. Furthermore, its independent solution capability adapts to local lighting variations in the photovoltaic array, effectively avoiding local optima and further enhancing the overall power generation efficiency of the photovoltaic array. This significantly improves the precision and intelligence of the entire tracking system.

[0035] In the specific implementation process, step 700 performs precise distribution and one-to-one binding of the optimal angle command. The structured optimal angle command package generated in step 600 is distributed through the industrial communication network of the photovoltaic array. During the command distribution process, a precise one-to-one binding is performed based on the unique identifier of the photovoltaic panel and the corresponding distributed actuator, ensuring that each actuator only receives and processes the angle command of the corresponding photovoltaic panel, avoiding command crosstalk or misexecution. At the same time, the communication network adopts an encrypted transmission method to ensure the stability and security of the command transmission process and prevent command loss or tampering. Next, the distributed actuator receives the command and parses the action. The distributed actuator associated with each photovoltaic panel receives the bound optimal angle command package through its own communication module, and sends speed adjustment drive signals to the azimuth drive motor and the pitch drive motor respectively according to the segmented action path parsed from the optimal angle command package. The drive motor drives the support structure of the photovoltaic panel to rotate stepwise in the azimuth and pitch angles through the reducer.

[0036] like Figure 2 As shown, the present invention also provides a photovoltaic array tracking system based on omnidirectional light intensity field sampling, comprising: The data acquisition module is used to perform spatiotemporal synchronous sampling of the omnidirectional light intensity detection network deployed in the photovoltaic array, and to collect the original light intensity data, spatial location coordinate data and operating status data of each detection unit, so as to obtain the original light intensity field dataset and the detection unit status set. The data filtering module is used to perform multi-scale spatiotemporal adaptive filtering on the original light intensity field dataset through the state set of the detection unit, and to remove invalid data according to the operating status of the detection unit to obtain an effective light intensity field dataset. The feature extraction module is used to extract time-domain dynamic features and spatial-domain gradient features from the effective light intensity field dataset to obtain a multi-dimensional feature set of the light intensity field. The multi-dimensional feature set of the light intensity field includes: spatiotemporal variation features of light intensity and spatial distribution features. The energy calculation module is used to perform spatial light intensity attenuation correction and three-dimensional spatial surface fitting on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector. The feature fusion module is used to perform vector fusion correction on the initial optimal energy direction vector using the environmental reflection correction factor and atmospheric scattering coefficient to obtain an effective solar vector. The instruction generation module is used to independently calculate the optimal azimuth and optimal pitch instructions for each photovoltaic panel based on the effective solar vector and the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. The instruction execution module is used to drive the distributed actuators of the photovoltaic panel to perform tracking actions based on the optimal azimuth angle instruction and the optimal pitch angle instruction.

[0037] The beneficial effects of this invention are as follows: 1) An omnidirectional light intensity detection network consisting of a main detection unit and an auxiliary detection unit is used to carry out spatiotemporal synchronous sampling. The sampling frequency can be adaptively adjusted according to the rate of change of light intensity. At the same time, the validity of the sampling data is determined by combining the working status of the detection unit, the cleanliness of the sensor, and the power supply status. Invalid data is eliminated from the source, which solves the problem of fixed sampling frequency and lack of data validity determination in traditional technology. It can accurately and completely capture the spatiotemporal distribution characteristics of the light intensity field of the entire photovoltaic array, and greatly improve the accuracy of light perception and the authenticity of light intensity field representation. 2) The original data is processed by multi-scale spatiotemporal adaptive filtering. The time window and spatial window parameters can be dynamically adjusted according to the light intensity time change rate and the spatial distribution density of the detection unit. Compared with the traditional single fixed window filtering, it fits the characteristics of the light intensity field better. At the same time, multi-dimensional features in the time domain and spatial domain are extracted from the effective light intensity field dataset, and factors such as atmospheric extinction and spatial occlusion are introduced to correct the spatial light intensity attenuation. This solves the problem of traditional data processing being coarse and ignoring the influence of light intensity attenuation. It provides a high-quality data source for energy direction vector solution, improves the effectiveness of data processing and feature extraction, and significantly reduces the basic error of solution. 3) After solving the initial optimal energy direction vector by fitting a three-dimensional surface, the theoretical solar vector calculated based on astronomical algorithms is fused, and environmental reflection correction factors and atmospheric scattering coefficients are introduced for dynamic vector fusion correction. This solves the problem that traditional techniques simply fuse theoretical and measured vectors and are easily misled by environmental factors such as broken clouds, strong reflections, and haze. The effective solar vector obtained by the solution has both theoretical stability and real-time measurement, and can accurately characterize the actual incident direction of sunlight under complex lighting conditions. This achieves high-precision solution of the effective solar vector and improves the ability to resist environmental interference. 4) Based on the effective solar vector, the optimal azimuth and pitch angles are independently calculated by combining the unique installation physical parameters and motion constraints of each photovoltaic panel, rather than the traditional unified synchronous adjustment. This can accurately adapt to the local illumination differences in different areas of the photovoltaic array and the mechanical characteristics of the photovoltaic panels themselves, avoiding the local tracking deviation and local optimization problems caused by unified adjustment. This maximizes the capture of light energy across the entire photovoltaic array and avoids the local optimization problem. 5) From the precise perception of light intensity field and the high-precision calculation of solar vector to the personalized optimal angle tracking of each photovoltaic panel, the core objective is to maximize light energy capture. This effectively solves the problem of light energy waste caused by insufficient accuracy and poor adaptability, and greatly improves the overall light energy capture efficiency and global power generation efficiency of the photovoltaic array. 6) Based on the automatic spatiotemporal synchronous sampling of light intensity field data, multi-scale intelligent processing, automatic extraction of multi-dimensional features, accurate calculation and correction of solar vectors, independent intelligent calculation of the optimal angle of photovoltaic panels, and automatic tracking and execution by distributed actuators, the method can adapt to the large-scale and refined development needs of the photovoltaic industry, replace the traditional extensive tracking mode of program control or simple perception feedback, and improve the intelligence and refinement level of photovoltaic array tracking system.

[0038] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0039] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A photovoltaic array tracking method based on omnidirectional light intensity field sampling, characterized in that, Includes the following steps: Spatiotemporal synchronous sampling is performed on the omnidirectional light intensity detection network deployed in the photovoltaic array, and the original light intensity data, spatial position coordinate data and operating status data of each detection unit are collected to obtain the original light intensity field dataset and the detection unit status set. The original light intensity field dataset is subjected to multi-scale spatiotemporal adaptive filtering through the state set of the detection unit, and invalid data is removed according to the operating state of the detection unit to obtain an effective light intensity field dataset. The effective light intensity field dataset is subjected to time-domain dynamic features and spatial-domain gradient features to obtain a multi-dimensional feature set of the light intensity field. The multidimensional feature set of the light intensity field includes: spatiotemporal variation features of light intensity and spatial distribution features; Spatial light intensity attenuation correction and three-dimensional spatial surface fitting are performed on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector; The initial optimal energy direction vector is fused and corrected using environmental reflection correction factor and atmospheric scattering coefficient to obtain an effective solar vector; Based on the effective solar vector, the optimal azimuth and elevation commands are obtained by independently calculating the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. The distributed actuators of the photovoltaic panel are driven to perform tracking actions according to the optimal azimuth angle command and the optimal pitch angle command.

2. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, The omnidirectional light intensity detection network includes a main detection unit and at least four auxiliary detection units, wherein both the main detection unit and the auxiliary detection units are standard polyhedral structures with light intensity sensors; the sampling frequency of the spatiotemporal synchronous sampling is adaptively adjusted according to the rate of change of illumination, and the adjustment formula for the sampling frequency is: ;in, For real-time sampling frequency, Based on the sampling frequency, This is the frequency adjustment factor. This represents the light intensity value at the current sampling time. This represents the light intensity value at the previous sampling time. This represents the relative rate of change of light intensity.

3. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, The original dataset of the light intensity field includes: light intensity value at the sampling time, sampling time, and spatial coordinates of the detection unit; the status set of the detection unit includes: working status identifier, sensor cleanliness, and power supply status.

4. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 3, characterized in that, The original light intensity field dataset is subjected to multi-scale spatiotemporal adaptive filtering using the state set of the detection unit, and invalid data is removed based on the operating state of the detection unit to obtain an effective light intensity field dataset, including: When the working status is 0, or the sensor cleanliness is less than a preset cleanliness threshold, or the power supply status is less than a preset power supply threshold, the original light intensity field data collected by the corresponding detection unit is removed, and missing values ​​are filled in according to the original light intensity field data collected by the adjacent and normal detection units to obtain the initial dataset. The time window length is adjusted based on the light intensity temporal change rate of the original light intensity field dataset; the formula for calculating the time window length is: ;in, The length of the time window. Adjust the coefficient for the time window. The rate of change of light intensity over time. The base length of the time window; The radius of the spatial window is adjusted according to the spatial distribution density of the state set of the detection unit; the formula for calculating the radius of the spatial window is: ;in, To adapt the spatial window radius, This is the adjustment coefficient for the spatial window. The average spatial distance between detection units. The basic radius of the space window; The initial dataset is subjected to Gaussian weighted filtering based on the time window length and the spatial window radius to obtain the effective light intensity field dataset.

5. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, The spatiotemporal variation characteristics of light intensity include: first-order difference of light intensity time series, second-order difference of light intensity time series, and moving average variance of light intensity; the spatial distribution characteristics include: spatial gradient magnitude of light intensity, spatial gradient direction of light intensity, and spatial gradient magnitude of light intensity.

6. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, Spatial intensity attenuation correction and three-dimensional spatial surface fitting are performed on the multidimensional feature set of the light intensity field to obtain the initial optimal energy direction vector, including: The spatiotemporal variation characteristics and spatial distribution characteristics of light intensity are weighted and assigned according to the spatial gradient contribution and temporal stability to obtain a weighted light intensity dataset. The weighted light intensity dataset is attenuated point-by-point using atmospheric extinction coefficient, spatial occlusion coefficient, and light propagation distance as correction factors to obtain the corrected light intensity dataset; the expression for the corrected light intensity dataset is: ;in, Let be the corrected light intensity value of the j-th detection unit at time i. Let be the weighted light intensity value of the j-th detection unit at time i. Atmospheric extinction coefficient, Let be the straight-line spatial distance from the j-th detection unit to the light incident reference point. Let be the spatial occlusion coefficient of the j-th detection unit; Using the spatial location coordinate data as the independent variable and the corrected light intensity value as the dependent variable, a three-dimensional spatial light intensity distribution surface model is constructed using cubic spline interpolation. The light intensity distribution surface model is fitted and solved using the least squares method to obtain the light intensity distribution surface. Find the extreme points of the light intensity distribution surface and normalize the surface normal vector of the extreme points to obtain the initial optimal energy direction vector.

7. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, The initial optimal energy direction vector is fused and corrected using an environmental reflection correction factor and an atmospheric scattering coefficient to obtain an effective solar vector, including: The theoretical solar vector is calculated based on the geographical location and UTC time of the photovoltaic array, combined with astronomical algorithms. The environmental reflection correction factor is determined based on the reflected light intensity value and the total light intensity value collected by the detection unit; The atmospheric scattering coefficient is determined by combining local real-time meteorological data and the temporal rate of change of light intensity. The theoretical solar vector is fused and corrected with the initial optimal energy direction vector by using the environmental reflection correction factor and the atmospheric scattering coefficient to obtain the effective solar vector.

8. The photovoltaic array tracking method based on omnidirectional light intensity field sampling according to claim 1, characterized in that, Based on the effective solar vector, and through the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array, independent calculations are performed for each photovoltaic panel to obtain the optimal azimuth and elevation commands, including: The coordinates of the photovoltaic panel mounting base and the initial value of the photovoltaic panel normal vector are used as the installation physical parameters, and the range of values ​​for the photovoltaic panel azimuth angle, the range of values ​​for the elevation angle, and the limit of the single-step rotation rate of the photovoltaic panel are used as the motion constraint conditions. The optimal solution objective function is constructed by taking the maximization of the dot product between the photovoltaic panel normal vector and the effective solar vector as the core objective. The optimal objective function is iteratively solved based on the motion constraints to obtain the optimal azimuth command and the optimal pitch command.

9. A photovoltaic array tracking system based on omnidirectional light intensity field sampling, characterized in that, include: The data acquisition module is used to perform spatiotemporal synchronous sampling of the omnidirectional light intensity detection network deployed in the photovoltaic array, and to collect the original light intensity data, spatial location coordinate data and operating status data of each detection unit, so as to obtain the original light intensity field dataset and the detection unit status set. The data filtering module is used to perform multi-scale spatiotemporal adaptive filtering on the original light intensity field dataset through the state set of the detection unit, and to remove invalid data according to the operating state of the detection unit to obtain an effective light intensity field dataset. The feature extraction module is used to extract time-domain dynamic features and spatial-domain gradient features from the effective light intensity field dataset to obtain a multi-dimensional feature set of the light intensity field. The multidimensional feature set of the light intensity field includes: spatiotemporal variation features of light intensity and spatial distribution features; The energy calculation module is used to perform spatial light intensity attenuation correction and three-dimensional spatial surface fitting on the multi-dimensional feature set of the light intensity field to obtain the initial optimal energy direction vector. The feature fusion module is used to perform vector fusion correction on the initial optimal energy direction vector using the environmental reflection correction factor and the atmospheric scattering coefficient to obtain an effective solar vector. The instruction generation module is used to independently calculate the optimal azimuth and optimal pitch instructions for each photovoltaic panel based on the effective solar vector and through the installation physical parameters and motion constraints of each photovoltaic panel in the photovoltaic array. The instruction execution module is used to drive the distributed actuator of the photovoltaic panel to perform tracking actions according to the optimal azimuth angle instruction and the optimal pitch angle instruction.

Images