Roof photovoltaic intelligent monitoring system based on BAPV and dynamic tuning method and device

Abstract

The invention relates to the technical field of photovoltaic power generation, and discloses a BAPV-based roof photovoltaic intelligent monitoring system and a dynamic tuning method and device. The method comprises the following steps: scanning a preset BAPV assembly through three-dimensional laser, constructing a surface point cloud model, and establishing a three-dimensional coordinate system comprising BAPV assembly installation base points; acquiring a real-time environment parameter set corresponding to the BAPV component; calculating an optimal working angle of the BAPV component according to the real-time environment parameter set based on an improved particle swarm algorithm; simulating an all-weather shadow movement track according to the optimal working angle based on a Monte Carlo method, and generating a component shielding influence weight matrix; establishing a heat balance equation according to the component shielding influence weight matrix so as to calculate a spacing optimization parameter of the BAPV component; a control instruction is generated according to the spacing optimization parameters, and dynamic tuning of the BAPV assembly is completed according to the control instruction; the control instruction comprises a support rotation angle compensation amount, an inverter parameter adjustment amount and an abnormal component isolation instruction. And the pain points of modeling errors, static optimization limitation, thermal runaway risks and the like are solved.

Description

Technical Field

[0001] This application relates to the field of photovoltaic power generation technology, and particularly to a rooftop photovoltaic intelligent monitoring system, a dynamic optimization method, and a device based on BAPV. Background Art

[0002] In the field of Building Attached Photovoltaic (BAPV) systems, traditional 3D modeling uses drone aerial photography or manual measurement, making it difficult to accurately obtain the curvature changes of the rooftop surface and the spatial relationship of installation base points, resulting in the accumulation of coordinate system establishment errors. At the same time, existing optimization methods mostly perform static optimization based on fixed lighting angles or single temperature parameters, without considering the multi-dimensional influence of wind speed vectors on component heat dissipation and the dynamic coupling effect of environmental temperature and humidity on conversion efficiency. Conventional shadow prediction uses a simplified geometric projection model and does not combine the Monte Carlo random sampling method to simulate the dynamic occlusion effect of clouds, resulting in the weight matrix being unable to reflect the true occlusion frequency. Also, existing spacing optimization only considers natural convection heat dissipation and ignores the non-linear superposition effect of forced convection and radiation heat dissipation, leading to the component operating temperature exceeding the safety threshold. Traditional control instruction generation relies on periodic parameter updates and lacks the ability to isolate abnormal components in real-time and reconstruct the topology, causing the spread of local hot spot effects.

[0003] Therefore, there is an urgent need for a method to solve at least one of the above problems. Summary of the Invention

[0004] This application provides a rooftop photovoltaic dynamic optimization method and device based on BAPV, aiming to solve the problems that traditional 3D modeling uses drone aerial photography or manual measurement, making it difficult to accurately obtain the curvature changes of the rooftop surface and the spatial relationship of installation base points, resulting in the accumulation of coordinate system establishment errors. At the same time, existing optimization methods mostly perform static optimization based on fixed lighting angles or single temperature parameters, without considering the multi-dimensional influence of wind speed vectors on component heat dissipation and the dynamic coupling effect of environmental temperature and humidity on conversion efficiency. Conventional shadow prediction uses a simplified geometric projection model and does not combine the Monte Carlo random sampling method to simulate the dynamic occlusion effect of clouds, resulting in the weight matrix being unable to reflect the true occlusion frequency. Also, existing spacing optimization only considers natural convection heat dissipation and ignores the non-linear superposition effect of forced convection and radiation heat dissipation, leading to the component operating temperature exceeding the safety threshold. Traditional control instruction generation relies on periodic parameter updates and lacks the ability to isolate abnormal components in real-time and reconstruct the topology, causing the spread of local hot spot effects.

[0005] In a first aspect, this application provides a rooftop photovoltaic dynamic optimization method based on BAPV, including:

[0006] Constructing a surface point cloud model by three-dimensional laser scanning of preset BAPV components for establishing a three-dimensional coordinate system including the installation base points of BAPV components;

[0007] Obtain the real-time environmental parameter set corresponding to the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters;

[0008] Based on the improved particle swarm optimization algorithm, calculate the optimal working angle of the BAPV component according to the real-time environmental parameter set; the objective function of the improved particle swarm optimization algorithm integrates the real-time power generation efficiency and predicted irradiance of the BAPV component;

[0009] Based on the Monte Carlo method, simulate the all-weather shadow movement trajectory according to the optimal working angle, and generate a component occlusion influence weight matrix;

[0010] Establish a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameter of the BAPV component; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term;

[0011] Generate a control instruction according to the spacing optimization parameter, and complete the dynamic tuning of the BAPV component according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount, and an abnormal component isolation instruction.

[0012] In some embodiments, the constructing the surface point cloud model for establishing a three-dimensional coordinate system including the installation base point of the BAPV component includes: performing point cloud preprocessing on the scan data corresponding to the three-dimensional laser scan, and adopting a noise point filtering algorithm and a Poisson surface reconstruction algorithm based on the K-D tree to generate a triangular mesh model of the roof surface of the BAPV component; extracting the geometric features of the installation base point of the BAPV component in the triangular mesh model of the roof surface; the geometric features include the center coordinates of the installation bolt positioning holes, the extreme points of the component edge curvature, and the contact surface normal vector; constructing the three-dimensional coordinate system according to the geometric features.

[0013] Exemplarily, the constructing the three-dimensional coordinate system according to the geometric features includes: obtaining the three-dimensional centroid of the installation base point and the direction of the average normal vector of the roof surface according to the geometric features; constructing a local coordinate system with the three-dimensional centroid of the installation base point as the origin of the coordinate system, the Z-axis perpendicular to the direction of the average normal vector of the roof surface, and the X-axis pointing to the true north direction; constructing the three-dimensional coordinate system by fusing the GPS positioning data and the inertial measurement unit data corresponding to the BAPV component in the local coordinate system.

[0014] In some embodiments, calculating the optimal working angle of the BAPV module according to the real-time environmental parameter set by the improved particle swarm optimization algorithm includes: introducing a simulated annealing factor into the velocity update equation of the improved particle swarm optimization algorithm to apply a random perturbation to jump out of the stagnation region when the particle falls into a local optimum; constructing the objective function according to the real-time power generation power, theoretical maximum power, predicted irradiance, and reference irradiance corresponding to the BAPV module; setting an angle constraint penalty term for the improved particle swarm optimization algorithm to increase the penalty factor of the fitness function when the shadow overlap area of adjacent modules exceeds a threshold, completing the configuration of the improved particle swarm optimization algorithm, and inputting the real-time environmental parameter set to calculate the optimal working angle of the BAPV module.

[0015] Exemplarily, before introducing the simulated annealing factor into the velocity update equation of the improved particle swarm optimization algorithm, it further includes: generating an inertia weight adaptive adjustment mechanism corresponding to the BAPV module, and the inertia weight adaptive adjustment mechanism is used to dynamically reduce the weight coefficient corresponding to the improved particle swarm optimization algorithm by using the hyperbolic tangent function when the particle velocity exceeds a threshold.

[0016] In some embodiments, simulating the all-weather shadow movement trajectory according to the optimal working angle by the Monte Carlo method to generate a component occlusion influence weight matrix includes: calculating the solar altitude angle, solar azimuth angle, time series, and wind speed vector parameters corresponding to the BAPV module to generate a dynamic cloud occlusion probability distribution; calculating the shadow projection profile of the BAPV module at the optimal working angle according to the dynamic cloud occlusion probability distribution; obtaining the cumulative time ratio of the BAPV module being occluded according to the shadow projection profile to generate the component occlusion influence weight matrix.

[0017] In some embodiments, establishing the heat balance equation according to the component occlusion influence weight matrix includes: constructing a three-dimensional heat conduction equation according to the component heating power and heat dissipation power of the BAPV module; using the component occlusion influence weight matrix as the boundary condition of the heat flux density distribution; obtaining the convective heat transfer coefficient, component surface emissivity, and surrounding environment view factor of the BAPV module; constructing a forced convection heat dissipation term according to the convective heat transfer coefficient; constructing a radiative heat dissipation term according to the component surface emissivity and surrounding environment view factor; constructing the heat balance equation according to the three-dimensional heat conduction equation, boundary condition, convective heat dissipation term, and radiative heat dissipation term; and solving the heat balance equation by the finite element method to iteratively calculate the spacing optimization parameter that minimizes the temperature gradient of the BAPV module.

[0018] In some embodiments, generating the control instruction according to the spacing optimization parameter includes: inputting the spacing optimization parameter into a preset genetic algorithm optimization model to output the compensation amount of the bracket rotation angle; dynamically adjusting the adjustment amount of the inverter parameters of the inverter corresponding to the BAPV component based on the mapping relationship between the working temperature and the conversion efficiency of the BAPV component; performing impedance spectrum analysis on the BAPV component, and if the characteristics of a hidden crack fault are detected, generating an isolation signal including a topology reconstruction instruction as the abnormal component isolation instruction.

[0019] In a second aspect, the present application provides a roof photovoltaic intelligent monitoring system based on BAPV. The system includes:

[0020] A three-dimensional laser scanning device for scanning the BAPV component;

[0021] An environmental parameter measurement device for measuring the real-time environmental parameter set of the BAPV component in real time; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters;

[0022] A control device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the method provided in any embodiment of the present application when executing the computer program.

[0023] In a third aspect, the present application provides a roof photovoltaic dynamic tuning device based on BAPV, including:

[0024] A coordinate establishment module for constructing a surface point cloud model by three-dimensionally laser scanning a preset BAPV component, and used for establishing a three-dimensional coordinate system including the installation base point of the BAPV component;

[0025] A parameter acquisition module for acquiring the real-time environmental parameter set corresponding to the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters;

[0026] An angle calculation module for calculating the optimal working angle of the BAPV component based on an improved particle swarm algorithm according to the real-time environmental parameter set; the objective function of the improved particle swarm algorithm fuses the real-time power generation efficiency and the predicted irradiance of the BAPV component;

[0027] A matrix generation module for simulating the all-weather shadow movement trajectory based on the Monte Carlo method according to the optimal working angle, and generating a component occlusion influence weight matrix;

[0028] A parameter calculation module, configured to establish a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameters of the BAPV components; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term;

[0029] An optimization completion module, configured to generate a control instruction according to the spacing optimization parameter, and complete the dynamic optimization of the BAPV components according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount, and an abnormal component isolation instruction.

[0030] The present application provides a roof photovoltaic intelligent monitoring system and a dynamic optimization method and device based on BAPV. By three-dimensionally laser scanning the BAPV components to generate a surface point cloud model and accurately establishing a three-dimensional coordinate system including installation base points, the problem of curvature error accumulation in traditional manual or aerial photography modeling is solved. By integrating distributed light intensity, component temperature distribution, ambient temperature, humidity, and wind speed vector parameters, a multi-dimensional real-time environmental parameter set is constructed, breaking through the limitation of a single parameter in static optimization. By integrating real-time power generation efficiency and predicted irradiance in the objective function, the optimal working angle of the BAPV components is dynamically calculated to improve power generation efficiency and environmental adaptability. Based on the optimal angle, the all-weather shadow movement trajectory is simulated, and the occlusion influence weight matrix is generated by combining the random sampling cloud occlusion effect to improve the authenticity of shadow prediction. By introducing the superposition effect of the forced convection heat dissipation term and the radiation heat dissipation term into the heat balance equation and combining the weight matrix to calculate the spacing parameter, the component temperature is prevented from exceeding the limit. According to the spacing optimization parameter, a control set including bracket rotation compensation, inverter parameter adjustment, and abnormal component isolation instructions is generated to realize real-time topology reconstruction and hot spot suppression.

[0031] Through three-dimensionally laser scanning and the point cloud model, the coordinate error of the installation base points is reduced, providing a high-precision spatial reference for subsequent optimization. By integrating multi-dimensional parameters such as wind speed vector, temperature, and humidity, the dynamic coupling problem of heat dissipation and efficiency in traditional static optimization is solved, improving the robustness of the system. By improving the particle swarm algorithm and combining real-time power generation data and irradiance prediction, the component angle is dynamically adjusted to maximize the energy conversion efficiency. By Monte Carlo simulating the random cloud occlusion effect, the weight matrix reflects the real occlusion frequency, reducing local power generation losses. The heat balance equation comprehensively considers forced convection, natural convection, and radiation heat dissipation to prevent the components from overheating and extend the service life. Through real-time abnormal component isolation and topology reconstruction instructions, the spread of hot spots is suppressed, ensuring the overall operation stability of the system.

[0032] In summary, the provided method realizes the full-link dynamic optimization from modeling, parameter calculation to control instruction generation through multi-dimensional environmental parameter dynamic coupling modeling, Monte Carlo random shadow simulation, improved particle swarm algorithm, and heat balance equation optimization, solving the pain points such as modeling error, static optimization limitation, and thermal runaway risk in traditional BAPV systems.

[0033] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of this application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0035] Figure 1 is a schematic block diagram of the structure of a rooftop photovoltaic intelligent monitoring system based on BAPV provided by an embodiment of this application;

[0036] Figure 2 is a schematic structural diagram of a BAPV component provided by an embodiment of this application;

[0037] Figure 3 is a schematic flowchart of the steps of a rooftop photovoltaic dynamic optimization method based on BAPV provided by an embodiment of this application;

[0038] Figure 4 is a schematic block diagram of the structure of a control device provided by an embodiment of this application.

[0039] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and do not limit this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] The following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in this application belong to the scope of protection of this application.

[0041] The flowchart shown in the drawings is only an example, and does not necessarily include all contents and operations / steps, nor does it necessarily execute in the described order. For example, some operations / steps can also be decomposed, combined, or partially merged. Therefore, the actual execution order may change according to the actual situation.

[0042] It should be understood that in order to facilitate the clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. Those skilled in the art can understand that the terms "first" and "second" do not limit the quantity and execution order, and the terms "first" and "second" do not necessarily mean different.

[0043] It should be understood that the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0044] It should also be understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0045] The following will, with reference to the accompanying drawings, elaborate on some embodiments of this application. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0046] In the field of building attached photovoltaic systems (BAPV), traditional 3D modeling uses drone aerial photography or manual measurement, making it difficult to accurately obtain the curvature change of the roof surface and the spatial relationship of the installation base points, resulting in the accumulation of coordinate system establishment errors. At the same time, existing optimization methods mostly perform static optimization based on fixed lighting angles or single temperature parameters, without considering the multi-dimensional influence of wind speed vectors on component heat dissipation and the dynamic coupling effect of environmental temperature and humidity on conversion efficiency. Conventional shadow prediction uses a simplified geometric projection model and does not combine the Monte Carlo random sampling method to simulate the dynamic occlusion effect of clouds, resulting in the weight matrix being unable to reflect the true occlusion frequency. And existing spacing optimization only considers natural convection heat dissipation, ignoring the non-linear superposition effect of forced convection and radiation heat dissipation, resulting in the operating temperature of the components exceeding the safety threshold. Traditional control instruction generation relies on periodic parameter updates and lacks the ability to isolate abnormal components in real time and reconstruct the topology, causing the spread of local hot spot effects.

[0047] Therefore, there is an urgent need for a method to solve at least one of the above problems.

[0048] To solve the above problems, please refer to Figure 1 , this application provides a roof photovoltaic intelligent monitoring system based on BAPV, including: a 3D laser scanning device for scanning BAPV components; an environmental parameter measurement device for real-time measurement of the real-time environmental parameter set of the BAPV components; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters; a control device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the method provided in any embodiment of this application.

[0049] Exemplarily, the control device is configured to execute the following method: constructing a surface point cloud model by three-dimensional laser scanning of a preset BAPV component for establishing a three-dimensional coordinate system including the installation base point of the BAPV component; acquiring the corresponding real-time environmental parameter set of the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters; calculating the optimal working angle of the BAPV component based on the improved particle swarm algorithm according to the real-time environmental parameter set; the objective function of the improved particle swarm algorithm fuses the real-time power generation efficiency and predicted irradiance of the BAPV component; simulating the all-weather shadow movement trajectory based on the Monte Carlo method according to the optimal working angle to generate a component occlusion influence weight matrix; establishing a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameter of the BAPV component; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term; generating a control instruction according to the spacing optimization parameter and completing the dynamic optimization of the BAPV component according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount, and an abnormal component isolation instruction.

[0050] Specifically, the three-dimensional laser scanning device uses a high-precision lidar (LiDAR) to construct a roof surface point cloud model, matches the installation base point through the ICP (Iterative Closest Point) algorithm, and controls the local coordinate system error within ±2 mm to solve the curvature distortion problem of traditional aerial photography modeling.

[0051] The dynamic perception of environmental parameters is achieved by deploying a distributed sensor network (such as a pyranometer light meter, a PT100 temperature array, and an ultrasonic anemometer), collecting multi-dimensional parameters at a frequency of 10 Hz, and forming a real-time environmental parameter set with spatio-temporal correlation.

[0052] The objective function can be designed as F = η * Gpred - λ * ΔT, where η is the real-time conversion efficiency, Gpred is the predicted irradiance based on the atmospheric transmittance model, λ is the temperature penalty coefficient, and ΔT is the temperature change amount corresponding to the BAPV component. By introducing a dynamic inertia weight w(t) = wmax - (wmax - wmin) * t / Tmax, it accelerates convergence to the optimal working angle (within ±5° of the inclination angle). The cloud occlusion simulation is a Poisson point process, generating 10,000 random sampling trajectories, counting the occlusion frequency of each component, and outputting a weight matrix such as Wij = Nshadow / Ntotal; Nshadow is the shadow occlusion frequency, and Ntotal is the total occlusion frequency.

[0053] The equation form corresponding to the heat balance equation modeling can be:

[0054] ;

[0055] where , the forced convection term significantly improves the heat dissipation accuracy under high-temperature conditions.

[0056] Abnormal component isolation detects components with a temperature deviation more than 3σ from the population mean based on the Z-score algorithm, triggers a topology reconstruction instruction, and switches the faulty area to the bypass mode.

[0057] The compensation amount of the bracket rotation angle Δθ = arctan(vwind / vnominal), and the angle is dynamically adjusted in combination with the wind speed vector to avoid wind vibration.

[0058] Exemplarily, the corresponding deployment process of the system may include: generating a digital twin model of the roof by laser scanning, fitting the roof plane equation ax + by + cz + d = 0 through the RANSAC algorithm, and calibrating the component installation reference. Sensor network calibration, fitting the light-temperature cross-sensitivity coefficient by the least squares method to eliminate the measurement coupling error. In the online optimization stage, the PSO parameters are updated every 5 minutes, and the shadow weight matrix is reconstructed every 30 minutes.

[0059] The provided system is applicable to complex scenarios such as corrugated metal roofs and curved glass curtain walls, and its wind pressure resistance performance (able to withstand a typhoon of level 15) has been verified by ANSYS Fluent fluid simulation. It supports linkage with the Building Energy Management System (BEMS) to achieve the goal of Net Zero Energy Building (NZEB). The technical advancement of this system lies in converting discrete environmental parameters into a continuous spatio-temporal field for coupled optimization, and its dynamic tuning mechanism provides a paradigm-level solution for the BAPV field.

[0060] In some embodiments, as Figure 2 shown, the BAPV component includes: the first enamel glass 1, the first encapsulant film 2, the solar cell 3, the second enamel glass 5, and the second encapsulant film 4. The solar cell 3 includes a first side and a second side arranged oppositely. The first enamel glass 1 is adhered to the first side through the first encapsulant film 2, and the second enamel glass 5 is adhered to the second side through the second encapsulant film 4; wherein, the first enamel glass 1 and the second enamel glass 5 can be high-temperature resistant enamel glass.

[0061] Please refer to Figure 3 , Figure 3 is a schematic flowchart of a roof photovoltaic dynamic tuning method based on BAPV provided by an embodiment of the present application. The execution device of the method is the control device of the roof photovoltaic intelligent monitoring system based on BAPV provided by any embodiment of the present application.

[0062] As Figure 3As shown, the provided method includes steps S101 to S106. Among them, the control device can be a handheld terminal, a laptop, a wearable device, a robot, etc. It is used to implement steps S101 to S106 and their corresponding embodiments.

[0063] Step S101. Construct a surface point cloud model by three-dimensional laser scanning a preset BAPV component, for establishing a three-dimensional coordinate system including the installation base points of the BAPV component.

[0064] Specifically, in this step, spatial data of millimeter-level accuracy of the preset BAPV component and its surrounding structures on the roof is collected through three-dimensional laser scanning technology, a point cloud model with geometric details and curvature features is constructed, and a three-dimensional coordinate system of the component installation base points is established based on this. The core innovation lies in eliminating the cumulative error of the coordinate system caused by manual errors or UAV image distortion in traditional measurement methods through the omnidirectional coverage and multi-view data fusion of the laser scanner, ensuring that the input space of the subsequent optimization algorithm has sub-centimeter-level accuracy.

[0065] By using a phase-type laser scanner with a single-point ranging accuracy of ±1 mm, a maximum scanning distance of 350 m, and a point cloud density set to 800 points / m² to capture the minute curvature changes on the roof surface. Target balls (diameter 10 cm) are arranged at the four corners and the center of the roof as reference points for multi-view scan data registration, and the coordinates of the target balls are pre-calibrated to an accuracy of ±0.5 mm by a total station.

[0066] Use the Iterative Closest Point (ICP) algorithm for multi-view point cloud registration. Specifically, the KD-Tree is used to accelerate the nearest neighbor search, and the rotation matrix and translation vector are iteratively calculated until the average registration error of the overlapping area of adjacent point clouds ≤ 2 mm. Voxel filtering (voxel size 5 mm³) and outlier removal (statistical filtering threshold: points outside 1.5 times the standard deviation of the mean distance are removed) are performed on the registered point cloud to generate a noise-free point cloud data set.

[0067] Select the highest point on the roof as the origin (O) of the global coordinate system, the Z-axis is perpendicular to the roof surface (the normal vector is determined by fitting the local point cloud plane), and the X / Y axes extend along the main curvature directions of the roof surface (the directions of the maximum variance are extracted using the Principal Component Analysis PCA). Manually select the three-dimensional coordinates of the installation base points of each BAPV component in the point cloud (such as the center of the bolt hole), and fit the installation plane through the RANSAC algorithm to ensure that the bracket installation surface matches the roof curvature.

[0068] Use Non-Uniform Rational B-Spline (NURBS) surface to fit the point cloud data, and by adjusting the weights of the control points and the knot vectors, a continuous and smooth expression of the roof surface is achieved (the fitting residual ≤ 3 mm). Extract the Gaussian curvature distribution map of the surface to identify the concave and convex areas of the roof (the absolute value of the curvature ≥ The marked area (referred to as the high-curvature area) is used for local mesh encryption in subsequent shadow simulation and heat dissipation analysis.

[0069] The positioning error of the installation base point coordinates is reduced from ±10 mm in the traditional method to ±2 mm, eliminating the risk of bracket installation misalignment caused by coordinate system deviation. The NURBS surface model accurately represents the geometric characteristics of the roof, providing a geometric basis for shadow occlusion simulation and thermal field distribution calculation. The ICP registration and RANSAC fitting algorithms achieve full-process automation, reducing manual intervention and increasing the modeling efficiency by more than 60%.

[0070] Step S102. Obtain the real-time environmental parameter set corresponding to the BAPV components; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters.

[0071] Specifically, in this step, multi-dimensional environmental parameters such as light, temperature, wind speed, and humidity are collected in real time through a distributed sensor network, and a data fusion technology is used to generate a spatio-temporally synchronized environmental parameter matrix, providing dynamic input for subsequent optimization algorithms. The innovation lies in integrating sensors such as photosensitive, infrared, and ultrasonic sensors to achieve full environmental dimension coverage. The sampling frequency is dynamically adjusted according to the parameter change rate (such as sudden wind speed change) to balance data timeliness and system load.

[0072] For example, high-resolution photosensitive sensors are integrated on the backplane of each BAPV component, with a measurement range of 0.01~100 klux, a sampling frequency of 1 Hz, and a spatial resolution of 0.1 W / m². The infrared thermal imager scans the surface of the component at a frame rate of 30 fps, generating a thermal map of the temperature distribution (spatial resolution 5 cm×5 cm, temperature measurement accuracy ±1°C). PT100 platinum resistance sensors (accuracy ±0.1°C) are installed at the component frame and junction box to compensate for the error of infrared temperature measurement affected by the surface reflectivity. Three-dimensional ultrasonic anemometers are deployed at the four corners of the roof to measure the three-dimensional wind speed components (range 0 to 60 m / s, accuracy ±1%), with a sampling frequency of 10 Hz. Digital sensors (temperature accuracy ±0.1°C, humidity ±1.5%RH) are deployed in the ventilation area between the component arrays.

[0073] Kalman filtering (process noise covariance Q = 0.01, observation noise covariance R = 0.1) is applied to the light and temperature data to eliminate instantaneous disturbances. The clocks of all sensors are synchronized based on the NTP protocol, and data with different sampling rates are unified to a 10 Hz timestamp through an interpolation algorithm. A parameter change rate index (such as the instantaneous wind speed change rate dv / dt) is defined. When dv / dt exceeds the threshold (such as 2 m / s²), the wind speed sampling frequency is increased to 20 Hz, and the data is preferentially transmitted to the high-priority data channel. A sliding window mean filter (window width 5 s) is used for high-variance parameters (such as irradiance), and a median filter is used for low-variance parameters (such as humidity).

[0074] Cover multiple physical fields such as light, heat, flow field, and humidity. The parameter acquisition dimension increases by 300% compared with traditional methods. The transmission delay of key parameters (such as wind speed) is ≤50 ms, meeting the real-time optimization requirements. Through multi-sensor redundancy and fusion algorithms, the data confidence level is increased to 99.8%.

[0075] Step S103. Calculate the optimal working angle of the BAPV component according to the real-time environmental parameter set based on the improved particle swarm algorithm; the objective function of the improved particle swarm algorithm fuses the real-time power generation efficiency and predicted irradiance of the BAPV component.

[0076] Specifically, this step proposes an improved particle swarm algorithm (IPSO), which calculates the optimal working angle of the BAPV component in real time through dynamic inertia weight, constraint handling mechanism, and multi-objective function design. The core innovations include: Objective function fusion: Combine real-time power generation efficiency and future irradiance prediction to balance short-term benefits and long-term performance. Constraint adaptability: Convert hard constraints such as mechanical limits of the bracket and occlusion relationship between components into the boundaries of the search space to avoid invalid solutions.

[0077] The objective function design can be f(θ)=α*Preal+(1-α)*Ppred; Preal is the current power generation efficiency, calculated by dividing the component output power (voltage × current) by the theoretical maximum power (under STC conditions). α is the dynamic weight coefficient, which is adjusted according to environmental stability (such as 0.7 when the wind speed is stable and 0.9 when there is a mutation to prioritize the response to the real-time state).

[0078] Dynamic inertia weight can include , where is the gradient of environmental parameter change (such as the temperature change rate), and β is the attenuation coefficient (default 0.1). When the environment changes violently, the inertia weight is reduced to accelerate convergence.

[0079] Constraint handling can include directly using the bracket rotation angle limit (such as ±45°) as the particle position boundary. The occlusion relationship between components is pre-calculated through a geometric projection model to generate a feasible solution domain, and the particles only search within this domain.

[0080] By setting the particle swarm size to 50, the maximum number of iterations to 100, and the convergence condition as the change in the optimal solution <0.1% for 10 consecutive iterations. Using GPU acceleration (NVIDIA Jetson AGX Xavier), the single optimization time consumption is <30 seconds.

[0081] Through the above, the angle adjustment delay under environmental mutations (such as cloud occlusion) is <1 minute, avoiding a sudden drop in power generation. By constraint handling, mechanical overrun is avoided, and the bracket failure rate is reduced by 70%.

[0082] Step S104. Based on the Monte Carlo method, simulate the all-weather shadow movement trajectory according to the optimal working angle, and generate a component occlusion influence weight matrix.

[0083] Specifically, this step is based on the Monte Carlo random sampling method to simulate the shadow influence of cloud dynamic occlusion and solar trajectory changes on the components, and generate an occlusion frequency weight matrix in the time-space dimension. The innovation points are as follows: Cloud random process modeling: Use the Poisson process to simulate the cloud appearance probability, and combine historical meteorological data to correct the model parameters. Dynamic weight update: Resample every 15 minutes and link with the real-time weather API (such as OpenWeatherMap) to adjust the cloud density.

[0084] For solar position calculation, if the SPA (Solar Position Algorithm) is used, input the GPS coordinates, timestamp, and altitude to calculate the solar altitude angle and azimuth angle (accuracy ±0.01°).

[0085] The Poisson process parameter is that the cloud arrival time interval follows an exponential distribution, and the density function is obtained by training with historical meteorological data (for example, the density function in a certain area in summer = 0.2 clouds / min). Randomly generate cloud size (uniform distribution U[1,10] m²), moving speed (normal distribution N(3,0.5) m / s), and light transmittance (Beta distribution α = 2, β = 5).

[0086] Based on the Monte Carlo simulation, conduct 10,000 random samplings for all-weather (such as 06:00 - 18:00). In each sampling: Generate a cloud sequence according to the Poisson process. Calculate the shadow coverage area of each component at each time point (based on the solar position and cloud projection).

[0087] By statistically counting the occlusion frequency of each component within each 15-minute time window, generate the weight matrix Wij (i = component number, j = time window index). Connect to the real-time weather API. If it is detected that the actual cloud amount exceeds the predicted value by 20%, trigger the recalculation of the weight matrix (the number of sampling times increases to 20,000 times).

[0088] The prediction error rate of the occlusion frequency is reduced from 20% of the traditional geometric model to less than 3%. Through the dynamic correction mechanism, the prediction stability of the model in extreme weather is improved by 40%.

[0089] Step S105. Establish a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameters of the BAPV components; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term.

[0090] Specifically, in this step, a three-dimensional unsteady-state heat balance equation including forced convection, natural convection, and radiative heat dissipation is established, and the component spacing parameters are optimized by combining with the occlusion weight matrix to ensure that the operating temperature is lower than the safety threshold. The innovation points include: Multi-physics field coupling modeling: Considering wind speed vector, component tilt angle, and spacing as variables to analyze their non-linear effects on heat dissipation. Safety constraint embedding: Taking the IEC 61215 standard (component temperature ≤ 85°C) as a hard constraint and using the gradient descent method to solve for the optimal spacing.

[0091] The heat balance equation can include:

[0092] ;

[0093] ρCp and k are the material density, specific heat capacity, and thermal conductivity of the BAPV component respectively. is the heat generation power density, representing the heat generated per unit volume, defined by the product of the occlusion weight matrix and the power loss of the solar cell. is the forced convection heat dissipation coefficient, obtained by multiplying the convective heat transfer coefficient by the difference between the temperature of the BAPV component and the ambient temperature. is the radiative heat dissipation (corrected according to the Stefan-Boltzmann law), obtained by multiplying the product of the surface emissivity (the radiation ability of the material surface, taking 0.85), the Stefan-Boltzmann constant (the blackbody radiation constant), and the environmental view factor (the proportion of the radiative heat transfer perspective between the surface and the environment, taking 0.6) by the difference between the fourth power of the temperature of the BAPV component and the fourth power of the effective sky temperature (the equivalent temperature of the atmospheric long-wave radiation).

[0094] The finite volume method (FVM) is used to spatially discretize the computational domain (component array), and the grid size is 5 cm × 5 cm × 5 cm.

[0095] The occlusion weight matrix is converted into an equivalent heat source term (the occluded component Q = 0). The conjugate gradient method is used to solve the linear equations, with the optimization variable being the component spacing d and the objective function being the maximum temperature difference ΔTmax. The optimization process includes setting the initial spacing to the industry standard value (such as 2 times the component height). Iteratively adjust d until ΔT_max ≤ 3°C and T_max ≤ 85°C are satisfied. Output the optimal spacing parameter dopt and the corresponding tilt angle compensation amount (the component angle needs to be slightly adjusted due to the change in spacing).

[0096] The maximum temperature difference of the component is compressed from ±10°C in the traditional scheme to ±3°C, and the hot spot incidence rate is reduced by 90%. The operating temperature is reduced by 5 - 8°C, and the component efficiency decay rate decreases by 2% per year.

[0097] Step S106. Generate control instructions according to the spacing optimization parameters, and complete dynamic tuning of the BAPV components according to the control instructions; the control instructions include bracket rotation angle compensation, inverter parameter adjustment and abnormal component isolation instructions.

[0098] Specifically, this step generates a control instruction set based on the optimization results, including bracket rotation compensation, inverter parameter adjustment, and abnormal component isolation, and achieves low-latency response through edge computing nodes. The innovation lies in: Real-time isolation of abnormal components: Based on IV curve distortion detection and topology reconstruction algorithm, hot spot effect is prevented from spreading. Instruction collaborative optimization: The bracket angle, spacing and inverter MPPT parameters are adjusted in conjunction to achieve global optimization.

[0099] Bracket rotation compensation can convert the optimal angle θ_opt output in step S103 into the number of stepper motor pulses (e.g. 1° = 200 pulses), and superimpose the PID control algorithm to compensate for mechanical vibration (proportional coefficient Kp = 0.8, Ki = 0.2, Kd = 0.1). Inverter parameter adjustment can dynamically correct the MPPT operating point according to the component temperature and irradiance (e.g. V_mpp is reduced by 0.3% for every 1°C increase in temperature). Abnormal component isolation can monitor the IV curve of the string in real time. If the current of a component drops by 30% and lasts for 5 seconds, the bypass diode is triggered to turn on, and the string topology is reconstructed at the same time (e.g. 3 strings and 10 parallels are changed to 4 strings and 7 parallels + 1 redundancy).

[0100] Edge computing deployment can use Raspberry Pi Compute Module 4 as an edge node to run a lightweight control model (TensorFlow Lite deployment), with a command generation delay of <50 ms. The data link includes sensor → edge node (processing control instructions) → cloud (long-term data analysis), and dual-link redundancy ensures communication reliability. When the temperature exceeds 85°C, the inverter output power is forced to be reduced to 50%, and an audible and visual alarm is triggered.

[0101] Generate daily health reports, predict component life (based on Arrhenius model) and maintenance recommendations.

[0102] The system availability rate has been increased from 92% of the traditional system to 98.5%, reducing power generation losses. The frequency of component replacement has been reduced by 40% through early isolation of abnormalities. The hot spot suppression action delay is less than 2 seconds, avoiding the risk of fire caused by local overheating.

[0103] This method has been applied to a 2.5 MW rooftop BAPV project in an industrial park. Empirical data shows that: the annual power generation has increased by 19.3% and reached 1,892 MWh. The maximum temperature of the components has dropped from 91°C to 79°C, and the operation and maintenance cost has decreased by 37%. The system payback period has been shortened from 6.2 years to 4.8 years. The above embodiments confirm the significant technological progress and economic benefits of the present invention, and can be widely applied to BAPV scenarios such as industrial and commercial buildings, and agricultural-photovoltaic complementary.

[0104] In some embodiments, the construction of the surface point cloud model for establishing a three-dimensional coordinate system including the installation base points of BAPV components includes: performing point cloud preprocessing on the scan data corresponding to the three-dimensional laser scan, and using a noise point filtering algorithm based on K-D tree and a Poisson surface reconstruction algorithm to generate a triangular mesh model of the rooftop surface of the BAPV components; extracting the geometric features of the installation base points of the BAPV components in the triangular mesh model of the rooftop surface; the geometric features include the center coordinates of the installation bolt positioning holes, the extreme points of the component edge curvature, and the contact surface normal vector; constructing the three-dimensional coordinate system according to the geometric features.

[0105] This embodiment details a high-precision three-dimensional coordinate system construction method based on three-dimensional laser scan point cloud data, specifically including four stages: point cloud preprocessing, Poisson surface reconstruction, geometric feature extraction, and coordinate system fusion, for accurately calibrating the spatial position and attitude of the installation base points of BAPV components.

[0106] The scan data is collected by using a phase-type laser scanner to perform a full-station scan on the rooftop surface. The scan density is set to 1200 points / m², and the scan radius covers a range of 5 m outside the rooftop extension to ensure the data integrity of the BAPV component installation area and its surrounding accessory structures (such as ventilation ducts, parapets).

[0107] Six target balls (diameter 8 cm) are pre-placed on the rooftop surface, and their absolute coordinates are pre-determined by a total station (Trimble S7) with an accuracy of ±0.3 mm for subsequent multi-viewpoint cloud registration.

[0108] Construct a K-D tree spatial index for the original point cloud, set the search radius r = 10 cm, count the distance distribution of the k-nearest neighbors (k = 50) of each point, and calculate the distance mean μ and standard deviation σ. If the average distance of the k-nearest neighbors of a certain point exceeds the threshold (μ + 3σ), it is determined as a noise point and removed. After this processing, the point cloud noise rate is reduced from the initial 15% to below 0.2%. Use a voxel grid filter, set the voxel size to 3 mm × 3 mm × 3 mm, and unify the point cloud density to 1000 points / m² to eliminate the density unevenness problem caused by scanning perspective differences.

[0109] The Poisson reconstruction parameter settings include the input filtered point cloud data. Set the solution depth of the Poisson equation to 10, the maximum side length of the reconstructed surface to 2 cm, and generate a closed triangular mesh model (the number of vertices is about 5×10 6 , and the number of facets is about 1×10 7 ). Perform Laplacian smoothing on the reconstructed surface (the number of iterations is 3 times, and the smoothing factor is 0.5) to eliminate stepped artifacts and ensure continuous surface curvature. Perform Edge Collapse and Vertex Split operations to compress the number of mesh facets to 30% of the original data while keeping the curvature feature error ≤ 0.1 mm.

[0110] In the triangular mesh model, manually select the bolt hole area (a circular hole with a diameter of 2 cm), and use the Random Sample Consensus (RANSAC) algorithm to fit a cylindrical model: randomly select 3 points to calculate the axis direction and radius of the cylinder, and set the inlier threshold to 0.5 mm. Iterate 1000 times, select the cylindrical model with the largest number of inliers, and take the midpoint of its axis as the center coordinate of the positioning hole.

[0111] Calculate the Gaussian curvature of the triangular mesh vertices, set the threshold K = 0.25 m⁻¹, extract the curvature extreme points (i.e., the corner points of the component edge), and form a closed contour line. Simplify the contour line through the Douglas - Peucker algorithm, retain the feature points with a curvature change ≥ 10%, and generate a set of component edge feature points.

[0112] The contact surface normal vector is calculated by selecting a local area of 100×100 mm² on the component installation surface and using Principal Component Analysis (PCA) to calculate the eigenvector corresponding to the minimum eigenvalue of the point cloud covariance matrix in this area as the average normal vector of the contact surface.

[0113] Set the three - dimensional centroid of the installation base point (calculated by weighted averaging the center coordinates of all bolt holes, with the weight being the square of the diameter of each hole) as the origin O_local of the local coordinate system. Axial calibration includes: Z - axis: perpendicular to the average normal vector of the roof surface (obtained by weighted averaging the contact surface normal vectors, with the weight being the proportion of the area of each region). X - axis: pointing to the true north direction (measured by the electronic compass HMC5883L, with an accuracy of ±1°). Y - axis: determined by the cross - product of the Z - axis and the X - axis according to the right - hand rule.

[0114] The GPS positioning data is accessed by deploying a dual-frequency GPS module (ublox ZED-F9P) at the center of the roof, which outputs the longitude and latitude (accuracy ±1 cm) and altitude (accuracy ±2 cm) in the WGS84 coordinate system. The inertial measurement unit (IMU) is calibrated by installing a 6-axis IMU (BMI160) at the origin of the local coordinate system to measure the three-axis acceleration (±16g) and angular velocity (±2000° / s) in real time. The GPS data is fused through Kalman filtering to compensate for the coordinate system drift caused by the roof vibration.

[0115] A transformation matrix T_local→global from the local coordinate system to the WGS84 coordinate system is established, which includes a translation vector t (determined by the GPS positioning data) and a rotation matrix R (obtained by converting the Euler angles output by the IMU). The parameters of the transformation matrix are optimized by the Levenberg-Marquardt algorithm to make the projection error of the feature points of the BAPV components in the global coordinate system ≤3 cm.

[0116] The positioning accuracy of the installation base point coordinates reaches ±1.5 mm (generally ±10 mm for traditional methods), eliminating the risk of bracket installation misalignment caused by coordinate system deviation. The global coordinate system fusion error ≤3 cm, meeting the requirements for large-scale cross-regional collaborative optimization of the BAPV system. The Poisson surface reconstruction algorithm accurately restores the tiny undulations on the roof surface (residual ≤0.5 mm), avoiding the problem of stress concentration on the component contact surface caused by the traditional plane assumption.

[0117] Exemplarily, constructing the three-dimensional coordinate system according to the geometric features includes: obtaining the three-dimensional centroid of the installation base point and the direction of the average normal vector of the roof surface according to the geometric features; constructing a local coordinate system with the three-dimensional centroid of the installation base point as the origin of the coordinate system, the Z-axis perpendicular to the direction of the average normal vector of the roof surface, and the X-axis pointing to the true north direction; and constructing the three-dimensional coordinate system by fusing the GPS positioning data and inertial measurement unit data corresponding to the BAPV components in the local coordinate system.

[0118] When constructing a three-dimensional coordinate system, firstly, the geometric feature data obtained by three-dimensional laser scanning is used to calculate the arithmetic mean of the center coordinates of all mounting bolt positioning holes as the three-dimensional centroid of the mounting base point. The normal vector data set of the triangular mesh model of the roof surface is extracted using the principal component analysis method, and its average normal vector direction is calculated as the reference direction of the average normal vector of the roof surface. The Z axis is set to be perpendicular to the direction of the average normal vector (i.e. parallel to the inclined surface of the roof), and a local coordinate system is established with the three-dimensional centroid of the mounting base point as the origin. The X axis is calibrated to point to the geographic north direction through an electronic compass, and the Y axis is automatically generated according to the right-hand rule. Subsequently, the longitude and latitude coordinates collected by the GPS module are converted into plane rectangular coordinates through UTM projection, and the attitude angle data provided by the inertial measurement unit (IMU) are fused by Kalman filtering to finally generate a three-dimensional coordinate system with absolute geographic coordinate attributes. The geometric accuracy of the coordinate system is significantly improved through center of mass positioning and normal vector orthogonalization processing; combined with GPS / IMU multi-source data fusion, the coordinate system has millimeter-level positioning accuracy and anti-vibration interference capabilities; the standardized coordinate axis definition method can adapt to architectural scenes with different roof inclinations and azimuths, providing a precise spatial reference for subsequent shadow trajectory simulation.

[0119] For example, when constructing a three-dimensional coordinate system, the center coordinate data set of the mounting bolt locating hole obtained by three-dimensional laser scanning is first preprocessed to remove outliers caused by scanning noise (using the 3σ criterion for filtering). The three-dimensional centroid of the installation base point is determined by calculating the weighted average of all valid locating hole coordinates, where the weight is the area ratio of the triangular mesh where each locating hole is located. The principal component analysis (PCA) dimensionality reduction processing is used for the normal vector data set of the triangular mesh model of the roof surface: the normal vectors of all triangular facets are extracted to form an n×3 matrix, the covariance matrix is ​​calculated and its eigenvalues ​​and eigenvectors are solved, and the eigenvector direction corresponding to the maximum eigenvalue is used as the average normal vector direction of the roof surface (with an accuracy of 0.01°).

[0120] The Z axis is set to be orthogonal to the average normal vector direction to ensure that the Z axis is parallel to the actual tilt plane of the roof. The X axis is calibrated using a high-precision electronic compass (HMC5883L chip) to measure the direction of the geomagnetic North Pole in real time, and the inclination compensation algorithm is used to eliminate the influence of the roof tilt on the magnetometer reading, and finally determine that the X axis points to the true north direction (accuracy ±0.3° after error compensation). The Y axis is automatically generated according to the right-hand rule to complete the preliminary construction of the local coordinate system.

[0121] To fuse the absolute geographic coordinates, a dual-frequency GPS module is used to obtain the WGS84 longitude and latitude coordinates of the installation base point, and it is converted into a plane rectangular coordinate through the Universal Transverse Mercator projection. At the same time, a six-axis inertial measurement unit is used to collect the attitude angle data (pitch angle, roll angle, yaw angle) of the coordinate system in real time, and the GPS plane coordinates and IMU attitude data are data-fused through a Kalman filter (state equation noise covariance Q = 0.01, observation noise covariance R = 0.1), and finally a three-dimensional coordinate system with geographical reference is generated. The coordinate system transformation matrix is expressed as:

[0122] ;

[0123] wherein, Rz(θ), , Rx(ψ) are the rotation matrices around the Z, Y, and X axes respectively, corresponding to the yaw angle, pitch angle, and roll angle.

[0124] In some embodiments, the calculating the optimal working angle of the BAPV component according to the real-time environment parameter set by the improved particle swarm algorithm includes: introducing a simulated annealing factor into the velocity update equation of the improved particle swarm algorithm to apply a random perturbation to jump out of the stagnation area when the particle falls into the local optimum; constructing the objective function according to the real-time generated power, theoretical maximum power, predicted irradiance, and reference irradiance corresponding to the BAPV component; setting an angle constraint penalty term for the improved particle swarm algorithm to increase the penalty factor of the fitness function when the shadow overlap area of adjacent components exceeds the threshold, completing the configuration of the improved particle swarm algorithm, and calculating the optimal working angle of the BAPV component by inputting the real-time environment parameter set.

[0125] In the improved particle swarm optimization algorithm, based on the velocity update equation \(v_i(t + 1)=wv_i(t)+c1r1(pbest_i - x_i)+c2r2(gbest - x_i)\), a simulated annealing factor \(\eta=\exp(-\Delta f / T)\) is introduced, where \(\Delta f\) is the difference between the current particle fitness and the global optimum, and \(T\) is the annealing temperature parameter. When the gbest has not been updated for 5 consecutive iterations, a perturbation mechanism is triggered: a Gaussian noise term \(\eta\cdot N(0,\sigma)\) is added to the velocity term, and \(\sigma\) decays linearly with the number of iterations. The objective function is constructed as \(F = \alpha(P_{actual} / P_{max})+\beta(G_{predict} / G_{ref})\), where \(\alpha+\beta = 1\) is the weight coefficient, \(P_{actual}\) is the real-time power generation, \(P_{max}\) is the theoretical maximum power, \(G_{predict}\) is the predicted irradiance, and \(G_{ref}\) is the AM1.5 standard irradiance. The angle constraint penalty term is set such that when the shadow overlap area \(S_{overlap}>S_{threshold}\) between adjacent components, the fitness function is corrected to \(F'=F-\gamma*(S_{overlap} / S_{threshold})\), where \(\gamma\) is the penalty coefficient.

[0126] The perturbation mechanism with dynamically adjusted simulated annealing factor increases the probability of the algorithm jumping out of the local optimum by 42%; the dual-objective weighted function balances the current efficiency and future irradiance trend, improving the time adaptability of the angle optimization result by 35%; the shadow overlap penalty term effectively avoids the hot spot effect caused by component occlusion, and experiments show that the component failure rate is reduced by 28%.

[0127] Exemplarily, before introducing the simulated annealing factor in the velocity update equation of the improved particle swarm optimization algorithm, it further includes: generating an inertia weight adaptive adjustment mechanism corresponding to the BAPV component, and the inertia weight adaptive adjustment mechanism is used to dynamically reduce the weight coefficient corresponding to the improved particle swarm optimization algorithm using the hyperbolic tangent function when the particle velocity exceeds the threshold.

[0128] The specific implementation of the inertia weight adaptive adjustment mechanism is: real-time monitoring of the particle velocity modulus \(\|v_i\|\), when it exceeds the threshold \(v_{max}\), the weight is dynamically adjusted using the hyperbolic tangent function \(w = w_{max}-(w_{max}-w_{min})\cdot anh(k\cdot(\|v_i\| / v_{max}))\), where \(k\) is the curvature coefficient (taking 2.5). In the algorithm initialization stage, \(w_{max}=0.9\), \(w_{min}=0.4\), and \(v_{max}=\pi / 6\ rad / s\). When the particle velocity approaches \(v_{max}\), the weight \(w\) smoothly decreases from 0.9 to 0.4, ensuring that high-speed particles perform refined local searches and low-speed particles maintain global exploration capabilities.

[0129] This mechanism increases the convergence speed of the algorithm by 19% and reduces the optimization error of the standard test function to 0.023 rad. The smooth characteristic of the hyperbolic tangent function avoids the oscillation phenomenon caused by the sudden change of weights. During the actual measurement of the photovoltaic array, the number of angle adjustments is reduced by 33%, and the mechanical loss decreases significantly.

[0130] In some embodiments, the Monte Carlo method is used to simulate the all-weather shadow movement trajectory according to the optimal working angle and generate the component occlusion influence weight matrix, including: calculating the solar altitude angle, solar azimuth angle, time series, and wind speed vector parameters corresponding to the BAPV components to generate the dynamic cloud occlusion probability distribution; calculating the shadow projection contour of the BAPV components at the optimal working angle according to the dynamic cloud occlusion probability distribution; obtaining the proportion of the cumulative occlusion time of the BAPV components according to the shadow projection contour to generate the component occlusion influence weight matrix.

[0131] The Monte Carlo method is used to perform one million random samplings, and calculate the solar altitude angle α = arcsin(sinδsinφ + cosδcosφcosω) and azimuth angle γ = arctan(sinω / (cosωsinφ - tanδcosφ)) per hour, where δ is the solar declination angle, φ is the latitude, and ω is the hour angle. The dynamic cloud occlusion probability is predicted by training an LSTM network with historical meteorological data, and the occlusion probability distribution on the time series is output. Substitute the optimal working angle into the three-dimensional coordinate system for coordinate transformation, calculate the projection coordinates of the component edge vertices on the horizontal plane, and generate the shadow polygon through the convex hull algorithm. Statistically calculate the proportion of the duration of each component covered by adjacent shadows, and construct an n×n weight matrix W, where W_ij represents the time weight coefficient of the i-th component occluded by the j-th component.

[0132] The Monte Carlo simulation enables the shadow prediction accuracy to reach the minute-level resolution, and the calculation error of the component occlusion time is <3%. The weight matrix quantifies the degree of occlusion influence between components, provides data support for the spacing optimization, and the actual measurement shows that the annual average power generation loss of the component array is reduced by 17%.

[0133] In some embodiments, establishing the heat balance equation according to the component occlusion influence weight matrix includes: constructing a three-dimensional heat conduction equation according to the component heating power and heat dissipation power of the BAPV components; using the component occlusion influence weight matrix as the boundary condition of the heat flux density distribution; obtaining the convective heat transfer coefficient, component surface emissivity, and surrounding environment view factor of the BAPV components; constructing a forced convection heat dissipation term according to the convective heat transfer coefficient; constructing a radiation heat dissipation term according to the component surface emissivity and surrounding environment view factor; constructing the heat balance equation according to the three-dimensional heat conduction equation, boundary condition, convective heat dissipation term, and radiation heat dissipation term; and solving the heat balance equation by the finite element method to iteratively calculate the spacing optimization parameters that minimize the temperature gradient of the BAPV components.

[0134] By establishing the three-dimensional heat conduction equation of the BAPV module , where Q_gen = ηP_max(1 - T / T_ref) is the heat generation power of the module, and Q_loss includes the forced convection term h_conv(T - T_amb) and the radiation term . After normalizing the occlusion influence weight matrix W, it is used as the heat flux density boundary condition. The surface heat flux density q of the occluded area is q = W_ij * G_predict * α_abs. COMSOL Multiphysics is used for finite element solution, setting the air velocity boundary condition and the radiation view factor, and iteratively calculating the temperature field distribution at different spacings. The optimization goal is to make the maximum temperature gradient ∇T_max ≤ 5 °C / m, and finally output the minimum spacing parameter of the module that satisfies the heat balance. This heat balance model accurately predicts the operating temperature of the module (error < 1.5 °C). The optimized spacing parameter reduces the temperature difference of the module by 42%, and the probability of hot spot formation drops to 0.3%; the collaborative consideration of the forced convection and radiation terms improves the heat dissipation efficiency by 28%, and extends the service life of the module by 3 - 5 years.

[0135] In some embodiments, generating the control instruction according to the spacing optimization parameter includes: inputting the spacing optimization parameter into a preset genetic algorithm optimization model to output the bracket rotation angle compensation amount; dynamically adjusting the inverter parameter adjustment amount of the inverter corresponding to the BAPV module based on the mapping relationship between the operating temperature and the conversion efficiency of the BAPV module; performing impedance spectrum analysis on the BAPV module, and if a hidden crack fault feature is detected, generating an isolation signal including a topology reconstruction instruction as the abnormal module isolation instruction.

[0136] The genetic algorithm optimization model uses real number coding, the population size is set to 200, the crossover probability is 0.8, and the mutation probability is 0.05. The spacing optimization parameter is encoded as a chromosome, and the fitness function comprehensively considers the power generation efficiency improvement rate and the mechanical adjustment energy consumption ratio. After 50 generations of iteration through the tournament selection strategy, the optimal bracket rotation angle compensation amount is output, with an accuracy of 0.1 °. The inverter parameter adjustment is based on the temperature - efficiency curve η(T) = η_ref[1 - β(T - T_ref)], β = 0.0045 / °C. When the infrared temperature measurement shows that the module temperature > 65 °C, the MPPT operating point voltage is automatically adjusted to decrease by 2%. The impedance spectrum analysis scans in the frequency band of 1 kHz - 1 MHz. When the phase angle mutation at the characteristic frequency point is detected to be > 5 °, it is determined as a hidden crack fault, and the topology reconstruction is performed through the intelligent circuit breaker and the matrix switch to isolate the faulty module.

[0137] Genetic algorithm optimization reduces the energy consumption of bracket adjustment by 41%; the temperature-adaptive MPPT adjustment strategy improves the power generation efficiency by 12% in high-temperature environments; impedance spectroscopy detection can accurately identify (with an accuracy of 98.7%) within 30 minutes of the occurrence of hidden cracks, effectively preventing the spread of hot spots, and increasing the system availability to 99.92%.

[0138] The embodiment of the present application also provides a roof photovoltaic dynamic tuning device based on BAPV. The roof photovoltaic dynamic tuning device based on BAPV is used to execute the steps of the roof photovoltaic dynamic tuning method shown in the above embodiments. The roof photovoltaic dynamic tuning device based on BAPV can be a single server or a server cluster, or the roof photovoltaic dynamic tuning device based on BAPV can be a terminal, and the terminal can be a handheld terminal, a laptop computer, a wearable device, or a robot, etc.

[0139] The roof photovoltaic dynamic tuning device based on BAPV includes:

[0140] A coordinate establishment module, configured to construct a surface point cloud model by three-dimensional laser scanning of a preset BAPV component, for establishing a three-dimensional coordinate system including the installation base points of the BAPV component;

[0141] A parameter acquisition module, configured to acquire a set of real-time environmental parameters corresponding to the BAPV component; the set of real-time environmental parameters includes distributed light intensity parameters, component temperature distribution parameters, ambient temperature and humidity parameters, and wind speed vector parameters;

[0142] An angle calculation module, configured to calculate the optimal working angle of the BAPV component based on the improved particle swarm algorithm according to the set of real-time environmental parameters; the objective function of the improved particle swarm algorithm fuses the real-time power generation efficiency and predicted irradiance of the BAPV component;

[0143] A matrix generation module, configured to simulate the all-weather shadow movement trajectory based on the Monte Carlo method according to the optimal working angle, and generate a component occlusion influence weight matrix;

[0144] A parameter calculation module, configured to establish a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameters of the BAPV component; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term;

[0145] A tuning completion module, configured to generate a control instruction according to the spacing optimization parameter, and complete the dynamic tuning of the BAPV component according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount, and an abnormal component isolation instruction.

[0146] It should be noted that those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described BAPV-based rooftop photovoltaic dynamic optimization device and each unit can refer to the corresponding processes in the embodiments of the BAPV-based rooftop photovoltaic dynamic optimization method described above, and will not be elaborated here.

[0147] The above-mentioned BAPV-based rooftop photovoltaic dynamic optimization method is implemented in the form of a computer program, and this computer program can run on the above-mentioned device.

[0148] Please refer to Figure 4 , Figure 4 which is a schematic block diagram of the structure of the control device provided by the embodiment of the present application. The control device includes a processor, a memory, and a network interface connected through a device bus. Among them, the memory can include a storage medium and an internal memory.

[0149] The storage medium can store an operating device and a computer program. This computer program includes program instructions, and when the program instructions are executed, the processor can execute any embodiment of the BAPV-based rooftop photovoltaic dynamic optimization method.

[0150] The processor is used to provide computing and control capabilities to support the operation of the entire control device.

[0151] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When this computer program is executed by the processor, the processor can execute any BAPV-based rooftop photovoltaic intelligent monitoring system method.

[0152] This network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that Figure 4 the structure shown in

[0153] It should be understood that the processor can be a Central Processing Unit (CPU), and the processor can also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0154] Among them, in one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0155] Construct a surface point cloud model by three-dimensional laser scanning of a preset BAPV component, which is used to establish a three-dimensional coordinate system including the installation base point of the BAPV component.

[0156] Obtain the real-time environmental parameter set corresponding to the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters.

[0157] Calculate the optimal working angle of the BAPV component based on the improved particle swarm algorithm according to the real-time environmental parameter set; the objective function of the improved particle swarm algorithm fuses the real-time power generation efficiency and predicted irradiance of the BAPV component.

[0158] Simulate the all-weather shadow movement trajectory based on the Monte Carlo method according to the optimal working angle, and generate a component occlusion influence weight matrix.

[0159] Establish a heat balance equation according to the component occlusion influence weight matrix to calculate the spacing optimization parameter of the BAPV component; the heat balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term.

[0160] Generate a control instruction according to the spacing optimization parameter, and complete the dynamic tuning of the BAPV component according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount, and an abnormal component isolation instruction.

[0161] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described processor can refer to the corresponding process in the method embodiments described in the above various embodiments, and will not be elaborated here.

[0162] An embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. The processor executes the program instructions to implement the steps of the method for dynamically optimizing rooftop photovoltaics based on BAPV provided in the above embodiments of the present application.

[0163] Among them, the computer-readable storage medium may be an internal storage unit of the control device described in the foregoing embodiment, such as the hard disk or memory of the control device. The computer-readable storage medium may also be an external storage device of the control device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the control device.

[0164] As described above, the above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or replacements, and these modifications or replacements should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A rooftop photovoltaic dynamic tuning method based on BAPV, characterized in that: include: By scanning the preset BAPV components with 3D laser, a surface point cloud model is constructed to establish a 3D coordinate system including the installation base point of the BAPV components; Acquire a real-time environmental parameter set corresponding to the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters; Based on the improved particle swarm algorithm, the optimal working angle of the BAPV component is calculated according to the real-time environmental parameter set; the objective function of the improved particle swarm algorithm integrates the real-time power generation efficiency and predicted irradiance of the BAPV component; Based on the Monte Carlo method, the all-weather shadow movement trajectory is simulated according to the optimal working angle to generate a component shading impact weight matrix; A thermal balance equation is established according to the component shielding influence weight matrix to calculate the spacing optimization parameters of the BAPV components; the thermal balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term; A control instruction is generated according to the spacing optimization parameter, and dynamic tuning of the BAPV component is completed according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount and an abnormal component isolation instruction.

2. The method according to claim 1, characterized in that The surface point cloud model is constructed to establish a three-dimensional coordinate system including the BAPV component installation base point, including: The point cloud preprocessing of the scanning data corresponding to the 3D laser scanning is performed, and the noise point filtering algorithm based on the KD tree and the Poisson surface reconstruction algorithm are used to generate the roof surface triangular mesh model of the BAPV component; Extracting geometric features of the BAPV component installation base point in the roof surface triangular mesh model; the geometric features include the center coordinates of the installation bolt positioning hole, the component edge curvature extreme point and the contact surface normal vector; The three-dimensional coordinate system is constructed according to the geometric features.

3. The method according to claim 2, characterized in that The step of constructing the three-dimensional coordinate system according to the geometric features comprises: According to the geometric features, the three-dimensional centroid of the installation base point and the average normal vector direction of the roof surface are obtained; A local coordinate system is constructed with the three-dimensional centroid of the installation base point as the origin of the coordinate system, the Z axis being perpendicular to the direction of the average normal vector of the roof surface, and the X axis pointing to the geographic north direction; The three-dimensional coordinate system is constructed by fusing the GPS positioning data and the inertial measurement unit data corresponding to the BAPV component in the local coordinate system.

4. The method according to claim 1, characterized in that: The method of calculating the optimal working angle of the BAPV component according to the real-time environmental parameter set based on the improved particle swarm algorithm includes: Introducing a simulated annealing factor into the speed update equation of the improved particle swarm algorithm, so as to apply random disturbance to jump out of the stagnant area when the particles fall into the local optimum; Constructing the objective function according to the real-time power generation, theoretical maximum power, predicted irradiance and reference irradiance corresponding to the BAPV assembly; The angle constraint penalty item of the improved particle swarm algorithm is set to increase the penalty factor of the fitness function when the overlapping area of ​​shadows of adjacent components exceeds a threshold, and the configuration of the improved particle swarm algorithm is completed to input the real-time environmental parameter set to calculate the optimal working angle of the BAPV component.

5. The method according to claim 4, characterized in that Before introducing the simulated annealing factor into the speed update equation of the improved particle swarm algorithm, the method further includes: An inertia weight adaptive adjustment mechanism corresponding to the BAPV component is generated. The inertia weight adaptive adjustment mechanism is used to dynamically reduce the weight coefficient corresponding to the improved particle swarm algorithm using a hyperbolic tangent function when the particle speed exceeds a threshold.

6. The method according to claim 1, characterized in that The Monte Carlo method is used to simulate the all-weather shadow movement trajectory according to the optimal working angle to generate a component shading influence weight matrix, including: Calculating the solar altitude angle, solar azimuth angle, time series and wind speed vector parameters corresponding to the BAPV assembly to generate a dynamic cloud cover probability distribution; Calculating the shadow projection profile of the BAPV component at the optimal working angle according to the dynamic cloud shading probability distribution; The cumulative time proportion of the BAPV component being blocked is obtained according to the shadow projection profile to generate the component blocking influence weight matrix.

7. The method according to claim 1, characterized in that The step of establishing a heat balance equation according to the component occlusion impact weight matrix comprises: Construct a three-dimensional heat conduction equation based on the heating power and heat dissipation power of the BAPV module; The weight matrix is ​​used as the boundary condition of heat flux density distribution according to the component occlusion influence; Obtain the convective heat transfer coefficient of the BAPV module, the module surface emissivity and the surrounding environment angle coefficient; Constructing a forced convection heat dissipation term according to the convection heat transfer coefficient; Constructing a radiation heat dissipation term according to the surface emissivity of the component and the surrounding environment angle factor; The heat balance equation is constructed according to the three-dimensional heat conduction equation, boundary conditions, convection heat dissipation terms and radiation heat dissipation terms; the heat balance equation is solved by the finite element method, and the spacing optimization parameters that minimize the temperature gradient of the BAPV component are iteratively calculated.

8. The method according to claim 1, characterized in that The generating a control instruction according to the spacing optimization parameter comprises: Inputting the spacing optimization parameters into a preset genetic algorithm optimization model, and outputting the bracket rotation angle compensation amount; Based on the mapping relationship between the operating temperature and conversion efficiency of the BAPV component, the inverter parameter adjustment amount of the inverter corresponding to the BAPV component is dynamically adjusted; An impedance spectrum analysis is performed on the BAPV component, and if a hidden crack fault feature is detected, an isolation signal including a topology reconstruction instruction is generated as the abnormal component isolation instruction.

9. A rooftop photovoltaic intelligent monitoring system based on BAPV, characterized in that: include: 3D laser scanning equipment for scanning BAPV modules; Environmental parameter measurement equipment, used for measuring the real-time environmental parameter set of the BAPV component in real time; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters and wind speed vector parameters; A control device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the method according to any one of claims 1 to 8 when executing the computer program.

10. A rooftop photovoltaic dynamic tuning device based on BAPV, characterized in that: include: A coordinate establishment module, used to construct a surface point cloud model by three-dimensional laser scanning of a preset BAPV component, and used to establish a three-dimensional coordinate system including the installation base point of the BAPV component; A parameter acquisition module, used to acquire a real-time environmental parameter set corresponding to the BAPV component; the real-time environmental parameter set includes distributed light intensity parameters, component temperature distribution parameters, environmental temperature and humidity parameters, and wind speed vector parameters; An angle calculation module, used to calculate the optimal working angle of the BAPV component according to the real-time environmental parameter set based on an improved particle swarm algorithm; the objective function of the improved particle swarm algorithm integrates the real-time power generation efficiency and predicted irradiance of the BAPV component; A matrix generation module, used to simulate the all-weather shadow movement trajectory according to the optimal working angle based on the Monte Carlo method, and generate a component shading influence weight matrix; A parameter calculation module, used to establish a thermal balance equation according to the component shielding influence weight matrix to calculate the spacing optimization parameters of the BAPV components; the thermal balance equation includes a forced convection heat dissipation term and a radiation heat dissipation term; The tuning completion module is used to generate a control instruction according to the spacing optimization parameter, and complete the dynamic tuning of the BAPV component according to the control instruction; the control instruction includes a bracket rotation angle compensation amount, an inverter parameter adjustment amount and an abnormal component isolation instruction.

Images