A tracking angle adaptive optimization method for a bifacial photovoltaic module of a horizontal single-axis tracker and a control system thereof

Abstract

This invention discloses an adaptive optimization method and system for the tracking angle of bifacial photovoltaic modules using a horizontal single-axis tracker, belonging to the field of photovoltaic power generation system tracking control technology. The method, within each control cycle, uses the effective irradiance G after incident angle correction and back-side mismatch correction. eff =G front +φ·(1-MRF)·G rear Maximizing the tracking tilt angle is the sole optimization objective, determining the optimal tracking tilt angle within a search interval constrained by slope-sensing anti-occlusion constraints. This objective function simultaneously rewards the real electrical conversion gains from direct sunlight, back reflection, and sky scattering. Its extreme point locations naturally emerge in three tracking modes depending on weather type: approaching the anti-occlusion baseline under direct sunlight dominance, generating dual-surface sensing fine-tuning under mixed weather conditions, and automatically leveling to a horizontal position under scattering dominance, without requiring any weather mode judgment branches or manual threshold switching. This invention merges three independent algorithms into a single optimizer, eliminating the need for site ray tracing pre-calibration, and can run independently on a low-cost embedded microcontroller.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation system tracking control technology, specifically to an adaptive optimization method and control system for the tracking angle of bifacial photovoltaic modules used in horizontal single-axis trackers. In this invention, the ground coverage ratio (GCR), which is the ratio of the module's width in the rotational direction to the row spacing, is a key geometric parameter determining the risk of shading between tracker rows. Background Technology

[0002] Traditional single-axis trackers rely on astronomical tracking to ensure the module's normal is always aligned with the sun. This strategy is near-optimal in single-sided photovoltaic systems, but when applied to bifacial modules, it fails to consider the reflected and scattered radiation received on the back of the module, thus not fully utilizing the bifaciality of 0.65~0.85 of bifacial modules.

[0003] To improve the power generation of bifacial modules, the industry has proposed three typical improvement algorithms: (1) Two-sided sensing and tracking. A representative scheme uses ray tracing to pre-calibrate a third-order polynomial model of the two-sided ratio, estimates the back face gain in a band-correlated manner, and optimizes within the tracking angle ±ψ range. This scheme only optimizes the two-sided gain and does not handle the scattering-dominated working condition, so it needs to be used in combination with other algorithms.

[0004] (2) Scattering Enhancement Algorithm. When the scattering ratio is higher than the threshold, the component is forced to flatten (β→0) to maximize the reception of scattered light. This scheme requires explicit determination of weather type and switching of control branches based on the scattering ratio threshold, which carries the risk of oscillation near the threshold.

[0005] (3) Slope sensing anti-shading algorithm. For periods of low daily altitude angle, actively reducing the tilt angle to avoid mutual shading between adjacent rows has become the industry standard baseline.

[0006] The three types of algorithms mentioned above are generally deployed as three relatively independent functional modules, and their collaborative operation requires complex switching logic. There has long been a technical bias in this field, namely that two-sided sensing tracking, scattering enhancement, and anti-occlusion are three different optimization problems that must be modeled separately and then superimposed. The inventors have discovered that the essence of these three behaviors is to maximize the effective irradiance received by the component under different weather conditions, which can be uniformly described by a single objective function, thus breaking the aforementioned technical bias. Summary of the Invention

[0007] The purpose of this invention is to provide a method and control system for adaptive optimization of tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers. This method naturally embeds the three behaviors of bifacial sensing and tracking, scattered light enhancement, and anti-shading into the same objective function optimization process. It does not require any weather pattern judgment branch, site pre-calibration, or computational load, and can run independently on a low-cost embedded controller.

[0008] The present invention provides an adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers, and the following steps are performed in each control cycle: Step 1: acquire the current direct normal irradiance, horizontal plane diffuse irradiance, solar zenith angle and solar azimuth angle, and read the current tracking angle β of the horizontal single-axis tracker current , and acquire the geometric parameters of the horizontal single-axis tracker; said geometric parameters at least include ground coverage ratio GCR and mechanical limit inclination angle β max , wherein the ground coverage ratio GCR is the ratio of the width of the module in the rotation direction to the row spacing. This step realizes the unified collection of all input parameters required by the optimization algorithm, and provides a complete data basis for subsequent steps.

[0009] Step 2: calculate the unshaded pure tracking angle β according to said solar zenith angle, said solar azimuth angle and said geometric parameters true and slope-aware anti-shading baseline inclination angle β bt . In this step, the anti-shading function is incorporated into the input of the optimization framework instead of being an independent algorithm module, which provides a basis for the working condition adaptive determination of the subsequent search interval.

[0010] Step 3: determine the search interval of the tracking inclination angle according to said mechanical limit inclination angle β max , said ground coverage ratio GCR and said unshaded pure tracking angle β true , the determination rule is: when said unshaded pure tracking angle β true satisfies |cosβ true |<GCR, it is determined that the current period is a period when inter-row shading may occur; if said baseline inclination angle β_bt is greater than or equal to zero, said search interval is [0, β bt ; if said baseline inclination angle β bt is less than zero, said search interval is [β bt , 0]; otherwise, said search interval is β max , +β max . This step makes the anti-shading function a boundary constraint of the optimization problem, and eliminates the threshold oscillation problem caused by switching between algorithms.

[0011] Step 4: generate a candidate inclination set within said search interval, and incorporate said baseline inclination angle β bt into said candidate inclination set. Forcibly incorporating β bt into the candidate set ensures that the optimization result is not inferior to the anti-shading baseline strategy under any circumstances.

[0012] Step 5: calculate the effective irradiance G corresponding to each candidate inclination angle respectively effThe effective irradiance G eff Calculate using the following formula: G eff =G front +φ·(1-MRF)·G rear ; Among them, G front G represents the effective irradiance of the front side of the photovoltaic module after incident angle correction. rear G represents the effective irradiance on the back of the photovoltaic module after incident angle correction, φ represents the nominal bifaciality of the module, and MRF is the back mismatch coefficient characterizing the current mismatch loss in the cell string caused by non-uniform irradiation on the back of the bifacial module. front and G rear It incorporates contributions from four sources: direct light, isotropic scattering, anisotropic scattering, and ground reflection, calculated using the field-of-view factor. MRF is a preset scalar constant with a value range of 0.05 ≤ MRF ≤ 0.15, calibrated once before shipment, eliminating the need for on-site ray tracing pre-calibration.

[0013] Step Six: Select G eff The largest candidate tilt angle is used as the target tracking angle β in the current control cycle. opt : β opt =argmax k G eff (β k ) Due to β opt G is always in the candidate set, therefore eff (β opt ) ≥ G eff (β bt Heng is established.

[0014] Step 7: Calculate the target tracking angle β opt With the current tracking angle β current If the absolute value of the difference is greater than a preset hysteresis threshold, then the drive motor rotates to the target tracking angle β. opt And update the current tracking angle β current For β opt Otherwise, maintain the current tracking angle β. current Unchanged. By using hysteresis comparison control, frequent small changes in the tracking angle are avoided, thus extending the mechanical life of the tracker.

[0015] G eff The extreme point location of (β) naturally exhibits three tracking modes depending on the weather type, and these three modes do not require any weather pattern judgment branch or manual threshold switching: Under the direct sunlight-dominated condition, β opt Approaching β bt Under mixed weather conditions, β opt relative to βbt Generates dual-sided sensing fine-tuning; under scattering-dominated operating conditions, β opt Approaching 0°.

[0016] The present invention has the following beneficial effects: (1) Simplify the system architecture by merging the three independent algorithms into a single optimizer; (2) The behavior is continuously differentiable and there is no jump near the threshold; no on-site calibration is required, and the MRF single scalar replaces the high-order polynomial pre-calibration; (3) It has strong adaptability across climates and significant simulation gains throughout the year; (4) Embedded independent deployment, which can run independently on low-cost MCUs. Attached Figure Description

[0017] Figure 1 This is a flowchart of the adaptive tracking angle optimization method of the present invention.

[0018] Figure 2 This is a geometric and optical path diagram of the horizontal single-axis tracking bifacial photovoltaic module of the present invention.

[0019] Figure 3 The effective irradiance G under three typical weather types according to the present invention eff A graph showing the tracking tilt angle β.

[0020] Figure 4 The actual tracking angle offset (β) throughout the year of this invention opt β baseline Histogram of distribution of ).

[0021] Figure 5 This is a typical monthly gain histogram for the present invention in Stockholm. Detailed Implementation

[0022] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0023] Example 1: System Composition This embodiment provides a bifacial photovoltaic module tracking angle adaptive optimization system for a horizontal single-axis tracker, comprising the following components: The input acquisition module is used to obtain the current direct normal irradiance, horizontal diffuse irradiance, solar zenith angle, and solar azimuth angle, read the current tracking angle of the horizontal single-axis tracker, and obtain the geometric parameters of the horizontal single-axis tracker. This module interfaces with the SCADA system of a weather station or photovoltaic power station to collect irradiance data such as DNI and DHI, calculates the instantaneous zenith angle and azimuth angle using the built-in NREL solar position algorithm, and reads site preset constants including latitude and longitude and ground albedo.

[0024] A geometric parameter storage module is used to store the geometric parameters and component parameters of the horizontal single-axis tracker. The geometric parameters include at least the ground cover rate (GCR) and the mechanical limit tilt angle β. max Specifically, persistent storage tracks geometric parameters such as GCR, hub height h, component width W, axis azimuth angle, and mechanical limit β. max , as well as component parameters φ and MRF.

[0025] The baseline calculation module is used to calculate the unobstructed pure tracking angle β based on the solar zenith angle, solar azimuth angle, and geometric parameters. true and slope-sensing anti-masking baseline tilt angle β bt This module uses an industry-standard slope-sensing anti-occlusion method to provide the anti-occlusion baseline tilt angle β at each moment. bt .

[0026] The search interval determination module is used to determine the mechanical limit tilt angle β based on the search interval. max The ground coverage rate GCR and the unobstructed pure tracking angle β true Determine the search range for the tracking tilt angle.

[0027] The candidate dip angle generation module is used to generate a set of candidate dip angles within the search interval and to set the baseline dip angle β. bt The candidate tilt angles are included in the set.

[0028] The effective irradiance calculation module is used to calculate the effective irradiance G corresponding to each candidate tilt angle. eff .

[0029] The target tracking angle selection module is used to select G. eff The largest candidate tilt angle is used as the target tracking angle β in the current control cycle. opt .

[0030] The drive execution module is used to calculate the target tracking angle β. opt If the absolute value of the difference between the current tracking angle and the target tracking angle β is greater than a preset hysteresis threshold, then the drive motor rotates to the target tracking angle β. opt And update the current tracking angle to β. optOtherwise, the current tracking angle remains unchanged. The hysteresis threshold is typically set to 1°, determining whether to actually drive the motor to rotate. This hysteresis comparison control avoids frequent small fluctuations in the tracking angle near the target value, thereby extending the mechanical life of the tracker.

[0031] The log and remote reporting module is used to record β at each time step. bt β opt G front G rear Power output to local CSV and optional SCADA uplink.

[0032] Example 2: Detailed Method Flow like Figure 1 As shown, the adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers of the present invention includes the following steps: Step S1, Input Acquisition: Obtain the current direct normal irradiance (DNI), horizontal diffuse irradiance (DHI), and solar zenith angle and solar azimuth angle calculated by the NREL solar position algorithm; read the current tracking angle β of the horizontal single-axis tracker. current ; Obtain the geometric parameters of the horizontal single-axis tracker; the geometric parameters include at least the ground cover rate (GCR) and the mechanical limit tilt angle β. max The ground coverage ratio (GCR) is the ratio of the component's width to the row spacing in the rotation direction. The ground albedo (ρ), the ratio of solar radiation reflected by the ground surface, is read and used in step S5 to calculate the ground reflected irradiance. DNI is the direct normal irradiance, and DHI is the horizontal diffuse irradiance. The solar zenith angle and azimuth angle are calculated using the NREL solar position algorithm based on the station's latitude, longitude, altitude, time zone, and time. The ground albedo (ρ) is set to 0.25 by default and is used for calculating the ground reflection term. Air temperature and wind speed are used for battery temperature and electrical post-processing, and are not required for optimizing the inner loop. This step achieves unified acquisition of all input parameters required by the optimization algorithm, providing a complete data foundation for subsequent steps; by reading the current tracking angle β... current This provides feedback input for the hysteresis comparison control in step S7, thus forming a closed loop for the entire control cycle.

[0033] Step S2, Anti-Obstruction Baseline Angle Calculation: Based on the solar zenith angle, solar azimuth angle, and geometric parameters, calculate the unobstructed pure tracking angle β. true and slope-sensing anti-masking baseline tilt angle β bt Specifically, the slope-sensing anti-shading algorithm is invoked, employing the singleaxis tracking function from the open-source photovoltaic simulation library pvlib, with the backtrack parameter set to True, and the baseline tilt angle β is given. bt; Call the same application programming interface, set the backtrack parameter to False, and obtain the unobstructed pure tracking angle β true . Said unobstructed pure tracking angle β true represents the ideal astronomical tracking angle when inter-row shading is not considered, and said slope perception anti-shading baseline inclination angle β bt represents the safe inclination angle after active retraction to avoid inter-row shading. In this step, the anti-shading function is incorporated into the input of the optimization framework instead of being an independent algorithm module, which provides a basis for the working condition adaptive determination of the subsequent search interval; β true and β bt paired calculation enables the subsequent interval determination step to accurately determine the shading risk period.

[0034] Step S3, Search interval determination: according to said mechanical limit inclination angle β max , said ground coverage ratio GCR and said unobstructed pure tracking angle β true , determine the search interval of the tracking inclination angle. The determination rule is: when said unobstructed pure tracking angle β true satisfies |cosβ true |<GCR, it is determined that the current period is a period when inter-row shading may occur. If said baseline inclination angle β bt is greater than or equal to zero, said search interval is [0,β bt , if said baseline inclination angle β bt is less than zero, said search interval is [β bt , 0]; otherwise, said search interval is β max , +β max . When β bt =0, the search interval given by the above rule is [0,0], that is, a single-point set. At this time, the candidate inclination set only contains the inclination angle 0°, step S6 selects G eff( 0°) as β opt , step S7 normally performs difference comparison and driving decision, and the control process is complete and executable. This step makes the anti-shading function a boundary constraint of the optimization problem instead of another set of independent algorithms, eliminating the threshold oscillation problem caused by switching between algorithms. When |cosβ true |<GCR, it indicates that the solar altitude angle is low, and mutual shading may occur between adjacent rows. At this time, it is necessary to limit the search interval between the anti-shading baseline inclination angle and the horizontal position to prevent the optimizer from selecting an inclination angle that may cause shading; when |cosβ true | ≥ GCR, inter-row shading will not occur, and the search interval can be extended to the mechanical limit range.

[0035] Step S4, Construct candidate angle set: generate a candidate inclination set within said search interval, and add said baseline inclination angle β btThe candidate tilt angles are included in the set.

[0036] The candidate tilt angle set can be generated in one of the following two ways: Method 1: Within the search interval, perform uniform scanning with a preset step size to generate a candidate tilt angle set, and then set the baseline tilt angle β... bt Forced inclusion into the candidate tilt angle set. In one specific embodiment, the step size is set to 1°, β bt =+25° represents the morning period, with the candidate set being {0°, 1°, 2°, …, 25°}, and β is mandatory. bt Afternoon β bt When the value is negative, the candidate set is [β]. bt [, …, 0°]; During the mid-to-high solar hours, the candidate set can be [ 60°, …, +60°).

[0037] Method 2: Employ the golden section search algorithm to converge to the optimal search inclination angle with an error of less than 1° within the search interval using no more than 12 forward calculations; use the optimal search inclination angle and the baseline inclination angle β... bt This serves as a set of candidate inclination angles for comparison. The golden section search algorithm approximates the objective function G by continuously narrowing the search interval. eff The extreme point of (β) has a fast convergence speed and constant computational cost, making it suitable for deployment in embedded environments.

[0038] β bt Forced inclusion into the candidate set, for each candidate tilt angle β k Calculate G eff Then, select G. eff The largest candidate is β opt Because of β bt G is always in the candidate set, therefore eff (β opt ) ≥G eff (β bt If this holds true, the optimization results are no worse than the baseline strategy.

[0039] Step S5, Forward evaluation: For each candidate tilt angle β k Perform the following sub-steps sequentially to calculate the effective irradiance G corresponding to each candidate tilt angle. eff : S5.1 The field of view factor, including geometric occlusion of the current row and its ±3 neighboring rows, is calculated using the Hottel cross-rope method. Specifically, a geometric model of seven rows (reference row and ±3 adjacent rows) is established. The Hottel cross-rope method is used for occlusion ray detection, and the field of view factors for direct ground stripe occlusion, scattering, direct panel stripe occlusion, sky scattering, and ground reflection are calculated. The selection of the ±3 neighboring rows is based on the following engineering considerations: In a tracker array, the geometric occlusion angle of the target row from adjacent rows with a distance greater than 3 rows has attenuated to a negligible level, typically less than 0.5°, while fewer than 3 rows are insufficient to accurately characterize the cumulative occlusion effect of multi-row arrays at low solar altitude angles. This selection achieves a reasonable balance between computational accuracy and computational efficiency. This method has been vectorized and can efficiently handle complex occlusion relationships in multi-row geometry.

[0040] S5.2. Based on the Perez 1990 model, the horizontal surface diffuse irradiance (DHI) is decomposed into isotropic and anisotropic scattering components. Isotropic scattering refers to the uniform distribution of DHI according to the Lambert sky and sky field of view factors, which is the main path. The anisotropic scattering component is calculated using the Perez 1990 model, which parametrically describes the enhancement effect of sky diffuse irradiance in the region around the sun and the concentration effect near the horizon under clear sky conditions. Its calculation accuracy is better than that of the simple isotropic assumption.

[0041] S5.3. Integrating incident angle correction to correct for incident angle loss in direct light. The Sjerps-Koomen physical incident angle correction model, with a refractive index n=1.526, is used to correct for incident angle loss in direct light. This model, based on Fresnel's equations describing the transmission loss of light at the glass-air interface, has a clear physical basis. A constant coefficient of 0.96 is multiplied between scattered and reflected light, representing the average transmittance correction coefficient for scattered light from the glass cover. The physical basis for this is that when scattered light is incident at different angles, the average transmittance of a single-layer glass-air interface is approximately 96%, meaning the average reflection loss is approximately 4%.

[0042] S5.4, Summing yields G front (β k ) and G rear (β k G front and G rear The average effective irradiance of the two sides before incident angle correction is expressed in W / m², and is obtained by combining the above field factor calculation, scattering decomposition, and incident angle correction.

[0043] S5.5, Applying the composite objective function of this invention: G eff =G front + φ·(1 MRF)·G rear ; Where φ is the nominal bifaciality of the component, preferably 0.75, and MRF is the back-side mismatch coefficient, preferably 0.10, based on the NREL Deline 2019 public dataset. The technical principle of this objective function is: G front and G rear It incorporates contributions from four sources: direct light, isotropic scattering, anisotropic scattering, and ground reflection. The field-of-view factor is calculated using the Hottel cross-rope method to determine the multi-row geometry. Unlike existing techniques that pre-calibrate the bifacial ratio BR(β) using high-order polynomials, this invention introduces a constant MRF as a back-side mismatch comprehensive correction coefficient. This equates the current mismatch loss caused by uneven back-side illumination in the module's cell strings to a scalar multiplier (1 / 2). MRF). The physical meaning of this scalar multiplier is: effective electrical contribution from the back side = optical contribution from the back side × (1 The back mismatch loss rate (MRF) is calculated, where MRF=0.10 represents that the electrical loss caused by uneven back illumination is approximately 10% of the back irradiance gain. This step uses a single scalar MRF to replace the high-order polynomial precalibration of existing technologies, eliminating the need for site ray tracing preprocessing and significantly reducing engineering implementation costs.

[0044] Step S6, Extreme Value Selection: Select G eff The largest candidate tilt angle is used as the target tracking angle β in the current control cycle. opt : β opt =argmax k G eff (β k ) Due to β bt G is always in the candidate set, therefore eff (β opt ) ≥ G eff (β bt This holds true consistently. This step uses maximizing effective irradiance as the sole selection criterion, and the optimizer automatically switches between three behaviors—direct irradiance-dominated β... opt Approximately equal to β bt β under mixed weather conditions opt Deviation from β bt Several fine-tunings are produced, and β is dominated by scattering. opt Approaching 0°, the three behaviors automatically emerge from the extreme point of the same objective function as the weather naturally shifts, without any human judgment or branching.

[0045] Step S7, Driving Decision: Calculate the target tracking angle β opt With the current tracking angle β current If the absolute value of the difference is greater than a preset hysteresis threshold, then the drive motor rotates to the target tracking angle β.opt And update the current tracking angle β current For β opt Otherwise, maintain the current tracking angle β. current Unchanged. If |β opt β current If the angle is greater than the motor hysteresis threshold (typically 1°), a target angle command is sent to the motor driver; otherwise, the current attitude is maintained. This step, through hysteresis comparison control, avoids frequent small fluctuations in the tracking angle near the target value, thereby extending the tracker's mechanical lifespan.

[0046] Step S8, Log and Report: Data β for this period bt β opt G front G rear Estimate AC power and write it to the local CSV file, then upload it to SCADA at set intervals.

[0047] Example 3: Geometric and Optical Path Diagram like Figure 2 As shown, multi-row single-axis trackers use a row spacing of pitch = collector. width / GCRs are arranged at equal intervals, with the central axis of each row of components located at height h. The tilt angle β is defined as the angle by which the component normal deviates from the zenith in the vertical plane. In this invention, β>0 indicates that the western edge is raised, i.e., the morning configuration.

[0048] The front of the module receives direct sunlight (DNI), sky scattering (DHI), and reflections from adjacent rows; the back of the module mainly receives ground reflections (ρ·GHI) and a small amount of scattering. The field-of-view factor calculation in this invention uses the Hottel cross-rope method, considering geometric occlusion of 7 rows (±3 rows of neighboring rows), therefore G... front With G rear The energy loss due to mutual occlusion between rows is already implicitly included.

[0049] Example 4: Three-state adaptive behavior of spontaneous emergence of a single objective Figure 3 Given the same solar position (zenith angle 55°, azimuth angle 110°), the G values ​​for three typical weather types are given. eff (β) curve.

[0050] The following three modes can be observed: Under sunny conditions: DNI = 750 W / m², DHI = 110 W / m², the curve reaches its extreme value at β≈+50°, β opt It is close to the classic tracking angle. The physical principle is that direct light dominates in frontal irradiation, while back reflection and scattering contribute relatively little. The extremum of the objective function is mainly determined by the frontal incident angle loss term.

[0051] Mixed weather conditions: DNI = 350 W / m², DHI = 280 W / m², extreme values ​​shift to the left to β≈+36°, β opt The system generates two-sided sensing fine-tuning relative to the baseline. The physical principle is that direct and scattering contributions are roughly equal, while the contribution weight of back reflection gains to the objective function increases. The optimizer seeks a balance between the frontal incident angle loss and the back reflection gain.

[0052] Under cloudy conditions: DNI=30 W / m², DHI=290 W / m², the extreme value of the curve appears at β≈+6°, the module tends to be horizontal, which is equivalent to scattering enhancement behavior. The physical principle is: direct light is almost gone, the cumulative irradiance of the horizontal surface from all-sky scattering is the highest, and the extreme value of the objective function naturally shifts to the horizontal position.

[0053] Therefore, it can be seen that the present invention does not require any weather type identification branch or manually set thresholds, and the three typical behaviors all emerge naturally from the geometric shape of the same physical objective function.

[0054] Example 5: Verification Data To verify the effectiveness of this invention, hourly simulations were conducted over 8760 hours throughout the year in three typical European climate zones, based on typical EPW meteorological year data. The configurations for each region were identical: GCR=0.35, h=1.5m, W=2.4m, max_angle=±60°, φ=0.75, MRF=0.10, ρ=0.25. Electrical conversion employed the pvlib ModelChain, where the CEC single-diode model was the photovoltaic module electrical model proposed by the California Energy Commission, the Sandia inverter was the inverter efficiency model proposed by Sandia National Laboratories, and the SAPM glass-to-glass open cooling system was the bifacial module temperature model from the System Consultant photovoltaic model. The AC outputs were calculated end-to-end for both the baseline strategy and the strategy of this invention.

[0055] The annual gain is summarized in Table 1: Table 1 Summary of Annual Gains

[0056] Gain structure based on scattering ratio: The hourly data for the entire year were further divided into three intervals based on the scattering ratio DR = DHI / (DNI·cosθ + DHI), as shown in Table 2: Table 2. Interval Division

[0057] The gains in all three regions are +0.18% to +0.22% in the direct-light-dominated region, which is the contribution of the two-sided sensing fine-tuning. The gains in the scattering-dominated region are +4.82% to +6.35%, which is the contribution of the scattering enhancement behavior. These gains are automatically determined by the single objective function of this invention.

[0058] Angular offset distribution: like Figure 4 As shown, Munich's hourly (β) throughout the year opt β baseline The distribution exhibits two distinct modes: the central peak (|Δβ|<5°) is the dual-sensing fine-tuning mode, which occurs during the direct-light-dominated period; the scattering-dominated mode occurs around ±60°, where the controller actively flattens the component from the near-maximum tracking angle of the baseline to near horizontal.

[0059] Typical daily power generation curve: Taking June 18, 2023 in Munich as an example (the day with the highest gain throughout the year), the baseline strategy generated 1.10 kWh / module, while the strategy of this invention generated 1.18 kWh / module, resulting in a daily gain of +7.59%. As can be seen from the tracking angle panel, during the morning (low solar eclipse and high scattering) and the afternoon (cloudy period), the algorithm actively pulls the tilt angle towards the horizontal to maximize the reception of scattered light; during the midday high DNI period, the tilt angle is consistent with the baseline. The gain is concentrated in the two periods of high scattering in the morning and evening.

[0060] Monthly gain distribution: like Figure 5 As shown, the monthly gain data for the Stockholm (Northern Europe) high scattering climate shows that the gain is positive in all months throughout the year, ranging from 1.30% to 2.53%, with no significant drop between months, proving that the invention remains stable and effective under seasonal changes. The gain is higher in summer (June-September) than in winter (December-March) because summer has longer daylight hours and a higher frequency of mixed weather events.

[0061] Industrial applicability This invention provides a method and system for adaptive optimization of tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers, which can be widely applied in horizontal single-axis trackers for bifacial photovoltaic modules in photovoltaic power plants. This invention requires no site pre-calibration, no weather pattern judgment, and can run independently on a low-cost embedded microcontroller, demonstrating significant industrial practical value and economic benefits.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive optimization of tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers, characterized in that, The method performs the following steps in each control cycle: Obtain the current direct normal irradiance, horizontal diffuse irradiance, solar zenith angle, and solar azimuth angle, and read the current tracking angle β of the horizontal single-axis tracker. current The geometric parameters of the horizontal single-axis tracker are obtained; the geometric parameters include at least the ground cover rate (GCR) and the mechanical limit tilt angle β. max The ground coverage rate (GCR) is the ratio of the width of the component in the rotation direction to the row spacing. Calculate the unobstructed pure tracking angle β based on the solar zenith angle, solar azimuth angle, and geometric parameters. true and slope-sensing anti-masking baseline tilt angle β bt ; According to the mechanical limit tilt angle β max The ground coverage rate GCR and the unobstructed pure tracking angle β true Determine the search range for the tracking tilt angle; A set of candidate dip angles is generated within the search interval, and the baseline dip angle β is set. bt Included in the candidate tilt angle set; Calculate the effective irradiance G corresponding to each candidate tilt angle. eff The effective irradiance G eff Calculate using the following formula: G eff =G front +φ·(1-MRF)·G rear ; Among them, G front G represents the effective irradiance of the front side of the photovoltaic module after incident angle correction. rear φ is the effective irradiance on the back of the photovoltaic module after incident angle correction, φ is the nominal bifaciality of the module, and MRF is the back mismatch coefficient characterizing the current mismatch loss of the cell string caused by non-uniform irradiation on the back of the bifacial module. Select G eff The largest candidate tilt angle is used as the target tracking angle β in the current control cycle. opt ; Calculate the target tracking angle β opt With the current tracking angle β current If the absolute value of the difference is greater than a preset hysteresis threshold, then the drive motor rotates to the target tracking angle β. opt And update the current tracking angle β current For β opt Otherwise, maintain the current tracking angle β. current constant.

2. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The effective irradiance G front and G rear The solution includes: calculating the field factor including the geometric occlusion of the current row and ±3 neighboring rows using the field factor calculation method; decomposing the horizontal scattering irradiance into scattering components based on the scattering decomposition model; and correcting the incident angle loss of direct light by combining the incident angle correction.

3. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 2, characterized in that, The field-of-view factor was calculated using the Hottel cross-rope method, and the scattering decomposition used the Perez 1990 model to decompose the horizontal surface scattering irradiance into isotropic and anisotropic scattering components.

4. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The back mismatch coefficient MRF is a preset scalar constant with a value range of 0.05≤MRF≤0.

15. The MRF is calibrated once before leaving the factory and does not require on-site ray tracing pre-calibration.

5. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The rule for determining the search range of the tracking tilt angle is as follows: When the unobstructed pure tracking angle β true satisfies |cosβ true |<GCR, it is determined that the current period is a period when inter-row occlusion may occur. If the base inclination β bt is greater than or equal to zero, the search interval is [0, β bt ; if the base inclination β bt is less than zero, the search interval is [β bt , 0]; otherwise, the search interval is β max , +β max .

6. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The effective irradiance G eff The location of extreme points naturally exhibits the following three tracking modes depending on the weather type, and none of the three modes require any weather pattern judgment branches or manual threshold switching: Under direct-fire dominant conditions, the target tracking angle β opt Approaching the baseline tilt angle β bt ; Under mixed weather conditions, the target tracking angle β opt relative to the baseline tilt angle β bt Generates dual-sensory fine-tuning; Under scattering-dominated conditions, the target tracking angle β opt Approaching 0°.

7. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The candidate tilt angle set is generated by uniformly scanning within the search interval at a preset step size, and the baseline tilt angle β is used as the baseline tilt angle β. bt Forced inclusion into the candidate tilt angle set; The preset step size is 1°.

8. The adaptive optimization method for tracking angle of bifacial photovoltaic modules for horizontal single-axis trackers according to claim 1, characterized in that, The golden section search algorithm is used to converge to the optimal search inclination angle with an error of less than 1° within the search interval with no more than 12 forward evaluations; the optimal search inclination angle and the baseline inclination angle β are then used as the basis for the search. bt As a set of candidate tilt angles for comparison, the effective irradiance corresponding to each tilt angle in the set of candidate tilt angles is calculated, and the one with the larger effective irradiance is selected as the target tracking angle.

9. A bifacial photovoltaic module tracking angle adaptive control system for a horizontal single-axis tracker, implementing the tracking angle adaptive control method for a bifacial photovoltaic module for a horizontal single-axis tracker according to any one of claims 1-8, characterized in that, include: The input acquisition module is used to acquire the direct normal irradiance, horizontal diffuse irradiance, solar zenith angle and solar azimuth angle at the current moment, read the current tracking angle of the horizontal single-axis tracker, and acquire the geometric parameters of the horizontal single-axis tracker. A geometric parameter storage module is used to store the geometric parameters and component parameters of the horizontal single-axis tracker. The geometric parameters include at least the ground cover rate (GCR) and the mechanical limit tilt angle β. max ; The baseline calculation module is used to calculate the unobstructed pure tracking angle β based on the solar zenith angle, solar azimuth angle, and geometric parameters. true and slope-sensing anti-masking baseline tilt angle β bt ; The search interval determination module is used to determine the mechanical limit tilt angle β based on the search interval. max The ground coverage rate GCR and the unobstructed pure tracking angle β true Determine the search range for the tracking tilt angle; The candidate dip angle generation module is used to generate a set of candidate dip angles within the search interval and to set the baseline dip angle β. bt Included in the candidate tilt angle set; The effective irradiance calculation module is used to calculate the effective irradiance G corresponding to each candidate tilt angle. eff The effective irradiance G eff Calculate using the following formula: Geff=G front +φ·(1-MRF)·G rear ; Among them, G front G represents the effective irradiance of the front side of the photovoltaic module after incident angle correction. rear φ is the effective irradiance on the back of the photovoltaic module after incident angle correction, φ is the nominal bifaciality of the module, and MRF is the back mismatch coefficient characterizing the current mismatch loss of the cell string caused by non-uniform irradiation on the back of the bifacial module. The target tracking angle selection module is used to select G. eff The largest candidate tilt angle is used as the target tracking angle β in the current control cycle. opt ; The drive execution module is used to calculate the target tracking angle β. opt If the absolute value of the difference between the current tracking angle and the target tracking angle β is greater than a preset hysteresis threshold, then the drive motor rotates to the target tracking angle β. opt And update the current tracking angle to β. opt Otherwise, maintain the current tracking angle.

Images