A double-sided photovoltaic tracking method and system based on a genetic algorithm

Abstract

The present application relates to the field of photovoltaic power generation, and more particularly to a double-sided photovoltaic tracking method and system based on a genetic algorithm. The method comprises: constructing a basic data structure based on multi-source data such as field station geographic time parameters and actuator constraint parameters; performing solar position and attitude geometric calculation to obtain solar geometric structure and candidate attitude geometric structure; inputting the solar geometric structure and the candidate attitude geometric structure into double-sided irradiation calculation to obtain front effective irradiation, back effective irradiation and shading factor; performing temperature calculation and electrical output calculation to obtain a candidate attitude power sequence and an energy index; performing fitness construction and feasible region processing to obtain a fitness evaluator and a feasible region processor; performing genetic algorithm calculation to output optimal strategy parameters and an angle curve, and generating a target angle instruction. The present application can realize collaborative evaluation and constraint optimization output of geometric, irradiation and electrical links, effectively improving the availability and stability of control instruction generation.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation, and in particular to a bifacial photovoltaic tracking method and system based on a genetic algorithm. Background Technology

[0002] In the field of photovoltaic power generation, existing solutions for bifacial photovoltaic tracking typically calculate the solar position and attitude geometry based on the site's geographic time parameters and array geometry parameters, and estimate irradiance by combining meteorological irradiance parameters and surface albedo parameters. Then, electrical output is evaluated based on the electrical parameters of the bifacial photovoltaic modules, leading to angle setting and control output. However, this approach suffers from limitations such as inconsistent basic data structure construction, a disconnect between the generation and constraint expression of solar geometry and candidate attitude geometry, and a lack of a unified interface for the coupling link between effective front irradiance, effective back irradiance, and shading factors. Existing methods often rely on fixed attitude scanning or local rule adjustments, and the processing of actuator constraint parameters often involves post-processing or local correction. This results in a lack of a unified engineering entry point for fitness construction and feasible domain processing, leading to issues such as drift in the evaluation criteria of candidate attitude power sequences and energy indicators, and inconsistencies in the time correlation between control parameter tables and angle curves during target angle command generation. Consequently, it is difficult to achieve stable output of optimal strategy parameters and angle curves and generation of target angle commands under actuator constraint parameters. For the joint processing of basic data structure, solar geometry, candidate attitude geometry, effective frontal irradiance, effective backal irradiance, shading factor, candidate attitude power sequence, and energy index in the same link, existing technologies generally have common shortcomings in the connection between data alignment, constraint expression, and optimization output. It is difficult to form a consistent process of acquisition-alignment-determination-control output-record update between double-sided irradiance calculation, temperature calculation and electrical output calculation, fitness construction and feasible domain processing, and genetic algorithm optimization. As a result, the mapping relationship between optimal strategy parameters, angle curves and control parameter tables lacks a unified standard, and the generation process of target angle commands is difficult to maintain verifiable consistency. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a bifacial photovoltaic tracking method based on a genetic algorithm, comprising:

[0004] S100: Obtain the site's geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters to construct the basic data structure;

[0005] S200. Input the basic data structure into the solar position and attitude geometry calculation to obtain the solar geometry structure and candidate attitude geometry structure;

[0006] S300. Input the solar geometry and the candidate attitude geometry into the bi-face irradiance calculation to obtain the effective front irradiance, the effective back irradiance, and the shading factor.

[0007] S400: Input the effective front irradiance, the effective back irradiance, the shading factor, and the basic data structure into temperature calculation and electrical output calculation to obtain the candidate attitude power sequence and energy index;

[0008] S500: Input the energy index and the actuator constraint parameters into the fitness construction and feasible region processing to obtain the fitness evaluator and feasible region processor;

[0009] S600: Input the fitness evaluator and the feasible domain processor into the genetic algorithm for optimization, output the optimal strategy parameters and angle curve, and write the angle curve into the control parameter table to generate the target angle instruction.

[0010] Furthermore, the array geometric parameters include the azimuth angle of the rotating shaft, the tilt angle of the rotating shaft, the height of the component above the ground, the row spacing, the component size, and the number of array rows and columns; the actuator constraint parameters include the tracking angle limit, the upper limit of angular velocity, the upper limit of angular acceleration, the control cycle, and the docking angle.

[0011] Furthermore, the bifacial irradiance calculation includes: calculating the incident angle based on the candidate attitude geometry and generating a frontal incident correction coefficient; calculating the back view factor based on the surface albedo parameter, the component's ground clearance, and the row spacing; calculating the occlusion discrimination factor based on the solar geometry and the row spacing and generating the occlusion factor; and outputting the effective frontal irradiance and the effective backal irradiance based on the frontal incident correction coefficient, the back view factor, and the occlusion factor.

[0012] Furthermore, the temperature calculation inputs the effective front irradiance, the effective back irradiance, the ambient temperature, and the wind speed into the thermal balance model, and outputs the component operating temperature; the thermal balance model includes irradiance absorption terms, convective heat transfer terms, and radiative heat transfer terms.

[0013] Furthermore, the electrical output calculation adopts a single diode five-parameter model, and the photocurrent, reverse saturation current, series resistance, parallel resistance and ideality factor are corrected for temperature based on the component operating temperature; the effective irradiance on the front and the effective irradiance on the back are combined according to the back gain coefficient to obtain the equivalent irradiance, and the candidate attitude power sequence is calculated based on the equivalent irradiance.

[0014] Furthermore, the fitness construction combines the energy index with the penalty term to obtain the fitness value; the penalty term includes limit violation penalty, angular velocity violation penalty, angular acceleration violation penalty, occlusion duration penalty and action frequency penalty; the feasible domain processing performs boundary projection repair on individuals that exceed the tracking angle limit.

[0015] Furthermore, the optimal strategy parameters include the axis tilt angle parameter, the backtracking correction coefficient parameter, and the segmented angle curve node parameter; the genetic algorithm optimization uses real number encoding to map the optimal strategy parameters to chromosomes, and inputs the chromosomes into the fitness evaluator to obtain fitness values.

[0016] Furthermore, the genetic algorithm optimization includes an elite retention operator, a tournament selection operator, a simulated binary crossover operator, and a Gaussian mutation operator; the Gaussian mutation operator updates the mutation rate based on the algebraic index, which is obtained by combining the current iteration algebra and the maximum iteration algebra.

[0017] Furthermore, the angle curve is composed of target angle nodes for multiple time periods; the control parameter table stores the target angle nodes for multiple time periods using a time index; the target angle instruction is generated by reading from the control parameter table according to the control cycle and then performing linear interpolation.

[0018] Furthermore, a bifacial photovoltaic tracking system based on a genetic algorithm includes: a basic data structure construction module, a solar position and attitude geometry calculation module, a bifacial irradiance calculation module, a temperature calculation and electrical output calculation module, a fitness construction and feasible region processing module, a genetic algorithm optimization and angle curve generation module, a control parameter table writing and target angle instruction generation module, and a running data structure and parameter correction and update module; the modules are connected in sequence to implement the method described in any of the above-mentioned embodiments.

[0019] The key innovations of this invention include:

[0020] (1) Organize the site geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters into a basic data structure, and establish a data object transfer link between solar position and attitude geometry calculation, bifacial irradiance calculation, temperature calculation and electrical output calculation, with solar geometry, candidate attitude geometry, front effective irradiance, back effective irradiance, shading factor, candidate attitude power sequence and energy index as the core.

[0021] (2) Connect the energy index and the constraint parameters of the actuator to the fitness construction and feasible domain processing, construct the fitness evaluator and form a feasible domain processor, make the evaluation input of the fitness evaluator consistent with the constraint processing input of the feasible domain processor, and perform boundary projection repair on the candidate posture related objects in the feasible domain processor.

[0022] (3) Connect the fitness evaluator and the feasible domain processor to the genetic algorithm for optimization. In the iterative process of the genetic algorithm optimization, the optimal strategy parameters and angle curves are jointly optimized and output. The angle curves are written into the control parameter table to generate the target angle command, so that the genetic algorithm optimization output and the control parameter table and the target angle command generation object are consistent with the mapping relationship.

[0023] The following are its main beneficial effects:

[0024] (1) Based on the basic data structure, the calculation of solar position and attitude geometry, double-sided irradiance, temperature and electrical output are integrated, so that the solar geometry and candidate attitude geometry drive the generation of front effective irradiance, back effective irradiance and shading factor under the same data caliber, and the candidate attitude power sequence and energy index are obtained continuously from the same link, avoiding the inconsistency of objects and link breakage caused by the segmented independence of geometry estimation, irradiance estimation and electrical output evaluation in the existing direct calculation tracking.

[0025] (2) The fitness evaluator and the feasible domain processor operate in concert under the unified input of energy index and actuator constraint parameters, so that the candidate pose related objects evaluated by the genetic algorithm are consistent with the candidate pose related objects under constraint processing. Furthermore, the boundary projection repair reduces the continuous propagation of individuals exceeding the actuator constraint parameters during the iteration process, thus alleviating the evaluation distortion and instability of feasible solutions caused by the post-constraint pruning in the existing scheme.

[0026] (3) The optimal strategy parameters and angle curves output by the genetic algorithm form a continuous data landing link with the control parameter table and the target angle command, so that the angle curve writing and the target angle command generation revolve around the same control object. This reduces the problem of mismatch between the optimization output and the control side input structure in the existing scheme, which leads to the deviation of the distribution caliber and the difficulty in aligning the operation record. It also provides a consistent data carrying form for subsequent traceability and updates based on the control parameter table. Attached Figure Description

[0027] Figure 1 A flowchart illustrating a bifacial photovoltaic tracking method based on a genetic algorithm, provided in an embodiment of this application;

[0028] Figure 2 This is a structural block diagram of a bifacial photovoltaic tracking system based on a genetic algorithm, provided in an embodiment of this application. Detailed Implementation

[0029] Example 1: Refer to Figure 1 This is a flowchart illustrating a bifacial photovoltaic tracking method based on a genetic algorithm provided in an embodiment of the present invention. The process may include at least steps S100-S600:

[0030] S100: Obtain the site's geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters to construct the basic data structure;

[0031] S200. Input the basic data structure into the solar position and attitude geometry calculation to obtain the solar geometry structure and candidate attitude geometry structure;

[0032] S300. Input the solar geometry and the candidate attitude geometry into the bi-face irradiance calculation to obtain the effective front irradiance, the effective back irradiance, and the shading factor.

[0033] S400: Input the effective front irradiance, the effective back irradiance, the shading factor, and the basic data structure into temperature calculation and electrical output calculation to obtain the candidate attitude power sequence and energy index;

[0034] S500: Input the energy index and the actuator constraint parameters into the fitness construction and feasible region processing to obtain the fitness evaluator and feasible region processor;

[0035] S600: Input the fitness evaluator and the feasible domain processor into the genetic algorithm for optimization, output the optimal strategy parameters and angle curve, and write the angle curve into the control parameter table to generate the target angle instruction.

[0036] The S100 acquires the site's geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters to construct the basic data structure;

[0037] This step is triggered by the data acquisition module after the tracking controller completes its power-on self-test, and is triggered again when three events occur: version switching of the control parameter table, change of sensor channel status, and parameter update command issued by the site maintenance terminal. The data acquisition module reads static configuration items from the site configuration library, real-time measurement items from field sensors and metering devices, and calibration items and missing measurement completion items from the historical database. After unified encapsulation, it generates a basic data structure and writes it to the local cache area and audit log area. The site's geographic time parameters constitute the minimum set required for calculating the solar position and attitude geometry. These parameters include the site's longitude, latitude, altitude, time zone, date and time, timestamp sequence, and calendar marker. The date and time are provided by the tracking controller's clock source, the time zone and latitude / longitude are provided by the site configuration library, and the timestamp sequence is generated by the data acquisition module according to the control cycle and aligned with the sampling time of the field-acquired data. When the tracking controller detects clock drift or a time source switch, the data acquisition module writes the time status into the time status field of the basic data structure and provides this field, along with the timestamp sequence, for reference in subsequent steps, preventing cross-day jumps or time misalignments in the subsequent solar geometry generation.

[0038] On the array structure side, the array geometric parameters are defined as the set of structures required for the spatial configuration and occlusion discrimination of the oblique single-axis array, including the pivot azimuth angle, pivot tilt angle, component height above ground, row spacing, component size, and number of array rows and columns, and further include row number mapping, support axis position, component installation offset, and terrain slope correction fields; wherein the pivot azimuth angle, pivot tilt angle, component size, and number of array rows and columns are from the site configuration library, the component height above ground and row spacing are from the design value and corrected by on-site sampling records, the row number mapping is used to bind the status of each row actuator to the array geometric parameters, the support axis position is used for subsequent occlusion discrimination quantity generation, and the terrain slope correction field is written from surveying data or construction acceptance data. The data acquisition module performs integrity and boundary checks on the array's geometric parameters. The checks include the range of values ​​for the axis azimuth and tilt angles, the physical feasible range of the component's height above the ground and row spacing, and the consistency of the mapping between the number of array rows and columns and the row number. When the check fails, the data acquisition module marks the corresponding field as a structural anomaly and records the anomaly reason code. At the same time, it keeps the valid fields from the previous version in the basic data structure to form an auditable parameter evolution link, supporting subsequent steps to continue to complete the calculation link within the same control cycle.

[0039] On the environmental input side, the surface albedo parameter is defined as a set of surface reflectance characteristics to incident radiation, including a surface albedo baseline value, surface type markers, a zonal albedo table, a time-varying albedo factor, and an albedo confidence marker. The surface albedo baseline value and surface type markers are derived from the site configuration database or construction acceptance data. The zonal albedo table describes the differences in surface cover in different areas of the array. The time-varying albedo factor is written by the operation and maintenance end when conditions change, such as snow cover, vegetation growth, or sand cover. The albedo confidence marker is generated by the data acquisition module based on the reliability of the acquisition source and the update time interval. The meteorological irradiance parameter is defined as a joint set of irradiance components and thermal environment components, including total irradiance, direct irradiance, diffuse irradiance, ambient temperature, wind speed, cloud cover markers, and a sampling time sequence. Total irradiance, direct irradiance, and diffuse irradiance are provided by a radiometer or meteorological station; ambient temperature and wind speed are provided by meteorological sensors; and cloud cover markers are output by the meteorological station or mapped from a historical database. The data acquisition module performs time alignment, missing data completion, anomaly removal, and unit normalization on meteorological irradiance parameters. Time alignment is achieved by aligning the sampled time sequence to the timestamp sequence. Missing data completion uses historical databases for similar time periods and writes the missing data source field. Anomaly removal uses a combination of physical boundaries and jump thresholds for discrimination and writes the anomaly flag field. Unit normalization unifies the outputs of different devices to the agreed units in the basic data structure and writes them to the unit flag field. When the field radiometer is offline, the data acquisition module calls the meteorological station's irradiance channel and synchronously writes the acquisition source flag to ensure continuous input for subsequent calculations of effective irradiance on the front and back sides.

[0040] On the component and mechanism side, the electrical parameters of the bifacial photovoltaic module are defined as a set of parameters for electrical output calculation, including nominal power parameters, open-circuit voltage parameters, short-circuit current parameters, temperature coefficient parameters, back-side gain coefficient, single-diode five-parameter model parameters, and parameter source markers. Among them, the nominal power parameters, open-circuit voltage parameters, short-circuit current parameters, temperature coefficient parameters, and back-side gain coefficient are written into the site configuration library from the component nameplate and factory data. The single-diode five-parameter model parameters are written into the historical database from offline calibration results and loaded in this step. The parameter source markers record three types of sources: factory, calibration, and operation and maintenance revision, supporting subsequent parameter correction traceability. The actuator constraint parameters are defined as a set of constraints for tracking control and feasible domain processing, including tracking angle limit, upper limit of angular velocity, upper limit of angular acceleration, control cycle, and docking angle. They are further extended to include zero-return trigger condition, wind load protection threshold, fault lockout flag, and actuator grouping flag. The tracking angle limit, upper limit of angular velocity, upper limit of angular acceleration, control cycle, and docking angle are derived from the control parameter table; the zero-return trigger condition is configured by the operations and maintenance department; the wind load protection threshold is loaded from the safety policy table; and the fault lockout flag is written by the actuator status diagnosis channel. The data acquisition module performs consistency checks on the actuator constraint parameters. The checks include the consistency between the control cycle and the timestamp sequence step size, the inclusion relationship between the docking angle and the tracking angle limit, and the linkage constraint between the upper limit of angular velocity and the upper limit of angular acceleration. When the check fails, the data acquisition module writes a constraint anomaly flag and freezes the policy update channel for the current cycle to prevent the introduction of unexecutable angle curves in subsequent genetic algorithm optimization.

[0041] After completing the above collection and processing, the data acquisition module encapsulates the site's geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters into a basic data structure according to a unified field specification, and adds parameter version number, acquisition batch number, data quality tag, acquisition source tag, and audit summary fields. The parameter version number is generated by combining the control parameter table version and the configuration library version; the acquisition batch number is generated by the tracking controller incrementing each time execution is triggered; the data quality tag is generated by summarizing the results of integrity verification, anomaly removal, and missing measurement completion; and the audit summary field records the triggering reason, anomaly reason code, and field update summary for this step. The basic data structure is used as input to S200 for solar position and attitude geometry calculation. The field of field geo-time parameter is used for solar geometry generation, and the fields of array geometry parameter and actuator constraint parameter are used for candidate attitude geometry generation. At the same time, the fields of meteorological irradiance parameter, surface albedo parameter and bifacial photovoltaic module electrical parameter are continued to be referenced in subsequent S300 and S400 and are associated with the same acquisition batch number, forming a consistent data link across steps.

[0042] The technical effects of this step can be summarized as follows: By unifying the encapsulation of site geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters into a unified basic data structure, and simultaneously writing version and audit fields, the generation of subsequent solar geometry and candidate attitude geometry structures has a consistent input boundary. Through time alignment, missing measurement completion, anomaly marking, and constraint consistency verification, the irradiance component, thermal environment component, and mechanism constraints maintain traceable correlation within the same acquisition batch. Through the linkage of triggering mechanisms and freeze strategy update channels, the basic data structure maintains a continuous operational link and an auditable state under parameter evolution and equipment anomaly scenarios.

[0043] S200 inputs the basic data structure into the solar position and attitude geometry calculation to obtain the solar geometry structure and candidate attitude geometry structure; this step is triggered by the tracking controller when each control cycle arrives, and is recalculated when the parameter version number of the basic data structure changes, the acquisition batch number changes, or the timing status field changes.

[0044] The tracking controller calls the solar position and attitude geometry calculation module to read the site's geographic time parameters and array geometry parameters from the basic data structure, and simultaneously reads the actuator constraint parameters to complete the feasible angle domain assembly. It then enters the solar position calculation link and attitude geometry generation link. The solar position and attitude geometry calculation module consists of a time axis management unit, a solar position calculation unit, a coordinate transformation unit, an attitude generation unit, a feasible domain projection unit, and a structure output unit. The time axis management unit performs continuity judgment on the timestamp sequence and performs time correction in conjunction with the time synchronization status field. When cross-day jumps, repeated times, or missing times occur, the time axis management unit writes the abnormal segment into the structure abnormal status field, selects the most recent continuous timestamp sequence for subsequent calculations, and writes the abnormal reason code into the audit summary field, forming a traceable record for the same batch.

[0045] Specifically, the solar position calculation unit extracts the longitude, latitude, altitude, time zone, and date / time from the site's geographic time parameters and performs calendar parsing on the date / time to obtain the day sequence information and local time information required for solar position calculation. Then, it calculates the solar altitude angle, solar azimuth angle, and solar incidence direction hour by hour according to the timestamp sequence, and performs horizon discrimination on the solar altitude angle to generate a nighttime marker field. The coordinate transformation unit extracts the rotation axis azimuth angle and rotation axis tilt angle from the array geometric parameters and constructs a rotation transformation parameter set from the site coordinate system to the rotation axis coordinate system by combining the site's longitude and latitude. Simultaneously, it generates an array spatial parameter set from component size, component height above ground, row spacing, array row and column number, and support axis position. When the array geometric parameters contain structural anomaly status fields, the coordinate transformation unit calls the rotation transformation parameter set of the previous version and locks the version number of the structural output unit for the current cycle to avoid multiple versions being used interchangeably within the same control cycle.

[0046] In the attitude geometry generation chain, the attitude generation unit receives the solar altitude angle, solar azimuth angle, solar incidence direction, and nighttime marker fields output by the solar position calculation unit, and receives the rotation transformation parameter set and array space parameter set output by the coordinate transformation unit. Simultaneously, it extracts the tracking angle limit, angular velocity upper limit, angular acceleration upper limit, control cycle, and docking angle from the actuator constraint parameters, and assembles candidate attitude enumeration rules and boundary constraints. The candidate attitude enumeration rules are defined as a set of rules for discretely sampling the tracking angle domain according to the control cycle to form a multi-time-period target angle node set. The boundary constraints are defined as a set of constraints that perform limit clipping, motion amplitude constraints, and docking angle coverage on the discretely sampled target angle nodes. When the nighttime marker field indicates the nighttime state, the attitude generation unit writes the docking angle into the target angle node set and adds a nighttime docking marker to it, forming a nighttime attitude segment and writing it into the candidate attitude geometry structure for direct reference in subsequent irradiance calculations.

[0047] The feasible region projection unit performs feasible region projection repair on the target corner node set output by the attitude generation unit. Projection repair includes two types of processing: limit clipping and motion continuity constraint. Limit clipping projects out-of-bounds target corner nodes to boundary corner nodes based on tracking angle limits and writes limit repair markers. Motion continuity constraint calculates the constraint discrimination value of the target angle change at adjacent times based on the upper limit of angular velocity, the upper limit of angular acceleration, and the control period. When the discrimination value exceeds the upper limit, smooth repair is performed on the target corner node set and motion repair markers are written. The structure output unit unifies the output of the solar position calculation link and the output of the attitude geometry generation link to generate a solar geometry structure and a candidate attitude geometry structure. The solar geometry structure includes a timestamp sequence, solar altitude angle, solar azimuth angle, solar incidence direction, night marker field, and timing status field. The candidate attitude geometry structure includes a timestamp sequence, target corner node set, rotation transformation parameter set, array space parameter set, limit repair markers, and motion repair markers. Parameter version numbers and acquisition batch numbers are written into both types of structures to maintain continuity within the same batch.

[0048] In the engineering embodiment, the tracking controller is deployed in the field control cabinet of the ground-based centralized bifacial photovoltaic single-axis power station. The basic data structure is written into the local buffer after the data acquisition module completes a synchronous sampling with the meteorological sensor, radiometer, angle encoder and configuration library. When the control cycle arrives, the solar position and attitude geometry calculation module reads the basic data structure and completes the solar geometry calculation. Then, it combines the azimuth angle and tilt angle of the rotating shaft to generate a set of rotation transformation parameters. Subsequently, under the tracking angle limit constraint, it generates a set of target angle nodes and writes them into the candidate attitude geometry structure. When the wind load protection threshold is triggered and the docking angle is covered, the attitude generation unit writes the docking angle into the set of target angle nodes and adds a docking mark. At the same time, it writes the change into the audit summary field and keeps the acquisition batch number unchanged, so that the subsequent bifacial irradiance calculation and temperature calculation are completed in the same batch for continuous link processing.

[0049] The solar geometry and candidate attitude geometry output in this step are written into the intermediate data area of ​​the tracking controller and used as input to the S300 for bi-directional irradiation calculation. The solar elevation angle, solar azimuth angle, solar incidence direction, and nighttime marker fields in the solar geometry are used for occlusion discrimination generation and incidence angle calculation. The target corner node set, rotation transformation parameter set, and array space parameter set in the candidate attitude geometry are used for frontal incidence correction coefficient calculation and back view factor calculation. At the same time, the limit repair marker and motion repair marker are used for reference in the abnormal record field of the subsequent occlusion factor link.

[0050] In summary, the technical effects of this step are as follows: By mapping the geographic time and array configuration information within the basic data structure to solar geometry and candidate attitude geometry, subsequent bifacial irradiation calculations have a unified description of solar incidence and executable attitude geometry input. By incorporating tracking angle limits, angular velocity limits, angular acceleration limits, and docking angles into the generation and repair chain of candidate attitude geometry, the attitude geometry output is kept consistent with the actuator constraints. By continuously writing parameter version numbers and acquisition batch numbers into both the solar geometry and candidate attitude geometry, cross-step data integration enables continuous recording and audit traceability.

[0051] S300. Input the solar geometry and the candidate attitude geometry into the bi-face irradiance calculation to obtain the effective front irradiance, the effective back irradiance, and the shading factor.

[0052] This step is automatically triggered by the tracking controller after the S200 output is written to the intermediate data area. Recalculation is triggered when the timing status field of the solar geometry changes, the limit repair mark or action repair mark of the candidate attitude geometry changes, or the acquisition batch number of the basic data structure increases. The bifacial irradiance calculation is performed by the bifacial irradiance calculation module, which consists of an incident angle calculation unit, a front irradiance synthesis unit, a back irradiance synthesis unit, a back viewpoint factor calculation unit, an occlusion discrimination unit, a quality control unit, and a structure output unit. The incident angle calculation unit reads the timestamp sequence, solar altitude angle, solar azimuth angle, solar incidence direction, and nighttime marker field from the solar geometry, and reads the target angle node set, rotation transformation parameter set, and array space parameter set from the candidate attitude geometry. Simultaneously, it loads the meteorological irradiance parameters and the surface albedo parameters from the basic data structure according to the acquisition batch number, forming a unified input frame for this step. This unified input frame is then written to the audit summary field for subsequent steps to trace the calculation link within the same batch.

[0053] Specifically, the incident angle calculation unit projects the solar incident direction onto the rotation axis coordinate system based on the rotation transformation parameter set, and generates candidate attitude normal directions time-by-time in conjunction with the target angle node set, thereby obtaining an incident angle sequence. The incident angle sequence is defined as the sequence of angles between the solar incident direction and the candidate attitude normal directions. The incident angle sequence includes an incident angle field of the same length as the timestamp sequence and an attitude identifier field. Further, the frontal irradiance synthesis unit reads direct irradiance, diffuse irradiance, and total irradiance from the meteorological irradiance parameters, performs incident angle projection processing on the direct irradiance, and performs sky anisotropic distribution mapping processing on the diffuse irradiance to generate a frontal incident correction coefficient. The frontal incident correction coefficient is defined as a correction field driven by the incident angle sequence, including a direct correction component, a diffuse correction component, and a nighttime suppression component. The nighttime suppression component is controlled by the nighttime marker field. When the nighttime marker field indicates the nighttime state, the incident angle calculation unit writes the incident angle sequence into the nighttime marker and writes the frontal incident correction coefficient into the zero-value marker, so that the subsequent synthesis link maintains time alignment and does not introduce invalid irradiance.

[0054] In the back-side irradiation link, the back-side view factor calculation unit reads the component ground height, row spacing, component size, array row and column number, and support axis position from the array spatial parameter set, and reads the surface albedo reference value, zonal albedo table, and time-varying albedo factor from the surface albedo parameters to construct a description of the ground visible domain. The back-side view factor is characterized by a view factor, which is defined as the geometric proportion field of the ground visible domain in the back-side facing half-space, including geometric view component, zonal albedo component, and attitude-related component. Further, the back-side view factor calculation unit updates the attitude-related component time-by-time according to the target corner node set, and maps the ground visible domain description to the array row number mapping to generate row-specific back-side view factors. When the zonal albedo table is missing or the albedo confidence flag is in a low-confidence state, the quality control unit uses the surface albedo reference value as a backoff input and writes it into the acquisition source flag and missing measurement source field to keep the back-side view factor field available. The backside irradiance synthesis unit converts the total irradiance and the scattered irradiance into ground-received irradiance components, and applies the ground surface albedo reference value and albedo time-varying factor to the ground-reflected irradiance components. Then, it combines the backside view factor to output a backside candidate irradiance sequence. The backside candidate irradiance sequence is aligned with the timestamp sequence and includes a row number identifier field for occlusion discrimination link association.

[0055] The occlusion discrimination unit reads the solar altitude angle, solar azimuth angle, and timestamp sequence from the solar geometry, and reads the row spacing, component height above ground, component size, and number of array rows and columns from the array spatial parameter set to generate an occlusion discrimination quantity. This occlusion discrimination quantity is defined as an intermediate field characterizing the occlusion geometric relationship between adjacent array rows, including a shadow projection length field, an occlusion start and end time field, and an occlusion coverage ratio field. Further, the occlusion discrimination unit compares the shadow projection length field with the row spacing using a threshold to obtain an occlusion state sequence, and maps the occlusion coverage ratio field to the front and back irradiance links to generate an occlusion factor. This occlusion factor is defined as the proportion of attenuation applied to the front and back candidate irradiance sequences, including a front occlusion factor component and a back occlusion factor component, and is consistent with the timestamp sequence and row number identifier field. The quality control unit performs continuity and jump checks on the occlusion factor. When a non-continuous segment is found in the occlusion start and end time field, the segment is written into the structural anomaly status field and the occlusion factor version number corresponding to the collection batch number is locked for subsequent steps to consistently reference the data of the same batch.

[0056] In the irradiation output link, the structure output unit fuses the frontal incident correction coefficient generated by the frontal irradiation synthesis unit with direct irradiation and diffuse irradiation to obtain a frontal candidate irradiation sequence, and applies the frontal shading factor component in the shading factor to the frontal candidate irradiation sequence to generate effective frontal irradiation; at the same time, the structure output unit fuses the backal candidate irradiation sequence generated by the backal irradiation synthesis unit with the backal shading factor component in the shading factor to generate effective backal irradiation. Both the effective frontal irradiance and the effective rearal irradiance are written into the same output data frame. The output data frame includes a timestamp sequence, a row number identifier field, an attitude identifier field, a frontal effective irradiance field, a rearal effective irradiance field, an occlusion factor field, a collection batch number field, and a parameter version number field. The output data frame is written into the intermediate data area and is used by the S400. The frontal effective irradiance, rearal effective irradiance, and occlusion factor are used as input fields for the S400's temperature calculation and electrical output calculation. The attitude identifier field and the row number identifier field are used by the S400 to associate the candidate attitude power sequence with the candidate attitude geometry. The collection batch number field is used by the S400 to perform batch consistency comparison with the basic data structure.

[0057] In the engineering embodiment, the tracking controller is deployed in the field control cabinet of the ground-based centralized bifacial photovoltaic single-axis power station. The meteorological irradiance parameters are collected in real time by the station's meteorological station and written into the basic data structure. The surface albedo parameters are written into the station configuration library and updated by the operation and maintenance terminal when the sand or snow cover status changes. After the control cycle arrives, the bifacial irradiance calculation module reads the solar geometry and candidate attitude geometry, and generates multiple sets of incident angle sequences according to the target angle node set. Then, it generates the front incident correction coefficient and synthesizes the front candidate irradiance sequence. The back view factor calculation unit generates the row back view factor according to the component's ground height and row spacing, and maps the partitioned albedo table to the array row number identifier field. The back irradiance synthesis unit generates the back candidate irradiance sequence accordingly. The shading discrimination unit generates the shading discrimination quantity and outputs the shading factor according to the solar altitude angle and row spacing. The structure output unit applies the shading factor to the front and back links simultaneously, outputs the front effective irradiance and the back effective irradiance and writes them into the output data frame, and simultaneously writes the audit summary field for subsequent verification.

[0058] In summary, the technical effects of this step are as follows: By driving the linkage calculation of the incident angle sequence, back view factor, and occlusion discrimination quantity through the solar geometry and candidate attitude geometry, the effective irradiance on the front and back sides is aligned with the occlusion factor in the same batch for output. By incorporating the surface albedo parameters, component height above ground, and row spacing into the back view factor calculation chain, the structured fields of the effective irradiance on the back side are kept consistent with the array geometry. By recording the attitude identifier field, row sequence identifier field, acquisition batch number field, and parameter version number field of the output data frame, the temperature calculation and electrical output calculation of the S400 obtain traceable input boundaries.

[0059] S400 calculates the effective front irradiance, the effective back irradiance, the shading factor, and the basic data structure input temperature and electrical output to obtain the candidate attitude power sequence and energy index;

[0060] This step is triggered by the tracking controller after the output data frame of S300 is written into the intermediate data area, and is triggered for recalculation when the acquisition batch number of the output data frame is incremented, the parameter version number of the basic data structure is incremented, or the attitude identifier field of the output data frame is changed. Temperature calculation and electrical output calculation are performed by a temperature calculation and electrical output calculation module, which consists of an input alignment unit, a thermal balance solution unit, an equivalent irradiance generation unit, a parameter temperature correction unit, a single diode five-parameter solution unit, a maximum power point search unit, an energy summarization unit, and a structure output unit. The input alignment unit reads the timestamp sequence, row number identifier field, attitude identifier field, front effective irradiance field, back effective irradiance field, and shading factor field from the output data frame. It then loads the ambient temperature and wind speed from the meteorological irradiance parameters, the back gain coefficient and temperature coefficient from the bifacial photovoltaic module electrical parameters, and the control cycle from the actuator constraint parameters from the basic data structure according to the acquisition batch number, thereby forming an aligned input frame under the same attitude identifier field. When the aligned input frame has missing timestamp sequences, missing ambient temperatures, or missing wind speeds, the input alignment unit writes the missing fragment into the structure abnormal state field and calls the corresponding field from the previous acquisition batch number for backfilling. Simultaneously, it writes the backfill source into the audit summary field, ensuring continuity of the subsequent solution chain within the same batch.

[0061] Specifically, the thermal balance solution unit receives the front effective irradiance field, the back effective irradiance field, the shading factor field, ambient temperature, and wind speed from the aligned input frame, and loads a thermal configuration subset from the electrical parameters of the bifacial photovoltaic module. This thermal configuration subset includes mapping parameters for module absorptivity, module emissivity, module heat capacity, and convective heat transfer coefficient. The module absorptivity is used to convert the front and back effective irradiance fields into irradiance absorption terms, the module emissivity is used to construct radiative heat transfer terms, and the convective heat transfer coefficient mapping parameters are used to map wind speed into convective heat transfer terms. Understandably, the shading factor field has already participated in the irradiance attenuation output in S300, but the thermal balance solution unit still reads the shading factor field and generates a shading thermal correction flag. This shading thermal correction flag is used to record the effective irradiance change and thermal inertia processing state within the shading duration, thereby establishing temperature evolution constraints between adjacent time stamps. The thermal balance solution unit calculates the component operating temperature step-by-step according to the timestamp sequence. The solution process includes three stages: initialization, iterative update, and convergence determination. In the initialization stage, the component operating temperature of the previous moment is used as the initial value for the current moment, and boundary clipping is performed in conjunction with the ambient temperature. In the iterative update stage, the temperature update is calculated based on the irradiance absorption term, convective heat transfer term, and radiative heat transfer term and written into the iteration status field. In the convergence determination stage, the stability of the temperature update is written into the convergence flag field. When the convergence flag field indicates that the iteration has not converged or an abnormal jump occurs, the thermal balance solution unit reverts the component operating temperature of the current moment to the component operating temperature of the previous stable moment and writes it into the revert flag field. At the same time, the abnormal reason code is written into the audit summary field to achieve traceable processing of abnormal operating conditions. The resulting component operating temperature is consistent with the timestamp sequence, attitude identifier field, and row number identifier field, and is used as the temperature input field for subsequent single diode five-parameter solution.

[0062] In the electrical output calculation link, the equivalent irradiance generation unit receives the front effective irradiance field, the back effective irradiance field, and the back gain coefficient, and generates the equivalent irradiance. The equivalent irradiance is defined as the irradiance input field after synthesizing the front effective irradiance field and the back effective irradiance field corrected by the back gain coefficient. The equivalent irradiance corresponds one-to-one with the attitude identification field, the row number identification field, and the timestamp sequence and is written into the equivalent irradiance mark field to identify the version number of the back gain coefficient used for synthesis. The parameter temperature correction unit reads the nominal parameters and temperature coefficient of the single diode five-parameter model from the electrical parameters of the bifacial photovoltaic module, and inputs the module operating temperature into the temperature correction link. The temperature correction link performs temperature correction on the photocurrent, reverse saturation current, series resistance, parallel resistance, and ideal factor and writes them into the correction parameter set. Among them, the photocurrent is related to the equivalent irradiance and is updated by the equivalent irradiance, the reverse saturation current is updated by the module operating temperature, the series resistance and parallel resistance are corrected by the temperature coefficient, and the ideal factor is corrected by the nominal parameters and the temperature coefficient. The correction parameter set is output step by step according to the timestamp sequence and remains consistent within the same batch. The single-diode five-parameter solving unit receives the corrected parameter set and equivalent irradiance, and performs current-voltage characteristic solving to generate candidate attitude voltage-current state fields. The maximum power point search unit performs power search on the candidate attitude voltage-current state fields and outputs candidate attitude DC power fields. The candidate attitude DC power fields are aligned with the timestamp sequence, attitude identifier field, and row number identifier field. When the maximum power point search unit detects an infeasible interval in the search state field, it backtracks the candidate attitude DC power field at that moment to the interpolated power of the adjacent moment and writes it into the interpolation mark field. At the same time, it writes the correction record of the infeasible interval into the audit summary field. Furthermore, the temperature calculation and electrical output calculation module is also equipped with an array-level aggregation sub-link. The array-level aggregation sub-link reads the component size and array row and column number from the basic data structure, and performs aggregation and array aggregation on the candidate attitude DC power field under the same attitude identifier field to generate a candidate attitude power sequence. The candidate attitude power sequence is defined as a power sequence field set arranged by timestamp sequence, including the attitude identifier field, timestamp sequence and candidate attitude DC power field, and includes the component operating temperature field and equivalent irradiance field as homogeneity verification fields for consistency verification and auditing in the subsequent fitness construction stage.

[0063] The energy aggregation unit performs energy aggregation processing on the candidate attitude power sequence. This processing performs time-weighted accumulation of the DC power fields of candidate attitudes at adjacent timestamps based on the control cycle, generating an energy index. The energy index is defined as a set of energy fields associated with the attitude identifier field, including a time-division energy field and a cumulative energy field. The time-division energy field is aligned with the timestamp sequence and records the energy increment for each control cycle. The cumulative energy field records the cumulative energy value within the coverage area of ​​the timestamp sequence. Both are written into the acquisition batch number field and the parameter version number field. Understandably, the energy aggregation unit simultaneously reads the occlusion factor field and generates an occlusion statistics field. This occlusion statistics field records the duration of occlusion in the occlusion state and the summary result of the coverage ratio, and is written as an accompanying field of the energy index into the same structured output packet for subsequent penalty term assembly by the feasible domain processor and fitness evaluator, without changing the accumulation method of the energy index. The structural output unit encapsulates the candidate attitude power sequence and energy index into an output data packet and writes it into the intermediate data area. The output data packet includes an attitude identifier field, a timestamp sequence, a candidate attitude power sequence field, an energy index field, a component operating temperature field, an equivalent irradiance field, an occlusion statistics field, a data acquisition batch number field, and a parameter version number field. The candidate attitude power sequence and the energy index are used as the energy index input and the basic evaluation input for fitness construction of the S500 in subsequent steps, respectively. The energy index field is directly input into fitness construction and feasible region processing, while the candidate attitude power sequence field is used by the fitness evaluator to perform alignment comparisons between different attitude identifier fields and generate fitness values.

[0064] In the engineering embodiment, the tracking controller is installed in the field control cabinet of the ground-based centralized bifacial photovoltaic single-axis power station. The output data frame is output by S300 in each control cycle. The basic data structure is maintained by the site configuration library and the meteorological station data access module and updated according to the collection batch number. When the control cycle arrives, the temperature calculation and electrical output calculation module reads the front effective irradiance field, the back effective irradiance field, and the shading factor field from the intermediate data area, and reads the ambient temperature and wind speed. After the thermal balance solution unit calculates the component operating temperature, the equivalent irradiance generation unit generates the equivalent irradiance according to the back gain coefficient. The temperature correction unit updates the correction parameter set of the single diode five-parameter model accordingly. The single diode five-parameter solution unit and the maximum power point search unit output the candidate attitude DC power field and form the candidate attitude power sequence. The energy aggregation unit outputs the energy index cumulatively according to the control cycle. When the sensor fails to measure for a short time or the timing jitter causes the timestamp sequence to be abnormal, the input alignment unit performs backfilling and records the audit summary field. The thermal balance solution unit performs back-off marking field recording for non-converged segments to ensure that the output data packets under the same acquisition batch number maintain structural consistency and enter the S500 fitness construction link.

[0065] In summary, the technical effects of this step are as follows: By linking the effective irradiance of the front and back sides with the shading factor in the basic data structure, along with ambient temperature, wind speed, and the electrical parameters of the bifacial photovoltaic modules, a candidate attitude power sequence and energy index aligned with the attitude identification field are formed. Through the connection between thermal balance solution and the temperature correction link of the single-diode five-parameter model, the module operating temperature, equivalent irradiance, and candidate attitude DC power fields are traceably associated under the same acquisition batch number and parameter version number. Through structured encapsulation of the output data packet and anomaly backfilling rollback records, the S500 obtains a stable and consistent energy index input boundary and completes subsequent fitness construction and feasible region processing.

[0066] S500: Input the energy index and the actuator constraint parameters into the fitness construction and feasible region processing to obtain the fitness evaluator and feasible region processor;

[0067] This step is triggered by the tracking controller after the output data packet of the S400 is written to the intermediate data area. The triggering conditions include the update of the acquisition batch number of the output data packet, the update of the parameter version number of the basic data structure, or the change of the actuator constraint parameters. Specifically, the tracking controller reads the energy index field associated with the attitude identification field from the output data packet and reads the actuator constraint parameter field from the basic data structure. Then, it sends the two types of inputs to the fitness construction and feasible domain processing module. The fitness construction and feasible domain processing module consists of an input alignment unit, an index normalization unit, a penalty term assembly unit, a fitness aggregation unit, a boundary projection repair unit, a dynamic consistency repair unit, a feasibility discrimination unit, and a structure output unit. The input alignment unit is used to align the energy index to the same evaluation domain according to the attitude identification field and bind the actuator constraint parameters to the evaluation context of the same acquisition batch number to form an evaluation context structure. The evaluation context structure will subsequently serve as the common input base for the fitness evaluator and the feasible domain processor and be written into the audit summary field for version consistency verification and traceability when the genetic algorithm calls for optimization.

[0068] Furthermore, the energy index is defined as a set of energy fields output by S400, including a time-division energy field aligned with the timestamp sequence and a cumulative energy field accumulated within the coverage area of ​​the timestamp sequence, and may include an occlusion statistics field associated with the occlusion factor as an accompanying field; the actuator constraint parameters are defined as a set of constraint fields describing the working boundary and discrete control cycle of the tracking actuator, including tracking angle limit, angular velocity limit, angular acceleration limit, control cycle and docking angle, wherein the tracking angle limit and control cycle constitute the core minimum set for feasible region processing, the angular velocity limit and angular acceleration limit are used for dynamic consistency repair and penalty term calculation, and the docking angle is used as an angle reference to mark the docking state in the evaluation context and participates in the judgment of action frequency and limit violation. The index normalization unit receives the energy index and performs scale alignment processing within the same acquisition batch number. The scale alignment processing includes interval mapping of the cumulative energy field within the attitude identification field, performing a timestamp sequence consistency check on the time-division energy field and generating a valid time window marker. When a timestamp sequence is missing, the missing segment is written into the structural anomaly state field and backfilled using the energy increment of adjacent valid time windows, so that the energy input on which fitness aggregation is based is structurally continuous.

[0069] In the fitness construction process, the penalty item assembly unit performs constraint compliance measurement on the candidate strategies corresponding to each attitude identifier field based on the actuator constraint parameters, and generates a penalty item field set. This set includes limit violation penalties, angular velocity violation penalties, angular acceleration violation penalties, occlusion duration penalties, and action frequency penalties. Specifically, the limit violation penalty is obtained by comparing the angle sequence corresponding to the candidate strategy with the tracking angle limit. When the angle sequence exceeds the tracking angle limit, the penalty item assembly unit encodes the excess amplitude and excess duration into a limit violation segment field and generates the limit violation penalty field accordingly. The angular velocity violation penalty is obtained by comparing the velocity sequence calculated by combining the angle change within adjacent control cycles with the control cycle with the angular velocity upper limit. When the velocity sequence contains an interval exceeding the angular velocity upper limit, the penalty item assembly unit writes the corresponding interval into the velocity violation segment field and generates the angular velocity violation penalty field. The angular acceleration violation penalty is obtained by comparing the acceleration sequence calculated by combining the velocity change within adjacent control cycles with the control cycle with the angular acceleration upper limit, and similarly records the acceleration violation segment field to form the angular acceleration violation penalty field. The occlusion duration penalty is calculated by extracting the occlusion duration segment from the occlusion statistics field in the energy index and jointly calculating it with the effective time window marker of the timestamp sequence. An occlusion duration penalty field is generated when the occlusion duration segment covers the evaluation time window. The action frequency penalty is obtained by statistically analyzing the effective adjustment counts of the angle sequence within a continuous control cycle. This statistical analysis is achieved by counting the number of angle change direction switches and adjustment segments in non-docked states, and then combining this with the docking angle to remove docking segments, thus forming the action frequency penalty field. The fitness aggregation unit uses the normalized cumulative energy field or time-division energy field as the energy base value input and performs a combination operation using the penalty item field set as the penalty input, outputting a fitness value field. The fitness value field corresponds one-to-one with the attitude identifier field and carries a collection batch number field and a parameter version number field, allowing it to be directly used for fitness comparison and elite retention judgment during genetic algorithm optimization.

[0070] In the feasible domain processing link, the boundary projection repair unit performs boundary projection repair on candidate strategies that exceed the tracking angle limit. The boundary projection repair includes performing limit boundary projection on the exceeding angle points, performing local smoothing on the discontinuous segments generated after projection, and writing the repair mark field. The local smoothing is achieved by symmetrically distributing the angle change within adjacent control cycles, so that the repaired angle sequence remains continuous under discrete control beats. After the boundary projection repair, the dynamic consistency repair unit further performs consistency repair on the angular velocity and angular acceleration constraints. The consistency repair includes performing amplitude clipping and backward propagation compensation on the angle change exceeding the upper limit of angular velocity, performing cycle-by-cycle amplitude limiting and forward backtracking adjustment on the velocity change exceeding the upper limit of angular acceleration, and recording the propagation depth field and backtracking depth field for each propagation and backtracking, so that the feasibility judgment unit can determine whether the repair has converged. The feasibility determination unit generates a feasibility marker field based on the repair marker field, propagation depth field, backtracking depth field, and structural anomaly state field. When the feasibility marker field indicates infeasibility, the candidate strategy is written into the infeasible individual marker field and simultaneously into the audit summary field, enabling the fitness evaluator to apply corresponding penalties or execute elimination strategies for infeasible individuals during subsequent evaluations. The structure output unit encapsulates the fitness value field and its calculation rules from the fitness construction link output into a fitness evaluator, and encapsulates the boundary projection repair rules, dynamic consistency repair rules, and feasibility determination rules from the feasible domain processing link output into a feasible domain processor. The fitness evaluator is recorded as a fitness evaluator field and used as the genetic algorithm optimization input of S600, and the feasible domain processor is recorded as a feasible domain processor field and also used as the genetic algorithm optimization input of S600. Both are bound to the acquisition batch number field and parameter version number field and written into the intermediate data area, thereby completing the consistent loading of the same batch when S600 is called.

[0071] In the engineering embodiment, the tracking controller is deployed within the site's edge computing gateway. The edge computing gateway periodically receives energy indicators output by the S400 and accesses actuator constraint parameter change events reported by the actuator driver. When changes in sunlight conditions cause updates to the candidate attitude power sequence and energy indicators, the tracking controller triggers the fitness construction and feasible domain processing module to rebuild the fitness evaluator and feasible domain processor, writing them into the optimization task queue for the genetic algorithm to read. When maintenance personnel adjust the tracking angle limit or control cycle, the parameter version number updates with the change event, and the fitness evaluator and feasible domain processor generate new versions while retaining the audit summary field of the previous version. During the genetic algorithm's optimization, the matching version is selected based on the acquisition batch number field and the parameter version number field to avoid cross-batch misuse. Furthermore, when the surface albedo parameter or the backside gain coefficient is updated and written back to the basic data structure, the incrementing parameter version number triggers this reconstruction step, ensuring that the constraint context and energy input upon which penalty term assembly and feasibility judgment depend remain traceable and consistent.

[0072] The technical effects of this step can be summarized as follows: By aligning the energy index and the actuator constraint parameters under the same acquisition batch number and parameter version number, a fitness evaluator and feasible region processor that can be directly invoked by the genetic algorithm for optimization are formed. Through the linkage of penalty term assembly and boundary projection repair, the fitness value field and the feasibility marker field maintain consistent evaluation criteria and feasible region constraint expression within the discrete control cycle. Through the audit summary field and version binding writing mechanism, the fitness evaluation and feasible region processing have traceable operation records and stable loading paths during parameter evolution.

[0073] S600: Input the fitness evaluator and the feasible region processor into the genetic algorithm for optimization, output the optimal strategy parameters and angle curve, and write the angle curve into the control parameter table to generate the target angle instruction. This step is triggered by the tracking controller after the fitness evaluator and feasible region processor are packaged and output in S500. The triggering conditions include the new round of energy index update, the change of the execution mechanism constraint parameters, the control cycle reaching the predetermined recalculation point, or the parameter version number field incrementing event. Specifically, the tracking controller reads the fitness evaluator and feasible region processor from the intermediate data area, and reads the parameter version number field and the acquisition batch number field bound to them to form an optimization task context. The optimization task context is used to limit the input consistency range of the current round of genetic algorithm. When input is missing, version mismatch, or batch inconsistency occurs, the exception is written to the running log field and a rollback strategy is triggered. The rollback strategy includes directly using the angle curve of the previous version or writing the docking angle into the control parameter table to enter the conservative control mode, so that the optimization link has an auditable state transition record during the project operation.

[0074] The genetic algorithm optimization uses real-number encoding to map the optimal strategy parameters to chromosomes. The optimal strategy parameters are defined as a set of strategy fields that can be written into the control parameter table by the tracking controller and can directly generate target angle commands. This minimum set includes axis tilt parameters, backtracking correction coefficient parameters, and piecewise angle curve node parameters. The axis tilt parameters describe the tilt configuration of the tracking axis in the array coordinate system and form a consistent geometric constraint entry with the array geometric parameters. The backtracking correction coefficient parameters describe the backtracking amplitude adjustment rules under row spacing and occlusion factor constraints. The piecewise angle curve node parameters describe the values ​​of target angle nodes at different time periods on the time index and the transition forms between nodes. Furthermore, chromosome generation is completed by an encoding and mapping unit. This unit concatenates the axis tilt parameters, backtracking correction coefficient parameters, and piecewise angle curve node parameters into a real-number sequence according to a predefined field order and writes it into the chromosome field. Simultaneously, it writes a boundary description field to indicate the value range corresponding to each field. The value range is at least constrained by the tracking angle limit and can be indirectly constrained by the upper limit of angular velocity and the upper limit of angular acceleration, forming implicit boundaries. For the generation of the initial population, after reading the historical angle curve segments of the control parameter table, the population initialization unit samples the nodes according to the time index and uses them as the source of initial values ​​for some individuals. At the same time, it randomly samples and generates the remaining individuals within the scope of the boundary description field. After generation, each chromosome is input into the feasible domain processor to perform boundary projection repair and dynamic consistency repair, outputs the feasible chromosome field and records the repair marker field. The feasible chromosome field serves as the input object for the subsequent fitness evaluator.

[0075] In each iteration, the genetic algorithm scheduling unit sequentially executes the Elite Retention operator, the Tournament Selection operator, the Simulated Binary Crossover operator, and the Gaussian Mutation operator, with the algebraic index field driving the adaptive update of the operator parameters. Specifically, the Elite Retention operator sorts the current population according to the fitness value field output by the fitness evaluator, directly retaining several individuals with higher fitness values ​​to enter the next generation elite pool, and writing the chromosome field and fitness value field of the elite pool individuals into the elite record field for subsequent convergence judgment; the Tournament Selection operator extracts candidate adversarial groups from the non-elite individual set according to a preset adversarial scale, calls the fitness evaluator to output the fitness value field of each candidate adversarial group, and selects the best to form the parent set, while writing the random seed summary of the selection process into the audit summary field to support the reproduction of the experiment. The simulated binary crossover operator performs real-field crossover on pairs of individuals in the parent set, resulting in offspring chromosome fields. When the crossover causes the offspring chromosome field to touch the boundary description field or trigger an implicit default related to the tracking angle limit, the genetic algorithm scheduling unit immediately inputs the offspring chromosome field into the feasible region processor to perform boundary projection repair and dynamic consistency repair, outputs the repaired chromosome field and updates the repaired label field, and then sends the repaired chromosome field to the fitness evaluator to output the fitness value field, thus preventing infeasible individuals from continuously spreading within the population. The Gaussian mutation operator performs perturbation updates on the selected offspring chromosome field. These updates are triggered by the mutation rate field, which is dynamically updated based on the generation index field, obtained by combining the current iteration generation and the maximum iteration generation. When the generation index field indicates a later stage of iteration, the mutation rate field is lowered, prioritizing small perturbations on local nodes in the piecewise angle curve node parameters. When the generation index field indicates an early stage of iteration, the mutation rate field is raised, allowing for larger perturbations on the backtracking correction coefficient parameters and node time positions. This allows for differentiated exploration and convergence rhythms at different stages of the search. After each generation completes crossover and mutation, the genetic algorithm scheduling unit inputs the new population into the feasible region processor, outputting the feasible chromosome field and writing the feasibility marker field into the population state field. Subsequently, the fitness evaluator is called to output fitness values ​​in batches and write them into the population evaluation field for reuse in the next generation selection and elite retention.

[0076] Convergence discrimination and termination control are performed by a termination determination unit. This unit reads the elite record field, population evaluation field, and iteration count field. It terminates the iteration when the maximum number of iterations is reached or the fitness value field changes less than a preset threshold over several consecutive generations. The unit then reverse-maps the chromosome field of the final elite individual to the optimal strategy parameters. Specifically, the reverse mapping is performed by a decoding mapping unit. This unit splits the chromosome field into rotation axis tilt parameters, backtracking correction coefficient parameters, and segmented angle curve node parameters according to the field order of the encoding mapping unit, and writes this parameter set into the optimal strategy parameter field. Subsequently, the angle curve generation unit generates an angle curve based on the segmented angle curve node parameters. The angle curve is defined as a time-series structure composed of multi-time-period target angle nodes. Each multi-time-period target angle node includes a time index field and a target angle field, and may also include a node interpolation mode field to indicate the transition mode between nodes. After the angle curve is generated, it is written into the control parameter table. The writing process is completed by the parameter table writing unit. The parameter table writing unit stores the target angle nodes of multiple time periods according to the time index field, and writes the parameter version number field and the acquisition batch number field into the table header metadata field, so that the control parameter table can be aligned with the input batch of the current optimization when read in subsequent cycles. The target angle command is generated by the command generation unit. The command generation unit reads the target angle field of the adjacent time index from the control parameter table according to the control cycle and performs linear interpolation to generate the target angle command. The target angle command is sent to the actuator driver and synchronously written into the command issuance record field. When the actuator driver reports that the upper limit of angular velocity or the upper limit of angular acceleration has triggered the amplitude limiting state, the command generation unit writes the amplitude limiting state into the abnormal state field and prioritizes calling the feasible domain processor in the next control cycle to perform fast repair on the local angle curve node, forming a locally repaired angle curve and writing it back into the control parameter table, so that the control link and the optimization link form a traceable closed loop connection in the engineering operation. Understandably, the optimal strategy parameter field and the angle curve field are recorded as reference inputs when constructing the running data structure. They are used to associate the measured angle with the target angle command in the same batch, and participate in the re-optimization after the basic data structure is updated as the basis for historical strategy when parameter correction is triggered.

[0077] In the engineering implementation, in the single-axis tracking scenario of a ground-based centralized bifacial photovoltaic power station, the tracking controller is deployed on the station's edge computing server and connected to the actuator driver via a fieldbus. The control parameter table is stored in a non-volatile storage medium and managed by the parameter version number field. When the weather monitoring equipment reports a significant change in meteorological irradiance parameters and the S400 outputs an energy index refreshed, the tracking controller triggers this step to execute a genetic algorithm for optimization. In the initialization phase, angle curve segments from the same time period of the previous day in the control parameter table are sampled to form a partial initial population. At the same time, the remaining individuals are generated by combining the tracking angle limit and the docking angle. Then, selection, crossover, and mutation are iteratively executed, and the feasible domain processor is called in each generation to repair infeasible offspring. After the iteration terminates, the optimal strategy parameter field is written to the strategy cache and the angle curve field is generated and written back to the control parameter table. When the control cycle arrives, the instruction generation unit reads the target angle node from the control parameter table and interpolates and outputs the target angle instruction to drive the actuator to rotate. If the actuator is detected to have entered the amplitude limiting state on site, the local repair process is triggered and the control parameter table is written back. All related events are recorded with the parameter version number field and the collection batch number field to facilitate subsequent operation and maintenance traceability and parameter correction link call.

[0078] In summary, the technical effects of this step are as follows: By incorporating the fitness evaluator and feasible domain processor into the iterative closed loop of the genetic algorithm optimization, the optimal policy parameters and angle curves are generated within the same input batch and version context and can be directly written into the control parameter table. By storing the angle curves by time index and reading interpolation according to the control cycle, the target angle command forms a continuous, engineering-executable output under the constraints of the execution mechanism. By introducing batch and version records into the optimization, table writing, and distribution links, the control parameter table and target angle command have a traceable generation path and form a cross-step connection with the subsequent running data structure construction.

[0079] Example 2: Figure 2 A structural block diagram of a bifacial photovoltaic tracking system based on a genetic algorithm according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include:

[0080] The basic data structure construction module 01 is used to acquire site geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters, and to perform field validation, unit normalization, and time base alignment. It then generates a basic data structure and outputs it to the solar position and attitude geometry calculation module. Simultaneously, it outputs the basic data structure to the temperature calculation and electrical output calculation module. Specifically, the basic data structure construction module receives the externally input site geographic time parameters, which include site latitude and longitude information, altitude information, time zone information, and date and time index information. It then performs format one on the date and time index information. Consistency checks and time zone consistency checks are performed to form a unified time index sequence. The basic data structure construction module receives the array geometric parameters, which include the azimuth angle of the rotation axis, the tilt angle of the rotation axis, the height of the component above the ground, the row spacing, the component size, and the number of array rows and columns. It performs consistency checks on the value range of the geometric parameters and the physical constraints, and performs combination relationship checks on the component size and the number of array rows and columns to form an array geometric constraint set. The basic data structure construction module receives the surface albedo parameters and checks their effective range, receives the meteorological irradiance parameters and checks the irradiance dimensions and sampling time index, and receives the electrical parameters of the bifacial photovoltaic modules and checks the completeness of the parameters. The system performs line verification, receiving the constraint parameters of the actuator and performing boundary consistency verification on the tracking angle limit, angular velocity upper limit, angular acceleration upper limit, control cycle, and docking angle. After the field verification passes, it performs unit normalization processing on the array geometric parameters, the meteorological irradiance parameters, and the electrical parameters of the bifacial photovoltaic module, and aligns the time index of the meteorological irradiance parameters with the control cycle to obtain a meteorological irradiance sequence aligned with the time reference. When there are missing fields, discontinuous time indexes, or inconsistent unit identifiers, the basic data structure construction module records the abnormal state and associates the abnormal state with the corresponding time index into the basic data structure. The basic data structure construction module will... By verifying and completing normalization and alignment processing, the site geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters are associated and mapped, and the basic data structure is encapsulated to generate the basic data structure. The basic data structure provides the site geographic time parameters and array geometric parameters as inputs to the solar position and attitude geometry calculation module, and simultaneously provides the meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters as inputs to the temperature calculation and electrical output calculation module.

[0081] The solar position and attitude geometry calculation module 02 is used to connect the basic data structure to the solar position and attitude geometry calculation, generate a solar geometric structure based on the site geographic time parameters and the array geometric parameters, and generate candidate attitude geometric structures in combination with the actuator constraint parameters. It then outputs the solar geometric structure and the candidate attitude geometric structures to the dual-sided irradiance calculation module. Specifically, the solar position and attitude geometry calculation module receives the basic data structure output by the basic data structure construction module, reads the date and time index information, time zone information, and site latitude and longitude information from the site geographic time parameters, and aligns them with the control cycle to generate a unified time index sequence. During the solar position and attitude geometry calculation process, it solves for the geometric position parameters corresponding to the sun at the site position moment by moment for the time index sequence, and binds and encapsulates the geometric position parameters with the date and time index information to generate the solar geometric structure. When the date and time index information has a jump or exceeds the valid range of the site geographic time parameters, the solar position and attitude geometry calculation module marks that moment as "none". The solar position and attitude geometry calculation module reads the rotation axis azimuth, rotation axis tilt, component height above ground, row spacing, component size, and number of array rows and columns from the array geometry parameters, constructs a geometric description of the array attitude, and establishes a correspondence with the time index of the solar geometry. Simultaneously, it reads the tracking angle limit, angular velocity limit, angular acceleration limit, control cycle, and docking angle from the actuator constraint parameters, and applies angle boundary and rate of change boundary constraints to the candidate attitude generation process. The candidate attitude geometry generates a set of candidate angle states on the time index sequence, and associates each candidate angle state with the array geometry description, rotation axis azimuth, and rotation axis tilt, forming a temporal encapsulation of the candidate attitude geometry. The solar position and attitude geometry calculation module provides the solar geometry and the candidate attitude geometry to the dual-sided irradiation calculation module. The dual-sided irradiation calculation module calls the time index and geometric position parameters of the solar geometry and the candidate angle state set of the candidate attitude geometry to complete the dual-sided irradiation calculation.

[0082] The bifacial irradiance calculation module 03 is used to connect the solar geometry and the candidate attitude geometry to bifacial irradiance calculation, calculate the incident angle based on the candidate attitude geometry and generate a frontal incident correction coefficient, calculate the back view factor based on the surface albedo parameter, the component's ground clearance and the row spacing, calculate the shading discrimination value based on the solar geometry and the row spacing and generate a shading factor, generate the frontal effective irradiance and the back effective irradiance and output the frontal effective irradiance, the back effective irradiance and the shading factor to the temperature calculation and electrical output calculation module; specifically, the bifacial irradiance calculation module... The block receives the solar geometry and candidate attitude geometry output by the solar position and attitude geometry calculation module, and reads the surface albedo parameters, meteorological irradiance parameters, component height above ground, and row spacing from the basic data structure. It aligns the meteorological irradiance parameters with the solar geometry on the time index sequence. The dual-sided irradiance calculation module calculates the frontal incidence angle for the candidate angle state set in the candidate attitude geometry, combining it with the geometric position parameters of the solar geometry, and generates a frontal incidence correction coefficient based on the frontal incidence angle. The frontal incidence correction coefficient is the same as the time index sequence. The index sequence is associated and stored and bound to the corresponding candidate angle state; the bifacial irradiance calculation module calculates the back view factor based on the surface albedo parameter and the component's ground clearance and row spacing. The back view factor is associated with the component size and array row and column number of the array geometry parameters to form a back geometric field of view description; the bifacial irradiance calculation module calculates the occlusion discrimination value based on the solar geometry and row spacing. The occlusion discrimination value is matched with the candidate angle state set time-by-time to generate an occlusion factor. When the solar geometry is marked as invalid or the meteorological irradiance parameter is missing, the occlusion factor is... The sub-item is registered as invalid at this moment and associated with an abnormal state; the dual-sided irradiance calculation module couples the meteorological irradiance parameters with the frontal incident correction coefficient, the back-side viewing angle factor and the shading factor to generate the frontal effective irradiance and the back-side effective irradiance, and outputs the frontal effective irradiance, the back-side effective irradiance and the shading factor to the temperature calculation and electrical output calculation module according to the time index sequence. The temperature calculation and electrical output calculation module calls the frontal effective irradiance, the back-side effective irradiance and the shading factor to complete the calculation of the component operating temperature and candidate attitude power sequence.

[0083] The temperature calculation and electrical output calculation module 04 is used to input the effective front irradiance, the effective back irradiance, the shading factor, and the basic data structure into the temperature calculation and electrical output calculation. It employs a thermal balance model and inputs the effective front irradiance, the effective back irradiance, ambient temperature, and wind speed into the irradiance absorption term, convection heat transfer term, and radiation heat transfer term, outputting the component operating temperature. It uses a single-diode five-parameter model and performs temperature corrections on the photocurrent, reverse saturation current, series resistance, parallel resistance, and ideality factor based on the component operating temperature. Finally, it synthesizes the effective front irradiance and the effective back irradiance according to the back gain coefficient to obtain the equivalent irradiance. The system generates candidate attitude power sequences and energy indices, and outputs these sequences to the fitness construction and feasible domain processing module. Specifically, the temperature calculation and electrical output calculation module receives the effective front irradiance, the effective back irradiance, and the shading factor output by the bifacial irradiance calculation module. It also reads the ambient temperature and wind speed from the meteorological irradiance parameters, as well as the electrical parameters of the bifacial photovoltaic module and the back gain coefficient, from the basic data structure. The module aligns the ambient temperature and wind speed with the effective front irradiance and the effective back irradiance on the time index sequence. The temperature calculation and electrical output calculation module then calls the thermal... The irradiance absorption, convection heat transfer, and radiation heat transfer terms of the equilibrium model are used, and the effective irradiance at the front, the effective irradiance at the back, ambient temperature, and wind speed are integrated into each calculation link to obtain the component operating temperature sequence. When the shading factor is invalid or the irradiance input is missing, the component operating temperature is registered as invalid at the corresponding time and associated with the abnormal state. During the electrical output calculation process, the temperature calculation and electrical output calculation module calls the single diode five-parameter model, and performs temperature correction on the photocurrent, reverse saturation current, series resistance, parallel resistance, and ideality factor based on the component operating temperature, and the effective irradiance at the front and back are calculated according to the specified parameters. The equivalent irradiance is obtained by synthesizing the back gain coefficient. The equivalent irradiance, together with the operating temperature of the component, drives the output power calculation result of the single diode five-parameter model. The temperature calculation and electrical output calculation module generates a candidate attitude power sequence according to the candidate angle state set of the candidate attitude geometry, and performs integration and convergence on the candidate attitude power sequence on the time index sequence to obtain an energy index. The candidate attitude power sequence and the energy index are output to the fitness construction and feasible region processing module. The fitness construction and feasible region processing module calls the energy index and combines it with the actuator constraint parameters to form a fitness evaluator and a feasible region processor.

[0084] The fitness construction and feasible region processing module 05 is used to input the energy index and the actuator constraint parameters into the fitness construction and feasible region processing module, combine the energy index and penalty terms to obtain a fitness value and form a fitness evaluator. The penalty terms include limit violation penalty, angular velocity violation penalty, angular acceleration violation penalty, occlusion duration penalty, and action frequency penalty. For individuals exceeding the tracking angle limit, boundary projection repair is performed to form a feasible region processor. The fitness evaluator and the feasible region processor are output to the genetic algorithm optimization and angle curve generation module. Specifically, the fitness construction and feasible region processing module receives the temperature... The module calculates the energy index and reads the actuator constraint parameters from the basic data structure. These constraint parameters include tracking angle limit, angular velocity limit, angular acceleration limit, control cycle, and docking angle. The fitness construction and feasible region processing module establishes a discrete discrimination method for angle change rate and angular acceleration under the control cycle constraint, maintaining time index consistency with the angle curve generation process of the genetic algorithm optimization and angle curve generation module. During fitness construction, the module combines the energy index with a penalty term to obtain the fitness value. The penalty items consist of limit violation penalty, angular velocity violation penalty, angular acceleration violation penalty, occlusion duration penalty, and action frequency penalty. The limit violation penalty is based on the degree of deviation from the tracking angle limit curve; the angular velocity violation penalty is based on the rate of change of the upper limit of angular velocity on the angle curve; the angular acceleration violation penalty is based on the difference in the rate of change of the upper limit of angular acceleration on the angle curve; the occlusion duration penalty is based on the invalid or occlusion markers of the occlusion factor on the time index sequence; and the action frequency penalty is based on the number of node changes on the angle curve. The fitness construction and feasible region processing module encapsulates the fitness values ​​into a fitness evaluator for genetic computation. The genetic algorithm optimization and angle curve generation module is invoked. During the feasible region processing, the fitness construction and feasible region processing module performs boundary projection repair on individuals that exceed the tracking angle limit. The boundary projection repair maps the out-of-bounds angle to the tracking angle limit boundary and verifies it in conjunction with the angle change rate constraint of adjacent control cycles to form a feasible region processor. The feasible region processor and the fitness evaluator are output to the genetic algorithm optimization and angle curve generation module. The genetic algorithm optimization and angle curve generation module calls the feasible region processor to perform boundary projection repair on the chromosome decoding result and calls the fitness evaluator to complete the fitness value calculation.

[0085] The genetic algorithm optimization and angle curve generation module 06 is used to connect the fitness evaluator and the feasible region processor to the genetic algorithm optimization. It uses real-number encoding to map the optimal strategy parameters to chromosomes and calls the fitness evaluator to obtain fitness values. It executes elite retention operators, tournament selection operators, simulated binary crossover operators, and Gaussian mutation operators to complete iterative updates. The Gaussian mutation operator updates the mutation rate based on the algebraic index. It outputs the optimal strategy parameters and angle curve to the control parameter table writing and target angle instruction generation module. Specifically, the genetic algorithm optimization and angle curve generation module receives the fitness values. The fitness evaluator and feasible region processor output by the feasible region processing module are constructed, and the composition of the optimal strategy parameters is read. The optimal strategy parameters include the axis tilt angle parameter, the backtracking correction coefficient parameter, and the piecewise angle curve node parameter. The genetic algorithm optimization and angle curve generation module uses real number encoding to map the axis tilt angle parameter, backtracking correction coefficient parameter, and piecewise angle curve node parameter to chromosomes. In the initialization phase, the feasible region processor is called to project the angle nodes obtained by decoding the chromosomes onto the tracking angle limit, while simultaneously constraining the rate of change corresponding to the upper limit of angular velocity and upper limit of angular acceleration. Consistency verification is performed; the genetic algorithm optimization and angle curve generation module performs fitness calculation on the population in each generation iteration. During the fitness calculation phase, the fitness evaluator is called and the evaluation data corresponding to the energy index and the angle curve obtained from chromosome decoding are input to obtain the fitness value and establish a correlation with the chromosome; the genetic algorithm optimization and angle curve generation module selects individuals based on fitness values ​​using the elite retention operator, generates the parent set using the tournament selection operator, generates the offspring set using the simulated binary crossover operator, and performs perturbation updates on the real-number encoded chromosomes using the Gaussian mutation operator. The Gaussian mutation operator is based on the generation... The mutation rate is updated using the number index, and the generation index is obtained by combining the current iteration generation and the maximum iteration generation. After the offspring generation is generated, the genetic algorithm optimization and angle curve generation module calls the feasible domain processor again to perform boundary projection repair on the angle curve and perform consistency verification on the docking angle constraint to obtain a set of candidate individuals that meet the actuator constraint parameters. When the iteration termination condition is met, the genetic algorithm optimization and angle curve generation module outputs the optimal strategy parameters corresponding to the fitness value and generates the angle curve. The optimal strategy parameters and the angle curve are output to the control parameter table writing and target angle instruction generation module.

[0086] The control parameter table writing and target angle command generation module 07 is used to parse the angle curve into multi-time period target angle nodes and write them into the control parameter table. The control parameter table stores the multi-time period target angle nodes with time indexes and reads the multi-time period target angle nodes according to the control cycle, generates target angle commands through linear interpolation, and outputs them to the tracking controller. Specifically, the control parameter table writing and target angle command generation module receives the angle curve output by the genetic algorithm optimization and angle curve generation module, and reads the control cycle and the docking angle from the basic data structure. During the writing stage, it performs consistency checks on the continuity of the time index and the order of the nodes of the angle curve. The control parameter table writing and target angle command generation module performs segmented parsing of the angle curve to obtain multi-time period target angle nodes. The multi-time period target angle nodes are bound to the corresponding time indexes and written into the control parameter table. The control parameter table stores the multi-time period target angle nodes with time indexes and writes them into the control parameter table. The target angle nodes for each time period are stored with a time index and the index granularity is consistent with the control cycle. During the instruction generation phase, the control parameter table writing and target angle instruction generation module reads adjacent target angle nodes from the control parameter table according to the control cycle, generates target angle instructions by linear interpolation, and outputs them to the tracking controller. When a time index is missing, a node is abnormal, or the reading exceeds the storage range, the control parameter table writing and target angle instruction generation module writes the docking angle into the target angle instruction and outputs it to the tracking controller, while registering the association record between the corresponding abnormal state and the time index. The tracking controller receives the target angle instruction and publishes the measured power, measured irradiance, measured temperature, and measured angle to the operation data structure and parameter correction update module. The operation data structure and parameter correction update module calls the measured power, measured irradiance, measured temperature, and measured angle to construct the operation data structure and trigger parameter correction.

[0087] The running data structure and parameter correction update module 08 is used to acquire the measured power, measured irradiance, measured temperature and measured angle collected by the tracking controller to construct the running data structure and complete the recording. The running data structure is then connected to the parameter correction update of the surface albedo parameters and the back gain coefficient to update the basic data structure. The updated basic data structure is then output to the solar position and attitude geometry calculation module. Specifically, the operational data structure and parameter correction update module receives the measured power, measured irradiance, measured temperature, and measured angle published by the tracking controller, and aligns them with the control cycle to generate an operational time index sequence. During the operational data structure construction phase, it performs time index integrity checks, value range checks, and abnormal jump checks on the measured power, measured irradiance, measured temperature, and measured angle. Data that passes the checks is bound and encapsulated into an operational data structure with the operational time index sequence and recorded. During the parameter correction phase, the operational data structure and parameter correction update module connects the operational data structure to the parameter correction link. Parameter correction reads the surface albedo parameter and the backside gain coefficient from the basic data structure and establishes a correspondence with the measured irradiance, measured power, and measured angle in the operational data structure to generate the surface albedo parameter. The updated values ​​and the updated values ​​of the backside gain coefficient are written into the basic data structure to form the updated basic data structure. When the running data structure has an invalid moment or an abnormal state, the parameter correction link registers an invalid flag at the corresponding moment and skips the update calculation at that moment, while retaining the association record between the invalid flag and the running time index sequence. The running data structure and parameter correction update module outputs the updated basic data structure to the solar position and attitude geometry calculation module. The solar position and attitude geometry calculation module receives the updated basic data structure and refreshes the calculation input of the solar geometry structure and the candidate attitude geometry structure. The double-sided irradiation calculation module, the temperature calculation and electrical output calculation module, the fitness construction and feasible region processing module, and the genetic algorithm optimization and angle curve generation module call the updated data object according to the link and form a closed-loop update.

Claims

1. A bifacial photovoltaic tracking method based on a genetic algorithm, characterized in that, include: S100: Obtain the site's geographic time parameters, array geometric parameters, surface albedo parameters, meteorological irradiance parameters, bifacial photovoltaic module electrical parameters, and actuator constraint parameters to construct the basic data structure; S200. Input the basic data structure into the solar position and attitude geometry calculation to obtain the solar geometry structure and candidate attitude geometry structure; S300. Input the solar geometry and the candidate attitude geometry into the bi-face irradiance calculation to obtain the effective front irradiance, the effective back irradiance, and the shading factor. S400: Input the effective front irradiance, the effective back irradiance, the shading factor, and the basic data structure into temperature calculation and electrical output calculation to obtain the candidate attitude power sequence and energy index; S500: Input the energy index and the actuator constraint parameters into the fitness construction and feasible region processing to obtain the fitness evaluator and feasible region processor; S600: Input the fitness evaluator and the feasible domain processor into the genetic algorithm for optimization, output the optimal strategy parameters and angle curve, and write the angle curve into the control parameter table to generate the target angle instruction.

2. The method according to claim 1, characterized in that, The array geometry parameters include the azimuth angle of the rotating shaft, the tilt angle of the rotating shaft, the height of the component above the ground, the row spacing, the component size, and the number of array rows and columns; the actuator constraint parameters include the tracking angle limit, the upper limit of angular velocity, the upper limit of angular acceleration, the control cycle, and the docking angle.

3. The method according to claim 1, characterized in that, The bifacial irradiance calculation includes: calculating the incident angle based on the candidate attitude geometry and generating a frontal incident correction coefficient; and calculating the back view factor based on the surface albedo parameter, the component's ground clearance, and the row spacing. The occlusion discrimination value is calculated based on the solar geometry and the row spacing, and the occlusion factor is generated; the effective front irradiance and the effective back irradiance are output based on the front incidence correction coefficient, the back view factor, and the occlusion factor.

4. The method according to claim 1, characterized in that, The temperature calculation inputs the effective front irradiance, the effective back irradiance, the ambient temperature, and the wind speed into the thermal balance model, and outputs the component operating temperature; the thermal balance model includes irradiance absorption terms, convective heat transfer terms, and radiative heat transfer terms.

5. The method according to claim 1, characterized in that, The electrical output calculation adopts a single diode five-parameter model, and the photocurrent, reverse saturation current, series resistance, parallel resistance and ideality factor are corrected for temperature based on the component operating temperature; the effective irradiance on the front and the effective irradiance on the back are combined according to the back gain coefficient to obtain the equivalent irradiance, and the candidate attitude power sequence is calculated based on the equivalent irradiance.

6. The method according to claim 1, characterized in that, The fitness construction combines the energy index with the penalty term to obtain the fitness value; the penalty term includes limit violation penalty, angular velocity violation penalty, angular acceleration violation penalty, occlusion duration penalty and action frequency penalty; the feasible domain processing performs boundary projection repair on individuals that exceed the tracking angle limit.

7. The method according to claim 1, characterized in that, The optimal strategy parameters include the axis tilt angle parameter, the backtracking correction coefficient parameter, and the segmented angle curve node parameter; the genetic algorithm optimization uses real number encoding to map the optimal strategy parameters to chromosomes, and inputs the chromosomes into the fitness evaluator to obtain fitness values.

8. The method according to claim 1, characterized in that, The genetic algorithm optimization includes an elite retention operator, a tournament selection operator, a simulated binary crossover operator, and a Gaussian mutation operator; the Gaussian mutation operator updates the mutation rate based on an algebraic index, which is obtained by combining the current iteration algebra and the maximum iteration algebra.

9. The method according to claim 1, characterized in that, The angle curve is composed of target angle nodes for multiple time periods; the control parameter table stores the target angle nodes for multiple time periods using a time index; the target angle instruction is generated by reading from the control parameter table according to the control cycle and then performing linear interpolation.

10. A bifacial photovoltaic tracking system based on a genetic algorithm, characterized in that, include: The system comprises a basic data structure construction module, a solar position and attitude geometry calculation module, a two-sided irradiation calculation module, a temperature calculation and electrical output calculation module, a fitness construction and feasible region processing module, a genetic algorithm optimization and angle curve generation module, a control parameter table writing and target angle instruction generation module, and a running data structure and parameter correction and update module; these modules are connected in sequence to implement the method as described in any one of claims 1-9.

Images