Method and system for measuring hemispherical sky radiation based on double-axis scanning

Abstract

The application discloses a hemispherical sky radiation measurement method and system based on double-axis scanning, and belongs to the technical field of solar radiation measurement and atmospheric radiation monitoring. In order to solve the problem that the existing equipment is difficult to synchronously obtain the hemispherical directional radiance, the horizontal total irradiance, the direct normal irradiance and the scattered horizontal irradiance under a unified space-time reference, and is affected by installation and shafting deviation, the application realizes the unified space-time reference measurement and reduces the mutual confusion of time variation, pointing error and radiation drift by the following technical effects: unified clock synchronous acquisition, reverse order paired scanning on both sides of the center time, solar centroid geometric calibration, hemispherical cosine solid angle integral and three-component closure, mutual switching in the geometric update, radiation update and all frozen states according to the quality and stability indexes and feedback of the next period.

Description

Technical Field

[0001] This invention relates to the field of solar radiation measurement and atmospheric radiation monitoring technology, and in particular to a method and system for measuring hemispherical sky radiation based on dual-axis scanning. Background Technology

[0002] Hemispherical sky radiance measurements are used to characterize the radiance distribution in different sky directions and provide fundamental data for solar energy resource assessment, atmospheric radiative transfer analysis, and ground-based radiation observation. Existing observations typically use horizontal total radiometers, direct radiometers, and diffuse radiometers to obtain irradiance components, while directional distribution is acquired point-by-point at different azimuth and elevation angles by sky-scanning equipment. Each channel often has different sampling periods, installation references, and calibration chains.

[0003] Because hemispherical scanning requires a certain amount of time, changes in cloud cover can cause measurements in different directions within the same scan cycle to correspond to different atmospheric conditions, making it difficult for the directional probe and the fixed reference channel to naturally align to the same central moment. Azimuth zero bias of the scanning mechanism, base roll and pitch deviations, non-orthogonality of the two axes, and pitch zero bias can further cause nodal orientation errors. Furthermore, if the geometric parameters and radiation gain are updated simultaneously using radiation closure residuals, cloud-induced time variations, pointing deviations, and scale drift can easily be confused, increasing the complexity of on-site calibration and the cost of configuring multiple devices.

[0004] Therefore, there is a need for a biaxial scanning-based hemispherical sky radiation measurement method and system that can overcome the shortcomings of the existing technologies. Summary of the Invention

[0005] One objective of this invention is to propose a hemispherical sky radiation measurement method and system based on dual-axis scanning. Addressing the challenges of existing technologies in simultaneously obtaining hemispherical radiance distribution, total horizontal irradiance, direct normal irradiance, and scattered horizontal irradiance under a unified time and spatial reference, and the potential for confusion due to scanning time variations, installation deviations, and radiation drift, this invention proposes a technical solution: unified clock synchronous acquisition; performing forward and reverse paired scanning of the same hemispherical nodes around the central time while interspersed with solar neighborhood scanning; correcting paired radiance values ​​to the central time; generating candidate values ​​for geometric errors using the solar centroid and ephemeris; integrating the corrected distribution to form a three-component closed residual; and then selecting, based on mutually exclusive quality and stability indicators, either updating geometric parameters, updating radiation parameters, or freezing all parameters, ensuring that the selected candidate values ​​only take effect in the next scanning cycle. This invention achieves the technical effects of improving spatiotemporal consistency, correcting directional positioning errors, and avoiding confusion between different error sources during calibration.

[0006] This invention provides a method for measuring hemispherical sky radiation based on dual-axis scanning, comprising:

[0007] S1. Set up a horizontal total radiation reference channel on a fixed base, and set up a directional radiation probe in the dual-axis scanning mechanism that can switch between sky radiance range and direct solar attenuation range, and collect data from both and the dual-axis encoder synchronously with a unified clock.

[0008] S2. Scan the same hemispherical nodes in reverse order around the central time, and scan the solar neighborhood between the two scans to form paired direction measurement groups and a central solar measurement group with each node located on both sides of the central time.

[0009] S3. Correct the radiance of the node direction to the center time according to the time distance of each measurement in the paired direction measurement group from the center time. Determine the candidate value of the geometric error parameter according to the central solar measurement group and the calculated solar vector. Correct the node direction with the currently effective geometric parameter to form the hemispherical radiance distribution at the center time.

[0010] S4. Exclude the solar regions in the distribution, obtain the scattered horizontal irradiance by integrating the cosine of the nodal solid angle and the zenith angle, and combine it with the total horizontal irradiance and the direct normal irradiance in the central solar measurement group to form a three-component closed residual.

[0011] S5. Based on solar measurement quality, node scanning stability, and the closed residual, mutually exclusive execution of geometric parameter update, radiation parameter update, or all parameter freeze: update the geometric parameters of the next cycle with the candidate values ​​and freeze the radiation parameters, or update the radiation parameters of the next cycle with the closed residual and freeze the geometric parameters, or keep the currently effective parameters for the next cycle to call.

[0012] Optionally, S1 includes: fixing a horizontal total radiation meter to the fixed base to form the horizontal total radiation reference channel, setting an azimuth axis and a pitch axis with a dual-axis encoder on the dual-axis scanning mechanism, setting the directional radiation probe at the scanning end of the pitch axis, and setting a temperature sensor at the directional radiation probe.

[0013] The unified clock provides synchronous trigger signals to each acquisition channel and the dual-axis drive controller, and writes the trigger time, reference irradiance, direction probe response value, range status, azimuth code value, elevation code value and probe temperature value into the same acquisition record.

[0014] The sky radiance range uses a non-attenuated optical path, while the direct solar attenuation range uses an attenuated optical path with a known transmittance. The direction probe response value is scaled based on the range state and the transmittance.

[0015] Optionally, S2 includes: constructing a hemispherical node table according to the pre-stored node numbers and node target directions, using the ascending order of node numbers as the scanning order before the center time, using the descending order of node numbers as the scanning order after the center time, and associating two measurements with the same node number before and after the center time as the paired direction measurement group.

[0016] The solar neighborhood consists of target points whose angular distance from the direction of the calendar sun is not greater than the radius of the solar neighborhood in the scanning configuration. The solar centroid is obtained by calculating the centroid of the direct solar attenuation range of the target points, and the time center of the solar neighborhood scan is used as the center time.

[0017] The horizontal total irradiance is obtained based on the synchronous measurement value of the horizontal total radiation reference channel at the central time. The direct normal irradiance is obtained based on the measurement value of the solar direct attenuation range in the direction of the solar barycenter and the known transmittance corresponding to the range. The solar barycenter, horizontal total irradiance, direct normal irradiance and their corresponding time stamps and dual-axis coded values ​​are associated with the central solar measurement group.

[0018] Optionally, S3 includes: multiplying the directional radiance of the same node before and after the center time by a weight proportional to the time distance on the other side, summing the results, and normalizing them with the sum of the time distances on both sides to obtain the interpolated radiance of the node at the center time;

[0019] According to the radiance normalization scale and curvature normalization scale in the scanning configuration, the absolute value of the bidirectional measurement difference at the same node and the absolute value of the time curvature of the horizontal total radiation reference channel in the corresponding time period are normalized and added together to obtain a dimensionless nodal time nonlinearity index.

[0020] When the node time nonlinearity index is greater than the dimensionless nonlinearity threshold in the scanning configuration, a supplementary scan table is established. The node and nodes whose target direction angular distance is not greater than the node neighborhood radius in the scanning configuration are written into the supplementary scan table. Before the supplementary scan passes the same criterion, the effective integration weight of the node is reset to zero. When the supplementary scan passes, the supplementary scan measurement value is re-interpolated and the node solid angle coefficient is restored as the effective integration weight. When the number of consecutive supplementary scan failures reaches the failure number threshold in the scanning configuration, the effective integration weight is kept at zero and the confidence level of the scanning cycle is set to the downgrade level.

[0021] When the node time nonlinearity index is not greater than the dimensionless nonlinearity threshold, the effective integral weight is obtained by multiplying the node solid angle coefficient by a coefficient obtained by subtracting the ratio of the node time nonlinearity index to the dimensionless nonlinearity threshold.

[0022] Furthermore, the solar centroid, corresponding time markers, and dual-axis encoded values ​​are extracted from the central solar measurement group of multiple scanning cycles, and the solar vectors for each ephemeris are calculated based on the corresponding time markers;

[0023] The angular residual between the solar centroid measurement vector after correction by geometric error parameters and the calculated solar vector is used as the target quantity. The identifiability condition is that the rank of the Jacobian matrix of the angular residual with respect to the five types of geometric error parameters is five and the condition number is not greater than the identifiability threshold in the calibration configuration. Candidate values ​​of geometric error parameters are obtained. The five types of geometric error parameters include the azimuth zero bias of the scanning mechanism, the roll angle of the fixed base, the pitch angle of the fixed base, the non-orthogonality angle between the azimuth axis and the pitch axis, and the pitch zero bias.

[0024] Substitute the currently effective geometric parameters into the coordinate transformation from the dual-axis encoded values ​​to the spatial direction to obtain the corrected hemispherical node direction.

[0025] Optionally, S4 includes: the solar region includes an angular domain centered on the calendar solar direction, with the sum of the solar disk angular radius and the circumstellar angular width in the scanning configuration as the exclusion radius;

[0026] For each hemispherical node outside the solar region, the product of the node interpolated radiance, the effective integral weight, and the cosine of the node zenith angle is summed to obtain the scattered horizontal irradiance.

[0027] The product of the direct normal irradiance and the cosine of the solar zenith angle is added to the scattered horizontal irradiance, and the summation is subtracted from the total horizontal irradiance to obtain the three-component closed residual.

[0028] Optionally, S5 includes: the solar measurement quality includes the solar barycenter signal-to-noise ratio and the solar position span, wherein the solar barycenter signal-to-noise ratio is determined by the ratio of the peak value of the barycenter in the solar neighborhood to the standard deviation of the background noise, and the solar position span is determined by the upper bound of the angle between the epoch-calculated solar vectors involved in the solution;

[0029] Set a full parameter freeze trigger flag and a radiation parameter update trigger flag. Select the state in the order of the full parameter freeze trigger flag, the radiation parameter update trigger flag, and the solar measurement quality. Select the full parameter freeze state when the full parameter freeze trigger flag is true. Select the radiation parameter update state when the full parameter freeze trigger flag is false and the radiation parameter update trigger flag is true. Select the geometric parameter update state when both trigger flags are false, the solar centroid signal-to-noise ratio is not less than the signal-to-noise ratio threshold in the calibration configuration, and the solar position span is not less than the position span threshold in the calibration configuration. Select the full parameter freeze state in all other cases.

[0030] In the updated geometric parameters state, the radiation gain and nodal solid angle coefficients are frozen, and the candidate values ​​of the geometric error parameters are updated to the geometric parameters that will take effect in the next scan cycle.

[0031] Furthermore, the node scanning stability is characterized by the change in geometric parameters, the bidirectional measurement difference statistic of the same node, the time curvature of the total horizontal irradiance, and the change in probe temperature.

[0032] A temperature sensor is installed at the directional radiation probe, and the temperature change of the probe is determined based on the difference between the synchronous measurement values ​​of the temperature sensor in adjacent scanning cycles.

[0033] When the absolute value of the change of each geometric parameter within the continuous scanning period window is not greater than the convergence threshold in the calibration configuration, the bidirectional measurement difference statistic is not greater than the bidirectional difference threshold, the absolute value of the horizontal total irradiance time curvature is not greater than the stable curvature threshold, and the probe temperature change is not greater than the temperature change threshold, the radiation parameter update trigger flag is set to true.

[0034] In the radiation parameter update state, the currently effective geometric parameters are frozen, and a unified radiation gain candidate value is determined based on the three-component closed residual of multiple scan cycles, or a grouped radiation gain candidate value and a node solid angle coefficient candidate value are determined based on the preset zenith angle ring group to which the hemispherical node belongs. The lower limit and upper limit of the parameters in the calibration configuration are used as safety constraint limit candidate values, and the candidate values ​​are updated to the radiation parameters effective in the next scan cycle.

[0035] Furthermore, an anomaly recovery branch is set up, comparing the absolute value of the bidirectional measurement difference of the same node with the node difference threshold in the scanning configuration, and comparing the absolute value of the horizontal total irradiance time curvature with the stable curvature threshold and cloud change curvature threshold in the calibration configuration, wherein the cloud change curvature threshold is greater than the stable curvature threshold.

[0036] When the absolute value of the bidirectional measurement difference is greater than the node difference threshold and the absolute value of the horizontal total irradiance time curvature is not greater than the stable curvature threshold, the node and its neighboring nodes are written into the supplementary scan table.

[0037] When the absolute value of the total horizontal irradiance time curvature is greater than the cloud curvature threshold, the all parameter freeze trigger flag, which has priority over other states, is set to true, the currently effective parameters are maintained, and the confidence level is set to the downgrade level in the calibration configuration. The same process is performed when the all parameter freeze state is selected in other cases in step S5.

[0038] The downgrade level ensures that the current effective parameters are maintained in the next scan cycle and that the supplementary scan table and solar neighborhood scan are executed first. If the freeze trigger flag for all parameters is not triggered within the continuous recovery window, at least one parameter update status is completed, and the absolute value of the three-component closed residual is not greater than the recovery threshold, the confidence level is restored to the normal level in the calibration configuration.

[0039] A freeze flag representing the selected state is generated, and the freeze flag, the geometric and radiometric parameters that will take effect in the next scan cycle, the supplementary scan table, and the confidence level are associated and stored for use in the next scan cycle.

[0040] On the other hand, the present invention also provides a hemispherical sky radiation measurement system based on dual-axis scanning, comprising:

[0041] The synchronous acquisition module is used to synchronously generate reference radiation data, directional radiation data, and axis angle data from a horizontal total radiation reference channel set on a fixed base, a directional radiation probe with switchable range set on a dual-axis scanning mechanism, and a dual-axis encoder. The paired scanning module is used to scan the same hemispherical nodes and the solar neighborhood in reverse order around the central time to form paired directional measurement groups and a central solar measurement group. The spatiotemporal correction module is used to correct the directional radiance of each hemispherical node to the central time, determine the candidate values ​​of geometric error parameters, and call the currently effective geometric parameters to correct the orientation of the hemispherical node. The radiation closure module is used to integrate the hemispherical nodes after excluding the solar region to obtain the scattered horizontal irradiance and form a three-component closure residual. The calibration control and output module is used to select a state from mutually exclusive parameter update states, generate a freeze flag, the next cycle effective parameter vector, a supplementary scan table, and a confidence level for use in the next scan cycle, and output the measurement results and calibration results with a unified spatiotemporal reference.

[0042] The beneficial effects of this invention are:

[0043] 1. By synchronously acquiring data from a fixed reference channel, dual-range direction probe, and dual-axis encoder using a unified clock, and performing forward and reverse pair scans on the same hemispherical nodes around the center time, the radiance of each direction can be corrected to the same center time, reducing the impact of scanning timing differences on the hemispherical distribution.

[0044] 2. By solving the angular residuals of multiple solar centroid measurement vectors and calendrical solar vectors, and distinguishing between candidate values ​​of geometric errors, currently effective parameters, and parameters effective in the next cycle, five types of geometric errors of the scanning mechanism and the base can be corrected, so that the node direction, hemispherical integral, and closure residual use the same parameter version.

[0045] 3. By integrating nodes excluding the solar disk and the circumpolar region using effective integration weights, and forming closed residuals with total horizontal irradiance, direct normal irradiance, and diffuse horizontal irradiance, and then performing geometric updates, radiative updates, or complete freezes according to priority and mutual exclusion, the erroneous attribution between cloud-induced changes, geometric pointing errors, and radiative gain drift can be reduced. Furthermore, a single switchable range directional probe can be used to cover both sky and solar measurements. Attached Figure Description

[0046] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0047] Figure 1 This is a flowchart of the hemispherical sky radiation measurement method based on dual-axis scanning according to the present invention.

[0048] Figure 2 This is a flowchart of the S5 mutual exclusion parameter update state selection process of the present invention. Detailed Implementation

[0049] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0050] refer to Figures 1-2 A method for measuring hemispherical sky radiation based on dual-axis scanning, comprising:

[0051] S1. Set up a horizontal total radiation reference channel on a fixed base, and set up a directional radiation probe in the dual-axis scanning mechanism that can switch between sky radiance range and direct solar attenuation range, and collect data from both and the dual-axis encoder synchronously with a unified clock.

[0052] S2. Scan the same hemispherical nodes in reverse order around the central time, and scan the solar neighborhood between the two scans to form paired direction measurement groups and a central solar measurement group with each node located on both sides of the central time.

[0053] S3. Correct the radiance of the node direction to the center time according to the time distance of each measurement in the paired direction measurement group from the center time. Determine the candidate value of the geometric error parameter according to the central solar measurement group and the calculated solar vector. Correct the node direction with the currently effective geometric parameter to form the hemispherical radiance distribution at the center time.

[0054] S4. Exclude the solar regions in the distribution, obtain the scattered horizontal irradiance by integrating the cosine of the nodal solid angle and the zenith angle, and combine it with the total horizontal irradiance and the direct normal irradiance in the central solar measurement group to form a three-component closed residual.

[0055] S5. Based on solar measurement quality, node scanning stability, and the closed residual, mutually exclusive execution of geometric parameter update, radiation parameter update, or all parameter freeze: update the geometric parameters of the next cycle with the candidate values ​​and freeze the radiation parameters, or update the radiation parameters of the next cycle with the closed residual and freeze the geometric parameters, or keep the currently effective parameters for the next cycle to call.

[0056] In this specific embodiment, S1 includes:

[0057] In this specific embodiment, the measuring device is installed on a horizontal concrete base in an unobstructed observation field. The horizontal total radiation meter is fixed near the center of the base by a three-point leveling seat and forms a horizontal total radiation reference channel. The azimuth axis of the dual-axis scanning mechanism is set vertically, and the pitch axis intersects the azimuth axis. The directional radiation probe is installed at the scanning end of the pitch axis and its optical axis passes through the nominal intersection of the two axes. The half angle of the probe's field of view is set to 2.5°. The temperature sensor inside the probe housing is set close to the photoelectric conversion circuit. The azimuth axis and the pitch axis are respectively equipped with absolute encoders, and the encoding resolution is no greater than 0.01°.

[0058] The unified clock uses a satellite timing module with second pulse output. The acquisition controller uses second pulse to correct the local microsecond counter and synchronously triggers the horizontal total radiation reference channel, directional radiation probe, temperature channel and two encoders at 20 Hz. The same acquisition record is latched on each trigger edge. The acquisition record includes at least the scan cycle identifier, node number, trigger time, original reference irradiance value, directional probe response value, range status, azimuth code value, elevation code value, probe temperature value, drive position flag and calibration configuration version. If the latching time of any channel exceeds 2 ms relative to the unified trigger edge, the record is marked as out of sync and will not be included in the subsequent node pairwise operation.

[0059] The directional radiation probe switches between a non-attenuated optical path and a direct solar attenuated optical path via an electric switching plate. The non-attenuated optical path is used for sky radiance measurement, while the direct solar attenuated optical path is equipped with a traceable calibrated neutral attenuator. In this specific embodiment, the transmittance calibration value of the attenuator in the probe's operating band is 0.0010 and the relative expansion uncertainty is no greater than 2%. After switching, wait 150 ms and continuously acquire three synchronous records. The last record is only accepted when the range status of the three records is consistent and the probe response is not saturated.

[0060] The data acquisition controller converts the probe response of the sky range into directional radiance, and the conversion relationship is as follows: ,in Indicates the first The directional radiance of each record is expressed in W·m⁻²·sr⁻¹. Indicates the sequence number of the collected record. Indicates the probe at temperature The sky range response coefficient is expressed in W·m⁻²·sr⁻¹·V⁻¹. Indicates the first The probe temperature recorded is in °C. Indicates the first The directional probe response voltage is recorded in V. The dark voltage of the sky range is expressed in volts (V); the conversion relationship for the scale of the direct solar attenuation range is as follows: ,in Indicates the first Each solar direction record corresponds to the direct normal irradiance, measured in W·m⁻². Indicates temperature The solar range response coefficient is expressed in W·m⁻²·V⁻¹. This represents the dark voltage of the solar range, measured in volts (V). The known transmittance of the attenuating optical path is a dimensionless positive number, and is taken as 0.0010 in this specific embodiment.

[0061] The sky range response coefficient and the solar range response coefficient are respectively calculated as follows: Temperature compensation is performed, among which Indicates the range type index and retrieves or , Indicates range At temperature The response coefficient under the following conditions Indicates range The response coefficient at the reference temperature has the same units as the response coefficient for the corresponding range. Indicates range The temperature coefficient, in °C⁻¹, The reference temperature is 25°C in this specific embodiment. The dark voltage, reference response coefficient and temperature coefficient of each range are calibrated by the shading dark field and the standard radiation source in the range of 5°C to 45°C and then written into the probe calibration table. The table entries use the probe serial number and range status as the joint search key and carry the calibration date, effective temperature range and version number.

[0062] The horizontal total radiation reference channel converts the original voltage into reference irradiance using the manufacturer's sensitivity and field comparison gain, with the unit uniformly set to W·m⁻². The controller writes the corresponding quality bit and prohibits the records that exceed the calibrated temperature range, are negative over the range, are saturated in analog-to-digital conversion, are not in place, or have encoder jumps exceeding 0.2° from participating in subsequent calculations. Records that meet the conditions are written to the periodic acquisition buffer in increments according to the trigger time.

[0063] A temperature sensor is installed at the directional radiation probe. The temperature sensor range is set to -20℃ to 60℃ and the resolution is no greater than 0.1℃. Its digital output is bound to the directional probe response using the same trigger number. A calibration point is set every 5℃ in the temperature range in the probe calibration table. When the difference between the compensation results of adjacent ranges exceeds 0.5% of the full scale, the temperature range is marked as needing recalibration and is prohibited from being used for parameter updates.

[0064] The sky radiance range uses a non-attenuated optical path, while the direct solar attenuation range uses an attenuated optical path with a known transmittance. The range switching actuator only operates after the dual-axis drive is in place. The switching status is read back by the Hall position switch and written into the acquisition record. When the command status is inconsistent with the read-back status, the current direction probe response is discarded and the switching fault code is recorded.

[0065] The field comparison gain of the horizontal total radiation reference channel is determined by the least squares slope passing through the origin after continuous synchronous observation of the same level standard total radiation table under clear sky and stable conditions for 24 hours. The gain record includes the reference table certificate number, slope, fitting residual, validity period and version number. When the fitting residual exceeds 2% or the certificate validity period expires, the channel will no longer be used for closed calibration, but the original observation output will continue to be retained.

[0066] When the device is started for the first time or the parameter snapshot loading fails, the acquisition controller first loads five types of geometric error parameters, uniform radiation gain or grouped radiation gain and the corresponding version number from the factory calibration record or field calibration record that is valid and has not expired. If no valid record is found, the five types of geometric error parameters are set to zero, the uniform radiation gain and the three grouped radiation gains are all set to 1.000, an initial snapshot with version number 0 and a check code is generated, and the first scan cycle is limited to measurement and rescanning without parameter updates, until the candidate availability conditions and stability conditions specified in S3 and S5 are met simultaneously.

[0067] At the end of each scan cycle, the acquisition controller verifies the continuity of trigger sequence numbers, the integrity of range switching, the monotonicity of encoders, and the validity of temperature. The cycle acquisition buffer that has passed the verification, along with the probe calibration table version, attenuator transmittance version, reference channel gain version, unified clock lock status, and quality bit used in this cycle, is transmitted to S2.

[0068] In this specific embodiment, S2 includes:

[0069] The paired scanning module reads the periodic acquisition buffer, scanning configuration, and current effective radiation parameter snapshot formed in S1. The hemispherical node table in the scanning configuration uses the node number as the primary key and records the node target azimuth, target zenith angle, reference node solid angle coefficient, current effective node solid angle coefficient, allowable positioning error, node absolute difference threshold, and range status. The reference node solid angle coefficient is pre-calculated by the difference between the azimuth width of the corresponding spherical grid boundary and the cosine difference of the zenith angle boundary. In this specific embodiment, the zenith angle is divided into 5° rings from 0° to 85°, and the nodes are arranged at azimuth intervals of no more than 10° in each ring. The zenith is numbered separately, and the sum of all reference node coefficients is equal to 2π sr. The current effective node solid angle coefficient is equal to the reference value in the initial state. It is replaced in the next scanning cycle only after the solid angle coefficient candidate mode is selected in S5 and the snapshot is confirmed, and its sum is still equal to 2π sr. The node absolute difference threshold represents the rescan threshold of the absolute difference of radiance on both sides before and after each node, and the unit is W·m⁻²·sr⁻¹. In this specific embodiment, it is taken as 8. W·m⁻²·sr⁻¹, whose value is determined by the 95th percentile of the absolute difference of each node for no less than 30 clear sky stable cycles and is bound to the scan configuration version; for the sky range directional radiance obtained by S1, the paired scan modules select the unified radiation gain or the grouped radiation gain of the zenith angle ring to which the node belongs according to the current snapshot mode before writing it into the node record and perform only one multiplicative correction. The corrected directional radiance and the actual radiation parameter version are written into the node record together, and the gain is not applied again in subsequent S3 and S4;

[0070] The dual-axis drive controller first executes the first hemispherical scan in ascending order of node number. After reaching each node, the encoder angular velocity is less than 0.02°·s⁻¹ for three consecutive trigger cycles and the difference between the actual axis angle and the target axis angle is not greater than 0.03° as the arrival condition. After the arrival condition is met, the sky radiance range record is collected. After the first segment is completed, the direct solar attenuation range is immediately switched and the solar neighborhood scan is executed. Then the sky radiance range is restored and the second hemispherical scan is executed in descending order of node number, so that the two effective measurements of the same node fall on the front and back sides of the solar neighborhood scan time center respectively.

[0071] For node number The paired directional measurement groups are written with front and rear records and pairing status. The front and rear records carry the trigger time, converted directional radiance, azimuth code value, elevation code value, probe temperature value, and mass position, respectively. This indicates the hemispherical node number; if one side's record is missing, out of sync, out of range, or not in place, the pairing status of that node is set to pending scan, the record of the other side is retained but the center time radiance is not directly generated, and the node number, the missing side and the reason for failure are written into the pending scan queue.

[0072] The solar neighborhood generates a target point centered on the calendar solar unit direction vector, and the angular distance between the target direction and the calendar solar direction is calculated as follows: Calculation, where Indicates the first The angular distance between the direction of the target in the solar neighborhood and the direction of the ephemeris is expressed in rad. This represents the index of the target point in the solar neighborhood. Indicates the first The unit direction vector of each target point in the local northeast-northeast coordinate system is a dimensionless three-dimensional column vector. This represents the ephemeris direction vector calculated from the latitude, longitude, altitude, and target time of the observation point. It is a dimensionless three-dimensional column vector with superscript. This indicates vector transpose; in this specific embodiment, the solar neighborhood radius is set to 3°, and azimuth and elevation offset grids are generated at 0.25° intervals, retaining only target points with an angular distance of no more than 3°;

[0073] The solar barycenter is spherically weighted using the attenuated range response after background subtraction, and the calculation relationship is as follows: as well as ,in This represents the unit direction vector of the Sun's center of mass, and is a dimensionless three-dimensional column vector. This represents the total number of valid target points in the solar neighborhood, and is a positive integer. Indicates the first The non-negative centroid weights of each target point are expressed in W·m⁻². Indicates the first The direct normal irradiance converted for each target point, in W·m⁻². This represents the median background value of the direct normal irradiance of samples at the outer edge of the solar neighborhood, expressed in W·m⁻². The vector L2 norm is used; when there are fewer than 9 valid target points or the denominator is not greater than 1 W·m⁻², the mass position of the solar barycenter is invalid and the ephemeris solar direction is retained as the scanning guide direction instead of being used as a geometric update observation.

[0074] For the The sub-solar neighborhood scan module also reads the azimuth encoding angles actually latched at each valid target point. and pitch coding angle Both units are in rad, and each pair of encoded angles is converted into a dimensionless three-dimensional column unit direction vector according to the same axis-angle convention as S3; the module subtracts weights from the background of the target point. Multiplying by the corresponding unit direction vector, summing over all valid target points, and then normalizing, yields the dimensionless centroid-encoded unit direction vector, whose first, second, and third components are denoted as... , and Equivalent centroid azimuth coding angle Take the zero point according to the same orientation. and The two-parameter arctangent value, equivalent centroid pitch coding angle Pick The arcsine value, and All units are rad; when there are fewer than 9 valid coded records, any participating record is of invalid quality, or the sum of weights is not greater than 1 W·m⁻² or the weighted vector norm is not greater than 1×10⁻ 6 When this happens, the equivalent centroid coding angle is not generated and the scan is prohibited from entering the geometric solution.

[0075] The center time of the solar neighborhood scan is based on Confirmed, among which Indicates the center time, in seconds. This indicates the trigger time of the first valid record in the solar neighborhood, in seconds. The trigger time of the last valid record in the solar neighborhood is indicated in seconds. The total horizontal irradiance of the reference channel at the center time is obtained by linear interpolation of two adjacent synchronous reference records at the center time. The direct normal irradiance is obtained by weighted average of the valid values ​​of the nearest target point of the solar barycenter and its eight neighborhoods. The weight decreases as the angular distance from the solar barycenter increases. If the angular distance from the nearest target point exceeds 0.4°, the solar measurement mass position is downgraded.

[0076] The matching threshold of the paired direction measurement group is limited by the unified clock trigger interval and the upper limit of the scanning cycle. In this specific implementation, the node numbers of the front and rear records are exactly the same, the range status is the sky radiance range, the two arrival errors are not greater than 0.03°, and the two measurements are located on both sides around the center time. If any matching condition is not met, a valid pairing is not generated, but the two records and the failure field are written to the pairing exception table.

[0077] The solar barycenter mass criterion simultaneously checks the number of valid target points, the sum of barycenter weights, the angular distance to the nearest target point, and the saturation level of the attenuation range. If there are no fewer than 9 valid target points, the sum of barycenter weights is greater than 1 W·m⁻², the angular distance to the nearest target point is no greater than 0.4°, and there is no saturation, it is marked as valid. If it fails, the center time and reference channel value are retained, but the solar barycenter and direct normal irradiance are marked as invalid for S5 to prioritize and freeze parameters.

[0078] The paired scanning module associates the solar centroid unit direction vector, equivalent centroid azimuth coding angle, equivalent centroid elevation coding angle, actual dual-axis coding values ​​and weights of each target point used to construct the equivalent coding angle, center time, total horizontal irradiance, direct normal irradiance, solar neighborhood peak, background noise statistics, range status, and mass potential into a central solar measurement group. It also writes the paired direction measurement groups of all nodes, the queue of nodes to be scanned, the paired anomaly table, the hemispherical node table version, the ephemeris solar direction, and the scan configuration version into the scan data package for this cycle, so that S3 can perform time correction and geometric calibration.

[0079] In this specific embodiment, S3 includes:

[0080] The spatiotemporal correction module reads the current-cycle scan data packet generated by S2 and performs linear correction on nodes with valid records on both sides according to their time distance from the center time. Center time interpolation radiance according to Calculation, where Represents a node The interpolated radiance at the center moment, in units of W·m⁻²·sr⁻¹. This represents the time distance between the center time and the preceding measurement time, expressed in seconds. This represents the time distance between the later measurement time and the center time, in seconds. Represents a node The radiance in the frontal direction, expressed in W·m⁻²·sr⁻¹. Represents a node The radiance in the rear direction, in units of W·m⁻²·sr⁻¹, is used to switch the node to the waiting-to-be-re-scan state when the sum of the time distances on both sides is less than 0.1 s or greater than 180 s as set in this specific embodiment;

[0081] The module uses the bidirectional measurement difference at the same node and the time curvature of a fixed reference channel to jointly determine the effectiveness of linear interpolation, and the node time nonlinearity index is based on... Calculation, where Represents a node It is a dimensionless time nonlinearity index, and the larger the value, the stronger the change during the scanning period. This represents the normalized scale of radiance, with units of W·m⁻²·sr⁻¹. This represents the time curvature of total horizontal irradiance, expressed in W·m⁻²·s⁻². The curvature normalization scale is expressed in W·m⁻²·s⁻²; the time curvature of the total horizontal irradiance is expressed as follows: Calculation, where Indicates time The synchronous horizontal total irradiance, in W·m⁻². Represents the time in a unified clock, with the unit being seconds (s). The time interval for curvature calculation is 10 s in this specific embodiment;

[0082] The radiance normalization scale is first taken as the median of the interpolated radiance of all effective nodes in the current period. When this median is less than 5 W·m⁻²·sr⁻¹, the radiance normalization scale is set to 5 W·m⁻²·sr⁻¹. The curvature normalization scale is calibrated to 1 W·m⁻²·s⁻² using stable clear sky data. The dimensionless nonlinear threshold in the scan configuration is taken as 0.25 in this specific embodiment. When the time nonlinearity index is not greater than this threshold, the effective integration weight is adjusted according to... Confirmed, among which Represents a node The effective integration weight, in units of sr. Represents the nodes in the hemispherical node table The solid angle coefficient of the currently effective node within this snapshot period, in units of sr, and carrying the solid angle coefficient version. Candidate values ​​may not replace this value midway through this period. This represents the dimensionless nonlinear threshold in the scanning configuration, and in this specific embodiment, it is set to 0.25; when the time nonlinearity index is greater than this threshold, the effective integration weight is reset to zero;

[0083] For nodes exceeding the nonlinear threshold and nodes marked as needing rescanning by S2, the module establishes a rescanning table. The rescanning table uses the scan cycle identifier and node number as a joint primary key, recording the target direction, failure reason, neighborhood radius, priority, number of consecutive failures, and current effective integral weight. Neighboring nodes whose angular distance from the target node is no greater than 2° are also written in. The next scan cycle prioritizes rescanning. When the rescanning record is used with the same pairwise interpolation and nonlinearity criterion, the interpolated radiance at the center time is recalculated using the rescanning value, and the corresponding solid angle coefficient is restored. If three consecutive rescanning failures occur, the weight is kept at zero and the cycle confidence is set to a downgraded level.

[0084] The geometric calibration module extracts the solar barycenter, central time, equivalent barycenter azimuth code angle, and equivalent barycenter elevation code angle from at least 12 recent central solar measurement groups with a time span of at least 30 minutes. It then calculates the corresponding calendrical solar unit direction vector using the observation point's latitude and longitude, altitude, unified clock time, and astronomical calendrical calculation program. If any central solar measurement group lacks an equivalent barycenter code angle, the records or weights used to construct the code angle cannot be verified, or the solar barycenter mass is invalid, it will not be included in the geometric solution window.

[0085] For the For each solar observation, the encoder direction is generated according to the following formula:

[0086]

[0087] Press again Transform to the local northeast-sky coordinate system, where This represents the solar observation index used in the geometric solution. Represents five types of geometric error parameter vectors, S2 represents the first step constructed according to the target point weights. The equivalent centroid azimuth coding angle for a solar observation, in rad. S2 represents the first S2 constructed with equal weights. The equivalent centroid elevation angle of a solar observation, in rad. Indicates zero azimuth offset, in rad. This represents the angle between two non-orthogonal axes, expressed in rad. This indicates zero pitch deflection, in rad. This represents a dimensionless three-dimensional column unit direction vector after taking the equivalent centroid encoding angle as input and correcting for axis system errors. Indicates the base roll angle The resulting dimensionless 3×3 rotation matrix The unit is rad. Indicates the pitch angle of the base. The resulting dimensionless 3×3 rotation matrix The unit is rad. This represents a dimensionless three-dimensional column unit direction vector that corresponds one-to-one with the S2 solar barycenter measurement and is corrected for five types of errors.

[0088] No. The angular residual of a solar observation is calculated according to Calculation, where Represents angular residuals, in rad. Indicates the first The epochal vector of the solar unit direction at each central moment is a dimensionless three-dimensional column vector; the candidate values ​​of the geometric error parameters are calculated according to... Solve for, where This represents the five candidate vectors of geometric error parameters output by the solution. express The five types of geometric error parameter trial vectors in the subscript are the optimization variables; they have different identities. This indicates the number of valid solar observations involved in the solution, and is not less than 12; the... The solar barycenter signal-to-noise ratio of each effective solar observation is calculated as follows: Calculation, where This represents the dimensionless solar center-of-mass signal-to-noise ratio. Indicates the first in the S2 central solar measurement group The peak value of the centroid after subtracting the background radiation from the solar neighborhood is expressed in W·m⁻². The standard deviation of background noise for samples at the outer edge of the same solar neighborhood is expressed in W·m⁻²; weights are calculated as follows: Normalization, in which Indicates the first The dimensionless nonnegative weight of an effective solar observation. This represents the summation index of valid solar observations participating in the same geometric solution window, with the sum of all weights being 1. When there are no valid solar observations or the sum of signal-to-noise ratios is not greater than zero, the geometric candidate state is set to unavailable and no iteration is performed. The solution uses Gauss-Newton iteration with upper and lower bounds for the parameters, with the upper limit of the absolute value of each of the five angle parameters set to 5°, and the iteration increment norm being less than 1 × 10⁻. 6 Stop when rad or iterations reach 30;

[0089] The module performs an identifiability check on the Jacobian matrix of the diagonal residuals with respect to the five types of geometric error parameters. Candidate values ​​are accepted only when the rank of the Jacobian matrix is ​​5 and the condition number is not greater than the identifiability threshold in the calibration configuration. In this specific embodiment, the identifiability threshold is 1×10^4. If the identifiability condition is not met, there are fewer than 12 solar observations, the solar barycenter mass is invalid, or the solution has not converged, the candidate state is set to unavailable and the currently effective geometric parameters are not rewritten.

[0090] In this period, the orientation of all hemispherical nodes is corrected using the currently effective candidate geometric parameters instead of the newly obtained ones. The dual-axis encoded values ​​of each node are substituted into the same coordinate transformation as the solar observation to obtain the corrected node unit orientation vector and its zenith angle. The candidate geometric parameters are only written into the candidate parameter area and carry the solution window, Jacobian matrix rank, condition number, root mean square of angular residual and version number, waiting for S5 to decide whether to take effect in the next scan cycle.

[0091] The supplementary scan scheduler generates priorities based on the failure reason and the node zenith angle. Nodes lacking records on one side have higher priority than nonlinear over-limit nodes. Nodes with the same priority are sorted in ascending order by the estimated rotation time from the node's target direction to the current axis angle. If the supplementary scan table is not empty before the solar neighborhood scan begins, the supplementary scan queue is executed once before entering the solar neighborhood scan. If supplementary scanning causes the center time to deviate from the planned value by more than 30 seconds, the unexecuted items are retained for the next cycle, and the corresponding effective integral weight is kept at zero.

[0092] The solar observation window for geometric candidate solutions is maintained incrementally over time. When a new observation is entered, old observations with a time span of more than 7 days are deleted. When there are more than 60 valid observations, 60 are retained by equal sampling of the two-dimensional grid of solar azimuth and zenith angle, so that the observation direction distribution of the five types of error parameters is not concentrated in a single solar trajectory segment.

[0093] Before accepting candidate parameters, the root mean square of the angular residuals before and after the solution is compared. If the candidate parameter reduces the root mean square by no less than 20% and the change of any parameter relative to the current value does not exceed 1°, the candidate state is set to available. Otherwise, the candidate state is set to frozen suggestion and the currently effective geometric parameters are retained. The candidate state and the reason for the decision are passed to S5.

[0094] The spatiotemporal correction module finally writes the hemispherical radiance distribution record at the center time according to the node number. Each record includes the interpolated radiance at the center time, the corrected node unit direction vector, the zenith angle, the node solid angle coefficient, the effective integral weight, the nonlinear index, the scan status and the quality bit, and passes the distribution, the candidate values ​​of the geometric error parameters, the currently effective geometric parameter version, the scan table and the period confidence to S4 and S5.

[0095] In this specific embodiment, S4 includes:

[0096] The radiation closure module reads the hemispherical radiance distribution at the center time formed by S3 and the central solar measurement group formed by S2, and establishes a solar exclusion angular domain centered on the epoch-calculated solar unit direction vector, with the exclusion radius set according to... Confirmed, among which This indicates the sun's excluding radius, measured in rad. This represents the radius of the solar disk angle, which is taken as 0.266° in this specific embodiment. The width of the circumstellar angle in the scanning configuration is 2.5° in this specific embodiment; when the angular distance between the node correction direction and the ephemeris direction is not greater than the solar exclusion radius, the node is marked as a solar region and prohibited from entering the scattering integral.

[0097] For nodes outside the solar region with an effective integral weight greater than zero, the scattered horizontal irradiance is calculated as follows: Calculation, where This represents the scattered horizontal irradiance at the center moment, expressed in W·m⁻². This represents the set of hemispherical node indices that are outside the solar region and have valid mass. Represents a node The interpolated radiance at the center moment, in units of W·m⁻²·sr⁻¹. Represents a node The effective integration weight, in units of sr. Represents a node The corrected zenith angle is in rad; if the sum of the effective node solid angles is less than the lower limit of the effective solid angle of the hemisphere set by the scanning configuration, which is 1.7π sr in this specific embodiment, the horizontal irradiance of the current period will still be stored but the quality position will be degraded.

[0098] The total horizontal irradiance in the central solar measurement group is denoted as the reference closed component, and the projection of the direct normal irradiance along the horizontal plane is... Calculation, where This represents direct horizontal irradiance, measured in W·m⁻². This represents the direct normal irradiance at the center moment, expressed in W·m⁻². The solar zenith angle is calculated at the center time, in rad. When the sun is below the horizon or the solar zenith angle is not less than 88°, the direct horizontal irradiance is set to zero and the current cycle is prohibited from being used to update radiation parameters.

[0099] Add the product of the direct normal irradiance and the cosine of the solar zenith angle to the scattered horizontal irradiance, then subtract this sum from the total horizontal irradiance to form a three-component closed residual. The calculation relationship is as follows: ,in This represents the three-component closure residual at the center time, expressed in W·m⁻². A positive value indicates that the sum of the scattered and direct horizontal components calculated based on the current radiation parameters is lower than the total horizontal irradiance of the fixed reference channel. This represents the total horizontal irradiance in the central solar measurement group, expressed in W·m⁻²; to compare the degree of closure at different irradiance levels, it is also calculated... ,in This represents a dimensionless closed relative residual. The horizontal total irradiance reference is used to prevent low irradiance from being divided by zero, and in this specific embodiment it is taken as 50 W·m⁻².

[0100] The module writes the number of nodes, total effective solid angles, number of excluded solar nodes, scattered horizontal irradiance, direct horizontal irradiance, total horizontal irradiance, closure residual, relative closure residual, and quality bits of each component into the radiation closure record, and uses the scan cycle identifier, center time, currently effective geometric parameter version, currently effective node solid angle coefficient version, and radiation parameter version actually involved in the directional irradiance correction of this cycle as a joint index; when any input component is invalid, a new synchronization loss or quality fault occurs in this cycle, the effective solid angle is insufficient, or the total horizontal irradiance is lower than 50 W·m⁻², the closure record is marked as output only and not used for parameter updates; the historical degradation confidence is only used as a marker for the recovery phase and does not individually prevent the closure record from entering the controlled recovery window when each input becomes valid again in this cycle;

[0101] Before writing to the closed window, the module deletes duplicate records according to the scan cycle identifier and center time, converts the three irradiance components into W·m⁻² and checks that the time mark difference is not greater than 2 ms. For records with non-finite values, negative horizontal total irradiance, negative direct normal irradiance or exceeding the sensor calibration limit, write the abnormal reason and transfer them to the output area only. Only records that pass the check are sorted by time increment.

[0102] The radiation closure record and the hemispherical radiance distribution at the center time together constitute the unified spatiotemporal reference measurement results for this period. The module writes the closure records that can be updated into the closure window of the most recent 20 scan periods in increments over time. The output records are saved separately with the reason for failure. Then, the closure window, the closure record of the current period and the distribution record are passed to S5 to perform mutual exclusion calibration state selection.

[0103] In this specific embodiment, S5 includes:

[0104] The calibration control and output module reads the geometric candidate parameters, supplementary scan table, and period confidence of S3, as well as the closed window and current period closed record of S4, and calculates the solar measurement quality from the central solar measurement group; the solar barycenter signal-to-noise ratio is calculated according to... Calculation, where This represents the dimensionless signal-to-noise ratio of the solar barycenter, with a larger value indicating better barycenter mass. This represents the peak value of the centroid after subtracting the background radiation from the solar neighborhood, expressed in W·m⁻². The standard deviation of background noise for samples at the outer edge of the solar neighborhood is expressed in W·m⁻², and is taken as 1 W·m⁻² if less than 1 W·m⁻²; the solar position span involved in the geometric solution is... Calculation, where This indicates the span of the sun's position, measured in rad. and Represents any two valid solar observation indices. and These represent the calendar solar unit direction vectors at the corresponding times;

[0105] Node scan stability is evaluated within the most recent five consecutive scan period windows. The change in geometric parameters is taken as the absolute difference between each component of adjacent candidate vectors, and the bidirectional measurement difference statistic is taken from all valid nodes. The 90th quantile, the horizontal total irradiance time curvature is taken as the maximum absolute value within the window, and the probe temperature change is calculated according to... Calculation, where Indicates the first The change in probe temperature between each scan cycle and the previous cycle, expressed in °C. Indicates the scan cycle number. Indicates the first The probe temperature at the center of each scan cycle, in °C. The probe temperature at the center of the previous cycle is expressed in °C.

[0106] The calibration configuration records, using the configuration version number as the primary key, the following thresholds: signal-to-noise ratio (SNR) threshold, location span threshold, convergence thresholds for five types of geometric parameters, bidirectional difference statistical threshold, node-by-node absolute difference threshold, stable curvature threshold, cloud curvature threshold, temperature change threshold, closure recovery threshold, parameter upper and lower limits, and recovery window. In this specific implementation, one configuration record sets the SNR threshold to 20, the location span threshold to 15°, the convergence thresholds for each angular parameter to 0.01°, the bidirectional difference statistical threshold to 8 W·m⁻²·sr⁻¹, the node-by-node absolute difference threshold to 8 W·m⁻²·sr⁻¹, the stable curvature threshold to 1 W·m⁻²·s⁻², and the cloud curvature threshold to 5. W·m⁻²·s⁻², temperature change threshold is set to 1℃, closed relative residual recovery threshold is set to 0.05 and continuous recovery window is set to 3 cycles. The node-by-node absolute difference threshold uses the same configuration item and version number as the node absolute difference threshold in the S2 scan configuration. The above thresholds are obtained by statistically analyzing the 95th percentile of no less than 30 clear sky stable cycles, 30 cloud change cycles and temperature chamber calibration data. The version number is incremented when the configuration is reviewed or the calibration equipment is replaced.

[0107] When the absolute value of the time curvature of the newly occurring horizontal total irradiance in this cycle is greater than the cloud curvature threshold, synchronous acquisition is lost, the effective solid angle is insufficient, the solar measurement quality is invalid, or the closed record cannot be used for updating, the fault flag and the freeze trigger flag for all parameters in this cycle are set to true; the historical degradation confidence level only retains the recovery phase mark and is not used as a new freeze trigger source; when the fault flag and the freeze trigger flag for all parameters in this cycle are both false, and the changes in the five types of geometric parameters in five consecutive cycles are not greater than their respective convergence thresholds, the bidirectional measurement difference statistic is not greater than the bidirectional difference threshold, the absolute value of the time curvature of the horizontal total irradiance is not greater than the stable curvature threshold, and the probe temperature change is not greater than the temperature change threshold, the radiation parameter update trigger flag is set to true; otherwise, it is set to false.

[0108] Before selecting the mutually exclusive state, the module independently compares the absolute value of the bidirectional measurement difference with the node absolute difference threshold in the scan configuration node by node. When the absolute value of the bidirectional measurement difference of the same node is greater than the threshold and the absolute value of the horizontal total irradiance time curvature is not greater than the stable curvature threshold, regardless of whether the subsequent selection is geometric parameter update, radiation parameter update or all parameter freeze state, the node and its neighboring nodes whose angular distance from its target direction is not greater than 2° are deduplicated and written into the supplementary scan table and the node-by-node stable out-of-tolerance reason code is recorded. This supplementary scan table is merged with the supplementary scan table formed by S3 based on the nonlinear criterion according to the scan cycle identifier and node number, retaining the higher priority, the larger number of consecutive failures and all reason codes of the two. The supplementary scan action itself does not change the state of the mutually exclusive parameters.

[0109] The mutual exclusion state selection strictly follows the order of the all-parameter freeze trigger flag, the radiation parameter update trigger flag, and the solar measurement quality. When the all-parameter freeze trigger flag is true, the all-parameter freeze state is selected. When the flag is false and the radiation parameter update trigger flag is true, the radiation parameter update state is selected. When both flags are false, the solar centroid signal-to-noise ratio is not less than 20, the solar position span is not less than 15°, and the geometric candidate state output by S3 is available, the geometric parameter update state is selected. In other cases, the all-parameter freeze state is selected. The available state also requires that the candidate is identifiable, the solution converges, the root mean square of the angular residual decreases by not less than 20%, and the change of any parameter relative to the current value does not exceed 1°. When the candidate state is a freeze suggestion or unavailable, the geometric update qualification is cleared and the corresponding reason code is written, thereby ensuring that only one parameter layer is open in each scan cycle.

[0110] During the geometric parameter update state, the module copies the candidate vectors of geometric error parameters obtained in S3 that are in an available state to the effective area of ​​the next cycle, and increments the version of the geometric parameters for the next cycle by 1. The current uniform radiation gain or grouped radiation gain and the solid angle coefficient of the currently effective node are copied to the corresponding parameter area of ​​the next cycle and marked as frozen. The candidate vectors only take effect at the parameter snapshot switching point at the beginning of the next scan cycle. The hemispherical distribution and closed records generated in this cycle continue to reference the current effective version to avoid using different geometric parameters for data before and after the same cycle.

[0111] When the radiation parameters are updated and the unified gain mode is selected, the candidate values ​​for unified radiation gain are as follows: Calculation, where This represents a candidate value for the uniform radiation gain in the next scan cycle, and is a dimensionless positive number. This represents the currently effective uniform radiation gain, which is a dimensionless positive number. This represents the dimensionless gain update step size, which is 0.2 in this specific implementation. This represents the truncated average calculated from the closed relative residuals available for updating within each period within the closed window. It is a dimensionless window statistic and distinct from the single-period closed relative residuals of S4. When truncating, delete the 10% of records with the largest absolute value; To increase the candidate gain, A negative value reduces the candidate gain, further limiting the candidate value to a safe range of 0.8 to 1.2;

[0112] When configuring the zenith ring grouping mode, hemispherical nodes are divided into three groups: 0° to 30°, 30° to 60°, and 60° to 85°. The configuration field allows selection between the grouped radiation gain candidate mode and the node solid angle coefficient candidate mode. In the grouped radiation gain candidate mode, a linear least squares equation is constructed using the contribution of each group to the closure residual, and a smoothing term is added to the square of the gain difference between adjacent groups to obtain three grouped radiation gain candidate values. Simultaneously, the original values ​​of all currently active node solid angle coefficients are copied to the candidate snapshot. In the node solid angle coefficient candidate mode, either the unified radiation gain or the three grouped radiation gains are fixed to the currently active values. Three ring coefficient correction factors are obtained based on the contribution of each zenith ring to the scattering integral and closure residual. The reference solid angle coefficients of each node within the same ring are multiplied by the corresponding correction factor and then normalized so that the sum of all node solid angle coefficient candidate values ​​equals 2π. sr, and restrict each candidate value to the range of 0.8 to 1.2 times the corresponding baseline value; this mode uses fixed radiation gain as an identifiable constraint, while the grouped radiation gain candidate mode uses fixed node solid angle coefficient as an identifiable constraint, prohibiting the simultaneous change of two types of parameters in the same closed correction; only when there are no less than five closed cycles available for updating, the solution matrix is ​​full rank, the normalization constraint is satisfied, and the candidate value does not touch the safety boundary, the selected candidate value, mode code, solution window and next version number are written into the radiation candidate area, otherwise the radiation update is abandoned and all parameters are frozen;

[0113] When all parameters are frozen, regardless of the reason for the freeze being triggered by cloud change, synchronization loss, insufficient effective solid angle, invalid solar measurement quality, inability to use closed records for updates, candidate solution failure, or snapshot submission failure, the module copies the original values ​​of the currently effective geometric parameters, radiation gain, and nodal solid angle coefficients to the effective area of ​​the next cycle while keeping the version numbers unchanged. The freeze flag is set to true, the cycle confidence level is uniformly set to the downgraded level, and the freeze reason code is saved. The next scan cycle immediately following the freeze only loads the currently effective snapshot and performs measurements, rescanning, and solar neighborhood scanning, without allowing any parameter updates to be submitted.

[0114] Recovery from historical degradation is achieved using a continuous cycle counter. The next scan cycle immediately following a freeze is the first controlled recovery cycle. As long as no new quality faults occur in this cycle, synchronous acquisition remains locked, the effective solid angle meets the lower limit, solar measurement quality is valid, and the absolute value of the closure relative residual is no greater than 0.05, the counter is recorded as 1, but all parameters remain frozen. From the next scan cycle onwards, only consecutive qualified cycles are allowed to select a parameter layer according to the aforementioned mutual exclusion criterion. The second consecutive qualified cycle completes the candidate review without degrading. When the cumulative count reaches the third consecutive qualified cycle, if the selected geometric parameter update or radiation parameter update has completed the atomic snapshot submission and confirmation record verification, the historical degradation is degraded and the confidence level is restored to the normal level. Any new fault, closure exceeding the limit, or snapshot confirmation failure will reset the counter to zero and re-execute the measurement-only rule of the immediately following freeze cycle. Therefore, any freeze cause applies the same constraint of maintaining the current parameters to the next cycle.

[0115] Before the parameter snapshot is submitted, the module performs consistency arbitration on the candidate regions, candidate states, version chains, and configuration check codes according to the selected mutually exclusive states. In the geometric parameter update state, only the geometric candidate regions are checked to be non-empty and the candidate states are available, and the current radiation parameters are copied and frozen. Empty radiation candidate regions do not affect the submission. In the radiation parameter update state, only the radiation candidate regions are checked to be non-empty, the mode codes are valid, and the solution constraints are satisfied, and the current geometric parameters are copied and frozen. Empty or unavailable geometric candidate regions do not affect the submission. In the all-parameter-frozen state, no candidate region is required to be valid and all current values ​​are copied. When two candidate regions are valid at the same time, only the candidate corresponding to the selected state is submitted and the other candidate is marked as unselected and frozen. When the candidate corresponding to the selected state does not meet the aforementioned conditions, the module switches to the all-parameter-frozen state. When the candidate versions are not continuous, the current version is maintained. When the supplementary scan table version lags behind the current cycle, its cycle identifier is updated. When the configuration check codes are inconsistent, the candidate parameters are rejected. When the atomic snapshot write fails, the module rolls back to the currently effective snapshot. When the load verification fails in the next cycle, the previous cycle snapshot is used and the confidence level is set to the downgraded level.

[0116] The snapshot writer submits parameter fields according to the status code. When the status code is geometric update, only the geometric parameters are updated. When the status code is radial update, only the radial parameters are updated. When the status code is frozen, all current values ​​are copied. When the status code is missing, the submission is rejected. When the submission is completed, an acknowledgment record containing parameter hash and period identifier is generated. When the acknowledgment record verification fails, the previous effective snapshot is rolled back immediately.

[0117] The module generates a freeze flag representing the selected mutually exclusive state, and stores the freeze flag, the next cycle effective geometric parameters, the next cycle effective unified radiation gain or grouped radiation gain, the next cycle effective node solid angle coefficient, the version of each parameter, the supplementary scan table, the cycle confidence level, the solar measurement quality, the node scan stability index, the closure record, and the state selection reason code in association with the scan cycle identifier. It is loaded as an atomic snapshot at the beginning of the next cycle. S2 reads the effective radiation gain at the snapshot switching point and applies it only once before the directional radiance is written into the node record. At the same time, it reads the currently effective node solid angle coefficient but does not multiply the coefficient into the radiance. S3 uses the currently effective node solid angle coefficient to form an effective integration weight. S4 binds the actual radiation parameter version and the node solid angle coefficient version in the closure record, thereby making the grouped radiation gain candidate value or the node solid angle coefficient candidate value have a unique, clear and traceable point of action in the next cycle. At the same time, it outputs the hemispherical directional radiance distribution, total horizontal irradiance, direct normal irradiance, scattered horizontal irradiance, three-component closure residual, and calibration status at the center time as the measurement results and calibration results of the unified spatiotemporal reference for this scan cycle.

[0118] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0119] This invention places fixed reference irradiance, directional irradiance, and axial angle data on the same time base using a unified clock. It cancels the time offset caused by the scanning order by scanning pairs of nodes in reverse order before and after the center time, and establishes a direct normal irradiance and geometric direction reference by combining solar neighborhood measurements. The hemispherical direction distribution corrected by the currently effective geometric parameters is used for the integration of scattered horizontal irradiance and the three-component closure, thereby forming a verification chain between the direction distribution and the independent irradiance components.

[0120] This invention sets up a mutually exclusive parameter freezing structure between the geometric calibration layer and the radiation calibration layer, and manages candidate parameters, currently effective parameters, and parameters effective in the next cycle separately. When solar mass, geometric convergence, bidirectional difference, horizontal total irradiance curvature, and probe temperature meet the corresponding thresholds, only the corresponding parameter layer is opened. When cloud conditions change or mass is insufficient, the current parameters are maintained and the confidence level is reduced. At the same time, the supplementary scan table and state products are fed back to the next cycle, thereby better suppressing the coupling between different error sources.

Claims

1. A method for measuring hemispherical sky radiation based on dual-axis scanning, characterized in that, include: S1. Set up a horizontal total radiation reference channel on a fixed base, and set up a directional radiation probe in the dual-axis scanning mechanism that can switch between sky radiance range and direct solar attenuation range, and collect data from both and the dual-axis encoder synchronously with a unified clock. S2. Scan the same hemispherical nodes in reverse order around the central time, and scan the solar neighborhood between the two scans to form paired direction measurement groups and a central solar measurement group with each node located on both sides of the central time. S3. Correct the radiance of the nodal direction to the center time according to the time distance of each measurement in the paired direction measurement group from the center time. Determine the candidate values ​​of the geometric error parameters according to the central solar measurement group and the calculated solar vector. Correct the nodal direction with the currently effective geometric parameters to form the hemispherical radiance distribution at the center time. S4. Excluding the solar regions in the distribution, the scattered horizontal irradiance is obtained by integrating the cosine of the nodal solid angle and the zenith angle, and combined with the total horizontal irradiance and the direct normal irradiance in the central solar measurement group to form a three-component closed residual. S5. Based on solar measurement quality, node scanning stability, and closure residual, mutually exclusive execution of geometric parameter update, radiation parameter update, or all parameter freeze: update the geometric parameters of the next cycle with candidate values ​​and freeze the radiation parameters, or update the radiation parameters of the next cycle with the closure residual and freeze the geometric parameters, or keep the currently effective parameters for the next cycle to call. S2 includes: A hemispherical node table is constructed according to the pre-stored node numbers and node target directions. The scanning order before the center time is in ascending order of node numbers, and the scanning order after the center time is in descending order of node numbers. Two measurements with the same node number before and after the center time are associated as the paired direction measurement group. The solar neighborhood consists of target points whose angular distance from the direction of the calendar sun is not greater than the radius of the solar neighborhood in the scanning configuration. The solar centroid is obtained by calculating the centroid of the direct solar attenuation range of the target points, and the time center of the solar neighborhood scan is used as the center time. The horizontal total irradiance is obtained based on the synchronous measurement value of the horizontal total radiation reference channel at the central time. The direct normal irradiance is obtained based on the measurement value of the solar direct attenuation range in the direction of the solar barycenter and the known transmittance corresponding to the range. The solar barycenter, horizontal total irradiance, direct normal irradiance and their corresponding time stamps and dual-axis coded values ​​are associated with the central solar measurement group. In step S3, the directional radiance of the same node before and after the center time is multiplied by a weight proportional to the time distance on the other side, and then summed. The sum of the time distances on both sides is then normalized to obtain the interpolated radiance of the node at the center time. According to the radiance normalization scale and curvature normalization scale in the scanning configuration, the absolute value of the bidirectional measurement difference at the same node and the absolute value of the time curvature of the horizontal total radiation reference channel in the corresponding time period are normalized and added together to obtain a dimensionless nodal time nonlinearity index. When the node time nonlinearity index is greater than the dimensionless nonlinearity threshold in the scanning configuration, a supplementary scan table is established. The node and nodes whose target direction angular distance is not greater than the node neighborhood radius in the scanning configuration are written into the supplementary scan table. Before the supplementary scan passes the same criterion, the effective integration weight of the node is reset to zero. When the supplementary scan passes, the supplementary scan measurement value is re-interpolated and the node solid angle coefficient is restored as the effective integration weight. When the number of consecutive supplementary scan failures reaches the failure number threshold in the scanning configuration, the effective integration weight is kept at zero and the confidence level of the scanning cycle is set to the downgrade level. When the node time nonlinearity index is not greater than the dimensionless nonlinearity threshold, the effective integral weight is obtained by multiplying the node solid angle coefficient by a coefficient obtained by subtracting the ratio of the node time nonlinearity index to the dimensionless nonlinearity threshold.

2. The hemispherical sky radiation measurement method based on dual-axis scanning according to claim 1, characterized in that, S1 includes: A horizontal total radiation meter is fixed to the fixed base to form the horizontal total radiation reference channel. An azimuth axis and a pitch axis with a dual-axis encoder are set on the dual-axis scanning mechanism. A directional radiation probe is set at the scanning end of the pitch axis, and a temperature sensor is set at the directional radiation probe. The unified clock provides synchronous trigger signals to each acquisition channel and the dual-axis drive controller, and writes the trigger time, reference irradiance, direction probe response value, range status, azimuth code value, elevation code value and probe temperature value into the same acquisition record. The sky radiance range uses a non-attenuated optical path, while the direct solar attenuation range uses an attenuated optical path with a known transmittance. The direction probe response value is scaled based on the range state and the transmittance.

3. The hemispherical sky radiation measurement method based on dual-axis scanning according to claim 1, characterized in that, In step S3, the solar centroid, corresponding time markers, and dual-axis encoded values ​​are extracted from the central solar measurement group of multiple scanning cycles, and the solar vectors of each ephemeris are calculated based on the corresponding time markers. The angular residual between the solar centroid measurement vector after correction by geometric error parameters and the calculated solar vector is used as the target quantity. The identifiability condition is that the rank of the Jacobian matrix of the angular residual with respect to the five types of geometric error parameters is five and the condition number is not greater than the identifiability threshold in the calibration configuration. Candidate values ​​of geometric error parameters are obtained. The five types of geometric error parameters include the azimuth zero bias of the scanning mechanism, the roll angle of the fixed base, the pitch angle of the fixed base, the non-orthogonality angle between the azimuth axis and the pitch axis, and the pitch zero bias. Substitute the currently effective geometric parameters into the coordinate transformation from the dual-axis encoded values ​​to the spatial direction to obtain the corrected hemispherical node direction.

4. The hemispherical sky radiation measurement method based on dual-axis scanning according to claim 3, characterized in that, In step S4, the solar region includes an angular region centered on the calendar solar direction, with the sum of the solar disk angular radius and the circumstellar angular width in the scanning configuration as the exclusion radius; For each hemispherical node outside the solar region, the product of the node interpolated radiance, the effective integral weight, and the cosine of the node zenith angle is summed to obtain the scattered horizontal irradiance. The product of the direct normal irradiance and the cosine of the solar zenith angle is added to the scattered horizontal irradiance, and the summation is subtracted from the total horizontal irradiance to obtain the three-component closed residual.

5. The hemispherical sky radiation measurement method based on dual-axis scanning according to claim 4, characterized in that, In step S5, the solar measurement quality includes the solar barycenter signal-to-noise ratio and the solar position span. The solar barycenter signal-to-noise ratio is determined by the ratio of the peak value of the barycenter in the solar neighborhood to the standard deviation of the background noise. The solar position span is determined by the upper bound of the angle between the epoch-calculated solar vectors involved in the solution. Set a full parameter freeze trigger flag and a radiation parameter update trigger flag. Select the state in the order of the full parameter freeze trigger flag, the radiation parameter update trigger flag, and the solar measurement quality. Select the full parameter freeze state when the full parameter freeze trigger flag is true. Select the radiation parameter update state when the full parameter freeze trigger flag is false and the radiation parameter update trigger flag is true. Select the geometric parameter update state when both trigger flags are false, the solar centroid signal-to-noise ratio is not less than the signal-to-noise ratio threshold in the calibration configuration, and the solar position span is not less than the position span threshold in the calibration configuration. Select the full parameter freeze state in all other cases. In the updated geometric parameters state, the radiation gain and nodal solid angle coefficients are frozen, and the candidate values ​​of the geometric error parameters are updated to the geometric parameters that will take effect in the next scan cycle.

6. The method for measuring hemispherical sky radiation based on dual-axis scanning according to claim 5, characterized in that, In step S5, the node scanning stability is characterized by the change in geometric parameters, the bidirectional measurement difference statistic of the same node, the time curvature of the total horizontal irradiance, and the change in probe temperature. A temperature sensor is installed at the directional radiation probe, and the temperature change of the probe is determined based on the difference between the synchronous measurement values ​​of the temperature sensor in adjacent scanning cycles. When the absolute value of the change of each geometric parameter within the continuous scanning period window is not greater than the convergence threshold in the calibration configuration, the bidirectional measurement difference statistic is not greater than the bidirectional difference threshold, the absolute value of the horizontal total irradiance time curvature is not greater than the stable curvature threshold, and the probe temperature change is not greater than the temperature change threshold, the radiation parameter update trigger flag is set to true. In the radiation parameter update state, the currently effective geometric parameters are frozen, and a unified radiation gain candidate value is determined based on the three-component closed residual of multiple scan cycles, or a grouped radiation gain candidate value and a node solid angle coefficient candidate value are determined based on the preset zenith angle ring group to which the hemispherical node belongs. The lower limit and upper limit of the parameters in the calibration configuration are used as safety constraint limit candidate values, and the candidate values ​​are updated to the radiation parameters effective in the next scan cycle.

7. The method for measuring hemispherical sky radiation based on dual-axis scanning according to claim 5, characterized in that, In step S5, an anomaly recovery branch is set up, the absolute value of the bidirectional measurement difference of the same node is compared with the node difference threshold in the scanning configuration, and the absolute value of the horizontal total irradiance time curvature is compared with the stable curvature threshold and cloud change curvature threshold in the calibration configuration, wherein the cloud change curvature threshold is greater than the stable curvature threshold. When the absolute value of the bidirectional measurement difference is greater than the node difference threshold and the absolute value of the horizontal total irradiance time curvature is not greater than the stable curvature threshold, the node and its neighboring nodes are written into the supplementary scan table. When the absolute value of the total horizontal irradiance time curvature is greater than the cloud curvature threshold, the all parameter freeze trigger flag, which has priority over other states, is set to true, the currently effective parameters are maintained, and the confidence level is set to the downgrade level in the calibration configuration. The same process is performed when the all parameter freeze state is selected in other cases in step S5. The downgrade level ensures that the current effective parameters are maintained in the next scan cycle and that the supplementary scan table and solar neighborhood scan are executed first. If the freeze trigger flag for all parameters is not triggered within the continuous recovery window, at least one parameter update status is completed, and the absolute value of the three-component closed residual is not greater than the recovery threshold, the confidence level is restored to the normal level in the calibration configuration. A freeze flag representing the selected state is generated, and the freeze flag, the geometric and radiometric parameters that will take effect in the next scan cycle, the supplementary scan table, and the confidence level are associated and stored for use in the next scan cycle.

8. A hemispherical sky radiation measurement system based on dual-axis scanning, used to perform the hemispherical sky radiation measurement method based on dual-axis scanning as described in any one of claims 1 to 7, characterized in that, include: The synchronous acquisition module is used to synchronously generate reference radiation data, directional radiation data and axis angle data from a horizontal total radiation reference channel set on a fixed base, a directional radiation probe set on a dual-axis scanning mechanism and capable of switching ranges, and a dual-axis encoder. The paired scanning module is used to scan the same hemispherical nodes and the solar neighborhood in reverse order around the central time to form a paired directional measurement group and a central solar measurement group; The spatiotemporal correction module is used to correct the directional radiance of each hemispherical node to the center time, determine candidate values ​​of geometric error parameters, and call the currently effective geometric parameters to correct the orientation of the hemispherical nodes. The radiation closure module is used to integrate the hemispherical nodes excluding the solar region to obtain the scattered horizontal irradiance and form a three-component closed residual. The calibration control and output module is used to select a state from mutually exclusive parameter update states, generate a freeze flag, a parameter vector effective for the next cycle, a supplementary scan table, and a confidence level for use in the next scan cycle, and output measurement and calibration results with a unified spatiotemporal reference.

Images