Intelligent controllable method and system for simulating dynamic change of solar radiation

Abstract

The invention relates to the technical field of solar simulation adjustment, and discloses an intelligent controllable method and system for simulating dynamic change of solar radiation. Long-term irradiation data are obtained from a database and a local weather station and fused, the discontinuity of the daily boundary is overcome by applying a circumferential mapping and clustering method, and the maximum daily radiation intensity, the typical sunrise moment and the sunset moment of each month are obtained; on the basis, constructing an attenuation model taking a monthly typical value as a benchmark, introducing random disturbance and physical constraints conforming to meteorological statistical characteristics, generating intra-month daily simulation parameters, further utilizing a trigonometric function improvement model to dynamically represent the daily simulation parameters of the current month as target irradiation intensities at all moments, standardizing the target irradiation intensities and mapping the target irradiation intensities into command control quantities, and calculating the target irradiation intensities. Precise reproduction of the natural solar radiation curve by the illumination intensity is realized; then actual light intensity is detected through an irradiation sensor, execution errors are calculated, control action quantity is generated in real time through a PID control law, and finally execution quality is evaluated through a control action quantity sequence.

Description

Technical Field

[0001] This invention relates to the field of solar simulation and regulation technology, specifically to an intelligent and controllable method for simulating dynamic changes in solar radiation. Background Technology

[0002] Solar radiation exhibits significant intra-diurnal dynamic variations in the natural environment. Its irradiance and spectral distribution change in real time with the solar altitude angle and meteorological conditions, displaying a continuous, non-steady-state pattern from sunrise to sunset. In key research areas such as photocatalytic degradation of pollutants, solar fuel synthesis, photoelectrochemical water splitting, and soil evaporation, the development of intelligent and controllable methods and systems capable of accurately reproducing the aforementioned dynamic and continuous changes in solar radiation in controlled indoor environments has become a crucial prerequisite for enhancing the scientific rigor and practicality of related research.

[0003] However, existing experiments generally use static light source systems with fixed output power as the simulated light source, typically using xenon lamps as the radiation source for solar simulation. Although this spectrum is similar to sunlight, the fixed output mode makes it difficult to reproduce the rhythmic changes under natural lighting conditions. This fails to meet the dynamic variation requirements of solar radiation intensity in indoor laboratories, especially during the low light periods of dawn and dusk, where the simulation accuracy is poor. Because the sunrise and sunset of the sun exhibit very obvious rhythmic changes, the sunrise and sunset times and radiation intensities vary significantly across different months. Consequently, the experimental results obtained from solar simulation experiments (dynamic variation characteristics of evaporation intensity, photochemical reaction rates, or energy conversion efficiency, etc.) often have systematic deviations from real-world observations, resulting in certain limitations in indoor simulation results. Summary of the Invention

[0004] In view of the problems in related technologies, the present invention provides an intelligent and controllable method for simulating the dynamic changes of solar radiation, so as to overcome the technical problems existing in the existing related technologies.

[0005] To solve the aforementioned technical problem, the present invention is achieved through the following technical solution:

[0006] On the one hand, this invention provides an intelligent and controllable method for simulating the dynamic changes of solar radiation, specifically including the following steps:

[0007] S101: Hourly irradiance data for the target area was obtained from the NASA POWER meteorological database for at least one consecutive year, and validated and corrected by integrating measured data from local weather stations. Based on the validated and corrected data, the daily total radiation intensity was calculated on a daily basis, and then divided by month to obtain the daily total radiation intensity for each month. The maximum daily radiation intensity Rmax was selected for each month. Simultaneously, the typical sunrise time T0 and typical sunset time T1 for the target area in each month were calculated using astronomical algorithms. Based on Rmax, T0, and T1, an attenuation model was constructed using the monthly typical values ​​as a benchmark. Random perturbations and physical constraints conforming to meteorological statistical characteristics were introduced to generate daily simulation parameters for the month, i.e., the maximum daily radiation intensity. Target sunrise time and the target sunset time , where d is the index of the simulated number of days in the month;

[0008] S102, Obtain the current month of the target region and match the daily simulation parameters for the corresponding month, i.e., the maximum daily solar radiation intensity. Target sunrise time and the target sunset time Rmax, T0, and T1 are used to dynamically represent the target irradiance intensity at time t in the target area using an improved trigonometric function model. This will be converted into a command control quantity to drive the xenon lamp electronic ballast to adjust the xenon lamp power in real time; the specific dynamic characterization formula for radiation intensity is: κ is the shape parameter, Sc(t) is the atmospheric correction factor, Sc(t)∈(0,1]; t is the current time of the target area;

[0009] S103, an irradiation sensor is arranged at the output end of the optical path to measure the actual light intensity and convert it into an actual control quantity. The execution error is calculated, and PID control is performed on the execution error to obtain the control action quantity. The xenon lamp power is adjusted according to the control action quantity to form a closed-loop control.

[0010] S104 performs execution quality assessment and fault diagnosis based on the control action at each moment.

[0011] Preferably, the typical sunrise time T0 and typical sunset time T1 for the target region each month are:

[0012] The sunrise and sunset times for the target region are calculated using astronomical algorithms for each day of each month. A 24-hour day is divided into several equal-length periods. The sunrise time for each day is then assigned to its corresponding equal-length period, and the number of sunrise times within each period is counted cumulatively. The period with the most sunrise times is selected as the typical sunrise period. Similarly, the typical sunset period is determined using the same method. Each sunrise time within the typical sunrise period is converted into minutes within a day as a data point. A circular mapping and two-dimensional vector space clustering are then used to process the data points for each day of the month to obtain the typical sunrise time T0. Similarly, the same processing is applied to all sunset time samples within the typical sunset period to obtain the typical sunset time T1.

[0013] Preferably, a circumferential mapping is introduced:

[0014] Each sunrise moment in a typical sunrise period is converted into a corresponding number of minutes, and with a zero-period constant T on a diurnal scale, the number of minutes corresponding to each sunrise moment is mapped to an angle θ. The mapping formula is as follows: Each sunrise time corresponds to an angle, forming an angle set. As can be seen from the mapping formula, the angle θ∈[0,2π). Each angle θ is mapped to a two-dimensional vector coordinate (cosθ, sinθ) using cosine and sine. Thus, each sunrise time in a typical sunrise period can be transformed into a two-dimensional vector coordinate.

[0015] Preferred method: Two-dimensional vector space clustering

[0016] The K-means clustering algorithm is adopted, and the number of clusters K=1 is set. Since the goal is to find the most representative center point from a dataset, setting K=1 is reasonable and efficient. In practical applications, the specific value can be adjusted according to whether minute-level or second-level data is used. A point is randomly selected from the two-dimensional vector coordinate dataset as the initial cluster center μ1. The algorithm enters the iterative process to optimize the position of the cluster center. The mean of all data points in cluster C1 is calculated, and the position of the cluster center μ1 is updated accordingly. The mean calculation is performed in a two-dimensional coordinate system, that is, the X coordinate of the new cluster center = the arithmetic mean of the X coordinates (i.e., cosθ) of all data points, and the Y coordinate of the new cluster center = the arithmetic mean of the Y coordinates (i.e., sinθ) of all data points. The formula is expressed as: N is the total number of sunrise times within a typical sunrise period, i.e., the total number of angles in the angle clustering. Repeat this update step until the position change of the cluster center μ1 is less than the preset convergence threshold, or the maximum number of iterations is reached. At this time, the algorithm determines that the cluster center has converged. The coordinates of the final cluster center after convergence are converted back to time according to the inverse process of time, and the typical sunrise time T0 is obtained.

[0017] Preferably, the coordinates of the final cluster centers after convergence are transformed back to time by the inverse process of converting time into coordinates:

[0018] The coordinates of the final cluster centers after convergence Calculate its angle θ using the arctangent function in the four quadrants. center : The atan(*) function ensures that the angle θ of the final cluster centers is obtained. center Located in the correct quadrant, its domain is (-π, π], if θ center If θ < 0, then let θ center =θ center +2π, thus ensuring that the angle value falls within the period of [0, 2π);

[0019] angle θ center Convert to minutes of a day Z center The conversion formula is: Then adjust the angle θ center The corresponding number of minutes Z center This is converted into time, which gives us the typical sunrise time T0.

[0020] Preferably, based on Rmax, T0, and T1, a decay model with monthly typical values ​​is constructed, and random disturbances and physical constraints conforming to meteorological statistical characteristics are introduced to generate daily simulation parameters within the month:

[0021] 1-5-1, Defined by sunrise and sunset times as boundaries: Extract the earliest sunrise and latest sunrise times, as well as the earliest sunset and latest sunset times for the current month, and denot them as follows: and ,as well as and Define the radiation intensity boundary: Calculate the minimum daily radiation intensity for the month from historical meteorological data. And the maximum solar radiation intensity of the month Constructing sunrise and sunset time baselines: Using the typical sunrise time T0 and typical sunset time T1 of the month as the baseline values ​​for the month, construct a baseline that passes through the typical values ​​and has a reasonable trend. The sunrise time baseline for the month is thus obtained and denoted as... The baseline for sunset is denoted as Where j represents the index of the actual number of days in the monthly baseline; similarly, construct the maximum daily radiation intensity baseline: using the typical monthly value Rmax as the mid-month baseline, construct a curve reflecting the monthly variation in radiation intensity: using the beginning and end of the month... and Using Rmax as the starting and ending point, and the peak value in the middle of the month as the reference point, a smooth curve is generated as the baseline, denoted as . ;

[0022] 1-5-2, add random perturbations and physical constraints to ensure that the daily parameters generated by the above attenuation and perturbation model are strictly within their physical possible boundaries, thereby obtaining the constrained sunrise time sequence, sunset time sequence and maximum solar radiation intensity sequence;

[0023] 1-5-3, Randomization: To simulate the nondeterministic correspondence between actual meteorological conditions and dates, the constrained sunrise time sequence, sunset time sequence, and maximum daily radiation intensity sequence are randomly rearranged to generate a random permutation from 1 to D, which serves as the new date order; where D is a positive integer representing the total number of actual or simulated days in the month; then, the reordered sequence is used as the daily simulation parameter for the month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time ), where d∈D, and d represents the simulated day index of the current month.

[0024] Preferably, the control action is obtained by performing PID control on the execution error:

[0025] An irradiation sensor is placed at the output end of the optical path to measure the actual output light intensity and convert it into an actual control quantity. The actual control quantity is then subtracted from the command control quantity to obtain the execution error e(t). Finally, a PID control law is applied to the execution error e(t) to calculate the real-time control action Y(t). The calculus calculation formula is as follows: e(t) represents the execution error at the current moment. K represents the cumulative total of all execution errors in the past, k represents the time before the current time t, and Ts is the sampling period; p K i and K d These are the proportional coefficient, integral coefficient, and differential coefficient, respectively.

[0026] Preferably, performance evaluation and fault diagnosis are performed based on the control action at each moment:

[0027] A two-dimensional coordinate system is constructed with time as the horizontal axis and the control action quantity as the vertical axis. Then, the control action quantity at each moment is input into the coordinate system to form data points. The data points are connected sequentially with a smooth curve to construct the control action quantity change curve. The polynomial least squares method is used to perform trend fitting on the control action quantity change curve to obtain the fitted curve function. The first derivative of the fitted curve function is obtained to obtain the instantaneous change trend degree of the control action quantity at each data point.

[0028] Fault diagnosis is performed based on the change curve of the control action and its instantaneous change trend to make early warning of fault risk.

[0029] Preferably, fault diagnosis is based on the control action change curve and its instantaneous change trend:

[0030] The oscillation value of the control action is calculated by using the sample standard deviation formula to determine the instantaneous change trend. The mean value of the control action at each time moment is calculated by taking the mean value. The oscillation value and the mean value are then normalized. A sensitivity coefficient is assigned to each of the normalized oscillation value and the mean value, and the sum of the two sensitivity coefficients is one. Finally, the normalized oscillation value, the mean value, and their corresponding sensitivity coefficients are linearly weighted and fused to obtain the fault diagnosis value.

[0031] A preset diagnostic threshold is set. If the fault diagnosis value is greater than the diagnostic threshold, a fault risk warning is generated.

[0032] On the other hand, the present invention provides an intelligent and controllable system for simulating the dynamic changes of solar radiation. The system specifically includes: a parameter analysis module, a model characterization module, a control adjustment module, and a fault diagnosis module.

[0033] The parameter analysis module obtains hourly horizontal irradiance data for the target area for at least one consecutive year from the NASA POWER meteorological database, and verifies and corrects it by integrating measured data from local weather stations. Based on the verified and corrected data, the daily total radiation intensity is calculated on a daily basis, and the daily total radiation intensity for each month is obtained by dividing it by month. The maximum daily radiation intensity Rmax is selected in each month. At the same time, the typical sunrise time T0 and typical sunset time T1 for the target area in each month are calculated based on astronomical algorithms.

[0034] The model representation module obtains the current month of the target region and matches it with the daily simulation parameters for that month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time The target irradiance intensity at each time t in the target area is obtained by dynamically representing it using a trigonometric function improved model. This will be converted into a command control quantity to drive the xenon lamp electronic ballast to adjust the xenon lamp power in real time; the specific dynamic characterization formula for radiation intensity is: κ is the shape parameter, Sc(t) is the atmospheric correction factor, Sc(t)∈(0,1]; t is the current time of the target area;

[0035] The control and adjustment module is used to place an irradiation sensor at the output end of the optical path to measure the actual light intensity and convert it into an actual control quantity, calculate the execution error, perform PID control on the execution error to obtain the control action quantity, and adjust the xenon lamp power according to the control action quantity to form a closed-loop control.

[0036] The fault diagnosis module performs execution quality assessment and fault diagnosis based on the control action at each moment.

[0037] The present invention has the following beneficial effects:

[0038] This invention acquires and integrates long-term radiation data from the NASA POWER database and local weather stations, and uses astronomical algorithms to calculate the monthly maximum daily radiation intensity Rmax, typical sunrise time T0, and sunset time T1. It then uses circumferential mapping and K-means clustering to overcome the discontinuity of the solar boundary, obtaining accurate solar motion time parameters that conform to seasonal variations. Based on this, an attenuation model is constructed using the monthly typical values ​​as a benchmark, and random perturbations and physical constraints conforming to meteorological statistics are introduced to generate specific simulated parameters (daily maximum radiation intensity) for each day of the month. Target sunrise time and the target sunset time This establishes a refined diurnal parameter set that reflects both seasonal patterns and natural inter-diurnal fluctuations. Furthermore, by matching the corresponding diurnal simulation parameters to the current date, and using trigonometric functions that integrate shape parameters and atmospheric correction factors to improve the model, the parameters are dynamically represented as the target irradiance at each time point. The light intensity is then standardized and mapped into command control quantities, which are then used to drive the xenon lamp electronic ballast via MODBUS communication, achieving accurate reproduction of the natural solar radiation curve. The actual light intensity is detected by an irradiance sensor, and the execution error e(t) is calculated. A PID control law is used to generate the control action Y(t) in real time, and the xenon lamp power is adjusted through the same communication path to form a closed-loop control, ensuring that the actual light intensity closely tracks the target curve. Finally, by analyzing the oscillation values ​​and mean values ​​of the control action sequence, normalization and weighted fusion are performed to generate a fault diagnosis value. When the statistical threshold is exceeded, an early warning is generated. This invention achieves full-link automation from real data-driven multi-scale parameter generation, high-fidelity dynamic illumination simulation, real-time closed-loop feedback control to intelligent fault diagnosis and early warning, improving the realism, data reliability, and system operational safety of solar simulation experiments. It effectively solves a series of key technical bottlenecks of traditional static solar simulators, such as the inability to reproduce the dynamic rhythm of natural illumination, poor accuracy in twilight simulation, weak seasonal adaptability, and insufficient system operational reliability.

[0039] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 The flowchart illustrates an intelligent and controllable method for simulating dynamic changes in solar radiation provided by this invention.

[0042] Figure 2 A block diagram of an intelligent and controllable system for simulating dynamic changes in solar radiation is provided by the present invention.

[0043] Figure 3 This is a comparison and optimization verification diagram of indoor and outdoor radiation for this invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Application Scenarios: Current experiments generally use static light source systems with fixed output power as simulated light sources, typically using xenon lamps as radiation sources for solar simulation. Although this spectrum is similar to sunlight, the fixed output mode makes it difficult to reproduce the rhythmic changes under natural lighting conditions. This fails to meet the continuous dynamic changes in solar radiation intensity required by indoor laboratories, especially during the low light periods of dawn and dusk, where the simulation accuracy is poor. This is because sunrise and sunset exhibit very obvious rhythmic changes, with significant differences in sunrise and sunset times and radiation intensities across different months. Consequently, the experimental results obtained from solar simulation experiments (intra-day variations in evaporation rate, photochemical reaction rates, or energy conversion efficiency, etc.) often show systematic deviations from real-world observations, thus limiting the effectiveness of indoor simulations.

[0046] To solve the above technical problems, such as Figure 1 As shown, this embodiment of the invention provides an intelligent and controllable method for simulating the dynamic changes of solar radiation, specifically including the following steps:

[0047] S101 uses NASA's POWER meteorological database to obtain hourly irradiance data for the target area for at least one consecutive year, and integrates it with measured data from local weather stations for verification and correction to ensure the integrity and accuracy of the data sequence. Based on the verified and corrected horizontal irradiance data, the daily total radiation intensity is accumulated on a daily basis, and then divided by month to obtain the daily total radiation intensity for each month of the year. The maximum daily radiation intensity Rmax is then selected for each month, because in practical applications, the daily total solar radiation intensity is highly dependent on the season (i.e., month). Simultaneously, based on astronomical algorithms, the typical sunrise time T0 and typical sunset time T1 for the target area in each month are calculated. Finally, attenuation and random allocation processing are applied to simulate the maximum daily daily radiation intensity for each month. Target sunrise time and the target sunset time , where d is the index of the number of days in January; the specific calculation process is as follows:

[0048] 1-1 Specifically, the sunrise and sunset times for the target region are first calculated using astronomical algorithms for each day of each month. The 24 hours of a day are then divided into several smaller, equal-length intervals, such as 5 minutes, 10 minutes, or 15 minutes. The sunrise time for each day is then assigned to its corresponding equal-length interval, and the number of sunrise times within each interval is counted. The interval with the most sunrise times is selected as the typical sunrise time. Similarly, the typical sunset time is determined using the same method.

[0049] 1-2. Each sunrise moment in a typical sunrise period is converted into minutes within a day as a data point. In practical applications, midnight 00:00 is used as the starting point. Each sunrise moment in a typical sunrise period is converted into the corresponding minutes, denoted as Z. The calculation formula is minutes = clock × 60 + minutes. The zero-period constant on the daily scale is T. In practical applications, T = 1440 for minutes and 86400 for seconds. This embodiment uses the minute level, i.e., T = 1440. To correctly handle the circumference of the data near the date line, the minutes corresponding to each sunrise moment are mapped to an angle θ. The mapping formula is: Each sunrise time corresponds to an angle, forming an angle set. As can be seen from the mapping formula, the angle θ∈[0,2π). To facilitate the subsequent use of clustering algorithms to obtain typical sunrise times, each angle θ is mapped to a two-dimensional vector coordinate (cosθ, sinθ) using cosine and sine. Thus, each sunrise time in the typical sunrise period can be transformed into a two-dimensional vector coordinate.

[0050] 1-3, the K-means clustering algorithm is used, and the number of clusters K=1 is set. Since the goal is to find the most representative center point from a dataset, setting K=1 is reasonable and efficient. In practical applications, the specific setting can be adjusted according to whether minute-level or second-level methods are used. A point is randomly selected from the two-dimensional vector coordinate dataset as the initial cluster center μ1. The algorithm enters the iterative process to optimize the position of the cluster center, calculates the mean of all data points in cluster C1, and updates the position of the cluster center μ1 accordingly. The mean calculation is performed in a two-dimensional coordinate system, that is, the X coordinate of the new cluster center = the arithmetic mean of the X coordinates (i.e., cosθ) of all data points, and the Y coordinate of the new cluster center = the arithmetic mean of the Y coordinates (i.e., sinθ) of all data points. The formula is expressed as: N is the total number of sunrise times within a typical sunrise period, i.e., the total number of angles in the angle clustering. Repeat this update step until the position change of the cluster center μ1 is less than the preset convergence threshold, or the maximum number of iterations is reached. At this point, the algorithm determines that the cluster center has converged.

[0051] 1-4, the coordinates of the final cluster centers after convergence. Calculate its angle θ using the arctangent function in the four quadrants. center : The atan(*) function ensures that the angle θ of the final cluster centers is obtained. center Located in the correct quadrant, its domain is (-π, π], if θ center If θ < 0, then let θ center =θ center +2π, thus ensuring that the angle value falls within the period [0, 2π), thereby allowing the angle θ of the final cluster center to converge. center Standardize to the interval [0, 2π); then use the mapping relationship from steps 1-2 to adjust the angle θ. center Convert to minutes of a day Z center The conversion formula is: Finally, the angle θ center The corresponding number of minutes Z center This is converted into time, which gives the typical sunrise time T0. Similarly, for all sunset time samples within the typical sunset period, the exact same steps 1-2 to 1-4 are performed to obtain the precise typical sunset time T1.

[0052] By introducing circular mapping and two-dimensional vector space clustering, the one-dimensional linear time axis is mapped onto a two-dimensional circle. This allows moments that are far apart on the original time axis (due to the boundary of the solar calendar) but actually close in the natural cycle to be correctly represented as adjacent points in the two-dimensional vector space, thus ensuring the physical rationality of the clustering algorithm. Furthermore, the cluster centers based on vector operations, compared to calculating the arithmetic mean on a discontinuous one-dimensional number axis, more accurately reflect the concentration trend of sunrise and sunset times in different months due to different seasons, and can more accurately simulate the solar motion time in different years and seasons.

[0053] 1-5. Based on the typical monthly parameters (maximum daily radiation intensity Rmax, typical sunrise time T0, and typical sunset time T1 in each month), the diurnal scale parameter attenuation and random allocation are used to obtain the daily simulation parameters (maximum daily radiation intensity) for each month. Target sunrise time and the target sunset time Specifically:

[0054] 1-5-1, Defined by sunrise / sunset times: Extract the earliest sunrise and latest sunrise times, as well as the earliest sunset and latest sunset times for the current month, and denot them as follows: and ,as well as and Define the radiation intensity boundary: Calculate the minimum daily radiation intensity for the month from historical meteorological data. And the maximum solar radiation intensity of the month Constructing a sunrise / sunset time baseline: Using the typical sunrise time T0 and typical sunset time T1 of the month as baseline values ​​for the middle of the month (around the 15th day), considering that the actual changes in sunrise and sunset times within a month are usually monotonic and approximately linear, a baseline that passes through the typical values ​​and has a reasonable trend can be constructed. Thus, the sunrise time baseline for the month is denoted as... The baseline for sunset is denoted as Where j represents the index of the actual number of days in the baseline for the current month; for example, taking T0 as the midpoint of the month and setting it to... and Linear variation within a range can be achieved by connecting three key points using piecewise linear or smooth curves (such as cubic splines): beginning of the month ( ), mid-month (T0) and end-of-month ( To ensure the baseline is smooth and passes through the typical value, a baseline for maximum solar radiation intensity is constructed: using the monthly typical value Rmax as the mid-month baseline, a curve reflecting the monthly variation in radiation intensity is built. Considering that solar radiation may show a trend of first increasing and then decreasing within the month (such as in summer months), a parabolic model can be used: using the values ​​at the beginning and end of the month... and Using Rmax as the starting and ending point, and the peak value in the middle of the month as the reference point, a smooth curve is generated as the baseline, denoted as . ;

[0055] 1-5-2, Add random perturbations and physical constraints to ensure that the daily parameters generated by the above decay and perturbation model are strictly within their physically possible boundaries:

[0056] Random perturbations are superimposed on the daily baseline values ​​to generate a random perturbation sequence for each parameter independently. The perturbation values ​​are drawn from a normal distribution with a mean of zero, and the standard deviation of the distribution is set according to the parameter characteristics (e.g., the standard deviation of the perturbation at sunrise / sunset is set to 10-30 minutes, and the standard deviation of the perturbation of radiation intensity is set to 5%-15% of the typical monthly value).

[0057] The formula for calculating the superposition of disturbances is:

[0058] (1)

[0059] (2)

[0060] (3)

[0061] in , and These are disturbances at sunrise, sunset, and radiation intensity, respectively.

[0062] To ensure that the perturbed parameters conform to physical reality and logical consistency, the sunrise time sequence obtained after superposition is... Sunset time sequence and maximum solar radiation intensity sequence Apply boundary constraints to enforce all sunrise time sequences exist and All sunset time sequences within the range exist and Within the range, the sequence of maximum solar radiation intensity exist and Within the range; for values ​​exceeding the boundary, they are truncated to the nearest boundary; thus, the constrained sunrise time sequence, sunset time sequence, and maximum solar radiation intensity sequence can be obtained;

[0063] 1-5-3, Randomization: To simulate the non-deterministic correspondence between actual meteorological conditions and dates (i.e., to avoid parameters for specific dates being strictly bound to a specific position in the trend), the constrained sunrise time sequence, sunset time sequence, and maximum daily radiation intensity sequence are randomly rearranged to generate a random permutation from 1 to D (i.e., shuffling), which serves as the new date order; where D is a positive integer representing the total number of actual or simulated days in the month. In practical applications, the total number of actual and simulated days in the month is consistent; then, the reshuffled order is used as the daily simulation parameter for the month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time ), where d∈D, d represents the simulated day index of the current month; for example, if the 5th element in the random order (i.e., d=5) corresponds to the original 12th day (i.e., i=12), then the 5th day will eventually obtain the parameters of the original 12th day; these parameters reflect the overall trend of monthly changes with the typical monthly value as the center, and also include inter-day random fluctuations that conform to meteorological statistical characteristics, and the random allocation enhances the diversity and realism of the simulation scenario;

[0064] By using typical parameter values ​​(T0, T1, and Rmax) of the current month as benchmark anchors to construct the intra-month variation trend, the generated daily parameters are ensured to be consistent with the typical values ​​in overall distribution. Then, through baseline trend, linear decay of controllable time parameters (i.e., baseline), and constrained random perturbation, the maximum daily radiation intensity sequence, sunrise time, and sunset time sequence of the current month are generated, which conform to historical data boundaries and satisfy physical consistency. This sequence retains both seasonal and monthly-scale statistical characteristics, as well as daily-scale randomness and adjacent-day correlation, making it suitable for driving solar simulators to conduct high-fidelity, repeatable, and realistically representative intra-day irradiance simulations. All generated parameters and random seeds are recorded and traceable to support experimental reproduction, result traceability, and long-term adaptive correction.

[0065] S102, Obtain the current month of the target region and compare it with all months of the target region to obtain the daily simulation parameters for the corresponding month, specifically: the maximum daily radiation intensity of the current month. Target sunrise time and the target sunset time The target irradiance at each time t in the target area is obtained by dynamically representing the daily simulation parameters using a trigonometric function-based improved model. The specific dynamic representation formula is as follows: κ is a shape parameter, which typically ranges from 1 to 2 in practical applications. When κ > 1, it indicates a sharper curve peak. Sc(t) is an atmospheric correction factor, where Sc(t) ∈ (0,1]; t is the current time in the target area; then the target irradiance intensity is... Standardization is set to the normalization target u(t), specifically u(t) = / Rmax, and map the normalized target u(t) to the field-drivable command control quantity, which is specifically represented as current or voltage. If the command control quantity is current, then the current mapping formula is used: If the command control quantity is voltage, then the voltage mapping formula is used: The command control quantity is packaged into a MODBUS write register command and written to the designated register of the PLC. After receiving the MODBUS command, the PLC converts the digital command into an analog quantity (0-10V or 4-20mA) to drive the xenon lamp electronic ballast, thereby realizing the xenon lamp power regulation.

[0066] like Figure 3 As shown, the model for simulating day and night solar radiation intensity is improved by using trigonometric functions. The simulation curve's peak value and short-term attenuation are adjusted using shape parameters and atmospheric correction factors to obtain a time-series target irradiance curve with clear physical meaning, consistent with local seasonal and atmospheric conditions. Field drive commands (current or voltage) are obtained through linear or calibration mapping, giving the simulation curve controllable and reproducible physical characteristics in terms of amplitude and diurnal smoothness. This enables smooth, continuous intraday simulation from zero to peak and back to zero, achieving scientific-level solar radiation simulation. This overcomes the limitations of existing static light source systems with fixed output power, which result in poor simulation accuracy of solar radiation.

[0067] S103, an irradiance sensor (specifically a planar or normal radiometer or a standard solar cell) is placed at the output end of the optical path to measure the actual output light intensity and convert it into an actual control quantity. Then, the command control quantity is subtracted from the actual control quantity to obtain the execution error e(t). Finally, a PID control law is used to calculate the real-time control action Y(t) on the execution error e(t). The calculus calculation formula is as follows: e(t) represents the execution error at the current moment. K represents the cumulative total of all execution errors in the past, k represents the time before the current time t, and Ts is the sampling period; p K i and K d These are the proportional coefficient, integral coefficient, and differential coefficient, respectively. The proportional coefficient K... p Used to amplify the instantaneous error e(t) at the current moment, achieving instantaneous and rapid error correction; proportional coefficient K p The larger the value, the stronger and faster the response to errors. In practical applications, the value is usually in the range of 0.1-5. Engineering experience suggests that if the xenon lamp and ballast response is slow, a slightly larger value (1-5) should be used to improve the response speed; if the noise is high or the target curve changes slowly, a smaller value (0.1-1) should be used. The integral coefficient K... iUsed to accumulate historical errors, eliminate steady-state deviations, and ensure that the system has no residual errors during long-term operation. K i The larger the value, the higher the steady-state accuracy. In practical applications, the value typically ranges from 0.001 to 0.5; the differential coefficient K d Predictive compensation is applied to the rate of error change to suppress rapidly changing disturbances, improve the dynamic characteristics of the system, and reduce overshoot. Its value ranges from 0 to 0.5; the proportional term in this formula... Used to immediately compensate for deviations within the current monitoring period, directly amplifying erroneous signals for output correction, primarily determining the instantaneous gain of the closed-loop response; integral term. Accumulate historical deviations to eliminate steady-state deviations; when errors persist for a long time, the output will be continuously adjusted until the errors are eliminated; differential term Feedforward suppression of the error rate of change is used to reduce overshoot and suppress rapid disturbances. From this, the control action Y(t) corresponding to each time t can be obtained. Similarly, it is packaged as a MODBUS write register command and written to the designated register of the PLC. After receiving the MODBUS command, the PLC converts the digital command into an analog quantity (0-10V or 4-20 mA) to drive the xenon lamp electronic ballast. It can convert the control quantity calculated by PID to eliminate errors into the actual light power output of the xenon lamp without loss and in a timely manner. This allows the output light intensity of the xenon lamp to closely and smoothly track the target solar radiation curve generated by the trigonometric function improved model, realizing real-time accurate illumination simulation of the xenon lamp.

[0068] S104, based on the control actions at each moment, performs an execution quality assessment of the solar illumination simulation to achieve fault diagnosis of solar illumination; specifically:

[0069] The control action is generated to eliminate execution errors. When the control action remains large or changes drastically, it indicates a risk of failure in the solar illumination simulation's actuator. If not addressed promptly, this can lead to distortion of experimental data, poor repeatability, and thermal damage to the optical system. A two-dimensional coordinate system is constructed with time as the x-axis and the control action as the y-axis. The control action at each moment is then input into the coordinate system to form data points. A smooth curve is used to connect these data points, constructing a control action variation curve. The polynomial least squares method is then used to fit the trend of the control action variation curve to obtain the fitted curve. The linear function is used to obtain the instantaneous trend of the control action at each data point by taking the first derivative of the fitted curve function. If the instantaneous trend is greater than zero, it indicates that the control action at this point is increasing; if the instantaneous trend is less than zero, it indicates that the control action at this point is decreasing. The oscillation values ​​of the control action are calculated using the sample standard deviation formula for each instantaneous trend value Q1, Q2...Qk and Qt. The mean value of the control action at each time point is calculated by taking the mean value. Then, the oscillation values ​​and the mean value are normalized by dividing the oscillation value by the historical maximum oscillation value and dividing the mean value by the values ​​at each time point. The maximum value of the control action is determined; then, a sensitivity coefficient is assigned to the normalized oscillation value and the action mean, and the sum of the two sensitivity coefficients is one; in practical applications, the sensitivity coefficients can be adjusted according to the actual situation to reflect different dominant fault modes; then, the normalized oscillation value, action mean, and their corresponding sensitivity coefficients are linearly weighted and fused to obtain the fault diagnosis value; a preset diagnostic threshold is obtained by statistically analyzing the historical data of fault diagnosis values ​​under normal operating conditions to determine their probability distribution characteristics, using the mean plus or minus a certain number of standard deviations (such as μ+3σ) as the threshold benchmark to ensure that low-probability abnormal events occur. It can trigger early warnings in a timely manner; at the same time, it combines physical constraints such as the thermal damage threshold of optical components, the critical value of experimental data distortion, and the safe operating range of actuators for comprehensive calibration, so that the thresholds not only conform to mathematical and statistical laws, but also meet the actual equipment protection requirements. Thus, it can achieve graded early warning and protection before it is about to exceed the normal fluctuation range or approach the physical limit, effectively avoiding systematic deviations in experimental data and damage to equipment hardware; if the fault diagnosis value is greater than the diagnosis threshold, it indicates that there is a relatively serious fault risk, and a fault risk warning is generated, which can respond in a timely and accurate manner, effectively avoiding experimental data distortion and equipment damage, and ensuring the smooth progress of scientific research activities.

[0070] This invention acquires and integrates long-term radiation data from the NASA POWER database and local weather stations, and uses astronomical algorithms to calculate the monthly maximum daily radiation intensity Rmax, typical sunrise time T0, and sunset time T1. It then uses circumferential mapping and K-means clustering to overcome the discontinuity of the solar boundary, obtaining accurate solar motion time parameters that conform to seasonal variations. Based on this, an attenuation model is constructed using the monthly typical values ​​as a benchmark, and random perturbations and physical constraints conforming to meteorological statistics are introduced to generate specific simulated parameters (daily maximum radiation intensity) for each day of the month. Target sunrise time and the target sunset time This establishes a refined diurnal parameter set that reflects both seasonal patterns and natural diurnal fluctuations; furthermore, by using trigonometric functions to improve the model, the daily simulation parameters for the current month are dynamically represented as the target irradiance at each moment. The light intensity is then standardized and mapped into command control quantities. These quantities are then used to drive the xenon lamp electronic ballast via MODBUS communication, achieving accurate reproduction of the natural solar radiation curve. Next, an irradiance sensor detects the actual light intensity and calculates the execution error e(t). A PID control law is used to generate the control action Y(t) in real time, and the xenon lamp power is adjusted through the same communication path to form a closed-loop control, ensuring the actual light intensity closely tracks the target curve. Finally, the oscillation values ​​and mean values ​​of the control action sequence are analyzed, normalized, and weighted to form a fault diagnosis value. An early warning is generated when the statistical threshold is exceeded. Based on these steps, full-link automation is achieved, from real data-driven modeling, accurate dynamic illumination simulation, closed-loop feedback control to intelligent fault diagnosis, improving the realism, data reliability, and system operational safety of the solar simulation experiment.

[0071] To make the implementation effect of the present invention more intuitive and reliable, a comparison table of simulated and actual data of solar solar radiation is attached below:

[0072] Time / interval 5min <![CDATA[Measured radiation intensity (W / m 2 )]]> <![CDATA[Simulated radiation intensity (W / m 2 ).]]> Simulation effect (%) 5-360min 0 0 Sampling time before sunrise 365min 0.8818276 10.09513451 8.74% 370min 5.264287 18.18415352 28.95% 375min 7.395555 22.26094399 33.22% 380min 11.62419 39.31939785 29.56% 390min 20.42815 52.35690306 39.02% 395min 25.82195 60.3237853 42.81% 400min 31.17565 69.2479976 45.02% 405min 36.70785 70.1234937 52.35% 410min 43.51416 74.9442471 58.06% 415min 51.09918 80.7042534 63.32% 420min 58.16346 83.397533 69.74% 425min 65.26714 90.018133 72.50% 430min 72.12791 99.5601303 72.45% 435min 81.85448 101.0176335 81.03% 440min 90.9959 137.3847853 66.23% 445min 107.2163 151.6557651 70.70% 450min 123.5659 178.8247914 69.10% 455min 134.5759 199.8861239 67.33% 460min 144.6293 208.8340658 69.26% 465min 156.3634 217.6629665 71.84% 470min 172.1231 221.3672233 77.75% 475min 186.4147 234.9412844 79.35% 480min 201.6341 248.3796503 81.18% 485min 213.4327 261.6768769 81.56% 490min 222.0088 274.827577 80.78% 495min 240.8073 287.8264228 83.66% 500min 255.6642 300.6681481 85.03% 505min 269.8027 313.3475504 86.10% 510min 284.3677 325.8594927 87.27% 515min 297.326 338.1989062 87.91% 520min 310.5028 350.3607918 88.62% 525min 323.2153 362.3402222 89.20% 530min 337.2523 374.1323443 90.14% 535min 354.4985 385.7323807 91.90% 540min 368.9542 397.1356319 92.90% 545min 380.0193 408.3374781 93.07% 550min 389.9156 419.333381 92.98% 555min 400.6238 430.1188858 93.14% 560min 475.651 480.6896231 98.95% 565min 484.3669 511.0413103 94.78% 570min 506.2152 521.1697536 97.13% 575min 527.2589 541.0708496 97.45% 580min 548.58 560.7405872 97.83% 585min 563.4273 580.1750487 97.11% 590min 575.0119 599.3704121 95.94% 595min 589.8248 608.322952 96.96% 600min 595.7291 617.0290414 96.55% 605min 607.237 625.4851532 97.08% 610min 614.75 633.6878616 97.01% 615min 626.1964 641.6338435 97.59% 620min 637.4988 649.3198796 98.18% 625min 644.7152 656.7428561 98.17% 630min 652.3604 663.8997658 98.26% 635min 664.0094 670.7877091 98.99% 640min 670.9749 677.4038956 99.05% 645min 679.6785 683.7456447 99.41% 650min 681.8 689.8103873 98.84% 655min 690.9651 695.5956664 99.33% 660min 695.1105 701.0991381 99.15% 665min 698.8935 706.3185729 98.95% 670min 701.6761 711.2518561 98.65% 675min 711.9899 715.8969892 99.45% 680min 718.677 720.2520903 99.78% 685min 720.5217 724.3153949 99.48% 690min 722.3331 728.0852569 99.21% 695min 729.4377 731.5601491 99.71% 700min 733.2491 734.7386635 99.80% 705min 736.1165 737.6195126 99.80% 710min 739.249 740.2015292 99.87% 715min 741.8344 742.4836673 99.91% 720min 741.9771 744.4650022 99.67% 725min 743.4777 746.1447313 99.64% 730min 747.2216 747.522174 99.96% 735min 744.4856 748.5967724 99.45% 740min 745.3214 749.368091 99.46% 745min 747.7689 749.8358175 99.72% 750min 749.4857 749.9997622 99.93% 755min 748.9832 749.8598588 99.88% 760min 733.4668 749.4161639 97.87% 765min 735.8196 748.6688574 98.28% 770min 731.2213 747.6182419 97.81% 775min 729.382 746.2647431 97.74% 780min 734.6788 744.6089094 98.67% 785min 731.0865 742.6514116 98.44% 790min 727.9111 740.3930426 98.31% 795min 716.8683 737.8347175 97.16% 800min 715.9949 734.9774728 97.42% 805min 716.335 731.8224658 97.88% 810min 721.6816 728.370975 99.08% 815min 705.2575 724.6243985 97.33% 820min 709.7002 720.5842543 98.49% 825min 710.575 716.252179 99.21% 830min 703.1027 711.6299278 98.80% 835min 689.6805 706.7193733 97.59% 840min 690.4987 701.5225049 98.43% 845min 699.6124 696.041428 99.49% 850min 686.6312 690.2783632 99.47% 855min 686.4015 684.2356452 99.68% 860min 675.9415 677.9157222 99.71% 865min 657.4136 671.3211545 97.93% 870min 653.047 664.4546139 98.28% 875min 650.6749 657.3188821 98.99% 880min 636.0945 649.91685 97.87% 885min 630.6295 643.2515166 98.04% 890min 628.3803 634.3259871 99.06% 895min 619.042 626.1434726 98.87% 900min 612.0646 617.707288 99.09% 905min 604.8569 609.020851 99.32% 910min 593.2555 600.0876808 98.86% 915min 589.1055 590.9113965 99.69% 920min 577.5638 581.4957157 99.32% 925min 562.9739 571.844453 98.45% 930min 554.986 561.9615183 98.76% 935min 545.2114 551.8509157 98.80% 940min 538.6735 541.5167412 99.47% 945min 522.3069 530.9631815 98.37% 950min 510.1374 520.1945121 98.07% 955min 489.8172 509.2150959 96.19% 960min 485.6834 498.0293808 97.52% 965min 460.9991 486.6418987 94.73% 970min 454.2957 475.0572628 95.63% 975min 443.2816 463.2801665 95.68% 980min 429.9996 451.3153811 95.28% 985min 406.5219 439.1677539 92.57% 990min 389.6398 426.8422061 91.28% 995min 379.6959 414.3437314 91.64% 1000min 367.4448 401.6773933 91.48% 1005min 359.9949 388.8483231 92.58% 1010min 337.4391 370.8617185 90.99% 1015min 328.7281 359.7228407 91.38% 1020min 307.0397 338.4370126 90.72% 1025min 293.10375 322.0096167 91.02% 1030min 271.60387 308.446093 88.06% 1035min 252.45498 284.7519363 88.66% 1040min 239.99441 271.9326947 88.26% 1045min 221.11203 260.9939667 84.72% 1050min 204.6263 238.9413994 85.64% 1055min 189.16579 215.7806858 87.67% 1060min 176.60393 203.517563 86.78% 1065min 166.46954 195.1578093 85.30% 1070min 152.40623 180.7072423 84.34% 1075min 124.29326 165.1717164 75.25% 1080min 114.59407 151.5571205 75.61% 1085min 101.91125 136.8693752 74.46% 1090min 83.7346 106.1144311 78.91% 1095min 73.67236 95.2982659 77.31% 1100min 64.21322 91.4268821 70.23% 1105min 45.00727 79.5063044 56.61% 1110min 34.66937 70.54257779 49.15% 1115min 30.28152 66.54176442 45.51% 1120min 26.64048 58.5099416 45.53% 1125min 22.74993 46.4531992 48.97% 1130min 21.58487 31.37763717 68.79% 1135min 18.82526 16.28936309 84.43% 1140min 14.21039 10.194489687 60.61% 1145min 9.317442 6.58763515 58.56% 1150min 4.931083 3.10293 41.08% 1155min 3.33868 0 / 1160min 2.530168 0 / 1165-1440min 0 0 Sampling time after sunset

[0073] As shown in the comparison table of simulated and actual data above, in the actual implementation of this scheme, 360 minutes is the sunrise time and 1140 minutes is the sunset time. The complete diurnal solar radiation change process from sunrise to sunset is simulated at a diurnal scale with a sampling time of 5 minutes. The specific fitting effect is calculated as: 1 - |difference between actual radiation intensity and simulated radiation intensity| / actual radiation intensity. According to the comparison of simulation effects in the experimental data table, this scheme shows high fitting accuracy in simulating diurnal solar radiation changes. Overall, the simulated radiation intensity and the measured radiation intensity gradually become consistent after sunrise. The simulation effect improves rapidly from a low level at the beginning of sunrise (e.g., 8.74%) and remains above 90% during the main daytime period (approximately 500–800 minutes), even approaching 99% at some times, indicating that the model has excellent predictive ability in the medium to high radiation intensity range. Before and after sunset, although the simulation effect decreases, it still generally remains between 70% and 90%, with the fitting effect only slightly lower during periods of extremely low radiation intensity. Overall, the simulation method can reflect the dynamic trend of solar radiation throughout the day well, especially the midday period when radiation is strong, which shows a significant fitting effect and verifies the reliability and practicality of the model in daytime radiation simulation.

[0074] like Figure 2 As shown, this embodiment of the invention provides an intelligent and controllable system for simulating the dynamic changes of solar radiation. The system specifically includes: a parameter analysis module, a model characterization module, a control adjustment module, and a fault diagnosis module.

[0075] The parameter analysis module obtains hourly horizontal irradiance data for the target area for at least one consecutive year from the NASA POWER meteorological database, and verifies and corrects it by integrating measured data from local weather stations. Based on the verified and corrected data, the daily total radiation intensity is calculated on a daily basis, and then divided by month to obtain the daily total radiation intensity for each month. The maximum daily radiation intensity Rmax is selected for each month. Simultaneously, the typical sunrise time T0 and typical sunset time T1 for the target area in each month are calculated based on astronomical algorithms. Based on this, diurnal parameter attenuation and random allocation are performed to obtain the daily simulated parameters (daily maximum radiation intensity) for each month. Target sunrise time and the target sunset time The model representation module is used to obtain the month to which the target region belongs at the current time, match the daily simulation parameters of the corresponding month, and use trigonometric functions to improve the model for dynamic representation to obtain the target irradiance of the target region at each time t of each day d. This will be converted into a command control quantity to drive the xenon lamp electronic ballast to adjust the xenon lamp power in real time; the specific dynamic characterization formula is as follows: κ is the shape parameter, Sc(t) is the atmospheric correction factor, Sc(t)∈(0,1]; t is the current time of the target area;

[0076] The control and adjustment module is used to place an irradiation sensor at the output end of the optical path to measure the actual light intensity and convert it into an actual control quantity, calculate the execution error, perform PID control on the execution error to obtain the control action quantity, and adjust the xenon lamp power according to the control action quantity to form a closed-loop control.

[0077] The fault diagnosis module performs execution quality assessment and fault diagnosis based on the control action at each moment.

[0078] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0079] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. An intelligent control method for simulating dynamic changes in solar radiation, characterized in that, Includes the following steps: S101: Obtain hourly horizontal irradiance data for the target area for at least one consecutive year from a meteorological database, and verify and correct it by integrating measured data from local meteorological stations. Based on the verified and corrected data, calculate the daily total radiation intensity on a daily basis, divide it by month to obtain the daily total radiation intensity for each month, and select the maximum daily radiation intensity Rmax for each month. At the same time, calculate the typical sunrise time T0 and typical sunset time T1 for the target area in each month based on astronomical algorithms. Based on Rmax, T0, and T1, construct an attenuation model with the monthly typical values ​​as the benchmark, and then introduce random disturbances and physical constraints that conform to meteorological statistical characteristics to generate daily simulation parameters for the month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time , where d is the index of the simulated number of days in the month; S102, Obtain the current month of the target region and match the daily simulation parameters for the corresponding month, i.e., the maximum daily solar radiation intensity. Target sunrise time and the target sunset time The target irradiance intensity at each time t in the target area is obtained by dynamically representing it using a trigonometric function improved model. This is then converted into command control quantities to drive the xenon lamp electronic ballast to adjust the xenon lamp power in real time. S103, an irradiation sensor is arranged at the output end of the optical path to measure the actual light intensity and convert it into an actual control quantity. The execution error is calculated, and PID control is performed on the execution error to obtain the control action quantity. The xenon lamp power is adjusted according to the control action quantity to form a closed-loop control. S104 performs execution quality assessment and fault diagnosis based on the control action at each moment.

2. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 1, characterized in that, Typical sunrise time T0 and typical sunset time T1 for the target area each month: The sunrise and sunset times for the target region are calculated based on astronomical algorithms for each day of each month. The 24 hours of a day are divided into several equal-length periods. The sunrise time of each day is then assigned to the corresponding equal-length periods. The number of sunrise times in each equal-length period is counted cumulatively, and the period with the most sunrise times is selected as the typical sunrise period. Similarly, the same method is used to determine typical sunset times; Each sunrise moment in the typical sunrise period is converted into a minute of the day as a data point. Then, a circular mapping and two-dimensional vector space clustering are introduced to process the data points of each day within a month to obtain the typical sunrise moment T0. Similarly, the same processing is performed on all sunset moment samples in the typical sunset period to obtain the typical sunset moment T1.

3. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 2, characterized in that, Introducing circumferential mapping: Convert each sunrise time in a typical sunrise period into its corresponding minutes, denoted as Z. Let T be the period constant on a diurnal scale. Map the minutes corresponding to each sunrise time to an angle θ using the following formula: Each sunrise time corresponds to an angle, forming an angle set. As can be seen from the mapping formula, the angle θ∈[0,2π). Each angle θ is mapped to a two-dimensional vector coordinate (cosθ, sinθ) using cosine and sine. Thus, each sunrise time in a typical sunrise period can be transformed into a two-dimensional vector coordinate.

4. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 3, characterized in that, Two-dimensional vector space clustering: The K-means clustering algorithm is adopted, and the number of clusters K=1 is set. A point is randomly selected from the two-dimensional vector coordinate dataset as the initial cluster center μ1. The algorithm enters the iterative process to optimize the position of the cluster center. The mean of all data points in cluster C1 is calculated and the position of cluster center μ1 is updated accordingly. The mean calculation is performed in the two-dimensional coordinate system, that is, the X coordinate of the new cluster center = the arithmetic mean of the X coordinates of all data points, and the Y coordinate of the new cluster center = the arithmetic mean of the Y coordinates of all data points. This update step is repeated until the change in the position of cluster center μ1 is less than the preset convergence threshold or the maximum number of iterations is reached. At this time, the algorithm determines that the cluster center has converged. The coordinates of the final cluster centers after convergence are converted back to time by the reverse process of converting time into coordinates, resulting in the typical sunrise time T0.

5. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 4, characterized in that, The process of converting the coordinates of the final cluster centers after convergence back to time is the reverse of the time-based coordinate conversion: The coordinates of the final cluster centers after convergence Calculate its angle θ using the arctangent function in the four quadrants. center : The atan(*) function ensures that the angle θ of the final cluster centers is obtained. center Located in the correct quadrant, its domain is (-π, π], if θ center If θ < 0, then let θ center =θ center +2π; angle θ center Convert to minutes of a day Z center Then adjust the angle θ center The corresponding number of minutes Z center This is converted into time, which gives us the typical sunrise time T0.

6. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 5, characterized in that, Based on Rmax, T0, and T1, a decay model with monthly typical values ​​is constructed. Random perturbations and physical constraints conforming to meteorological statistical characteristics are introduced to generate daily simulation parameters within the month: 1-5-1, Defined by sunrise and sunset times as boundaries: Extract the earliest sunrise and latest sunrise times, as well as the earliest sunset and latest sunset times for the current month, and denot them as follows: and ,as well as and ; Define the radiation intensity boundary: Obtain the minimum daily radiation intensity for the month from historical meteorological data. And the maximum solar radiation intensity of the month ; Constructing sunrise and sunset time baselines: Using the typical sunrise time T0 and typical sunset time T1 of the month as the baseline values ​​for the month, construct a baseline. The sunrise time baseline for the month is thus obtained and denoted as... The baseline for sunset is denoted as Where j represents the index of the actual number of days in the monthly baseline; similarly, construct the maximum daily radiation intensity baseline: using the typical monthly value Rmax as the mid-month baseline, construct a curve reflecting the monthly variation in radiation intensity: using the beginning and end of the month... and Using Rmax as the starting and ending point, and the peak value in the middle of the month as the reference point, a smooth curve is generated as the baseline, denoted as . ; 1-5-2, add random perturbations and physical constraints to ensure that the daily parameters generated by the above attenuation and perturbation model are strictly within their physical possible boundaries, thereby obtaining the constrained sunrise time sequence, sunset time sequence and maximum solar radiation intensity sequence; 1-5-3, Randomization: To simulate the nondeterministic correspondence between actual meteorological conditions and dates, the constrained sunrise time sequence, sunset time sequence, and maximum daily radiation intensity sequence are randomly rearranged to generate a random permutation from 1 to D, which serves as the new date order; where D is a positive integer representing the total number of actual or simulated days in the month; then, the reordered sequence is used as the daily simulation parameter for the month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time ), where d∈D, and d represents the simulated day index of the current month.

7. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 6, characterized in that, The control action is obtained by applying PID control to the execution error: An irradiation sensor is placed at the output end of the optical path to measure the actual output light intensity and convert it into an actual control quantity. The actual control quantity is then subtracted from the command control quantity to obtain the execution error e(t). Finally, a PID control law is applied to the execution error e(t) to calculate the real-time control action Y(t). The calculus calculation formula is as follows: e(t) represents the execution error at the current moment. K represents the cumulative total of all execution errors in the past, k represents the time before the current time t, and Ts is the sampling period; p K i and K d These are the proportional coefficient, integral coefficient, and differential coefficient, respectively.

8. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 7, characterized in that, Execution quality assessment and fault diagnosis are performed based on the control action at each moment: A two-dimensional coordinate system is constructed with time as the horizontal axis and the control action quantity as the vertical axis. Then, the control action quantity at each moment is input into the coordinate system to form data points. The data points are connected sequentially with a smooth curve to construct the control action quantity change curve. The polynomial least squares method is used to perform trend fitting on the control action quantity change curve to obtain the fitted curve function. The first derivative of the fitted curve function is obtained to obtain the instantaneous change trend degree of the control action quantity at each data point. Fault diagnosis is performed based on the change curve of the control action and its instantaneous change trend to make early warning of fault risk.

9. The intelligent and controllable method for simulating dynamic changes in solar daily radiation according to claim 8, characterized in that, Fault diagnosis based on the change curve of the control action and its instantaneous change trend: The oscillation value of the control action is obtained by calculating the instantaneous change trend degree using the sample standard deviation formula. The mean value of the control action at each time moment is calculated by averaging the control action. The oscillation value and the mean value are then normalized. A sensitivity coefficient is assigned to the normalized oscillation value and the mean value, and the sum of the two sensitivity coefficients is one. Finally, the normalized oscillation value, the mean value of action, and the corresponding sensitivity coefficient are linearly weighted and fused to obtain the fault diagnosis value. A preset diagnostic threshold is set. If the fault diagnosis value is greater than the diagnostic threshold, a fault risk warning is generated.

10. An intelligent and controllable system for simulating dynamic changes in solar radiation, characterized in that... An intelligent and controllable method for simulating dynamic changes in solar radiation as described in any one of claims 1-9, comprising: a parameter analysis module, a model characterization module, a control adjustment module, and a fault diagnosis module; The parameter analysis module obtains hourly horizontal irradiance data for the target area for at least one consecutive year from the meteorological database, and verifies and corrects it by integrating measured data from the local meteorological station. Based on the verified and corrected data, the total daily radiation intensity is calculated on a daily basis, and the total daily radiation intensity for each month is obtained by dividing it by month. The maximum daily radiation intensity Rmax is selected in each month. At the same time, the typical sunrise time T0 and typical sunset time T1 for the target area in each month are calculated based on astronomical algorithms. The model representation module obtains the current month of the target region and matches it with the daily simulation parameters for that month, i.e., the daily maximum radiation intensity. Target sunrise time and the target sunset time The target irradiance intensity at each time t in the target area is obtained by dynamically representing it using a trigonometric function improved model. This will be converted into a command control quantity to drive the xenon lamp electronic ballast to adjust the xenon lamp power in real time; The control adjustment module is used to place an irradiation sensor at the output end of the optical path to measure the actual light intensity and convert it into an actual control quantity, calculate the execution error, perform PID control on the execution error to obtain the control action quantity, and adjust the xenon lamp power according to the control action quantity to form a closed-loop control. The fault diagnosis module performs execution quality assessment and fault diagnosis based on the control action at each moment.

Images