Solar cell maximum power point adaptive tracking method, system and storage medium

Abstract

This invention discloses an adaptive maximum power point tracking method, system, and storage medium for solar cells, belonging to the field of photovoltaic power generation technology. The method includes: extracting mechanical correction data and temperature regulation data of the solar cell for the current day according to a preset time slot, and constructing a multi-dimensional feature vector; combining the multi-dimensional feature vector and the corresponding recorded electrical operation data to form the daily operation record; based on the daily operation record, clustering and grouping the multi-dimensional feature vector by fusing the feature distance and time-series distance to obtain multiple control intervals; calculating corresponding control parameters based on the multi-dimensional feature vector and electrical operation data respectively; and upon arrival of the next day, reading the corresponding control parameters according to the control interval to which the current time belongs to complete the tracking of the solar cell's maximum power point. This invention can improve the accuracy of maximum power point tracking.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power generation technology, specifically relating to a method, system, and storage medium for adaptive maximum power point tracking of solar cells. Background Technology

[0002] The output characteristics of solar cells are affected by light intensity and operating temperature, and their voltage-current characteristic curves have a maximum power point. To ensure that solar cells always operate at maximum power output to improve photoelectric conversion efficiency, a DC-DC converter is usually placed between the solar cell and the load. By adjusting the duty cycle of the DC-DC converter, the operating voltage of the solar cell is changed, allowing it to operate at the maximum power point under the current environmental conditions.

[0003] Currently, the most widely used maximum power point tracking (MPPT) method in engineering is the perturbation-observation method. Its basic principle is to apply a fixed perturbation increment to the duty cycle of the DC-DC converter in each control cycle, and determine the direction of duty cycle adjustment by comparing the output power change before and after the perturbation. If the power increases, the perturbation continues in the current direction; if the power decreases, the perturbation direction is reversed. However, in practical applications, the perturbation-observation method uses a fixed perturbation step size and a fixed detection period. If the perturbation step size is set too large, the duty cycle oscillates around the maximum power point, leading to increased steady-state power loss. If the perturbation step size is set too small, the tracking speed is insufficient to keep up with the drift speed of the maximum power point when illumination conditions change rapidly, resulting in increased dynamic power loss. These problems result in low MPPT tracking accuracy in existing technologies. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a method, system, and storage medium for adaptive maximum power point tracking of solar cells, thereby resolving the issues present in the background art.

[0005] To achieve the aforementioned objective, this invention proposes an adaptive maximum power point tracking method for solar cells, comprising: According to the preset time slot, extract the mechanical correction data and temperature regulation data of the solar cell for the day, construct a multi-dimensional feature vector, and combine the multi-dimensional feature vector and the corresponding recorded electrical operation data into the operation record for the day; Based on the daily operation records, clustering and grouping are performed by fusing the feature distance and temporal distance of multidimensional feature vectors. Time slots belonging to the same cluster and with adjacent temporal sequences are merged, and time slots with discontinuous temporal sequences are split or migrated to obtain multiple control intervals. For each control interval, the corresponding control parameters are calculated based on the multi-dimensional feature vector and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters. Based on the control parameters of the previous day and the current day, the control parameters of each control interval for the next operating day are calculated. When the next day arrives, the corresponding control parameters are read according to the control interval to which the current time belongs. Based on the mechanical tracking parameters, the spot position is corrected and the temperature is adjusted. An adjustment locking window mechanism is used to shield interference. The DC-DC converter is adjusted based on electrical tracking parameters to achieve maximum power point tracking of the solar cell.

[0006] The present invention also provides a solar cell maximum power point adaptive tracking system for implementing the above-described method, the system comprising: The data construction module extracts the mechanical correction data and temperature regulation data of the solar cells for the day according to the preset time slots, and constructs a multi-dimensional feature vector. The multi-dimensional feature vector and the corresponding recorded electrical operation data are combined into the operation record for the day. The clustering and partitioning module, based on the daily operation records, integrates the feature distance and temporal distance of multi-dimensional feature vectors to perform clustering and grouping, merges time slots that belong to the same cluster and are temporally adjacent, and splits or migrates time slots that are not temporally discontinuous to obtain multiple control intervals. The parameter calculation module calculates the corresponding control parameters for each control interval based on multi-dimensional feature vectors and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters. The tracking execution module calculates the control parameters for each control interval of the next operating day based on the control parameters of the previous day and the current day. When the next day arrives, it reads the corresponding control parameters according to the control interval to which the current time belongs, corrects the spot position and adjusts the temperature based on the mechanical tracking parameters, and uses an adjustment locking window mechanism to shield interference. Based on the electrical tracking parameters, it adjusts the DC-DC converter to complete the tracking of the maximum power point of the solar cell.

[0007] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the method described above.

[0008] The beneficial effects of this invention are as follows: This invention constructs a multi-dimensional feature vector and combines it with corresponding recorded electrical operation data to form a daily operation record. Then, based on this daily operation record, it performs clustering by fusing the feature distance and temporal distance of the multi-dimensional feature vectors. Time slots belonging to the same cluster and with adjacent time sequences are merged, while time slots with discontinuous time sequences are split or migrated, resulting in multiple control intervals that precisely match the actual operating conditions. Subsequently, for each control interval, corresponding control parameters containing mechanical and electrical tracking parameters are calculated. Based on the control parameters of the previous day and the current day, the control parameters for each control interval of the next operating day are calculated. Upon arrival of the next day, the corresponding control parameters are read according to the control interval to which the current time belongs. Based on the mechanical tracking parameters, spot position correction and temperature adjustment are performed. Finally, the DC-DC converter is adjusted based on the electrical tracking parameters, thereby handling the mutual interference between mechanical actions and electrical adjustments and achieving adaptive and precise tracking of the maximum power point of the solar cell. This invention enables adaptive tracking of the maximum power point. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the steps of an adaptive maximum power point tracking method for solar cells according to the present invention. Figure 2 This is a schematic diagram of the overall process of the present invention; Figure 3 This is a schematic diagram of the structure of the concentrated photovoltaic system of the present invention; Figure 4 This is a schematic diagram of the structure of a solar cell maximum power point adaptive tracking system according to the present invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0011] like Figure 1 and Figure 2 As shown, a solar cell maximum power point adaptive tracking method includes: S1: Extract the mechanical correction data and temperature regulation data of the solar cell for the day according to the preset time slot, construct a multi-dimensional feature vector, and combine the multi-dimensional feature vector and the corresponding recorded electrical operation data into the operation record for the day.

[0012] Before introducing the technical solution of this invention, the concentrated photovoltaic system used in this application will first be described, such as... Figure 3As shown, the concentrated photovoltaic system includes a convex lens, a solar panel, a two-axis tracking platform, a light spot adjustment mechanism, a temperature sensor (not shown), a photosensor module (not shown), a DC-DC converter, and a lithium battery pack. Sunlight is focused by the convex lens to form a light spot that illuminates the solar panel. The two-axis tracking platform includes an azimuth motor and a pitch motor. The azimuth motor controls the horizontal rotation of the solar panel, and the pitch motor controls the vertical rotation. The two-axis tracking platform drives the solar panel through the azimuth and pitch motors to keep the light spot centered on the panel. The light spot adjustment mechanism adjusts the distance between the solar panel and the lens via a lead screw stepper motor, changing the light spot size to control the panel temperature. The temperature sensor and the photosensor module collect data on the battery surface temperature and the light spot position, respectively. The solar panel converts the focused light energy into direct current, which is then converted into a stable voltage by the DC-DC converter to charge the lithium battery. The controller is used to control these actions.

[0013] Within each operating day of the concentrated photovoltaic system, the preset sunshine period is divided into N time slots according to the preset time slot division method. In each time slot, the mechanical correction data, temperature regulation data, and electrical operation data of the concentrated photovoltaic system are recorded. Among them, the mechanical correction data includes the number of times the azimuth motor is started and the total operating time, which are defined here as the number of azimuth corrections and the cumulative azimuth operating time, respectively; the number of times the pitch motor is started and the total operating time, which are defined here as the number of pitch corrections and the cumulative pitch operating time, respectively; and the cumulative number of times the azimuth and pitch dimensions continuously perform correction actions until the light spot is corrected during each correction, which are defined as the number of consecutive azimuth corrections and the number of consecutive pitch corrections, respectively.

[0014] Temperature regulation data includes the number of temperature adjustments and the average temperature value. The number of temperature adjustments is determined by counting the number of times the lead screw stepper motor in the solar panel lifting and adjusting device is started, and the average temperature value within the time slot is obtained by the solar cell surface temperature sensor.

[0015] Within each time slot, the output voltage and output current of the solar cell are collected to calculate the average output power within the time slot. The duty cycle of the DC-DC converter when the output power reaches its maximum value within the time slot is recorded as the optimal duty cycle of the time slot. The average output power and the optimal duty cycle are used as electrical data.

[0016] The azimuth correction frequency is calculated based on the number of azimuth corrections and the duration of the time slot; the pitch correction frequency is calculated based on the number of pitch corrections and the duration of the time slot; and the temperature adjustment frequency is calculated based on the number of temperature adjustments and the duration of the time slot. The azimuth correction frequency, pitch correction frequency, temperature adjustment frequency, and average temperature value are combined to form a four-dimensional feature vector, which serves as the multi-dimensional feature vector for the time slot. The multi-dimensional feature vector is combined with electrical operation data to form operation record entries. All operation record entries for the time slots together constitute the daily operation record.

[0017] S2: Based on the daily operation records, clustering is performed by fusing the feature distance and time distance of multi-dimensional feature vectors. Time slots belonging to the same cluster and with adjacent time sequences are merged, and time slots with discontinuous time sequences are split or migrated to obtain multiple control intervals.

[0018] Based on the daily operation records, the multidimensional feature vector of each time slot is extracted and normalized. An improved K-Means clustering algorithm is used to cluster all time slots based on the multidimensional feature vectors. Based on the clustering results, the time slots are merged or split to obtain multiple control intervals.

[0019] S3: For each control interval, calculate the corresponding control parameters based on the multi-dimensional feature vector and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters.

[0020] Mechanical tracking parameters include detection interval, correction step size, and dimension priority mode, while electrical tracking parameters include initial duty cycle value and historical average output power.

[0021] S4: Calculate the control parameters for each control interval of the next operating day based on the control parameters of the previous day and the current day. When the next day arrives, read the corresponding control parameters according to the control interval to which the current time belongs. Based on the mechanical tracking parameters, perform spot position correction and temperature adjustment, and use the adjustment locking window mechanism to shield interference.

[0022] S5: Adjust the DC-DC converter based on electrical tracking parameters to achieve maximum power point tracking of the solar cell.

[0023] Because daily data may be affected by accidental factors and produce outliers, merging the control parameters from the previous day and the current day can smooth out short-term disturbances. Based on the foregoing, the mechanical tracking parameters include dimension priority modes. If the dimension priority mode is azimuth priority mode, then in each detection cycle, the photosensitive detection module first detects the offset of the light spot in the azimuth dimension. If an azimuth offset is detected, the azimuth motor is driven to perform azimuth correction according to the correction step size. After the azimuth motor stops running, a preset stabilization time is waited, and then the photosensitive detection module detects the offset of the light spot in the pitch dimension and performs pitch correction. If the dimension priority mode is pitch priority mode, the process is reversed. If the dimension priority mode is balanced mode, in odd-numbered detection cycles, azimuth detection and correction are performed first, followed by pitch detection and correction; in even-numbered detection cycles, pitch detection and correction are performed first, followed by azimuth detection and correction.

[0024] After position correction is completed, the current temperature value of the solar panel surface temperature sensor is read and compared with a preset upper temperature threshold and a lower temperature threshold. If the current temperature value exceeds the upper temperature threshold, the motor is driven to raise the solar panel by a preset lifting / lowering step. If the current temperature value is lower than the lower temperature threshold and the current lifting / lowering pulse count of the solar panel is not zero, the motor is driven to lower the solar panel back to the focal position by the lifting / lowering step. If the current temperature value is between the lower temperature threshold and the upper temperature threshold, no temperature adjustment is performed. This step avoids frequent adjustments to the solar panel when there are slight temperature fluctuations.

[0025] On the next operating day, the corresponding mechanical and electrical tracking parameters are read according to the control interval at the current time. Spot position detection and correction, as well as temperature adjustment, are performed according to the mechanical tracking parameters. Position correction and temperature adjustment are executed sequentially in a fixed order, and this adjustment process is called the mechanical adjustment sequence. After the entire mechanical adjustment sequence is completed, if at least one of the position correction or temperature adjustment actions has been performed, an adjustment lock window is activated. During the duration of the adjustment lock window, detection, correction, and duty cycle adjustment actions of other loops are prohibited. Normal operation resumes after the adjustment lock window ends, and any corrections caused by the previous action after the adjustment lock window ends are marked as passive corrections. After both position correction and temperature adjustment are completed and the system is in a steady state, the duty cycle of the DC-DC converter is adjusted according to the electrical tracking parameters, starting from the initial duty cycle value and using the duty cycle perturbation step size as the adjustment amount, to track the maximum power point of the solar cell under the current illumination and temperature conditions.

[0026] In this embodiment, clustering and grouping are performed by fusing the feature distance and temporal distance of multidimensional feature vectors, including: Clustering is performed based on the K-Means clustering algorithm, and time slots are numbered according to time order. In the allocation stage of the clustering algorithm, the Euclidean distance between the multidimensional feature vector of the time slot and the multidimensional feature vector of the centroid of the target cluster is calculated as the feature distance.

[0027] Obtain the temporal center of the target cluster, and calculate the temporal distance based on the time slot number and the temporal center number. In the initial allocation stage, the temporal center is the time slot number corresponding to the initial centroid, and in the iterative update stage, it is the average of the allocated time slot numbers within the target cluster.

[0028] The composite distance is obtained by weighted summation of the feature distance and the temporal distance. Based on the composite distance, the time slots are assigned to the nearest clusters and the centroids are updated until the convergence condition is met to obtain the final clustering result.

[0029] In the final clustering results, consecutive time slots within each cluster are merged into a control interval.

[0030] This embodiment uses the K-Means clustering algorithm for clustering. During clustering, a composite distance is used, comprising feature distance and temporal distance. The feature distance is the Euclidean distance between the multidimensional feature vector of the time slot and the multidimensional feature vector of the target cluster centroid. When calculating the temporal distance, time slots are numbered chronologically. The absolute value of the difference between the time slot number and the time center number of the target cluster is calculated as the temporal distance. The composite distance is the product of the feature distance and a first weight, plus the product of the temporal distance and a second weight, where the first weight is greater than the second weight to ensure that the feature distance dominates the composite distance. The temporal distance serves only as an auxiliary constraint to prioritize clustering temporally adjacent time slots into the same cluster when feature distances are close. Specifically, the time center in the initial allocation phase is the time slot number corresponding to the initial centroid.

[0031] During clustering, all time slots are arranged according to their time slot numbers, and the multidimensional feature vector of the time slot with the smallest number is selected as the first initial centroid. Subsequently, the minimum composite distance between each of the remaining time slots and all the selected initial centroids is calculated, and the multidimensional feature vector of the time slot with the largest minimum value is selected as the next initial centroid. This process is repeated until a predetermined number of initial centroids are selected, and each time slot is assigned to the cluster containing the centroid with the closest composite distance.

[0032] In the centroid update phase of the clustering algorithm, the arithmetic mean of the multidimensional feature vectors of all time slots within a cluster is calculated, and this arithmetic mean is used as the updated multidimensional feature vector of the centroid. Simultaneously, the arithmetic mean of the time slot indices of all time slots within a cluster is calculated, and this arithmetic mean is used as the updated time series center. This allocation and update process is repeated until the convergence condition is met, yielding the clustering result corresponding to the current number of clusters.

[0033] The number of clusters is changed to obtain multiple clustering results. For each clustering result, the average feature distance from the multidimensional feature vectors of all time slots in each cluster to the cluster centroid is calculated as the intra-cluster compactness. The average feature distance between the centroids of all clusters is calculated as the inter-cluster separation. The ratio of inter-cluster separation to intra-cluster compactness is used as the clustering evaluation index. The clustering result corresponding to the largest clustering evaluation index is selected as the final clustering result.

[0034] In this embodiment, time slots belonging to the same cluster and with adjacent time sequences are merged, and time slots with discontinuous time sequences are split or migrated, including: The time slots within the target cluster are sorted according to time sequence. The time slot sequences with discontinuous time sequence and the breakpoints are identified. The first feature difference between the time slots on both sides of the breakpoint and the second feature difference between the missing time slot at the breakpoint and the feature reference value of the target cluster are calculated. If both the first feature difference and the second feature difference meet the preset compatibility conditions, the missing time slot is migrated to the target cluster to fill the breakpoint. Otherwise, the target cluster is split into multiple sub-clusters at the breakpoint.

[0035] Time slots that belong to the same cluster and are temporally adjacent, target clusters that have not been split and have completed migration processing, or sub-clusters generated after splitting, with temporally consecutive time slots within them are merged into control intervals.

[0036] In the final clustering result, time slots within each cluster are sorted by their time slot numbers. If the time slot numbers are discontinuous, the Euclidean distance between the multidimensional feature vectors of adjacent time slots on both sides of the discontinuity breakpoint is calculated, and this Euclidean distance is used as the first feature difference between the time slots on both sides of the breakpoint. If the Euclidean distance is less than a preset merging feature distance threshold, the Euclidean distance between the multidimensional feature vector of each missing time slot at the breakpoint and the multidimensional feature vector of the current centroid of the target cluster is calculated, and this Euclidean distance is used as the second feature difference between the missing time slot at the breakpoint and the feature reference value of the target cluster. If the Euclidean distances between all missing time slots at the breakpoint and the centroid of the target cluster are less than a preset migration compatibility distance threshold, that is, if both the first and second feature differences satisfy the preset compatibility conditions, then the missing time slot at the breakpoint is migrated from its current cluster to this cluster, and the centroid of the target cluster is recalculated. Otherwise, no migration operation is performed, and the cluster is split into multiple sub-clusters at the breakpoint. If the Euclidean distance between adjacent time slots on both sides of the breakpoint is greater than or equal to the merging feature distance threshold, then the cluster is split into multiple sub-clusters at the breakpoint.

[0037] Finally, time slots that belong to the same cluster and are temporally adjacent, target clusters that have not been split and have completed migration processing, or sub-clusters generated after splitting, with temporally consecutive time slots within them, are merged into control intervals.

[0038] In this embodiment, the corresponding mechanical tracking parameters and electrical tracking parameters are calculated based on the multidimensional feature vector and electrical operation data, respectively, including: The mechanical tracking parameters include detection interval, correction step size, and dimension priority mode. The detection interval is calculated based on the position correction frequency and pitch correction frequency in the multi-dimensional feature vector, and the detection interval of adjacent control intervals is smoothly corrected.

[0039] The detection interval defines the time interval at which the photosensitive detection module detects the position offset of the light spot. In other words, it determines how often the light spot is checked for offset. When a offset is detected, the drive motor needs to correct the position. The correction step size is the amount of drive the motor performs in a single operation during each correction action, which determines the adjustment range of the light spot position in each correction action. Since the correction includes two independent correction axes, azimuth and pitch, both dimensions need to be detected and corrected separately in each detection cycle. The dimension priority mode is used to determine the order of correction execution for the azimuth and pitch dimensions within the same detection cycle. When the correction requirement for one dimension in historical operation is higher than that for another dimension, prioritizing the correction of that dimension can eliminate the main offset more quickly and reduce cross-interference and cumulative deviation caused by improper correction order.

[0040] When calculating the detection interval, the active correction frequency is first calculated based on the azimuth correction frequency and the pitch correction frequency. During calculation, the azimuth correction frequency and pitch correction frequency of all time slots within the control interval are extracted, and the average values ​​of the azimuth correction frequency and pitch correction frequency are calculated separately. These two average values ​​are then added together to obtain the active correction frequency. The azimuth correction frequency is calculated by dividing the number of azimuth motor starts within the corresponding time slot by the duration of that time slot, and the pitch correction frequency is calculated by dividing the number of pitch motor starts within the corresponding time slot by the duration of that time slot. These two values ​​respectively characterize the frequency of spot offset in two dimensions. In addition, the locking mechanism in this embodiment generates passive correction actions, which also include azimuth correction frequency and pitch correction frequency. Therefore, the passive correction frequency also needs to be calculated based on the passive azimuth correction frequency and the passive pitch correction frequency. The calculation method for the passive correction frequency is the same as that for the active correction frequency. The active correction frequency and the passive correction frequency are then added together to obtain the total correction frequency.

[0041] The system determines whether the total correction frequency is less than a preset minimum frequency threshold. If the total correction frequency is less than the minimum frequency threshold, the detection interval is set to a preset upper bound value, indicating that the illumination conditions are highly stable during this period and the tracking platform does not require correction. In this case, the detection interval is directly set to the upper bound value. If the total correction frequency is not less than the minimum frequency threshold, the ratio of the preset reference frequency to the total correction frequency is calculated. This ratio is multiplied by the preset reference interval, and the product is limited to the preset upper and lower bounds to obtain the detection interval.

[0042] Based on the comparison between the average number of consecutive corrections performed by the motor in each detection cycle within the control range and the preset step size threshold, the correction step size is determined, and the dimension priority mode is determined based on the average ratio of the azimuth correction frequency to the pitch correction frequency.

[0043] For the correction step size, the average number of consecutive corrections for azimuth and pitch angles within the control range is calculated. The average number of consecutive corrections is the average of the number of consecutive corrections for azimuth and pitch angles, respectively. If the average number of consecutive corrections for azimuth is greater than a preset upper threshold, the correction step size is increased to the product of the current value and a preset increase coefficient, and limited to not exceeding the preset upper limit. If the average number of consecutive corrections for azimuth is less than a preset lower threshold, the correction step size is reduced to the product of the current value and a preset reduction coefficient, and limited to not falling below the preset lower threshold. If the average number of consecutive corrections is between the lower and upper thresholds, the correction step size remains unchanged. The adjustment process for pitch angle is similar and will not be described here.

[0044] When the average number of consecutive corrections exceeds the upper threshold of the step size, it indicates that multiple motor starts are required in each detection cycle to correct the light spot to the correct position. This suggests that the light spot offset is relatively large compared to the current correction step size. Increasing the correction step size in this case can increase the adjustment range of each correction action, reduce the number of consecutive corrections, and accelerate the offset elimination speed. When the average number of consecutive corrections is less than the lower threshold of the step size, it indicates that only a very small number of corrections or even a single correction is needed to complete the correction in each detection cycle. This suggests that the light spot offset is small and the system is already in a high-precision tracking state. Reducing the correction step size in this case can decrease the adjustment range of each correction action, avoid over-adjustment and oscillation caused by excessively large step sizes, and allow the light spot position to converge more accurately to the target position, thereby improving focusing accuracy and power output stability.

[0045] For the dimension priority mode, the ratio of the average azimuth correction frequency to the average pitch correction frequency within the control interval is calculated. If the ratio is greater than a preset priority threshold, it is set to azimuth priority mode; if the ratio is less than the reciprocal of the priority threshold, it is set to pitch priority mode; otherwise, it is set to balanced mode. The balanced mode avoids either dimension being delayed for a long time by alternately prioritizing the detection of the two dimensions, ensuring that both dimensions receive equal correction opportunities.

[0046] The electrical tracking parameters include the initial duty cycle value and the historical average output power. Specifically, the duty cycle of the DC-DC converter that maximizes the output power in each time slot of the electrical operation data is extracted as the optimal duty cycle of the time slot. The standard deviation and average value of the optimal duty cycles of all time slots in the control interval are calculated. If the standard deviation is less than the preset duty cycle discrete threshold, the average value is used as the initial duty cycle value. Otherwise, the optimal duty cycles of all time slots in the control interval are sorted by value, and the median of the sorted values ​​is used as the initial duty cycle value.

[0047] Electrical tracking parameters include the initial duty cycle value and historical average output power. When the standard deviation is less than the duty cycle dispersion threshold, it indicates that the optimal duty cycles of each time slot within the control interval are concentrated and their variation is small. In this case, using the average of the optimal duty cycles of all time slots as the initial duty cycle value can effectively suppress the influence of abnormal duty cycles caused by instantaneous disturbances in individual time slots on the initial value. When the standard deviation is greater than or equal to the duty cycle dispersion threshold, it indicates that the optimal duty cycles of each time slot within the control interval are more dispersed. Using the median as the initial duty cycle value can effectively resist the influence of extreme outliers. For the historical average output power, the average output power of all time slots within the control interval is extracted, and its arithmetic mean is calculated as the historical average output power of the control interval.

[0048] After calculating the corresponding control parameters based on multidimensional feature vectors and electrical operation data, if the difference in detection interval between two control intervals exceeds the preset interval jump threshold, the number of extractable time slots in the previous control interval and the next control interval are calculated respectively. The last number of extractable time slots in the previous control interval and the first number of extractable time slots in the next control interval are removed and merged into a transition control interval. The numerical control parameters of the transition control interval are the arithmetic mean of the corresponding parameters of the previous control interval and the next control interval.

[0049] For two temporally adjacent control intervals, if the difference in detection interval exceeds a preset interval transition threshold, a transition control interval is inserted between the two control intervals. When generating the transition control interval, the number of extractable time slots in the preceding control interval is first calculated. The total number of time slots contained in the control interval is subtracted from the preset minimum number of retained slots to obtain the first number of slots. This first number is compared with the preset second number of slots, and the smaller value is taken as the number of extractable time slots. If the number of extractable time slots is less than or equal to zero, no time slots are extracted from the control interval. The same method is used to calculate the number of extractable time slots for the following control interval. If the number of extractable time slots in both the preceding and following control intervals is greater than zero, the transition control interval consists of the last number of extractable time slots from the preceding control interval and the first number of extractable time slots from the following control interval. The extracted time slots are removed from their original control intervals and merged into the transition control interval.

[0050] The detection interval of the transition control interval is set to the arithmetic mean of the detection intervals of the preceding and following control intervals. The remaining control parameters of the transition control interval are the arithmetic mean of the corresponding parameters of the preceding and following control intervals. For dimension priority modes, if the dimension priority modes of the preceding and following control intervals are the same, the transition control interval adopts that same mode; otherwise, a balanced mode is adopted. The transition control interval participates as an independent control interval in subsequent cross-day parameter mapping and runtime parameter reading to achieve a smooth transition between adjacent control intervals while maintaining the uniformity of parameters within each control interval.

[0051] If the number of extractable time slots in either the preceding or following control interval is less than or equal to zero, no transition control interval is inserted. In this case, the last time slot of the preceding control interval and the first time slot of the following control interval are each retained within their original control intervals. The detection interval of the preceding control interval is adjusted to the weighted average of the original detection interval of the preceding control interval and the detection interval of the following control interval. The weight of the original detection interval of the preceding control interval is the preset weight of the preceding interval retention, and the weight of the detection interval of the following control interval is 1 minus the weight of the preceding interval retention. The weight of the preceding interval retention is greater than 0.5, so that the adjusted detection interval is biased towards the original value of the preceding control interval. This achieves a moderate convergence to the following control interval without inserting a transition interval, while maintaining the uniformity of the parameters within the preceding control interval.

[0052] In this embodiment, an adjustment locking window mechanism is used to shield against interference, including: Position correction and temperature adjustment are executed in a fixed order to form a mechanical adjustment sequence. The mechanical adjustment sequence does not activate the locking mechanism. When the entire mechanical adjustment sequence is completed, if at least one of the position correction action or temperature adjustment action is executed, the corresponding locking flag is set and a timer is started. Before the timer reaches the preset locking duration, the execution of other adjustment actions is prohibited. After the timer ends and the locking state is released, the subsequent correction actions caused by the preceding actions after the recovery detection are marked as passive corrections. Passive corrections are included in the correction frequency statistics.

[0053] At the start of a detection cycle, position correction and temperature adjustment are constructed as a mechanical adjustment sequence and performed sequentially in a fixed order, with position correction performed first and temperature adjustment performed later. No locking mechanism is activated during the execution of the mechanical adjustment sequence to ensure that temperature adjustment can be performed immediately after position correction is completed.

[0054] Once the entire mechanical adjustment sequence is completed, the detection cycle ends. If at least one of the position correction or temperature adjustment actions was actually performed, the adjustment lock window is activated. Specifically, the position tracking lock flag and the temperature adjustment lock flag are set to the locked state. The first timer is started, and its lock duration is equal to the product of the sum of all motor running times within the detection cycle and a preset lock multiple coefficient. Before the first timer reaches the lock duration, new spot offset detection and position correction actions, as well as new temperature detection and temperature adjustment actions, are prohibited. Simultaneously, the electrical tracking lock flag is set to the locked state. After the first timer reaches the lock duration, both the position tracking lock flag and the temperature adjustment lock flag remain locked. The second timer is started, and its duration is a preset electrical stabilization waiting time. Before the second timer reaches the electrical stabilization waiting time, the DC-DC converter is prohibited from performing duty cycle adjustment actions, and new spot offset detection, position correction, temperature detection, and temperature adjustment actions are also prohibited. After the second timer reaches the electrical stability waiting period, the position tracking lock flag, temperature regulation lock flag, and electrical tracking lock flag are simultaneously unlocked, and duty cycle adjustment is performed. If no motor action is actually performed during the position correction and temperature regulation in this cycle, the regulation lock window is not activated.

[0055] By taking the above steps, it is ensured that the mechanical and electrical systems are locked during the entire adjustment locking window (including the entire duration of the first and second timers), thus avoiding the disruption of the system steady state caused by the reintroduction of disturbances due to mechanical actions during the waiting period for electrical stabilization, thereby eliminating the race condition between the end of the first and second timers.

[0056] In the initial detection after unlocking, if the photosensitive detection module detects a shift in the position of the light spot, it performs a position correction and marks this correction as a passive correction. When updating the daily operation record, the passive correction is recorded separately as the number of passive azimuth corrections and the number of passive pitch corrections, along with the corresponding cumulative passive operating time. The passive correction frequency is then calculated based on this. When calculating the total correction frequency used in the detection interval, the active correction frequency and the passive correction frequency are directly added together to obtain the total correction frequency. This ensures that the calculation of the detection interval can fully reflect the actual tracking demand intensity accumulated by the system during the locking period.

[0057] The detection interval determines the system's response speed to spot offset. Passive correction reflects the actual offset accumulated during the locking period. Including it in the total correction frequency can ensure that the detection interval matches the actual tracking requirements and avoids a decrease in tracking accuracy due to an underestimation of the correction requirements and an extended detection interval.

[0058] In this embodiment, adjusting the DC-DC converter using electrical tracking parameters includes: Calculate the power deviation ratio between the current output power and the historical average output power. If the absolute value of the power deviation ratio does not exceed the preset threshold, set the duty cycle of the DC-DC converter to the initial duty cycle value. Perform bidirectional disturbance test based on the initial duty cycle value. Determine the duty cycle that makes the output power reach the maximum value as the optimal duty cycle based on the disturbance test results.

[0059] After position correction and temperature adjustment are completed and the electrical tracking lock flag is unlocked, the current output voltage and current of the solar panel are collected, and the current output power is calculated. For the historical average output power, the historical average output power in the electrical tracking parameters of the current control range is read.

[0060] If the power deviation ratio between the current output power and the historical average output power within the control range exceeds a preset deviation threshold, the initial duty cycle value is replaced with the actual duty cycle currently being used by the DC-DC converter, and the duty cycle perturbation step size is temporarily increased to the product of the original value and a preset acceleration coefficient. This situation indicates that historical values ​​are unreliable, and it is necessary to use the current actual duty cycle as the starting point and temporarily increase the perturbation step size to quickly respond to sudden environmental changes and recapture the maximum power point.

[0061] If the absolute value of the power deviation ratio does not exceed the deviation threshold, it indicates that the current illumination and temperature conditions are similar to historical records. In this case, the duty cycle is directly set to the initial value, and the duty cycle perturbation step size is kept unchanged.

[0062] During bidirectional perturbation testing, the initial duty cycle value is used as a reference, and the corresponding output power is collected as the reference power. The duty cycle is increased and decreased by one perturbation step, and the corresponding output power in the increasing direction and decreasing direction is collected. If the output power in the increasing direction is the maximum, the duty cycle is gradually adjusted in the increasing direction with perturbation step size from the increased duty cycle until the output power decreases. The duty cycle with the maximum power is taken as the optimal duty cycle. If the output power in the decreasing direction is the maximum, the duty cycle is gradually adjusted in the decreasing direction with perturbation step size from the decreased duty cycle until the output power decreases. The duty cycle with the maximum power is taken as the optimal duty cycle. If the reference power is not less than the other two, the initial duty cycle value is maintained as the optimal duty cycle.

[0063] In this embodiment, the control parameters for each control interval of the next operating day are calculated based on the control parameters of the previous day and the current day, including: The control parameters of each control interval of the previous operating day and the current day are assigned to all time slots contained in the corresponding control interval. The numerical control parameters of the previous operating day and the numerical control parameters of the current day are weighted and fused to obtain the fused numerical control parameters of the time slots. For the dimension priority mode, if the dimension priority mode corresponding to the time slot of the previous operating day and the current day is the same, the fusion result takes the same mode; otherwise, the fusion result takes the dimension priority mode of the current day.

[0064] Since the sunshine duration of each operating day is divided into the same number of N time slots, the time slot number is used as a unified time coordinate. For the previous operating day and the current day, the control parameters of each control interval are assigned to all time slots contained in that interval, so that each time slot of the previous operating day and the current day corresponds to a set of control parameters.

[0065] For each time slot number, the control parameters corresponding to that number on the previous running day are weighted and fused with the control parameters corresponding to that number on the current day. The weight of the previous running day is a preset historical weight coefficient, and the weight of the current day is a preset daily weight coefficient. The sum of the historical weight coefficient and the daily weight coefficient is 1, resulting in the fused control parameters for that time slot. For dimension priority modes, if the dimension priority modes corresponding to that time slot on the previous running day and the current day are the same, the fusion result adopts the same mode. If the dimension priority modes corresponding to that time slot on the previous running day and the current day are different, the fusion result adopts the dimension priority mode of the current day.

[0066] After obtaining the fusion control parameters for all time slots, if the differences between the numerical components in the fusion control parameters of multiple consecutive time slots are all less than the preset merging tolerance, and the dimension priority modes are the same, then these time slots are merged into one control interval. For numerical parameters, the arithmetic mean of the corresponding components in the fusion control parameters of the included time slots is taken. Through the above-described slot-by-slot alignment and re-merging mechanism, the fusion result is used as the control parameters for the next operating day, and the control interval division for the next operating day is determined.

[0067] like Figure 4 As shown, the present invention also provides a solar cell maximum power point adaptive tracking system for implementing the above-described method. The system includes: The data construction module extracts the mechanical correction data and temperature regulation data of the solar cells for the day according to the preset time slots, and constructs a multi-dimensional feature vector. The multi-dimensional feature vector and the corresponding recorded electrical operation data are combined to form the operation record for the day.

[0068] The clustering and partitioning module, based on the daily operation records, integrates the feature distance and temporal distance of multi-dimensional feature vectors to perform clustering and grouping. It merges time slots that belong to the same cluster and are temporally adjacent, and splits or migrates time slots that are not temporally discontinuous to obtain multiple control intervals.

[0069] The parameter calculation module calculates the corresponding control parameters for each control interval based on multi-dimensional feature vectors and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters.

[0070] The tracking execution module calculates the control parameters for each control interval of the next operating day based on the control parameters of the previous day and the current day. When the next day arrives, it reads the corresponding control parameters according to the control interval to which the current time belongs, corrects the spot position and adjusts the temperature based on the mechanical tracking parameters, and uses an adjustment locking window mechanism to shield interference. Based on the electrical tracking parameters, it adjusts the DC-DC converter to complete the tracking of the maximum power point of the solar cell.

[0071] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the method described above.

[0072] It should be specifically noted that the preset parameters, threshold settings, and basic data processing methods mentioned in the specific embodiments of this invention, including but not limited to the normalization processing method for multi-dimensional feature vectors, as well as the preset time slot division length, convergence conditions and number of clusters in the clustering algorithm, various weights and thresholds, etc., are all specific values ​​or formulas that can be reasonably determined by those skilled in the art based on their understanding of the core inventive concept of this invention, according to the actual concentrated photovoltaic system hardware model (such as motor response speed, solar panel characteristics, DC-DC converter specifications, etc.), the operating environment of the region (such as light intensity variation patterns, temperature fluctuation range, etc.), and historical operating data, through conventional experimental testing methods (such as trial and error, controlled variable method, grid search optimization, etc.). The values ​​of the above parameters and the selection of basic methods are all conventional technical means in this field, and those skilled in the art can determine them according to specific application scenarios and experimental tests.

Claims

1. A method for adaptive maximum power point tracking of solar cells, characterized in that, include: According to the preset time slot, extract the mechanical correction data and temperature regulation data of the solar cell for the day, construct a multi-dimensional feature vector, and combine the multi-dimensional feature vector and the corresponding recorded electrical operation data into the operation record for the day; Based on the daily operation records, clustering and grouping are performed by fusing the feature distance and temporal distance of multidimensional feature vectors. Time slots belonging to the same cluster and with adjacent temporal sequences are merged, and time slots with discontinuous temporal sequences are split or migrated to obtain multiple control intervals. For each control interval, the corresponding control parameters are calculated based on the multi-dimensional feature vector and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters. Based on the control parameters of the previous day and the current day, the control parameters of each control interval for the next operating day are calculated. When the next day arrives, the corresponding control parameters are read according to the control interval to which the current time belongs. Based on the mechanical tracking parameters, the spot position is corrected and the temperature is adjusted. An adjustment locking window mechanism is used to shield interference. The DC-DC converter is adjusted based on electrical tracking parameters to achieve maximum power point tracking of the solar cell.

2. The method according to claim 1, characterized in that, Clustering and grouping are performed by fusing feature distance and temporal distance from multidimensional feature vectors, including: Clustering is performed based on the K-Means clustering algorithm, and time slots are numbered according to time order. In the allocation stage of the clustering algorithm, the Euclidean distance between the multidimensional feature vector of the time slot and the multidimensional feature vector of the centroid of the target cluster is calculated as the feature distance. Obtain the temporal center of the target cluster, and calculate the temporal distance based on the time slot number and the time center number. In the initial allocation stage, the time center is the time slot number corresponding to the initial centroid, and in the iterative update stage, it is the average of the allocated time slot numbers within the target cluster. The composite distance is obtained by weighted summation of the feature distance and the temporal distance. Based on the composite distance, the time slots are assigned to the nearest clusters and the centroids are updated until the convergence condition is met to obtain the final clustering result.

3. The method according to claim 2, characterized in that, Merge time slots belonging to the same cluster and with adjacent time sequences, and split or migrate time slots with discontinuous time sequences, including: The time slots within the target cluster are sorted according to time sequence. The time slot sequences with discontinuous time sequence and the breakpoints are identified. The first feature difference between the time slots on both sides of the breakpoint and the second feature difference between the missing time slot at the breakpoint and the feature reference value of the target cluster are calculated respectively. If both the first feature difference and the second feature difference meet the preset compatibility conditions, the missing time slot is migrated to the target cluster to fill the breakpoint. Otherwise, the target cluster is split into multiple sub-clusters at the breakpoint. Time slots that belong to the same cluster and are temporally adjacent, target clusters that have not been split and have completed migration processing, or sub-clusters generated after splitting, with temporally consecutive time slots within them are merged into control intervals.

4. The method according to claim 2, characterized in that, The corresponding mechanical tracking parameters and electrical tracking parameters are calculated based on the multidimensional feature vector and electrical operation data, respectively, including: The mechanical tracking parameters include detection interval, correction step size, and dimension priority mode. The detection interval is calculated based on the orientation correction frequency and pitch correction frequency in the multi-dimensional feature vector. Based on the comparison between the average number of consecutive corrections performed by the motor in each detection cycle within the control range and the preset step size threshold, the correction step size is determined, and the dimension priority mode is determined based on the average ratio of the azimuth correction frequency to the pitch correction frequency. The electrical tracking parameters include the initial duty cycle value and the historical average output power. Specifically, the duty cycle of the DC-DC converter that maximizes the output power in each time slot of the electrical operation data is extracted as the optimal duty cycle of the time slot. The standard deviation and average value of the optimal duty cycles of all time slots in the control interval are calculated. If the standard deviation is less than the preset duty cycle discrete threshold, the average value is used as the initial duty cycle value. Otherwise, the optimal duty cycles of all time slots in the control interval are sorted by value, and the median of the sorted values ​​is used as the initial duty cycle value.

5. The method according to claim 4, characterized in that, After calculating the corresponding control parameters based on multidimensional feature vectors and electrical operation data, if the difference in detection interval between two control intervals exceeds the preset interval jump threshold, the number of extractable time slots in the previous control interval and the next control interval are calculated respectively. The last number of extractable time slots in the previous control interval and the first number of extractable time slots in the next control interval are removed and merged into a transition control interval. The numerical control parameters of the transition control interval are the arithmetic mean of the corresponding parameters of the previous control interval and the next control interval.

6. The method according to claim 1, characterized in that, Interference is shielded by adjusting the locking window mechanism, including: Position correction and temperature adjustment are executed in a fixed order to form a mechanical adjustment sequence. The mechanical adjustment sequence does not activate the locking mechanism. When the entire mechanical adjustment sequence is completed, if at least one of the position correction action or temperature adjustment action is executed, the corresponding locking flag is set and a timer is started. Before the timer reaches the preset locking duration, the execution of other adjustment actions is prohibited. After the timer ends and the locking state is released, the subsequent correction actions caused by the preceding actions after the recovery detection are marked as passive corrections. Passive corrections are included in the correction frequency statistics.

7. The method according to claim 1, characterized in that, Adjusting a DC-DC converter using electrical tracking parameters includes: Calculate the power deviation ratio between the current output power and the historical average output power. If the absolute value of the power deviation ratio does not exceed the preset threshold, set the duty cycle of the DC-DC converter to the initial duty cycle value. Perform bidirectional disturbance test based on the initial duty cycle value. Determine the duty cycle that makes the output power reach the maximum value as the optimal duty cycle based on the disturbance test results.

8. The method according to claim 4, characterized in that, Calculate the control parameters for each control interval for the next operating day based on the control parameters of the previous day and the current day, including: The control parameters of each control interval of the previous operating day and the current day are assigned to all time slots contained in the corresponding control interval. The numerical control parameters of the previous operating day and the numerical control parameters of the current day are weighted and fused to obtain the fused numerical control parameters of the time slots. For the dimension priority mode, if the dimension priority mode corresponding to the time slot of the previous operating day and the current day is the same, the fusion result takes the same mode; otherwise, the fusion result takes the dimension priority mode of the current day.

9. A solar cell maximum power point adaptive tracking system, used to implement the method as described in any one of claims 1-8, characterized in that, The system includes: The data construction module extracts the mechanical correction data and temperature regulation data of the solar cells for the day according to the preset time slots, and constructs a multi-dimensional feature vector. The multi-dimensional feature vector and the corresponding recorded electrical operation data are combined into the operation record for the day. The clustering and partitioning module, based on the daily operation records, integrates the feature distance and temporal distance of multi-dimensional feature vectors to perform clustering and grouping, merges time slots that belong to the same cluster and are temporally adjacent, and splits or migrates time slots that are not temporally discontinuous to obtain multiple control intervals. The parameter calculation module calculates the corresponding control parameters for each control interval based on multi-dimensional feature vectors and electrical operation data. The control parameters include mechanical tracking parameters and electrical tracking parameters. The tracking execution module calculates the control parameters for each control interval of the next operating day based on the control parameters of the previous day and the current day. When the next day arrives, it reads the corresponding control parameters according to the control interval to which the current time belongs, corrects the spot position and adjusts the temperature based on the mechanical tracking parameters, and uses an adjustment locking window mechanism to shield interference. Based on the electrical tracking parameters, it adjusts the DC-DC converter to complete the tracking of the maximum power point of the solar cell.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements the method as claimed in any one of claims 1-8.

Images