An AI-based photovoltaic tracking support control method and system

Abstract

This invention relates to the field of photovoltaic tracking bracket technology, and more particularly to an AI-based photovoltaic tracking bracket control method and system. The method acquires light-sensing data from various angles of the photovoltaic module, real-time power generation, bracket tracking energy consumption, environmental wind and rain parameters, and the current tilt angle of the bracket. It collects dust accumulation response signals and tracking energy efficiency response signals within a preset tracking cycle and generates a dust accumulation reference signal. It also acquires tracking operation response signals and constructs a set of dust accumulation energy efficiency response curves. The curve set is analyzed to determine the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value, generating a photovoltaic dust accumulation energy efficiency time series diagram, divided into stable and fluctuating sub-graphs. The duration of the fluctuating sub-graph determines the dust accumulation anomaly category. Based on different categories, it generates graded control commands to achieve differentiated control such as tilt angle adjustment, frequency switching, pulse oscillation, or triggered cleaning. This invention improves self-cleaning capability and optimizes power generation efficiency by adaptively adjusting the tracking strategy based on the dust accumulation state.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic tracking bracket technology, and in particular to an AI-based photovoltaic tracking bracket control method and system. Background Technology

[0002] In actual operation, dust, dirt, and other contaminants inevitably accumulate on the surface of photovoltaic modules, leading to decreased light transmittance and reduced power generation efficiency. Existing technologies for addressing dust accumulation mainly fall into two categories: one is to remove dust through periodic manual cleaning or the installation of automatic cleaning devices, using a fixed cycle or a simple threshold-based method; the other is to introduce dust accumulation sensors into the support control strategy, triggering an alarm or shutdown for cleaning when dust accumulation exceeds a set threshold. However, both of these methods treat dust accumulation as an external disturbance independent of tracking control, failing to consider the dynamic coupling relationship between dust accumulation and the tracking strategy.

[0003] On the one hand, different tracking tilt angles directly affect the deposition and distribution of dust on the module surface and the path of rainwater erosion. A reasonable tilt angle design can leverage natural wind and rain to achieve self-cleaning of dust, reducing the frequency of manual cleaning. On the other hand, changes in the degree of dust accumulation affect the actual power generation gain of the module, thereby altering the energy efficiency benefit boundary of tracking control. If the tracking strategy does not consider the dust accumulation status, high-frequency tracking may still be executed during periods of severe dust accumulation, resulting in a situation where motor energy consumption exceeds power generation gain. Therefore, there is an urgent need for a control method that can sense the dust accumulation status in real time, assess the self-cleaning potential of dust accumulation, and dynamically adjust the tracking strategy according to the degree of dust accumulation. Summary of the Invention

[0004] To address this, the present invention provides an AI-based photovoltaic tracking bracket control method and system to overcome the problems in the prior art where dust accumulation detection and tracking strategies are disconnected, and the tracking mode cannot be dynamically adjusted according to the self-cleaning potential of dust accumulation. This results in difficulties in simultaneously taking into account natural self-cleaning utilization and power generation efficiency optimization under dust accumulation conditions, and easily leads to increased energy consumption from ineffective tracking or excessively high cleaning costs.

[0005] To achieve the above objectives, in one aspect, the present invention provides an AI-based photovoltaic tracking bracket control method, comprising:

[0006] Acquire light-sensing data of photovoltaic modules from various angles, real-time power generation, energy consumption of the support system, environmental wind and rain parameters, and the current tilt angle of the support system;

[0007] The dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module are collected during the preset tracking period, and the characteristic signals corresponding to the light sensing data at each angle are recorded as dust accumulation reference signals.

[0008] Acquire a tracking operation response signal corresponding to the time of the ash accumulation reference signal. The tracking operation response signal includes a transmittance attenuation sequence and a power generation gain change sequence. Generate a set of ash accumulation energy efficiency response curves based on the ash accumulation reference signal and the tracking operation response signal.

[0009] The set of ash accumulation energy efficiency response curves is analyzed, and the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module are determined based on the analysis results. A photovoltaic ash accumulation energy efficiency time series diagram is generated based on the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value.

[0010] The photovoltaic ash accumulation energy efficiency time series diagram is divided into a stable subgraph and a fluctuating subgraph, and the ash accumulation anomaly characteristic category is determined according to the duration of the fluctuating subgraph. The ash accumulation anomaly characteristic category includes self-cleaning category, semi-self-cleaning category and non-self-cleaning category.

[0011] Based on the category of abnormal dust accumulation characteristics and the current tilt angle of the support, a target tracking and control strategy is determined, and corresponding hierarchical control instructions are generated. The hierarchical control instructions include adjusting the tracking tilt angle of the support, switching the tracking frequency, and applying pulsed tilt angle reciprocating oscillation in combination with the environmental wind and rain parameters to enhance the self-cleaning state or trigger component cleaning.

[0012] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, within the preset tracking period, the ratio of incident irradiance to reflected irradiance of each angle light sensing data at each sampling time is extracted as an irradiance comparison feature value, and the ratio of edge transmittance to incident irradiance is extracted as a transmittance attenuation feature value.

[0013] The irradiance comparison feature values ​​of all sampling times within the same preset tracking period are arranged in chronological order to form an irradiance comparison feature sequence. The light transmittance attenuation feature values ​​of all sampling times within the same preset tracking period are arranged in chronological order to form a light transmittance attenuation feature sequence. The irradiance comparison feature sequence and the light transmittance attenuation feature sequence are used together as the ash accumulation reference signal.

[0014] Each sampling time corresponds one-to-one with the acquisition time of the ash accumulation response signal and the tracking energy efficiency response signal within the preset tracking period.

[0015] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, the tracking operation response signal is determined to include:

[0016] The edge transmittance at each sampling time is extracted from the ash accumulation response signal and arranged in chronological order to form the transmittance attenuation sequence.

[0017] The power generation time series at each sampling time is extracted from the tracking energy efficiency response signal, the change in power generation at each sampling time relative to the reference time is calculated, and the power generation gain change sequence is formed by arranging them in chronological order.

[0018] The transmittance attenuation sequence and the power generation gain change sequence are aligned in the time dimension and combined to form the tracking operation response signal.

[0019] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, the process of generating a set of ash accumulation energy efficiency response curves based on the ash accumulation reference signal and the tracking operation response signal includes:

[0020] The irradiation comparison feature sequence in the ash accumulation reference signal and the power generation gain change sequence in the tracking operation response signal are plotted with time as the horizontal axis as irradiation comparison curve and power generation gain change curve, respectively. The irradiation comparison curve and the power generation gain change curve are combined into an irradiation energy efficiency response curve group.

[0021] The transmittance attenuation characteristic sequence in the ash accumulation reference signal, the transmittance attenuation sequence in the tracking operation response signal, and the power generation gain change sequence are plotted with time as the horizontal axis as the transmittance attenuation reference curve, the real-time transmittance attenuation curve, and the power generation gain change curve, respectively, and are denoted as the transmittance energy efficiency response curve group.

[0022] The irradiation energy efficiency response curve set and the light transmission energy efficiency response curve set are used together as the ash accumulation energy efficiency response curve set.

[0023] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, the process of analyzing the set of ash accumulation energy efficiency response curves includes:

[0024] Calculate the difference between the power generation gain change curve and the irradiation comparison curve in the irradiation energy efficiency response curve group at the same sampling time, and generate an irradiation energy efficiency difference sequence;

[0025] Calculate the difference between the real-time transmittance decay curve and the transmittance decay reference curve in the transmittance energy efficiency response curve group at the same sampling time, and generate a transmittance decay difference sequence.

[0026] Calculate the difference between the power generation gain change curve and the power generation gain baseline in the light transmission energy efficiency response curve group at the same sampling time, and generate a power generation energy efficiency difference sequence;

[0027] The irradiance ratio dispersion characterization value, light transmission attenuation dispersion characterization value, and power generation gain dispersion characterization value are calculated based on the irradiance energy efficiency difference sequence, light transmission attenuation difference sequence, and power generation energy efficiency difference sequence, respectively.

[0028] As a preferred technical solution for AI-based photovoltaic tracking bracket control, the process of determining the ash self-cleaning characteristic value and the tracking energy consumption and power generation gain characterization value based on the analysis results includes:

[0029] The self-cleaning characteristic value of ash accumulation is determined based on the dispersion characterization value of the irradiance ratio and the dispersion characterization value of the light transmittance attenuation.

[0030] The power generation gain dispersion characterization value is used as the power generation gain characterization value for tracking energy consumption.

[0031] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, the process of dividing the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determining the ash accumulation anomaly characteristic category based on the duration of the fluctuating subgraph includes:

[0032] The curve segments of the self-cleaning characteristic value curve of the photovoltaic ash accumulation energy efficiency time series diagram that are within the preset stable benchmark range are marked as stable subgraphs, and the curve segments that are outside the preset stable benchmark range are marked as fluctuating subgraphs.

[0033] Obtain the duration of the fluctuation subgraph and compare the duration with a preset duration threshold;

[0034] Wherein, when the duration of the fluctuation subgraph is less than the preset duration threshold, it is determined to be a self-cleaning category;

[0035] When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold and the peak value of the self-cleaning feature value is less than the preset feature value threshold, it is determined to be a semi-self-cleaning category.

[0036] When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold and the peak value of the self-cleaning feature value is greater than or equal to the preset feature value threshold, it is determined to be a non-self-cleaning category.

[0037] As a preferred technical solution for AI-based photovoltaic tracking bracket control methods, the process of determining the target tracking control strategy based on the category of dust accumulation anomaly characteristics combined with the current tilt angle of the bracket, and generating corresponding hierarchical control commands, includes:

[0038] When the dust accumulation abnormality feature category is a self-cleaning category, a first control command is generated. The first control command includes adjusting the support tracking tilt angle to a preset self-cleaning tilt angle range and switching the tracking frequency to a low-frequency tracking mode.

[0039] When the dust accumulation abnormality feature category is semi-self-cleaning, a second control command is generated. The second control command includes determining the pulse swing parameters based on the current tilt angle of the support and the environmental wind and rain parameters, and applying a pulse tilt angle reciprocating swing. The pulse swing parameters include swing amplitude, swing period and swing number.

[0040] When the dust accumulation abnormality characteristic category is non-self-cleaning, a third control command is generated. The third control command includes locking the bracket at a preset safe tilt angle and triggering a component cleaning command.

[0041] As a preferred technical solution for the AI-based photovoltaic tracking bracket control method, the pulse swing parameter is determined as follows when generating the second control command:

[0042] The swing amplitude is determined based on the difference between the current tilt angle of the bracket and the preset self-cleaning tilt angle range, and the swing amplitude increases as the difference increases;

[0043] The oscillation period is determined based on the wind speed in the environmental wind and rain parameters, and the oscillation period decreases as the wind speed increases;

[0044] The number of oscillations can be set to a preset fixed value, or determined according to the duration of the oscillation subgraph.

[0045] On the other hand, the present invention also provides a system comprising:

[0046] The data acquisition module is used to acquire light-sensing data of photovoltaic modules from various angles, real-time power generation, energy consumption of the support tracking, environmental wind and rain parameters, and the current tilt angle of the support.

[0047] The intelligent feature generation module is used to collect the dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module at a preset tracking period, record the feature signals corresponding to the light sensing data at each angle as dust accumulation reference signals, obtain the tracking operation response signal corresponding to the time of the dust accumulation reference signal, generate a set of dust accumulation energy efficiency response curves based on the dust accumulation reference signal and the tracking operation response signal, analyze the set of dust accumulation energy efficiency response curves, determine the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module, and generate a photovoltaic dust accumulation energy efficiency time series diagram based on the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value.

[0048] The intelligent time series analysis module is used to divide the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determine the ash accumulation anomaly feature category based on the duration of the fluctuating subgraph;

[0049] The intelligent decision-making module is used to determine the target tracking and control strategy based on the category of the abnormal ash accumulation characteristics and the current tilt angle of the support, and to generate corresponding hierarchical control instructions;

[0050] The execution module is used to execute the graded control commands, control the tilt adjustment, frequency switching and pulse oscillation of the support, and output a cleaning start signal when the graded control command triggers component cleaning.

[0051] Compared with existing technologies, the beneficial effects of this invention are as follows: By simultaneously collecting data on the light sensitivity of photovoltaic modules at various angles, power generation, energy consumption of the tracking bracket, environmental wind and rain, and tilt angle, this invention constructs a dust accumulation reference signal and a tracking operation response signal, and generates a set of dust accumulation energy efficiency response curves. Through quantitative analysis, it obtains the dust accumulation self-cleaning characteristic value and the tracking energy consumption and power generation gain characterization value. Based on time series diagram division and threshold judgment, it achieves three-level classification identification of dust accumulation: self-cleaning, semi-self-cleaning, and non-self-cleaning. Then, it matches the graded control strategies of tilt angle adjustment, tracking frequency switching, pulse reciprocating oscillation, locking the safe tilt angle, and cleaning triggering, realizing the dynamic coupling between dust accumulation status and photovoltaic tracking control. It can make full use of natural wind and rain to achieve dust accumulation self-cleaning, effectively reducing the energy consumption of the tracking bracket and unnecessary cleaning frequency, improving power generation gain and system operation stability, and taking into account power generation efficiency, equipment safety, and operation and maintenance cost optimization. Attached Figure Description

[0052] Figure 1 This is a flowchart of an AI-based photovoltaic tracking bracket control method according to an embodiment of the present invention;

[0053] Figure 2 This is a flowchart illustrating the analysis of the ash accumulation energy efficiency response curve set according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the system structure of the AI-based photovoltaic tracking bracket control method according to an embodiment of the present invention. Detailed Implementation

[0055] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0056] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0057] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0058] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0059] Please see Figure 1 and Figure 2 As shown, this invention provides an AI-based photovoltaic tracking bracket control method, comprising:

[0060] Step S1: Obtain light sensing data of photovoltaic modules at various angles, real-time power generation, energy consumption of support tracking, environmental wind and rain parameters, and the current tilt angle of the support.

[0061] Step S2: Collect the dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module during the preset tracking period, and record the characteristic signals corresponding to the light sensing data at each angle as dust accumulation reference signals.

[0062] Step S3: Obtain the tracking operation response signal corresponding to the time of the ash accumulation reference signal. The tracking operation response signal includes a transmittance attenuation sequence and a power generation gain change sequence. Generate a set of ash accumulation energy efficiency response curves based on the ash accumulation reference signal and the tracking operation response signal.

[0063] Step S4: Analyze the set of ash accumulation energy efficiency response curves, determine the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module based on the analysis results, and generate a photovoltaic ash accumulation energy efficiency time series diagram based on the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value.

[0064] Step S5: Divide the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determine the ash accumulation anomaly feature category according to the duration of the fluctuating subgraph. The ash accumulation anomaly feature category includes self-cleaning category, semi-self-cleaning category and non-self-cleaning category.

[0065] Step S6: Determine the target tracking and control strategy based on the category of abnormal dust accumulation characteristics and the current tilt angle of the support, and generate corresponding hierarchical control instructions. The hierarchical control instructions include adjusting the tracking tilt angle of the support, switching the tracking frequency, and applying pulsed tilt angle reciprocating oscillation in combination with the environmental wind and rain parameters to enhance the self-cleaning state or trigger component cleaning.

[0066] This invention synchronously collects data on light sensitivity, power generation, support tracking energy consumption, environmental wind and rain, and tilt angle from various angles of photovoltaic modules. It constructs a dust accumulation reference signal and a tracking operation response signal, and generates a set of dust accumulation energy efficiency response curves. Through quantitative analysis, it obtains the dust accumulation self-cleaning characteristic value and the tracking energy consumption and power generation gain characterization value. Based on time series diagram division and threshold judgment, it realizes three-level classification of dust accumulation: self-cleaning, semi-self-cleaning, and non-self-cleaning. Then, it matches a graded control strategy of tilt angle adjustment, tracking frequency switching, pulse reciprocating oscillation, locking the safe tilt angle, and cleaning triggering. This realizes the dynamic coupling between dust accumulation status and photovoltaic tracking control. It can make full use of natural wind and rain to achieve dust accumulation self-cleaning, effectively reduce support tracking energy consumption and unnecessary cleaning frequency, improve power generation gain and system operation stability, and take into account power generation efficiency, equipment safety, and operation and maintenance cost optimization.

[0067] In implementation, the process of collecting the dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module includes: controlling the irradiance sensor and transmittance sensor that rotate synchronously with the support shaft; at each sampling moment of the preset tracking period, simultaneously collecting the incident irradiance intensity on the front of the photovoltaic module, the reflected irradiance intensity on the back, and the transmittance at the edge; arranging the three types of light sensing data at all sampling moments within the same preset tracking period in chronological order to form the dust accumulation response signal; at each sampling moment identical to the above light sensing data collection, simultaneously collecting the real-time power generation of the photovoltaic module and the tracking energy consumption of the support; arranging them according to the sampling moment to form a power generation time series and an energy consumption time series; calculating the power generation gain change at the corresponding sampling moment based on the power generation time series; arranging them in chronological order to form a power generation gain change series; and combining the power generation time series, energy consumption time series, and power generation gain change series together to form the tracking energy efficiency response signal.

[0068] Understandably, when photovoltaic modules are clean, the ratio of incident irradiance on the front to reflected irradiance on the back, and the ratio of edge transmittance to incident irradiance, are all within a stable baseline range. When dust or mud stains accumulate on the module surface after rain, the incident light is scattered and absorbed by the dust, the effective transmission ratio of incident irradiance on the front decreases, and the reflected irradiance on the back decreases simultaneously. The decrease in edge transmittance is directly related to the thickness and range of the dust accumulation. By using ratio-based feature extraction, the influence of the natural fluctuations in incident light intensity on the detection results can be eliminated, retaining only the feature changes caused by dust accumulation, thus achieving accurate characterization of the dust accumulation state.

[0069] In this embodiment, the preset tracking period is set to 10-30 minutes. This duration range strikes a balance between the rate of irradiance change and the rate of dust accumulation in a typical photovoltaic power station, ensuring sensitivity to changes in dust accumulation while avoiding unnecessary system noise introduced by frequent sampling. Within this period, the sampling interval must meet the basic requirements of the Nyquist sampling theorem, i.e., the sampling frequency must be no less than twice the highest effective frequency component in the dust accumulation reference signal. In actual implementation, sampling can be set to once every 30 to 60 seconds based on the response bandwidth of the support motor to ensure complete acquisition and reconstruction of the disturbance signal.

[0070] It should be understood that the light-sensing characteristics, ash accumulation status, and power generation efficiency data at the same moment have a direct causal relationship. Timing misalignment will lead to the break of the causal relationship. Therefore, by triggering the synchronous acquisition of the three types of signals by the same clock source, the acquisition time is fully aligned, and the subsequently generated feature sequence and response signal are fully matched in the time dimension, providing a comparable unified timing benchmark for the generation of the ash accumulation energy efficiency response curve group.

[0071] Understandably, the transmittance decay sequence directly reflects the impact of dust accumulation on the light transmittance performance of the module, while the power generation gain change sequence reflects the change in power generation revenue brought about by the tracking action. The alignment of the two in the time dimension is the basis for subsequent correlation analysis. In implementation, when extracting the transmittance decay sequence from the dust accumulation response signal, the edge transmittance at each sampling moment is directly collected by the transmittance sensor that rotates with the support, and the sequence can be formed by arranging them in chronological order without additional calculation.

[0072] When extracting the power generation gain change sequence from the tracking energy efficiency response signal, the power generation at the first sampling moment within the preset tracking period is used as the reference moment to calculate the change in power generation at each subsequent sampling moment relative to the reference moment. The reference moment is selected based on the fact that at the beginning of the period, the support has completed the previous period's regulation and entered a stable operating state, and its power generation can be used as the zero point of energy efficiency reference within that period. After a limited number of tests, this reference moment setting method can maintain the stability of the power generation gain calculation under different irradiation conditions, avoiding calculation deviations introduced by periodic fluctuations.

[0073] By combining the transmittance attenuation sequence and the power generation gain change sequence according to the sampling time, a tracking operation response signal is formed. The ash accumulation characterization and energy efficiency characterization are determined on the same time axis to ensure that the causal relationship between the two sequences is not destroyed by time misalignment in subsequent curve analysis.

[0074] In practice, the purpose of generating a set of ash accumulation energy efficiency response curves is to establish a visual correlation between ash accumulation characterization parameters and energy efficiency response parameters, so as to provide a data basis for subsequent feature extraction and classification.

[0075] The irradiation energy efficiency response curve set is generated by plotting the irradiation comparison feature sequence and the power generation gain change sequence on the same coordinate system with time as the horizontal axis, forming the irradiation comparison curve and the power generation gain change curve. The irradiation comparison feature value reflects the ratio of irradiation on the front and back of the module. This ratio changes under the influence of dust accumulation. Presenting it side by side with the power generation gain change curve allows for a direct identification of the correlation between changes in irradiation conditions and changes in power generation revenue.

[0076] The transmission efficiency response curve set is generated by plotting the transmission attenuation characteristic sequence, the transmittance attenuation sequence, and the power generation gain change sequence on the same coordinate system with time as the horizontal axis, forming the transmission attenuation baseline curve, the real-time transmittance attenuation curve, and the power generation gain change curve. The transmission attenuation characteristic sequence, as a component of the dust accumulation reference signal, reflects the edge dust accumulation characteristics after environmental irradiation normalization, serving as a benchmark characterization of the dust accumulation degree. The transmittance attenuation sequence, as a component of the tracking operation response signal, reflects the real-time raw measurement value of the edge transmittance, serving as a real-time monitoring value of the dust accumulation degree.

[0077] The two sets of curves together constitute the ash accumulation energy efficiency response curve set, presenting the correlation between ash accumulation status and power generation energy efficiency from two dimensions: irradiation comparison-energy efficiency and light transmission attenuation benchmark-real-time light transmission-energy efficiency. This dual-dimensional design enables subsequent analysis to distinguish whether the energy efficiency decline is caused by changes in environmental irradiation or by ash accumulation. Furthermore, by comparing the benchmark curve and the real-time curve in the light transmission energy efficiency response curve set, drifts or anomalies in the light transmission attenuation characteristic sequence can be identified, providing a basis for the accurate determination of graded control strategies.

[0078] In implementation, the purpose of analyzing the ash accumulation energy efficiency response curve set is to quantify the dynamic correlation between the curves into a calculable deviation index, providing a numerical basis for subsequent feature value extraction. In the irradiation energy efficiency response curve set, the irradiation comparison curve represents the ash accumulation state after environmental normalization, and the power generation gain change curve represents the dynamic response of the tracking energy efficiency. The difference between the two at the same sampling time reflects the degree of deviation of the current energy efficiency relative to the current ash accumulation state. This deviation is arranged chronologically to form an irradiation energy efficiency difference sequence. In the light transmittance energy efficiency response curve set, the real-time transmittance attenuation curve comes from the tracking operation response signal, reflecting the original measured value of edge transmittance, and the transmittance attenuation reference curve comes from the ash accumulation reference signal, reflecting the edge ash accumulation characteristics after environmental normalization. The difference between the two at the same sampling time reflects the deviation between the real-time transmittance measurement value and the theoretical ash accumulation characteristics. This deviation is arranged chronologically to form a transmittance attenuation difference sequence. The difference between the power generation gain change curve and the power generation gain reference line at the same sampling time reflects the fluctuation of the actual energy efficiency relative to the stable operation energy efficiency reference. This fluctuation is arranged chronologically to form a power generation energy efficiency difference sequence.

[0079] When calculating the dispersion characterization value, the standard deviation of each difference sequence is used as the dispersion characterization value. The standard deviation is calculated as follows: calculate the arithmetic mean of the difference sequence, then calculate the sum of squares of the deviations of each difference from this mean, divide by the sequence length, and take the square root. This calculation method can reflect the fluctuation range of the difference sequence around its mean; the greater the fluctuation, the larger the standard deviation, indicating that the current state is more unstable. The length of the difference sequence within the preset tracking period is determined by the number of sampling times, usually between 20 and 60 sampling points. This range can balance the stability of data statistics and the response speed to changes in dust accumulation.

[0080] In implementation, the purpose of determining the ash self-cleaning characteristic value and the energy consumption power generation gain characterization value is to integrate multi-dimensional dispersion characterization values ​​into a comprehensive index that can be used for classification judgment. The irradiance ratio dispersion characterization value reflects the degree of fluctuation in energy efficiency deviation under ash accumulation conditions, while the light transmittance attenuation dispersion characterization value reflects the degree of fluctuation in the deviation between real-time light transmittance measurement values ​​and theoretical ash accumulation characteristics. These two values ​​characterize the ash accumulation state from two dimensions: energy efficiency-ash accumulation correlation and light transmittance measurement consistency, respectively. The two values ​​are integrated using a square root method: the square of the sum of the squares of the irradiance ratio dispersion characterization value and the square of the light transmittance attenuation dispersion characterization value is calculated, and the square root is taken as the ash self-cleaning characteristic value. This integration method can comprehensively reflect the combined influence of the two dimensions; when the dispersion of either dimension increases, the characteristic value will increase and will not be overwhelmed by excessively small values.

[0081] The power generation gain dispersion characterization value directly reflects the fluctuation range of actual energy efficiency relative to the stable operating energy efficiency benchmark, and is used as the characterization value for tracking energy consumption power generation gain. The larger the value, the more severe the energy efficiency fluctuation, the worse the stability of the tracking strategy, and the more necessary the regulatory intervention. The calculation of the above characteristic value does not involve weight setting, avoiding the introduction of subjective parameters, and the calculation result depends only on the deviation distribution between the measured data and the benchmark.

[0082] When the duration of the fluctuation subgraph is less than the preset duration threshold, it is determined to be a self-cleaning category;

[0083] When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold and the peak value of the self-cleaning feature value is less than the preset feature value threshold, it is determined to be a semi-self-cleaning category.

[0084] When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold and the peak value of the self-cleaning feature value is greater than or equal to the preset feature value threshold, it is determined to be a non-self-cleaning category.

[0085] In implementation, a photovoltaic ash accumulation energy efficiency time series diagram is generated based on the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value. With time as the unified horizontal axis, the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value are plotted as corresponding time series curves according to the sampling time sequence. This allows the two types of characterization values ​​to intuitively present the changing trend on the same time dimension, providing a visual analysis carrier for subsequent regional division and state determination.

[0086] When dividing the photovoltaic ash accumulation energy efficiency time series diagram into stable and fluctuating sub-diagrams, a preset stable benchmark range is used as the criterion. Curve segments within the preset stable benchmark range are marked and distinguished; those outside this range are marked as fluctuating sub-diagrams. This allows for a direct distinction between stable and abnormal fluctuations in ash accumulation energy efficiency. The preset stable benchmark range is determined by obtaining the numerical distribution of the ash self-cleaning characteristic value curve during historical periods without ash accumulation or after ash removal. The mean and standard deviation of the ash self-cleaning characteristic value during this period are calculated. The range of the mean plus or minus twice the standard deviation is used as the preset stable benchmark range. This eliminates interference from normal environmental fluctuations and accurately identifies abnormal fluctuations caused by ash accumulation. Curve segments within this preset stable benchmark range are marked as stable sub-diagrams, while those outside are marked as fluctuating sub-diagrams.

[0087] In implementation, the preset duration threshold is the typical duration for dust to naturally detach under natural rainfall and wind. This threshold is determined by statistically analyzing the actual self-cleaning time of dust accumulation in photovoltaic power plants in different regions, using linear regression fitting to distinguish between short-term fluctuations and persistent dust accumulation anomalies. Generally, under no human intervention, mild dust accumulation can usually be cleared by natural wind and rain within 1 to 3 hours; therefore, the preset duration threshold is preferably set to 2 hours. When the duration of the fluctuation sub-graph is less than 2 hours, it indicates that the dust accumulation is temporary and is classified as self-cleaning.

[0088] The preset characteristic value threshold is determined as follows: Data on the self-cleaning characteristic values ​​of the photovoltaic power station during its historical operation, specifically before cleaning or during periods with more than 15 consecutive days without rainfall, are collected. The 95th percentile of the self-cleaning characteristic values ​​during this period is calculated, and this 95th percentile is used as the preset characteristic value threshold. The selection of the 95th percentile eliminates the interference of extreme outliers, ensuring that the threshold reflects the typical upper limit of the severity of dust accumulation. For newly built power stations or power stations without historical data, the initial threshold can be determined by artificially simulating dust accumulation during the first tracking period or by referring to historical data from similar power stations in the same region. Subsequently, as operational data accumulates, the threshold is recalculated and updated quarterly or semi-annually based on newly added data.

[0089] In this invention, an automatic classification and determination of ash accumulation anomalies is achieved by quantifying thresholds, avoiding the subjectivity and lag of manual judgment. The principle is that the duration and severity of ash accumulation anomalies directly reflect the ability to remove ash. By combining time-series fluctuation characteristics and characteristic value peaks, the ash accumulation anomaly category can be accurately classified. After implementation, it can provide a reliable basis for subsequent classification and control strategies, ensuring that the stent control actions match the actual ash accumulation state.

[0090] During implementation, the target tracking and control strategy is determined based on the abnormal characteristics of ash accumulation and the current tilt angle of the support, and corresponding hierarchical control instructions are generated. The purpose is to take differentiated control actions for different levels of ash accumulation, maximize the use of natural conditions to achieve self-cleaning while ensuring power generation efficiency, and avoid unnecessary cleaning operations or ineffective tracking.

[0091] When the dust accumulation anomaly is classified as self-cleaning, it indicates a relatively light dust accumulation and the potential for natural removal. At this point, the first control command is generated. This command adjusts the support tracking angle to a preset self-cleaning angle range. This range is determined based on aerodynamic principles and calibrated through a limited number of tests. When the support angle is between 30° and 45°, the airflow velocity on the module surface is maximized, and the rainwater scouring path is longest, which is conducive to the natural shedding of dust. Simultaneously, the tracking frequency is switched to a low-frequency tracking mode, adjusting the tracking angle once per hour, replacing the original minute-level continuous tracking. This reduces motor energy consumption while maintaining basic light exposure for the module during the main irradiation period.

[0092] When the dust accumulation anomaly category is semi-self-cleaning, it indicates that the dust accumulation is persistent but has not yet reached a severe level. Simply adjusting the tilt angle is insufficient for removal, requiring the introduction of active disturbance to enhance the self-cleaning effect. At this point, a second control command is generated. Based on the current tilt angle of the support and environmental wind and rain parameters, pulse oscillation parameters are determined, and pulsed tilt angle reciprocating oscillation is applied. The principle of pulse oscillation is to create airflow impact and vibration on the component surface through periodic, large-amplitude tilt angle changes, disrupting the adhesion of the dust accumulation and enabling self-cleaning with the help of natural wind and rain. The purpose of this command is to improve self-cleaning efficiency and extend the cleaning cycle through active control without triggering manual cleaning.

[0093] When the dust accumulation anomaly is classified as "non-self-cleaning," it indicates that the dust accumulation is severe and persistent, making effective self-cleaning impossible through tilt adjustment or pulse oscillation. In this case, a third control command is generated to lock the support at a preset safe tilt angle. The preset safe tilt angle is set to 5° to 10°. This angle range prevents the components from being subjected to excessive wind loads in severe weather and facilitates subsequent manual or robotic cleaning operations. After triggering the component cleaning command, the system enters a waiting-to-clean state until a cleaning completion signal is received, at which point it resumes normal tracking mode.

[0094] In implementation, the determination of pulse swing parameters when generating the second control command is based on the following principle: the swing amplitude determines the intensity of airflow disturbance on the component surface, the swing period determines the disturbance frequency, and the number of swings determines the total duration of the disturbance. The three work together to achieve the best self-cleaning effect with minimal energy consumption.

[0095] The swing amplitude is determined by the difference between the current tilt angle of the support and the preset self-cleaning tilt angle range. A larger difference indicates a greater deviation of the current tilt angle from the optimal self-cleaning range, requiring a higher disturbance intensity; therefore, the swing amplitude increases with the increase of the difference. After a limited number of tests, the swing amplitude was set to ±5° when the difference was less than 10°; ±10° when the difference was between 10° and 20°; and ±15° when the difference was greater than 20°. This tiered setting provides sufficient disturbance intensity under different degrees of deviation, preventing self-cleaning failure due to insufficient swing amplitude or excessive motor wear due to excessive swing amplitude.

[0096] The oscillation period is determined based on wind speed from the environmental wind and rain parameters, decreasing as wind speed increases. The principle is that when the natural wind speed is high, the component surface already has sufficient conditions for airflow scouring; increasing the oscillation frequency at this time can create a wind-turbine synergistic effect, enhancing the disturbance effect. When the wind speed is low, the oscillation period needs to be appropriately extended to allow sufficient time for the component surface to form an airflow boundary layer before being disturbed and destroyed. After a limited number of tests, the oscillation period was set to 120 seconds when the wind speed was below 2 m / s; 60 seconds when the wind speed was between 2 m / s and 5 m / s; and 30 seconds when the wind speed was above 5 m / s.

[0097] The number of oscillations is determined either by setting a preset fixed value or by determining it based on the duration of the oscillation sub-diagram. The preset fixed value method is suitable for normal semi-self-cleaning conditions. After a limited number of tests and calibrations, it is set to 10 to 20 oscillations. This range allows for initial dust removal without significantly increasing motor losses. The method based on the duration of the oscillation sub-diagram is suitable for situations where the dust accumulation is prolonged and firmly adhered. For every hour the duration of the oscillation sub-diagram increases, the number of oscillations increases by 5, thereby enhancing the disturbance intensity until the dust accumulation improves. This adaptive adjustment mechanism dynamically matches the disturbance intensity according to the actual degree of dust accumulation, avoiding excessive or insufficient disturbance.

[0098] Please see Figure 3 As shown, the present invention also provides a system comprising:

[0099] The data acquisition module is used to acquire light-sensing data of photovoltaic modules from various angles, real-time power generation, energy consumption of the support tracking, environmental wind and rain parameters, and the current tilt angle of the support.

[0100] The intelligent feature generation module is used to collect the dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module at a preset tracking period, record the feature signals corresponding to the light sensing data at each angle as dust accumulation reference signals, obtain the tracking operation response signal corresponding to the time of the dust accumulation reference signal, generate a set of dust accumulation energy efficiency response curves based on the dust accumulation reference signal and the tracking operation response signal, analyze the set of dust accumulation energy efficiency response curves, determine the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module, and generate a photovoltaic dust accumulation energy efficiency time series diagram based on the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value.

[0101] The intelligent time series analysis module is used to divide the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determine the ash accumulation anomaly feature category based on the duration of the fluctuating subgraph;

[0102] The intelligent decision-making module is used to determine the target tracking and control strategy based on the category of the abnormal ash accumulation characteristics and the current tilt angle of the support, and to generate corresponding hierarchical control instructions;

[0103] The execution module is used to execute the graded control commands, control the tilt adjustment, frequency switching and pulse oscillation of the support, and output a cleaning start signal when the graded control command triggers component cleaning.

[0104] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0105] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A photovoltaic tracking bracket control method based on AI, characterized in that, include: Acquire light-sensing data of photovoltaic modules from various angles, real-time power generation, energy consumption of the support system, environmental wind and rain parameters, and the current tilt angle of the support system; The dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module are collected during the preset tracking period, and the characteristic signals corresponding to the light sensing data at each angle are recorded as dust accumulation reference signals. Acquire a tracking operation response signal corresponding to the time of the ash accumulation reference signal. The tracking operation response signal includes a transmittance attenuation sequence and a power generation gain change sequence. Generate a set of ash accumulation energy efficiency response curves based on the ash accumulation reference signal and the tracking operation response signal. The set of ash accumulation energy efficiency response curves is analyzed, and the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module are determined based on the analysis results. A photovoltaic ash accumulation energy efficiency time series diagram is generated based on the ash accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value. The photovoltaic ash accumulation energy efficiency time series diagram is divided into a stable subgraph and a fluctuating subgraph, and the ash accumulation anomaly characteristic category is determined according to the duration of the fluctuating subgraph. The ash accumulation anomaly characteristic category includes self-cleaning category, semi-self-cleaning category and non-self-cleaning category. Based on the category of abnormal dust accumulation characteristics and the current tilt angle of the support, a target tracking and control strategy is determined, and corresponding hierarchical control instructions are generated. The hierarchical control instructions include adjusting the tracking tilt angle of the support, switching the tracking frequency, and applying pulsed tilt angle reciprocating oscillation in combination with the environmental wind and rain parameters to enhance the self-cleaning state or trigger component cleaning.

2. The AI-based photovoltaic tracking bracket control method according to claim 1, characterized in that, Within the preset tracking period, the ratio of incident irradiance to reflected irradiance of each angle light-sensing data at each sampling time is extracted as an irradiance comparison feature value, and the ratio of edge transmittance to incident irradiance is extracted as a transmittance attenuation feature value. The irradiance comparison feature values ​​of all sampling times within the same preset tracking period are arranged in chronological order to form an irradiance comparison feature sequence. The light transmittance attenuation feature values ​​of all sampling times within the same preset tracking period are arranged in chronological order to form a light transmittance attenuation feature sequence. The irradiance comparison feature sequence and the light transmittance attenuation feature sequence are used together as the ash accumulation reference signal. Each sampling time corresponds one-to-one with the acquisition time of the ash accumulation response signal and the tracking energy efficiency response signal within the preset tracking period.

3. The AI-based photovoltaic tracking bracket control method according to claim 2, characterized in that, Determining the tracking operation response signal includes: The edge transmittance at each sampling time is extracted from the ash accumulation response signal and arranged in chronological order to form the transmittance attenuation sequence. The power generation time series at each sampling time is extracted from the tracking energy efficiency response signal, the change in power generation at each sampling time relative to the reference time is calculated, and the power generation gain change sequence is formed by arranging them in chronological order. The transmittance attenuation sequence and the power generation gain change sequence are aligned in the time dimension and combined to form the tracking operation response signal.

4. The AI-based photovoltaic tracking bracket control method according to claim 3, characterized in that, The process of generating a set of ash accumulation energy efficiency response curves based on the ash accumulation reference signal and the tracking operation response signal includes: The irradiation comparison feature sequence in the ash accumulation reference signal and the power generation gain change sequence in the tracking operation response signal are plotted with time as the horizontal axis as irradiation comparison curve and power generation gain change curve, respectively. The irradiation comparison curve and the power generation gain change curve are combined into an irradiation energy efficiency response curve group. The transmittance attenuation characteristic sequence in the ash accumulation reference signal, the transmittance attenuation sequence in the tracking operation response signal, and the power generation gain change sequence are plotted with time as the horizontal axis as the transmittance attenuation reference curve, the real-time transmittance attenuation curve, and the power generation gain change curve, respectively, and are denoted as the transmittance energy efficiency response curve group. The irradiation energy efficiency response curve set and the light transmission energy efficiency response curve set are used together as the ash accumulation energy efficiency response curve set.

5. The AI-based photovoltaic tracking bracket control method according to claim 4, characterized in that, The process of analyzing the set of ash accumulation energy efficiency response curves includes: Calculate the difference between the power generation gain change curve and the irradiation comparison curve in the irradiation energy efficiency response curve group at the same sampling time, and generate an irradiation energy efficiency difference sequence; Calculate the difference between the real-time transmittance decay curve and the transmittance decay reference curve in the transmittance energy efficiency response curve group at the same sampling time, and generate a transmittance decay difference sequence. Calculate the difference between the power generation gain change curve and the power generation gain baseline in the light transmission energy efficiency response curve group at the same sampling time, and generate a power generation energy efficiency difference sequence; The irradiance ratio dispersion characterization value, light transmission attenuation dispersion characterization value, and power generation gain dispersion characterization value are calculated based on the irradiance energy efficiency difference sequence, light transmission attenuation difference sequence, and power generation energy efficiency difference sequence, respectively.

6. The AI-based photovoltaic tracking bracket control method according to claim 5, characterized in that, The process of determining the ash self-cleaning characteristic value and the energy consumption power generation gain characterization value based on the analysis results includes: The self-cleaning characteristic value of ash accumulation is determined based on the dispersion characterization value of the irradiance ratio and the dispersion characterization value of the light transmittance attenuation. The power generation gain dispersion characterization value is used as the power generation gain characterization value for tracking energy consumption.

7. The AI-based photovoltaic tracking bracket control method according to claim 6, characterized in that, The process of dividing the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determining the category of ash accumulation anomaly characteristics based on the duration of the fluctuating subgraph includes: The curve segments of the self-cleaning characteristic value curve of the photovoltaic ash accumulation energy efficiency time series diagram that are within the preset stable benchmark range are marked as stable subgraphs, and the curve segments that are outside the preset stable benchmark range are marked as fluctuating subgraphs. Obtain the duration of the fluctuation subgraph and compare the duration with a preset duration threshold; Wherein, when the duration of the fluctuation subgraph is less than the preset duration threshold, it is determined to be a self-cleaning category; When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold, and the peak value of the self-cleaning feature value is less than the preset feature value threshold, it is determined to be a semi-self-cleaning category. When the duration of the fluctuation subgraph is greater than or equal to the preset duration threshold, and the peak value of the self-cleaning feature value is greater than or equal to the preset feature value threshold, it is determined to be a non-self-cleaning category.

8. The AI-based photovoltaic tracking bracket control method according to claim 7, characterized in that, The process of determining the target tracking and control strategy based on the ash accumulation anomaly characteristics and the current tilt angle of the support, and generating corresponding hierarchical control commands, includes: When the dust accumulation abnormality feature category is a self-cleaning category, a first control command is generated. The first control command includes adjusting the support tracking tilt angle to a preset self-cleaning tilt angle range and switching the tracking frequency to a low-frequency tracking mode. When the dust accumulation abnormality feature category is semi-self-cleaning, a second control command is generated. The second control command includes determining the pulse swing parameters based on the current tilt angle of the support and the environmental wind and rain parameters, and applying a pulse tilt angle reciprocating swing. The pulse swing parameters include swing amplitude, swing period and swing number. When the dust accumulation abnormality characteristic category is non-self-cleaning, a third control command is generated. The third control command includes locking the bracket at a preset safe tilt angle and triggering a component cleaning command.

9. The AI-based photovoltaic tracking bracket control method according to claim 8, characterized in that, When generating the second control command, the pulse swing parameter is determined as follows: The swing amplitude is determined based on the difference between the current tilt angle of the bracket and the preset self-cleaning tilt angle range, and the swing amplitude increases as the difference increases; The oscillation period is determined based on the wind speed in the environmental wind and rain parameters, and the oscillation period decreases as the wind speed increases; The number of oscillations can be set to a preset fixed value, or determined according to the duration of the oscillation subgraph.

10. A system applied to the AI-based photovoltaic tracking bracket control method according to any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire light-sensing data of photovoltaic modules from various angles, real-time power generation, energy consumption of the support tracking, environmental wind and rain parameters, and the current tilt angle of the support. The intelligent feature generation module is used to collect the dust accumulation response signal and tracking energy efficiency response signal of the photovoltaic module at a preset tracking period, record the feature signals corresponding to the light sensing data at each angle as dust accumulation reference signals, obtain the tracking operation response signal corresponding to the time of the dust accumulation reference signal, generate a set of dust accumulation energy efficiency response curves based on the dust accumulation reference signal and the tracking operation response signal, analyze the set of dust accumulation energy efficiency response curves, determine the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value of the photovoltaic module, and generate a photovoltaic dust accumulation energy efficiency time series diagram based on the dust accumulation self-cleaning characteristic value and the tracking energy consumption power generation gain characterization value. The intelligent time series analysis module is used to divide the photovoltaic ash accumulation energy efficiency time series diagram into a stable subgraph and a fluctuating subgraph, and determine the ash accumulation anomaly feature category based on the duration of the fluctuating subgraph; The intelligent decision-making module is used to determine the target tracking and control strategy based on the category of the abnormal ash accumulation characteristics and the current tilt angle of the support, and to generate corresponding hierarchical control instructions; The execution module is used to execute the graded control commands, control the tilt adjustment, frequency switching and pulse oscillation of the support, and output a cleaning start signal when the graded control command triggers component cleaning.

Images