Method for dynamic adjustment of photovoltaic array based on artificial intelligence shadow detection and prediction

Abstract

The application provides a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction. The method forms a multi-modal data set by collecting photovoltaic array operation and maintenance related data, uses quantumized feature enhancement and dynamic spatial attention mechanism to realize accurate identification of shadow area, determines shadow boundary coordinates and calculates shadow coverage. Further, according to the shadow boundary coordinates, the solar elevation angle and the shelter parameter, the shadow expansion speed and the propagation time are calculated, and the shadow coverage range is predicted. At the same time, the risk of hot spot is quantified by the temperature difference between the shadow area and the non-shadow area and the shadow duration, and the graded warning is triggered. Based on the shadow coverage rate, the photovoltaic array output voltage is dynamically adjusted, and the maximum power point is tracked through the particle swarm optimization algorithm. In addition, combined with the shadow prediction result, the angle of the photovoltaic panel is adjusted mechanically to avoid the shadow area. The application improves the power generation efficiency of the photovoltaic array under the condition of shadow shielding, and enhances the stability and safety of the system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent photovoltaic technology, in particular to a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction. BACKGROUND

[0002] With the increasing global energy demand and the growing emphasis on environmental protection, solar energy as a clean and renewable energy resource has received worldwide attention. Photovoltaic arrays, as an important device for converting solar energy into electrical energy, have been widely used in distributed power generation, large-scale photovoltaic power stations and other fields. However, in actual operation, photovoltaic arrays are often affected by surrounding environmental factors, such as shadowing by trees, buildings and other obstructions, which can cause local shading of photovoltaic cells and a series of problems.

[0003] Specifically, shadowing leads to reduced power generation efficiency, and local shading can change the output current and voltage of photovoltaic cells, resulting in a decrease in the output power of photovoltaic arrays. In severe cases, it may even cause hot spot effect, damaging photovoltaic cells and reducing their service life.

[0004] Currently, traditional shadow detection methods mainly rely on manual inspection or simple sensor monitoring, which have low efficiency, low detection accuracy and cannot predict in real time. For example, manual inspection requires a lot of manpower and time, and it is difficult to find dynamic changes in shadow conditions in time; simple sensor monitoring cannot accurately identify the boundaries and range of shadows, making it difficult to achieve accurate shadow detection.

[0005] Existing photovoltaic arrays lack effective dynamic adjustment mechanisms when facing shadowing. Most photovoltaic systems can only operate in fixed modes and cannot adjust the working state of photovoltaic arrays in real time according to changes in shadow, thus failing to minimize the impact of shadow on power generation efficiency.

[0006] Shadowing can also cause local temperature rise in photovoltaic cells, resulting in hot spot effect. Hot spot effect not only reduces the power generation efficiency of photovoltaic cells, but also accelerates the aging of the cells, and even causes fire hazards and other safety hazards. However, traditional hot spot detection and warning methods are often not timely and accurate enough to effectively prevent the occurrence of hot spot effect. SUMMARY

[0007] The present application aims to at least solve one of the above technical problems in the prior art.

[0008] To this end, the present application provides a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction.

[0009] This invention provides a method for dynamic adjustment of a photovoltaic array based on artificial intelligence shadow detection and prediction, comprising:

[0010] A multimodal dataset is formed by collecting photovoltaic array operation and maintenance related data; the photovoltaic array operation and maintenance related data includes the photovoltaic array's current, voltage, battery temperature, solar altitude angle, obstruction parameters, and infrared thermal values;

[0011] Based on a multimodal dataset, shadow regions are identified, shadow boundary coordinates are determined, and shadow coverage is calculated through quantized feature enhancement and dynamic spatial attention mechanisms.

[0012] The shadow expansion speed and propagation time are calculated based on the shadow boundary coordinates, solar altitude angle, and shading object parameters to predict the shadow coverage area;

[0013] By measuring the temperature difference between shaded and unshaded areas and the duration of shadow, hot spot risk can be quantified and graded early warnings can be triggered.

[0014] The output voltage of the photovoltaic array is dynamically adjusted based on the shadow coverage, and the maximum power point is tracked using a particle swarm optimization algorithm.

[0015] Based on the shadow prediction results, the angle of the photovoltaic panels is mechanically adjusted to avoid the shadow area.

[0016] The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to the above-described technical solution of the present invention may also have the following additional technical features:

[0017] In the above technical solution, the step of identifying shadow regions, determining shadow boundary coordinates, and calculating shadow coverage based on a multimodal dataset through quantized feature enhancement and dynamic spatial attention mechanisms includes:

[0018] Acquire image data of the photovoltaic array;

[0019] Quantum feature maps are obtained from image data based on photovoltaic arrays using a quantum feature extraction method, including:

[0020]

[0021] in, Represents the quantized eigenvalue, i.e., the th The quantum state, the first The shadow feature matrix of each channel is used to form the quantized feature map; Indicates the first The fusion weights of the quantum states; Represents the first input image One channel; Indicates the first a plurality of quantum state convolution kernels for extracting corresponding shadow patterns, each quantum convolution kernel corresponding to a shadow pattern, the shadow pattern including a lateral shadow, a longitudinal shadow, and an edge shadow; denotes a convolution operation;

[0022] based on a dynamic spatial attention mechanism, a spatial attention weight map is calculated according to the quantized feature map, the spatial attention weight being used to represent a probability that a coordinate point is in shadow, including:

[0023]

[0024] wherein, denotes a spatial attention weight at a coordinate , i.e., a probability that the coordinate is in shadow; denotes a Sigmoid activation function; denotes a 1x1 convolution weight; denotes a feature map height; denotes a feature map width; denotes a feature map pixel value, i.e., a quantized feature value of the coordinate , the first channel; denotes an infrared fusion coefficient; denotes an infrared thermal value at a coordinate ;

[0025] statistics of coordinate points with a spatial attention weight greater than a set threshold, and then calculating a shadow coverage rate, including:

[0026]

[0027] wherein, denotes a shadow coverage rate; denotes a spatial attention weight set threshold;

[0028] by an edge detection algorithm, region edge pixel coordinates satisfying a spatial attention weight greater than a set threshold are extracted from the spatial attention weight map to form a shadow boundary coordinate set.

[0029] In the above technical solution, the image data based on the photovoltaic array adopts a quantum feature extraction method to obtain a quantized feature map, including:

[0030] initializing quantum convolution kernels, so that each quantum convolution kernel corresponds to a shadow pattern;

[0031] performing quantum kernel convolution on RGB channels of the input image respectively to obtain a preliminary feature map;

[0032] Based on the preliminary feature map, multi-channel features are enhanced by quantum state fusion weighted fusion.

[0033] In the above technical solution, the step of calculating the spatial attention weight map based on the quantized feature map using the dynamic spatial attention mechanism includes:

[0034] Perform global average pooling on the quantized feature map to extract channel-level global information;

[0035] Initial spatial weights are generated through 1×1 convolution, and then normalized to [0,1] using the Sigmoid activation function;

[0036] By fusing infrared thermal values ​​and correcting the weight distribution, shadows and interference can be distinguished.

[0037] In the above technical solution, the step of calculating the shadow expansion speed and propagation time based on the shadow boundary coordinates, solar altitude angle, and occlusion parameters, and predicting the shadow coverage area, includes:

[0038] Based on the shadow boundary coordinate set, solar altitude angle, and occlusion parameters, the shadow spread rate is calculated, including:

[0039]

[0040] in, Indicates the speed at which the shadow expands; Represents the coordinate set of the shadow boundary; Representing coordinates The solar projection vector at that location;

[0041] The method for calculating the solar projection vector includes:

[0042]

[0043] in, Indicates the solar altitude angle; Indicates the horizontal distance from the obstruction; Indicates the coordinates of the base point of the obstruction;

[0044] The shadow propagation time is calculated based on the shadow boundary coordinate set and the shadow spread velocity, including:

[0045]

[0046] in, Indicates the time it takes for the shadow to spread; This indicates the center coordinates of the photovoltaic panel.

[0047] In the above technical solution, the step of quantifying hot spot risk and triggering graded early warnings by measuring the temperature difference between the shaded and unshaded areas and the duration of shadowing includes:

[0048] The temperature difference between the shadow area and the normal area is calculated to obtain a basic risk index, including:

[0049]

[0050]

[0051] wherein, represents the shadow temperature difference; represents the battery temperature at the coordinates represents the number of shadow area pixels; represents the number of non-shadow area pixels; represents the basic risk index; represents the maximum temperature for safe operation of the photovoltaic cell; represents the ambient temperature around the photovoltaic panel; The shadow duration and the propagation time are combined to amplify the risk value by a time weight to obtain a dynamic risk index, including:

[0052]

[0053] wherein,

[0054] represents the dynamic risk index; represents the time influence coefficient; represents the current shadow duration; The basic risk index and the dynamic risk index are compared with corresponding set threshold values, respectively, to determine the current hot spot risk level.

[0055] In the above technical solution, the photovoltaic array output voltage is dynamically adjusted based on the shadow coverage rate, and a particle swarm optimization algorithm is used to track the maximum power point, including:

[0056] When the shadow coverage rate exceeds a reference value, a voltage compensation amount is calculated, and the calculation method of the voltage compensation amount includes:

[0057]

[0058]

[0059] wherein, represents the voltage compensation amount; represents the rated voltage; represents the compensation intensity coefficient; represents the current shadow coverage rate; represents the reference value of the shadow coverage rate;

[0060] A particle swarm algorithm is used to dynamically adjust the working voltage at the next time point through the difference between the local optimal voltage and the global optimal voltage, including: ​

[0061]

[0062] wherein, represents the optimal voltage at the next moment; represents the current voltage; represents the learning rate; represents the total number of particles in the particle swarm algorithm; represents the influence coefficient of the jth particle; represents the locally optimal voltage searched by the jth particle; represents the globally optimal voltage; The learning rate and particle weight are updated in real time to improve the maximum power point tracking efficiency.

[0063] In the above technical solution, the shadow prediction result is used to adjust the angle of the photovoltaic panel mechanically to avoid the shadow area, including:

[0064] The target angle of the photovoltaic panel is calculated in combination with the height of the shelter, the solar elevation angle, and the length of the photovoltaic support, including:

[0065]

[0066]

[0067] wherein, represents the target angle of the photovoltaic panel; represents the height of the shelter; represents the solar elevation angle; represents the length of the photovoltaic support; represents the safety margin angle;

[0068] The rotation angle is calculated based on the current angle and the required angle of the photovoltaic panel, and the motor rotation instruction is generated based on the rotation angle, including:

[0069]

[0070] wherein, represents the rotation angle; represents the current angle of the photovoltaic panel;

[0071] The photovoltaic panel rotates to the target angle based on the rotation instruction to avoid the shadow.

[0072] In the above technical solution, it further includes:

[0073] The efficiency improvement rate is calculated by comparing the power output of the photovoltaic panel before and after voltage compensation and mechanical adjustment, and the dynamic adjustment result is evaluated.

[0074] In the above technical solution, it further includes:

[0075] ​Based on historical operating data, the quantum state weights are updated using gradient descent, including:

[0076]

[0077] in, Indicates the updated number The fusion weights of the quantum states; Indicates the parameter update rate; The gradient of the loss function is represented by L, where L is the error between the predicted result and the true value.

[0078] In summary, due to the adoption of the above-mentioned technical features, the beneficial effects of the present invention are:

[0079] First, this invention collects multimodal data related to the operation and maintenance of photovoltaic arrays, including current, voltage, cell temperature, solar altitude angle, shading object parameters, and infrared thermal values. It then utilizes quantum feature enhancement and dynamic spatial attention mechanisms to achieve accurate identification of shadowed regions, accurately determining shadow boundary coordinates and calculating shadow coverage. Compared to traditional methods, this approach significantly improves the accuracy and efficiency of shadow detection, enabling rapid and accurate identification of shadowed areas in complex shading environments, providing reliable data support for subsequent dynamic adjustments.

[0080] Secondly, this invention can calculate the shadow expansion speed and propagation time based on the shadow boundary coordinates, solar altitude angle, and shading object parameters, thereby predicting the shadow coverage area in advance. This predictive function allows photovoltaic systems to take preventative measures in advance, avoiding sudden impacts of shadows on power generation efficiency and enhancing system stability and reliability. By predicting the shadow propagation path and range in advance, the system can more rationally plan the layout and adjustment strategies of photovoltaic panels, minimizing the impact of shading on power generation efficiency.

[0081] Furthermore, this invention quantifies hot spot risk by measuring the temperature difference between shaded and unshaded areas and the duration of shadowing, and triggers tiered early warnings. This hot spot risk assessment method can promptly identify potential hot spot problems and take corresponding early warning measures based on the risk level. This not only helps protect photovoltaic cells from the damage caused by hot spot effects and extends their lifespan, but also effectively prevents safety accidents caused by hot spot effects, improving the safety and reliability of photovoltaic systems.

[0082] Furthermore, this invention dynamically adjusts the output voltage of the photovoltaic array based on shading coverage and tracks the maximum power point using a particle swarm optimization algorithm. This method can adjust the operating state of the photovoltaic array in real time under shading conditions, ensuring that the system always operates near the maximum power point, thereby maximizing power generation efficiency. Compared to the traditional fixed operating mode, this dynamic adjustment mechanism can better adapt to shading changes, improving the overall performance and economy of the system.

[0083] Finally, the application also avoids the shadow area by mechanically adjusting the angle of the photovoltaic panel, further improving the power generation efficiency of the photovoltaic system. The target angle of the photovoltaic panel is calculated by combining the height of the shelter, the solar elevation angle and the length of the photovoltaic support, and a motor rotation instruction is generated, so that the photovoltaic panel can be automatically adjusted to the optimal angle to avoid shadow blocking. This mechanical adjustment combined with electrical adjustment realizes the all-round optimization of the photovoltaic system, further improving the power generation efficiency and operation stability of the system.

[0084] In summary, the application realizes efficient operation of the photovoltaic array under the condition of shadow blocking, improves the power generation efficiency, enhances the stability and safety of the system, has significant economic and social benefits, and provides strong technical support for the development of the photovoltaic industry.

[0085] Additional aspects and advantages of the application will become apparent from the following description with reference to the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0086] The above and / or additional aspects and advantages of the application will become apparent and be readily understood by a person of ordinary skill in the art from the following description, taken in conjunction with the accompanying drawings in which:

[0087] Figure 1 is a flowchart of a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to an embodiment of the application;

[0088] Figure 2 is a dynamic adjustment schematic diagram of a photovoltaic panel movable support in a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to an embodiment of the application. DETAILED DESCRIPTION

[0089] In order to more clearly understand the above-mentioned purposes, features and advantages of the application, the application will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that the embodiments of the application and the features in the embodiments can be combined with each other without conflict.

[0090] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, however, the application can also be implemented in other ways different from those described herein, therefore, the scope of protection of the application is not limited by the specific embodiments disclosed below.

[0091] The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to some embodiments of the application will be described below with reference to Figure 1 and Figure 2 .

[0092] Some embodiments of the present application provide a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction.

[0093] As shown in Figure 1 The first embodiment of the present application proposes a photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction, including the following steps S1-S6.

[0094] S1, collect photovoltaic array operation related data to form a multi-modal data set; the photovoltaic array operation related data includes current, voltage, battery temperature, solar elevation angle, shelter parameter and infrared thermal value of the photovoltaic array.

[0095] Specifically, the multi-modal data can be acquired by a sensor network, which can be composed of current sensors, voltage sensors, temperature sensors, solar angle sensors, infrared thermographs, etc., and several kinds of multi-modal data are fused into a multi-modal data set containing time stamps through a data fusion module , wherein, is the output current of the photovoltaic array, i.e. the real-time output current value of the photovoltaic panel, which is used to determine whether the circuit is working normally, and an abnormal value can trigger a short circuit / disconnection warning; is the output voltage of the photovoltaic array, i.e. the real-time output voltage value of the photovoltaic panel, which is used as the core input for power calculation and MPPT (maximum power point tracking) optimization; is the surface temperature of the photovoltaic cell, i.e. the temperature distribution data of the surface of the photovoltaic panel, which can be used for hot spot risk assessment; is the parameter of the shelter (such as trees, buildings, etc.), which records the height and straight-line distance of the shelter to the photovoltaic panel; is the infrared thermal radiation intensity, i.e. the thermal radiation energy per unit area of the surface of the photovoltaic panel.

[0096] In one specific embodiment, in a large ground photovoltaic power station, data is collected by a distributed sensor network every 100ms to form a real-time state matrix covering the entire array.

[0097] S2, based on the multi-modal data set, realize the identification of the shadow area through quantumized feature enhancement and dynamic spatial attention mechanism, determine the shadow boundary coordinates and calculate the shadow coverage rate.

[0098] In some embodiments, step S2 includes the following steps S21-S25.

[0099] S21, obtain image data of the photovoltaic array; the image data of the photovoltaic array is the real-time collected surface image of the photovoltaic panel, which is used as the input image of the subsequent steps, usually an RGB image.

[0100] S22. Quantum feature maps are obtained from image data based on photovoltaic arrays using a quantum feature extraction method, including:

[0101]

[0102] in, Represents the quantized eigenvalue, i.e., the th The quantum state, the first The shadow feature matrix of each channel is used to form the quantized feature map; Indicates the first The fusion weights of the quantum states; Represents the first input image Each channel has an RGB image input. Corresponding to R, G, B; Indicates the first A 3×3 convolution kernel for each quantum state, typically eight quantum convolution kernels, is used to extract the corresponding shadow pattern. Each quantum convolution kernel corresponds to a shadow pattern, which includes horizontal shadow, vertical shadow, and edge shadow, etc. This indicates a convolution operation.

[0103] In one specific embodiment, the execution process of step S22 is as follows:

[0104] Initialize 8 quantum convolution kernels, so that each quantum convolution kernel corresponds to a shadow mode;

[0105] Quantum kernel convolution is performed on the RGB channels of the input image to obtain preliminary feature maps;

[0106] Based on the preliminary feature map, multi-channel features are fused by quantum state fusion weighting to enhance weak shadow signals (such as shadows on cloudy days).

[0107] S23. Based on the dynamic spatial attention mechanism, calculate the spatial attention weight map according to the quantized feature map, wherein the spatial attention weight is used to represent the probability that the corresponding coordinate point is shaded, including:

[0108]

[0109] in, Representing coordinates Spatial attention weights at a location, i.e., coordinates The probability that a certain area is shaded; This represents the Sigmoid activation function; Indicates 1×1 convolution weights; This represents the feature map height, i.e., the number of vertical pixels in the quantized feature map; This represents the width of the feature map, which is the number of horizontal pixels in the quantized feature map. represents a feature image pixel value, that is, a coordinate , the first quantized feature value of the channel; represents an infrared fusion coefficient for adjusting the proportion of the infrared thermal value in the attention weight; represents the infrared thermal value at the coordinate .

[0110] In one specific embodiment, the execution process of step S23 is as follows:

[0111] Global average pooling is performed on the quantumized feature map to extract channel-level global information;

[0112] An initial spatial weight is generated by 1x1 convolution, and the initial spatial weight is normalized to [0, 1] by a Sigmoid activation function;

[0113] Based on the low thermal value characteristics of the shadow area, the infrared thermal value is fused to correct the weight distribution and distinguish shadows and interference objects. The interference objects can be bird droppings, dust, etc.

[0114] S24, statistics of coordinate points with spatial attention weight greater than a set threshold, and then calculation of shadow coverage, including:

[0115]

[0116] wherein, represents the shadow coverage; represents the spatial attention weight set threshold, which can be set to 0.7.

[0117] Specifically, the shadow coverage is mainly used to identify the range of a single photovoltaic panel currently covered by a shadow, and the proportion is controlled within a specified range, usually an absolute safety threshold (0%~5%), a warning threshold (5%~15%), and a danger threshold (>15%).

[0118] S25, by an edge detection algorithm, extracting region edge pixel coordinates satisfying the spatial attention weight greater than the set threshold from the spatial attention weight map to form a shadow boundary coordinate set.

[0119] During overcast or dawn, the contrast between the shadow and non-shadow areas on the surface of the photovoltaic panel is low (traditional methods are prone to missed detection), and the recognition accuracy can be significantly improved by quantumized feature enhancement, while the shadow boundary coordinates are accurately extracted. In one specific embodiment, based on the above detection method, the shadow detection accuracy is improved from 82% of the traditional method to more than 98%, the misjudgment rate is reduced from 15% to less than 3%, the extracted shadow boundary coordinates provide key data support for subsequent shadow propagation prediction, and the daily power loss of a single module is reduced by 0.5kWh.

[0120] S3, calculating the shadow propagation speed and the shadow propagation time according to the shadow boundary coordinates, the sun elevation angle and the shelter parameter, and predicting the shadow coverage range.

[0121] In some embodiments, step S3 comprises steps S31 and S32.

[0122] S31, calculating the shadow propagation speed according to the shadow boundary coordinate set, the sun elevation angle and the shelter parameter, comprising:

[0123]

[0124] wherein, represents the shadow propagation speed, i.e. the moving distance of the shadow edge per unit time; represents the shadow boundary coordinate set; represents the sun projection vector at the coordinate , i.e. the projection direction vector of the sunray on the photovoltaic panel plane;

[0125] The calculation method of the sun projection vector comprises:

[0126]

[0127] wherein, represents the sun elevation angle, i.e. the angle between the sunray and the horizontal plane; represents the shelter horizontal distance, i.e. the straight-line distance from the shelter to the photovoltaic panel; represents the shelter base point coordinate, i.e. the pixel coordinate of the center of the bottom of the shelter in the image.

[0128] S32, calculating the shadow propagation time according to the shadow boundary coordinate set and the shadow propagation speed, comprising:

[0129]

[0130] wherein, represents the shadow propagation time; represents the photovoltaic panel center coordinate.

[0131] In one specific embodiment, the execution flow of step S3 is as follows:

[0132] calculating the sun projection vector based on the sun elevation angle;

[0133] combining the shelter horizontal distance and the shadow boundary coordinates obtained in step S2, calculating the moving speed of each point of the shadow edge, i.e. the shadow propagation speed;

[0134] taking the shortest distance from the shadow boundary obtained in step S2 to the photovoltaic panel center, dividing by the average shadow propagation speed, to obtain the shadow propagation time.

[0135] In one specific embodiment, during sunrise (7-9 AM), tree shadows spread into the photovoltaic array. Using the shadow boundary coordinates extracted in step S2, the shadow expansion speed (e.g., 0.15 m / s) and propagation time (e.g., 180 s) can be predicted, allowing for a response time for photovoltaic panel adjustments. Specifically, the shadow coverage area can be predicted 30-60 seconds in advance to reduce power fluctuations. When the shadow propagation time is less than 60 seconds, an emergency adjustment is triggered, which can reduce power loss by 5%-8%.

[0136] S4. Quantify hot spot risk and trigger graded early warning by measuring the temperature difference between shaded and unshaded areas and the duration of shadow.

[0137] In some embodiments, step S4 includes steps S41-S43.

[0138] S41. Calculate the temperature difference between the shaded area and the normal area to obtain the basic risk index, including:

[0139]

[0140]

[0141] in, This represents the shade temperature difference, which is the average temperature difference between the shaded area and the unshaded area. Representing coordinates Battery temperature at the location; Indicates the number of pixels in the shadow area; Indicates the number of pixels in the non-shaded area; This represents the basic risk index, which is the quantitative value of the hot spot risk caused by static shading; Indicates the highest temperature at which photovoltaic cells can operate safely; This indicates the ambient temperature around the photovoltaic panel;

[0142] S42. Combining the shadow duration and propagation time, amplifying the risk value through time weighting yields a dynamic risk index, including:

[0143]

[0144] in, This represents a dynamic risk index, which is a comprehensive risk quantification value that takes into account the duration of the shadow. This represents the time-related influence coefficient, which can be taken as 0.5. Indicates the current duration of the shadow;

[0145] S43. Compare the basic risk index and dynamic risk index with the corresponding set thresholds to determine the current hot spot risk level.

[0146] For example, when Cooling measures are triggered at times. High-priority alerts are triggered at certain times.

[0147] In one specific embodiment, if a tree branch shades 1 / 4 of the area of ​​a single module (for 60 seconds), and the calculated shade temperature difference is 20°C, with a base risk index of 1.2 and a dynamic risk index of 1.8, a high-priority warning is triggered. Based on this, it is possible to prevent the cells from burning due to localized high temperatures (>85°C) and reduce the aging rate of the module (a hot spot lasting for 1 hour can shorten the module's lifespan by 3 months).

[0148] S5. The output voltage of the photovoltaic array is dynamically adjusted based on the shadow coverage, and the maximum power point is tracked through the particle swarm optimization algorithm.

[0149] In some embodiments, step S5 includes the following steps S51-S53.

[0150] S51. When the shading coverage exceeds the reference value, calculate the voltage compensation amount. The calculation method for the voltage compensation amount includes:

[0151]

[0152] in, Indicates the voltage compensation amount; Indicates the rated voltage; This represents the compensation strength coefficient, which can be taken as 0.4; Indicates the current shadow coverage; The baseline value representing the shading coverage rate, i.e. the minimum shading coverage rate for starting voltage compensation, can be taken as 0.2;

[0153] S52. Using the particle swarm optimization algorithm, the operating voltage at the next moment is dynamically adjusted based on the difference between the local optimum and the global optimum voltage, including:

[0154]

[0155] in, This indicates the optimal voltage at the next moment, i.e., the adjusted MPPT operating point voltage; This indicates the current voltage, i.e., the current operating voltage of the MPPT; Indicates the learning rate; This represents the total number of particles in the particle swarm optimization algorithm. This represents the influence coefficient of the j-th particle; Indicates the first The local optimal voltage found by each particle; Indicates the globally optimal voltage;

[0156] It should be noted that the specific details of the MPPT algorithm are well known to those skilled in the art and will not be elaborated upon here.

[0157] S53, updating the learning rate and particle weight in real time to improve the efficiency of maximum power point tracking.

[0158] Specifically, when the shadow coverage is 25% (exceeding the benchmark value 0.2), the voltage is adjusted from 30V to 32.5V, combined with MPPT optimization to restore the power to more than 80% of the rated value. Each 1V compensation can improve the power output by 3%-5%, the MPPT tracking efficiency is improved from 85% of the traditional method to 95%, and the daily average power generation of a single component is about 2kWh.

[0159] S6, based on the shadow prediction result, adjusting the angle of the photovoltaic panel by mechanical adjustment to avoid the shadow area.

[0160] In some embodiments, step S6 includes steps S61-S63.

[0161] S61, combining the height of the shelter, the solar elevation angle and the length of the photovoltaic support, calculating the target angle of the photovoltaic panel, including:

[0162]

[0163] wherein, represents the target angle of the photovoltaic panel; represents the height of the shelter; represents the solar elevation angle; represents the length of the photovoltaic support; represents the safety margin angle;

[0164] S62, based on the current angle and the required angle of the photovoltaic panel, calculating the rotation angle, and generating a motor rotation instruction based on the rotation angle, including:

[0165]

[0166] wherein, represents the rotation angle; represents the current angle of the photovoltaic panel;

[0167] S63, the photovoltaic panel rotates to the target angle based on the rotation instruction to avoid the shadow. Specifically, as shown in Figure 2 the existing photovoltaic support can already meet the dynamic adjustment of the photovoltaic panel, and the control instruction is transmitted to the corresponding stepping motor or hydraulic device, so that the photovoltaic panel rotating shaft is adjusted to the target position. It can be understood that the adjustment of the photovoltaic panel is not only through rotation, but also through height adjustment (up and down movement) and position translation (cross movement) and the like.

[0168] For example, when a tree's shadow covers a photovoltaic panel, if the target angle of the panel is calculated to be 25° and the current angle is 12.7°, the drive motor will rotate 12.3° to completely avoid the shadow. This mechanical adjustment can restore 90% of the light-receiving area, increasing the daily power generation of a single module by approximately 1.2 kWh, with a response time of less than 2 seconds, making it 30 times more efficient than manual adjustment.

[0169] In some embodiments, the photovoltaic array dynamic adjustment method further includes:

[0170] By comparing the power output of the photovoltaic panel before and after voltage compensation and mechanical adjustment, the efficiency improvement rate is calculated, and the dynamic adjustment results are evaluated.

[0171] Specifically, the calculation method for efficiency improvement rate is as follows:

[0172]

[0173] in, Indicates the efficiency improvement rate; This represents the output power after compensation, i.e., the optimized actual power generation. This represents the output power before compensation, i.e., the actual power before optimization. This refers to the nominal maximum power of the photovoltaic system.

[0174] In some embodiments, the photovoltaic array dynamic adjustment method further includes:

[0175] Based on real historical operating data, the quantum state weights are updated using the gradient descent method, including:

[0176]

[0177] in, Indicates the updated number The fusion weights of the quantum states; Indicates the parameter update rate; The gradient of the loss function is represented by L, where L is the error between the predicted result and the true value.

[0178] After one month of system operation, the quantum state weights were optimized using real data, maintaining the shadow detection accuracy above 95% and increasing annual power generation by 5% to 10%. Simultaneously, the model can adapt to seasonal changes (such as changes in shadow patterns after trees shed their leaves in winter), maintaining high detection accuracy over the long term.

[0179] In this specification, the illustrative expressions of the terms used do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0180] Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for dynamic adjustment of a photovoltaic array based on artificial intelligence shadow detection and prediction, characterized in that, include: A multimodal dataset is formed by collecting photovoltaic array operation and maintenance related data; the photovoltaic array operation and maintenance related data includes the photovoltaic array's current, voltage, battery temperature, solar altitude angle, obstruction parameters, and infrared thermal values; Based on a multimodal dataset, shadow regions are identified, shadow boundary coordinates are determined, and shadow coverage is calculated through quantized feature enhancement and dynamic spatial attention mechanisms. The shadow expansion speed and propagation time are calculated based on the shadow boundary coordinates, solar altitude angle, and shading object parameters to predict the shadow coverage area; By measuring the temperature difference between shaded and unshaded areas and the duration of shadow, hot spot risk can be quantified and graded early warnings can be triggered. The output voltage of the photovoltaic array is dynamically adjusted based on the shadow coverage, and the maximum power point is tracked using a particle swarm optimization algorithm. Based on the shadow prediction results, the angle of the photovoltaic panels is mechanically adjusted to avoid the shadow area; The method, based on a multimodal dataset, utilizes quantized feature enhancement and dynamic spatial attention mechanisms to identify shadow regions, determine shadow boundary coordinates, and calculate shadow coverage, including: Acquire image data of the photovoltaic array; Quantum feature maps are obtained from image data based on photovoltaic arrays using a quantum feature extraction method, including: in, Represents the quantized eigenvalue, i.e., the th The quantum state, the first The shadow feature matrix of each channel is used to form the quantized feature map; Indicates the first The fusion weights of the quantum states; Represents the first input image One channel; Indicates the first A quantum convolution kernel is used to extract the corresponding shadow pattern. Each quantum convolution kernel corresponds to a shadow pattern, which includes horizontal shadow, vertical shadow and edge shadow. Indicates the convolution operation; Based on the dynamic spatial attention mechanism, a spatial attention weight map is calculated from the quantized feature map. The spatial attention weights represent the probability that a corresponding coordinate point is shaded, including: in, Representing coordinates Spatial attention weights at a location, i.e., coordinates The probability that a certain area is shaded; This represents the Sigmoid activation function; Indicates 1×1 convolution weights; Indicates the feature map height; Indicates the width of the feature map; Represents the pixel values ​​of the feature map, i.e., coordinates. , No. The quantized eigenvalues ​​of the channel; Indicates the infrared fusion coefficient; Representing coordinates Infrared thermal value at the location; The coordinates of points whose spatial attention weight is greater than a set threshold are used to calculate the shadow coverage, including: in, Indicates shadow coverage; This indicates that the spatial attention weights are set to a threshold. Using an edge detection algorithm, the edge pixel coordinates of regions whose spatial attention weights are greater than a set threshold are extracted from the spatial attention weight map to form a set of shadow boundary coordinates. The calculation of shadow expansion speed and propagation time based on shadow boundary coordinates, solar altitude angle, and occlusion parameters, and the prediction of shadow coverage area, includes: Based on the shadow boundary coordinate set, solar altitude angle, and occlusion parameters, the shadow spread rate is calculated, including: in, Indicates the speed at which the shadow expands; Represents the coordinate set of the shadow boundary; Representing coordinates The solar projection vector at that location; The method for calculating the solar projection vector includes: in, Indicates the solar altitude angle; Indicates the horizontal distance from the obstruction; Indicates the coordinates of the base point of the obstruction; The shadow propagation time is calculated based on the shadow boundary coordinate set and the shadow spread velocity, including: in, Indicates the time it takes for the shadow to propagate; Indicates the center coordinates of the photovoltaic panel; The method of quantifying hot spot risk and triggering graded early warnings by measuring the temperature difference between shaded and unshaded areas and the duration of shadowing includes: Calculate the temperature difference between the shaded area and the normal area to obtain the basic risk index, including: in, Indicates the temperature difference between the shade and the ground; Representing coordinates Battery temperature at the location; Indicates the number of pixels in the shadow area; Indicates the number of pixels in the non-shaded area; Indicates the basic risk index; Indicates the highest temperature at which photovoltaic cells can operate safely; This indicates the ambient temperature around the photovoltaic panel; By combining the duration of the shadow with its propagation time, and amplifying the risk value through time weighting, a dynamic risk index is obtained, including: in, Indicates a dynamic risk index; Indicates the time influence coefficient; Indicates the current duration of the shadow; The basic risk index and dynamic risk index are compared with the corresponding set thresholds to determine the current hot spot risk level.

2. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, The image data based on the photovoltaic array is used to obtain a quantized feature map using a quantized feature extraction method, including: Initialize the quantum convolution kernels so that each quantum convolution kernel corresponds to a shadow mode; Quantum kernel convolution is performed on the RGB channels of the input image to obtain preliminary feature maps; Based on the preliminary feature map, multi-channel features are enhanced by quantum state fusion weighted fusion.

3. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, The method based on dynamic spatial attention mechanism, which calculates spatial attention weight map according to quantized feature map, includes: Perform global average pooling on the quantized feature map to extract channel-level global information; Initial spatial weights are generated through 1×1 convolution, and then normalized to [0,1] using the Sigmoid activation function; By fusing infrared thermal values ​​and correcting the weight distribution, shadows and interference can be distinguished.

4. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, The method of dynamically adjusting the output voltage of the photovoltaic array based on shading coverage and tracking the maximum power point using a particle swarm optimization algorithm includes: When the shading coverage exceeds the reference value, the voltage compensation amount is calculated. The calculation method for the voltage compensation amount includes: in, Indicates the voltage compensation amount; Indicates the rated voltage; Indicates the compensation strength coefficient; Indicates the current shadow coverage; A baseline value representing shadow coverage; The particle swarm optimization algorithm is used to dynamically adjust the operating voltage at the next time step by measuring the difference between the local optimum and the global optimum voltage. This includes: in, Indicates the optimal voltage at the next moment; Indicates the current voltage; Indicates the learning rate; This represents the total number of particles in the particle swarm optimization algorithm. This represents the influence coefficient of the j-th particle; Indicates the first The local optimal voltage found for each particle; Indicates the globally optimal voltage; The learning rate and particle weights are updated in real time to improve the efficiency of maximum power point tracking.

5. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, The method of avoiding shadowed areas by mechanically adjusting the angle of the photovoltaic panels based on shadow prediction results includes: By combining the height of the obstruction, the solar altitude angle, and the length of the photovoltaic support frame, the target angle of the photovoltaic panel is calculated, including: in, Indicates the target angle of the photovoltaic panel; Indicates the height of the obstruction; Indicates the solar altitude angle; Indicates the length of the photovoltaic support structure; Indicates the safety margin angle; Based on the current angle and the required angle of the photovoltaic panel, calculate the rotation angle, and generate motor rotation commands based on the rotation angle, including: in, Indicates the rotation angle; Indicates the current angle of the photovoltaic panel; The photovoltaic panel rotates to the target angle based on a rotation command to avoid shadows.

6. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, Also includes: By comparing the power output of the photovoltaic panel before and after voltage compensation and mechanical adjustment, the efficiency improvement rate is calculated, and the dynamic adjustment results are evaluated.

7. The photovoltaic array dynamic adjustment method based on artificial intelligence shadow detection and prediction according to claim 1, characterized in that, Also includes: Based on historical operating data, the quantum state weights are updated using gradient descent, including: in, Indicates the updated number The fusion weights of the quantum states; Indicates the parameter update rate; The gradient of the loss function is represented by L, where L is the error between the predicted result and the true value.

Images