A folding photovoltaic module deployment control method based on environmental perception

Abstract

The application discloses a kind of based on environmental perception's foldable photovoltaic module deployment control method, it is related to photovoltaic module control technical field, the steps of this method include: obtaining environmental data and photovoltaic module mechanical state by sensor network;Satellite image data in the environmental data is extracted, to obtain environmental occlusion feature, output future environmental occlusion loss coefficient;By the photovoltaic module mechanical state, assess future deployment structure risk index;Based on the environmental occlusion loss coefficient and deployment structure risk index, establish benefit-loss model, the benefit-loss model is used to output future deployment benefit;Set photovoltaic panel action control algorithm, future deployment benefit is taken as input, and the deployment control parameter of foldable photovoltaic module is output.The application solves the problem that the existing control strategy is mostly to maximize instantaneous power generation as target, ignores equipment loss and safety risk.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic module control technology, specifically to a method for controlling the unfolding of a foldable photovoltaic module based on environmental perception. Background Technology

[0002] With the increasing global demand for renewable energy, photovoltaic power generation, as a clean and carbon-free energy technology, has attracted much attention. However, traditional rigid photovoltaic modules have significant limitations in terms of size, weight, and deployment flexibility, often requiring substantial site support and transportation costs, making them difficult to popularize in highly mobile, irregularly shaped, or multifunctional application scenarios.

[0003] In photovoltaic control scenarios, it is necessary to quickly, online or in near real-time predict cloud shadow movement and its shading impact on specific sites. Existing photovoltaic shading assessments mostly use static geometry or simplified view factor models, relying only on empirical height or two-dimensional projection, ignoring the three-dimensional deformation of surrounding tree canopies, buildings and other elements, as well as the dynamic shadow caused by the change of the sun's position throughout the day. This results in a large deviation in the estimation of shading loss, which is particularly evident in complex terrain or urban environments, easily causing power generation prediction errors and bringing opportunity costs or safety risks.

[0004] Environmental changes (such as temporary cloud cover or wind speed fluctuations) can occur within seconds or minutes. If the control system adjusts its speed instantly for each minute change, it may cause frequent movements (jitter), resulting in mechanical wear and energy waste. Furthermore, for mobile photovoltaic modules, the wear caused by the deployment process is irreversible. If the current scenario is geared towards short-term power generation needs, and the power generation gain from continued deployment is not balanced with the jitter loss during deployment, the power generation cost will be less than the revenue, resulting in an imbalance between power generation gain and losses.

[0005] Therefore, the present invention provides a method for controlling the unfolding of foldable photovoltaic modules based on environmental perception. Summary of the Invention

[0006] The purpose of this invention is to provide a method for controlling the unfolding of foldable photovoltaic modules based on environmental perception, so as to solve the existing problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling the unfolding of a foldable photovoltaic module based on environmental perception, comprising the following steps:

[0008] S1. Acquire environmental data and the mechanical status of photovoltaic modules through a sensor network;

[0009] S2. Extract satellite imagery data from the environmental data to obtain environmental occlusion features and output future time. Environmental shading loss coefficient ;

[0010] S3. Assess future time based on the mechanical state of the photovoltaic modules. Expanding structural risk indicators ;

[0011] S4. Based on the environmental shading loss coefficient and the deployed structure risk index, establish a revenue-loss model, which is used to output future time... The benefits of expansion;

[0012] S5. Set the photovoltaic panel motion control algorithm, and use future time... The unfolding benefits are taken as input, and the unfolding control parameters of the foldable photovoltaic module are output.

[0013] A further improvement of the present invention is that the environmental occlusion features include local static occlusion features and cloud dynamic occlusion features;

[0014] The local static shading prediction features include acquiring potential shading bodies, which include tree canopy features and building features; the tree canopy features include extracting multispectral data from satellite imagery, calculating the vegetation index within m meters centered on the photovoltaic module, and obtaining the canopy height by inverting vegetation height information through an SVM model; the building features are obtained by extracting building heights by combining high-resolution image classification and DSM data; outputting a three-dimensional surface feature set to form a potential shading body height model; and calculating the changes in solar altitude angle and azimuth angle over time based on the coordinates of the target photovoltaic module.

[0015] For each potential shading object, calculate its shadow area on the plane where the target photovoltaic module is located based on the sun's position; repeatedly project every 15 minutes throughout the day to obtain the spatiotemporal distribution of the shadow, and output the static shading shadow layer corresponding to different time points within the day; extract the predicted future time. Static occlusion shadow layer and future time The overlapping area of ​​the photovoltaic module's unfolded area is calculated, and its proportion in the total unfolded area is denoted as the local static shading feature.

[0016] A further improvement of this invention is that the dynamic cloud occlusion feature includes training satellite cloud images using a pre-trained cloud detection model to obtain the cloud mask at the current time t, the cloud mask area, and marking the pixel region where the cloud is located; and applying an optical flow algorithm to extract the cloud motion field from two consecutive frames of satellite cloud images. The extracted cloud motion vectors are smoothed and filtered to remove outliers; large-scale cloud cluster motion direction and velocity are obtained through region clustering; and for the center location of each cloud cluster... Cloudtop height and cloud motion vector To obtain the future location of the cloud shadow on the ground. The cloud cover area is recorded. If the center or boundary path of the cloud shadow intersects with the coordinates of the photovoltaic module, dynamic cloud shading is determined to exist. When the coordinates of the photovoltaic module are shaded, the time is recorded. The proportion of time the cloud is covered is used to obtain the dynamic occlusion characteristics of the cloud.

[0017] A further improvement of the present invention is that the environmental occlusion loss coefficient is obtained by weighted summation of the local static occlusion features and the cloud dynamic occlusion features.

[0018] A further improvement of this invention lies in that the unfolding structural risk index is calculated by extracting the mechanical state of the photovoltaic module; the new round of photovoltaic panel unfolding action is taken as the current unfolding action, and the mechanical state of the photovoltaic module includes the vibration threat index and temperature change index of the photovoltaic panel, and historical actions and the number of historical actions are recorded; the mechanical state of the photovoltaic module is concatenated by feature dimensions to form a dual-channel damage input tensor, and the historical dual-channel damage input tensor and the corresponding historical unfolding structural risk index are used as training data. The XGBoost model outputs the predicted value of the risk index at the current unfolding angle and outputs the model confidence score. ;

[0019] Get the current unfolded ratio of the photovoltaic panel ; Combine the current operational risk indicators with the deployment ratio The ratio of the two values ​​is used as the predicted deployment risk value; and the sum of the vibration threat indicators of historical deployment actions is extracted, standardized, and then weighted and summed with the predicted deployment risk value to obtain the predicted deployment structural risk index.

[0020] A further improvement of this invention is that the benefit-loss model includes extracting meteorological data from the environmental data to obtain future time data. Solar irradiance And calculate future time Environmental characteristics With unfolded area Through solar irradiance Environmental characteristics With unfolded area The product of time and future time is obtained. Power generation forecast Calculate the current power generation in the same way. Calculate future time Power generation forecast Compared with current power generation Integral of the difference Obtain the predicted power generation gain By combining the predicted power generation gain with the corresponding environmental shading loss coefficient and the unfolded structural risk index, a revenue-loss model is generated, denoted as the future time... Expanding benefits ,in, , , and They represent the weights, , .

[0021] A further improvement of this invention is that the photovoltaic panel motion control algorithm is based on the annealing algorithm, combined with the unfolding structure risk index, to output photovoltaic panel unfolding motion commands. The specific construction process includes:

[0022] S51. Define each state in the solution space as... , Indicates the deployment speed; and obtains the predicted value of the deployment structure risk index. and model confidence in step S3 Set a hard threshold for risk. ;

[0023] S52. Use the revenue-loss model from step S4 as the objective function. ;

[0024] S53, in time state Randomly sample m candidate states from the vicinity Calculate the corresponding expanded structural risk indicators Input the sample risk dataset and calculate the spread return difference. ;

[0025] S54. Set the initial temperature adaptive strategy and obtain the initial temperature. ;

[0026] S55. Set a risk-driven adaptive cooling strategy and update the temperature iteratively;

[0027] S56. When the current temperature T is less than the set minimum temperature threshold, the operation will terminate safely and output the state corresponding to the maximum value of the objective function. If the increase value of the objective function is less than the set profit increase threshold in three consecutive temperature updates and the average unfolded structure risk index of the candidate state does not decrease, the current state is determined to be in an invalid or high-risk area, and the unfolding action will be terminated.

[0028] A further improvement of this invention is that the initial temperature adaptive strategy includes setting the expanded structural risk index of the sample risk dataset to be less than or equal to the risk hard threshold. The state samples are sent to the low-risk group, and the structural risk index will be greater than the risk hard threshold. State samples are sent to the high-risk group; acceptance probabilities are assigned to state samples in the low-risk group. Assign acceptance probabilities to state samples of the high-risk group. ;

[0029] Calculate the average absolute value of the spread returns of the low-risk group. And set the initial temperature of the low-risk state sample to be Then the initial temperature of all state samples is expressed as: ;

[0030] A further improvement of this invention is that the risk-driven adaptive cooling strategy includes the following temperature update formula in the (k+1)th round: ,in This represents the initial cooling coefficient; with each temperature update, the unfolded structural risk index of the candidate states from the previous round is calculated. Difference from average unfolded revenue The cooling coefficient during the update process Follow these rules:

[0031]

[0032] in, This indicates the threshold for expanding the profit difference; , This represents the adjustment coefficient. This indicates a zero-factor.

[0033] A further improvement of this invention is that the vibration threat index is obtained by acquiring the raw vibration signal through a miniature MEMS accelerometer mounted on the surface of the photovoltaic module in a sensor network. The current spectrum is obtained by using a sliding window fast Fourier transform. The main excitation frequency is obtained. For each current unfolding angle Retrieve the structure's inherent frequency group from the database The vibration threat index is obtained by calculating the proximity of the current main excitation frequency to all modal frequencies; the temperature change index is obtained by obtaining the current actual temperature of the hinge through a temperature sensor installed at the hinge connection of the photovoltaic module, and calculating the difference between the current actual temperature of the hinge and the chain temperature threshold.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] 1. This invention first calculates the future power generation gain and deducts the risk cost through a benefit-loss model to obtain the deployment benefit. It can make a dynamic trade-off between benefit and risk, control the number of photovoltaic panels deployed and the speed of the deployment process, so that the decision-making pursues both power generation improvement and avoids structural damage or loss of life caused by high risk.

[0036] 2. Secondly, based on the annealing algorithm, the unfolded structural risk index is incorporated into the initial temperature, adaptive cooling coefficient and acceptance criteria. It can converge quickly in high-risk or uncertain scenarios to ensure safety, and fully explore high-yield solutions in low-risk scenarios. It overcomes the problems of traditional annealing algorithms, such as sensitivity to parameters, susceptibility to local conditions, slow convergence or premature convergence. Under the conditions of limited computational budget and real-time decision requirements, it improves the quality and stability of decision-making. Attached Figure Description

[0037] Figure 1 This invention provides a flow chart for a foldable photovoltaic module deployment control method based on environmental perception.

[0038] Figure 2 This is a schematic diagram of the photovoltaic module deployment process, illustrating the environmental perception-based foldable photovoltaic module deployment control method of the present invention.

[0039] Figure 3 This is a flowchart of the photovoltaic panel motion control algorithm in the environmental perception-based foldable photovoltaic module deployment control method of the present invention. Detailed Implementation

[0040] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0041] The term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three cases: A exists alone, A and B exist simultaneously, and B exists alone.

[0042] Example 1

[0043] Figure 1 The flowchart of a foldable photovoltaic module deployment control method based on environmental perception disclosed in this embodiment is shown, and the steps are as follows:

[0044] S1. Acquire environmental data and the mechanical status of photovoltaic modules through a sensor network;

[0045] S2. Extract satellite imagery data from the environmental data to extract environmental occlusion features and output future time. Environmental shading loss coefficient ;

[0046] S3. Assess future time based on the mechanical state of the photovoltaic modules. Expanding structural risk indicators ;

[0047] S4. Based on the environmental shading loss coefficient and the deployed structure risk index, establish a revenue-loss model, which is used to output future time... The benefits of expansion;

[0048] S5. Set the photovoltaic panel motion control algorithm, and use future time... The unfolding benefits are taken as input, and the unfolding control parameters of the foldable photovoltaic module are output.

[0049] The environmental occlusion features include local static occlusion features and cloud dynamic occlusion features;

[0050] The local static shading prediction features include acquiring potential shading bodies, which include tree canopy features and building features; converting pixel coordinates in satellite images to ground latitude and longitude coordinates using satellite metadata (such as satellite altitude, top view, etc.) and a geographic projection model (e.g., orthographic projection or latitude and longitude mapping); mapping the latitude and longitude of the target photovoltaic site to the corresponding pixels in the satellite image to extract cloud information; the tree canopy features include extracting multispectral data from satellite imagery, calculating the vegetation index within m meters centered on the photovoltaic module, and inverting vegetation height information using an SVM model (e.g., combining GEDI / ICESat-2 point cloud, Sentinel-1 SAR data, or an existing LiDAR point cloud training model) to obtain the canopy height; the building features extract building height by combining high-resolution image classification and DSM data; outputting a three-dimensional surface feature set, including tree canopy, buildings, elevation, etc., to form a potential shading body height model; and calculating the changes in solar altitude angle and azimuth angle over time using NOAA based on the coordinates of the target photovoltaic module.

[0051] For each potential shading object, calculate its shadow area on the plane where the target photovoltaic module is located based on the sun's position; repeatedly project every 15 minutes throughout the day to obtain the spatiotemporal distribution of the shadow, and output the static shading shadow layer corresponding to different time points within the day; extract the predicted future time. Static occlusion shadow layer and future time The overlapping area of ​​the photovoltaic module's unfolded area is calculated, and its proportion in the total unfolded area is denoted as the local static shading feature.

[0052] The dynamic cloud occlusion features include training satellite cloud images using a pre-trained cloud detection model to obtain the cloud mask at the current time t, and the cloud mask area, which can be represented by pixels, and marking the pixel region where the cloud is located; applying optical flow algorithms, such as Lucas-Kanade and Horn-Schunck, to two consecutive frames of satellite cloud images to extract the cloud motion field. The extracted cloud motion vectors are smoothed and filtered to remove outliers; large-scale cloud cluster motion direction and velocity are obtained through region clustering; and for the center location of each cloud cluster... Cloudtop height and cloud motion vector To obtain the future location of the cloud shadow on the ground. and cloud cover area;

[0053] Horizontal offset is By utilizing the sun's position, the shadow of the cloud top is projected onto the ground: if the sun's altitude angle... The horizontal offset distance of the shadow relative to the cloud base point is approximately The direction is opposite to the sun's position. Therefore, considering both motion and projection, at some point in the future... The center of the cloud shadow is represented as ;

[0054] If the center or boundary path of the cloud shadow intersects with the coordinates of the photovoltaic module, then dynamic cloud shading is determined to exist; when the coordinates of the photovoltaic module are shaded, the location of the photovoltaic module is determined by the real-time data of the photovoltaic module. coordinate boundaries and time The overlapping area of ​​the cloud mask, recording time The proportion of time the cloud is covered is used to obtain the dynamic occlusion characteristics of the cloud.

[0055] The environmental occlusion loss coefficient is obtained by weighted summation of the local static occlusion features and the cloud dynamic occlusion features.

[0056] The deployment structure risk index is calculated by extracting the mechanical state of the photovoltaic modules. The next round of photovoltaic panel deployment is taken as the current deployment action, i.e., the deployment process of the next two photovoltaic panels. The mechanical state of the photovoltaic modules includes vibration threat indicators and temperature change indicators. The vibration threat indicators are obtained by collecting raw vibration signals using miniature MEMS accelerometers installed on the surface of the photovoltaic modules in a sensor network. The current spectrum is obtained by using a sliding window fast Fourier transform. The main excitation frequency is obtained. For each current unfolding angle ratio, the structure's inherent frequency group is retrieved from the database. The vibration threat index is obtained by calculating the proximity between the current main excitation frequency and all modal frequencies. The proximity is expressed as... The nonlinearity increases, reflecting the resonance tendency when the frequencies are close. Approaching a certain stage When the denominator approaches 1, the vibration threat index approaches 1, indicating a high risk of resonance. The temperature change index is obtained by using a temperature sensor installed at the hinge connection of the photovoltaic module to obtain the current actual temperature of the hinge, and calculating the difference between the current actual temperature of the hinge and the chain temperature threshold.

[0057] Record historical actions and the number of historical actions. Since the photovoltaic panel module is formed by two panels unfolding at a time, the number of historical actions is half the number of photovoltaic panels that have been fully unfolded. The mechanical state of the photovoltaic module is spliced ​​along with its feature dimensions to form a dual-channel damage input tensor. The historical dual-channel damage input tensor and the corresponding historical unfolded structural risk indicators are used as training data. The XGBoost model outputs the predicted risk indicator value for the current unfolding angle and the model confidence score. ;

[0058] In order to obtain future time The unfolded structure is the predicted risk index value after the two photovoltaic panels are fully unfolded, and the unfolding ratio of the current photovoltaic panels is obtained. The angle of the currently unfolded photovoltaic panel Total angle after unfolding The proportion, such as Figure 2 As shown, Figure 2 This invention illustrates the photovoltaic panel module unfolding process; it also compares the current unfolding action risk indicators with the unfolding ratio. The ratio is used as the risk prediction value;

[0059] Vibration can generate periodic or random stress on a structure. Stress cycles can lead to the initiation and propagation of microcracks, eventually causing material fatigue failure. Fatigue damage is cumulative. In this invention, the photovoltaic panels are located on both sides of the box and unfold in a wave-like manner. This means that each time two photovoltaic panels are unfolded, the remaining lifespan of the entire structure is consumed, increasing the unfolding cost. The cost increases cumulatively for each unfolded photovoltaic panel. Therefore, the damage is gradually accumulated. Thus, the sum of vibration threat indicators from historical unfolding actions is extracted, standardized, and then weighted and summed with the unfolding risk prediction value to obtain the predicted unfolding structural risk index.

[0060] The benefit-loss model includes extracting meteorological data from the environmental data to obtain future time data. Solar irradiance And calculate future time Environmental characteristics With unfolded area Through solar irradiance Environmental characteristics With unfolded area The product of time and future time is obtained. Power generation forecast Calculate the current power generation in the same way. Calculate future time Power generation forecast Compared with current power generation Integral of the difference Obtain the predicted power generation gain By combining the predicted power generation gain with the corresponding environmental shading loss coefficient and the unfolded structural risk index, a revenue-loss model is generated, denoted as the future time... Expanding benefits ,in, , , and They represent the weights, , .

[0061] Figure 3 The flowchart of the photovoltaic panel motion control algorithm of the present invention is shown. The photovoltaic panel motion control algorithm is based on the annealing algorithm and combined with the unfolding structure risk index to output the photovoltaic panel unfolding action command. The specific construction process includes:

[0062] S51. Define each state in the solution space as... , Indicates the deployment speed; obtains the predicted value of the deployment structure risk index. and model confidence in step S3 Set a hard threshold for risk. ;

[0063] S52. Use the revenue-loss model from step S4 as the objective function. ;

[0064] S53, in time state Randomly sample m candidate states from the vicinity Calculate the corresponding expanded structural risk indicators Input the sample risk dataset and calculate the spread return difference. ;

[0065] S54. Set the initial temperature adaptive strategy and obtain the initial temperature. ;

[0066] The initial temperature adaptive strategy includes setting the structural risk index in the sample risk dataset to be less than or equal to the risk hard threshold. The state samples are sent to the low-risk group, and the structural risk index will be greater than the risk hard threshold. State samples are sent to the high-risk group; it is hoped that they will have a higher acceptance probability in the low-risk group, therefore acceptance probabilities are assigned to state samples in the low-risk group. A lower acceptance probability can be assigned to the high-risk group, therefore, an acceptance probability is assigned to the state samples of the high-risk group. ;

[0067] Calculate the average absolute value of the spread returns of the low-risk group. And set the initial temperature of the low-risk state sample to be High-risk samples, due to their excessively high risk, are not included in the initial temperature calculation. When mechanical health is poor or the risk model confidence is low, the initial temperature should be lowered to reduce high-risk exploration; when health is good and confidence is high, the initial temperature can be appropriately increased to enhance exploration capabilities. Therefore, the initial temperature for all state samples is expressed as... ;

[0068] S55. Set a risk-driven adaptive cooling strategy and update the temperature iteratively;

[0069] The risk-driven adaptive cooling strategy includes the following temperature update formula in the (k+1)th round: ,in This represents the initial cooling coefficient; with each temperature update, the unfolded structural risk index of the candidate states from the previous round is calculated. Difference from average unfolded revenue The cooling coefficient during the update process Follow these rules:

[0070]

[0071] in, This indicates the threshold for expanding the profit difference; , This represents the adjustment coefficient. This indicates a zero-prevention factor (to prevent division by zero).

[0072] High structural risk during deployment And expand the profit difference With minimal improvements, through To achieve rapid cooling, the probability of accepting high-risk solutions is forcibly reduced, significantly decreasing the possibility of the algorithm getting trapped in local optima and improving global search capabilities; this is achieved in situations where the unfolded structure has low risk. And the improvements are significant. In the case of, through Achieving slow cooling can further expand the exploration range of low-risk high-quality solutions, making the final output solution more robust and reliable, and enhancing the practical application value of the system.

[0073] Based on the above, the system can autonomously determine the cooling strategy according to the actual performance of the current optimization process, thereby dynamically adjusting the temperature decay rate, i.e., the cooling coefficient. Compared to a fixed annealing rate strategy, this method can accelerate the escape from inefficient regions in the early stages and facilitate detailed searching in the later stages, thereby improving the overall convergence speed.

[0074] This embodiment improves global search capability and reduces the risk of getting stuck by introducing an "expanded structure risk index" to adaptively adjust key parameters throughout the annealing algorithm; it balances exploration and utilization under limited computational budget and real-time decision-making requirements; it converges to a safe and feasible solution faster in high-risk or uncertain scenarios and fully explores high-yield solutions in low-risk scenarios; and it enhances the stability of results and the ability to adapt to dynamic environmental changes.

[0075] S56. When the current temperature T is less than the set minimum temperature threshold, the search enters the final stage and can be considered to have converged (at this time, the acceptance probability is almost zero, and only better solutions are accepted). Then, the search is safely terminated and the state corresponding to the maximum value of the objective function is output. When the increase value of the objective function in three consecutive temperature updates is less than the set profit increase threshold and the average unfolding structure risk index of the candidate states has not decreased, the current state is determined to be in an invalid or high-risk area, indicating that the damage is greater than the profit. Then, the unfolding action is terminated.

[0076] The threshold and weight settings and preset models can be set by default according to the present invention, or they can be set by those skilled in the art.

[0077] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0078] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0079] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0080] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0081] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for controlling the deployment of foldable photovoltaic modules based on environmental perception, characterized in that: Includes the following steps: S1. Acquire environmental data and the mechanical status of photovoltaic modules through a sensor network; S2. Extract satellite imagery data from the environmental data to obtain environmental occlusion features and output future time. Environmental shading loss coefficient ; S3. Assess future time based on the mechanical state of the photovoltaic modules. Expanding structural risk indicators ; S4. Based on the environmental shading loss coefficient and the deployed structure risk index, establish a revenue-loss model, which is used to output future time... The benefits of expansion; S5. Set the photovoltaic panel motion control algorithm, and use future time... The unfolding benefits are taken as input, and the unfolding control parameters of the foldable photovoltaic module are output. The benefit-loss model includes extracting meteorological data from the environmental data to obtain future time data. Solar irradiance And calculate future time Environmental characteristics With unfolded area Through solar irradiance Environmental characteristics With unfolded area The product of time and future time is obtained. Power generation forecast Calculate the current power generation in the same way. Calculate future time Power generation forecast Compared with current power generation Integral of the difference Obtain the predicted power generation gain By combining the predicted power generation gain with the corresponding environmental shading loss coefficient and the unfolded structural risk index, a revenue-loss model is generated, denoted as the future time... Expanding benefits ,in, , , and They represent the weights, , .

2. The method for controlling the unfolding of a foldable photovoltaic module based on environmental perception according to claim 1, characterized in that: The environmental occlusion features include local static occlusion features and cloud dynamic occlusion features; The local static shading prediction features include acquiring potential shading bodies, which include tree canopy features and building features; the tree canopy features include extracting multispectral data from satellite imagery, calculating the vegetation index within m meters centered on the photovoltaic module, and obtaining the canopy height by inverting vegetation height information through an SVM model; the building features are obtained by extracting building heights by combining high-resolution image classification and DSM data; outputting a three-dimensional surface feature set to form a potential shading body height model; and calculating the changes in solar altitude angle and azimuth angle over time based on the coordinates of the target photovoltaic module. For each potential shading object, calculate its shadow area on the plane where the target photovoltaic module is located based on the sun's position; repeatedly project every 15 minutes throughout the day to obtain the spatiotemporal distribution of the shadow, and output the static shading shadow layer corresponding to different time points within the day; extract the predicted future time. Static occlusion shadow layer and future time The overlapping area of ​​the photovoltaic module's unfolded area is calculated, and its proportion in the total unfolded area is denoted as the local static shading feature.

3. The method for controlling the deployment of a foldable photovoltaic module based on environmental perception according to claim 2, characterized in that: The cloud dynamic occlusion feature includes training satellite cloud images using a pre-trained cloud detection model to obtain the cloud mask at the current time t, the cloud mask area, and marking the pixel region where the cloud is located. The optical flow algorithm is applied to extract the cloud motion field from two consecutive frames of satellite cloud images. The extracted cloud motion vectors are smoothed and filtered to remove outliers; large-scale cloud cluster motion direction and velocity are obtained through region clustering; and for the center location of each cloud cluster... Cloudtop height Given the cloud motion vector v, we can obtain the position of the cloud shadow on the ground at the future time t. The location and cloud cover area are considered. If the center or boundary path of the cloud shadow intersects with the coordinates of the photovoltaic module, dynamic cloud shading is determined to exist. When the coordinates of the photovoltaic module are shaded, the time is recorded. The proportion of time the cloud is covered is used to obtain the dynamic occlusion characteristics of the cloud.

4. The method for controlling the deployment of a foldable photovoltaic module based on environmental perception according to claim 3, characterized in that: The environmental occlusion loss coefficient is obtained by weighted summation of the local static occlusion features and the cloud dynamic occlusion features.

5. The method for controlling the deployment of a foldable photovoltaic module based on environmental perception according to claim 1, characterized in that: The unfolding structural risk index is calculated by extracting the mechanical state of the photovoltaic module. The current unfolding action is taken as the next unfolding action. The mechanical state of the photovoltaic module includes vibration threat index and temperature change index, and historical actions and the number of historical actions are recorded. The mechanical state of the photovoltaic module is concatenated by feature dimensions to form a dual-channel damage input tensor. The historical dual-channel damage input tensor and the corresponding historical unfolding structural risk index are used as training data. The XGBoost model outputs the predicted value of the risk index at the current unfolding angle and the model confidence score. ; Get the current unfolded ratio of the photovoltaic panel ; Combine the current operational risk indicators with the deployment ratio The ratio is used as the risk prediction value; The vibration threat index of historical deployment actions is extracted, standardized, and then weighted and summed with the predicted deployment risk value to obtain the predicted deployment structural risk index.

6. The method for controlling the deployment of a foldable photovoltaic module based on environmental perception according to claim 5, characterized in that: The photovoltaic panel motion control algorithm is based on the annealing algorithm and combines the unfolding structure risk index to output photovoltaic panel unfolding motion commands. The specific construction process includes: S51. Define each state in the solution space as... , Indicates the deployment speed; and obtains the predicted value of the deployment structure risk index. and model confidence in step S3 Set a hard threshold for risk. ; S52. Use the revenue-loss model from step S4 as the objective function. ; S53, in time state Randomly sample m candidate states from the vicinity Calculate the corresponding expanded structural risk indicators Input the sample risk dataset and calculate the spread return difference. ; S54. Set the initial temperature adaptive strategy and obtain the initial temperature. ; S55. Set a risk-driven adaptive cooling strategy and update the temperature iteratively; S56. When the current temperature T is less than the set minimum temperature threshold, the operation will terminate safely and output the state corresponding to the maximum value of the objective function. If the increase value of the objective function is less than the set profit increase threshold in three consecutive temperature updates and the average unfolded structure risk index of the candidate state does not decrease, the current state is determined to be in an invalid or high-risk area, and the unfolding action will be terminated.

7. The method for controlling the unfolding of a foldable photovoltaic module based on environmental perception according to claim 6, characterized in that: The initial temperature adaptive strategy includes setting the structural risk index in the sample risk dataset to be less than or equal to the risk hard threshold. The state samples are sent to the low-risk group, and the structural risk index will be greater than the risk hard threshold. State samples are sent to the high-risk group; acceptance probabilities are assigned to state samples in the low-risk group. Assign acceptance probabilities to state samples of the high-risk group. ; Calculate the average absolute value of the spread returns of the low-risk group. And set the initial temperature of the low-risk state sample to be Then the initial temperature of all state samples is expressed as: .

8. The method for controlling the deployment of a foldable photovoltaic module based on environmental perception according to claim 7, characterized in that: The risk-driven adaptive cooling strategy includes the following temperature update formula in the (k+1)th round: ,in Indicates the initial cooling coefficient; At each temperature update, the unfolded structural risk index of the candidate states from the previous round is calculated. Difference from average unfolded revenue The cooling coefficient during the update process Follow these rules: ； in, This indicates the threshold for expanding the profit difference; , This represents the adjustment coefficient. This indicates a zero-factor.

9. The method for controlling the unfolding of a foldable photovoltaic module based on environmental perception according to claim 5, characterized in that: The vibration threat indicator is detected by miniature sensors mounted on the surface of the photovoltaic module within a sensor network. Accelerometers collect raw vibration signals The current spectrum is obtained by using a sliding window fast Fourier transform. The main excitation frequency is obtained. For each current unfolding angle Retrieve the structure's inherent frequency group from the database The vibration threat index is obtained by calculating the proximity of the current main excitation frequency to all modal frequencies; the temperature change index is obtained by obtaining the current actual temperature of the hinge through a temperature sensor installed at the hinge connection of the photovoltaic module, and calculating the difference between the current actual temperature of the hinge and the chain temperature threshold.

Images