A double-cable-rigid beam cooperative adjustment system for photovoltaic modules suitable for sloping terrain

Abstract

The present application relates to a kind of photovoltaic module double cable-rigid beam coordinated regulation system suitable for slope terrain, including side slope anchoring foundation, flexible support subsystem and rigid beam subsystem, belong to photovoltaic power generation technical field.There is also including: terrain perception unit obtains photovoltaic module inclination deviation, cable tension and sliding component displacement deviation, inversion equivalent terrain influence factor;Predictive and calculation unit is based on historical data and wind field model to generate predicted load, and calculate cable clamp mechanism target clamping force, sliding component target rotation angle and target coordinated stiffness distribution;Independent drive unit is independently controlled by remote control program Each actuator, so that rigid beam and double cable constitute variable stiffness coordinated bearing network;Self-adapting feedback unit updates wind field model parameters on-line, and when load deviates from prediction, trigger re-optimization.Adaptable coordinated regulation to slope terrain and wind, snow load is realized, effectively disperses stress, keeps the best inclination of component, improves system stability and terrain adaptability.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power generation technology, specifically relating to a dual-cable-rigid-beam coordinated adjustment system for photovoltaic modules suitable for sloping terrain. Background Technology

[0002] In photovoltaic (PV) power generation systems, the support structure of PV modules has a significant impact on their stability, power generation efficiency, and terrain adaptability. Traditional PV support systems mostly employ rigid structures, using columns and beams with fixed tilt angles to support the modules, making them suitable for flat terrain. However, in practical applications, many PV power plants need to be built on sloping terrain such as mountains, hills, and slopes. Rigid supports are difficult to adapt to surface undulations, often requiring extensive earthwork leveling or the use of columns of varying lengths for adjustment. This results in high construction costs, long construction periods, and significant damage to the ecological environment.

[0003] In recent years, flexible photovoltaic (PV) support technology has gradually attracted attention. Flexible supports are not entirely soft; rather, they utilize high-strength flexible materials (such as prestressed steel cables and composite fibers) and modular connectors to construct a bendable and adjustable support structure. Unlike the fixed angle of traditional rigid supports, flexible supports can dynamically adjust their angle according to terrain undulations and module arrangement requirements. They can even disperse stress through elastic deformation under external forces such as wind and snow loads, reducing the risk of module damage. This design overcomes terrain limitations, enabling PV systems to efficiently cover complex environments such as mountains, water surfaces, and agricultural greenhouses.

[0004] However, existing control strategies for flexible supports still have significant shortcomings. First, most control methods are open-loop or based on fixed rules, failing to adaptively optimize according to real-time terrain undulations and dynamically changing wind and snow loads. Second, existing systems lack load prediction capabilities, typically responding passively after external forces occur, resulting in limited stress dispersion and difficulty in effectively maintaining the optimal tilt angle of components under strong winds or blizzards. Third, the connection stiffness between steel cables and rigid beams in different sections is fixed, unable to dynamically adjust with load distribution changes, leading to localized stress concentration and affecting the long-term stability and service life of the system. Furthermore, existing control programs lack self-learning and model update mechanisms, failing to adapt to seasonal changes in the local microclimate of the slope, resulting in decreased control accuracy after long-term operation.

[0005] Therefore, there is an urgent need for a photovoltaic module support system that can coordinate the adjustment of flexible supports and rigid beams according to the characteristics of slope terrain and dynamic environmental loads, so as to improve the system's stability, terrain adaptability and power generation efficiency. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides a dual-cable-rigid-beam coordinated adjustment system for photovoltaic modules suitable for sloping terrain. The objective of this invention can be achieved through the following technical solution: A dual-cable-rigid-beam coordinated adjustment system for photovoltaic modules suitable for sloping terrain includes a slope anchoring foundation, a flexible support subsystem, and a rigid beam subsystem; it also includes: The terrain sensing unit is used to acquire in real time the tilt angle deviation of photovoltaic modules, the tension of steel cables in each section, the relative displacement and angle deviation of sliding components and sliding guide rails, and to invert the equivalent terrain influence factor based on this. The prediction and calculation unit includes a multi-objective optimization module. With tilt angle maintenance accuracy, peak cable tension, and adjustment energy consumption as optimization objectives, it uses a rolling time-domain optimization algorithm to calculate the target clamping force of the cable clamp mechanism in each section, the target rotation angle of each sliding component, and the target cooperative stiffness distribution based on the equivalent terrain influence factor and the predicted load generated based on historical data and wind field model. An independent drive unit is used to independently control each cable clamp mechanism to achieve the target clamping force through a remote control program, and independently drive each sliding component to rotate to the target angle. A variable stiffness collaborative bearing network is formed by rigid beams and double cables, while real-time feedback of actual position and force information is provided. The adaptive feedback unit includes a data-driven update module, which triggers the prediction and calculation unit to re-optimize when the actual load deviates from the predicted value, and stores the adjustment error and environmental response data in the historical database to update the parameters of the wind field model.

[0007] Specifically, the flexible support subsystem includes two sets of steel cables arranged parallel to the slope terrain and sliding guide rails fixed to each steel cable; the rigid beam subsystem includes a rigid beam arranged perpendicular to the direction of the steel cables, with both ends of the rigid beam connected to the sliding guide rails on the two sets of steel cables through sliding components, and the rigid beam is provided with a cable clamping mechanism for clamping or releasing the steel cables; the slope anchoring foundation is connected to the end of the steel cables.

[0008] Specifically, the multi-objective optimization module uses a weighted summation method to construct a comprehensive optimization objective function, wherein the weight coefficients of each optimization objective are dynamically adjusted according to the slope grade of the slope terrain, historical wind load statistical characteristics, or user-defined priorities; the optimization indicators include the root mean square of the tilt angle deviation, the maximum value and variance of the tension in each section of the steel cable, and the cumulative sum of the total adjustment energy consumption or adjustment amplitude of all actuators.

[0009] Specifically, the rolling time-domain optimization algorithm solves an open-loop optimization problem with a preset prediction time-domain length based on the current sensing data and predicted load in each control cycle, and executes only the clamping force and rotation angle commands corresponding to the first control step in the optimization sequence after solving the problem. When the next cycle arrives, the algorithm re-solves the problem based on the latest feedback state, wherein the prediction time-domain length can be adaptively shortened or extended according to the current load change rate.

[0010] Specifically, the wind field model is a time-series neural network model or an autoregressive moving average model trained based on historical local microclimate data of the slope. The model input includes wind speed, wind direction, temperature, humidity and photovoltaic module backsheet temperature at several past moments. The model output is the instantaneous wind speed sequence, the main wind direction angle and the predicted equivalent wind pressure distribution of each cable section within a preset time window.

[0011] Specifically, the predicted load includes the equivalent nodal force or distributed force of each section of the steel cable caused by wind load, snow load, or a combination of both. The wind load part is calculated based on the equivalent wind pressure output by the wind field model combined with the windward area and drag coefficient of the photovoltaic module. The snow load part is calculated based on the snow depth preset model or on-site snow sensor data. The predicted load is input into the rolling time domain optimization algorithm in the form of a spatial distribution vector, and the predicted load is set with a confidence interval, which is used to characterize the prediction uncertainty.

[0012] As a preferred embodiment of the present invention, the data-driven update module employs one of the following methods: recursive least squares, Kalman filtering, or gradient descent with momentum term. It takes the error vector between the actual load and the predicted load as input and updates the internal parameters of the wind field model online. The update operation is triggered after each control cycle or after accumulating a fixed amount of new data.

[0013] Specifically, the independent drive unit includes a pneumatic actuator or an electric actuator, which is connected to each cable clamping mechanism and each sliding component respectively. Each actuator has a built-in position sensor and force sensor for real-time feedback. The remote control program runs in an industrial-grade programmable logic controller or an embedded edge computing gateway, and sends independent control commands to each actuator through a wired industrial Ethernet, 4G / 5G or LoRa wireless communication network. It also supports local caching and retry mechanisms in the event of network outage.

[0014] Specifically, the variable stiffness cooperative bearing network is configured such that when the external load increases, the independent drive unit increases the clamping force of the cable clamping mechanism on the windward side or high load section and decreases the clamping force on the leeward side or low load section according to the rolling time domain optimization results. In this case, the elastic deformation section of the steel cable dynamically migrates from the high stress area to the low stress area. The peak stress that was originally concentrated in a local area is dispersed into multiple smaller stress peaks and transmitted to the slope anchoring foundations at different locations. Furthermore, the change in the tilt angle of the photovoltaic module is kept below a preset threshold.

[0015] Specifically, the terrain sensing unit also includes an initial calibration mode, which is used during the installation and commissioning phase. Under windless and unloaded conditions, with the rigid beam and sliding component in their initial zero position, the actual initial tilt angle of each photovoltaic module or each group of photovoltaic modules is recorded, and the initial tilt angle is stored as a reference value in a non-volatile memory. During normal operation, the real-time tilt angle deviation is calculated as the measured tilt angle minus the reference initial tilt angle.

[0016] Specifically, the adaptive feedback unit is also used to send a wind field model failure warning signal to the remote monitoring platform when the actual load deviates from the predicted value by more than a preset relative error threshold for a consecutive preset number of times, and automatically switch to the standby control mode, while retaining the original control log for offline analysis; when the error falls back to within the threshold after the wind field model parameters are updated, it can automatically switch back to the normal predictive control mode, or request confirmation from the operator before switching.

[0017] Specifically, the independent drive unit is also used to perform periodic self-checks under windless and unloaded conditions. The self-check process includes: driving all cable clamping mechanisms to traverse at least three stroke positions from release to clamping in a preset sequence, while driving the sliding component to traverse at least five angular positions from its lower limit to its upper limit of angular stroke, recording the actual response time, positioning accuracy, and force feedback curve of each actuator; generating a health status report after the self-check is completed; if it is found that the response time of an actuator exceeds the limit or the positioning deviation exceeds the allowable value, the actuator is automatically marked as a state to be maintained and a maintenance plan is reported.

[0018] The beneficial effects of this invention are as follows: Achieving adaptive and coordinated adjustment of slope terrain: The slope perception unit acquires tilt angle deviation, cable tension and displacement deviation in real time, and inverts the equivalent terrain influence factor; combined with predicted load and rolling time domain multi-objective optimization, the clamping force, rotation angle and coordinated stiffness distribution of each section are dynamically calculated to achieve closed-loop adaptive adjustment of slope terrain, without the need for a large amount of earthwork leveling, reducing construction costs and ecological damage.

[0019] Constructing a variable stiffness collaborative load-bearing network effectively disperses stress: When the external load increases, the independent drive unit adjusts the clamping force of each section differently according to the optimization results, so that the elastic deformation section of the steel cable dynamically migrates, dispersing the concentrated peak stress into multiple smaller peaks and transferring them to different anchor foundations. At the same time, it keeps the component tilt angle change within the threshold, significantly reducing the risk of component microcracks and extending the system life.

[0020] Introducing load prediction and feedforward active control: Based on historical data and wind field models (temporal neural networks or ARIMA), predictive loads are generated, and feedforward adjustment is achieved by using rolling time domain optimization. The length of the prediction time domain is adaptively adjusted according to the load change rate, and the support attitude is adjusted before strong winds or blizzards arrive, thereby improving the ability to resist extreme weather.

[0021] It has data-driven self-learning capabilities: the adaptive feedback unit updates the wind field model parameters online, enabling the system to continuously learn the local microclimate characteristics of the slope, and the control accuracy improves with the running time, solving the problem of long-term performance degradation.

[0022] Enhanced system reliability and maintainability: Built-in sensors provide real-time feedback, supporting local caching and retries during network outages; features include initial calibration to eliminate installation errors, periodic self-testing to diagnose actuator health, and automatic switching to standby mode with warnings when prediction errors exceed limits, significantly improving robustness and reducing maintenance costs. Attached Figure Description

[0023] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0024] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is the control flowchart of the present invention; Figure 3 This is a schematic diagram of stress dispersion in the variable stiffness cooperative bearing network of the present invention; Figure 4 This is the system installation and initial calibration process of the present invention; Figure 5 This is a flowchart of the periodic self-test process of the present invention. Detailed Implementation

[0025] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] Please see Figures 1-5 A dual-cable-rigid-beam coordinated adjustment system for photovoltaic modules suitable for sloping terrain includes a slope anchoring foundation, a flexible support subsystem, and a rigid beam subsystem; it also includes: The terrain sensing unit is used to acquire in real time the tilt angle deviation of photovoltaic modules, the tension of steel cables in each section, the relative displacement and angle deviation of sliding components and sliding guide rails, and to invert the equivalent terrain influence factor based on this. The prediction and calculation unit includes a multi-objective optimization module. With tilt angle maintenance accuracy, peak cable tension, and adjustment energy consumption as optimization objectives, it uses a rolling time-domain optimization algorithm to calculate the target clamping force of the cable clamp mechanism in each section, the target rotation angle of each sliding component, and the target cooperative stiffness distribution based on the equivalent terrain influence factor and the predicted load generated based on historical data and wind field model. An independent drive unit is used to independently control each cable clamping mechanism to achieve the target clamping force through a remote control program, and independently drive each sliding component to rotate to the target angle, so that the rigid beam and the double cables form a variable stiffness cooperative bearing network, while providing real-time feedback on actual position and force information. The adaptive feedback unit includes a data-driven update module, which triggers the prediction and calculation unit to re-optimize when the actual load deviates from the predicted value, and stores the adjustment error and environmental response data in the historical database to update the parameters of the wind field model.

[0027] Specifically, the flexible support subsystem includes two sets of steel cables arranged parallel to the slope terrain and sliding guide rails fixed to each steel cable; the rigid beam subsystem includes a rigid beam arranged perpendicular to the direction of the steel cables, with both ends of the rigid beam connected to the sliding guide rails on the two sets of steel cables through sliding components, and the rigid beam is provided with a cable clamping mechanism for clamping or releasing the steel cables; the slope anchoring foundation is connected to the end of the steel cables.

[0028] Specifically, the multi-objective optimization module uses a weighted summation method to construct a comprehensive optimization objective function, wherein the weight coefficients of each optimization objective are dynamically adjusted according to the slope grade of the slope terrain, historical wind load statistical characteristics, or user-defined priorities; the optimization indicators include the root mean square of the tilt angle deviation, the maximum value and variance of the tension in each section of the steel cable, and the cumulative sum of the total adjustment energy consumption or adjustment amplitude of all actuators.

[0029] Specifically, the rolling time-domain optimization algorithm solves an open-loop optimization problem with a preset prediction time-domain length (represented by multiple time steps) based on the current sensing data and predicted load in each control cycle, and executes only the clamping force and rotation angle commands corresponding to the first control step in the optimization sequence after solving the problem. When the next cycle arrives, the algorithm re-solves the problem based on the latest feedback state, wherein the prediction time-domain length can be adaptively shortened or extended according to the current load change rate.

[0030] Specifically, the wind field model is a time-series neural network model or an autoregressive moving average model trained based on historical local microclimate data of the slope. The model input includes wind speed, wind direction, temperature, humidity and photovoltaic module backsheet temperature at several past moments. The model output is the instantaneous wind speed sequence, the main wind direction angle and the predicted equivalent wind pressure distribution of each cable section within a preset time window.

[0031] Specifically, the predicted load includes the equivalent nodal force or distributed force of each section of the steel cable caused by wind load, snow load, or a combination of both. The wind load part is calculated based on the equivalent wind pressure output by the wind field model combined with the windward area and drag coefficient of the photovoltaic module. The snow load part is calculated based on the snow depth preset model or on-site snow sensor data. The predicted load is input into the rolling time domain optimization algorithm in the form of a spatial distribution vector, and the predicted load is set with a confidence interval, which is used to characterize the prediction uncertainty.

[0032] Specifically, the data-driven update module employs one of the following methods: recursive least squares, Kalman filtering, or gradient descent with momentum term. It takes the error vector between the actual load and the predicted load as input and updates the internal parameters of the wind field model online. The update operation is triggered after each control cycle or after accumulating a fixed number of new data.

[0033] Specifically, the independent drive unit includes a pneumatic actuator or an electric actuator, which is connected to each cable clamping mechanism and each sliding component respectively. Each actuator has a built-in position sensor and force sensor for real-time feedback. The remote control program runs in an industrial-grade programmable logic controller or an embedded edge computing gateway, and sends independent control commands to each actuator through a wired industrial Ethernet, 4G / 5G or LoRa wireless communication network. It also supports local caching and retry mechanisms in the event of network outage.

[0034] Specifically, the variable stiffness cooperative bearing network is configured such that when the external load increases, the independent drive unit increases the clamping force of the cable clamping mechanism on the windward side or high load section and decreases the clamping force on the leeward side or low load section according to the rolling time domain optimization results. In this case, the elastic deformation section of the steel cable dynamically migrates from the high stress area to the low stress area. The peak stress that was originally concentrated in a local area is dispersed into multiple smaller stress peaks and transmitted to the slope anchoring foundations at different locations. Furthermore, the change in the tilt angle of the photovoltaic module is kept below a preset threshold.

[0035] Specifically, the terrain sensing unit also includes an initial calibration mode, which is used during the installation and commissioning phase. Under windless and unloaded conditions, with the rigid beam and sliding component in their initial zero position, the actual initial tilt angle of each photovoltaic module or each group of photovoltaic modules is recorded, and the initial tilt angle is stored as a reference value in a non-volatile memory. During normal operation, the real-time tilt angle deviation is calculated as the measured tilt angle minus the reference initial tilt angle.

[0036] Specifically, the adaptive feedback unit is also used to send a wind field model failure warning signal to the remote monitoring platform when the actual load deviates from the predicted value by more than a preset relative error threshold for a consecutive preset number of times, and automatically switch to the standby control mode, while retaining the original control log for offline analysis; when the error falls back to within the threshold after the wind field model parameters are updated, it can automatically switch back to the normal predictive control mode, or request confirmation from the operator before switching.

[0037] Specifically, the independent drive unit is also used to perform periodic self-checks under windless and unloaded conditions. The self-check process includes: driving all cable clamping mechanisms to traverse at least three stroke positions from release to clamping in a preset sequence, while driving the sliding component to traverse at least five angular positions from its lower limit to its upper limit of angular stroke, recording the actual response time, positioning accuracy, and force feedback curve of each actuator; generating a health status report after the self-check is completed; if it is found that the response time of an actuator exceeds the limit or the positioning deviation exceeds the allowable value, the actuator is automatically marked as a state to be maintained and a maintenance plan is reported.

[0038] Example I. Application Scenario Setting A mountain photovoltaic power station is located in a hilly area at 25°N, 110°E. The site is a south-facing slope with an average slope of 25° and local undulations of 5° to 15°. The power station has an installed capacity of 5MW and adopts the dual-cable rigid beam coordinated regulation system described in this invention. A total of 20 independent regulation units are arranged, each covering approximately 500㎡. Each unit includes two sets of parallel prestressed steel cables (60m long, 4m apart), 12 rigid beams (perpendicular to the steel cables, 5m apart), and corresponding sliding components and cable clamping mechanisms. The slope anchoring foundation uses concrete anchor piers, and both ends of the steel cables are fixed to the anchor piers via cable end connectors.

[0039] II. System Installation and Initial Calibration After the system is installed, technicians start the initial calibration mode through the remote monitoring platform.

[0040] On a clear, windless, and snow-free day, all rigid beams and sliding components are in their factory zero position. Tilt sensors in the terrain sensing unit (installed in the middle of each rigid beam, corresponding to each group of photovoltaic modules) measure the actual initial tilt angle of each photovoltaic module. Due to the undulating slope, the initial tilt angle at different locations fluctuates between 22° and 28° (the target tilt angle is 25°). The system automatically stores these measured values ​​in non-volatile memory as reference values.

[0041] Simultaneously, the cable tensioning adjustment device adjusts the cable pretension to the design value (80kN per cable), the displacement sensor on the sliding guide rail records the initial position of the sliding component, and the force sensor records the initial clamping force of the cable clamping mechanism (set to 0, i.e., not clamped). After calibration, the system enters normal operation mode.

[0042] III. Coordinated Adjustment Process During Normal Operation 3.1 Terrain Sensing and Data Acquisition During real-time operation, the terrain sensing unit collects the following data at a frequency of 0.5Hz: Photovoltaic module tilt angle deviation: The measured tilt angle of the tilt sensor on each rigid beam is subtracted from the initial reference tilt angle. For example, if the measured tilt angle at a certain location is 24.2° and the reference is 25°, then the deviation is -0.8°.

[0043] Tension in each section of the steel cable: A fiber optic strain sensor is placed every 10m along the steel cable to calculate the tension value. Currently, the tension in section 3 is 85kN (slightly higher than the preload), and in section 5 it is 76kN.

[0044] Relative displacement and angular deviation of sliding components: The displacement sensor on the sliding guide rail shows that a certain sliding component has slid 120mm from the zero position to the lower side; the angle encoder on the sliding component shows that its rotation angle relative to the guide rail plane is +3.5° (indicating that the end of the rigid beam is raised to adapt to the local terrain).

[0045] Based on the above data, the terrain perception unit outputs the equivalent terrain influence factor through the built-in inversion algorithm (such as the equivalent slope calculation based on least squares fitting). In this example, it is a three-dimensional vector [0.32, 0.15, -0.05], which corresponds to the local slope correction coefficient, terrain relief coefficient, and terrain distortion coefficient, respectively.

[0046] 3.2 Load Prediction and Strategy Calculation The prediction and computation unit performs a rolling time-domain optimization every 10 seconds.

[0047] First, the wind field model (using an LSTM neural network, which is a temporal neural network and has been trained on historical data) receives weather station data from the past 30 minutes: wind speed 4.2~6.8 m / s, wind direction southwest (angle with the slope direction approximately 60°), temperature 28℃, humidity 75%, and photovoltaic module backsheet temperature 52℃. The model outputs a predicted wind speed sequence for the next 15 minutes (peak 7.5 m / s at the 8th minute), prevailing wind direction (southwest), and predicted equivalent wind pressure distribution for each cable section (wind pressure approximately 180 Pa on the windward side and approximately 60 Pa on the leeward side).

[0048] Then, based on the equivalent wind pressure and the windward area of ​​the photovoltaic modules (2.2m² per module), 2 ×20 blocks = 44m 2 The equivalent nodal force of each cable segment was calculated using the resistance coefficient (1.2) and the resistance coefficient (2.8kN). For example, the nodal force of segment 13 increased to 2.8kN, while that of segment 79 was only 0.9kN. Snow load sensor data showed that there was currently no snow accumulation, and the snow load was 0.

[0049] The multi-objective optimization module constructs a comprehensive objective function using a weighted summation method: Tilt holding accuracy: The goal is to minimize the sum of the absolute values ​​of the tilt angle deviations of all components (weight 0.5). Peak cable tension: The maximum tension is expected to not exceed 100kN (weight 0.3). Energy consumption adjustment: The goal is to minimize the total adjustment range of all cable clamping mechanisms and sliding components (weight 0.2). The current slope is 25°. Historical wind load statistics show that the maximum wind speed in this area is 12 m / s. Therefore, the weighting coefficients are dynamically adjusted values ​​(0.45, 0.35, 0.20).

[0050] The optimization algorithm (rolling time domain, prediction time domain length N=12 steps, 10 seconds per step) yields the following results: The target clamping force of the cable clamping mechanism in each section is as follows: Section 13 (windward side) increases from 0kN to 15kN; Section 46 remains at 5kN; Section 79 (leeward side) remains at 0kN.

[0051] Target rotation angle for each sliding component: Sliding components located at higher elevations need to rotate +2.5°, and sliding components located at lower elevations need to rotate -1.8°.

[0052] Target co-stiffness distribution: The axial equivalent stiffness of the 13th section is set to high (1800 N / mm), the 46th section to medium (1000 N / mm), and the 79th section to low (400 N / mm).

[0053] 3.3 Independent Drive and Dynamic Adjustment The independent drive unit sends instructions to each actuator via a remote control program (running on the edge gateway, 4G communication).

[0054] For the clamping mechanism in section 13, the pneumatic actuator receives the command "clamping force 15kN," and uses a built-in force sensor for closed-loop control. The actual clamping force reaches the target within 0.3 seconds, with an error of ±0.2kN. For section 46, the clamping force command is 5kN, and the same action is taken. For section 79, the system remains in the released state.

[0055] Simultaneously, the servo motor of the sliding component receives rotation angle commands, driving the end of the rigid beam to rotate relative to the sliding guide rail. For example, if the original angle of a sliding component located in the middle of the slope is +3.5° and the target angle is +2.5°, the motor rotates in the opposite direction by 1.0°, achieving an actual positioning accuracy of ±0.1° in 0.8 seconds.

[0056] During the adjustment process, the rigid beam and the two sets of steel cables form a variable stiffness collaborative load-bearing network: the windward section has a large clamping force and high stiffness, directly transferring wind loads to the steel cables; the leeward section is released, allowing the steel cables to absorb energy through elastic deformation; the middle section is moderately clamped, providing stable constraints. Overall, the concentrated wind pressure is dispersed into multiple stress peaks, which are then transferred to the slope anchorage foundations at different locations.

[0057] Real-time feedback information (actual position, force, angle) is transmitted back to the control unit via sensors at a frequency of 50Hz for optimization in the next cycle.

[0058] 3.4 Adaptive Feedback and Model Update When the actual external load deviates from the predicted value, the adaptive feedback unit intervenes.

[0059] For example, if the actual wind speed suddenly increases to 9.2 m / s (the predicted peak is 7.5 m / s), the measured tension in section 13 of the steel cable will reach 92 kN, higher than the predicted value of 82 kN, with a relative error of 12% (not exceeding the preset threshold of 30%). The data-driven update module records this error but does not trigger re-optimization. This error threshold is determined based on the structural safety requirements of photovoltaic modules and the design specifications of photovoltaic support systems.

[0060] On another occasion, if the actual wind pressure exceeded the predicted value by 35% (exceeding the 30% threshold) for three consecutive cycles, the adaptive feedback unit would perform the following operations: Send a "Wind Farm Model Failure Warning" signal to the remote monitoring platform and display a suggestion on the interface to check the weather station or retrain the model. Automatically switch to standby control mode: employ PID control, using tilt angle deviation as input and outputting an additional clamping force (e.g., proportional coefficient 0.5, integral time 5s) to ensure component tilt angle deviation is controlled within ±1°. Retain the error data and environmental response data for the most recent 24 hours to the historical database.

[0061] Subsequently, the data-driven update module (using recursive least squares with a forgetting factor of 0.98) updates the output layer weights of the LSTM model online using the most recent 100 error vectors. After 5 updates, the prediction error drops back to 18%, and the system automatically switches back to normal prediction control mode.

[0062] IV. Periodic Self-Check On the first windless night of each month, the system automatically performs a periodic self-check. The self-check sequence is set according to the cooperative stress logic of the rigid beam and the steel cable.

[0063] The self-test program drives the following sequentially: For all cable clamping mechanisms: from fully released (0kN) to fully clamped (20kN), passing through 30%, 60%, and 90% of the stroke (a total of 4 positions), the response time (normal ≤0.5s), positioning accuracy (error ≤0.3kN), and force feedback curve of each actuator were recorded.

[0064] All sliding parts: from the lower limit of the angle travel (-5°) to the upper limit (+10°), there are 6 positions at 3° intervals. Record the response time (normal ≤1s) and the positioning accuracy (error ≤0.15°).

[0065] A health report was generated after the self-test. One self-test revealed that the response time of the No. 8 cable clamp mechanism was 1.2 seconds, and its positioning deviation was 0.6 kN. The system automatically marked it as "needing maintenance" and reported a maintenance plan (recommending lubrication or replacement of seals). All other actuators were functioning normally.

[0066] V. Performance under extreme operating conditions One day, a sudden severe convective weather event occurred, with an average wind speed of 14 m / s over 10 minutes and gusts of 18 m / s (exceeding the design value of 12 m / s).

[0067] The system was pre-adjusted before the arrival of strong winds (based on an early warning system using a wind field model): the clamping force of the cable clamping mechanism on the windward side was increased to 18kN, while the leeward side remained unloaded, and the rotation angle of the sliding components was adjusted to the optimal windward angle. When gusts struck, the steel cable underwent elastic deformation (maximum elongation of 35mm), and the rigid beam absorbed the impact energy through the rotation and sliding of the sliding components. The short-term tension of the steel cable reached 108kN, which was still within the safe operating range.

[0068] Due to the effect of the variable stiffness cooperative bearing network, the stress that might have been concentrated in the middle section was distributed to sections 13 and 46, and then transferred to the steel cables through the cable clamp mechanism, and finally borne by the slope anchoring foundation. The maximum change in the tilt angle of the photovoltaic module was +2.3°, and the tilt angle threshold was set at ±2.5° according to the optimal working tilt angle range of the photovoltaic module and the slope topographic stability requirements. It did not exceed the allowable range, and no hidden cracks or detachment of the module were found.

[0069] After the storm, the system automatically performs a quick re-optimization to restore the clamping force and angle to normal.

[0070] VI. Summary of Implementation Results In this embodiment, the system of the present invention achieves the following technical effects: It adapts to 25° slope terrain and local undulations without the need for land leveling; Under strong winds, peak stress decreased by about 40%, and the rate of microcracks in the components decreased significantly. Predictive control keeps the tilt angle deviation of the modules within ±2.5°, thereby improving power generation efficiency; The self-learning capability improved control accuracy by 20% after 6 months of operation; The self-test function can detect actuator abnormalities in advance and avoid sudden failures.

[0071] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A photovoltaic module dual-cable-rigid beam coordinated adjustment system suitable for sloping terrain, comprising a slope anchoring foundation, a flexible support subsystem, and a rigid beam subsystem; characterized in that, Also includes: The terrain sensing unit is used to acquire in real time the tilt angle deviation of photovoltaic modules, the tension of steel cables in each section, the relative displacement and angle deviation of sliding components and sliding guide rails, and to invert the equivalent terrain influence factor based on this. The prediction and calculation unit includes a multi-objective optimization module. With tilt angle maintenance accuracy, peak cable tension, and adjustment energy consumption as optimization objectives, it uses a rolling time-domain optimization algorithm to calculate the target clamping force of the cable clamp mechanism in each section, the target rotation angle of each sliding component, and the target cooperative stiffness distribution based on the equivalent terrain influence factor and the predicted load generated based on historical data and wind field model. An independent drive unit is used to independently control each cable clamp mechanism to achieve the target clamping force through a remote control program, and independently drive each sliding component to rotate to the target angle. A variable stiffness collaborative bearing network is formed by rigid beams and double cables, while real-time feedback of actual position and force information is provided. The adaptive feedback unit includes a data-driven update module, which triggers the prediction and calculation unit to re-optimize when the actual load deviates from the predicted value, and stores the adjustment error and environmental response data in the historical database to update the parameters of the wind field model.

2. The system according to claim 1, characterized in that, The flexible support subsystem includes two sets of steel cables arranged parallel to the slope terrain and sliding guide rails fixed to each steel cable; the rigid beam subsystem includes a rigid beam arranged perpendicular to the direction of the steel cables, with both ends of the rigid beam connected to the sliding guide rails on the two sets of steel cables through sliding components, and the rigid beam is provided with a cable clamping mechanism for clamping or releasing the steel cables; the slope anchoring foundation is connected to the end of the steel cables.

3. The system according to claim 1, characterized in that, The multi-objective optimization module uses a weighted summation method to construct a comprehensive optimization objective function, wherein the weight coefficients of each optimization objective are dynamically adjusted according to the slope grade of the slope terrain, historical wind load statistical characteristics, or user-defined priorities; the optimization indicators include the root mean square of the tilt angle deviation, the maximum value and variance of the tension in each section of the steel cable, and the cumulative sum of the total adjustment energy consumption or adjustment amplitude of all actuators.

4. The system according to claim 1, characterized in that, The rolling time-domain optimization algorithm solves an open-loop optimization problem with a preset prediction time-domain length based on the current sensing data and predicted load in each control cycle. After solving the problem, it executes only the clamping force and rotation angle commands corresponding to the first control step in the optimization sequence. When the next cycle arrives, it re-solves the problem based on the latest feedback state. The prediction time-domain length can be adaptively shortened or extended according to the current load change rate.

5. The system according to claim 1, characterized in that, The wind field model is a time-series neural network model or an autoregressive moving average model trained based on historical local microclimate data of the slope. The model input includes wind speed, wind direction, temperature, humidity and photovoltaic module backsheet temperature at several past moments. The model output is the instantaneous wind speed sequence, the prevailing wind direction angle and the predicted equivalent wind pressure distribution of each cable section within a preset time window.

6. The system according to claim 1, characterized in that, The predicted load includes the equivalent nodal force or distributed force of each section of the steel cable caused by wind load, snow load, or a combination of both. The wind load part is calculated based on the equivalent wind pressure output by the wind field model combined with the windward area and drag coefficient of the photovoltaic module. The snow load part is calculated based on the snow depth preset model or the data from the on-site snow sensor. The predicted load is input into the rolling time domain optimization algorithm in the form of a spatial distribution vector, and the predicted load is set with a confidence interval, which is used to characterize the prediction uncertainty.

7. The system according to claim 1, characterized in that, The data-driven update module employs one of the following methods: recursive least squares, Kalman filtering, or gradient descent with momentum term. It takes the error vector between the actual load and the predicted load as input and updates the internal parameters of the wind field model online. The update operation is triggered after each control cycle or after accumulating a fixed number of new data.

8. The system according to claim 1, characterized in that, The independent drive unit includes a pneumatic actuator or an electric actuator, which is connected to each cable clamping mechanism and each sliding component respectively. Each actuator has a built-in position sensor and force sensor for real-time feedback. The remote control program runs in an industrial-grade programmable logic controller or an embedded edge computing gateway, and sends independent control commands to each actuator through a wired industrial Ethernet, 4G / 5G or LoRa wireless communication network. It also supports local caching and retry mechanisms in the event of a network outage.

9. The system according to claim 1, characterized in that, The variable stiffness cooperative bearing network is configured such that when the external load increases, the independent drive unit increases the clamping force of the cable clamping mechanism on the windward side or high load section and decreases the clamping force on the leeward side or low load section according to the rolling time domain optimization results. In this case, the elastic deformation section of the steel cable dynamically migrates from the high stress area to the low stress area. The peak stress that was originally concentrated in a local area is dispersed into multiple smaller stress peaks and transmitted to the slope anchoring foundations at different locations. Furthermore, the change in the tilt angle of the photovoltaic module is kept below a preset threshold.

10. The system according to claim 1, characterized in that, The terrain sensing unit also includes an initial calibration mode, which is used during the installation and commissioning phase. Under windless and unloaded conditions, and with the rigid beam and sliding component in their initial zero position, the actual initial tilt angle of each photovoltaic module or each group of photovoltaic modules is recorded, and the initial tilt angle is stored as a reference value in a non-volatile memory. During normal operation, the real-time tilt deviation is calculated as the measured tilt angle minus the initial tilt angle of the reference.

11. The system according to claim 1, characterized in that, The adaptive feedback unit is also used to send a wind field model failure warning signal to the remote monitoring platform when the actual load deviates from the predicted value by more than a preset relative error threshold for a consecutive preset number of times, and automatically switch to the standby control mode, while retaining the original control log for offline analysis; when the error falls back to within the threshold after the wind field model parameters are updated, it can automatically switch back to the normal predictive control mode, or request confirmation from the operator before switching.

12. The system according to claim 1, characterized in that, The independent drive unit is also used to perform periodic self-checks under windless and unloaded conditions. The self-check process includes: driving all cable clamping mechanisms to travel through at least three stroke positions from release to clamping in a preset sequence, while driving the sliding component to travel through at least five angular positions from its lower limit to its upper limit, and recording the actual response time, positioning accuracy, and force feedback curve of each actuator; generating a health status report after the self-check is completed; if it is found that the response time of an actuator exceeds the limit or the positioning deviation exceeds the allowable value, the actuator is automatically marked as a maintenance-needed state and a maintenance plan is reported.

Images