Photovoltaic panel self-energized dust removal method and system based on radiation cooling and thermoelectric effect

Abstract

The invention discloses a photovoltaic panel self-energized dust removal system based on radiation cooling and a thermoelectric effect. The core innovation of the photovoltaic panel self-energized dust removal system lies in that a self-sustaining intelligent ecological system in which energy and substances are completely closed is constructed. The system overturns the dependence of a traditional cleaning mode on an external power grid and a water source, and autonomous operation is realized through two environmental energy capture mechanisms: firstly, power is generated through a thermoelectric module by utilizing the temperature difference of the back surface of a photovoltaic panel; and secondly, atmospheric moisture is condensed by means of a panel radiation cooling coating, and is stored and released as required through an intelligent material. The control system adopts strict AND logic, and cleaning operation is triggered only when electric energy and condensate water are sufficient at the same time. The executing mechanism integrates large-stroke stable movement of the rigid portal frame and microcosmic force control of the flexible scraping piece, and efficient and lossless cleaning is achieved. After operation, the system automatically returns to a monitoring state, and a complete negative feedback control loop is formed after resources are supplemented by environment energy. According to the scheme, cost consumption of operation and maintenance of the photovoltaic power station is converted into an intelligent unit which autonomously creates value, and a fundamental breakthrough of an automatic cleaning technology is represented.

Description

Technical Field

[0001] This invention belongs to the field of intelligent automatic control applications and relates to a method and device for dust removal of photovoltaic panels in a photovoltaic power station capable of self-harvesting energy and self-cleaning at night. This invention eliminates the reliance on external robotic brushes for photovoltaic panel cleaning, instead integrating replacement capabilities into the photovoltaic array system itself, achieving true self-repair. This invention represents a cross-disciplinary fusion of environmental energy harvesting and ultra-low-power electromechanical systems, providing a completely self-consistent, off-grid operating solution. It is highly suitable as a supplement to existing cleaning solutions and can be applied to distributed rooftop photovoltaic scenarios that are extremely sensitive to noise, water consumption, and maintenance costs. With advancements in materials science (such as more efficient flexible thermoelectric generators and smart responsive materials) and micro-energy technologies, such self-sustaining "environmentally powered" systems are expected to see even more applications. Background Technology

[0002] Photovoltaic power plants, as a new generation of energy systems, are widely used. These plants typically operate continuously outdoors in strong sunlight, making fault detection and repair, particularly for those caused by dirt on the photovoltaic panels, a significant maintenance challenge. Therefore, the most basic maintenance task is cleaning the photovoltaic panel surfaces. Traditional methods rely on external robots or manual labor. External robots primarily use roller brush systems for cleaning. This invention, however, is a highly ingenious and cutting-edge development that successfully integrates energy harvesting, resource recovery, and intelligent control. It overturns the traditional approach of consuming energy for cleaning, instead collecting and utilizing two types of environmental energy naturally dissipated by the photovoltaic panels at night: utilizing the temperature difference between the back of the photovoltaic panel (hot) and the ambient air (cold), a thermoelectric generator module generates a weak but continuous flow of electricity; this collected electricity then drives a micro-motor that powers a low-damping, hydrophilic scraper mechanism. The scraper uses the condensate naturally generated when the photovoltaic panel surface radiates below its dew point temperature at night as a cleaning medium, performing a horizontal scraping motion. This invention embodies the core concept of achieving self-sustaining cleaning by utilizing the natural diurnal energy cycle. Summary of the Invention

[0003] The invented photovoltaic panel self-powered dust removal method and system based on radiative cooling and thermoelectric effects consists of four highly coordinated core subsystems: an energy harvesting and storage subsystem, a moisture harvesting and management subsystem, an execution subsystem, and an intelligent control system. The energy harvesting and storage subsystem is characterized by its use of the Seebeck effect to directly convert the temperature difference (typically 10-15°C) between the module's nighttime heat dissipation and the ambient cold source into electrical energy through a thermoelectric generator array installed on the back of the photovoltaic panel. The generated milliwatt-level power is managed by efficient circuitry and stored in a hybrid energy storage unit combining a solid-state supercapacitor with excellent low-temperature characteristics and a thin-film lithium battery. This provides off-grid, self-sustaining driving energy for the entire system and is the cornerstone of this invention's energy autonomy. The moisture harvesting and management subsystem utilizes the moisture that naturally condenses on the photovoltaic glass surface due to radiative cooling below the dew point. This moisture is efficiently adsorbed, directionally transported, and temporarily stored through a hydrophilic capillary network and smart materials (such as temperature-sensitive hydrogels). This provides a zero-water-consumption, zero-pollution in-situ cleaning medium, replacing the external water source required for traditional cleaning methods. The core technical feature of the execution subsystem lies in its rigid gantry structure, which combines vertical rail lifting with horizontal beam scanning. High-rigidity double columns and heavy-duty guide rails enable stable lifting over a long stroke, fundamentally solving the vibration and insufficient rigidity problems inherent in long cantilever designs. Simultaneously, it incorporates an intelligent scraper with an adaptive floating mechanism and six-dimensional force sensing. Through real-time closed-loop force control and servo coordination, it achieves microscopic flexible cleaning of the photovoltaic panel surface while ensuring precise macroscopic positioning, thus achieving the dual goals of efficient operation and zero damage to the photovoltaic panel surface. The intelligent control system adopts a hybrid architecture of "centralized management and decentralized execution," consisting of onboard intelligent units deployed on each photovoltaic panel, a regional controller coordinating the entire array, and a gantry control system. This balances independent monitoring of individual panels with coordinated operation of the entire array. Its operation is based on intelligent decision-making through multi-source information fusion: the onboard intelligent unit continuously collects and reports energy, moisture and environmental data of the board surface, and the onboard intelligent unit integrates and generates an array status map. When the conditions of sufficient energy, water storage readiness and dew point are met at the same time, the gantry system is triggered to execute a precise "zigzag" cleaning path. During the scraping process, the contact pressure is controlled in real time through a six-dimensional force sensor, thus forming an adaptive negative feedback working loop that relies on natural energy supply and prevents over-cleaning.

[0004] The invented photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effects is a typical environmental energy-driven closed-loop control system. The energy subsystem and water molecule system, as parallel front-end acquisition units, convert waste heat (temperature difference) and atmospheric humidity in the environment into usable electricity and water resources, respectively. The control subsystem, as the central processing and scheduling unit, performs state assessment and joint criteria on the resources acquired from the front end. Its output strictly depends on the common readiness state of the two front-end inputs, forming an AND logic: instructions are sent to the actuator only when both conditions—sufficient electrical energy and available condensate water—are met simultaneously. After the cleaning action is completed, the system returns to the monitoring and acquisition state, thus forming a completely self-sustaining, externally dependent negative feedback control loop. In summary, this system, through sophisticated thermoelectric conversion, condensate water collection, and ultra-low power consumption control technology, captures, converts, and intelligently schedules discrete, low-grade energy and material resources in the environment, ultimately achieving the goal of self-sustaining cleaning of the photovoltaic panel surface, embodying the core design concepts of energy self-consistency and material closed-loop. Without affecting the daytime power generation of the photovoltaic panel, all components of the energy harvesting and storage subsystem are compactly integrated within the photovoltaic panel's own frame, forming an independent intelligent module unit. The glass surface of the photovoltaic panel is the main working area. To enhance nighttime condensation, the surface is functionally divided into zones: Imagine a standard photovoltaic panel with two transparent strips (approximately 10-15 cm wide) running vertically across its surface. These strips appear identical to the surrounding glass, but their surface is coated with a special radiative cooling coating. During the day, it does not obstruct sunlight transmission; at night, it acts like a radiator, radiating heat from the panel surface more efficiently to the cold external space, thus making these two areas cooler than other parts of the panel and more prone to condensation. The cleaning arm's scraping path is designed to target these two condensation-rich zones. The back of the photovoltaic panel is the main installation location for the energy harvesting system, the core of which is a non-uniform arrangement of a thermoelectric generator array. The heat distribution characteristics on the back of the photovoltaic panel (aluminum frame and backsheet) are as follows: the central area dissipates heat slowly, resulting in relatively high temperatures at night; the edge and frame areas dissipate heat quickly, resulting in relatively low temperatures at night. The thermoelectric generator array arrangement strategy is to cover the central area of ​​the back panel with a high-density area to utilize its high-temperature hot end; and to place a low-density area near the frame area to match its low-temperature cold end, generating an effective temperature difference ΔT. The thermoelectric generator array converts ΔT into electrical energy. The thermoelectric generator modules, resembling tiny square modules (e.g., 25mm x 25mm), are tightly adhered to the photovoltaic panel backsheet using thermally conductive adhesive. All thermoelectric generator modules on the back of the same photovoltaic panel are first connected in series to increase the output voltage and reduce transmission losses. Then, the thermoelectric generator arrays of multiple photovoltaic panels are connected in parallel in a DC combiner box, and the combined output charges a shared energy storage battery. A narrow, waterproof box is installed on the lower frame of the photovoltaic panel, integrating the corresponding microcontroller, sensor interface, and cleaning arm drive circuitry.

[0005] The moisture collection and management subsystem is divided into four functional layers, with all components arranged around the lower frame of the photovoltaic panel: the top layer is the photovoltaic panel glass surface coated with a radiation-cooling coating, which is the condensate generation area; the middle layer is the frame water collection system, integrated inside the standard photovoltaic aluminum frame, without occupying additional space; the bottom layer is the cleaning arm water storage unit, integrated with the cleaning actuator; the bottom layer is the composite scraper assembly, which directly contacts the photovoltaic panel surface for cleaning; the core component of the condensation generation layer (top layer) is the photovoltaic glass panel coated with a radiation-cooling coating. At night, the coating radiates heat from the panel surface to the cold space through the "atmospheric window" (8-13μm band), making the panel surface temperature lower than the ambient air dew point. Water vapor in the atmosphere condenses into liquid water droplets upon contact with the cold surface. The core component of the collection and transmission layer (middle layer) is the hydrophilic capillary water collection belt and manifold integrated within the lower frame of the photovoltaic panel. Under the influence of gravity, the condensate flows along the panel surface to the lower edge, where it is instantly adsorbed by the super-hydrophilic surface layer of the water collection belt. Moisture is transported laterally to the collection point at the center of the frame through the porous structure and capillary action within the material, and then delivered to the cleaning arm via pipes. The core component of the storage and release layer (bottom layer) is the water storage chamber built into the cleaning arm, the core of which is a temperature-sensitive hydrogel module. At night when the temperature is low, the hydrogel is in a hydrophilic swelling state, actively absorbing and locking in moisture from the pipes. When the temperature rises during the day or frictional heat is generated by the cleaning action, the hydrogel temperature exceeds its critical point, the physical structure changes from hydrophilic to hydrophobic, a contraction phase transition occurs, and the stored moisture is squeezed out. The core component of the cleaning execution layer (bottom layer) is the composite scraper. Moisture released from the water storage chamber is evenly distributed through the porous base plate and continuously supplied to the hydrophilic layer of the scraper. The scraper remains moist during scraping, achieving a "water-washing" cleaning of the photovoltaic panel surface, significantly improving the cleaning efficiency of dry dust. The complete flow path of moisture in the system is a closed loop of water circulation from the atmosphere to the atmosphere. The first step involves water vapor condensing on the low-temperature plate surface, following a nucleation-growth mechanism. The second step involves liquid water flowing towards the frame by gravity, where it is captured and transported by the multi-level capillary structure of the water collection belt. The third step involves water entering the storage chamber, where it is absorbed and stored by the temperature-sensitive hydrogel through osmosis. The fourth step involves a temperature-triggered phase transition of the hydrogel when cleaning is needed, releasing the stored water. The fifth step involves the released water being evenly distributed to the scraper blades through a distribution system, achieving moist cleaning. The sixth step involves some of the water evaporating back into the atmosphere after cleaning, some being removed with the dirt, and a small amount remaining for the next cleaning cycle. Throughout this process, the water undergoes multiple phase transitions: from gas to liquid, from liquid to gel, from gel to liquid, and from liquid to gas. The entire process requires no external energy input and is driven entirely by natural physical processes. This system, through ingenious multi-layered structural design and material innovation, achieves a complete closed loop for the collection, storage, and on-demand utilization of atmospheric water, providing a sustainable solution for photovoltaic cleaning with zero water and zero energy consumption.

[0006] The execution subsystem employs a rigid gantry structure with vertical rail lifting and horizontal beam scanning, specifically designed for large-scale, high-array photovoltaic power plants. Its core is the robust vertical rail bearing the main load and positioning, achieving stable and reliable large-range vertical movement, fundamentally solving problems such as insufficient rigidity and excessive vibration associated with long cantilever horizontal telescopic solutions. Its mechanical skeleton consists of the vertical rail and the horizontal beam. The vertical rail, as the system's main spine, comprises double columns, heavy-duty linear guides, and a rack and pinion drive. It provides a large stroke and high rigidity lifting motion along the Z-axis, forming the foundation for the entire execution mechanism to achieve height coverage. The horizontal beam system is mounted on the lifting platform, acting as an extended arm. A traveling trolley is mounted on it, driving the scraper to move horizontally along the X-axis to cover the width of the photovoltaic array. The beam adopts a box-shaped structure with a pre-set camber to ensure rigidity over the span and prevent sagging. The intelligent scraper assembly has an adaptive floating mechanism: the scraper is connected to the beam trolley via a spring or airbag mechanism, forming a "flexible joint" that allows it to adaptively conform to the uneven surface of the photovoltaic panels, ensuring uniform cleaning pressure. Its sensing and force control system integrates a six-dimensional force sensor and a distance sensor to sense the contact pressure and distance to the board surface in real time, and feeds the data back to the control system to achieve closed-loop precise control of the scraper pressure, protecting the board surface while cleaning efficiently. Its servo drive system uses servo motors for both vertical lifting (Z-axis) and horizontal movement (X-axis), combined with an absolute encoder, to achieve precise positioning and speed control throughout the entire stroke, eliminating the need for repeated zeroing. The central intelligent control is based on an industrial PLC, connecting all drives and sensors via a PROFINET / EtherCAT bus network. It performs zigzag path planning and coordinates dual-axis interpolation motion, while simultaneously adjusting the scraper pressure in real time based on sensor feedback, achieving fully automatic intelligent cleaning. Multi-layered safety protection includes a hardware emergency stop circuit, software limits, anti-fall mechanical clamps, and real-time wind speed monitoring (automatic shutdown if exceeding limits), ensuring system safety in adverse weather conditions. During system operation, the vertical track first drives the crossbeam to the target height, and then the horizontal trolley moves the scraper laterally. During this process, the scraper's sensing system provides real-time feedback, and the control system dynamically adjusts the pressure and attitude. This highly collaborative approach, where a rigid subsystem handles macroscopic precision positioning while a flexible subsystem handles microscopic fine-tuning, enables the system to not only cover extremely large areas but also achieve high-quality, adaptive, and zero-damage cleaning operations. This gantry-type solution solves the engineering challenges of high-altitude cleaning through a rigid vertical lifting main structure, achieves fine-tuning through flexible intelligent scraper terminals, and deeply integrates the two through an integrated bus control system. It represents an industrial-grade solution that moves from active to intelligent, reliable, and efficient operation, providing a solid technical foundation for the automated operation and maintenance of large-scale photovoltaic power plants.

[0007] The intelligent control system adopts a hybrid architecture of centralized management and distributed execution, including onboard intelligent units and area controllers and gantry control systems, to address the needs of independent management of individual photovoltaic panels and coordinated cleaning of the entire array. The onboard intelligent units are integrated into a waterproof box on the lower frame of each photovoltaic panel and use ultra-low power microcontrollers (such as those based on the ARM Cortex-M series). The onboard intelligent unit functional module has an energy management interface that monitors the voltage / current output of the thermoelectric array on the board, as well as the voltage (SOC state) of the hybrid energy storage unit (supercapacitor + thin-film lithium battery); the environmental sensing module integrates a high-precision temperature and humidity sensor (monitoring ambient temperature and humidity, calculating dew point temperature) and a back-side temperature sensor (indirectly reflecting the thermoelectric power generation potential); the moisture management interface monitors the "water level sensor in the storage chamber" integrated in the frame or indirectly determines the water storage volume through the state (temperature) of the temperature-sensitive hydrogel; the local drive circuit drives the small actuators attached to the board (such as future expandable micro pre-wetting nozzles), currently mainly for status reporting; the communication module uses low-power wireless (such as LoRa, Zigbee) or wired (such as RS-485) methods to upload the board's status data to the area controller; the onboard intelligent unit local control algorithm (running on the microcontroller) performs dew point calculation and condensation prediction status assessment, and the onboard intelligent unit global decision and scheduling algorithm (running on the PLC) receives and processes the status reports of all onboard intelligent units in the area, generating an array-level status map. The cleaning task trigger criteria (core AND logic) are as follows.

[0008] The area controller and gantry control system of the intelligent control system are deployed at one end of the photovoltaic array, integrated with or adjacent to the gantry-type cleaning robot main unit, and use an industrial-grade programmable logic controller (PLC) or a high-performance embedded industrial computer. Its functional modules include: a main control PLC, the system's "decision center" and "motion control core"; a communication gateway that collects data uploaded by all onboard intelligent units within its jurisdiction (such as a row or an array); a motion control unit that controls the gantry's vertical lifting servo motor and horizontal beam traveling servo motor via PROFINET / EtherCAT bus; a sensor interface that receives feedback signals from the six-dimensional force sensor and distance sensor of the gantry's intelligent scraper assembly, as well as wind speed sensor signals; a safety loop integrating hardware emergency stop buttons, software limit switches, safety relays, etc.; and an execution terminal—the integrated sensing and control module of the intelligent scraper assembly—that measures the contact force and torque between the scraper and the plate surface in real time through a six-dimensional force sensor; performs distance measurement for anti-collision and floating height reference during scraper lifting and lowering; processes local sensor data, and communicates in real time with the area controller's PLC via bus to achieve closed-loop pressure control. The operation of this part is as follows: Path planning is performed to plan the overall movement path of the gantry; during horizontal scanning, precise positioning ensures that the scraper cleaning trajectory coincides with the two radiative cooling coating strips (condensate enrichment strips) on the photovoltaic panel surface, maximizing the utilization of condensate; motion control is performed using dual-axis interpolation to coordinate the Z-axis (vertical) and X-axis (horizontal) servo motors to achieve smooth and efficient continuous path movement; the scraping speed is dynamically adjusted according to cleaning needs and energy status; the optimal cleaning pressure Ftarget is set according to the photovoltaic panel surface type (glass, coating), and a scraper force closed-loop control algorithm is run (running on the scraper module microprocessor and PLC); the actual pressure Factual is measured in real time by a six-dimensional force sensor; the error e=Ftarget−Factual is calculated, and a control signal is output through a PID algorithm to adjust the scraper floating mechanism (such as controlling the airbag pressure through a proportional valve or adjusting the spring preload) to stabilize Factual near Ftarget, achieving "gentle yet effective" cleaning.

[0009] The closed-loop control circuit workflow is as follows: First, in the monitoring and acquisition phase (normal state), all onboard intelligent units are in ultra-low power monitoring mode. The thermoelectric array continuously generates and stores electricity, and at night, radiative cooling continuously condenses and stores water. Onboard intelligent units periodically evaluate and report the status. Next, in the decision-making and triggering phase, the onboard intelligent units continuously analyze the status map. When the above "cleaning task triggering criteria" are met, they are activated from "sleep" mode. Next, in the execution and cleaning phase, the onboard intelligent units activate the gantry system, vertically lifting and lowering it to the top of the array. The horizontal beam controls the wet intelligent scraper, scanning along the planned path (aligned with the condensation zone). Real-time force control is implemented; throughout the scraping process, the scraper force closed-loop control algorithm continuously operates, adapting to the surface undulations. Water release is synchronized; during cleaning, the temperature-sensitive hydrogel in the water storage chamber releases water due to friction or environmental temperature rise, wetting the scraper. Finally, in the reset and feedback phase, after completing a single cleaning cycle, the gantry resets to the standby position. The system re-enters the monitoring and data acquisition phase. Each cleaning action consumes stored electrical and water resources, causing the system to fall out of the "ready" state until environmental energy replenishes it above the threshold, thus naturally forming a negative feedback control loop to prevent ineffective or excessive cleaning.

[0010] The intelligent control system organically combines onboard ultra-low power sensing, regional intelligent decision-making, and high-precision motion control. It is not only the trigger for cleaning actions but also the scheduling center of the entire self-powered ecosystem. Through rigorous logical judgment and closed-loop control, it ensures precise coordination and dynamic balance among energy, water, and motion, ultimately achieving the ultimate goal of effective cleaning at the right time, using its own resources, and in the optimal way. This perfectly supports the invention as a truly self-maintaining intelligent skin.

[0011] The invented photovoltaic panel self-powered dust removal method and system based on radiative cooling and thermoelectric effects, an innovative solution combining environmental energy self-driven technology with an industrial-grade gantry execution system, demonstrates revolutionary comprehensive advantages. Its advantages are not simply additive, but rather a qualitative leap achieved through fundamental principle innovation and system-level collaborative design. In short, this solution solves the survival problem of automation through environmental energy self-drive, the efficiency problem of automation through precise gantry execution, and the intelligence problem of automation by integrating the two through intelligent closed-loop control. It is not merely a "cleaning tool," but a "self-maintaining intelligent skin" that empowers photovoltaic power plants. This system transforms power plant operation and maintenance from a "cost center" that continuously consumes external resources and relies on manual intervention into a "value creation unit" capable of self-sustaining, self-optimizing, and continuously improving power generation revenue, representing the future direction of intelligent and autonomous operation and maintenance of photovoltaic power plants. Attached Figure Description

[0012] Appendix Figure 1This is a schematic diagram of the energy harvesting and storage subsystem of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0013] Appendix Figure 2 This is a schematic diagram of the moisture collection and management subsystem of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0014] Appendix Figure 3 This is a schematic diagram of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0015] Appendix Figure 4 This is a schematic diagram of the overall function of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0016] Appendix Figure 5 This is a schematic diagram of the control relationship of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0017] Appendix Figure 6 This is a functional analysis diagram of the front side of the photovoltaic panel of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0018] Appendix Figure 7 This is a functional analysis diagram of the back of a photovoltaic panel of the self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0019] Appendix Figure 8 This is a schematic diagram of the water collection system structure of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0020] Appendix Figure 9 This is a flowchart of the state machine of the main control system of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0021] Appendix Figure 10 This is a flowchart of the cleaning decision algorithm (AND logic) of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0022] Appendix Figure 11 This invention relates to the closed-loop control process of scraper pressure in a photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect.

[0023] Appendix Figure 12 This is a flowchart of the gantry motion control (zigzag path) of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0024] Appendix Figure 13 This is an energy management and scheduling flowchart of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0025] Appendix Figure 14 This is a flowchart of the fault diagnosis and fault-tolerant processing of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention.

[0026] Appendix Figure 15 This is a LoRa communication and data recording flowchart of the photovoltaic panel self-powered dust removal method and system based on radiation cooling and thermoelectric effect described in this invention. Detailed Implementation

[0027] According to the aforementioned photovoltaic panel self-powered dust removal method and system based on radiative cooling and thermoelectric effect, its specific implementation involves capturing electrical energy and water resources from the environment through thermoelectric effect and radiative cooling, respectively; the controller triggers cleaning only when electrical energy is sufficient and condensate water is available; after cleaning consumes resources, the system re-enters the acquisition state, waiting for environmental energy replenishment, forming a fully self-sustaining negative feedback closed loop. The parameters of the implemented system are shown in Table 1, which quantitatively defines all key parameters: from overall size (2.5 x 2.0 x 0.5 meters), cleaning pressure (5 N ± 1 N) to verification targets (completion of closed loop, energy self-sufficiency).

[0028] Table 1 Model Specifications and Objectives project Specifications Remark Number of photovoltaic panels 4 pieces (arranged in a 2x2 pattern) Standard 60-cell component Overall dimensions 2.5m (W) × 2.0m (H) × 0.5m (D) movable frame The energy harvesting and storage subsystem of this implementation demonstrates that the temperature difference caused by waste heat on the back of the photovoltaic panel can be converted into sufficient off-grid electricity to drive the entire system. Four commercial TEC1-12706 thermoelectric modules are attached to the central area of ​​the back of each model photovoltaic panel to simulate a non-uniform thermal field. At night, the natural temperature difference between the back of the panel and the environment (simulating approximately 10-15°C) drives these modules to generate milliwatt-level power. This arrangement directly verifies the engineering feasibility of the core idea of ​​"utilizing the photovoltaic panel's own heat dissipation." The customized ultra-low power management uses a custom PCB circuit board based on the TI BQ25570 chip. This circuit is specifically designed to efficiently harvest energy from extremely low voltage and weak current (such as electricity generated by TEG). It converts unstable thermoelectric power generation with an efficiency exceeding 85% into a stable power supply suitable for charging energy storage components, and its own operating power consumption is extremely low, which is key to achieving a positive energy surplus in the model. The hybrid energy storage unit used employs a supercapacitor (5.5V / 1F): simulating the pulsed high current demand during the start-up of the cleaning motor to ensure stable operation. Thin-film lithium batteries (3.7V / 500mAh) simulate long-term, stable energy storage, and their excellent performance ensures reliability during multiple demonstration cycles. The combination of these two components provides buffering against high-power pulses and continuous power delivery, acting as both an energy "buffer" and "battery" for stable system operation. The composition of its energy harvesting and storage subsystem is shown in Table 2.

[0029] Table 2 Energy Harvesting and Storage Subsystem Component Module Technical Specifications Function Description Installation location Thermoelectric generator (TEG) Model: TEC1-12706 Size: 40×40mm Output at ΔT=10℃: ~50mW / piece Based on the Seebeck effect, the temperature difference between the back of the photovoltaic panel and the environment is converted into electrical energy. High-density arrangement in the central area on the back of the photovoltaic panels Energy storage unit Supercapacitor: 5.5V / 1F; Thin-film lithium battery: 3.7V / 500mAh Hybrid energy storage: Supercapacitors handle pulsed loads, while lithium batteries provide stable energy. Integrated into the lower frame control box Energy management circuit Chip: BQ25570 Efficiency: >85% MPPT (Maximum Power Point Tracking) and Ultra-Low Power Charging Management Custom PCB Output performance Estimated total power output at night: 50-100mW; Energy storage charging time: 2-3 nights Power the entire control system and actuators The control hardware architecture is based on an ultra-low-power MCU, integrating multiple high-precision sensors via the I2C bus and controlling actuators through dedicated drivers. Remote monitoring is achieved via LoRa (see Table 3). All core components are low-power models, ensuring the system's standby power consumption is significantly lower than the thermoelectric power generation power. Sensors cover comprehensive information on the environment (temperature and humidity), the system itself (power and water volume), and the execution process (blade pressure). Industrial-grade connectors and isolation design ensure long-term stability during outdoor operation. Expansion interfaces: The MCU has ample GPIO and communication interfaces for future expansion of more functions (such as rain and snow sensors, dust detection sensors, etc.). All component selection prioritizes milliwatt-level power consumption, ensuring the system's standby power consumption (estimated <2mW) is significantly lower than the minimum nighttime output of the thermoelectric power generation module (>50mW), fundamentally guaranteeing energy self-sufficiency through hardware.

[0030] The perception layer is equipped with a triple perception network consisting of the environment (temperature and humidity, infrared thermometry), the system itself (voltage and current), and the execution end (six-dimensional force sensor), providing comprehensive and high-precision data input for the decision-making algorithm. The execution layer adopts cost-effective stepper motors and drive solutions, which reduce motion energy consumption while meeting the model's accuracy and torque requirements through efficient drive circuitry.

[0031] The integrated LoRa wireless module enables remote status monitoring, while the MCU's rich interfaces reserve space for future functional expansion (such as visual dust recognition).

[0032] Table 3 Intelligent Control and Execution Subsystem - Hardware Architecture Configuration Table Module Categories Component Name Model / Specification quantity Key parameters and performance Main functions Interface / Communication Method Control Core Main Microcontroller (MCU) STM32L452RET6 1 ARM Cortex-M4, 80MHz, ultra-low power consumption (< 100μA / MHz), integrated FPU The system acts as the central controller, running decision-making algorithms and coordinating the work of each module. GPIO, I2C, SPI, UART, ADC Auxiliary power management IC BQ25570 1 Ultra-low power energy harvesting manager, startup voltage 0.1V, integrated MPPT Manage the TEG array inputs to efficiently charge the energy storage units. I2C (configurable) Sensing unit Ambient temperature and humidity sensor SensirionSHT35-DIS 2 Accuracy: ±0.2°C, ±2%RH, low power consumption (1.2μA average) Monitor ambient temperature and humidity, and calculate dew point temperature. I2C Infrared surface temperature sensor MelexisMLX90614ESF 1 Non-contact temperature measurement, accuracy ±0.5°C, field of view 90°. Non-contact measurement of photovoltaic panel glass surface temperature I2C Six-dimensional force / torque sensor ATI Mini40 1 Measuring range: Fx,y: ±40N, Fz: ±120N, Tx,y,z: ±2Nm Real-time sensing of the force and torque at contact between the scraper blade and the plate surface is used for closed-loop force control. Analog voltage output / EtherCAT (optional) Current / voltage monitor TI INA219 1 I2C interface, 0-26V, ±1% accuracy Monitor the voltage, current, and power of energy storage units (batteries, capacitors). I2C Execution driver Dual-channel DC motor driver TB6612FNG 2 Each channel provides a continuous output of 1.2A, with built-in standby and protection circuitry. Drive vertical (Z-axis) and horizontal (X-axis) stepper motors PWM, GPIO Bipolar stepper motor (Z-axis) 42HSC3401N 1 Step angle 1.8°, holding torque 0.4 Nm, current 1.5 A / phase Drive the vertical gantry lifting motion Four-wire system, connected to the driver. Bipolar stepper motor (X-axis) 42HSC3401N 1 Step angle 1.8°, holding torque 0.4 Nm, current 1.5 A / phase Drive the horizontal sweeper carriage to move Four-wire system, connected to the driver. Communication and Interaction Low power wireless module Ra-02(SX1278) 1 LoRa modulation, 433MHz, communication distance >2km, receiving current <10mA Upload system status (energy, water consumption, cleaning records) to the remote monitoring terminal. SPI Local status display 0.96-inch OLED 1 SSD1306 driver, 128x64 resolution, I2C interface Local real-time display of key system status parameters (voltage, water volume, mode) I2C Debugging and Programming Interface SWD & UART 1 Standard JTAG / SWD interface, USART to USB Used for firmware burning, debugging, and runtime log output. SWD, UART Auxiliary and Interface Level conversion and isolation TXS0108E, ADuM3151 1 each 5V / 3.3V bidirectional level conversion, high-speed digital isolator Ensure stable communication between devices in different voltage domains and isolate motor interference. I2C, SPI Waterproof connectors JST, M8 / M12 series several IP67 protection rating For connecting all external sensors, motors, and power supplies, ensuring outdoor reliability. Custom wire harness power supply System power input From hybrid energy storage units 1 Input: 3.0V - 5.5V DC Provide operating power to the entire control subsystem Power terminals Local LDO / DC-DC TPS7A series, TPS63020 several High-efficiency, low-noise voltage regulation, such as 3.3V and 5V output. Provides a stable and clean voltage rail for MCUs, sensors, and communication modules. The mechanical execution system combines a rigid frame with a flexible terminal, as shown in Table 4. The "rigid" frame, achieved through a double-column heavy-duty vertical track and a pre-arched box-shaped horizontal beam, solves the structural deformation and vibration problems under long spans and large strokes, ensuring the stability and precision of the motion platform—the foundation for reliable cleaning. The "flexible" terminal, the intelligent scraper assembly, is the highlight of the entire system. Its integrated floating mechanism and six-dimensional force sensor form a pressure closed-loop control, allowing the scraper to adaptively conform to the panel surface like a human hand, ensuring cleaning power while eliminating the risk of scratching the photovoltaic glass. Integrated design: The table illustrates the high degree of integration between mechanics, water supply, and electrical control. The routing paths of the water storage chamber, pipes, and cables are pre-planned in the structural design, resulting in a clean and reliable model appearance, rather than a temporary patchwork. Safety redundancy: Multi-layered safety design, from hardware emergency stop to mechanical anti-fall clamps, ensures the system can safely fail even if anomalies occur during demonstrations or tests—a key manifestation of engineering thinking. This specification defines a high-performance two-dimensional precision motion platform that perfectly combines macroscopic positioning rigidity with microscopic operational flexibility, serving as the physical foundation for high-quality and highly reliable cleaning operations.

[0033] Table 4 Specifications of Gantry Actuator Models Components / assemblies Design parameters and specifications Material / Model quantity Functional Description and Design Considerations Main framework External dimensions: 2.5m (width) × 2.0m (height) × 0.5m (depth) 4040 aluminum profile (anodized) 1 set The system features a rigid base that provides overall support; pre-set mounting holes and four universal casters with brakes for easy movement and positioning. Vertical track (Z-axis) Travel: 1.5m Maximum speed: 0.1m / s Positioning accuracy: ±1mm Guide rail: MGN12H heavy-duty linear guide rail; Drive: 42 stepper motor + 16GT synchronous belt pulley; Transmission: 16GT synchronous belt and tensioning mechanism. 2 sets The system's main spindle. A symmetrical dual-column layout bears the entire vertical load. Guide rails ensure high rigidity and low-friction movement; synchronous belt drive enables smooth lifting and lowering. Horizontal beam (X-axis) Span: 2.0m Maximum speed: 0.2m / s Positioning accuracy: ±0.5mm Preset camber: 3mm Beam: 2020 aluminum profile (internal reinforcing ribs) Guide rail: MGN9C linear guide Drive: 42 stepper motor + 1.5 module rack and pinion 1 set The actuator's extension arm features a box-shaped structure with a pre-set camber to resist downward deflection caused by its own weight, ensuring rigidity and levelness over the span. A rack and pinion drive provides high-precision, long-distance linear motion. Walking car Dimensions: 150mm×200mm×120mm Integrated features: Floating scraper mechanism, sensor interface, water supply interface Main body: 3D printed (carbon fiber PLA) or aluminum plate processed slider: MGN9C matching slider 1 Mounted on a horizontal beam, it supports the scraper assembly. It integrates a mounting base and adjustment mechanism for the internal scraper floating mechanism, and provides sensor and water pipe interfaces. Intelligent scraper assembly Scraper width: 300mm; Floating stroke: ±20mm; Target cleaning pressure: 5N ± 1N; Water supply method: Porous ceramic uniform water permeation. Scraper: Silicone rubber body + hydrophilic non-woven fabric outer layer; Floating mechanism: Linear bearing + precision spring (or miniature airbag); Force sensor: Six-dimensional force sensor (Mini40); Distributor: Porous ceramic plate. 1 set The core execution terminal adapts to the undulations of the plate surface through a floating mechanism; a force sensor provides real-time feedback of contact force to achieve precise closed-loop pressure control; and a multi-hole distributor ensures that the scraper is wetted across its entire width. Lifting Platform Dimensions: 400mm × 500mm Load Capacity: ≥30kg 5mm thick aluminum sheet (laser-cut and bent) 1 It connects the vertical track slider to the horizontal crossbeam, serving as a movable base for horizontal movement. Reinforcing ribs and leveling screws are provided to ensure the mounting surface is level. Drive and feedback systems Z-axis motor: 42HSC3401N (1.5A, 0.4Nm) X-axis motor: 42HSC3401N (1.5A, 0.4Nm) Driver: TB6612FNG Position feedback: Origin / limit sensor (Hall effect or photoelectric) Stepper motor, motor driver, limit switch 2 sets each Provides motion power. The stepper motor features open-loop control, coupled with a high-precision limit sensor to achieve reliable origin return and travel protection. The driver is controlled by the main controller via pulse direction signals. Water storage and supply system Water storage chamber volume: 50mL; Hydrogel module: PNIPAM thermosensitive gel (LCST≈32℃); Water supply pipe: 3mm inner diameter silicone tubing. 3D-printed water storage cavity, custom-made temperature-sensitive hydrogel, food-grade silicone tubing 1 set Integrated design with the cleaning arm. It stores condensate overnight and releases it during cleaning when temperature triggers. A multi-hole distributor connected to the scraper via piping enables a "washing" function. Safety protection devices Emergency stop button; Mushroom-shaped self-locking hardware limit switch; One anti-fall clamp at each end of the travel; Power-off self-locking (Z-axis) wind speed monitoring interface; Reserved installation position. Industrial emergency stop switches, micro limit switches, safety clamps 1 set Multi-layered safety redundancy. The hardware emergency stop circuit has the highest priority; mechanical limit switches prevent overtravel; the anti-fall clamp automatically locks the vertical rail in the event of power failure or malfunction; and the wind speed sensor interface is used for linkage protection. Control box Dimensions: 250mm × 200mm × 120mm Protection Rating: IP65 Waterproof plastic or metal electrical box 1 It integrates motor drivers, relays, terminal blocks, controller interfaces, etc. Mounted on the gantry column for easy wiring and maintenance. Design of Control Algorithm for Photovoltaic Self-Powered Dust Removal System. The main control system state machine flowchart adopts a hierarchical design, with the system in the SYS_MONITOR ultra-low power monitoring state (current <500μA) for 90% of the time. Only when all cleaning conditions are met simultaneously will the system sequentially enter the preparation, execution, and completion states, forming a complete cleaning loop. In abnormal situations, the system enters the ENERGY_CRITICAL or SYS_HIBERNATE state to maximize system survival time. This "long sleep-short work" mode is key to the system's energy self-sufficiency. The cleaning decision algorithm (AND logic) flowchart uses strict "AND gate" logic, requiring six conditions to be met simultaneously to trigger cleaning. This design ensures: dual checks of batteries and capacitors to prevent power outages during cleaning; sufficient moisture for wet cleaning; selection of morning hours to avoid nighttime condensation and peak power generation; automatic shutdown during strong winds and rain to protect equipment; and 12-hour intervals to prevent ineffective cleaning and save resources. The algorithm returns specific reasons for failure, facilitating system optimization and debugging. The scraper pressure closed-loop control process adopts the classic PID algorithm, with a six-dimensional force sensor measuring the contact force between the scraper and the plate surface in real time for PID adjustment. It features a fast-response pressure change integral term (I) to eliminate steady-state error, an anti-saturation differential term (D) to prevent integral overshoot, and a low-pass filter to reduce noise. Output limiting ensures that the control quantity is adaptively adjusted within the range of the physical actuator. The target pressure control cycle is dynamically adjusted by 10ms based on the plate surface temperature and dust thickness to ensure real-time pressure response while avoiding excessively frequent adjustments. The gantry motion control (zigzag path) flowchart uses a zigzag path planning to ensure full coverage cleaning: mechanical zeroing is performed, and the position reference is ensured to be accurate before each cleaning. Path calculation is performed: Z-axis (vertical), descending by one scraper width (300mm) each time; X-axis (horizontal) is adjusted, odd passes from left to right, even passes from right to left; S-curve speed planning, acceleration 50mm / s², maximum speed 100mm / s, flexible start and stop to reduce mechanical shock and vibration; motion synchronization, horizontal movement and pressure control are synchronized in real time to ensure cleaning quality; abnormal handling, in case of abnormal pressure, the movement is paused, and after adjustment, the total stroke continues to cover a 2m wide × 1.5m high photovoltaic panel area (model dimensions). The energy management scheduling flowchart employs a four-level priority scheduling to ensure the system's survival under energy-constrained conditions: Emergency Mode (red): Energy < minimum survival value, all non-core functions are shut down, only RTC timing is maintained; Charging Priority Mode (yellow): Energy is insufficient to support cleaning, focusing on energy harvesting and reducing monitoring frequency; Normal Monitoring Mode (green): Energy is sufficient to maintain monitoring, but insufficient to support cleaning; Ready Mode (blue): Energy is sufficient, allowing the cleaning algorithm to be triggered to dynamically switch modes based on the ratio of available energy to the energy required for a single cleaning cycle (820mWh), achieving optimal energy usage.The fault diagnosis and fault tolerance process flowchart adopts a layered inspection strategy to identify common faults and implement corresponding fault tolerance measures: Motor overload: immediately reduce speed; if overload continues, stop immediately to prevent hardware damage. Sensor inconsistency: activate redundant sensors or data fusion algorithms to ensure the system continues to operate and charge; Circuit fault: switch to supercapacitor-only power supply mode, and the system degrades its operation; Cleaning efficiency decreases: adjust the scraper pressure or cleaning speed to adapt to the board surface condition. All faults are recorded in detail and uploaded to the monitoring center via LoRa for remote diagnosis and maintenance. LoRa communication and data recording flowcharts illustrate the adaptive interval strategy employed by the communication system to balance data real-time performance and energy consumption: Dynamic interval adjustment: 30-minute interval for low battery (energy saving); 1-minute interval for cleaning (real-time monitoring); 2-minute interval for fault detection (timely alarm); 5-minute interval for normal monitoring (balancing energy consumption). Data packet design: Includes complete information such as device ID, timestamp, energy, moisture, environment, and status; CRC check ensures data integrity. Reliable transmission: Onboard intelligent unit K-acknowledgment mechanism, 3 retries on failure to ensure no loss of important data. Local backup: All data is simultaneously stored on an SD card to prevent data loss during communication interruptions. The communication module is only activated during transmission and immediately enters sleep mode after transmission is complete, minimizing communication energy consumption. The above seven flowcharts illustrate the internal logic of the system's core algorithms. In actual operation, these algorithms work collaboratively through FreeRTOS task scheduling: see Table 5 for the status table. The system ensures the real-time performance of critical tasks (such as stress control) through carefully designed task priorities and scheduling strategies, while allowing non-critical tasks (such as communication) to execute when the system is idle, achieving the optimal balance between performance and energy consumption.

[0034] Table 5 Task Scheduling Table Task Name Priority Period / Trigger Main Algorithm State machine master controller high Event-driven State machine transition Cleaning Decisions middle 1 minute AND logical judgment Pressure control Highest 10ms PID closed-loop control Motion control high 20ms Path planning + S-curve Energy Management middle 30 seconds Energy Dispatch Fault Diagnosis middle 5 minutes Health check Communication tasks Low Adjustable interval LoRa communication Through practical verification: The system successfully achieved a self-sustaining balance between energy harvesting and consumption (see Table 6). The system can collect sufficient water and significantly improve the power generation performance of photovoltaic panels (see Table 7). The operation results of the intelligent control system prove that the system makes accurate decisions and operates stably and reliably (see Table 8). The feasibility of the principle has been fully verified: The model system operated successfully, fully demonstrating the closed loop from "environmental energy harvesting (electricity + water) → intelligent storage and judgment → triggering and executing cleaning → returning to harvesting after resource consumption". The three core technologies (thermoelectric self-powered, radiative cooling condensate, and low-power intelligent control) have all been proven effective. Core design indicators were exceeded: positive energy surplus: harvesting > consumption, achieving the designed energy self-sufficiency target. Water self-sufficiency: atmospheric condensate production meets cleaning needs, achieving a material closed loop. Significant effects: data on increased light transmittance and power generation confirm the economic value of cleaning. Preliminary system reliability verification: During several weeks of continuous testing, the system performed stably, made accurate decisions, and did not experience any failures. This provides important confidence and data foundation for subsequent engineering scale-up. Significant economic potential: Although the model itself has costs, its "zero water consumption, zero electricity consumption" characteristics during operation mean that the savings in water, electricity, and labor costs during the 25-year lifespan of the photovoltaic power plant will be substantial, with a short expected payback period and significant long-term economic benefits. Final conclusion: The successful operation of this 1:4 scale model system is not only a technical demonstration but also powerful evidence of the future feasibility of "autonomous intelligent operation and maintenance of photovoltaic power plants." It proves that through ingenious multidisciplinary integrated design, a completely environment-driven, self-sufficient, intelligent, and efficient clean system can be created, providing a revolutionary and sustainable technological path to solve the pain points of global photovoltaic operation and maintenance. Future work will focus on engineering scale-up, enhanced environmental adaptability, and large-scale grid control.

[0035] Table 6 Self-sustaining balance results of energy harvesting and consumption Test Project Actual measurement results Analysis and Explanation Thermoelectric power generation performance Average nighttime temperature difference (ΔT): 11.5°C; Average output power per TEG cell: 6.2 mW; Total system power generation (16 TEG cells): ~99 mW The design performance was achieved and exceeded (>50mW). An effective temperature difference was formed between the center and the edge of the back of the photovoltaic panel, verifying the correctness of the non-uniform arrangement strategy. System operating power consumption Average standby power consumption: 1.8 mW; Average energy consumption per cleaning operation: 820 mWh The ultra-low power design is a success. Standby power consumption is extremely low, ensuring that energy collected at night is primarily used for energy storage rather than being consumed by the device itself. Energy balance of payments Average energy stored per night (10 hours): ~990 mWh; Energy consumed per cleaning cycle: ~820 mWh; Energy surplus: ~170 mWh / cycle Key findings: The system's energy harvesting exceeds its consumption, resulting in a positive surplus. This means the model system can operate continuously entirely based on ambient temperature differences without any external charging, empirically demonstrating the feasibility of energy self-sufficiency. The energy storage unit (supercapacitor + battery) can be recharged approximately 24-36 hours after cleaning. Table 7 Quantitative Results of Water Circulation and Cleaning Effect Test Project Actual measurement results Analysis and Explanation Radiative cooling and condensate The lowest nighttime temperature in the coated area was 6.3°C lower than the ambient temperature. The average condensation rate per night was 13.5 g / m². The radiation cooling coating is highly effective, successfully reducing the panel temperature below the dew point. The condensation amount is sufficient for a single cleaning cycle (approximately 10-15g). Hydrogel performance Water absorption and swelling ratio: 320%; Critical phase transition temperature (LCST): 31.5°C; Single trigger water release volume: ~12 mL The thermosensitive hydrogel has reliable performance, enabling intelligent "temporary storage" and "on-demand release" of moisture. During cleaning, the friction generated by the scraper is sufficient to trigger its phase change and release of water. Cleaning effectiveness (key performance indicator) Before cleaning, the light transmittance of the board surface was 89.2%. After cleaning, the light transmittance of the board surface was 96.7%. The light transmittance increased by 7.5 percentage points. Wet squeegee cleaning delivers superior results, far exceeding those of dry squeegee cleaning. Light transmittance is restored to near-new glass levels, proving that this cleaning method effectively removes stubborn dust that hinders power generation. Increased power generation Instantaneous power generation increased by 8.1% after cleaning (under the same sunlight conditions). This is the ultimate value of the system. A single cleaning operation directly brings a considerable increase in electricity generation, validating the core economic logic of "exchanging clean energy for power generation". Table 8. Operation Results of the Intelligent Control System Test Project Actual measurement results Analysis and Explanation Decision logic trigger During the two-week test, the system automatically triggered cleaning 7 times. All triggers met the following conditions simultaneously: energy storage > 85%, water storage > 10g, occurred in the morning, and wind speed met the standard. False triggers: 0. The "energy-water" dual-condition AND logic was executed perfectly. The system only acts when resources are plentiful, avoiding ineffective or damaging operations and demonstrating the effectiveness of intelligent decision-making. scraper force closed-loop control During the cleaning process, the blade pressure was maintained at 5.1 N ± 0.7 N. The pressure curve was stable without any violent fluctuations. The closed-loop control system composed of a six-dimensional force sensor and a floating mechanism has a good effect, achieving "gentle yet powerful" cleaning while protecting the photovoltaic panels. System Communication and Monitoring The local OLED status display is accurate, and the LoRa wireless data packet upload success rate is >99.5%. The remote monitoring platform successfully received all operational data and generated logs. Stable hardware and software communication provides a reliable foundation for unattended operation and maintenance and remote status monitoring.

Claims

1. A method and system for self-powered dust removal of photovoltaic panels based on radiative cooling and thermoelectric effect, characterized by: It constructs an intelligent ecosystem that relies entirely on low-grade environmental energy, achieving a self-sustaining closed-loop system of energy and matter. The core innovation of the system is not the stacking of components, but a self-sustaining negative feedback logic of "environmental energy capture - intelligent resource scheduling - precise on-demand execution." Specifically, the system innovatively utilizes the waste heat emitted by the photovoltaic panels at night, directly converting the temperature difference into electricity through a non-uniformly distributed thermoelectric generator array on the back, achieving in-situ, parasitic energy harvesting. Simultaneously, it uses a radiation cooling coating on the photovoltaic glass surface to lower the temperature below the dew point, actively condensing moisture from the air. This moisture is then collected, intelligently stored, and released on demand through a hydrophilic capillary network and temperature-sensitive hydrogel, forming a closed-loop atmospheric water cycle that requires no external supply. Based on this physical foundation, the system employs a combination of "onboard sensing and regional decision-making." The hybrid intelligent control architecture employs a strict "AND" logic in its control center, triggering a cleaning command only when both sufficient energy storage and available condensate water are simultaneously met. This eliminates ineffective operations and ensures that every action is based on the dual self-sufficiency of energy and materials. The cleaning action is completed by a set of macro-micro collaborative actuators that combine rigidity and flexibility. The rigid gantry structure ensures stability and precision across a wide range of array movements, while the intelligent scraper, integrating a six-dimensional force sensor and an adaptive floating mechanism at the end, achieves micro-flexible contact and constant force cleaning against the panel surface, ensuring zero damage to the photovoltaic panel while efficiently removing dust. Finally, after cleaning consumes resources, the system automatically returns to environmental monitoring and energy harvesting status, waiting for natural processes to replenish electricity and water to the threshold, thus forming a completely self-driven, self-judgment, and self-execution negative feedback control loop. This system transforms photovoltaic power plant operation and maintenance from a task that consumes external resources into a self-sustaining and self-optimizing "value creation unit." Its absolute energy and material closed loop, achieved through the deep integration of multidisciplinary principles, represents a fundamental breakthrough in the evolution of photovoltaic cleaning technology towards complete autonomy.

Images