Intelligent eave deicing system and method based on multi-mode cooperation

Abstract

The invention belongs to the technical field of building maintenance, and particularly discloses an intelligent eave deicing system and method based on multi-mode collaboration.The system comprises an eave and a support, the support is located on one side of the eave, a plurality of scrapers are arranged on the support, self-adaptive springs are connected between the scrapers, spraying assemblies are arranged on the eave, and a motor is arranged at the bottom of the support; an optical fiber sensor, a laser radar, a ToF camera and a plurality of infrared thermometers are arranged on the edge of the eave, a carbon nano tube heating wire is laid on the eave, a solar thin film cell panel is arranged on the top of the eave, and the optical fiber sensor, the laser radar, the ToF camera and the motor are all electrically connected with the PLC. According to the system, a heating wire pre-melting, mechanical scraping and spraying melting triple-combination deicing mechanism is adopted, compared with a traditional mode, the deicing efficiency is improved by more than 50%, the energy consumption is reduced by 30%, the modular design is compatible with various building scenes such as residences, factory buildings and bridge cornices, and the system is environmentally friendly and high in universality.

Description

Technical Field

[0001] This invention relates to the field of building maintenance technology, specifically to a system and method for intelligent roof de-icing based on multimodal collaboration. Background Technology

[0002] In winter, especially in cold regions, eaves face a severe challenge in removing ice after snowfall. During the day, sunlight melts some of the snow, which flows down the eaves. However, at night, as temperatures drop sharply, this melted snow quickly freezes at the edges of the eaves, forming sharp and heavy icicles. Icicles not only severely affect the appearance of buildings, ruining their overall aesthetics, but also pose a significant safety hazard. Icicles falling from heights are like "high-altitude bombs," potentially injuring pedestrians, damaging parked vehicles, and causing serious damage to surrounding facilities.

[0003] Currently, roof de-icing mainly relies on manual knocking or chemical de-icing agents. However, these methods have several drawbacks: workers at heights are prone to slipping and falling due to ice; falling ice fragments can injure people below; and the low temperatures slow worker reaction times, posing a significant safety risk. Sometimes, due to inclement weather and limited visibility at night, manual removal is time-consuming, requiring repeated climbing and repositioning, resulting in poor continuity. High-altitude de-icing requires hiring professionals, leading to high costs for equipment maintenance and material replacement, and potential indirect losses due to temporary road closures. Manual de-icing typically involves cracking the ice, but frequent knocking can cause cracks or leaks in eaves and pipes; and the use of salt-based chemical de-icing agents can corrode structural components. Therefore, safe and efficient removal of roof icing has become an urgent need to ensure building safety and maintain normal living order during winter. Summary of the Invention

[0004] The purpose of this invention is to provide a system and method for intelligent roof de-icing based on multimodal collaboration, in order to solve the problems of high safety risks, low work efficiency, high economic costs, and easy damage to building structures caused by manual high-altitude roof de-icing.

[0005] To achieve the above objectives, the basic solution provided by this invention is as follows: a smart eaves de-icing system based on multimodal collaboration, comprising an eaves and a support frame, the support frame being located on one side of the eaves, the support frame being equipped with several scrapers, the scrapers being connected by adaptive springs, the eaves being equipped with spraying components, the support frame having a motor at its bottom, the eaves edge being equipped with fiber optic sensors, lidar, ToF cameras and several infrared thermometers, carbon nanotube heating wires being laid on the eaves, and a solar thin-film battery panel being installed on the top of the eaves, the fiber optic sensors, lidar, ToF cameras and motors being electrically connected to a PLC controller.

[0006] The working principle of this invention is as follows: The intelligent eaves de-icing system achieves automated de-icing through a collaborative mechanism of multimodal sensor fusion detection, intelligent decision control, and composite de-icing execution. It utilizes multimodal sensors such as fiber optic sensors, lidar, ToF cameras, and infrared thermometers along the eaves edge to collect multidimensional data such as ice layer stress, thickness, distribution, and temperature in real time and transmits them to the PLC controller. After analysis and judgment by a preset algorithm, once the de-icing threshold is reached, carbon nanotube heating wires are activated according to priority to preheat and reduce the adhesion of the ice layer. The scraper assembly with adaptive springs is driven by a bracket motor to peel off the ice layer by adhering to the curved surface of the eaves. At the same time, the spray assembly is activated to spray warm water or de-icing agent to accelerate the melting of residual ice slag. Meanwhile, the solar thin-film battery panel provides power to achieve a green energy closed loop.

[0007] The beneficial effects of this invention are as follows: This system significantly improves the safety, efficiency, and economy of eaves de-icing through the integration of multiple technologies and intelligent control. Specific advantages include: utilizing multi-modal sensors such as lidar, ToF cameras, infrared thermometers, and fiber optics to achieve accurate perception of the ice layer's presence, thickness, distribution, and temperature across all dimensions, avoiding misjudgments by a single sensor; combining a scraper assembly with an adaptive spring to fit eaves of different curvatures, reducing mechanical damage while achieving a de-icing coverage rate of over 95%; employing a triple-layer de-icing mechanism of pre-melting with heating wires, mechanical scraping, and spray melting, which improves de-icing efficiency by over 50% and reduces energy consumption by 30% compared to traditional methods; and its solar power supply design is suitable for remote areas or high-altitude buildings without external power sources; fully automated control without human intervention eliminates the risks of working at heights; real-time monitoring and early warning functions can prevent injuries from falling icicles and potential overloading of the eaves structure; the spray assembly can switch between clean water and environmentally friendly de-icing agents to reduce chemical pollution; and its modular design is compatible with various building scenarios such as residences, factories, and bridge eaves, making it environmentally friendly and highly versatile.

[0008] Option 2, which is the preferred option of the basic option, has guide rails along the edge of the eaves, and the bracket is slidably connected to the guide rails.

[0009] Option 3, which is the preferred option of the basic option, has a magnetic adsorption component on one side of the eaves, and the bracket is magnetically connected to the magnetic adsorption component.

[0010] Option 4, a method for intelligent roof de-icing based on multimodal collaboration, includes the following steps: S1: Utilize fiber optic sensors, infrared thermometers, lidar, and ToF cameras to collect data on roof structure stress changes, ice temperature, ice thickness, and a 3D model of the ice layer, and transmit the collected data to the edge computing chip. S2: After receiving the data, the edge computing chip runs the CNN-LSTM hybrid algorithm to predict the risk of icing and generate the optimal de-icing strategy, and sets the ice thickness threshold according to the actual environmental conditions. S3: Determine whether the ice layer thickness has reached the threshold. If it has not reached the threshold, no action is taken. If it has reached the threshold, the carbon nanotube heating wire is automatically turned on to heat the ice layer on the eaves and the PTC temperature control technology is used to maintain the surface temperature of the eaves at 0-5℃. S4: Once the infrared thermometer detects that the surface temperature of the eaves has reached 0-5℃, the motor or electric motor will be automatically turned on to drive several scrapers to scrape the ice surface of the eaves.

[0011] Option 5, which is a preferred option of Option 4, involves the following steps in step S1 for transmitting data to the edge computing chip: 1) Use fiber optic sensors, infrared thermometers, lidar, and ToF cameras to collect historical and real-time data on roof structure stress changes, ice temperature, ice thickness, and three-dimensional ice models, and store the collected data in a database; 2) The collected data is denoised and normalized, the temperature data is converted into standardized Celsius values, and the ice thickness data is smoothed. 3) Use a CNN network to process each frame of the 3D model image of the ice layer captured by the ToF camera separately, extract spatial features, output feature vectors, arrange the feature vectors of all frames in chronological order to form a time sequence, and transmit the time sequence to the LSTM network; 4) The LSTM network receives the formed time sequence, analyzes the trend of feature vector changes over time, captures the inter-frame correlation, learns the dynamic law of ice thickness increasing or melting over time, and predicts the freezing trend in the future. 5) Prepare iterators for loading training and validation sets in batches, and perform multiple loop training on the entire CNN-LSTM hybrid algorithm model; 6) Deploy the trained CNN-LSTM algorithm model onto the edge computing chip.

[0012] Option 6, which is a preferred option of Option 4, involves spraying the roof with a 30% concentration of food-grade propylene glycol every 45 seconds to form an antifreeze film on the roof.

[0013] Option 7, which is the preferred option of Option 5, involves establishing a nonlinear relationship model between ice thickness, vibration frequency, and heating power during the training process. The formula is as follows: Where h represents the ice thickness, T represents the ambient temperature, and k1 and k2 are empirical coefficients. Attached Figure Description

[0014] Figure 1 This is a perspective view of Embodiment 1 of the intelligent roof de-icing system and method based on multimodal collaboration of the present invention; Figure 2 yes Figure 1Enlarged view of point A in the middle; Figure 3 This is a perspective view of Embodiment 2 of the intelligent roof de-icing system and method based on multimodal collaboration of the present invention. Detailed Implementation

[0015] The present invention will be further described in detail below through specific embodiments: The reference numerals in the accompanying drawings include: 1. Eaves, 2. Bracket, 3. Scraper, 4. Adaptive spring, 5. Spray assembly, 6. Motor, 7. Fiber optic sensor, 8. LiDAR, 9. ToF camera, 10. Infrared thermometer, 11. Carbon nanotube heating wire, 12. Guide rail, 13. Magnetic adsorption assembly.

[0016] Example 1 like Figure 1 and Figure 2 As shown: A smart eaves de-icing system based on multimodal collaboration includes an eaves 1 and a support 2. The support 2 is located on one side of the eaves 1. A guide rail 12 is provided along the edge of the eaves 1. The support 2 is slidably connected to the guide rail 12. Several scrapers 3 are provided on the support 2. Adaptive springs 4 are connected between the scrapers 3. Spraying components 5 are provided on each eaves 1. A motor 6 is provided on one side of the eaves 1. Fiber optic sensors 7, lidar 8, ToF cameras 9 and several infrared thermometers 10 are provided along the edge of the eaves 1. The ToF cameras 9 have an accuracy of ±2mm. Carbon nanotube heating wires 11 are laid on the eaves 1. The fiber optic sensors 7, lidar 8, ToF cameras 9 and motor 6 are all electrically connected to a PLC controller.

[0017] The implementation method of this embodiment is as follows: This method is suitable for tiled roofs or other sloping eaves. A guide rail 12 is laid horizontally along the bottom edge of the eaves 1 and fixed to the tiles by a custom buckle. A slider is provided at the bottom of the bracket 2 and slidably connected to the guide rail 12. Adjacent scrapers 3 are hinged by an adaptive spring 4. Distributed fiber optic sensors 7 are embedded along the edge of the eaves 1. The top of the vertical arm of the bracket 2 integrates a lidar 8, a ToF camera 9, and an infrared thermometer 10. Carbon nanotube heating wires 11 are laid in a serpentine pattern on the surface of the tiles, with a spacing of 5-8cm, and covered with a transparent waterproof membrane. The PLC controller communicates with each sensor and actuator through an RS485 bus, has a built-in ice layer recognition algorithm, prioritizes solar power, and switches to AC power when the battery pack is low. When the system detects that the ice layer thickness is >5mm and the temperature is <0℃ for 2 hours or the fiber stress is >50με, it sequentially starts the heating wire preheating, the motor 6 drives the scraper 3 for mechanical de-icing, and the spray assembly 5 assists in melting. The sensor monitors in real time during the de-icing process, and switches to heat preservation mode when the thickness is <2mm.

[0018] Example 2 like Figure 3As shown: A smart roof de-icing system based on multimodal collaboration, which differs from Embodiment 1 in that: a magnetic adsorption component 13 is provided on one side of the roof 1, and the bracket 2 is magnetically connected to the magnetic adsorption component 13.

[0019] The implementation method of this embodiment is as follows: The difference from Embodiment 1 is that this method is applicable to metal structure eaves. Angle steel is used to fix the bracket 2 to the magnetic adsorption component 13. The magnetic adsorption component 13 is magnetically attracted to the metal eaves 1 to fix the position of the bracket 2 and the scraper 3.

[0020] Example 3 A method for intelligent roof de-icing based on multimodal collaboration includes the following steps: S1: Using fiber optic sensor 7, infrared thermometer 10, lidar 8, and ToF camera 9, data on roof structure stress changes, ice temperature, ice thickness, and a 3D model of the ice layer are collected respectively. The collected data is then transmitted to the edge computing chip. The specific steps for transmitting data to the edge computing chip are as follows: 1) Use fiber optic sensor 7, infrared thermometer 10, lidar 8, and ToF camera 9 to collect historical and real-time data on the structural stress changes of the eaves 1, ice temperature, ice thickness, and three-dimensional model of the ice layer, and store the collected data in the database. 2) The collected data is denoised and normalized, the temperature data is converted into standardized Celsius values, and the ice thickness data is smoothed. 3) Use a CNN network to process each frame of the 3D model image of the ice layer acquired by the ToF camera 9 separately, extract spatial features, output feature vectors, arrange the feature vectors of all frames in chronological order to form a time sequence, and transmit the time sequence to the LSTM network; 4) The LSTM network receives the formed time sequence, analyzes the trend of feature vector changes over time, captures the inter-frame correlation, learns the dynamic law of ice thickness increasing or melting over time, and predicts the freezing trend in the future. 5) Prepare iterators for the training and validation sets, loading them in batches, and perform multiple iterations of training on the entire CNN-LSTM hybrid algorithm model. During training, establish a nonlinear relationship model between ice thickness, vibration frequency, and heating power, using the following formula: Where h represents the ice thickness, T represents the ambient temperature, and k1 and k2 are empirical coefficients. In actual testing, h > 5 mm and T < 0 °C. 6) Deploy the trained CNN-LSTM algorithm model onto the edge computing chip; S2: After receiving the data, the edge computing chip runs the CNN-LSTM hybrid algorithm to predict the risk of icing and generate the optimal de-icing strategy, and sets the ice thickness threshold according to the actual environmental conditions. S3: Determine whether the ice layer thickness has reached the threshold. If it has not reached the threshold, no action is taken. If it has reached the threshold, the carbon nanotube heating wire 11 is automatically turned on to heat the ice layer on the eaves 1 and the PTC temperature control technology is used to maintain the surface temperature of the eaves 1 at 0-5℃. S4: After the infrared thermometer 10 detects that the surface temperature of the eaves 1 reaches 0-5℃, the motor 6 is automatically turned on, driving several scrapers 3 to scrape the ice surface of the eaves 1. During the process, the spraying component 5 sprays 30% food-grade propylene glycol on the roof every 45 seconds to form an antifreeze film on the roof.

[0021] The above descriptions are merely embodiments of the present invention, and common knowledge regarding specific structures and characteristics is not elaborated upon here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the structure of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A smart roof de-icing system based on multimodal collaboration, characterized in that, Includes an eaves (1) and a support (2). The support (2) is located on one side of the eaves (1). The support (2) is provided with several scrapers (3). Adaptive springs (4) are connected between the scrapers (3). Spraying components (5) are provided on each of the eaves (1). A motor (6) is provided on one side of the eaves (1). Fiber optic sensors (7), lidar (8), ToF cameras (9) and several infrared thermometers (10) are provided along the edge of the eaves (1). Carbon nanotube heating wires (11) are laid on the eaves (1). The fiber optic sensors (7), lidar (8), ToF cameras (9), and motors (6) are all electrically connected to a PLC controller.

2. The intelligent roof de-icing system based on multimodal collaboration according to claim 1, characterized in that, The edge of the eaves (1) is provided with a guide rail (12), and the bracket (2) is slidably connected to the guide rail (12).

3. The intelligent roof de-icing system based on multimodal collaboration according to claim 1, characterized in that, A magnetic adsorption component (13) is provided on one side of the eaves (1), and the bracket (2) is magnetically connected to the magnetic adsorption component (13).

4. A method for intelligent roof de-icing based on multimodal collaboration, characterized in that, Includes the following steps: S1: Use fiber optic sensor (7), infrared thermometer (10), lidar (8), and ToF camera (9) to collect data on roof structure stress changes, ice temperature, ice thickness, and three-dimensional ice model, and transmit the collected data to the edge computing chip. S2: After receiving the data, the edge computing chip runs the CNN-LSTM hybrid algorithm to predict the risk of icing and generate the optimal de-icing strategy, and sets the ice thickness threshold according to the actual environmental conditions. S3: Determine whether the ice layer thickness has reached the threshold. If it has not reached the threshold, no action will be taken. If it has reached the threshold, the carbon nanotube heating wire (11) will be automatically turned on to heat the ice layer on the eaves (1) and the PTC temperature control technology will be used to maintain the surface temperature of the eaves (1) at 0-5℃. S4: After the infrared thermometer (10) detects that the surface temperature of the eaves (1) reaches 0-5℃, the motor (6) is automatically turned on, driving several scrapers (3) to scrape the ice surface of the eaves (1).

5. The method for intelligent roof de-icing based on multimodal collaboration according to claim 4, characterized in that, In step S1, the specific steps for transmitting data to the edge computing chip are as follows: 1) Use fiber optic sensor (7), infrared thermometer (10), lidar (8), and ToF camera (9) to collect historical and real-time data on roof structure stress changes, ice temperature, ice thickness, and three-dimensional ice model, and store the collected data in the database. 2) The collected data is denoised and normalized, the temperature data is converted into standardized Celsius values, and the ice thickness data is smoothed. 3) Use a CNN network to process each frame of the three-dimensional model image of the ice layer acquired by the ToF camera (9) separately, extract spatial features, output feature vectors, arrange the feature vectors of all frames in time order to form a time sequence, and transmit the time sequence to the LSTM network; 4) The LSTM network receives the formed time sequence, analyzes the trend of feature vector changes over time, captures the inter-frame correlation, learns the dynamic law of ice thickness increasing or melting over time, and predicts the freezing trend in the future. 5) Prepare iterators for loading training and validation sets in batches, and perform multiple loop training on the entire CNN-LSTM hybrid algorithm model; 6) Deploy the trained CNN-LSTM algorithm model onto the edge computing chip.

6. The method for intelligent roof de-icing based on multimodal collaboration according to claim 4, characterized in that, The spray assembly (5) sprays food-grade propylene glycol at a concentration of 30% onto the roof every 45 seconds to form an antifreeze film on the roof.

7. The method for intelligent roof de-icing based on multimodal collaboration according to claim 5, characterized in that, In step 5), during the training process, a nonlinear relationship model is established between ice thickness, vibration frequency, and heating power. The formula is: Where h represents the ice thickness, T represents the ambient temperature, and k1 and k2 are empirical coefficients.

Images