Photovoltaic curtain wall intelligent safety monitoring system and dynamic decision-making method

Abstract

The invention discloses a photovoltaic curtain wall intelligent safety monitoring system and a dynamic decision-making method. The photovoltaic curtain wall intelligent safety monitoring system comprises a hardware system, a software system and a decision-making output terminal. The hardware system comprises a sensing layer, a transmission layer and a decision layer; the software system comprises data acquisition, feature extraction and risk assessment; through deep fusion of multi-modal sensing and digital twinning, the running state of the photovoltaic curtain wall can be comprehensively sensed from multiple dimensions, analysis is performed by using an advanced algorithm and model, and the fault detection accuracy is greatly improved; according to the dynamic decision-making method, the photovoltaic curtain wall can still keep certain power generation efficiency by reasonably adjusting the operation strategy under the fault working condition, and the power generation loss caused by the fault is reduced.

Description

Technical Field

[0001] The present invention relates to the intersection of building integrated photovoltaic (BIPV) and intelligent operation and maintenance, and in particular to an intelligent safety monitoring system for photovoltaic curtain walls and a dynamic decision-making method. Background Art

[0002] As the core component of building integrated photovoltaics (BIPV), photovoltaic curtain walls are exposed to complex environments for a long time and are susceptible to risks such as hot spot effects, structural damage, and electrical failures, which threaten building safety and power generation efficiency.

[0003] In its early stages, this technology relied on manual inspections, visual inspections, and handheld thermal imagers, which were unable to identify microcracks and texture damage within hidden structures. Existing monitoring technologies suffer from the following drawbacks: a. Detection is limited in scope, often relying on a single sensor, such as electrical detection, which is insensitive to early hot spots and has a high rate of missed detections. b. It has poor environmental adaptability, with imaging distorted by strong light interference and unable to perform effective detection on rainy days. c. Decision-making lags are significant, with an average time from data collection to maintenance response exceeding 48 hours, lacking predictive maintenance capabilities. d. Deployment and maintenance costs are high, with sensor wiring damaging the curtain wall structure and manual inspections accounting for 58% of total O&M costs. Furthermore, it lacks collaborative perception of multiple dimensions, including mechanical, electrical, and environmental conditions, and fault diagnosis relies on manual experience, making real-time dynamic decision-making difficult. While digital twin technology has found some application in the industrial sector, solutions integrating it with multimodal sensing to address photovoltaic curtain wall safety issues have yet to be reported. Summary of the Invention

[0004] The technical problem to be solved by the present invention is: in order to overcome the deficiencies in the prior art, an intelligent safety monitoring system for photovoltaic curtain walls and a dynamic decision-making method are provided.

[0005] The technical solution adopted by the present invention is: a photovoltaic curtain wall intelligent safety monitoring system, including a hardware system, a software system and a decision output terminal;

[0006] The hardware system includes a perception layer, a transmission layer, and a decision-making layer; the software system includes data acquisition, feature extraction, and risk assessment;

[0007] The sensing layer is composed of a tri-spectral imager, a graphene temperature sensor array, a thin-film microcurrent sensor, and a laser displacement meter. It uses a conformal sensing array to integrate the thin-film microcurrent sensor with the photovoltaic glass to achieve accurate perception of the electrical parameters of the photovoltaic curtain wall.

[0008] Among them, the graphene temperature sensor array has high sensitivity and fast response characteristics, realizing high-density temperature monitoring, and the high sampling frequency can capture temperature changes in a timely manner;

[0009] Thin-film microcurrent sensors can accurately measure microcurrent changes in photovoltaic curtain walls, while laser displacement meters are used to monitor minute displacements of the curtain wall structure, providing comprehensive awareness of the operating status of the photovoltaic curtain wall.

[0010] The transmission layer consists of a TDMA-LoRa gateway and anti-interference relay nodes. Through dynamic power control and time slot allocation algorithm optimization, the TDMA-LoRa gateway can dynamically adjust the transmission power and time slot allocation according to the network load and signal strength.

[0011] The decision-making layer consists of edge computing units, digital twin platforms and autonomous maintenance robots.

[0012] At the architectural level, edge-cloud collaborative architecture technology is introduced. The Jetson Orin module at the edge achieves 200ms-level real-time processing, enabling rapid preliminary analysis and screening of collected multimodal data, enabling timely detection of abnormal signals. In the future, the module will further integrate a photonic computing unit, significantly improving the energy efficiency of multimodal data processing.

[0013] In the cloud, a digital twin engine performs multi-physics simulation, simulating the photovoltaic curtain wall's multi-physics processes, including light, electricity, heat, and force, to deeply analyze system operating status and predict potential failures. By synergizing real-time edge response with deep cloud simulation, a closed-loop system from data collection to decision optimization is established, providing low-latency, high-precision intelligent solutions for the intersection of building integrated photovoltaics (BIPV) and intelligent operation and maintenance.

[0014] In terms of algorithms, a physically constrained deep learning model is employed. By embedding a PINN network that incorporates photovoltaic thermodynamic equations and combining it with a coupled solver for the four fields of light, electricity, heat, and force, the model fully leverages physical knowledge, improving its ability to simulate and predict complex physical processes. Furthermore, an autonomous decision-making engine is employed to generate maintenance strategies based on deep reinforcement learning (PPO algorithm). This algorithm dynamically adjusts maintenance plans based on real-time monitoring data and system operating status, enabling intelligent decision-making. Furthermore, a three-level early warning linkage mechanism is implemented that, upon detecting a high-risk fault, disconnects the circuit within milliseconds, ensuring system safety.

[0015] Furthermore, the tri-spectral imager described in the present invention has three channels: visible, infrared, and ultraviolet, which facilitates the acquisition of surface information of photovoltaic curtain walls from multiple spectral dimensions, effectively improving the detection capability of defects such as hot spots and cracks;

[0016] By adopting the above technical solution, the tri-spectral imager has the function of polarization filtering and anti-strong light interference, and can obtain clear images under complex lighting conditions.

[0017] Furthermore, the edge computing unit described in the present invention has the ability to process four channels of 4K video in parallel, and can quickly process large amounts of image data;

[0018] Furthermore, the digital twin platform of the present invention constructs a virtual model corresponding to the physical entity through real-time simulation, thereby achieving real-time monitoring and prediction of the operating status of the photovoltaic curtain wall;

[0019] Furthermore, the autonomous maintenance robot described in the present invention is equipped with a vacuum adsorption walking mechanism, which can move freely on the photovoltaic curtain wall to perform inspection and maintenance tasks.

[0020] Furthermore, the laser displacement meter of the present invention establishes a reference point through laser positioning to ensure the accuracy of the installation position, making it suitable for installation work in a curved curtain wall environment.

[0021] Furthermore, the conformal sensing array described in the present invention utilizes low-temperature plasma surface treatment technology to improve the surface properties of the material and enhance the adhesion between the materials. Transparent conductive adhesive is laminated and cured to achieve conformal packaging of the sensor and photovoltaic glass, ensuring the stability of the electrical connection without affecting the light transmittance and power generation performance of the photovoltaic glass.

[0022] Self-aligning universal joint bracket adjustment is used to achieve adaptive installation of sensors and equipment on curved photovoltaic curtain walls, ensuring close fit and stable operation of the equipment and curtain wall;

[0023] By adopting the above technical solution and using conformal sensing array, thin-film microcurrent sensors are integrated with photovoltaic glass to achieve accurate perception of the electrical parameters of the photovoltaic curtain wall without affecting the overall structure and aesthetics of the curtain wall.

[0024] A dynamic decision-making method based on a photovoltaic curtain wall intelligent safety monitoring system includes the following steps:

[0025] Step 1: Real-time data is collected through the multimodal perception layer. The edge computing node performs threshold detection and fast Fourier transform, filters abnormal signals, and uploads them to the cloud.

[0026] Step 2: The digital twin cloud platform uses digital thread technology to update the virtual model status based on the uploaded abnormal signal. It then uses the fault diagnosis model to analyze the abnormal situation and output the risk probability and impact range.

[0027] Step 3: The dynamic decision engine generates a multi-objective optimization plan based on multiple factors, including risk level, grid demand (such as peak load regulation instructions), and building energy consumption patterns.

[0028] When the risk level is low, the MPPT algorithm is dynamically adjusted to avoid the impact of local shadows and improve power generation efficiency. When the risk is medium, drones are activated for fixed-point inspections, and the load balance of adjacent units is adjusted to promptly identify and resolve potential problems. When the risk is high, circuit breakers are triggered to isolate the faulty unit, and structural reinforcement plans are pushed to the operation and maintenance terminal to ensure system safety.

[0029] Step 4: Use blockchain to store key decision logs to support post-event traceability and model iterative optimization.

[0030] Compared with the prior art, the present invention has the following advantages:

[0031] (1) Through the deep integration of multimodal perception and digital twins, the present invention can fully perceive the operating status of photovoltaic curtain walls from multiple dimensions, and use advanced algorithms and models for analysis, greatly improving the accuracy of fault detection. Compared with traditional methods, the warning time is significantly advanced, and faults can be discovered and warned in time at the early stage, buying valuable time for maintenance measures and effectively reducing failure losses;

[0032] (2) The dynamic decision-making method enables the photovoltaic curtain wall to maintain a certain power generation efficiency by reasonably adjusting the operation strategy under fault conditions, thereby reducing the power generation loss caused by the fault. At the same time, the precise fault detection and predictive maintenance capabilities reduce the frequency and blindness of manual inspections and reduce unnecessary maintenance costs. According to calculations, the average annual operation and maintenance cost is significantly reduced compared to traditional methods, which improves the economy and sustainability of the photovoltaic curtain wall system.

[0033] (3) This invention supports seamless integration with smart grids and building energy management systems (BEMS) to achieve efficient energy allocation and management. Through real-time monitoring and optimized operation, it provides a stable power supply for smart grids and works in conjunction with building energy management systems to reduce overall building energy consumption. This not only promotes the development of zero-carbon buildings, but also provides strong support for smart city construction and promotes technological progress and industrial upgrading in the fields of building photovoltaic integration and intelligent operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a structural principle block diagram of the present invention;

[0035] Figure 2 It is a flow chart of the dynamic decision-making method in the present invention. DETAILED DESCRIPTION

[0036] The embodiments of the present invention are described in detail below. The embodiments are implemented based on the technical solutions of the present invention, and detailed implementation methods and specific operating processes are given. However, the protection scope of the present invention is not limited to the following embodiments.

[0037] like Figure 1 The photovoltaic curtain wall intelligent safety monitoring system shown includes a hardware system, a software system and a decision output terminal;

[0038] The hardware system includes the perception layer, transmission layer and decision-making layer; the software system includes data acquisition, feature extraction and risk assessment;

[0039] The sensing layer consists of a tri-spectral imager, a graphene temperature sensor array, a thin-film microcurrent sensor, and a laser displacement meter. Using a conformal sensing array, the thin-film microcurrent sensor is integrated with the photovoltaic glass to achieve precise perception of the electrical parameters of the photovoltaic curtain wall.

[0040] Among them, the graphene temperature sensor array has high sensitivity and fast response characteristics, realizing high-density temperature monitoring, and the high sampling frequency can capture temperature changes in a timely manner;

[0041] Thin-film microcurrent sensors can accurately measure microcurrent changes in photovoltaic curtain walls, while laser displacement meters are used to monitor minute displacements of the curtain wall structure, providing comprehensive awareness of the operating status of the photovoltaic curtain wall.

[0042] The transmission layer consists of TDMA-LoRa gateways and anti-interference relay nodes. Through dynamic power control and time slot allocation algorithm optimization, the TDMA-LoRa gateway can dynamically adjust the transmission power and time slot allocation according to the network load and signal strength.

[0043] The decision-making layer consists of edge computing units, digital twin platforms and autonomous maintenance robots.

[0044] Among them, the tri-spectral imager has three channels: visible, infrared, and ultraviolet, which facilitates the acquisition of surface information of photovoltaic curtain walls from multiple spectral dimensions, effectively improving the detection capabilities of defects such as hot spots and cracks;

[0045] By adopting the above technical solution, the tri-spectral imager has the function of polarization filtering and anti-strong light interference, and can obtain clear images under complex lighting conditions.

[0046] The edge computing unit is capable of processing four channels of 4K video in parallel, enabling rapid processing of large amounts of image data. The digital twin platform uses real-time simulation to build a virtual model corresponding to the physical entity, enabling real-time monitoring and prediction of the operating status of the photovoltaic curtain wall.

[0047] The autonomous maintenance robot is equipped with a vacuum adsorption walking mechanism and can move freely on the photovoltaic curtain wall to perform inspection and maintenance tasks.

[0048] The laser displacement meter establishes a reference point through laser positioning to ensure the accuracy of the installation position, making it suitable for installation work in curved curtain wall environments.

[0049] Conformal sensing arrays use low-temperature plasma surface treatment technology to improve material surface properties and enhance adhesion between materials. Transparent conductive adhesive is laminated and cured to achieve conformal packaging of the sensor and photovoltaic glass, ensuring electrical connection stability without affecting the light transmittance and power generation performance of the photovoltaic glass.

[0050] By using self-aligning universal joint bracket adjustment, sensors and equipment can be adaptively installed on the curved photovoltaic curtain wall, ensuring a close fit and stable operation of the equipment and the curtain wall; by adopting conformal sensing display, thin-film microcurrent sensors are integrated with photovoltaic glass to achieve accurate perception of the electrical parameters of the photovoltaic curtain wall without affecting the overall structure and appearance of the curtain wall.

[0051] A dynamic decision-making method based on a photovoltaic curtain wall intelligent safety monitoring system includes the following steps: Figure 2 As shown:

[0052] Step 1: Real-time data is collected through the multimodal perception layer. The edge computing node performs threshold detection and fast Fourier transform, filters abnormal signals, and uploads them to the cloud.

[0053] Step 2: The digital twin cloud platform uses digital thread technology to update the virtual model status based on the uploaded abnormal signal. It then uses the fault diagnosis model to analyze the abnormal situation and output the risk probability and impact range.

[0054] Step 3: The dynamic decision engine generates a multi-objective optimization plan based on multiple factors, including risk level, grid demand (such as peak load regulation instructions), and building energy consumption patterns.

[0055] When the risk level is low, the MPPT algorithm is dynamically adjusted to avoid the impact of local shadows and improve power generation efficiency. When the risk is medium, drones are activated for fixed-point inspections, and the load balance of adjacent units is adjusted to promptly identify and resolve potential problems. When the risk is high, circuit breakers are triggered to isolate the faulty unit, and structural reinforcement plans are pushed to the operation and maintenance terminal to ensure system safety.

[0056] Step 4: Use blockchain to store key decision logs to support post-event traceability and model iterative optimization.

Claims

1. A photovoltaic curtain wall intelligent safety monitoring system, characterized by: Including hardware system, software system and decision output terminal; The hardware system includes a perception layer, a transmission layer, and a decision-making layer; the software system includes data acquisition, feature extraction, and risk assessment; The sensing layer is composed of a tri-spectral imager, a graphene temperature sensor array, a thin-film microcurrent sensor, and a laser displacement meter. It uses a conformal sensing array to integrate the thin-film microcurrent sensor with the photovoltaic glass to achieve accurate perception of the electrical parameters of the photovoltaic curtain wall. The transmission layer consists of a TDMA-LoRa gateway and anti-interference relay nodes. Through dynamic power control and time slot allocation algorithm optimization, the TDMA-LoRa gateway can dynamically adjust the transmission power and time slot allocation according to the network load and signal strength. The decision-making layer consists of edge computing units, digital twin platforms and autonomous maintenance robots.

2. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The tri-spectral imager has three channels: visible, infrared, and ultraviolet, which facilitates obtaining surface information of the photovoltaic curtain wall from multiple spectral dimensions.

3. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The edge computing unit is capable of processing four channels of 4K video in parallel.

4. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The digital twin platform constructs a virtual model corresponding to the physical entity through real-time simulation, realizing real-time monitoring and prediction of the operating status of the photovoltaic curtain wall.

5. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The autonomous maintenance robot is equipped with a vacuum adsorption walking mechanism and can move freely on the photovoltaic curtain wall to perform inspection and maintenance tasks.

6. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The laser displacement meter establishes a reference point through laser positioning to ensure the accuracy of the installation position, making it suitable for installation work in a curved curtain wall environment.

7. The photovoltaic curtain wall intelligent safety monitoring system according to claim 1, characterized in that: The conformal sensing array adopts low-temperature plasma surface treatment technology to improve the surface properties of the material and enhance the adhesion between the materials. The conformal packaging of the sensor and the photovoltaic glass is achieved through lamination and curing of transparent conductive adhesive.

8. A dynamic decision-making method based on the photovoltaic curtain wall intelligent safety monitoring system according to any one of claims 1 to 7, comprising the following steps: Step 1: Real-time data is collected through the multimodal perception layer. The edge computing node performs threshold detection and fast Fourier transform, filters abnormal signals, and uploads them to the cloud. Step 2: The digital twin cloud platform uses digital thread technology to update the virtual model status based on the uploaded abnormal signal. It then uses the fault diagnosis model to analyze the abnormal situation and output the risk probability and impact range. Step 3: The dynamic decision engine combines risk level, grid demand, and building energy consumption patterns to generate a multi-objective optimization solution. When the risk level is low, the MPPT algorithm is dynamically adjusted to avoid the impact of local shadows and improve power generation efficiency. When the risk is medium, drones are activated for fixed-point inspections, and the load balance of adjacent units is adjusted to promptly identify and resolve potential problems. When the risk is high, circuit breakers are triggered to isolate the faulty unit, and structural reinforcement plans are pushed to the operation and maintenance terminal to ensure system safety. Step 4: Use blockchain to store key decision logs to support post-event traceability and model iterative optimization.

Images