A monitoring system and method for a solar thermal system reflector

Abstract

The present invention discloses a monitoring system and method for a reflector of a solar thermal system, and relates to the technical field of optical element monitoring. The present invention monitors the health status and cleanliness of the reflector in real time based on sensors and cameras, and transmits data to a cloud computing platform using wireless networks to avoid frequent manual inspections. Detailed data of the reflector is obtained in real time through high-precision sensors, optical sensors and high-resolution cameras, and accurate monitoring of the health status and cleanliness of the reflector is achieved by combining big data analysis and machine learning algorithms. An automatic alignment system based on computer vision captures the image of the reflector in real time and analyzes its alignment status, automatically adjusts the angle of the reflector, ensures that the reflector is always in the best alignment state, improves the operating stability of the system, and warns and handles potential problems in advance through predictive maintenance and intelligent monitoring, reduces system failures and downtime, and reduces maintenance costs.

Description

Technical Field

[0001] The present invention relates to the technical field of optical element monitoring, and in particular to a monitoring system and method for a reflector of a solar thermal system. Background Art

[0002] Solar thermal power generation refers to the use of large-scale arrays of parabolic or dish-shaped mirrors to collect solar thermal energy, provide steam through heat exchange devices, and combine with the technology of traditional steam turbine generators to achieve the purpose of power generation; the use of solar thermal power generation technology avoids the expensive silicon crystal photoelectric conversion process and can greatly reduce the cost of solar power generation; moreover, this form of solar energy utilization has an advantage that other forms of solar energy conversion cannot match, that is, water heated by solar energy can be stored in huge containers and can still drive steam turbines to generate electricity several hours after the sun sets.

[0003] For example, the patent publication number CN105785552A discloses a detection device and detection method based on a reflector of a butterfly-type solar thermal system. A beam meter that can be easily moved to the top of any reflector coordinate on the concentrator is set on the reflective surface of the concentrator. The beam meter emits a vertically downward visible light beam. When the vertically downward visible light beam irradiates the reflective mirror surface, the angle of the reflector is adjusted so that the light spot of the visible light beam falls on the focal position set by the light receiver after reflection. The position of the reflector at this time is the calibration position, and the reflector is tightened to achieve the purpose of installing and adjusting the position of the reflector. The detection device of the invention has a simple structure, low cost, and easy installation. The detection method is simple and the detection efficiency is high. It solves the problems existing in the installation, adjustment and repeated detection of each reflector with an optical curved surface in the dish-type solar thermal system.

[0004] In the process of monitoring the reflectors of solar thermal systems, the main difficulties are the alignment accuracy and cleanliness of the reflectors. Ensuring that the reflectors always maintain the best alignment and that there is no dust, dirt or other contaminants on the surface is the key to improving system efficiency. The current traditional reflector alignment method relies on regular manual inspection and adjustment, which is inefficient and prone to errors. At the same time, regular maintenance is generally used for maintenance. It is impossible to monitor in real time whether the surface of the reflector is affected by dust and dirt, and it is impossible to detect and deal with aging, wear and dirt problems in time. Therefore, a monitoring system and method for solar thermal system reflectors is urgently needed to solve such problems. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a monitoring system and method for the reflectors of a solar thermal system, which solves the problems in the prior art that the system relies on regular manual inspection and adjustment, is inefficient and prone to errors, and generally adopts regular maintenance, and cannot monitor in real time whether the surface of the reflector is affected by dust and dirt, and cannot promptly detect and deal with aging, wear, and dirt problems.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] The present invention provides a method for monitoring a reflector of a solar thermal system, comprising:

[0008] Step 1. Perform predictive maintenance on the health of the reflectors. High-precision sensors are evenly installed on each reflector and its supporting structure. The sensor data is transmitted to the data analysis and prediction module in real time via a wireless network. Big data analysis and machine learning technologies are used to analyze the sensor data in real time to predict the health of the reflectors. Based on the prediction results, early warning signals are issued in advance to arrange preventive maintenance.

[0009] In step 1, strain gauges, temperature sensors, and humidity sensors are used to monitor the health of the reflector and supporting structure in real time;

[0010] Utilize LPWAN technology to transmit sensor data; analyze sensor data in real time on a cloud computing platform and use machine learning algorithms to predict the health of the reflector;

[0011] Step 2. Intelligently monitor the cleanliness of the reflector surface. Install an optical sensor on the reflector surface to detect changes in the scattering and absorption characteristics of light on the reflector surface. The optical sensor collects data in real time and performs preliminary processing through edge computing devices to determine the cleanliness. When it is detected that the contamination on the reflector surface reaches a threshold, the cleaning robot and mechanical arm are automatically started for cleaning.

[0012] Step 3. Automatic alignment based on computer vision, arrange cameras around the reflector to capture the image of the reflector; the camera collects the image of the reflector in real time, and analyzes it through computer vision technology to determine the alignment status of the reflector; according to the analysis results, the automatic alignment device adjusts the angle of the reflector in real time to keep it in the best alignment state at all times.

[0013] The present invention is further configured as follows: the real-time analysis method of the sensor data in step 1 is:

[0014] Let the sensor on each reflector be Si, i represents different sensor types, including temperature sensor, strain gauge and humidity sensor, and the data collected by each sensor at time t is recorded as Di(t);

[0015] Assume the temperature sensor data is T(t), the strain gauge data is ε(t), and the humidity sensor data is H(t);

[0016] Using LPWAN technology, the collected data Di(t) is wirelessly transmitted to the cloud computing platform, i.e., the real-time analysis unit;

[0017] Data preprocessing is performed on the cloud computing platform, including data cleaning, denoising and standardization:

[0018] For each sensor data Di(t), normalization is performed: Among them, μi and σi are the mean and standard deviation of sensor Si data respectively;

[0019] Extract features from preprocessed data and use sliding window technique to extract time series features:

[0020] Set the window size to w and the step size to s: in, Identify the number of time steps within the window;

[0021] The present invention is further configured as follows: the sensor data real-time analysis method further includes:

[0022] Use the long short-term memory network LSTM model to process time series data and perform model training:

[0023] The LSTM network architecture includes:

[0024] Input layer: input feature Fi(t);

[0025] Hidden layer: multi-layer LSTM unit;

[0026] Output layer: health status prediction results;

[0027] Let ht be the hidden state of the LSTM unit, Ct be the cell state, σ be the sigmoid activation function, and tanh be the tanh activation function;

[0028] Forget Gate:

[0029] Input Gate:

[0030] Candidate cell states:

[0031] Cell status update:

[0032] Output Gate:

[0033] Hidden state update: ht = ot·tanh(Ct);

[0034] Among them, Wf,Wi,WC,Wo and bf,bi,bC,bo are weight and bias matrices;

[0035] Use the trained LSTM model to predict health status based on real-time data;

[0036] The present invention is further configured as follows: the health status prediction method is:

[0037] set up Predicted health status at time t: Among them, F(t) is the data feature of the current time step;

[0038] According to the prediction results Conduct real-time monitoring and early warning;

[0039] Set the health threshold θ, when When the threshold value θ is exceeded, an early warning is triggered;

[0040] The present invention is further configured as follows: in step 2, a high-precision optical sensor is installed on the surface of the reflector to monitor the cleanliness in real time;

[0041] Use edge computing technology to perform preliminary processing and analysis of sensor data;

[0042] Combined with cleanliness monitoring data, cleaning robots and robotic arms are automatically started;

[0043] The present invention is further configured as follows: the initial processing and analysis of sensor data using edge computing technology in step 2 is as follows:

[0044] A high-precision optical sensor is installed on the surface of each reflector to monitor the cleanliness of the surface. The sensor measures the scattering and absorption characteristics of light to reflect the degree of surface contamination. The data collected by the optical sensor is set to L(t), which represents the optical reflection intensity at time t.

[0045] The sensor data L(t) is collected in real time by edge computing devices, and the initial processing includes noise filtering and data standardization:

[0046] Use the moving average method to smooth the data and set the window size to w: Standardization: Where i represents the time step, L(ti) represents the optical reflection intensity data collected at time ti, and μ L and σ L are the mean and standard deviation of historical data respectively;

[0047] Use edge computing devices to perform real-time analysis to determine whether the cleanliness of the reflector has reached the cleanliness threshold;

[0048] A cleanliness threshold θL is set. When the normalized optical reflection intensity L'(t) is lower than the threshold, it is considered that the reflector needs to be cleaned.

[0049] Based on the pollution assessment results, the edge computing device decides whether to start the cleaning robot and robotic arm;

[0050] When it is detected that the cleanliness of the reflector is lower than the threshold value θL, a start signal is sent to the cleaning robot and the robotic arm;

[0051] The present invention is further configured as follows: in step 3, a high-resolution camera is installed around the reflector to capture the image of the reflector in real time;

[0052] Analyze the alignment status of the reflector in real time using computer vision technology and machine learning algorithms;

[0053] Automatically adjust the angle of the reflector based on the visual analysis results;

[0054] The present invention is further configured as follows: the automatic alignment method in step 3 is:

[0055] Arrange high-resolution cameras around the reflector, and let the image collected by each camera at time t be recorded as I(t);

[0056] Convert the collected image I(t) into a grayscale image Ig(t) to reduce the amount of calculation: Ig(t) = 0.299·R(t) + 0.587·G(t) + 0.114·B(t); where R(t), G(t), and B(t) are the red, green, and blue channel pixel values ​​of the image at time t, respectively;

[0057] The edge of the reflector is extracted using the Canny edge detection algorithm, E(t) = Canny(Ig(t)), where E(t) is the edge detection result at time t;

[0058] The contour line of the reflector is extracted from the edge detection result using Hough transform, H(t)=HoughLine(sE(t)), where H(t) is the Hough transform result at time t, including the detected straight line parameters;

[0059] The present invention is further configured as follows: the automatic alignment method in step 3 further includes:

[0060] Alignment state analysis: compare the reflector contour in the current image with the preset standard alignment state, calculate the offset, assume the preset standard alignment state is Href, the detection result at the current time t is H(t), and the offset calculation formula between the two is: Δθ(t)=H(t)-Href, where Δθ(t) is the alignment offset at time t;

[0061] The angle to be adjusted is calculated based on the offset Δθ(t), and an adjustment instruction is generated. The adjustment angle is set to α(t), and the adjustment instruction of the controller is A(t): Among them, kp, ki, kd are the parameters of proportional, integral, and differential controllers respectively;

[0062] Sending an adjustment instruction A(t) to the reflector drive system to adjust the angle of the reflector;

[0063] The present invention also provides a monitoring system for a solar thermal system reflector, comprising:

[0064] Sensor network module, including:

[0065] Temperature sensor unit, which monitors the temperature changes of the reflector and its supporting structure to prevent overheating damage;

[0066] Strain gauge units detect stress and strain in the mirror and its supporting structure to identify potential structural damage;

[0067] Humidity sensor unit, which monitors the ambient humidity to prevent the performance degradation of the reflector caused by humidity changes;

[0068] Data acquisition and transmission module, including:

[0069] Data acquisition unit, collects sensor data in real time to ensure data accuracy and timeliness;

[0070] The wireless transmission unit uses low-power wide area network (LPWAN) technology to transmit data to the computer vision module;

[0071] Data analysis and prediction module, including:

[0072] The real-time analysis unit uses big data technology to analyze sensor data in real time and identify abnormal situations;

[0073] The prediction algorithm unit uses machine learning algorithms to predict the health of the reflector and warn of possible failures in advance;

[0074] Optical sensing module, including:

[0075] An optical detection unit, which installs an optical sensor on the surface of the reflector to detect changes in the optical characteristics of the reflector surface in real time;

[0076] A data processing unit performs preliminary processing on the optical sensor data to determine the degree of contamination on the reflector surface;

[0077] Edge computing module, including:

[0078] Edge computing device unit: The edge computing device is deployed locally to process sensor data and make preliminary analysis, and perform cleanliness assessment. It uses edge computing technology to evaluate the cleanliness of the reflector in real time and provide cleaning recommendations.

[0079] Automatic cleaning system module, including:

[0080] Cleaning robot unit, a cleaning robot equipped with a flexible brush and a spray device, automatically cleans according to the cleanliness assessment results;

[0081] The automatic scheduling unit schedules the cleaning robot to clean the mirror according to the contamination status and cleaning priority of the mirror;

[0082] Image acquisition module, including:

[0083] A camera unit, which has high-resolution cameras mounted around the reflector to capture images of the reflector in real time;

[0084] An image transmission unit transmits the collected image data to a computer vision module for processing;

[0085] Computer vision modules, including:

[0086] An image processing unit processes the image of the reflector based on computer vision technology to extract key features;

[0087] A machine learning analysis unit that uses machine learning algorithms to analyze the alignment of the reflectors and identify small deviations;

[0088] Automatic alignment module, including:

[0089] The alignment control unit automatically adjusts the angle of the reflector to provide the best alignment state based on the visual analysis results;

[0090] Real-time feedback unit monitors the alignment effect of the reflector in real time, feeds back the adjustment results, and optimizes the alignment process.

[0091] Compared with the prior art, the present invention has the following beneficial effects:

[0092] The present invention monitors the health status and cleanliness of the reflector in real time based on sensors and cameras, and transmits data to a cloud computing platform using wireless networks, thereby avoiding frequent manual inspections and greatly improving monitoring efficiency. The present invention obtains detailed data of the reflector in real time through high-precision sensors, optical sensors and high-resolution cameras, and combines big data analysis and machine learning algorithms to achieve accurate monitoring of the health status and cleanliness of the reflector.

[0093] The present invention is an automatic alignment system based on computer vision, which captures the image of the reflector in real time and analyzes its alignment state, automatically adjusts the angle of the reflector, ensures that the reflector is always in the best alignment state, and improves the operation stability of the system;

[0094] The present invention uses predictive maintenance and intelligent monitoring to warn and handle potential problems in advance, reduce system failures and downtime, and reduce maintenance costs. At the same time, based on cloud computing and edge computing technology, it processes and analyzes a large amount of sensor data in real time, predicts health conditions through machine learning models, optimizes maintenance and adjustment strategies for reflectors, and realizes intelligent and data-driven monitoring.

[0095] The invention solves the problems in the prior art that the mirror relies on regular manual inspection and adjustment, which is inefficient and error-prone. Meanwhile, regular maintenance is generally adopted for maintenance, and it is impossible to monitor in real time whether the surface of the mirror is affected by dust and dirt, and it is impossible to detect and deal with aging, wear and dirt problems in time. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 The frame diagram of the monitoring system of the solar thermal system reflector of the present invention. DETAILED DESCRIPTION

[0097] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0098] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0099] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0100] Example 1

[0101] See also Figure 1In order to significantly improve the monitoring effect of the reflector of the solar thermal system, the present application combines the predictive maintenance of the health of the reflector, the intelligent monitoring of the cleanliness of the reflector surface, and the automatic alignment system based on computer vision to perform comprehensive monitoring, aiming to achieve real-time monitoring, automatic adjustment and predictive maintenance, and provides a monitoring method for the reflector of the solar thermal system, including:

[0102] Step 1. Perform predictive maintenance on the health of the reflectors. High-precision sensors are evenly installed on each reflector and its supporting structure. The sensor data is transmitted to the data analysis and prediction module in real time via a wireless network. Big data analysis and machine learning technologies are used to analyze the sensor data in real time to predict the health of the reflectors. Based on the prediction results, early warning signals are issued in advance to arrange preventive maintenance.

[0103] Use strain gauges, temperature sensors, and humidity sensors to monitor the health of the mirrors and supporting structures in real time;

[0104] Utilize LPWAN technology to transmit sensor data; analyze sensor data in real time on a cloud computing platform and use machine learning algorithms to predict the health of the reflector;

[0105] The real-time analysis method of sensor data is as follows:

[0106] Let the sensor on each reflector be Si, i represents different sensor types, including temperature sensor, strain gauge and humidity sensor, and the data collected by each sensor at time t is recorded as Di(t);

[0107] Assume the temperature sensor data is T(t), the strain gauge data is ε(t), and the humidity sensor data is H(t);

[0108] Using LPWAN technology, the collected data Di(t) is wirelessly transmitted to the cloud computing platform, i.e., the real-time analysis unit;

[0109] Data preprocessing is performed on the cloud computing platform, including data cleaning, denoising and standardization:

[0110] For each sensor data Di(t), normalization is performed: Among them, μi and σi are the mean and standard deviation of sensor Si data respectively;

[0111] Extract features from preprocessed data and use sliding window technique to extract time series features:

[0112] Set the window size to w and the step size to s: in, Identify the number of time steps within the window;

[0113] Use the long short-term memory network LSTM model to process time series data and perform model training:

[0114] The LSTM network architecture includes:

[0115] Input layer: input feature Fi(t);

[0116] Hidden layer: multi-layer LSTM unit;

[0117] Output layer: health status prediction results;

[0118] Let ht be the hidden state of the LSTM unit, Ct be the cell state, σ be the sigmoid activation function, and tanh be the tanh activation function;

[0119] Forget Gate:

[0120] Input Gate:

[0121] Candidate cell states:

[0122] Cell status update:

[0123] Output Gate:

[0124] Hidden state update: ht = ot·tanh(Ct);

[0125] Among them, Wf,Wi,WC,Wo and bf,bi,bC,bo are weight and bias matrices;

[0126] Use the trained LSTM model to predict health status based on real-time data:

[0127] set up Predicted health status at time t: Among them, F(t) is the data feature of the current time step;

[0128] According to the prediction results Conduct real-time monitoring and early warning;

[0129] Set the health threshold θ, when When the threshold value θ is exceeded, an early warning is triggered;

[0130] Step 2. Intelligently monitor the cleanliness of the reflector surface. Install an optical sensor on the reflector surface to detect changes in the scattering and absorption characteristics of light on the reflector surface. The optical sensor collects data in real time and performs preliminary processing through edge computing devices to determine the cleanliness. When it is detected that the contamination on the reflector surface reaches a threshold, the cleaning robot and mechanical arm are automatically started for cleaning.

[0131] Install high-precision optical sensors on the reflector surface to monitor cleanliness in real time;

[0132] Use edge computing technology to perform preliminary processing and analysis of sensor data;

[0133] Combined with cleanliness monitoring data, cleaning robots and robotic arms are automatically started;

[0134] The initial processing and analysis of sensor data using edge computing technology is as follows:

[0135] A high-precision optical sensor is installed on the surface of each reflector to monitor the cleanliness of the surface. The sensor measures the scattering and absorption characteristics of light to reflect the degree of surface contamination. The data collected by the optical sensor is set to L(t), which represents the optical reflection intensity at time t.

[0136] The sensor data L(t) is collected in real time by edge computing devices, and the initial processing includes noise filtering and data standardization:

[0137] Use the moving average method to smooth the data and set the window size to w: Standardization: Where i represents the time step, L(ti) represents the optical reflection intensity data collected at time ti, and μ L and σ L are the mean and standard deviation of historical data respectively;

[0138] Use edge computing devices to perform real-time analysis to determine whether the cleanliness of the reflector has reached the cleanliness threshold;

[0139] A cleanliness threshold θL is set. When the normalized optical reflection intensity L'(t) is lower than the threshold, it is considered that the reflector needs to be cleaned.

[0140] Based on the pollution assessment results, the edge computing device decides whether to start the cleaning robot or robotic arm;

[0141] When it is detected that the cleanliness of the reflector is lower than the threshold value θL, a start signal is sent to the cleaning robot and the robotic arm;

[0142] Step 3. Automatic alignment based on computer vision, arranging cameras around the reflector to capture images of the reflector; the camera collects the image of the reflector in real time, and analyzes it through computer vision technology to determine the alignment state of the reflector; according to the analysis result, the automatic alignment device adjusts the angle of the reflector in real time to keep it in the best alignment state at all times;

[0143] Install high-resolution cameras around the reflector to capture images of the reflector in real time;

[0144] Analyze the alignment status of the reflector in real time using computer vision technology and machine learning algorithms;

[0145] Automatically adjust the angle of the reflector based on the visual analysis results;

[0146] The automatic alignment method is:

[0147] Arrange high-resolution cameras around the reflector, and let the image collected by each camera at time t be recorded as I(t);

[0148] Convert the collected image I(t) into a grayscale image Ig(t) to reduce the amount of calculation: Ig(t) = 0.299·R(t) + 0.587·G(t) + 0.114·B(t); where R(t), G(t), and B(t) are the red, green, and blue channel pixel values ​​of the image at time t, respectively;

[0149] The edge of the reflector is extracted using the Canny edge detection algorithm, E(t) = Canny(Ig(t)), where E(t) is the edge detection result at time t;

[0150] The contour line of the reflector is extracted from the edge detection result using Hough transform, H(t)=HoughLine(sE(t)), where H(t) is the Hough transform result at time t, including the detected straight line parameters;

[0151] Alignment state analysis: compare the reflector contour in the current image with the preset standard alignment state, calculate the offset, assume the preset standard alignment state is Href, the detection result at the current time t is H(t), and the offset calculation formula between the two is: Δθ(t)=H(t)-Href, where Δθ(t) is the alignment offset at time t;

[0152] The angle to be adjusted is calculated based on the offset Δθ(t), and an adjustment instruction is generated. The adjustment angle is set to α(t), and the adjustment instruction of the controller is A(t): Among them, kp, ki, kd are the parameters of proportional, integral, and differential controllers respectively;

[0153] The adjustment instruction A(t) is sent to the mirror drive system to adjust the angle of the mirror.

[0154] The present invention also discloses a monitoring system for a solar thermal system reflector, comprising:

[0155] Sensor network module, including:

[0156] Temperature sensor unit, which monitors the temperature changes of the reflector and its supporting structure to prevent overheating damage;

[0157] Strain gauge units detect stress and strain in the mirror and its supporting structure to identify potential structural damage;

[0158] Humidity sensor unit, which monitors the ambient humidity to prevent the performance degradation of the reflector caused by humidity changes;

[0159] Data acquisition and transmission module, including:

[0160] Data acquisition unit, collects sensor data in real time to ensure data accuracy and timeliness;

[0161] The wireless transmission unit uses low-power wide area network (LPWAN) technology to transmit data to the computer vision module;

[0162] Data analysis and prediction module, including:

[0163] The real-time analysis unit uses big data technology to analyze sensor data in real time and identify abnormal situations;

[0164] The prediction algorithm unit uses machine learning algorithms to predict the health of the reflector and warn of possible failures in advance;

[0165] Optical sensing module, including:

[0166] An optical detection unit, which installs an optical sensor on the surface of the reflector to detect changes in the optical characteristics of the reflector surface in real time;

[0167] A data processing unit performs preliminary processing on the optical sensor data to determine the degree of contamination on the reflector surface;

[0168] Edge computing module, including:

[0169] Edge computing device unit: The edge computing device is deployed locally to process sensor data and make preliminary analysis, and perform cleanliness assessment. It uses edge computing technology to evaluate the cleanliness of the reflector in real time and provide cleaning recommendations.

[0170] Automatic cleaning system module, including:

[0171] Cleaning robot unit, a cleaning robot equipped with a flexible brush and a spray device, automatically cleans according to the cleanliness assessment results;

[0172] The automatic scheduling unit schedules the cleaning robot to clean the mirror according to the contamination status and cleaning priority of the mirror;

[0173] Image acquisition module, including:

[0174] A camera unit, which has high-resolution cameras mounted around the reflector to capture images of the reflector in real time;

[0175] An image transmission unit transmits the collected image data to a computer vision module for processing;

[0176] Computer vision modules, including:

[0177] An image processing unit processes the image of the reflector based on computer vision technology to extract key features;

[0178] A machine learning analysis unit that uses machine learning algorithms to analyze the alignment of the reflectors and identify small deviations;

[0179] Automatic alignment module, including:

[0180] The alignment control unit automatically adjusts the angle of the reflector to provide the best alignment state based on the visual analysis results;

[0181] Real-time feedback unit monitors the alignment effect of the reflector in real time, feeds back the adjustment results, and optimizes the alignment process.

[0182] This application performs predictive maintenance on the health status of the reflector. High-precision sensors, including temperature sensors, strain gauges, and humidity sensors, are installed on each reflector and its supporting structure to monitor the health status of the reflector and its supporting structure in real time. The collected data is wirelessly transmitted to the cloud computing platform through LPWAN technology. The sensor data is preprocessed on the cloud computing platform, and the time series features are extracted using the sliding window technology. The long short-term memory network LSTM model is used to process the time series data and train the model; the sensor data is analyzed in real time and the health status of the reflector is predicted; when the prediction result exceeds the set health status threshold, an early warning signal is triggered, and preventive maintenance is performed to effectively avoid system failures and downtime;

[0183] In addition, the cleanliness of the reflector surface is intelligently monitored. A high-precision optical sensor is installed on the reflector surface to monitor the cleanliness of the reflector surface in real time. The optical sensor is used to measure the scattering and absorption characteristics of light to reflect the degree of surface contamination. The edge computing device analyzes the sensor data in real time to determine whether the cleanliness of the reflector reaches the cleaning threshold. When the standardized optical reflection intensity is lower than the set threshold, the reflector is considered to need cleaning, and the cleaning robot or robotic arm is automatically started to perform the cleaning operation. Based on the pollution assessment results, the edge computing device decides whether to start the cleaning robot or robotic arm to ensure that the reflector surface is always kept in the best clean state.

[0184] In step 3, the reflector is automatically aligned based on computer vision technology. A high-resolution camera is installed around the reflector to capture the image of the reflector in real time. The image captured by the camera is converted into a grayscale image. The edge of the reflector is then extracted using the Canny edge detection algorithm, and the contour of the reflector is extracted from the edge detection result using the Hough transform. The offset is calculated by comparing the contour of the reflector in the current image with the preset standard alignment state. Based on the offset, the PID control algorithm is used to calculate the angle to be adjusted and generate an adjustment instruction, which is sent to the reflector drive system to achieve real-time adjustment. This enables automatic monitoring and adjustment of the reflector angle and improves the alignment accuracy of the reflector.

[0185] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for monitoring a reflector of a solar thermal system, characterized in that: include: Step 1. Perform predictive maintenance on the health of the reflectors. High-precision sensors are evenly installed on each reflector and its supporting structure. The sensor data is transmitted to the data analysis and prediction module in real time via a wireless network. Big data analysis and machine learning technologies are used to analyze the sensor data in real time to predict the health of the reflectors. Based on the prediction results, an early warning signal is issued in advance. In step 1, strain gauges, temperature sensors, and humidity sensors are used to monitor the health of the reflector and supporting structure in real time; Utilize LPWAN technology to transmit sensor data; analyze sensor data in real time on a cloud computing platform and use machine learning algorithms to predict the health of the reflector; Step 2. Intelligently monitor the cleanliness of the reflector surface by installing an optical sensor on the reflector surface to detect changes in the scattering and absorption characteristics of light on the reflector surface. The optical sensor collects data in real time and performs preliminary processing through the edge computing device to determine the cleanliness. When it is detected that the contamination on the reflector surface reaches a threshold, the cleaning robot and mechanical arm are automatically started for cleaning; Step 3. Automatically align based on computer vision, and arrange cameras around the reflector to capture images of the reflector; The camera collects images of the reflector in real time and analyzes them through computer vision technology to determine the alignment status of the reflector; based on the analysis results, the automatic alignment device adjusts the angle of the reflector in real time.

2. A method for monitoring a solar thermal system reflector according to claim 1, characterized in that: The real-time analysis method of sensor data in step 1 is: Let the sensor on each reflector be S i , i represents different sensor types, including temperature sensors, strain gauges and humidity sensors. The data collected by each sensor at time t is recorded as D i (t); Assume the temperature sensor data is T(t), the strain gauge data is ε(t), and the humidity sensor data is H(t); Using LPWAN technology, the collected data i (t) wireless transmission to the cloud computing platform, i.e., the real-time analysis unit; Data preprocessing is performed on the cloud computing platform, including data cleaning, denoising and standardization: For each sensor data D i (t), and perform standardization: Among them, μ i and σ i Sensor S i The mean and standard deviation of the data; Extract features from preprocessed data and use sliding window technique to extract time series features: Set the window size to w and the step size to s: F i (t) = [D' i (t),D' i (t+s),...,D' i (t+nw)], where The number of time steps in the identification window, F i (t) represents the feature data collected by the sensor.

3. A method for monitoring a solar thermal system reflector according to claim 2, characterized in that: Real-time analysis of sensor data also includes: Use the long short-term memory network LSTM model to process time series data and perform model training: The LSTM network architecture includes: Input layer: input feature F i (t); Hidden layer: multi-layer LSTM unit; Output layer: health status prediction results; Assume h t is the hidden state of the LSTM unit, C t is the cell state, σ is the sigmoid activation function, and tanh is the tanh activation function; Forget Gate: Input Gate: Candidate cell states: Cell status update: Output Gate: Hide status update:h t =o t tanh(C t ); Among them, W f ,W i ,W C ,W o and b f ,b i ,b C ,b o are weight and bias matrices, x t represents the input features of the LSTM model input layer, i.e., the standardized feature vector collected from multiple sensors, f t Represents the forget gate output in the LSTM model; Use the trained LSTM model to predict health status based on real-time data.

4. A method for monitoring a solar thermal system reflector according to claim 3, characterized in that: Health status is predicted by: set up Predicted health status at time t: Among them, F(t) is the data feature of the current time step; According to the prediction results Conduct real-time monitoring and early warning; Set the health threshold θ, when When the threshold value θ is exceeded, an early warning is triggered.

5. A method for monitoring a solar thermal system reflector according to claim 4, characterized in that: In step 2, a high-precision optical sensor is installed on the surface of the reflector to monitor the cleanliness in real time; Use edge computing technology to perform preliminary processing and analysis of sensor data; Combined with cleanliness monitoring data, cleaning robots and robotic arms are automatically started.

6. A method for monitoring a solar thermal system reflector according to claim 5, characterized in that: The initial processing and analysis of sensor data using edge computing technology in step 2 is as follows: A high-precision optical sensor is installed on the surface of each reflector to monitor the cleanliness of the surface. The sensor measures the scattering and absorption characteristics of light to reflect the degree of surface contamination. The data collected by the optical sensor is set to L(t), which represents the optical reflection intensity at time t. The sensor data L(t) is collected in real time by edge computing devices, and the initial processing includes noise filtering and data standardization: Use the moving average method to smooth the data and set the window size to w: Standardization: Where i represents the time step, L(ti) represents the optical reflection intensity data collected at time ti, and μ L and σ L are the mean and standard deviation of historical data respectively; Use edge computing devices to perform real-time analysis to determine whether the cleanliness of the reflector has reached the cleanliness threshold; Set the cleanliness threshold θ L , when the normalized optical reflection intensity L'(t) is lower than the threshold, it is considered that the reflector needs to be cleaned; Based on the pollution assessment results, the edge computing device decides whether to start the cleaning robot or robotic arm; When the cleanliness of the reflector is detected to be lower than the threshold value θ L When the cleaning robot and the robotic arm are in operation, a start signal is sent.

7. A method for monitoring a solar thermal system reflector according to claim 6, characterized in that: In step 3, a high-resolution camera is installed around the reflector to capture the image of the reflector in real time; Analyze the alignment status of the reflector in real time using computer vision technology and machine learning algorithms; Combined with the visual analysis results, the angle of the reflector is automatically adjusted.

8. A method for monitoring a solar thermal system reflector according to claim 7, characterized in that: The automatic alignment method in step 3 is: Arrange high-resolution cameras around the reflector, and let the image collected by each camera at time t be recorded as I(t); Convert the acquired image I(t) into a grayscale image I g (t), reduce the amount of calculation: I g (t)=0.299·R(t)+0.587·G(t)+0.114·B(t), where R(t), G(t), and B(t) are the red, green, and blue channel pixel values ​​of the image at time t, respectively. g (t) represents the image data after preprocessing at time t; The Canny edge detection algorithm is used to extract the edge of the reflector, E(t) = Canny(I g (t)), where E(t) is the edge detection result at time t; The Hough transform is used to extract the contour line of the reflector from the edge detection result, H(t)=HoughLines(E(t)), where H(t) is the Hough transform result at time t, including the detected straight line parameters, and HoughLines represents the Hough straight line transform of the edge detection result.

9. A method for monitoring a reflector of a solar thermal system according to claim 8, characterized in that: The automatic alignment method in step 3 also includes: Alignment state analysis: compare the reflector contour in the current image with the preset standard alignment state, calculate the offset, and set the preset standard alignment state as H ref , the detection result at the current time t is H(t), and the offset calculation formula between the two is: Δθ(t)=H(t)-H ref , where Δθ(t) is the alignment offset at time t; The angle to be adjusted is calculated based on the offset Δθ(t), and an adjustment instruction is generated. The adjustment angle is set to α(t), and the adjustment instruction of the controller is A(t): Among them, k p ,k i ,k d are the parameters of proportional, integral and differential controllers respectively; The adjustment instruction A(t) is sent to the mirror drive system to adjust the angle of the mirror.

10. A monitoring system for a solar thermal system reflector, characterized in that: A method for monitoring a solar thermal system reflector according to any one of claims 1 to 9, comprising: Sensor network module, including: A temperature sensor unit to monitor temperature changes of the mirror and its supporting structure; Strain gauge units to detect stress and strain in the mirror and its supporting structure; Humidity sensor unit, monitoring ambient humidity; Data acquisition and transmission module, including: Data acquisition unit, collects sensor data in real time; The wireless transmission unit uses wide area network technology to transmit data to the computer vision module; Data analysis and prediction module, including: The real-time analysis unit uses big data technology to analyze sensor data in real time and identify abnormal situations; The prediction algorithm unit uses machine learning algorithms to predict the health of the reflector and warn of possible failures in advance; Optical sensing module, including: An optical detection unit, in which an optical sensor is installed on the surface of the reflector to detect changes in the optical characteristics of the reflector surface in real time; A data processing unit performs preliminary processing on the optical sensor data to determine the degree of contamination on the reflector surface; Edge computing module, including: Edge computing device unit: The edge computing device is deployed locally to process sensor data and make preliminary analysis, and perform cleanliness assessment. It uses edge computing technology to evaluate the cleanliness of the reflector in real time and provide cleaning recommendations. Automatic cleaning system module, including: Cleaning robot unit, a cleaning robot equipped with a flexible brush and a spray device, automatically cleans according to the cleanliness assessment results; The automatic scheduling unit schedules the cleaning robot to clean the mirror according to the contamination status and cleaning priority of the mirror; Image acquisition module, including: A camera unit, which has high-resolution cameras mounted around the reflector to capture images of the reflector in real time; An image transmission unit transmits the collected image data to a computer vision module for processing; Computer vision modules, including: An image processing unit processes the image of the reflector based on computer vision technology to extract key features; A machine learning analysis unit that uses machine learning algorithms to analyze the alignment of the reflectors and identify small deviations; Automatic alignment module, including: The alignment control unit automatically adjusts the angle of the reflector according to the visual analysis results; The real-time feedback unit monitors the alignment effect of the reflector in real time and provides feedback on the adjustment results.

Images