Equipment monitoring method based on big data

Abstract

The invention discloses an equipment monitoring method based on big data, and relates to the field of equipment monitoring, firstly, the method can calculate the video ambiguity Qi, the picture jitter degree Sshake, the dust coverage rate Djust and the camera angle drift rate Arift through a related algorithm formula by using a big data analysis technology; based on the parameters, the system can comprehensively calculate the comprehensive state score SPF of the inspection camera, and compares and analyzes the comprehensive state score SPF with a preset running state threshold Y to assess the current health state of the equipment. And when the comprehensive state score SPF is lower than Y, the system generates early warning information to prompt that the equipment is possibly abnormal and needs to be checked, cleaned or recalibrated in time. Through the process, multi-dimensional intelligent state monitoring of the camera equipment can be achieved, it is ensured that the operation condition of the equipment can be evaluated timely and accurately, and potential faults are prevented from occurring.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment monitoring, and in particular to an equipment monitoring method based on big data. Background Art

[0002] Equipment monitoring is a very important link, especially in the intelligent inspection system in property management. By real-time monitoring of security equipment such as cameras, the safety and efficiency of property management can be improved. In specific application scenarios, property inspection cameras, as key equipment for intelligent monitoring, can achieve all-weather monitoring and safety management of property sites by collecting data and video information of the surrounding environment.

[0003] In the Chinese invention application with application publication number CN108810526A, a camera fault intelligent prediction method, device, camera management server and computer storage medium are disclosed. The camera fault intelligent prediction method includes: establishing a fault prediction model based on the camera state sequence information before the fault of multiple faults determined and the historical fault information before the corresponding fault; when predicting the fault, using the camera state sequence information before the prediction, the historical fault information before the prediction and the fault prediction model to predict whether the fault will occur. The camera fault intelligent prediction method of the present invention autonomously predicts the fault of the camera based on the historical information, so that the camera maintenance personnel can prepare the maintenance work in advance, maintain the faulty camera in time, and reduce the safety hazards caused by the fault.

[0004] In combination with the prior art, the above application still has the following deficiencies:

[0005] Although property inspection cameras have basic video surveillance functions, they have many deficiencies in equipment status monitoring, making it difficult to provide comprehensive and real-time fault warnings. When cameras are exposed to outdoor environments for a long time, image quality may deteriorate due to dust accumulation, vibration, ambient light changes, and other problems that are not immediately discovered when they occur. Due to the inability to conduct timely status assessment and early warning, the monitoring screen may appear blurred, the viewing angle may shift, and other abnormal conditions may occur, which in turn affects the security of the entire property management system, and may even lose the monitoring function at a critical moment, posing serious safety hazards. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a device monitoring method based on big data, which solves the problems mentioned in the background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A device monitoring method based on big data, comprising the following steps:

[0008] S1. Through the sensor group and built-in functions installed on the property inspection camera, the camera equipment operation data and video data are collected in real time, and the collected camera equipment operation data and video data are transmitted to the cloud server through the wireless gateway;

[0009] S2. receiving the collected camera device operation data and video data in real time in the cloud server, performing data processing on the camera device operation data, extracting video quality data from the video data by image processing technology, fusing the camera device operation data and the video quality data to obtain a camera efficiency data set, and storing the camera efficiency data set;

[0010] S3. Use big data analysis to build relevant algorithm formulas, extract camera performance data sets and input them into relevant algorithm formulas to calculate and output video blur Qi, image shake Sshake, dust coverage Ddust and camera angle drift Adrift;

[0011] S4, based on the obtained video blur Qi, image jitter Sshake, dust coverage Ddust and camera angle drift rate Adrift, and combined with the camera performance data set, a comprehensive calculation is performed to obtain the comprehensive status score SPF of the inspection camera, and a preset operation status threshold Y is used to perform a preliminary comparative evaluation and analysis of the health status of the camera with the obtained comprehensive status score SPF;

[0012] S5. Based on the preliminary comparative evaluation results, the fault prediction and early warning mechanism is triggered. By constructing a fault prediction algorithm formula and combining the obtained comprehensive status score SPF into the fault prediction algorithm formula, the camera failure probability Pf is calculated and output. The preset fault prediction threshold G is compared with the camera failure probability Pf for a second comparative evaluation, and early warning information is generated based on the evaluation results.

[0013] Preferably, S1 includes S11 and S12;

[0014] S11. By adding sensor groups and built-in functions to property inspection cameras, real-time data on camera equipment operation is collected, and real-time video data is collected through the built-in functions of the cameras;

[0015] The sensor set includes an ambient temperature sensor, a light sensor, and a gyroscope;

[0016] Device operation data includes device temperature Ti, network delay Li, ambient light change rate Elight, and camera angle Jd;

[0017] The equipment temperature Ti is acquired through the ambient temperature sensor;

[0018] Ambient light change rate Elight uses an ambient light sensor equipped with a camera to directly measure changes in ambient light intensity;

[0019] The network delay Li is obtained through the network diagnostic function of the camera system, which is used to detect the network status;

[0020] The camera angle Jd is acquired through the gyroscope;

[0021] S12. The inspection camera and the installed sensor group are integrated and connected with the cloud server by setting up a LoRa wireless gateway, and then the collected equipment operation data and video data are transmitted to the cloud server through the LoRa communication protocol.

[0022] Preferably, S2 includes S21, S22 and S23;

[0023] S21. Receive the collected device operation data and video data in real time in the cloud server, and process the device operation data, wherein the data processing includes data cleaning, data format standardization, and time synchronization and alignment;

[0024] Data cleaning is used to remove erroneous, missing, and unreasonable data points. For example, network interruption may result in missing data for some periods of time, or abnormal data caused by abnormal instantaneous fluctuations in equipment.

[0025] Data format standardization is used to convert device operation data from different sources into a unified format. For example, hardware sensor data and image analysis data may have different timestamp formats or unit standards.

[0026] Time synchronization and alignment are used to synchronize and align the inconsistent device operation data collection time caused by network delay Li;

[0027] S22. Based on the collected video data, video quality data is extracted from the video data by image processing technology, where the image processing technology includes a Sobel operator and a Gaussian blur algorithm for edge detection, and the video quality data includes a brightness value I(x, y) at the pixel point (x, y), a brightness gradient change ▽I(x, y) at the pixel point (x, y), a brightness value Iblur(x, y) of the pixel point (x, y) in a blurred video frame, and a brightness value Isharp(x, y) of the pixel point (x, y) in a clear video frame;

[0028] Using the Sobel operator of edge detection, the gradient of the image is estimated by calculating the brightness changes in the horizontal and vertical directions. After calculation, the gradient value of each pixel will reflect the changes in the brightness of its surroundings. A clear image will have a higher gradient value in the edge area.

[0029] Use the Gaussian blur algorithm to blur the image. Gaussian blur will reduce the details in the image and make the edges smooth, imitating the effect of dust covering the lens. By comparing the pixel brightness values ​​of the blurred image and the original image, the change in image clarity can be judged.

[0030] S23. By constructing a cloud server database, a historical database table and a real-time database table are constructed at the same time, and a write port and a write port are set. After executing step S21 and step S22, the acquired device operation data and video data at the same time point are horizontally integrated to obtain a real-time camera efficiency data set; the acquired device operation data and video data are vertically integrated to obtain a historical camera efficiency data set; and the real-time camera efficiency data set is stored in the real-time database table through the write port, and the historical camera efficiency data set is written into the historical database table.

[0031] Preferably, S3 includes S31, S32, S33 and S34;

[0032] S31, by constructing a video clarity algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) and the brightness gradient change ▽I(x, y) at the pixel point (x, y) through the write port of the cloud server database, inputting them into the video clarity algorithm formula, calculating and outputting the video blur Qi, and capturing the local blur phenomenon caused by focus offset;

[0033] The video blur Qi is calculated and output by the following video clarity algorithm formula;

[0034]

[0035] Where N represents the total number of pixels in each frame of the video, which is obtained by a known resolution, w(x, y) represents the pixel point (x, y), which is dynamically set according to the image edge strength by the Sobel operator of edge detection, and I max Indicates the upper limit of the brightness of the video frame, which is used to normalize the effect of brightness on the gradient.

[0036] Preferably, S32, by constructing a picture jitter analysis algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) through the write port of the cloud server database and inputting it into the picture jitter analysis algorithm formula, calculating and outputting the picture jitter degree Sshake, and analyzing the jitter of the inspection camera in different time periods;

[0037] The image jitter degree Sshake is calculated and obtained by the following image jitter analysis algorithm formula;

[0038]

[0039] In the formula, T represents the total number of frames in the time window, t represents the time, and α t Represents the time weight, which represents the weight value at time t. It is specifically set by the property user. The time weight is set according to the application requirements. Usually, an exponential weight or a linear attenuation function is used to assign a larger weight to the most recent frame. I(x, y, t) represents the brightness value of the pixel point (x, y) at time t. I(x, y, t-1) represents the brightness value of the pixel point (x, y) at time t-1. It represents the brightness value of the pixel point in the previous time frame and is used to calculate the change between frames. V(x, y, t) represents the velocity vector of the pixel point (x, y) at time t. It is calculated by analyzing the pixel motion in continuous frames of the video using the optical flow method. V max Indicates the upper limit of the velocity vector in the video frame.

[0040] Preferably, S33, by constructing a dust coverage analysis algorithm formula, extracting the brightness value Iblur(x, y) of the pixel point (x, y) in the blurred video frame and the brightness value Isharp(x, y) of the pixel point (x, y) in the clear video frame through the write port of the cloud server database, inputting them into the dust coverage analysis algorithm formula, calculating and outputting the dust coverage rate Ddust, and analyzing the dust coverage of the inspection camera;

[0041] The dust coverage rate Ddust is calculated and output by the following dust coverage analysis algorithm formula;

[0042]

[0043] Where Iblur(x, y, σ) represents the brightness value of the pixel (x, y) after Gaussian blur algorithm processing, σ represents the standard deviation of the blur kernel, and V(x, y) represents the motion vector of the pixel (x, y).

[0044] Preferably, S34, by constructing an angle drift analysis algorithm formula, extracting the camera angle Jd through the write port of the cloud server database and inputting it into the angle drift analysis algorithm formula, calculating and outputting the camera angle drift rate Adrift, and analyzing the speed of the inspection camera angle change and the camera's violent drift situation;

[0045] The camera angle drift rate Adrift is calculated and output by the following angle drift analysis algorithm formula;

[0046]

[0047] In the formula, Jd0 represents the initial installation angle of the inspection camera, d represents the differential operator, represents the rate of change of the camera angle Jd over time t, that is, the angular velocity, It represents the rate of change of angular velocity, that is, angular acceleration. γ and δ represent the preset weight values ​​of angular velocity and angular acceleration respectively. The specific values ​​are set by the user according to the application scenario, and γ+δ=1.

[0048] Preferably, S4 includes S41 and S42;

[0049] S41, extracting the obtained video blur Qi, image shake degree Sshake, dust coverage rate Ddust and camera angle drift rate Adrift, and combining them with the equipment operation data, performing comprehensive calculation to obtain the comprehensive status score SPF of the inspection camera, and scoring the overall operation status of the camera;

[0050] The comprehensive status score SPF is calculated by the following algorithm formula;

[0051] SPF=a1·Qi+a2·Sshake+a3·Ddust+a4·Adrift+a5·Ti+a6·Li+a7·Elight+ε;

[0052] Wherein, a1, a2, a3, a4, a5, a6 and a7 represent the preset weight values ​​of video blur Qi, image shake Sshake, dust coverage Ddust, camera angle drift Adrift, device temperature Ti, network delay Li and ambient light change rate Elight, respectively. The specific values ​​are set by the user according to the application scenario, and a1+a2+a3+a4+a5+a6+a7=1, ε represents the comprehensive scoring error term;

[0053] S42, based on the historical data and performance standards of the inspection camera, a preset operation status threshold Y is set, and then a preliminary comparative evaluation is performed with the obtained comprehensive status score SPF to analyze the operation status of the current inspection camera. The specific evaluation contents are as follows;

[0054] When the comprehensive status score SPF> the operating status threshold Y, it means that the equipment is in normal operating state and does not need maintenance at present. The system is working within the normal range. At this time, step S5 is executed to perform fault prediction and generate a maintenance mechanism;

[0055] When the comprehensive status score SPF ≤ the operating status threshold Y, it means that the equipment is abnormal. At this time, the first warning information is generated to prompt the property to immediately check, clean and recalibrate the patrol camera.

[0056] Preferably, S5 includes S51 and S52;

[0057] S51. When the preliminary comparison and evaluation shows that the equipment is in normal operation, the equipment fault prediction is triggered. By constructing a fault prediction algorithm formula and extracting the comprehensive status score SPF combined with the equipment temperature Ti and network delay Li of the inspection camera, the camera fault probability Pf is calculated and output, and the fault occurrence of the inspection camera is predicted;

[0058] The camera failure probability Pf is calculated using the following fault prediction algorithm formula:

[0059]

[0060] Where SPF t-1 represents the comprehensive status score at the previous time point t-1, κ represents the prediction error term, b1, b2 and b3 represent the preset weight values ​​of the comprehensive status score change rate, device temperature Ti and network delay Li respectively. The specific values ​​are set by the user according to the application scenario, and b1+b2+b3=1.

[0061] Preferably, S52, preset a fault prediction threshold G by the frequency of occurrence of patrol camera faults, and then perform a secondary comparative evaluation with the obtained camera fault occurrence probability Pf to predict the fault condition of the patrol camera, and generate warning information based on the evaluation result. The specific evaluation content is as follows;

[0062] When the probability of camera failure Pf>fault prediction threshold G, it indicates that the inspection camera has a predicted failure. At this time, the second warning information is generated to prompt the property to carry out planned maintenance;

[0063] When the camera failure probability Pf ≤ the failure prediction threshold G, it means that there is no predicted failure for the predicted inspection camera. At this time, maintenance is performed according to the original maintenance plan.

[0064] Beneficial Effects

[0065] The present invention provides a device monitoring method based on big data. It has the following beneficial effects:

[0066] (1) This method adds a sensor group to the property inspection camera and combines it with the camera's own acquisition function. The system can collect equipment operation data and video data in real time. After transmitting this data to the cloud server through the LoRa wireless gateway, it is received and processed in real time. The accuracy and consistency of the data are ensured through processing steps such as cleaning, format standardization, and time synchronization of the operation data. The video data uses image processing technology to extract video quality data, which is fused with the equipment operation data to generate and store the camera performance data set. This process effectively improves the real-time monitoring capability and data fusion efficiency of the equipment, ensures the integrity and accuracy of the data, and provides a solid data foundation for subsequent status evaluation and fault prediction.

[0067] (2) This method uses big data analysis technology. Through relevant algorithm formulas, this method can calculate the video blur Qi, the image jitter Sshake, the dust coverage Ddust, and the camera angle drift rate Adrift. Based on these parameters, the system can comprehensively calculate the comprehensive status score SPF of the inspection camera, and compare and analyze it with the preset operating status threshold Y to evaluate the current health status of the equipment. When the comprehensive status score SPF is lower than Y, the system will generate an early warning message to indicate that the equipment may be abnormal and needs to be inspected, cleaned or recalibrated in time. This process can realize multi-dimensional intelligent status monitoring of camera equipment, ensure that the equipment operation status can be evaluated in a timely and accurate manner, and prevent the occurrence of potential failures.

[0068] (3) Based on the preliminary health status assessment results, the system can trigger the fault prediction mechanism, and combine the comprehensive status score SPF, equipment temperature Ti, and network delay Li factors to calculate the camera failure probability Pf through the fault prediction algorithm. The failure probability Pf will be compared and evaluated with the preset failure prediction threshold G for a second time. When the failure probability Pf exceeds the failure prediction threshold G, the system will generate a second-level warning information, indicating that a failure may occur and recommending planned maintenance. This intelligent fault prediction mechanism not only improves the reliability of monitoring equipment, but also reduces the risk of sudden equipment failure. Through the optimized maintenance mechanism, the property management party can take timely measures before the equipment fails, effectively reduce maintenance costs, and ensure the continuous and stable operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 The figure is a schematic diagram of the steps of a device monitoring method based on big data of the present invention. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0071] Example 1

[0072] The present invention provides a device monitoring method based on big data, please refer to Figure 1 , including the following steps:

[0073] S1. Through the sensor group and built-in functions installed on the property inspection camera, the camera equipment operation data and video data are collected in real time, and the collected camera equipment operation data and video data are transmitted to the cloud server through the wireless gateway;

[0074] S2. receiving the collected camera device operation data and video data in real time in the cloud server, performing data processing on the camera device operation data, extracting video quality data from the video data by image processing technology, fusing the camera device operation data and the video quality data to obtain a camera efficiency data set, and storing the camera efficiency data set;

[0075] S3. Use big data analysis to build relevant algorithm formulas, extract camera performance data sets and input them into relevant algorithm formulas to calculate and output video blur Qi, image shake Sshake, dust coverage Ddust and camera angle drift Adrift;

[0076] S4, based on the obtained video blur Qi, image jitter Sshake, dust coverage Ddust and camera angle drift rate Adrift, and combined with the camera performance data set, a comprehensive calculation is performed to obtain the comprehensive status score SPF of the inspection camera, and a preset operation status threshold Y is used to perform a preliminary comparative evaluation and analysis of the health status of the camera with the obtained comprehensive status score SPF;

[0077] S5. Based on the preliminary comparative evaluation results, the fault prediction and early warning mechanism is triggered. By constructing a fault prediction algorithm formula and combining the obtained comprehensive status score SPF into the fault prediction algorithm formula, the camera failure probability Pf is calculated and output. The preset fault prediction threshold G is compared with the camera failure probability Pf for a second comparative evaluation, and early warning information is generated based on the evaluation results.

[0078] In this embodiment, the method collects multi-dimensional data of the device in real time by utilizing the sensor group and the built-in functions of the camera, and transmits it to the cloud through the wireless gateway, thereby ensuring the real-time transmission and storage of the operating data and video data of the camera device. This enables the entire monitoring system to perform efficient and low-latency device status monitoring over a large range. Secondly, by cleaning, standardizing and time synchronizing the collected data, and extracting video quality data in combination with image processing technology, the camera performance data set is further fused. This data fusion process greatly improves the accuracy and comprehensiveness of the camera device performance monitoring, ensuring that the system can capture subtle anomalies in the operation of the device. The key to this method is that the relevant algorithm constructed through big data analysis can accurately output the video blur Qi, the image jitter degree Sshake, the dust coverage rate Ddust, and the camera angle drift rate Adrift. Combined with these parameters, the system generates a comprehensive status score SPF and compares and analyzes it with the preset operating status threshold Y to further evaluate the health status of the camera. Through this analysis, the system can not only detect equipment anomalies in a timely manner, but also perform intelligent early warnings to prompt maintenance personnel to check, clean and adjust. Compared with traditional technologies, the significant improvement of this method lies in its intelligent fault prediction and early warning mechanism. After completing the initial status assessment, the system calculates the probability of equipment failure Pf through the fault prediction algorithm and compares it with the fault prediction threshold G. Based on the evaluation results, the system can generate accurate early warning information and formulate maintenance plans for potential problems. This intelligent early warning mechanism greatly improves the maintenance efficiency of camera equipment, avoids safety hazards caused by sudden equipment failures, and reduces the frequency and errors of human intervention.

[0079] Example 2

[0080] This embodiment is explained in Example 1, please refer to Figure 1 , specifically: S1 includes S11 and S12;

[0081] S11. By adding sensor groups and built-in functions to property inspection cameras, real-time data on camera equipment operation is collected, and real-time video data is collected through the built-in functions of the cameras;

[0082] The sensor set includes an ambient temperature sensor, a light sensor, and a gyroscope;

[0083] Device operation data includes device temperature Ti, network delay Li, ambient light change rate Elight, and camera angle Jd;

[0084] The equipment temperature Ti is acquired through the ambient temperature sensor;

[0085] Ambient light change rate Elight uses an ambient light sensor equipped with a camera to directly measure changes in ambient light intensity;

[0086] The network delay Li is obtained through the network diagnostic function of the camera system, which is used to detect the network status;

[0087] The camera angle Jd is acquired through the gyroscope;

[0088] S12. The inspection camera and the installed sensor group are integrated and connected with the cloud server by setting up a LoRa wireless gateway, and then the collected equipment operation data and video data are transmitted to the cloud server through the LoRa communication protocol.

[0089] In this embodiment, the method uses the added sensor group and the camera's built-in functions, and the system can collect multi-dimensional device operation data and video data in real time. These data not only cover the physical state of the device, but also cover the environment and network operation status, forming comprehensive monitoring information. Secondly, by utilizing the low-power wide-area communication capability of the LoRa wireless gateway, efficient and remote transmission of camera device data is achieved, ensuring real-time upload and processing of data. This solution deeply integrates the physical and network status information of the device through a combination of multiple sensors and real-time data collection, achieving more accurate device monitoring and status analysis.

[0090] Example 3

[0091] This embodiment is explained in Example 2. Please refer to Figure 1 , specifically: S2 includes S21, S22 and S23;

[0092] S21. Receive the collected device operation data and video data in real time in the cloud server, and process the device operation data, wherein the data processing includes data cleaning, data format standardization, and time synchronization and alignment;

[0093] Data cleaning is used to remove erroneous, missing, and unreasonable data points. For example, network interruption may result in missing data for some periods of time, or abnormal data caused by abnormal instantaneous fluctuations in equipment.

[0094] Data format standardization is used to convert device operation data from different sources into a unified format. For example, hardware sensor data and image analysis data may have different timestamp formats or unit standards.

[0095] Time synchronization and alignment are used to synchronize and align the inconsistent device operation data collection time caused by network delay Li;

[0096] S22. Based on the collected video data, video quality data is extracted from the video data by image processing technology, where the image processing technology includes a Sobel operator and a Gaussian blur algorithm for edge detection, and the video quality data includes a brightness value I(x, y) at the pixel point (x, y), a brightness gradient change ▽I(x, y) at the pixel point (x, y), a brightness value Iblur(x, y) of the pixel point (x, y) in a blurred video frame, and a brightness value Isharp(x, y) of the pixel point (x, y) in a clear video frame;

[0097] Using the Sobel operator of edge detection, the gradient of the image is estimated by calculating the brightness changes in the horizontal and vertical directions. After calculation, the gradient value of each pixel will reflect the changes in the brightness of its surroundings. A clear image will have a higher gradient value in the edge area.

[0098] Use the Gaussian blur algorithm to blur the image. Gaussian blur will reduce the details in the image and make the edges smooth, imitating the effect of dust covering the lens. By comparing the pixel brightness values ​​of the blurred image and the original image, the change in image clarity can be judged.

[0099] S23. By constructing a cloud server database, a historical database table and a real-time database table are constructed at the same time, and a write port and a write port are set. After executing step S21 and step S22, the acquired device operation data and video data at the same time point are horizontally integrated to obtain a real-time camera efficiency data set; the acquired device operation data and video data are vertically integrated to obtain a historical camera efficiency data set; and the real-time camera efficiency data set is stored in the real-time database table through the write port, and the historical camera efficiency data set is written into the historical database table.

[0100] In this embodiment, the method eliminates erroneous data and unifies the data format through data cleaning and format standardization, ensuring that data from different sources can be processed in a consistent manner, and solving the problems of incomplete data and errors caused by network interruption, time asynchrony, etc. Through time synchronization and alignment technology, the time difference between data sources is eliminated, ensuring the synchronization of equipment operation data and video data, thereby improving the accuracy of analysis. In terms of video quality analysis, the system uses image processing technology to accurately extract the edge and blur information of the video, helping to better evaluate the clarity and overall efficiency of the camera image. In addition, by fusing real-time and historical data, the system can simultaneously monitor the current operating status of the equipment, and use historical data for longitudinal analysis to help detect the changing trend of equipment performance and support long-term monitoring and fault prediction. This method significantly improves the accuracy of data processing and the intelligent level of monitoring. The system achieves a more comprehensive equipment status evaluation through the fusion of multi-dimensional data, and the longitudinal analysis capability based on historical data provides a solid foundation for the prediction of equipment performance trends, thereby effectively improving the fault prediction capability and overall stability of the system.

[0101] Example 4

[0102] This embodiment is explained in Example 3, please refer to Figure 1 , specifically: S3 includes S31, S32, S33 and S34;

[0103] S31, by constructing a video clarity algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) and the brightness gradient change ▽I(x, y) at the pixel point (x, y) through the write port of the cloud server database, inputting them into the video clarity algorithm formula, calculating and outputting the video blur Qi, and capturing the local blur phenomenon caused by focus offset;

[0104] The video blur Qi is calculated and output by the following video clarity algorithm formula;

[0105]

[0106] Where N represents the total number of pixels in each frame of the video, which is obtained by a known resolution, w(x, y) represents the pixel point (x, y), which is dynamically set according to the image edge strength by the Sobel operator of edge detection, and I max Indicates the upper limit of the brightness of the video frame, which is used to normalize the effect of brightness on the gradient.

[0107] S32, by constructing a picture jitter analysis algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) through the write port of the cloud server database and inputting it into the picture jitter analysis algorithm formula, calculating and outputting the picture jitter degree Sshake, and analyzing the jitter of the inspection camera in different time periods;

[0108] The image jitter degree Sshake is calculated and obtained by the following image jitter analysis algorithm formula;

[0109]

[0110] In the formula, T represents the total number of frames in the time window, t represents the time, and α t Represents the time weight, which represents the weight value at time t. It is specifically set by the property user. The time weight is set according to the application requirements. Usually, an exponential weight or a linear attenuation function is used to assign a larger weight to the most recent frame. I(x, y, t) represents the brightness value of the pixel point (x, y) at time t. I(x, y, t-1) represents the brightness value of the pixel point (x, y) at time t-1. It represents the brightness value of the pixel point in the previous time frame and is used to calculate the change between frames. V(x, y, t) represents the velocity vector of the pixel point (x, y) at time t. It is calculated by analyzing the pixel motion in continuous frames of the video using the optical flow method. V max Indicates the upper limit of the velocity vector in the video frame.

[0111] S33, by constructing a dust coverage analysis algorithm formula, extracting the brightness value Iblur(x, y) of the pixel point (x, y) in the blurred video frame and the brightness value Isharp(x, y) of the pixel point (x, y) in the clear video frame through the write port of the cloud server database, inputting them into the dust coverage analysis algorithm formula, calculating and outputting the dust coverage rate Ddust, and analyzing the dust coverage of the inspection camera;

[0112] The dust coverage rate Ddust is calculated and output by the following dust coverage analysis algorithm formula;

[0113]

[0114] Where Iblur(x, y, σ) represents the brightness value of the pixel (x, y) after Gaussian blur algorithm processing, σ represents the standard deviation of the blur kernel, and V(x, y) represents the motion vector of the pixel (x, y).

[0115] S34, constructing an angle drift analysis algorithm formula, extracting the camera angle Jd through the write port of the cloud server database and inputting it into the angle drift analysis algorithm formula, calculating and outputting the camera angle drift rate Adrift, and analyzing the speed of the inspection camera angle change and the camera's violent drift;

[0116] The camera angle drift rate Adrift is calculated and output by the following angle drift analysis algorithm formula;

[0117]

[0118] In the formula, Jd0 represents the initial installation angle of the inspection camera, d represents the differential operator, represents the rate of change of the camera angle Jd over time t, that is, the angular velocity, It represents the rate of change of angular velocity, that is, angular acceleration. γ and δ represent the preset weight values ​​of angular velocity and angular acceleration respectively. The specific values ​​are set by the user according to the application scenario, and γ+δ=1.

[0119] In this embodiment, the method accurately calculates the camera image blur Qi through the video clarity algorithm, captures the local blur caused by focus offset, and ensures continuous monitoring of image quality. Secondly, through the image jitter analysis algorithm, the system can dynamically detect the jitter degree Sshake of the image, analyze the stability of the camera in different time periods, and then identify the image instability caused by external vibration or loose equipment. Through the calculation of the dust coverage rate Ddust, the system can monitor the cleanliness of the camera lens surface in real time and warn of the situation where dust coverage affects the image quality. In addition, the angle drift analysis algorithm monitors the change speed and drastic deviation of the camera angle through the calculation of the angle drift rate Adrift, ensuring that the camera always remains at the predetermined monitoring angle. Compared with the prior art, this method not only enhances the self-diagnosis and feedback capabilities of the equipment through multi-dimensional analysis of video clarity, image jitter, dust coverage and angle drift, but also makes the fault prediction of the inspection camera more intelligent and automated.

[0120] Example 5

[0121] This embodiment is explained in Example 4. Please refer to Figure 1 , specifically: S4 includes S41 and S42;

[0122] S41, extracting the obtained video blur Qi, image shake degree Sshake, dust coverage rate Ddust and camera angle drift rate Adrift, and combining them with the equipment operation data, performing comprehensive calculation to obtain the comprehensive status score SPF of the inspection camera, and scoring the overall operation status of the camera;

[0123] The comprehensive status score SPF is calculated by the following algorithm formula;

[0124] SPF=a1·Qi+a2·Sshake+a3·Ddust+a4·Adrift+a5·Ti+a6·Li+a7·Elight+ε;

[0125] Wherein, a1, a2, a3, a4, a5, a6 and a7 represent the preset weight values ​​of video blur Qi, image shake Sshake, dust coverage Ddust, camera angle drift Adrift, device temperature Ti, network delay Li and ambient light change rate Elight, respectively. The specific values ​​are set by the user according to the application scenario, and a1+a2+a3+a4+a5+a6+a7=1, ε represents the comprehensive scoring error term;

[0126] S42, based on the historical data and performance standards of the inspection camera, a preset operation status threshold Y is set, and then a preliminary comparative evaluation is performed with the obtained comprehensive status score SPF to analyze the operation status of the current inspection camera. The specific evaluation contents are as follows;

[0127] When the comprehensive status score SPF> the operating status threshold Y, it means that the equipment is in normal operating state and does not need maintenance at present. The system is working within the normal range. At this time, step S5 is executed to perform fault prediction and generate a maintenance mechanism;

[0128] When the comprehensive status score SPF ≤ the operating status threshold Y, it means that the equipment is abnormal. At this time, the first warning information is generated to prompt the property to immediately check, clean and recalibrate the patrol camera.

[0129] In this embodiment, the method extracts the video blur Qi, the image shake degree Sshake, the dust coverage rate Ddust, and the camera angle drift rate Adrift, and combines the equipment operation data to generate the comprehensive status score SPF of the inspection camera, and compares it with the preset operation status threshold Y, which brings significant beneficial effects. The method comprehensively evaluates the overall operation status of the camera through the comprehensive calculation of multi-dimensional data, and accurately obtains the current health level of the equipment. Through the dynamic setting of the weight value, the system can flexibly adjust the importance of each parameter according to different scenarios, making the scoring more targeted and practical. In addition, the system sets the operation status threshold Y through historical data and performance standards, and realizes the automatic evaluation of the camera operation status. When the comprehensive status score SPF is lower than the operation status threshold Y, the system automatically generates an early warning message, indicating that the equipment may be abnormal, thereby triggering the maintenance mechanism. Compared with the traditional inspection method, the system significantly improves the accuracy and response speed of the evaluation. Traditional monitoring and maintenance usually rely on manual judgment, and it is difficult to find potential problems in time. However, this method can identify problems in advance and automatically warn by scoring the comprehensive status of the equipment, reducing the workload and errors of manual inspection. The system's automated early warning mechanism not only improves the efficiency of equipment maintenance, but also effectively avoids further expansion of equipment failures.

[0130] Example 6

[0131] This embodiment is explained in Example 5. Please refer to Figure 1 , specifically: S5 includes S51 and S52;

[0132] S51. When the preliminary comparison and evaluation shows that the equipment is in normal operation, the equipment fault prediction is triggered. By constructing a fault prediction algorithm formula and extracting the comprehensive status score SPF combined with the equipment temperature Ti and network delay Li of the inspection camera, the camera fault probability Pf is calculated and output, and the fault occurrence of the inspection camera is predicted;

[0133] The camera failure probability Pf is calculated using the following fault prediction algorithm formula:

[0134]

[0135] Where SPF t-1 represents the comprehensive status score at the previous time point t-1, κ represents the prediction error term, b1, b2 and b3 represent the preset weight values ​​of the comprehensive status score change rate, device temperature Ti and network delay Li respectively. The specific values ​​are set by the user according to the application scenario, and b1+b2+b3=1.

[0136] S52, preset a fault prediction threshold G by the frequency of inspection camera faults, and then perform a secondary comparative evaluation with the obtained camera fault probability Pf to predict the fault condition of the inspection camera, and generate warning information based on the evaluation results. The specific evaluation content is as follows;

[0137] When the probability of camera failure Pf>fault prediction threshold G, it indicates that the inspection camera has a predicted failure. At this time, the second warning information is generated to prompt the property to carry out planned maintenance;

[0138] When the camera failure probability Pf ≤ the failure prediction threshold G, it means that there is no predicted failure for the predicted inspection camera. At this time, maintenance is performed according to the original maintenance plan.

[0139] In this embodiment, the method calculates the probability of camera failure Pf by using the comprehensive status score SPF, device temperature Ti, network delay Li and other key parameters in the fault prediction algorithm when the system evaluates that the device is operating normally. This algorithm can dynamically predict the possible future failure of the camera and help identify potential problems in advance. Secondly, by setting the fault prediction threshold G, the system can perform a secondary evaluation of the probability of camera failure Pf and generate specific warning information based on the evaluation results. When the probability of camera failure Pf exceeds the fault prediction threshold G, the system will remind the property to take planned maintenance measures; when the probability of camera failure Pf is lower than the fault prediction threshold G, the system allows the device to continue to operate according to the original maintenance plan. This fault prediction mechanism greatly improves the intelligent level of monitoring equipment management. By predicting the possible failure of the equipment in advance, the system can actively generate warning information, avoiding the limitation of traditional technology that relies on remedial measures after the equipment fails. This predictive maintenance not only reduces the risk of sudden equipment failure and extends the service life of the equipment, but also effectively reduces maintenance costs and downtime.

[0140] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A device monitoring method based on big data, characterized in that: The following steps are involved: S1. Through the sensor group and built-in functions installed on the property inspection camera, the camera equipment operation data and video data are collected in real time, and the collected camera equipment operation data and video data are transmitted to the cloud server through the wireless gateway; S2. receiving the collected camera device operation data and video data in real time in the cloud server, performing data processing on the camera device operation data, extracting video quality data from the video data by image processing technology, fusing the camera device operation data and the video quality data to obtain a camera efficiency data set, and storing the camera efficiency data set; S3. Use big data analysis to build relevant algorithm formulas, extract camera performance data sets and input them into relevant algorithm formulas to calculate and output video blur Qi, image shake Sshake, dust coverage Ddust and camera angle drift Adrift; S4, based on the obtained video blur Qi, image jitter Sshake, dust coverage Ddust and camera angle drift rate Adrift, and combined with the camera performance data set, a comprehensive calculation is performed to obtain the comprehensive status score SPF of the inspection camera, and a preset operation status threshold Y is used to perform a preliminary comparative evaluation and analysis of the health status of the camera with the obtained comprehensive status score SPF; S5. Based on the preliminary comparative evaluation results, the fault prediction and early warning mechanism is triggered. By constructing a fault prediction algorithm formula and combining the obtained comprehensive status score SPF into the fault prediction algorithm formula, the camera failure probability Pf is calculated and output. The preset fault prediction threshold G is compared with the camera failure probability Pf for a second comparative evaluation, and early warning information is generated based on the evaluation results.

2. The device monitoring method based on big data according to claim 1, characterized in that: Said S1 includes S11 and S12; S11. By adding sensor groups and built-in functions to property inspection cameras, real-time data on camera equipment operation is collected, and real-time video data is collected through the built-in functions of the cameras; The sensor group includes an ambient temperature sensor, a light sensor and a gyroscope; The device operation data includes device temperature Ti, network delay Li, ambient light change rate Elight and camera angle Jd; S12. The inspection camera and the installed sensor group are integrated and connected with the cloud server by setting up a LoRa wireless gateway, and then the collected equipment operation data and video data are transmitted to the cloud server through the LoRa communication protocol.

3. The device monitoring method based on big data according to claim 1 is characterized in that: The S2 includes S21, S22 and S23; S21, receiving the collected device operation data and video data in real time in the cloud server, and performing data processing on the device operation data, wherein the data processing includes data cleaning, data format standardization, and time synchronization and alignment; The data cleaning is used to remove erroneous, missing and unreasonable data points; The data format standardization is used to convert equipment operation data from different sources into a unified format; The time synchronization and alignment is used to synchronize and align the inconsistent device operation data collection time caused by the network delay Li; S22, based on the collected video data, extracting video quality data from the video data by image processing technology, wherein the image processing technology includes a Sobel operator and a Gaussian blur algorithm for edge detection, and the video quality data includes a brightness value I(x, y) at a pixel point (x, y), a brightness gradient change ▽I(x, y) at a pixel point (x, y), a brightness value Iblur(x, y) of a pixel point (x, y) in a blurred video frame, and a brightness value Isharp(x, y) of a pixel point (x, y) in a clear video frame; S23, by building a cloud server database, building a historical database table and a real-time database table at the same time, and setting a write port and a write port, and then performing step S21 and step S22, horizontally integrating the acquired device operation data and video data at the same time point to obtain a real-time camera performance data set; Then, the acquired equipment operation data and video data are vertically integrated to obtain a historical camera performance data set; The real-time camera performance data set is stored in the real-time database table through the write port, and the historical camera performance data set is written into the historical database table.

4. The device monitoring method based on big data according to claim 3 is characterized in that: The S3 includes S31, S32, S33 and S34; S31, by constructing a video clarity algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) and the brightness gradient change ▽I(x, y) at the pixel point (x, y) through the write port of the cloud server database, inputting them into the video clarity algorithm formula, calculating and outputting the video blur Qi, and capturing the local blur phenomenon caused by focus offset; The video blur Qi is calculated and output by the following video clarity algorithm formula; Where N represents the total number of pixels in each frame of the video, w(x, y) represents the pixel point (x, y), and the Sobel operator for edge detection is dynamically set according to the image edge strength. max Indicates the upper limit of the brightness of the video frame, which is used to normalize the effect of brightness on the gradient.

5. The device monitoring method based on big data according to claim 4 is characterized in that: S32, by constructing a picture jitter analysis algorithm formula, extracting the brightness value I(x, y) at the pixel point (x, y) through the write port of the cloud server database and inputting it into the picture jitter analysis algorithm formula, calculating and outputting the picture jitter degree Sshake, and analyzing the jitter of the inspection camera in different time periods; The image jitter degree Sshake is calculated and obtained by the following image jitter analysis algorithm formula; In the formula, T represents the total number of frames in the time window, t represents the time, and α t Represents the time weight, which represents the weight value at time t, which is set by the property user. I(x, y, t) represents the brightness value of the pixel point (x, y) at time t, I(x, y, t-1) represents the brightness value of the pixel point (x, y) at time t-1, V(x, y, t) represents the velocity vector of the pixel point (x, y) at time t, V max Indicates the upper limit of the velocity vector in the video frame.

6. The device monitoring method based on big data according to claim 4 is characterized in that: S33, by constructing a dust coverage analysis algorithm formula, extracting the brightness value Iblur(x, y) of the pixel point (x, y) in the blurred video frame and the brightness value Isharp(x, y) of the pixel point (x, y) in the clear video frame through the write port of the cloud server database, inputting them into the dust coverage analysis algorithm formula, calculating and outputting the dust coverage rate Ddust, and analyzing the dust coverage of the inspection camera; The dust coverage rate Ddust is calculated and outputted by the following dust coverage analysis algorithm formula; Where Iblur(x, y, σ) represents the brightness value of the pixel (x, y) after Gaussian blur algorithm processing, σ represents the standard deviation of the blur kernel, and V(x, y) represents the motion vector of the pixel (x, y).

7. The device monitoring method based on big data according to claim 4 is characterized in that: S34, constructing an angle drift analysis algorithm formula, extracting the camera angle Jd through the write port of the cloud server database and inputting it into the angle drift analysis algorithm formula, calculating and outputting the camera angle drift rate Adrift, and analyzing the speed of the inspection camera angle change and the camera's violent drift; The camera angle drift rate Adrift is calculated and output by the following angle drift analysis algorithm formula; In the formula, Jd0 represents the initial installation angle of the inspection camera, d represents the differential operator, represents the rate of change of the camera angle Jd over time t, that is, the angular velocity, It represents the rate of change of angular velocity, that is, angular acceleration. γ and δ represent the preset weight values ​​of angular velocity and angular acceleration respectively. The specific values ​​are set by the user according to the application scenario, and γ+δ=1.

8. The device monitoring method based on big data according to claim 7, characterized in that: The S4 includes S41 and S42; S41, extracting the obtained video blur Qi, image shake degree Sshake, dust coverage rate Ddust and camera angle drift rate Adrift, and combining them with the equipment operation data, performing comprehensive calculation to obtain the comprehensive status score SPF of the inspection camera, and scoring the overall operation status of the camera; The comprehensive status score SPF is calculated by the following algorithm formula: SPF=a1·Qi+a2·Sshake+a3·Ddust+a4·Adrift+a5·Ti+a6·Li+a7·Elight+ε; Wherein, a1, a2, a3, a4, a5, a6 and a7 represent the preset weight values ​​of video blur Qi, image shake Sshake, dust coverage Ddust, camera angle drift Adrift, device temperature Ti, network delay Li and ambient light change rate Elight, respectively. The specific values ​​are set by the user according to the application scenario, and a1+a2+a3+a4+a5+a6+a7=1, ε represents the comprehensive scoring error term; S42, based on the historical data and performance standards of the inspection camera, a preset operation status threshold Y is set, and then a preliminary comparative evaluation is performed with the obtained comprehensive status score SPF to analyze the operation status of the current inspection camera. The specific evaluation contents are as follows; When the comprehensive status score SPF> the operating status threshold Y, it means that the equipment is in normal operating state and does not need maintenance at present. The system is working within the normal range. At this time, step S5 is executed to perform fault prediction and generate a maintenance mechanism; When the comprehensive status score SPF ≤ the operating status threshold Y, it means that the equipment is abnormal. At this time, the first warning information is generated to prompt the property to immediately check, clean and recalibrate the patrol camera.

9. The device monitoring method based on big data according to claim 8, characterized in that: The S5 includes S51 and S52; S51. When the preliminary comparison and evaluation shows that the equipment is in normal operation, the equipment fault prediction is triggered. By constructing a fault prediction algorithm formula and extracting the comprehensive status score SPF combined with the equipment temperature Ti and network delay Li of the inspection camera, the camera fault probability Pf is calculated and output, and the fault occurrence of the inspection camera is predicted; The camera failure probability Pf is calculated and obtained by the following failure prediction algorithm formula; In the formula, represents the rate of change of the comprehensive status score, SPFt-1 represents the comprehensive status score at the previous time point t-1, κ represents the prediction error term, b1, b2 and b3 represent the preset weight values ​​of the comprehensive status score change rate, device temperature Ti and network delay Li respectively. The specific values ​​are set by the user according to the application scenario, and b1+b2+b3=1.

10. The device monitoring method based on big data according to claim 9, characterized in that: S52, preset a fault prediction threshold G by the frequency of inspection camera faults, and then perform a secondary comparative evaluation with the obtained camera fault probability Pf to predict the fault condition of the inspection camera, and generate warning information based on the evaluation results. The specific evaluation content is as follows; When the probability of camera failure Pf>fault prediction threshold G, it indicates that the inspection camera has a predicted failure. At this time, the second warning information is generated to prompt the property to carry out planned maintenance; When the camera failure probability Pf ≤ the failure prediction threshold G, it means that there is no predicted failure for the predicted inspection camera. At this time, maintenance is performed according to the original maintenance plan.

Images