Low-altitude meteorological observation equipment operation and maintenance management method and system based on Internet of Things

Abstract

The invention relates to the field of meteorological observation management, and discloses a low-altitude meteorological observation equipment operation and maintenance management method and system based on the Internet of Things, and the system comprises a meteorological sensor array module which is used for collecting low-altitude atmospheric environment parameters as observation data, including temperature, humidity, air pressure, wind speed, wind direction and particle concentration; the Internet of Things communication module is used for transmitting observation data to the edge computing node module in real time; the edge computing node module is used for performing local preprocessing on original observation data, including data filtering, abnormal value elimination and compression; the cloud operation and maintenance management platform is used for storing and analyzing the preprocessed observation data, judging an equipment state according to an analysis result, and generating an operation and maintenance instruction according to the equipment state; and the control module is used for receiving the alarm information and remotely controlling start and stop of equipment. According to the technical scheme, the safety of the low-altitude meteorological observation equipment can be improved.

Description

Technical Field

[0001] The present invention relates to the field of meteorological observation management, and particularly to an operation and maintenance management method and system for low-altitude meteorological observation equipment based on the Internet of Things. Background Art

[0002] In the field of meteorological observation, low-altitude meteorological observation is of crucial significance for urban meteorological services, aviation safety guarantee, environmental monitoring, etc. With the rapid development of Internet of Things technology, low-altitude meteorological observation equipment based on the Internet of Things has been widely used. These devices can collect various parameters in the low-altitude atmospheric environment in real time, such as temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration.

[0003] However, in the actual operation process, low-altitude meteorological observation equipment faces many operation and maintenance management problems. On the one hand, the number of devices is large and they are widely distributed. The traditional manual inspection method is inefficient and costly, and it is difficult to meet the real-time and comprehensive operation and maintenance requirements.

[0004] On the other hand, the meteorological observation environment is complex and changeable, and the equipment is easily affected by natural environmental factors such as extreme weather, electromagnetic interference, and its own aging, resulting in data anomalies. For example, in strong wind weather, the wind speed sensor may have measurement deviations; after long-term use, the battery power decline may cause the equipment to work intermittently, affecting the continuity of data.

[0005] In addition, existing operation and maintenance management systems often lack in-depth analysis and effective utilization of observation data. Although a large amount of data can be collected, it is impossible to evaluate the data quality in a timely and accurate manner, difficult to accurately judge the equipment status, and unable to provide a scientific basis for operation and maintenance decisions. This makes the equipment maintenance work have a certain blindness, and there may be situations of over-maintenance or untimely maintenance, which not only wastes resources but also affects the reliability of observation data. Summary of the Invention

[0006] The purpose of the present invention is to provide an operation and maintenance management method and system for low-altitude meteorological observation equipment based on the Internet of Things to solve the above technical problems.

[0007] The purpose of the present invention can be achieved through the following technical solutions: An operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things, comprising: A meteorological sensor array module for collecting low-altitude atmospheric environment parameters as observation data, including temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration; An Internet of Things communication module for real-time transmitting the observation data to the edge computing node module; An edge computing node module for locally preprocessing the original observation data, including data filtering, outlier removal, and compression; A cloud operation and maintenance management platform is used to store and analyze the pre - processed observation data, judge the device status according to the analysis results, and generate operation and maintenance instructions according to the device status. A control module is used to receive alarm information and remotely control the start and stop of the device.

[0008] As a further technical solution, the cloud operation and maintenance management platform includes: A data quality assessment module is used to evaluate the quality of the observation data based on an abnormal data analysis model. If there are quality problems, the abnormal data is marked. A device health score module is used to predict the maintenance cycle according to the historical failure rate, signal stability and battery life.

[0009] As a further technical solution, the abnormal data analysis model includes: A dynamic threshold detection unit is used to calculate the dynamic threshold range of each meteorological parameter in real - time. When the observation data exceeds the threshold range, it is marked as potentially abnormal. A time - series prediction unit is used to predict each meteorological parameter through a pre - trained time - series prediction model to obtain a predicted value, and compare the actual observed value with the predicted value. When the deviation exceeds the preset tolerance, it is marked as a time - series anomaly; the time - series prediction model is a long short - term memory (LSTM) neural network. A spatial consistency detection unit is used to calculate the spatial correlation between the observation data of the current device and the observation data of neighboring devices when observing in a multi - device network. When the Mahalanobis distance exceeds the set threshold, it is marked as a spatial anomaly. A comprehensive decision - making unit is used to perform weighted fusion on the output results of the dynamic threshold detection unit, the time - series prediction unit and the spatial consistency detection unit. When the following conditions are met, it is determined as the final abnormal data: ; Wherein, , , , , are the dynamic, time - series, and spatial anomaly indication values respectively. The anomaly indication value takes the value of 0 when judged to be normal and 1 when judged to be abnormal. is the preset judgment threshold. is the comprehensive anomaly indication judgment function. When , it means that at time t, the current observed data is determined to be abnormal data; when , it means that at time t, the current observed data is determined to be normal data.

[0010] As a further technical solution, the process of obtaining the dynamic anomaly indication value is as follows: ; wherein, is the observed value at time t, is the average value obtained by backtracking a set period from time t, is the standard deviation value obtained by backtracking a set period from time t, is an adjustable sensitivity coefficient, determined based on historical data analysis.

[0011] As a further technical solution, the acquisition process of the time series anomaly indication value is as follows:

[0012] wherein, is the predicted value, is the preset error tolerance.

[0013] As a further technical solution, the acquisition process of the spatial anomaly indication value is as follows:

[0014] wherein, is the preset threshold, is the Mahalanobis distance between the current observed value and the data of neighboring sensors, The calculation formula of is: is the observed mean of neighboring devices, is the spatial correlation coefficient matrix between the current device and the data of neighboring devices, represents the transpose.

[0015] As a further technical solution, the spatial correlation coefficient matrix is generated by a graph neural network, including: Input layer: Receiving the observed data and geographical coordinates of each device; Graph convolutional layer: Aggregating the features of neighboring devices through a message passing mechanism; Output layer: Generating the spatial correlation coefficient matrix , and satisfying ; represents the correlation coefficient between the th atmospheric environment parameter and the th atmospheric environment parameter; The training loss function of the graph neural network is: ; wherein, is the benchmark matrix calculated from historical data, , are weight factors, determined based on historical data analysis, Represents a matrix The sum of the elements on the main diagonal, Represents taking the natural logarithm of the determinant of the spatial correlation coefficient matrix .

[0016] As a further technical solution, the process of predicting the maintenance cycle based on historical failure rate, signal stability, and battery life is as follows: Through the formula: The predicted maintenance cycle is calculated ; where, is the standard maintenance cycle, is the equipment health score; the specific calculation formula is: ; where, , , are the weight coefficients, is the historical failure rate, calculated by the ratio of the number of failures of the current device in the past period to the total operating time, is the signal stability, obtained by calculating the standard deviation of the signal strength of the current device over a period of time, is the battery life, obtained by calculating the ratio of the remaining battery power of the current device to the total battery capacity.

[0017] An operation and maintenance management method for low-altitude meteorological observation equipment based on the Internet of Things, including the following steps: S1. Data collection: The meteorological sensor array module continuously collects low-altitude atmospheric environment parameters, including temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration, to form original observation data; S2. Data transmission: The Internet of Things communication module transmits the original observation data collected by the meteorological sensor array module to the edge computing node module in real time; S3. Local preprocessing: The edge computing node module performs local preprocessing on the received original observation data, removes noise interference through data filtering, uses an outlier rejection algorithm to remove obvious errors or abnormal data points, and at the same time compresses the data using a compression algorithm to reduce the data volume; S4. Data upload and storage: The preprocessed data is uploaded to the cloud operation and maintenance management platform through the network, and the cloud operation and maintenance management platform stores the data in the database; S5. Data quality assessment and anomaly detection: Use an anomaly data analysis model to evaluate the quality of the observation data; S6. Device status judgment and operation and maintenance instruction generation: Based on the data quality assessment results and the maintenance cycle prediction given by the device health score module, judge the device status; if the device has anomalies or is about to reach the maintenance cycle, generate corresponding operation and maintenance instructions, which include device inspection, component replacement, or parameter adjustment. S7. Instruction execution: The control module receives the alarm information and operation and maintenance instructions sent by the cloud operation and maintenance management platform, and remotely controls the start and stop of the device according to the instructions, or notifies the operation and maintenance personnel to perform corresponding device maintenance operations.

[0018] Advantages of the present invention: (1) Through the collaborative work of the meteorological sensor array module, the edge computing node module, and the cloud operation and maintenance management platform, the present invention can perform multi-level processing and analysis of the observed data; the local preprocessing of the edge computing node module can effectively remove noise, eliminate outliers, and compress data, reducing the data transmission burden while improving the data quality. The anomaly data analysis model in the cloud operation and maintenance management platform can accurately identify abnormal data by using various means such as dynamic threshold detection, time series prediction, and spatial consistency detection, mark potential anomalies, time series anomalies, and spatial anomalies, ensuring the authenticity and reliability of the observed data stored and used, and providing high-quality data support for meteorological research, forecasting, and other work; (2) The device health score module based on historical failure rate, signal stability, and battery life can accurately predict the maintenance cycle of the device, avoiding the problems of over-maintenance or untimely maintenance in the traditional operation and maintenance methods. The operation and maintenance personnel can arrange device inspection, component replacement, etc. in advance according to the predicted maintenance cycle, improving the operation stability and reliability of the device, extending the service life of the device, and reducing the operation and maintenance cost; (3) In view of the complex and changeable characteristics of the meteorological observation environment, through the spatial consistency detection unit, when multiple devices are networked for observation, the present invention can timely detect anomalies in the device observation data caused by environmental factors, ensuring that accurate meteorological data can still be obtained in a complex environment, achieving the purpose of real-time monitoring of the operation status of the device and changes in environmental parameters; for the problem that devices in different positions in mountainous areas and other terrain complex areas may be affected by the terrain, resulting in large differences in meteorological parameters, it can effectively identify normal spatial differences and abnormal data, ensuring the validity of the data. Description of the drawings

[0019] The present invention will be further described below with reference to the drawings.

[0020] Figure 1 It is the system logic block diagram of the present invention. Detailed implementation manners

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0022] Please refer to Figure 1 As shown, the present invention is a low-altitude meteorological observation equipment operation and maintenance management system based on the Internet of Things, including: A meteorological sensor array module for collecting low-altitude atmospheric environment parameters as observation data, including temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration; it consists of multiple meteorological sensors distributed at different geographical locations, and the meteorological sensors collect data according to a preset sampling frequency, and the sampling frequency can be adjusted according to actual needs; An Internet of Things communication module for real-time transmitting the observation data to the edge computing node module; it is connected to the meteorological sensor array module through wired or wireless communication methods to receive the collected meteorological observation data; An edge computing node module for locally preprocessing the original observation data, including data filtering, outlier removal, and compression; the preprocessing can adopt existing technologies and will not be elaborated here; A cloud operation and maintenance management platform for storing and analyzing the preprocessed observation data, judging the device status according to the analysis results, and generating operation and maintenance instructions according to the device status; if the data is abnormal and it is judged that it is caused by a device failure, a device repair instruction is generated; if the device health score is low and close to the maintenance cycle, a device maintenance instruction is generated; A control module for receiving alarm information and remotely controlling the start and stop of the device.

[0023] The cloud operation and maintenance management platform includes: A data quality assessment module for assessing the quality of the observation data based on an abnormal data analysis model, and marking the abnormal data if there are quality problems; A device health score module for predicting the maintenance cycle according to the historical failure rate, signal stability, and battery life. The process of predicting the maintenance cycle according to the historical failure rate, signal stability, and battery life is as follows: Through the formula: The predicted maintenance cycle is calculated ; Wherein, Is the standard maintenance cycle, Is the device health score; the specific calculation formula is: ; Wherein, , , is the weight coefficient, is the historical failure rate, which is calculated by the ratio of the number of failures of the current device in the past period to the total operating time. is the signal stability, which is obtained by calculating the standard deviation of the signal strength of the current device over a period of time. is the battery life, which is obtained by calculating the ratio of the remaining battery power of the current device to the total battery capacity. When the device health score is low, it indicates that the device status is poor and the maintenance cycle needs to be shortened; when the device health score is high, it indicates that the device status is good and the maintenance cycle can be appropriately extended.

[0024] The abnormal data analysis model includes: A dynamic threshold detection unit, which is used to calculate the dynamic threshold range of each meteorological parameter in real time and mark it as a potential anomaly when the observed data exceeds the threshold range; A time series prediction unit, which predicts each meteorological parameter through a pre-trained time series prediction model to obtain a predicted value, compares the actual observed value with the predicted value, and marks it as a time series anomaly when the deviation exceeds the preset tolerance; the time series prediction model is a long short-term memory (LSTM) neural network; the LSTM model is a recurrent neural network that can handle long-term dependencies in sequence data and is suitable for predicting meteorological data; the LSTM model is trained using historical meteorological observation data. After training, the meteorological data before the current moment is input into the model to predict the meteorological parameter value at the next moment; the predicted value is compared with the actual observed value, and when the deviation exceeds the preset error tolerance, the data is marked as time series anomaly data; the error tolerance can be adjusted according to the characteristics of the meteorological parameter and actual requirements, and is generally set to 5%-10% of the predicted value.

[0025] A spatial consistency detection unit, which is used to calculate the spatial correlation between the observed data of the current device and the observed data of neighboring devices when multiple devices are networked for observation, and marks it as a spatial anomaly when the Mahalanobis distance exceeds the set threshold; the Mahalanobis distance takes into account the covariance structure of the data and can more accurately measure the similarity between data; A comprehensive decision-making unit, which is used to perform weighted fusion on the output results of the dynamic threshold detection unit, the time series prediction unit, and the spatial consistency detection unit, and determines it as the final abnormal data when the following conditions are met: ; where, , , , , are the dynamic, time series, and spatial anomaly indication values respectively. The anomaly indication value takes the value of 0 when it is determined to be normal and 1 when it is determined to be abnormal. is a preset judgment threshold value, is a comprehensive anomaly indication judgment function. When , it indicates that at time t, the current observed data is determined to be abnormal data; when , it indicates that at time t, the current observed data is determined to be normal data.

[0026] In this embodiment, through the high-precision data acquisition of the meteorological sensor array module, the local preprocessing of the edge computing node module, and the anomaly data analysis model of the cloud operation and maintenance management platform, it is possible to effectively remove noise, eliminate outliers, accurately mark potential anomaly data, time-series anomaly data, and spatial anomaly data, ensure the accuracy and reliability of meteorological observation data, and provide high-quality data support for meteorological research, forecasting, and other work; at the same time, the device health score module based on the historical failure rate, signal stability, and battery life of the device can accurately predict the maintenance cycle of the device, making the operation and maintenance work more scientific and reasonable. It avoids the problems of over-maintenance and untimely maintenance, improves the operation stability and reliability of the device, extends the service life of the device, and reduces the operation and maintenance cost.

[0027] The process of obtaining the dynamic anomaly indication value is as follows: ; where is the observed value at time t, which is used to reflect the actual measurement situation of the current meteorological parameter, is the average value obtained by looking back a set time period from time t, which represents the average level of the meteorological parameter within a certain historical period and can reflect the normal change range of this parameter, is the standard deviation value obtained by looking back a set time period from time t, which is used to measure the dispersion degree of historical data, that is, the fluctuation situation of the data, is an adjustable sensitivity coefficient, which is determined based on historical data analysis and can adjust the judgment sensitivity to data anomalies according to the characteristics of different meteorological parameters and actual requirements.

[0028] In this embodiment, the retrospective time period can be set according to the change characteristics of meteorological parameters and the sampling frequency of the device. For example, for meteorological parameters such as temperature and humidity that change relatively slowly, and the device sampling frequency is once per minute, the retrospective time period can be set to the past 1 hour, that is, 60 sampling points; for parameters such as wind speed and wind direction that change more frequently, the retrospective time period can be shortened to the past 15 minutes, that is, 15 sampling points; when , mark the observed data at this moment as dynamically abnormal; when , it indicates that the observed data at this moment is normal in the dynamic range; By combining the average value, standard deviation of historical data, and adjustable sensitivity coefficients, the normal range of observed values can be dynamically determined at each moment. Compared with the detection method with a fixed threshold, this dynamic threshold detection method can better adapt to the characteristics of meteorological parameters changing with factors such as time, season, and geographical location. For example, in different seasons, the normal fluctuation range of temperature is different, and the dynamic threshold detection can automatically adjust the threshold according to historical data to more accurately identify abnormal data. Since in low-altitude meteorological observations, the meteorological environment is complex and changeable, sensors may be interfered by various factors, such as local microclimate, equipment failures, etc. Through this dynamic anomaly detection mechanism, data points deviating from the normal fluctuation range in the observed data can be detected in a timely manner, providing an important basis for subsequent data processing and equipment fault diagnosis.

[0029] The process of obtaining the time-series anomaly indication value is as follows:

[0030] Among them, is the predicted value, which reflects the expectation of the meteorological parameter at the current moment based on the trend of historical data, is the preset error tolerance, which is determined according to the actual application requirements of the meteorological parameter and the prediction error situation of historical data, and is used to measure the acceptable deviation range between the actual observed value and the predicted value.

[0031] In this embodiment, a number of historical observed values before the current moment t are used as inputs and input into the trained time-series prediction model. The length of the input data is determined according to the change law of the meteorological parameter and the training effect of the model. For example, for the temperature parameter, 1440 sampling points in the past 24 hours, assuming the sampling frequency is once per minute, can be selected as the input, and the model outputs the predicted value at time t , when , mark the observed data at this moment as a time-series anomaly; when , it means that the observed data at this moment is normal in terms of time series; By using the time-series prediction model to learn and predict the change trend of meteorological parameters, and by comparing the actual observed value with the predicted value, time-series anomalies in the observed data can be found. Meteorological parameters usually have certain time-series characteristics, such as periodic changes (diurnal changes, seasonal changes, etc.) and trend changes. The time-series prediction model can capture these characteristics. When the deviation between the actual observed value and the predicted value is too large, it indicates that there may be abnormal situations, such as sensor failures, sudden meteorological events, etc. In meteorological observations, continuous and accurate time series data is crucial for analyzing meteorological change patterns. By promptly detecting abnormal time series data, meteorological analysis errors caused by such abnormal data can be avoided, and the accuracy and reliability of meteorological forecasts can be improved. For example, in weather forecasting models, accurate input of time series data can enhance the prediction accuracy of the model for weather change trends, providing more precise meteorological services to the public. At the same time, it also helps operation and maintenance personnel promptly discover problems in the acquisition of time series data by equipment, such as sensor drift and data transmission interruption, so as to carry out equipment maintenance and repair in a timely manner and ensure the normal operation of observation equipment.

[0032] The process of obtaining the spatial anomaly indication value is as follows:

[0033] Among them, is a preset threshold, is the Mahalanobis distance between the current observation value and the data of neighboring sensors, The calculation formula of is: is the mean observation of neighboring devices, which reflects the average meteorological conditions measured by surrounding devices and is used to compare with the observation value of the current device. is the spatial correlation coefficient matrix between the current device and the data of neighboring devices, which describes the spatial correlation between the observation data of different devices and reflects the connection of meteorological parameters in spatial distribution. represents the transpose, which is used to meet the matrix operation rules. The selection of neighboring devices needs to meet both: geographical distance ≤ 20 km; altitude difference ≤ 100 m; data update time difference ≤ 5 minutes.

[0034] Accurate C can make the Mahalanobis distance more accurately measure the spatial difference between the current device and the observation data of neighboring devices. For example, in an urban environment, meteorological observation devices in different regions are affected by buildings, terrain, etc. C can reflect the correlation of device data under these complex factors, avoid misjudging normal spatial differences as anomalies, improve the accuracy of spatial anomaly detection, and ensure the reliability of meteorological data.

[0035] The spatial correlation coefficient matrix is generated by a graph neural network and includes: Input layer: Receives the observation data and geographical coordinates of each device; Graph convolutional layer: Aggregates the features of neighboring devices through a message passing mechanism; Output layer: Generates the spatial correlation coefficient matrix , and satisfies ; represents the th atmospheric environment parameter and the The correlation coefficients between atmospheric environment parameters; Through the above technical solution, the spatial correlation of meteorological parameters between different devices can be effectively captured, and the connection of each meteorological parameter in spatial distribution can be accurately described; For example, in mountain meteorological observations, the correlation of parameters such as temperature and wind speed measured by devices at different altitudes can be clearly presented, providing strong support for analyzing local meteorological changes.

[0036] The training loss function of the graph neural network is: ; where is the reference matrix calculated from historical data, which provides a reference standard for training, enabling the generated C to conform to historical laws, , are weight factors, determined based on historical data analysis, and are used to adjust the importance of each part in the loss function, represents the sum of the main diagonal elements of matrix Constraining it (controlled by ) can affect certain properties of the matrix, such as the stability of the matrix, represents taking the natural logarithm of the determinant of the spatial correlation coefficient matrix . By reasonably setting these parameters, the generated spatial correlation coefficient matrix can better conform to the spatial correlation characteristics of actual meteorological data, improving the performance of the model.

[0037] In this embodiment, according to the geographical coordinate information of each meteorological sensor, devices with a geographical distance ≤ 20 km from the current device are screened out. Using the altitude data of the devices, devices with an altitude difference ≤ 100 m are further screened out from the above devices. At the same time, by comparing the data update timestamps of each device, devices with a data update time difference ≤ 5 minutes are ensured to be selected as neighboring devices; When observing with a multi-device network, considering that meteorological parameters have a certain continuity and correlation in space, the Mahalanobis distance is calculated to measure the spatial difference between the observed data of the current device and the observed data of neighboring devices; The Mahalanobis distance not only considers the mean difference of the data but also the covariance structure of the data, that is, the correlation between different meteorological parameters, and can more comprehensively reflect the spatial distribution characteristics of the data; In the low-altitude meteorological observation network, the observed data of each device should have a certain consistency in space; When the observed data of a certain device shows spatial anomalies, it may mean that the device has a fault, such as sensor damage, device calibration error, etc., or a special meteorological phenomenon has occurred in this area; Through spatial anomaly detection, devices that may have problems can be quickly located, providing a direction for device fault troubleshooting, and at the same time, it is also helpful to discover some special meteorological events, such as local microclimate anomalies, etc.; This is of great significance for ensuring the overall accuracy of the meteorological observation network, timely discovering device problems, and improving the refinement level of meteorological services.

[0038] A method for operation and maintenance management of low-altitude meteorological observation equipment based on the Internet of Things, comprising the following steps: S1. Data collection: The meteorological sensor array module continuously collects low-altitude atmospheric environment parameters, including temperature, humidity, air pressure, wind speed, wind direction and particulate matter concentration, to form original observation data; S2. Data transmission: The Internet of Things communication module transmits the original observation data collected by the meteorological sensor array module to the edge computing node module in real time; S3. Local preprocessing: The edge computing node module performs local preprocessing on the received original observation data, removes noise interference through data filtering, uses an outlier rejection algorithm to remove obvious error or abnormal data points, and at the same time compresses the data using a compression algorithm to reduce the data volume; S4. Data upload and storage: The preprocessed data is uploaded to the cloud operation and maintenance management platform through the network, and the cloud operation and maintenance management platform stores the data in the database; S5. Data quality evaluation and anomaly detection: Use an anomaly data analysis model to evaluate the quality of the observation data; S6. Equipment status judgment and operation and maintenance instruction generation: Judge the equipment status according to the data quality evaluation result and the maintenance cycle prediction given by the equipment health score module; if the equipment has an anomaly or is about to reach the maintenance cycle, generate corresponding operation and maintenance instructions, and the operation and maintenance instructions include equipment inspection, component replacement or parameter adjustment; S7. Instruction execution: The control module receives the alarm information and operation and maintenance instructions sent by the cloud operation and maintenance management platform, remotely controls the start and stop of the equipment according to the instructions, or notifies the operation and maintenance personnel to perform corresponding equipment maintenance operations.

[0039] It should be noted that: The calculation formulas and each parameter participating in the operation in the present invention have been pre-dimensionless processed, and the process of dimensionless processing is well known in the industry and will not be described here.

[0040] The above has described a specific embodiment of the present invention in detail, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. An operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things, characterized in that, Including: A meteorological sensor array module for collecting low-altitude atmospheric environment parameters as observation data, including temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration; An Internet of Things communication module for real-time transmitting the observation data to the edge computing node module; An edge computing node module for locally preprocessing the original observation data, including data filtering, outlier removal, and compression; A cloud operation and maintenance management platform for storing and analyzing the preprocessed observation data, judging the device status according to the analysis results, and generating operation and maintenance instructions according to the device status; A control module for receiving alarm information and remotely controlling the start and stop of the device.

2. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 1, wherein The cloud operation and maintenance management platform includes: A data quality evaluation module for evaluating the quality of the observation data based on an abnormal data analysis model, and marking the abnormal data if there are quality problems; A device health score module for predicting the maintenance cycle according to the historical failure rate, signal stability, and battery life.

3. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 2, characterized in that, The abnormal data analysis model includes: A dynamic threshold detection unit for calculating the dynamic threshold range of each meteorological parameter in real time, and marking it as potentially abnormal when the observation data exceeds the threshold range; A time series prediction unit for predicting each meteorological parameter through a pre-trained time series prediction model to obtain a predicted value, comparing the actual observation value with the predicted value, and marking it as a time series anomaly when the deviation exceeds a preset tolerance; the time series prediction model is a long short-term memory (LSTM) neural network; A spatial consistency detection unit for calculating the spatial correlation between the observation data of the current device and the observation data of neighboring devices when observing in a multi-device network, and marking it as a spatial anomaly when the Mahalanobis distance exceeds a set threshold; A comprehensive decision-making unit for weighted fusion of the output results of the dynamic threshold detection unit, the time series prediction unit, and the spatial consistency detection unit, and determining it as the final abnormal data when the following conditions are met: ； Among them, , , are weight coefficients, determined based on historical data analysis. , , are dynamic, temporal, and spatial anomaly indication values respectively. The anomaly indication value takes 0 when judged normal and 1 when judged abnormal. is a preset judgment threshold. is a comprehensive anomaly indication judgment function. When , it means that at time t, the current observed data is judged as abnormal data; when , it means that at time t, the current observed data is judged as normal data.

4. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 3, characterized in that, The process of obtaining the dynamic exception indication value is as follows: ； wherein, is the observed value at time t, is the average value obtained by backtracking a set time period from time t, is the standard deviation value obtained by backtracking a set time period from time t, is an adjustable sensitivity coefficient determined based on historical data analysis.

5. The low-altitude meteorological observation equipment operation and maintenance management system based on the Internet of Things according to claim 4, wherein The acquisition process of the timing anomaly indication value is as follows: ； Among them, is the predicted value, is the preset error tolerance.

6. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 5, characterized in that, The spatial anomaly indication value is obtained as follows: ； Among them, is a preset threshold value, is the Mahalanobis distance between the current observed value and the data of neighboring sensors, The calculation formula of is: is the observed mean value of neighboring devices, is the spatial correlation coefficient matrix between the data of the current device and neighboring devices, represents the transpose.

7. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 6, characterized in that, The spatial correlation coefficient matrix is generated by a graph neural network, including: An input layer: receiving the observation data and geographical coordinates of each device; A graph convolutional layer: aggregating the features of neighboring devices through a message passing mechanism; Output layer: Generate a spatial correlation coefficient matrix , and satisfy ; Indicates the correlation coefficient between the th atmospheric environment parameter and the th atmospheric environment parameter; The training loss function of the graph neural network is: ; wherein, is the reference matrix calculated from historical data, , are the weight factors, determined based on historical data analysis, represents the sum of the main diagonal elements of matrix , represents taking the natural logarithm of the determinant of the spatial correlation coefficient matrix .

8. The operation and maintenance management system for low-altitude meteorological observation equipment based on the Internet of Things according to claim 7, characterized in that, The process of predicting the maintenance cycle according to the historical failure rate, signal stability, and battery life is: Through the formula: The predicted maintenance cycle is calculated ; Among them, is the standard maintenance period, is the device health score; the specific calculation formula is: ; among them, , , are the weight coefficients, is the historical failure rate, calculated by the ratio of the number of failures of the current device in the past period to the total operating time, is the signal stability, obtained by calculating the standard deviation of the signal strength of the current device over a period of time, is the battery life, obtained by calculating the ratio of the remaining battery power of the current device to the total battery capacity.

9. A method for operation and maintenance management of low-altitude meteorological observation equipment based on the Internet of Things, characterized in that, This method is implemented based on the Internet of Things-based low-altitude meteorological observation device operation and maintenance management system described in claim 2, and includes the following steps: S1. Data collection: The meteorological sensor array module continuously collects low-altitude atmospheric environment parameters, including temperature, humidity, air pressure, wind speed, wind direction, and particulate matter concentration, to form the original observation data; S2. Data transmission: The Internet of Things communication module transmits the original observation data collected by the meteorological sensor array module to the edge computing node module in real time; S3. Local preprocessing: The edge computing node module performs local preprocessing on the received original observation data, removes noise interference through data filtering, removes obvious error or abnormal data points by using an outlier removal algorithm, and compresses the data by using a compression algorithm to reduce the data volume; S4. Data Upload and Storage: The preprocessed data is uploaded to the cloud operation and maintenance management platform through the network, and the cloud operation and maintenance management platform stores the data in the database; S5. Data Quality Assessment and Anomaly Detection: Use the anomaly data analysis model to evaluate the quality of the observed data; S6. Equipment Status Judgment and Maintenance Instruction Generation: Judge the equipment status according to the data quality assessment results and the maintenance cycle prediction given by the equipment health score module; if there are abnormalities in the equipment or the maintenance cycle is approaching, generate corresponding maintenance instructions, and the maintenance instructions include equipment inspection, component replacement or parameter adjustment; S7. Instruction Execution: The control module receives the alarm information and maintenance instructions sent by the cloud operation and maintenance management platform, and remotely controls the start and stop of the equipment according to the instructions, or notifies the maintenance personnel to perform corresponding equipment maintenance operations.

Images