Low-temperature cloudy and rainy weather monitoring and early warning method and system based on Internet of Things

Abstract

The invention relates to the technical field of disaster prediction, in particular to a low-temperature cloudy and rainy weather monitoring and early warning method and system based on the Internet of Things. Comprising the following steps: S1, acquiring the number of sensors, the types of the sensors, the service life of the sensors and the historical temperature anomaly gradient of the whole monitoring area; based on the number of the sensors, the types of the sensors, the service life of the sensors and the historical temperature anomaly gradient, dynamic space clustering analysis is executed, and a dynamically partitioned monitoring area is obtained; s2, on the basis of the monitoring area, the sensor abnormal data in the monitoring area are divided according to the temperature phase change type, and the sensor abnormal data after division is obtained. According to the method, sensor errors and latent heat data are accurately distinguished through a multi-factor verification model, monitoring area division is dynamically optimized, risk conduction is rapidly responded in combination with a cascade early warning mechanism, self-adaptive threshold adjustment is performed to suppress false alarms, and the low-temperature cloudy and rainy disaster defense capability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster prediction, and particularly to a monitoring and early warning method and system for low-temperature and rainy weather based on the Internet of Things. Background Art

[0002] With the wide application of Internet of Things technology in the field of meteorological monitoring, the real-time early warning ability for low-temperature and rainy weather has been significantly improved. However, the existing technology relies on fixed weather stations and a single sensor network, and has significant defects in high-humidity and low-temperature environments: the sensors are interfered by icing or latent heat effects, resulting in inflated or distorted temperature data, making it difficult to distinguish real meteorological anomalies from equipment errors. Traditional models rely on static thresholds and numerical weather forecasts, lack a dynamic adjustment mechanism, have a lag in response to short-term sudden low-temperature and rainy events, and cannot solve the problem of missed detection of microclimates caused by insufficient density of the sensor network in complex terrains. In addition, the existing methods do not effectively integrate dynamic parameters such as sensor aging, type differences, and historical abnormal temperature gradients, misjudge sensor fault data (such as intermittent icing errors) as latent heat effects, frequently trigger false positive alarms, and weaken user trust. The problem of isolated early warnings between regions is prominent, and there is a lack of a cascaded early warning mechanism based on spatio-temporal priorities, resulting in insufficient risk conduction response between adjacent regions.

[0003] Therefore, there is an urgent need for a monitoring and early warning scheme that integrates multi-source data collaborative correction, dynamic partition optimization, and adaptive threshold adjustment to break through the technical bottlenecks of sensor error correction, risk conduction modeling, and false alarm suppression. Summary of the Invention

[0004] In order to overcome the disadvantages of data distortion and misjudgment, the present invention provides a monitoring and early warning method and system for low-temperature and rainy weather based on the Internet of Things.

[0005] The technical implementation solution of the present invention is: a monitoring and early warning method for low-temperature and rainy weather based on the Internet of Things, comprising the following steps: S1: Obtain the number of sensors, sensor types, sensor service life, and historical temperature anomaly gradient of the entire monitoring area; based on the number of sensors, sensor types, sensor service life, and the historical temperature anomaly gradient, perform dynamic spatial clustering analysis to obtain the dynamically partitioned monitoring area; S2: Based on the monitoring area, divide the abnormal sensor data in the monitoring area according to the temperature phase change type to obtain the divided abnormal sensor data, where the temperature phase change type includes temperature inflation caused by icing artifacts of the sensor and regional temperature rise caused by latent heat of phase change, verify the data credibility by combining the humidity time derivative, and separate the monitoring error and the real meteorological anomaly; S3: Based on the sensor anomaly data that has been partitioned, trigger a warning signal according to the spatio-temporal priority, and generate auxiliary data for risk assessment triggered by the warning signal; combine the neighborhood temperature covariance to quantitatively evaluate the risk of low-temperature and rainy disasters in the monitoring areas where the warning signal has not been triggered, calculate the comprehensive risk index value, and generate the risk assessment result for triggering the warning signal of low-temperature and rainy weather. The monitoring areas where the warning signal has not been triggered are the monitoring areas that have not reached the warning signal trigger threshold currently. S4: Based on the risk assessment result for triggering the warning signal of low-temperature and rainy weather, dynamically adjust the warning signal trigger threshold for the monitoring areas where the warning signal has not been triggered. The dynamic adjustment of the warning signal trigger threshold means that based on the neighborhood propagation effect, a gradient decay of the warning threshold is performed on the adjacent areas of the triggered areas.

[0006] Preferably, the obtaining of the number of sensors, sensor types, sensor service life, and historical temperature anomaly gradient of the entire monitoring area includes: Based on the number of sensors, obtain the sensor density factor of the entire monitoring area; Based on the sensor types, obtain the sensor efficiency factor of the entire monitoring area; Based on the sensor service life, obtain the sensor accuracy factor of the entire monitoring area; Based on the historical temperature anomaly gradient, obtain the historical temperature anomaly gradients of different monitoring areas in the entire monitoring area.

[0007] Preferably, the performing of dynamic spatial clustering analysis based on the number of sensors, sensor types, sensor service life, and the historical temperature anomaly gradient to obtain the monitored areas after dynamic partitioning includes: Calculate the weights of each sub-area using the monitoring area partitioning formula, and perform dynamic spatial clustering analysis based on the weights to obtain the monitored areas after dynamic partitioning. The monitoring area partitioning formula is as follows, where, is the monitoring area partitioning weight factor, is the sensor density factor, is the sensor efficiency factor, is the sensor accuracy factor, is the historical temperature anomaly gradient, is the smoothing coefficient.

[0008] Preferably, the partitioning of the sensor anomaly data within the monitoring area according to the temperature phase change type based on the monitoring area to obtain the partitioned sensor anomaly data includes: Take the monitoring area where the earliest low-temperature and rainy weather occurred in the historical records as the central monitoring area, collect the abnormal sensor data of the central monitoring area and its adjacent areas, and define the abnormal sensor data as the first abnormal sensor data; Based on the first abnormal sensor data, extract the proportion of temperature data below 0 degrees Celsius and above 0 degrees Celsius in the first abnormal sensor data, and define the proportion of temperature data below 0 degrees Celsius as the first temperature data proportional relationship. At the same time, define the proportion of temperature data above 0 degrees Celsius as the second temperature data proportional relationship; Based on the first temperature data proportional relationship and the second temperature data proportional relationship, obtain the temperature phase change type.

[0009] Preferably, the obtaining of the temperature phase change type based on the first temperature data proportional relationship and the second temperature data proportional relationship includes: Based on the first temperature data proportional relationship, if there is temperature data above 0 degrees Celsius during the monitoring period corresponding to the first temperature data proportional relationship, then define the temperature data above 0 degrees Celsius as the first error data; If the continuous duration of the first error data exceeds the preset persistence threshold, then determine the first error data as the phase change latent heat effect data; If the single duration of the first error data is less than or equal to the preset single duration threshold, and the number of occurrences per unit time is less than or equal to the preset number threshold, then determine the first error data as the sensor icing artifact error data; Based on the second temperature data proportional relationship, use a multi-factor verification model to verify the phase change latent heat effect data and the sensor icing artifact error data.

[0010] Preferably, the verifying of the phase change latent heat effect data and the sensor icing artifact error data based on the second temperature data proportional relationship using a multi-factor verification model includes: The multi-factor verification model is as follows, where, is the verification value, is the duration of temperature anomaly, is the persistence determination threshold, is the spatial consistency weight factor, is the current sensor temperature anomaly amplitude, is the average temperature anomaly amplitude of adjacent sensors, is the smoothing coefficient, is the humidity synergy weight factor, is the humidity time derivative.

[0011] Preferably, based on the divided sensor abnormal data, a warning signal is triggered according to the spatio-temporal priority, and risk assessment auxiliary data for triggering the warning signal is generated, including: If a warning signal is triggered first in the central monitoring area, the sensor abnormal data in the central monitoring area is used as the risk assessment auxiliary data for triggering the warning signal in the adjacent area of the central monitoring area; If a warning signal is triggered first in the adjacent area of the central monitoring area, the sensor abnormal data in the adjacent area of the central monitoring area is used as the risk assessment auxiliary data for triggering the warning signal in the central monitoring area.

[0012] Preferably, by combining the neighborhood temperature covariance, a quantitative assessment of the risk of low-temperature and rainy disasters is carried out on the monitoring areas where the warning signal has not been triggered, the comprehensive risk index value is calculated, and the risk assessment result for triggering the warning signal in low-temperature and rainy weather is generated, including: Based on the risk assessment auxiliary data for triggering the warning signal and the sensor abnormal data in the monitoring area, a risk assessment for triggering the warning signal in low-temperature and rainy weather is carried out on the monitoring areas where the warning signal has not been triggered by using the warning signal triggering risk assessment formula; the warning signal triggering risk assessment formula is as follows. Among them, is the comprehensive risk index value, is the latent heat effect intensity, is the sensor error, is the neighborhood temperature covariance, is the error correction coefficient, is the risk assessment auxiliary data for triggering the warning signal, 、 、 are the weight factors, is the correlation weight factor.

[0013] Preferably, based on the risk assessment result for triggering the warning signal in low-temperature and rainy weather, a dynamic adjustment of the warning signal triggering threshold is carried out on the monitoring areas where the warning signal has not been triggered, including: Based on the occurrence frequency and duration of the phase change latent heat effect data and the sensor icing artifact error data, the weight factors in the warning signal triggering risk assessment formula are reversely adjusted through the weight factor adjustment formula; the weight factor adjustment formula is as follows. Among them, is the adjusted weight factor, is the initial weight factor.

[0014] An Internet of Things-based low-temperature and rainy weather monitoring and early warning system for implementing the above-mentioned Internet of Things-based low-temperature and rainy weather monitoring and early warning method, including: A sensor data integration and dynamic partitioning module that dynamically divides monitoring sub-regions by collecting the number, type, service life, and historical temperature anomaly gradients of sensors in the monitoring area, and combines sensor density, efficiency, and accuracy factors to provide basic data support for subsequent anomaly monitoring; A phase change type intelligent identification module that takes the area where low-temperature and rainy weather occurs earliest as the center, collects temperature data of it and adjacent areas, analyzes the temperature distribution ratios below and above 0°C, combines the humidity change characteristics, and separates abnormal data sources based on the physical models of latent heat of phase change effect and sensor icing artifacts; A cascaded early warning signal triggering and risk assessment module that establishes an inter-regional association relationship according to the order of occurrence of early warning signal triggers, calculates the comprehensive risk index of the area where the early warning signal has not been triggered based on the intensity of the latent heat effect, sensor error, and temperature influence of adjacent areas, and dynamically reduces the early warning signal trigger threshold of adjacent areas; An adaptive threshold optimization module that reversely adjusts the weight parameters in the risk assessment model according to the occurrence frequency and duration of the latent heat of phase change effect data and sensor errors, and synchronously updates the early warning signal trigger threshold of the terminal device.

[0015] Beneficial effects: By constructing a multi-factor verification model, the present invention accurately distinguishes sensor icing artifact errors and latent heat of phase change effect data, combines a dynamic spatial clustering analysis algorithm to optimize the monitoring area division, and significantly improves the data analysis accuracy of low-temperature and rainy weather. Based on the cascaded early warning mechanism of spatio-temporal priority, with the earliest triggered area as the center, dynamically associates the risk conduction paths of adjacent areas, and realizes the pre-triggering and rapid response of early warning signals. By integrating sensor density factors, efficiency factors, and historical temperature anomaly gradients, dynamically optimizes the weight of monitoring area division, preferentially allocates resources to high-risk areas, and solves the problem of microclimate monitoring blind spots under complex terrains. The unique threshold adaptive optimization mechanism reversely adjusts the weight factors of the risk assessment model according to the occurrence frequency of the latent heat of phase change effect data and sensor errors, and synchronously reduces the early warning thresholds of adjacent areas, effectively suppressing the false alarm rate. In addition, the present invention enhances the physical credibility of latent heat effect identification through the collaborative analysis of the humidity time derivative and the duration of temperature anomaly, provides highly reliable technical support for agricultural frost damage prevention, traffic icing early warning, and power icing prevention and control, and comprehensively improves the active defense ability against low-temperature and rainy disasters. Description of the Drawings

[0016] Figure 1 It is a flowchart of the Internet of Things-based low-temperature and rainy weather monitoring and early warning method of the present invention; Figure 2 It is a structure diagram of the Internet of Things-based low-temperature and rainy weather monitoring and early warning system of the present invention. Detailed implementation manners

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to 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 creative efforts shall fall within the protection scope of the present invention.

[0018] Embodiment 1: A method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things, as Figure 1 shown, includes the following steps: S1: Obtain the number of sensors, sensor types, sensor service life in the entire monitoring area, and the historical temperature anomaly gradient in the entire monitoring area; based on the number of sensors, sensor types, sensor service life, and the historical temperature anomaly gradient, perform dynamic spatial clustering analysis to obtain the dynamically partitioned monitoring area; S2: Based on the monitoring area, divide the abnormal sensor data in the monitoring area according to the temperature phase change type to obtain the divided abnormal sensor data. The temperature phase change types include the temperature virtual increase caused by ice formation artifacts of the sensor and the regional temperature rise caused by the latent heat of phase change. Combine the humidity time derivative to verify the data credibility, and separate the monitoring error and the real meteorological anomaly; S3: Based on the divided abnormal sensor data, trigger a warning signal according to the spatio-temporal priority to generate auxiliary data for risk assessment of warning signal triggering; combine the neighborhood temperature covariance to quantitatively evaluate the risk of low-temperature and rainy disasters in the monitoring area where the warning signal is not triggered, calculate the comprehensive risk index value, and generate the risk assessment result of low-temperature and rainy weather warning signal triggering. The monitoring area where the warning signal is not triggered is the monitoring area that has not reached the warning signal triggering threshold currently; S4: Based on the risk assessment result of low-temperature and rainy weather warning signal triggering, dynamically adjust the warning signal triggering threshold for the monitoring area where the warning signal has not been triggered. The dynamic adjustment of the warning signal triggering threshold refers to performing a gradient attenuation of the warning threshold on the adjacent areas of the triggered area based on the neighborhood propagation effect.

[0019] For further explanation, the neighborhood propagation effect refers to the physical law of the spread of low-temperature and rainy weather along geographical or meteorological conditions (such as wind direction, terrain); the gradient attenuation adjusts the threshold through an exponential function, and the formula is: , where is the reference threshold, is the attenuation coefficient, is the Euclidean distance from the triggered area.

[0020] Obtain the number of sensors, sensor types, years of sensor use, and historical temperature anomaly gradients in the entire monitoring area, including: Based on the number of sensors, obtain the sensor density factor for the entire monitoring area; Based on the sensor types, obtain the sensor efficiency factor for the entire monitoring area; Based on the years of sensor use, obtain the sensor accuracy factor for the entire monitoring area; Based on the historical temperature anomaly gradients, obtain the historical temperature anomaly gradients for different monitoring areas in the entire monitoring area.

[0021] For further illustration, the sensor efficiency factor ( ): Based on the calibration parameters of sensor types (such as thermocouples, infrared sensors), determine the data acquisition efficiency through laboratory tests. The formula is: , the sensor accuracy factor ( ): Based on the years of sensor use, calculate the sensor performance decay rate through the following formula. The formula is: ( is the aging coefficient, calibrated by the equipment life experiment), the historical temperature anomaly gradient ( ): By statistically analyzing the temperature mutation amplitude and frequency in the historical data of the monitoring area, calculate the normalized gradient value. The formula is: . Supplementary note, The normalization of adopts the maximum-minimum value method, mapping the historical gradient to the [0, 1] interval. The formula is: and are the minimum and maximum values of the historical gradient, avoiding the denominator dominance in the high-gradient area.

[0022] Based on the number of sensors, sensor types, years of sensor use, and the historical temperature anomaly gradients, perform dynamic spatial clustering analysis to obtain the dynamically partitioned monitoring area, including: Calculate the weights of each sub-region using the monitoring area division formula and perform dynamic spatial clustering analysis based on the weights to obtain the dynamically partitioned monitoring area. The monitoring area division formula is as follows, where, is the monitoring area division weight factor, is the sensor density factor, is the sensor efficiency factor, is the sensor accuracy factor, is the historical temperature anomaly gradient, is the smoothing coefficient.

[0023] To further explain, the dynamic spatial clustering analysis formula calculates the weights of each sub-region by integrating sensor density (coverage ability), efficiency (type difference), accuracy (aging effect), and the normalized historical temperature anomaly gradient. , where a higher weight indicates a higher monitoring priority. Based on the weight distribution, a density clustering algorithm (such as DBSCAN) is used to divide the high, medium, and low-risk monitoring regions. Technical effect: Suppress the over-sensitivity of historical anomaly regions and avoid misjudging regions with dense but aging sensors as high-priority. Example: Sensor density in a certain region in the Northeast = 0.9, aging coefficient = 0.3, efficiency factor = 0.8 (based on type calibration), normalized historical gradient, then: = (0.9 * 0.8 * 0.3) / 0.5 + 0.3 = 0.36, the weight is reduced to avoid false triggering.

[0024] Based on the monitoring regions, the sensor anomaly data within the monitoring regions is divided according to the temperature phase change type to obtain the divided sensor anomaly data, including: Taking the monitoring region where the earliest low-temperature and rainy weather occurred in the historical records as the central monitoring region, collecting the sensor anomaly data of the central monitoring region and its adjacent regions, and defining the sensor anomaly data as the first sensor anomaly data; Based on the first sensor anomaly data, extracting the proportion of temperature data below 0 degrees Celsius and above 0 degrees Celsius in the first sensor anomaly data, and defining the proportion of temperature data below 0 degrees Celsius as the first temperature data proportion relationship, and at the same time, defining the proportion of temperature data above 0 degrees Celsius as the second temperature data proportion relationship; Based on the first temperature data proportion relationship and the second temperature data proportion relationship, obtaining the temperature phase change type.

[0025] Further explanation is as follows. The technical logic of temperature phase change type division (the data ratio above and below 0°C) is that during the freezing process, latent heat is released (the temperature briefly rises) and there are sensor icing artifact errors (the temperature is falsely high). By statistically analyzing the data ratio below / above 0°C, two types of anomalies can be distinguished. The technical effect is that based on the persistence threshold and the multi-factor verification model, the latent heat effect (an environmental physical phenomenon) and the sensor icing artifact error (a device failure) can be distinguished, significantly reducing the false alarm rate. Example: In a certain area, the data below 0°C accounts for 70% (the first ratio), but 80% of the persistent anomalies in the data above 0°C are detected. It is determined as a latent heat effect rather than a sensor icing. Supplementary explanation: The preset persistence threshold is set based on the statistical analysis of historical low-temperature and rainy events (such as 30 minutes), and it supports dynamic adjustment according to the regional climate characteristics; the intermittent threshold is defined as the single anomaly duration ≤ t minutes (such as 5 minutes) and the number of occurrences per hour ≤ x times (such as 3 times).

[0026] Based on the first temperature data ratio relationship and the second temperature data ratio relationship, obtain the temperature phase change type, including: Based on the first temperature data ratio relationship, if there is temperature data above 0 degrees Celsius during the monitoring period corresponding to the first temperature data ratio relationship, then define the temperature data above 0 degrees Celsius as the first error data; If the continuous duration of the first error data exceeds the preset persistence threshold, then determine the first error data as the phase change latent heat effect data; If the single duration of the first error data is less than or equal to the preset single duration threshold and the number of occurrences per unit time is less than or equal to the preset number threshold, then determine the first error data as the sensor icing artifact error data; Based on the second temperature data ratio relationship, use the multi-factor verification model to verify the phase change latent heat effect data and the sensor icing artifact error data.

[0027] Further explanation is as follows. The second temperature data ratio relationship (the proportion of data above 0°C) is used to initially distinguish the latent heat effect from the sensor error, while the multi-factor verification model verifies the physical credibility of the initial classification results through time persistence, spatial consistency, and humidity synergy. For example, if the proportion of data above 0°C in a certain area is high (the second ratio relationship), but the humidity derivative If it is negative (humidity decreases), it may be a sensor error and needs to be rejudged; the technical logic of the phase change type judgment rule (persistent vs intermittent), the persistent data above 0°C reflects the latent heat of phase change (slow heat release from the ice layer), and the intermittent anomaly reflects the temporary malfunction of the sensor after icing. Technical effect, distinguish environmental physical phenomena from equipment failures and handle them accordingly. Example, if a sensor shows readings above 0°C for 5 minutes every hour (intermittent), it is judged as icing error; if it is above 0°C for 3 consecutive hours, it is judged as latent heat effect.

[0028] Based on the second temperature data proportional relationship, use a multi-factor verification model to verify the phase change latent heat effect data and the sensor icing artifact error data, including: The multi-factor verification model is as follows, Among them, is the verification value, is the duration of temperature anomaly, is the persistent determination threshold, is the spatial consistency weight factor, is the current sensor temperature anomaly amplitude, is the average temperature anomaly amplitude of adjacent sensors, is the smoothing coefficient, is the humidity synergy weight factor, is the humidity time derivative.

[0029] For further explanation, the technical effect of the multi-factor verification model is to suppress the influence of abnormal data of isolated sensors through the spatial consistency weight ( ), and improve the reliability of anomaly verification. Example, for a certain sensor = 2 hours ( = 1 hour), = 3°C (neighborhood average = 1°C), = 5% / h ( = 0.2), then = 2 / 1 + 0.5×|3 - 1| / (1 + 0.1) + 0.2×5 = 2 + 0.91 + 1 = 3.91. If it is higher than the threshold, it is confirmed as an effective early warning. Supplementary explanation, the physical meaning of the humidity derivative is to reflect the freezing / melting rate (such as > 0 indicates the freezing process).

[0030] Based on the divided sensor abnormal data, trigger warning signals according to the spatio-temporal priority, and generate auxiliary data for risk assessment of warning signal triggering, including: If a warning signal is triggered first in the central monitoring area, the abnormal sensor data in the central monitoring area is used as the auxiliary data for the risk assessment of the warning signal trigger in the adjacent areas of the central monitoring area. If a warning signal is triggered first in the adjacent areas of the central monitoring area, the abnormal sensor data in the adjacent areas of the central monitoring area is used as the auxiliary data for the risk assessment of the warning signal trigger in the central monitoring area.

[0031] For further explanation, according to the order of warning trigger, the data reference weights are dynamically adjusted, and the data in the earliest triggered area is trusted first. Technical effect: Based on the spatio-temporal priority trigger rule, the data in the core area is given priority to be responded to, ensuring the logical consistency of the spatio-temporal propagation of the warning signal. Example: A certain core area in Hangzhou triggers a warning first, and its adjacent areas use the data in the core area for auxiliary evaluation, and the threshold is reduced by 10%. Supplementary explanation: The time accuracy of "triggered first" is defined as a time window (for example, triggering successively within 5 minutes is regarded as simultaneous), to prevent logical oscillation caused by tiny delays.

[0032] Combined with the neighborhood temperature covariance, a quantitative assessment of the risk of low-temperature and rainy disasters is carried out for the monitoring areas where the warning signal has not been triggered, the comprehensive risk index value is calculated, and the risk assessment result of the warning signal trigger for low-temperature and rainy weather is generated, including: Based on the auxiliary data for the risk assessment of the warning signal trigger and the abnormal sensor data in the monitoring area, the risk assessment of the warning signal trigger for low-temperature and rainy weather is carried out for the monitoring areas where the warning signal has not been triggered by using the warning signal trigger risk assessment formula; the warning signal trigger risk assessment formula is as follows. Where, is the comprehensive risk index value, is the intensity of the latent heat effect, is the sensor error, is the neighborhood temperature covariance, is the error correction coefficient, is the auxiliary data for the risk assessment of the warning signal trigger, 、 、 are the weight factors, is the correlation weight factor.

[0033] For further explanation, the technical logic of the risk assessment formula comprehensively considers the latent heat intensity ( ), the sensor error ( ), and the neighborhood temperature covariance ( ) to quantify the risk of the untriggered area. Technical effect: By comprehensively considering the latent heat intensity, the sensor error, and the neighborhood synergy effect, the risk of the untriggered area is quantified, and the early warning of high-risk areas is realized. Example: A certain area =0.8 (strong latent heat), =1.2℃ (large error), =0.6, =0.5, =0.3, =0.2, then =0.5×0.8+0.3×1.2 / 0.9+0.2×0.6=0.4+0.4+0.12=0.92, when the comprehensive risk index value = Exceeding a preset risk threshold (e.g. =≥0.9), it is determined that the area needs to trigger an early warning signal. Supplementary explanation: Error correction coefficient The source is based on the sensor age calibration (such as =1-0.05×years of use); neighborhood temperature covariance ( ) is obtained by calculating the covariance of the temperature anomaly amplitude between the current area and the adjacent area, and the formula is: ,in, is the number of adjacent regions, is the mean value, reflecting the coordination of temperature changes among regions.

[0034] Based on the risk assessment result of the low temperature and rainy weather warning signal triggering, the warning signal triggering threshold is dynamically adjusted for the monitoring area where the warning signal has not been triggered, including: Based on the occurrence frequency and duration of the phase change latent heat effect data and the sensor icing artifact error data, the weight factor in the early warning signal trigger risk assessment formula is reversely adjusted through a weight factor adjustment formula; the weight factor adjustment formula is as follows: in, is the adjusted weight factor, is the initial weight factor. Further explanation is that the technical logic of dynamic adjustment of thresholds (lowering the thresholds of adjacent areas) is that the adjacent areas of the triggered area are at higher risk of being affected by physical conduction, and early warning is achieved by lowering the threshold. Technical effect: dynamically adjust the threshold according to the neighborhood propagation effect, and give early warning to potential risk areas on the diffusion path of meteorological disasters, such as suburbs on the moving path of cold fronts. Supplementary explanation: the scope of "adjacent areas" is defined as a spatial radius (such as 5 kilometers) or administrative division to avoid over-generalization.

[0035] Technical logic of reverse adjustment of weight factor (influence of frequency and duration). For high-frequency and long-duration latent heat effects or sensor failures, the weight of the risk assessment formula needs to be reduced. Technical effect: Through the reverse adjustment mechanism of the weight factor, the influence of multiple risk parameters is balanced, overfitting caused by a single factor is prevented, and the generalization ability of the model is enhanced. Example: The sensors in a certain area frequently freeze (10 times per month on average), and is reduced from 0.3 to 0.1 to reduce the false alarm rate. Supplementary note: The failure frequency is the proportion of the occurrence times of sensor icing artifact errors or latent heat effect data in the total monitoring times, and the total sample number is the total monitoring times or the number of valid data samples in the same time period.

[0036] Embodiment 2: On the basis of Embodiment 1, an Internet of Things-based low-temperature and rainy weather monitoring and early warning system, as Figure 2 shown, includes: Sensor data integration and dynamic partitioning module, which dynamically partitions the monitoring sub-areas by collecting the number, type, service life and historical temperature anomaly gradient of sensors in the monitoring area, and combines the sensor density, efficiency and accuracy factors to provide basic data support for subsequent anomaly monitoring; Intelligent phase change type identification module, centered on the area where low-temperature and rainy weather occurs earliest, collects temperature data of it and adjacent areas, analyzes the temperature distribution ratio below and above 0°C, combines the humidity change characteristics, and separates abnormal data sources based on the physical models of latent heat effect of phase change and sensor icing artifacts; Cascaded early warning signal triggering and risk assessment module, which establishes an inter-regional association relationship according to the sequence of occurrence of early warning signal triggering, calculates the comprehensive risk index of the area where the early warning signal has not been triggered based on the latent heat effect intensity, sensor error and temperature influence of adjacent areas, and dynamically reduces the early warning signal triggering threshold of adjacent areas; Adaptive threshold optimization module, which reversely adjusts the weight parameters in the risk assessment model according to the occurrence frequency and duration of latent heat effect data of phase change and sensor error, and synchronously updates the early warning signal triggering threshold of the terminal device.

[0037] The above has introduced this application in detail. Specific examples are used in this article to elaborate on the principle and implementation method of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A monitoring and early warning method for low-temperature and rainy weather based on the Internet of Things, characterized in that It includes the following steps: S1: Obtain the number of sensors, sensor types, sensor service life, and historical temperature anomaly gradient of the entire monitoring area in the entire monitoring area; based on the number of sensors, sensor types, sensor service life, and the historical temperature anomaly gradient, perform dynamic spatial clustering analysis to obtain the dynamically partitioned monitoring area; S2: Based on the monitoring area, divide the sensor anomaly data in the monitoring area according to the temperature phase change type to obtain the divided sensor anomaly data. The temperature phase change types include the temperature over-increase caused by icing artifacts of the sensor and the regional temperature rise caused by the latent heat of phase change. Combine the humidity time derivative to verify the data credibility, and separate the monitoring error and the real meteorological anomaly; S3: Based on the divided sensor anomaly data, trigger the warning signal according to the spatio-temporal priority, and generate the auxiliary data for the risk assessment triggered by the warning signal; Combine the neighborhood temperature covariance to quantitatively evaluate the risk of low-temperature and rainy disasters in the monitoring areas where the warning signal is not triggered, calculate the comprehensive risk index value, and generate the risk assessment result for triggering the warning signal for low-temperature and rainy weather. The monitoring areas where the warning signal is not triggered are the monitoring areas that have not reached the warning signal trigger threshold currently; S4: Based on the risk assessment result for triggering the warning signal for low-temperature and rainy weather, dynamically adjust the warning signal trigger threshold for the monitoring areas where the warning signal is not triggered. The dynamic adjustment of the warning signal trigger threshold means that based on the neighborhood propagation effect, perform the warning threshold gradient attenuation on the adjacent areas of the triggered areas.

2. The method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 1, characterized in that, The obtaining the number of sensors, sensor types, sensor service life, and historical temperature anomaly gradient of the entire monitoring area in the entire monitoring area includes: Based on the number of sensors, obtain the sensor density factor of the entire monitoring area; Based on the sensor types, obtain the sensor efficiency factor of the entire monitoring area; Based on the sensor service life, obtain the sensor accuracy factor of the entire monitoring area; Based on the historical temperature anomaly gradient, obtain the historical temperature anomaly gradients of different monitoring areas in the entire monitoring area.

3. The method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 1, characterized in that, The performing dynamic spatial clustering analysis based on the number of sensors, sensor types, sensor service life, and the historical temperature anomaly gradient to obtain the dynamically partitioned monitoring area includes: Calculate the weights of each sub-region using the monitoring area division formula, and perform dynamic spatial clustering analysis based on the weights to obtain the dynamically partitioned monitoring area; the monitoring area division formula is as follows, Among them, is the weight factor for monitoring area division, is the sensor density factor, is the sensor efficiency factor, is the sensor accuracy factor, is the historical temperature anomaly gradient, is the smoothing coefficient.

4. The method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 1, characterized in that The dividing the sensor anomaly data in the monitoring area according to the temperature phase change type based on the monitoring area to obtain the divided sensor anomaly data includes: Take the monitoring area where the earliest low-temperature and rainy weather occurred in the historical records as the central monitoring area, collect the sensor anomaly data of the central monitoring area and the adjacent areas of the central monitoring area, and define the sensor anomaly data as the first sensor anomaly data; Based on the abnormal data of the first sensor, extract the proportion of temperature data below 0°C and above 0°C in the abnormal data of the first sensor, and define the proportion of temperature data below 0°C as the first temperature data proportional relationship. At the same time, define the proportion of temperature data above 0°C as the second temperature data proportional relationship; Based on the first temperature data proportional relationship and the second temperature data proportional relationship, obtain the temperature phase change type.

5. The method for monitoring and early warning of low-temperature and rainy weather based on the Internet of Things according to claim 4, wherein, The obtaining of the temperature phase change type based on the first temperature data proportional relationship and the second temperature data proportional relationship includes: Based on the first temperature data proportional relationship, if there is temperature data above 0°C during the monitoring period corresponding to the first temperature data proportional relationship, then define the temperature data above 0°C as the first error data; If the continuous duration of the first error data exceeds the preset persistence threshold, then determine the first error data as the phase change latent heat effect data; If the single duration of the first error data is less than or equal to the preset single duration threshold and the number of occurrences per unit time is less than or equal to the preset number threshold, then determine the first error data as the sensor icing artifact error data; Based on the second temperature data proportional relationship, use a multi-factor verification model to verify the phase change latent heat effect data and the sensor icing artifact error data.

6. The method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 5, characterized in that The verification of the phase change latent heat effect data and the sensor icing artifact error data using a multi-factor verification model based on the second temperature data proportional relationship includes: The multi-factor verification model is as follows, wherein, is the verification value, is the duration of abnormal temperature persistence, is the persistence determination threshold, is the spatial consistency weight factor, is the abnormal temperature amplitude of the current sensor, is the average abnormal temperature amplitude of adjacent sensors, is the smoothing coefficient, is the humidity synergy weight factor, is the time derivative of humidity.

7. A method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 1, characterized in that, Based on the divided abnormal sensor data, trigger a warning signal according to the spatio-temporal priority, and generate auxiliary data for risk assessment of warning signal triggering, including: If the warning signal trigger occurs first in the central monitoring area, use the abnormal sensor data of the central monitoring area as the auxiliary data for risk assessment of warning signal triggering in the adjacent area of the central monitoring area; If the warning signal trigger occurs first in the adjacent area of the central monitoring area, use the abnormal sensor data of the adjacent area of the central monitoring area as the auxiliary data for risk assessment of warning signal triggering in the central monitoring area.

8. The method for monitoring and warning of low-temperature and rainy weather based on the Internet of Things according to claim 1, characterized in that, The combination of neighborhood temperature covariance is used to quantitatively evaluate the risk of low-temperature and rainy disasters in the monitoring area where the warning signal has not been triggered, calculate the comprehensive risk index value, and generate the risk assessment result of warning signal triggering for low-temperature and rainy weather, including: Based on the auxiliary data for risk assessment of warning signal triggering and the abnormal sensor data of the monitoring area, use the warning signal trigger risk assessment formula to conduct a risk assessment of warning signal triggering for low-temperature and rainy weather in the monitoring area where the warning signal has not been triggered; The warning signal trigger risk assessment formula is as follows, Among them, is the comprehensive risk index value, is the latent heat effect intensity, is the sensor error, is the neighborhood temperature covariance, is the error correction coefficient, is the auxiliary data for triggering risk assessment of early warning signals, 、 、 are weight factors, is the associated weight factor.

9. The method for monitoring and early warning of low-temperature and rainy weather based on the Internet of Things according to claim 8, characterized in that, Based on the risk assessment result of warning signal triggering for low-temperature and rainy weather, conduct a dynamic adjustment of the warning signal trigger threshold for the monitoring area where the warning signal has not been triggered, including: Based on the occurrence frequency and duration of the phase change latent heat effect data and the sensor icing artifact error data, the weight factor in the early warning signal trigger risk assessment formula is adjusted inversely through the weight factor adjustment formula; the weight factor adjustment formula is as follows, Among them, is the adjusted weight factor, is the initial weight factor.

10. An Internet of Things-based low-temperature and rainy weather monitoring and early warning system for implementing the Internet of Things-based low-temperature and rainy weather monitoring and early warning method according to any one of claims 1-9, characterized in that, including: The sensor data integration and dynamic partitioning module dynamically divides the monitoring sub-regions by collecting the number, type, service life, and historical temperature anomaly gradient of sensors in the monitoring area, and combines the sensor density, efficiency, and accuracy factors to provide basic data support for subsequent anomaly monitoring; The intelligent phase change type identification module takes the area with the earliest occurrence of low-temperature and rainy weather as the center, collects the temperature data of it and adjacent areas, analyzes the temperature distribution ratios below and above 0°C, combines the humidity change characteristics, and separates the abnormal data sources based on the physical models of the phase change latent heat effect and sensor icing artifacts; The cascaded early warning signal trigger and risk assessment module establishes the inter-regional correlation relationship according to the sequence of occurrence of early warning signal triggers, calculates the comprehensive risk index of the area where the early warning signal is not triggered based on the latent heat effect intensity, sensor error, and the temperature influence of adjacent areas, and dynamically reduces the early warning signal trigger threshold of adjacent areas; The adaptive threshold optimization module inversely adjusts the weight parameters in the risk assessment model according to the occurrence frequency and duration of the phase change latent heat effect data and sensor errors, and synchronously updates the early warning signal trigger threshold of the terminal device.

Images