Forest remote monitoring and early warning system based on Internet of Things

Abstract

The invention relates to the technical field of forest monitoring, in particular to a forest remote monitoring and early warning system based on the Internet of Things, which is used for solving the technical problem of low forest environment monitoring accuracy in the prior art. The system comprises a data acquisition unit used for acquiring monitoring data of a forest environment, a fault identification unit used for determining fault weight factors of monitoring equipment, and identifying fault monitoring equipment and non-fault monitoring equipment according to distribution of the fault weight factors of the monitoring equipment; the abnormal value correction unit is used for determining the abnormal degree of electromagnetic interference based on the distribution characteristics of the abnormal values in the data sequence of each non-fault monitoring device, and correcting the abnormal values in the data sequence according to the abnormal degree of electromagnetic interference to obtain a corrected data sequence; the monitoring result generation unit is used for generating a forest environment monitoring result according to the corrected data sequence; and the early warning unit is used for carrying out early warning based on the environment monitoring result.

Description

Technical Field

[0001] This invention relates to the field of forest monitoring technology, and more specifically to a forest remote monitoring and early warning system based on the Internet of Things. Background Technology

[0002] Currently, to avoid frequent forest fires, various types of sensors can be deployed in forests to collect data such as temperature, humidity, and smoke concentration in real time, and this data can be used for early detection and warning of forest fires.

[0003] Existing forest remote monitoring systems typically deploy multiple sensor nodes in the monitoring area, transmitting the collected data to a cloud platform via an IoT communication module. The platform then builds predictive models based on historical data, determining environmental anomalies by comparing real-time data with predicted thresholds. However, due to the complex and variable nature of forest environments and the long-term operation of monitoring equipment in the field, the collected data is susceptible to various factors that can cause anomalies. This results in insufficient accuracy of monitoring results generated from such data, poor overall system early warning performance, and consequently, low accuracy in forest environmental monitoring. Summary of the Invention

[0004] To address the low accuracy of existing forest environment monitoring technologies, the present invention aims to provide an Internet of Things-based remote forest monitoring and early warning system. The specific technical solution adopted is as follows: The data acquisition unit is used to collect real-time monitoring data of the forest environment through multiple monitoring devices deployed in the forest. The monitoring data includes temperature data, humidity data, smoke concentration data, light intensity data, and monitoring images. The monitoring devices communicate with the outside world through an Internet of Things (IoT) module. The fault identification unit is used to determine the fault weight factor of each monitoring device based on the data sequence of each monitoring device, and to identify faulty monitoring devices and non-faulty monitoring devices according to the distribution of the fault weight factors of the monitoring devices; the fault weight factor is used to characterize the probability of the monitoring device malfunctioning. The outlier correction unit is used to determine the electromagnetic interference anomaly degree of each non-fault monitoring device based on the distribution characteristics of outliers in the data sequence of each non-fault monitoring device, and to correct the outliers in the data sequence according to the electromagnetic interference anomaly degree to obtain the corrected data sequence; the electromagnetic interference anomaly degree is used to characterize the degree to which data points are affected by electromagnetic interference. The monitoring result generation unit is used to generate forest environment monitoring results based on the corrected data sequence; The early warning unit is used to issue early warnings based on environmental monitoring results.

[0005] In one possible implementation, the fault identification unit includes a fault weight factor determination subunit: the fault weight factor determination subunit is configured to: determine a first fluctuation index based on a first data sequence from a first monitoring device within a first time period; determine a second fluctuation index based on a second data sequence from the first monitoring device within a second time period; determine a third fluctuation index based on a third data sequence from multiple second monitoring devices within the first time period; wherein the duration of the second time period is longer than the duration of the first time period, the first and second monitoring devices are monitoring devices that collect the same type of data, and the fluctuation index is used to characterize the degree of data fluctuation; determine a fluctuation difference degree based on the difference between the first and second fluctuation indices, and the difference between the first fluctuation index and each third fluctuation index; determine a deviation component based on the degree of deviation between the data in the first data sequence and the average value of the second data sequence; and determine a fault weight factor based on the fluctuation difference degree and the deviation component.

[0006] In one possible implementation, the fault identification unit includes a fluctuation index determination subunit, which is used to: take the average of the absolute values ​​of all data in the first difference sequence of the first data sequence as the first fluctuation index; take the average of the absolute values ​​of all data in the first difference sequence of the second data sequence as the second fluctuation index; and take the average of the absolute values ​​of all data in the first difference sequence of the third data sequence as the third fluctuation index.

[0007] In one possible implementation, the fault identification unit includes a deviation component determination subunit, which is used to: calculate the absolute value of the difference between each first fluctuation index and the second fluctuation index to obtain a first difference component; obtain a second difference component based on the sum of the absolute values ​​of the differences between each first fluctuation index; use the sum of the first difference component and the second difference component as the fluctuation difference degree; and use the sum of the absolute differences between each data point in the first data sequence and the average value of the second data sequence as the deviation component.

[0008] In one possible implementation, the fault identification unit is further configured to: generate a statistical sequence including fault weight factors for each monitoring device; calculate the Z-score value of each fault weight factor in the statistical sequence, wherein the Z-score value is used to characterize the degree of deviation of the fault weight factor in the statistical sequence; and identify monitoring devices with Z-score values ​​greater than a preset first threshold as fault monitoring devices.

[0009] In one possible implementation, the outlier correction unit includes an electromagnetic interference anomaly determination subunit. This subunit is used to: calculate the local reachability density of each data point in the data sequence of the target non-fault monitoring device based on a preset anomaly detection algorithm; the local reachability density characterizes the outlier degree of the data point's distribution in the data sequence; determine the illumination influence factor based on the ratio of the illumination intensity data of each data point at the same time to a preset threshold; the illumination influence factor characterizes the probability of an electromagnetic interference source operating; calculate the absolute difference between each data point and the average value of its preceding and following adjacent data points to obtain an anomaly factor, which characterizes the degree of abrupt change in the data point; and determine the electromagnetic interference anomaly degree based on the local reachability density, the illumination influence factor, and the anomaly factor.

[0010] In one possible implementation, the outlier correction unit includes a light influence factor determination subunit, which is used to: determine the value of the light influence factor as a first value when the light intensity data is greater than a preset threshold; the first value is used to characterize that the electromagnetic interference source is in a working state; and determine the value of the light influence factor as a second value when the light intensity data is less than or equal to the preset threshold; the second value is used to characterize that the electromagnetic interference source is in a non-working state.

[0011] In one possible implementation, the outlier correction unit is further configured to: take the electromagnetic interference anomaly degree of all data points in the target data sequence as an anomaly degree sequence; determine the segmentation threshold of the anomaly degree sequence based on a preset threshold segmentation algorithm, the segmentation threshold being used to distinguish normal data and abnormal data in the anomaly degree sequence; delete data points in the target data sequence whose electromagnetic interference anomaly degree is greater than the segmentation threshold, to obtain a target data sequence with missing values; and interpolate the target data sequence with missing values ​​based on a preset data interpolation algorithm to obtain a corrected data sequence.

[0012] In one possible implementation, the monitoring result generation unit is used to: predict the corrected data sequence of each non-fault monitoring device based on a preset time series prediction algorithm to obtain predicted data; compare the predicted data of each non-fault monitoring device with the corresponding preset warning threshold to determine the monitoring sub-result; identify the monitoring image based on a preset image recognition algorithm to determine the image analysis result of the target object; and perform decision fusion on the monitoring sub-result and the image analysis result to obtain the forest environment monitoring result.

[0013] In one possible implementation, the system also includes a monitoring and management cloud platform; wherein, the early warning unit is specifically used to compare the forest environment monitoring results with the preset alarm threshold, and determine whether to issue an alarm based on the comparison results; the monitoring and management cloud platform is used to receive and store the monitoring data collected by the data acquisition unit, the corrected data sequence output by the outlier correction unit, and the forest environment monitoring results generated by the monitoring result generation unit, and display the received data in a visual form.

[0014] The present invention has the following beneficial effects: This application provides a forest remote monitoring and early warning system based on the Internet of Things. The system first acquires multi-source environmental data through a distributed monitoring network, providing a rich data foundation for subsequent analysis; then, it effectively filters out unreliable monitoring equipment through an intelligent fault identification mechanism, eliminating interference from faulty data at the source; furthermore, it performs refined anomaly detection and correction on the data collected by normal equipment, eliminating the influence of instantaneous interference factors; finally, it generates accurate monitoring results based on high-quality data. Based on this scheme, the present application significantly improves the reliability and early warning accuracy of the entire system, effectively controlling the false alarm rate and missed alarm rate of forest fire monitoring, providing more reliable technical support for forest protection work, and greatly improving the accuracy of forest environmental monitoring. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A system architecture for a forest remote monitoring and early warning system based on the Internet of Things (IoT) is provided as an embodiment of the present invention. Figure 1 ; Figure 2 A system architecture for a forest remote monitoring and early warning system based on the Internet of Things (IoT) is provided as an embodiment of the present invention. Figure 2 ; Figure 3 A system architecture for a forest remote monitoring and early warning system based on the Internet of Things (IoT) is provided as an embodiment of the present invention. Figure 3 ; Figure 4 A system architecture for a forest remote monitoring and early warning system based on the Internet of Things (IoT) is provided as an embodiment of the present invention. Figure 4 . Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a forest remote monitoring and early warning system based on the Internet of Things proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] Unless otherwise specified, the normalization function Norm() mentioned in this application uses maximum and minimum value normalization. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculation result exceeds the [0,1] interval, a truncation function is used to limit it to the [0,1] range (i.e., if the result is less than 0, it is taken as 0; if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0020] The following description, in conjunction with the accompanying drawings, details a specific solution for a forest remote monitoring and early warning system based on the Internet of Things provided by this invention.

[0021] Please see Figure 1 This illustrates a system architecture diagram of a forest remote monitoring and early warning system based on the Internet of Things, according to an embodiment of the present invention. Figure 1 As shown, the Internet of Things-based forest remote monitoring and early warning system includes a data acquisition unit 101, a fault identification unit 102, an outlier correction unit 103, a monitoring result generation unit 104, and an early warning unit 105.

[0022] The data acquisition unit 101 is used to collect real-time monitoring data of the forest environment through multiple monitoring devices deployed in the forest. The monitoring data includes temperature data, humidity data, smoke concentration data, light intensity data, and monitoring images. The monitoring devices communicate with the outside world through an Internet of Things (IoT) module.

[0023] The fault identification unit 102 is used to determine the fault weight factor of the monitoring device based on the data sequence of each monitoring device, and to identify fault monitoring devices and non-fault monitoring devices according to the distribution of the fault weight factors of the monitoring devices; the fault weight factor is used to characterize the probability of the monitoring device malfunctioning.

[0024] The outlier correction unit 103 is used to determine the electromagnetic interference anomaly degree of each non-fault monitoring device based on the distribution characteristics of outliers in the data sequence of each non-fault monitoring device, and to correct the outliers in the data sequence according to the electromagnetic interference anomaly degree to obtain the corrected data sequence; the electromagnetic interference anomaly degree is used to characterize the degree of influence of electromagnetic interference on the data points.

[0025] The monitoring result generation unit 104 is used to generate forest environment monitoring results based on the corrected data sequence.

[0026] Early warning unit 105 is used to issue early warnings based on environmental monitoring results.

[0027] One possible approach is to collect forest environment monitoring data as follows: The forest to be monitored is evenly divided into multiple areas, and monitoring equipment is installed in each area to collect real-time monitoring data. This monitoring equipment includes temperature sensors, humidity sensors, smoke concentration sensors, light intensity sensors, and surveillance cameras. The monitoring data includes data collected by the aforementioned equipment, such as temperature, air humidity, soil moisture, smoke concentration, light intensity, and surveillance images. Optionally, all types of data are collected at the same sampling frequency of once per minute. After collecting the monitoring data, the monitoring equipment sends the collected data to the data acquisition unit via its built-in IoT module.

[0028] Based on the above technical solutions, this application provides an Internet of Things-based remote forest monitoring and early warning system. This system first acquires multi-source environmental data through a distributed monitoring network, providing a rich data foundation for subsequent analysis. Then, it effectively filters out unreliable monitoring equipment through an intelligent fault identification mechanism, eliminating interference from faulty data at the source. Furthermore, it performs refined anomaly detection and correction on data collected by normal equipment, eliminating the influence of instantaneous interference factors. Finally, it generates accurate monitoring results based on high-quality data. Based on this solution, this application significantly improves the reliability and early warning accuracy of the entire system, effectively controlling the false alarm and missed alarm rates in forest fire monitoring, providing more reliable technical support for forest protection work, and greatly improving the accuracy of forest environmental monitoring.

[0029] Combination Figure 1 ,like Figure 2As shown, in one possible implementation, the fault identification unit 102 includes a fault weight factor determination subunit 1021: the fault weight factor determination subunit 1021 is used to: determine a first fluctuation index based on a first data sequence from a first monitoring device within a first time period; determine a second fluctuation index based on a second data sequence from the first monitoring device within a second time period; determine a third fluctuation index based on a third data sequence from multiple second monitoring devices within a first time period; wherein the duration of the second time period is longer than the duration of the first time period, the first and second monitoring devices are monitoring devices that collect the same type of data, and the fluctuation index is used to characterize the degree of data fluctuation; determine a fluctuation difference degree based on the difference between the first and second fluctuation indices, and the difference between the first fluctuation index and each third fluctuation index; determine a deviation component based on the degree of deviation between the data in the first data sequence and the average value of the second data sequence; and determine a fault weight factor based on the fluctuation difference degree and the deviation component.

[0030] Optionally, the fault identification unit 102 further includes a fluctuation index determination subunit 1022, which is used to: take the average of the absolute values ​​of all data in the first difference sequence of the first data sequence as the first fluctuation index; take the average of the absolute values ​​of all data in the first difference sequence of the second data sequence as the second fluctuation index; and take the average of the absolute values ​​of all data in the first difference sequence of the third data sequence as the third fluctuation index.

[0031] Optionally, the fault identification unit 102 further includes a deviation component determination subunit 1023, which is used to: calculate the absolute value of the difference between each first fluctuation index and the second fluctuation index to obtain a first difference component; obtain a second difference component based on the sum of the absolute values ​​of the differences between each first fluctuation index; use the sum of the first difference component and the second difference component as the fluctuation difference degree; and use the sum of the absolute differences between each data point in the first data sequence and the average value of the second data sequence as the deviation component.

[0032] Based on this, this application can determine each fluctuation index through the fluctuation index determination subunit 1022 and determine the deviation component based on the deviation component determination subunit 1023, thereby enabling the aforementioned fault weight factor determination subunit 1021 to determine the fault weight factor based on each fluctuation index and the deviation component.

[0033] As an example, taking temperature data collected by a temperature sensor as an example, the specific implementation process for determining the fault weight factor described above can be as follows: Let the temperature data of the k-th temperature sensor for the most recent A days (e.g., the second time period mentioned above) be denoted as ,in, , and This represents the temperature data of the k-th temperature sensor at the 1st, 2nd, and 3rd sampling times within the most recent A days. This represents the temperature data at the nth sampling time. For example, in this embodiment, A is taken as an empirical value of 30. (The last sentence appears to be incomplete and possibly refers to a separate concept.) Then, linear interpolation was used to... The missing values ​​in the data are filled in, and the filled data sequence is denoted as . Due to sunlight, daytime temperatures are relatively high, while nighttime temperatures are relatively low. It will exhibit regular fluctuations. When the sequence When anomalies occur in the monitoring data, they could be caused by electromagnetic interference from solar panels or by a malfunctioning temperature sensor. If the anomaly is due to electromagnetic interference, it is usually rare, and the anomaly can be corrected to use the corrected data for forest fire prediction. However, if the anomaly is caused by a malfunctioning temperature sensor, there will typically be a large number of consecutive anomalies. Temperature data collected in this case has low reliability and cannot be used as forest fire monitoring data. Therefore, it is essential to first identify any faulty temperature sensors in the forest to eliminate the impact of faulty data on the accuracy of forest monitoring results.

[0034] Understandably, when a temperature sensor malfunctions, it can no longer detect changes in the actual temperature. Recent data may remain consistently stable or exhibit significant fluctuations, with outliers potentially far exceeding or falling below the actual temperature. Based on these characteristics, a fluctuation difference metric is constructed to determine the difference between the recent v-day temperature data and historical temperature data from the k-th temperature sensor. As an example, the fluctuation difference metric for the k-th temperature sensor... Satisfy the following formula: in, The first fluctuation index is the average of the absolute values ​​of all data in the first difference sequence of the temperature data sequence of the k-th temperature sensor over the past v days (e.g., the first time period mentioned above). Here, v is much smaller than A, and in this embodiment, v is taken as an empirical value of 2. It represents the average of the absolute values ​​of all data in the first-order difference sequence of the temperature data sequence of the k-th temperature sensor over the past A days (i.e., the second fluctuation index). This represents the average of the absolute values ​​of all data points in the first-order difference sequence of the temperature data sequence from the m-th temperature sensor over the past v days (i.e., a... (Corresponding to a third fluctuation index); w represents the total number of temperature sensors in the forest. Let be the standard deviation of the temperature data series over the past A days, used to measure the normal range of long-term fluctuations in temperature data. is the standard deviation of the temperature data sequence of the m-th temperature sensor over the past v days, used to measure the normal range of short-term fluctuations of the sensor.

[0035] The larger the value, the more significant the difference between the temperature data of the k-th temperature sensor over the past v days and the historical temperature data. The larger the value, the greater the difference in temperature fluctuation between the k-th temperature sensor and the m-th temperature sensor over the past v days.

[0036] A fault weighting factor is constructed to characterize the probability of the k-th temperature sensor failing. As an example, the fault weighting factor for the k-th temperature sensor is... Satisfy the following formula: in, This represents the degree of fluctuation difference of the k-th temperature sensor; Represents a sequence The temperature value at time i; j represents the time v days away from the current time; n represents the total number of sampling times in the last A days; Represents a sequence The average value of all temperature data.

[0037] when When the value is large, it indicates that the temperature data of the k-th temperature sensor over the past v days not only differs significantly from the temperature data of the k-th temperature sensor over the past A days, but also differs significantly from the temperature data of temperature sensors in other areas of the forest over the past v days, thus suggesting that the k-th temperature sensor is highly likely to malfunction; when When the value is large, it indicates that the temperature data of the k-th temperature sensor at each moment in the past v days differs significantly from the average temperature of the past A days, suggesting that the temperature data of the k-th temperature sensor in the past v days may be abnormal data caused by a fault.

[0038] In one possible implementation, the fault identification unit 102 is further configured to: generate a statistical sequence including fault weight factors for each monitoring device; calculate the mean μ and standard deviation σ of the statistical sequence; and if the standard deviation σ is less than a preset minimum positive number ε (e.g., ε = 1 × 10⁻⁶), then... -8 If the Z-score value is greater than the preset first threshold, then all monitoring devices are determined to be non-fault monitoring devices; otherwise, the Z-score value of the sequence is calculated to characterize the degree of deviation of the fault weight factor in the statistical sequence; monitoring devices with a Z-score value greater than the preset first threshold are identified as fault monitoring devices.

[0039] In other words, taking the aforementioned temperature sensor as an example, since temperature sensor failure is a low-probability event, typically only a few temperature sensors in the forest will fail within the same time period. Therefore, to determine whether any temperature sensor in the forest has failed at the current moment, the failure weight factor of all temperature sensors is calculated, and the sequence of all failure weight factors is denoted as U. The Z-score of each element in sequence U is calculated. If the Z-score of the failure weight factor of the k-th temperature sensor is greater than the threshold b (i.e., the aforementioned preset first threshold), then this temperature sensor is considered a failed sensor. If no failure weight factor has a Z-score greater than the threshold b, then there are no failed temperature sensors in the forest at the current moment. It should be noted that the threshold b can be 2 or 3, corresponding to the 95% or 99.7% confidence interval in statistics, respectively. Specifically, the value of the threshold b can be determined based on specific circumstances (such as statistical significance analysis based on historical failure data of forest monitoring equipment), and this application does not impose any limitations on this. It should be noted that, to avoid misjudging abnormal data in the early stages of a forest fire as sensor malfunctions, for candidate faulty sensors with a Z-score greater than threshold b, the system further performs multi-sensor fusion and spatial consistency checks: 1. Neighboring sensor collaborative analysis: Taking the candidate sensor as the center, calculate the average fault weight factor of all similar sensors within its radius R (e.g., 500 meters). If this average exceeds the global average by a certain percentage (e.g., 30%), it is determined that there is a general anomaly in the area, possibly indicating a regional fire, and the candidate sensor is not marked as faulty. 2. Smoke concentration verification: Simultaneously check whether the smoke concentration data of the candidate sensor in the same time period exceeds a preset warning threshold (e.g., 50 ppm). If the smoke concentration is abnormally high, it is preferentially regarded as a fire signal and not listed as a faulty device. 3. Comprehensive judgment: If the candidate sensor passes the above checks (i.e., there is no general anomaly in neighboring sensors and the smoke concentration is normal), it is finally judged as a faulty device, and its data is excluded; otherwise, it is marked as a fire candidate signal, triggering the fire review process, and the sensor's data is retained for fire monitoring.

[0040] Using the same method, fault diagnosis is performed on other types of sensors in the forest to identify humidity sensors, smoke concentration sensors, and light intensity sensors that are currently malfunctioning. The fault monitoring results of each sensor are transmitted to the monitoring and management cloud platform to remind forest management personnel to repair the sensors promptly.

[0041] Based on the above technical solution, this application comprehensively considers the consistency and deviation of data fluctuations, enabling the fault identification unit to more accurately assess the operating status of each monitoring device, which helps to discover potential faulty devices in advance, avoids the distortion of monitoring data caused by equipment failure, and improves the efficiency of system maintenance and the reliability of monitoring data.

[0042] Combination Figure 2 ,like Figure 3 As shown, in one possible implementation, the outlier correction unit 103 includes an electromagnetic interference anomaly determination subunit 1031. The electromagnetic interference anomaly determination subunit 1031 is used to: calculate the local reachability density of each data point in the data sequence of the target non-fault monitoring device based on a preset anomaly detection algorithm; the local reachability density is used to characterize the outlier degree of the data point's distribution in the data sequence; determine the illumination influence factor based on the ratio of the illumination intensity data of each data point at the same time to a preset threshold; the illumination influence factor is used to characterize the possibility of an electromagnetic interference source operating; calculate the absolute difference between each data point and the average value of its preceding and following adjacent data points to obtain an anomaly factor; the anomaly factor is used to characterize the degree of abrupt change in the data point; and determine the electromagnetic interference anomaly degree based on the local reachability density, the illumination influence factor, and the anomaly factor.

[0043] In one possible implementation, the outlier correction unit 103 includes an illumination influence factor determination subunit 1032, which is used to: determine the illumination influence factor to a first value when the illumination intensity data is greater than a preset threshold; the first value is used to characterize that the electromagnetic interference source is in a working state; and determine the illumination influence factor to a second value when the illumination intensity data is less than or equal to the preset threshold; the second value is used to characterize that the electromagnetic interference source is in a non-working state.

[0044] Based on this, this application can determine each fluctuation index through the fluctuation index determination subunit 1022 and determine the deviation component based on the deviation component determination subunit 1023, thereby enabling the aforementioned fault weight factor determination subunit 1021 to determine the fault weight factor based on each fluctuation index and the deviation component.

[0045] As one implementation method, taking data collected by a temperature sensor as an example, the process of determining the electromagnetic interference anomaly can be specifically as follows: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] As input to the Local Outlier Factor (LOF) anomaly detection algorithm (one of the aforementioned preset anomaly detection algorithms), the nearest neighbor number k is taken as an empirical value of 10, and the local reachability density of each data point is calculated. Based on the ratio of the illumination intensity data of each data point at the same time to a preset threshold, the illumination influence factor is determined; the illumination influence factor is used to characterize the possibility of an electromagnetic interference source operating; the absolute difference between each data point and the average value of its preceding and following adjacent data points is calculated to obtain the anomaly factor.

[0046] As an example, temperature series The illumination influence factor at time p satisfies the following formula: in, Temperature sequence The light intensity at time p; The minimum light intensity at which the solar panel inverter can operate (i.e., the aforementioned preset threshold; for example, the value of B can refer to the common photovoltaic inverter start-up threshold such as 2000 lux, or the value of B can be determined by combining the technical parameters provided by the selected inverter manufacturer or the average light intensity range of the forest area; this application does not limit this). When If the light intensity at time p is sufficient for the solar panel inverter to operate, then the temperature data at that time will be affected by electromagnetic interference. If the light intensity at time p is insufficient to power the inverter of the solar panel, then the temperature data at time p will not be affected by electromagnetic interference.

[0047] As yet another example, temperature sequences outlier at time p Satisfy the following formula: in, , and Representing temperature series The temperature data at the p-th, p-1-th, and p+1-th sampling times are used. If there is less than one data point before or after the data at the p-th sampling time, the outlier factor is not calculated. The larger the value, the greater the difference between the temperature at time p and the average temperature at times before and after it.

[0048] Based on the above characteristics, an electromagnetic interference anomaly degree is constructed to characterize the temperature sequence. The outlier of the data at time p. As an example, consider a temperature series. The electromagnetic interference anomaly degree at time p satisfies the following formula: in, Temperature sequence The illumination influence factor at the p-th time point; Temperature sequence Local reachability density of temperature data at time p. Temperature sequence The outlier factor at time p. In this formula... For parameter tuning coefficients, if If it is 0, then set it to 0.001. If it is not 0, then set it to 0.

[0049] It should be pointed out that, The larger the value, the higher the temperature series The greater the difference between the data at time p and the data at previous and subsequent times, the larger the temperature sequence... The data at time p is more likely to be outliers. When the value is 1, it indicates that the temperature sequence The data at time p may be affected by electromagnetic interference and could be outliers. When the value is 0, it indicates that the temperature sequence The data at time p is not subject to electromagnetic interference and cannot be an outlier. The smaller the value, the better the temperature series. The smaller the local reachability density of the data at time p, the better the temperature sequence. The data at time p is more likely to be outliers.

[0050] Based on the above technical solution, this application combines local reachability density, illumination influence factor and anomaly factor to construct electromagnetic interference anomaly degree, which can accurately identify instantaneous anomalies caused by electromagnetic interference, avoid the limitations of a single detection method, and improve the accuracy of data correction and environmental adaptability.

[0051] In one possible implementation, the outlier correction unit 103 is further configured to: use the electromagnetic interference anomaly degree of all data points in the target data sequence as an anomaly degree sequence; determine the segmentation threshold of the anomaly degree sequence based on a preset threshold segmentation algorithm, the segmentation threshold being used to distinguish between normal data and abnormal data in the anomaly degree sequence; for data points in the target data sequence whose electromagnetic interference anomaly degree is greater than the segmentation threshold, perform fire signal verification: if the smoke concentration data at the corresponding time of the data point exceeds the preset smoke concentration threshold, or if the temperature data of the sensor corresponding to the data point and more than half of the same type of sensors within a preset radius all show synchronous abrupt changes at the same time (for example, the anomaly factor Tp of each sensor at that time exceeds its historical normal fluctuation range), then retain the data point and not consider it an anomaly; otherwise, mark it as an anomaly point to be deleted. Deleting data points in the target data sequence whose electromagnetic interference anomaly degree is greater than the segmentation threshold yields a target data sequence with missing values; interpolating the target data sequence with missing values ​​based on a preset data interpolation algorithm yields a corrected data sequence.

[0052] As one implementation method, temperature sequences are calculated from temperature data. For each data point, the electromagnetic interference (EMI) anomaly degree is calculated. All EMI anomalies are used as input to the Otsu threshold segmentation algorithm to obtain a segmentation threshold. Temperature data at times corresponding to EMI anomalies exceeding the segmentation threshold are considered anomaly data. The process of determining the threshold based on the Otsu threshold segmentation algorithm can be referenced from existing technologies and will not be elaborated upon here.

[0053] After identifying the outliers, all outliers were removed from the temperature series. The abnormal data is deleted, and the temperature sequence after deleting the abnormal data is used as the input of the linear interpolation method to fill the missing values ​​in the sequence, so as to obtain the corrected temperature sequence of the k-th temperature sensor for the past A days.

[0054] Using the same method, outlier correction can be performed on the data sequences of other types of fault-free sensors in the forest to obtain the corrected data sequences of each fault-free humidity sensor, smoke concentration sensor, and light intensity sensor for nearly A days.

[0055] Based on the above technical solution, this application uses threshold segmentation and interpolation algorithms to correct outliers, which can automatically identify and repair outliers in the data sequence, ensuring the continuity and integrity of the data, providing a high-quality data foundation for subsequent prediction and analysis, and effectively improving the accuracy of monitoring results and system stability.

[0056] In one possible implementation, the monitoring result generation unit 104 is used to: predict the corrected data sequence of each non-fault monitoring device based on a preset time series prediction algorithm to obtain predicted data; compare the predicted data of each non-fault monitoring device with the corresponding preset warning threshold to determine the monitoring sub-result; identify the monitoring image based on a preset image recognition algorithm to determine the image analysis result of the target object; and perform decision fusion on the monitoring sub-result and the image analysis result to obtain the forest environment monitoring result. Optionally, the temperature warning threshold can be set to 40℃, the humidity warning threshold can be set to 30%RH, and the smoke concentration warning threshold can be set to 50ppm. The specific warning thresholds for each data can be determined with reference to the fire risk critical values ​​in relevant technical specifications, and this application does not limit this. For example, the decision fusion adopts the following rules: If the image analysis result detects fire characteristics such as open flame or continuous dense smoke, then regardless of the monitoring sub-result, it is determined that there is a fire risk, and a fire alarm monitoring result is generated; if the image analysis result does not detect fire characteristics, but at least two different types of environmental parameters (such as temperature and smoke concentration) in the monitoring sub-result simultaneously exceed their preset warning thresholds, then it is determined that there is a fire risk, and a fire alarm monitoring result is generated; if the image analysis result does not detect fire characteristics, and fewer than two environmental parameters in the monitoring sub-result exceed the warning threshold, then it is determined to be normal, and a normal monitoring result is generated. Optionally, to reduce the false negative rate, rule 2 can be adjusted to: if the image analysis result does not detect fire characteristics, but any environmental parameter in the monitoring sub-result exceeds its preset warning threshold, and this parameter continues to exceed the standard for multiple consecutive sampling periods, then it is also determined that there is a fire risk.

[0057] Optionally, after obtaining the corrected data sequences, the corrected data sequences and the monitoring images are transmitted to the monitoring result generation unit.

[0058] The monitoring result generation unit first uses the exponential moving average algorithm to predict the data for the next time step of all corrected data sequences, obtaining the predicted data for the next time step. The predicted data from each fault-free sensor at the next time step is then used as the monitoring result for the current time step. Next, image data collected by the monitoring camera is used as input to a pre-trained target detection model to obtain the forest environment monitoring result for the current time step.

[0059] Based on the above technical solution, this application integrates sensor data prediction and image recognition results, and uses decision generation to generate comprehensive monitoring results. It combines the advantages of different data sources, improves the comprehensiveness and redundancy of fire monitoring, reduces the risk of failure of a single data source, and enhances the system's early warning capability and robustness in complex scenarios.

[0060] Combination Figure 1 ,like Figure 4In one possible implementation, the system further includes a monitoring and management cloud platform 106. The early warning unit 105 is specifically used to compare forest environment monitoring results with preset alarm thresholds and determine whether to issue an alarm based on the comparison results. The monitoring and management cloud platform 106 is used to receive and store monitoring data collected by the data acquisition unit, corrected data sequences output by the outlier correction unit, and forest environment monitoring results generated by the monitoring result generation unit, and to display the received data in a visual format. Optionally, the temperature alarm threshold can be set to 45℃, and the smoke concentration alarm threshold can be set to 80ppm. The specific alarm thresholds for each sensor can be set based on the critical environmental conditions for forest fire occurrence or on the peak values ​​of monitoring data from historical fire cases; this application does not limit this.

[0061] Based on the above technical solution, this application integrates alarm units and a monitoring and management cloud platform to achieve automatic alarm and visualization of monitoring results, providing an intuitive user interface and decision support functions. This enables the system to respond to abnormal situations in a timely manner and facilitates monitoring by management personnel, thereby improving the system's practicality and operational efficiency, and ensuring the real-time nature and effectiveness of forest monitoring.

[0062] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A forest remote monitoring and early warning system based on the Internet of Things, characterized in that, The system includes: The data acquisition unit is used to collect real-time monitoring data of the forest environment through multiple monitoring devices deployed in the forest. The monitoring data includes temperature data, humidity data, smoke concentration data, light intensity data, and monitoring images. The monitoring devices communicate with the outside world through an Internet of Things (IoT) module. The fault identification unit is used to determine the fault weight factor of each monitoring device based on the data sequence of each monitoring device, and to identify faulty monitoring devices and non-faulty monitoring devices according to the distribution of the fault weight factors of the monitoring devices; the fault weight factor is used to characterize the probability of the monitoring device malfunctioning. An outlier correction unit is used to determine the electromagnetic interference anomaly degree of each non-fault monitoring device based on the distribution characteristics of outliers in the data sequence of each non-fault monitoring device, and to correct the outliers in the data sequence according to the electromagnetic interference anomaly degree to obtain a corrected data sequence; the electromagnetic interference anomaly degree is used to characterize the degree to which data points are affected by electromagnetic interference. The monitoring result generation unit is used to generate forest environment monitoring results based on the corrected data sequence; An early warning unit is used to issue early warnings based on the environmental monitoring results.

2. The forest remote monitoring and early warning system based on the Internet of Things according to claim 1, characterized in that, The fault identification unit includes a fault weight factor determination subunit: the fault weight factor determination subunit is used for: The first fluctuation index is determined based on the first data sequence from the first monitoring device within the first time period; The second fluctuation index is determined based on the second data sequence from the first monitoring device within the second time period; A third fluctuation index is determined based on the third data sequence from multiple second monitoring devices within the first time period; wherein the duration of the second time period is longer than the duration of the first time period, the first monitoring device and the second monitoring device are monitoring devices that collect the same type of data, and the fluctuation index is used to characterize the degree of data fluctuation. The degree of volatility difference is determined based on the difference between the first volatility index and the second volatility index, and the difference between the first volatility index and each of the third volatility indices. The deviation component is determined based on the degree of deviation between the data in the first data sequence and the average value of the second data sequence; The fault weight factor is determined based on the fluctuation difference and the deviation component.

3. The forest remote monitoring and early warning system based on the Internet of Things according to claim 2, characterized in that, The fault identification unit includes a fluctuation index determination subunit, which is used for: The average of the absolute values ​​of all data in the first difference sequence of the first data sequence is used as the first volatility index; The average of the absolute values ​​of all data in the first-order difference sequence of the second data sequence is used as the second volatility index; The average of the absolute values ​​of all data in the first-order difference sequence of the third data sequence is used as the third volatility index.

4. The forest remote monitoring and early warning system based on the Internet of Things according to claim 2, characterized in that, The fault identification unit includes a deviation component determination subunit, which is used for: Calculate the absolute value of the difference between each of the first volatility index and the second volatility index to obtain the first difference component; The second difference component is obtained by summing the absolute values ​​of the differences between the various first fluctuation indicators. The sum of the first difference component and the second difference component is taken as the fluctuation difference degree; The sum of the absolute differences between each data point in the first data sequence and the average value of the second data sequence is taken as the deviation component.

5. The forest remote monitoring and early warning system based on the Internet of Things according to claim 1, characterized in that, The fault identification unit is further configured to: Generate a statistical sequence including the fault weight factor for each of the monitoring devices; Calculate the Z-score value for each fault weight factor in the statistical sequence. The Z-score value is used to characterize the degree of deviation of the fault weight factor in the statistical sequence. Monitoring devices with Z-score values ​​greater than a preset first threshold are identified as fault monitoring devices.

6. The forest remote monitoring and early warning system based on the Internet of Things according to claim 1, characterized in that, The outlier correction unit includes an electromagnetic interference anomaly determination subunit, which is used for: Based on a preset anomaly detection algorithm, the local reachability density of each data point in the data sequence of the target non-fault monitoring device is calculated. The local reachability density is used to characterize the outlier degree of the data point distribution in the data sequence. The illumination influence factor is determined based on the ratio of the light intensity data of each data point at the same time to a preset threshold. The illumination influence factor is used to characterize the possibility of an electromagnetic interference source operating. The absolute difference between each data point and the average of its preceding and following data points is calculated to obtain the anomaly factor, which is used to characterize the degree of mutation of the data points; The electromagnetic interference anomaly degree is determined based on the local reachability density, the illumination influence factor, and the anomaly factor.

7. The forest remote monitoring and early warning system based on the Internet of Things according to claim 6, characterized in that, The outlier correction unit includes a light intensity influence factor determination subunit, which is used for: If the light intensity data is greater than the preset threshold, the value of the light influence factor is determined to be a first value; the first value is used to characterize that the electromagnetic interference source is in a working state. If the light intensity data is less than or equal to the preset threshold, the value of the light influence factor is determined to be a second value; the second value is used to characterize that the electromagnetic interference source is in a non-working state.

8. The forest remote monitoring and early warning system based on the Internet of Things according to claim 6, characterized in that, The outlier correction unit is further used for: The electromagnetic interference anomaly degree of all data points in the target data sequence is taken as the anomaly degree sequence; Based on a preset threshold segmentation algorithm, a segmentation threshold for the anomaly sequence is determined, and the segmentation threshold is used to distinguish between normal data and abnormal data in the anomaly sequence. Delete data points in the target data sequence whose electromagnetic interference anomaly degree is greater than the segmentation threshold to obtain a target data sequence with missing values; The target data sequence with missing values ​​is interpolated based on a preset data interpolation algorithm to obtain a corrected data sequence.

9. The forest remote monitoring and early warning system based on the Internet of Things according to claim 1, characterized in that, The monitoring result generation unit is used for: Based on a preset time series prediction algorithm, the corrected data sequence of each of the non-fault monitoring devices is predicted to obtain the predicted data. The predicted data of each non-fault monitoring device is compared with the corresponding preset warning threshold to determine the monitoring sub-result; The monitoring image is identified based on a preset image recognition algorithm to determine the image analysis results of the target object; The forest environment monitoring results are obtained by performing decision fusion on the monitoring sub-results and the image analysis results.

10. The forest remote monitoring and early warning system based on the Internet of Things according to claim 1, characterized in that, The system also includes a monitoring and management cloud platform; Specifically, the early warning unit is used to compare the forest environment monitoring results with a preset alarm threshold, and determine whether to issue an alarm based on the comparison results. The monitoring and management cloud platform is used to receive and store the monitoring data collected by the data acquisition unit, the corrected data sequence output by the outlier correction unit, and the forest environment monitoring results generated by the monitoring result generation unit, and to display the received data in a visual form.

Images