Internet of Things data early warning method based on multi-sensor cooperative monitoring

Abstract

The invention relates to the technical field of data processing, in particular to an Internet of Things data early warning method based on multi-sensor cooperative monitoring. The method comprises the following steps: acquiring data in a warehouse by using each sensor, the data comprising temperature, CO concentration and smoke concentration, and respectively constructing various data acquired every day into a group of sequences; and calculating the sum of the difference values between the monitoring values of the sensors at the current monitoring moment and the alarm threshold value. According to the method, temperature, CO concentration and smoke concentration are constructed into a sequence through multi-sensor data fusion, data difference value weighting is carried out by combining sensor credibility (based on historical data fluctuation variance) and monitoring time importance degree (historical alarm frequency normalization), and an early warning time period is dynamically delimited based on data curve inflection points. Therefore, potential signals of the fire risk can be found earlier, the accuracy of calculating the danger degree at the current moment is improved, and the false report and missing report rate is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing. More specifically, the present invention relates to an Internet of Things data early warning method for multi-sensor collaborative monitoring. Background Art

[0002] With the increasing complexity of the global supply chain, warehouse storage, as a core link in the supply chain, faces challenges in multi-source data management and security early warning. Traditional warehouse management mainly relies on manual inspections and single-sensor monitoring, making it difficult to effectively integrate multi-dimensional data such as temperature, humidity, and smoke concentration, resulting in a lag in early warnings for potential risks (such as fires and goods deterioration). With the rapid development of Internet of Things technology, warehouse storage management is gradually shifting towards intelligence and dataization. Modern warehouses not only need to store a large amount of goods but also need to monitor various environmental parameters in real time, such as temperature, humidity, and pressure, to ensure the safety of goods and the stability of storage conditions. Through multi-sensor collaborative monitoring, the Internet of Things can collect multi-source heterogeneous data in the warehouse, providing a rich information basis for storage management. These data include the environmental status collected by sensors, the operating conditions of equipment, and the locations of goods, jointly constituting the multi-data system of warehouse storage. The introduction of multi-data enables warehouse management to upgrade from traditional passive inspections to an active early warning mode, thereby improving efficiency and reducing risks.

[0003] Currently, when conducting fire early warnings for warehouses in the supply chain, most use a single sensor (such as only a temperature or smoke sensor) for fire early warning, which has significant limitations. First, a single sensor can only capture fire characteristics in a single dimension and is easily interfered by the environment (such as high-temperature weather or dust), resulting in false alarms, or missed alarms due to the lack of obvious early fire characteristics. Second, the lack of multi-dimensional data verification makes it impossible to comprehensively analyze the complex signs of fires (such as the coordinated changes in temperature, smoke, and carbon monoxide), reducing the accuracy and reliability of early warnings. Summary of the Invention

[0004] The present invention provides an Internet of Things data early warning method for multi-sensor collaborative monitoring, aiming to solve the problem of low accuracy of fire early warning using a single sensor in related technologies.

[0005] In a first aspect, the present invention provides an Internet of Things data early warning method for multi-sensor collaborative monitoring, including: collecting data in the warehouse using each sensor, where the data includes temperature, CO concentration, and smoke concentration, and constructing each type of data collected every day into a set of sequences; calculating the sum of the differences between the monitoring values of each sensor at the current monitoring moment and the alarm thresholds, and weighting the differences using the credibility of each sensor and the importance level of the current monitoring moment to obtain the risk level at the current monitoring moment, and giving an early warning based on the magnitude of the risk level; where the credibility of each sensor reflects the degree of fluctuation of the data collected by the sensor; calculating the importance level of each monitoring moment, including: counting the frequency of occurrence of the corresponding early warning time period at this monitoring moment for all alarm moments, and the importance level is positively correlated with the frequency, where the alarm moment refers to the alarm moment of the temperature sensor. This method dynamically calculates the risk level by weighting the difference between the monitoring value and the alarm threshold with the sensor credibility and the moment importance level. This method can adaptively adjust the early warning threshold according to real-time data and historical rules, reducing the false alarm rate and missed alarm rate.

[0006] Further, calculating the importance level of the monitoring moment includes: determining the early warning time period corresponding to each alarm moment; counting the frequency of occurrence of the early warning time period corresponding to the alarm moment at this monitoring moment, and taking the normalized frequency as the importance level of this monitoring moment. By counting the frequency of occurrence of the alarm moment or the early warning time period at each monitoring moment, the importance level of the moment is determined. This method is based on historical alarm rules and gives priority to high-risk time periods (such as moments with frequent alarms), making the early warning system more sensitive to critical moments and shortening the response time.

[0007] Further, determining the early warning time period includes: taking the time period between the alarm moment and the Nth monitoring moment as the early warning time period corresponding to this alarm moment, where N is a natural constant.

[0008] Further, determining the early warning time period includes: determining the early warning time period corresponding to each alarm moment according to the data change trend of each alarm moment.

[0009] Further, determining the early warning time periods corresponding to each alarm time includes: for any alarm time, obtaining the temperature sequence, smoke concentration sequence, and carbon monoxide sequence constructed from the data respectively before this alarm time; based on the change curves of temperature over time in the temperature sequence, smoke concentration over time in the smoke concentration sequence, and carbon monoxide concentration over time in the carbon monoxide sequence, obtaining the inflection points (points where the curve slope changes) in the three change curves; and determining the early warning time periods corresponding to each alarm time according to the inflection points. By analyzing the change curves of the temperature, smoke concentration, and carbon monoxide sequences, identifying the inflection points and determining the early warning time periods. This method can capture significant changes in the data trend (such as sudden slope changes), thereby detecting potential signals of fire risk earlier.

[0010] Further, determining the early warning time periods corresponding to each alarm time according to the inflection points includes: screening out the inflection point closest to this alarm time among all the inflection points of the three change curves as the target inflection point, and taking the time period between the monitoring time corresponding to the target inflection point and this alarm time as the early warning time period for this alarm time.

[0011] Further, determining the early warning time periods corresponding to each alarm time according to the inflection points includes: respectively screening out the inflection point closest to this alarm time from the three change curves, with one inflection point corresponding to one type of data, obtaining three final inflection points, namely the temperature inflection point, the smoke inflection point, and the carbon monoxide inflection point; constructing the initial time periods between each final inflection point and the alarm time, and calculating the mean of the three target time periods to obtain the target time period; and obtaining the time period equal in length to the target time period before this alarm time as the early warning time period for this alarm time.

[0012] Further, calculating the credibility of the temperature sensor includes: taking the historical temperature sequence when no fire occurs as the target sequence, obtaining multiple target sequences before the current day, and calculating the credibility of the temperature sensor according to the variances of the multiple target sequences. The credibility of the temperature sensor is inversely correlated with the difference between the variances of the multiple target sequences and the overall variance, where the variance can reflect the degree of fluctuation of the data collected by the sensor. This method calculates the credibility according to the degree of fluctuation of the sensor data (such as through variance analysis of the historical temperature sequence), and reduces the weight of the sensor data with lower stability. This mechanism effectively reduces the interference of noise or abnormal data on the early warning result and further improves the reliability of the early warning.

[0013] Further, determining the historical temperature sequence when no fire occurs includes: manually assigning labels to the temperature sequences of each historical day, where the labels include whether a fire has occurred or not.

[0014] Further, warnings are issued based on the degree of danger, including: if the degree of danger is greater than the alarm threshold, warnings are issued to relevant personnel. The abnormal degrees of temperature, CO concentration, and smoke concentration are combined with the credibility of the sensors and the importance of time periods to be converted into a unified degree-of-danger value. When this value exceeds the preset alarm threshold, a warning is triggered, avoiding complex decision-making for independent judgment of multiple parameters.

[0015] Beneficial effects: Through multi-sensor data fusion, temperature, CO concentration, and smoke concentration are constructed into sequences. Combining the credibility of the sensors (based on the variance of historical data fluctuations) with the importance degree of the monitoring moment (normalization of historical alarm frequencies), the data differences are weighted, and the warning time period is dynamically delimited based on the inflection points of the data curves, thereby discovering potential signals of fire risks earlier, improving the accuracy of calculating the degree of danger at the current moment, and reducing the false alarm and missed alarm rates. Description of the Drawings

[0016] Figure 1 It schematically shows a flowchart for calculating the degree of danger at the current monitoring moment according to an embodiment of the present invention. Detailed Embodiments

[0017] The following will describe in detail the specific embodiments of the present invention with reference to the drawings.

[0018] As Figure 1 shown, step S101: Collect data in the warehouse using multi-sensors.

[0019] In one embodiment, a variety of sensors are installed at appropriate positions in the warehouse, and a variety of sensors are used to monitor the storage environment in the warehouse. Specifically, the variety of sensors include temperature sensors, smoke sensors, and carbon monoxide sensors. The temperature, smoke concentration, and carbon monoxide concentration in the warehouse are respectively collected in real time using the temperature sensors, smoke sensors, and carbon monoxide sensors. Among them, the collection frequencies of the carbon monoxide sensors, temperature sensors, and smoke sensors are all once every 10 seconds. In other embodiments, the collection frequency can be once every 20 seconds or once every 30 seconds, which can be adjusted according to specific implementation situations. And the data collected every day is constructed into three groups of sequences, with one data type corresponding to one group of sequences. For example: For any day, the temperature sequence, carbon monoxide sequence, and smoke concentration sequence of that day can be obtained, and the lengths of the temperature sequence, carbon monoxide sequence, and smoke concentration sequence and the number of internal data are all the same. It should be noted that it is necessary to perform standardization processing on the collected variety of data to unify the dimension, facilitating analysis on a unified scale.

[0020] S102: Calculate the credibility of each sensor.

[0021] In one embodiment, taking the temperature sensor as an example, the credibility of the temperature sensor is calculated. Specifically, the temperature sequences of each historical day are obtained to get each historical temperature sequence, and labels are manually assigned to each historical temperature sequence. The labels are divided into two categories: fire has occurred or fire has not occurred. Therefore, the historical temperature sequences when the fire has occurred and the historical temperature sequences when the fire has not occurred can be obtained. Then, the historical temperature sequences when the fire has not occurred are screened out, and the credibility of the temperature sensor is calculated based on the historical temperature sequences when the fire has not occurred. The reason is as follows: Avoid using the sequences containing fire data to calculate the normal fluctuation range, and ensure the purity of the reference data for credibility evaluation. That is to say, the temperature sequence at the current time when the fire has occurred has a large fluctuation. Therefore, using the historical temperature sequences when the fire has occurred to calculate will result in a lower accuracy of the calculation result.

[0022] In one embodiment, the historical temperature sequences when the fire has not occurred are used as the target sequences, and multiple target sequences before the current day are obtained. The credibility of the temperature sensor is calculated according to the variance magnitudes of the multiple target sequences. The credibility of the temperature sensor is inversely correlated with the variance difference magnitudes of the multiple target sequences. Among them, the number of multiple target sequences can be 5 or 7, which can be adjusted according to specific circumstances. The greater the variance difference of the multiple target sequences, the greater the fluctuation of the data collected by the temperature sensor, and the lower the credibility of the temperature sensor. On the contrary, the higher the credibility of the temperature sensor.

[0023] In one embodiment, a calculation method for calculating the credibility of a temperature sensor is provided. The calculation method is as follows: . In the formula, represents the credibility of the temperature sensor, represents the variance of the target sequence on the th day before the current day, represents the mean value of the variances of all target sequences before the current day, represents the number of multiple target sequences before the current day. represents the difference between the variance of the target sequence on the th day before the current day and the mean value of the variances of all target sequences before the current day. The greater the difference, the greater the deviation degree of the variance of the target sequence on the th day from the overall variance mean value, which indicates that the fluctuation of the data collected by the temperature sensor on the th day is large, indicating that the temperature sensor has anomalies or a lot of noise. Therefore, the credibility of this sensor is lower. It should be noted that the reason for selecting the number of multiple target sequences before the current day for calculation is that the time distance of these days from the current day is relatively close, which can accurately reflect the operating state of the temperature sensor and avoid the influence of too old data on the current judgment result.

[0024] S103: Calculate the importance degree of each monitoring moment.

[0025] In one embodiment, according to the data acquisition frequencies of multiple sensors, monitoring times are determined. One acquisition frequency corresponds to one monitoring time, and the number of monitoring times per day is equal. For example, when the acquisition frequency is once a minute, 0:01, 0:02, and 0:03 are all monitoring times. By this method, all the monitoring times of the day can be obtained, and each monitoring time corresponds to a monitoring value. Thus, a sequence composed of the detection values of each day can be obtained. When using a temperature sensor to determine whether there is a fire hazard in a warehouse, early warning is carried out based on the monitoring value of the temperature sensor. When the monitoring value of the temperature sensor is greater than the alarm threshold, an alarm will be issued to warn the staff, so as to prevent the occurrence of a fire or detect a fire in a timely manner. Since the current time is recorded each time the temperature sensor in the warehouse alarms, the alarm times of the temperature sensor can be obtained. Although the alarm times of the temperature sensor may be different each time, some alarm times of fires are close. For any moment, if the number of times all alarm times are close to this monitoring time is large, it indicates that the coincidence degree between this monitoring time and all alarm times is high, which means that the possibility of a fire occurring at this monitoring time is large, and this monitoring time needs to be focused on subsequently. Therefore, the importance levels of each monitoring time can be calculated for subsequent key attention.

[0026] Specifically, when the temperature sensor triggers an alarm, the temperature in the warehouse is often significantly higher than the normal value. However, when a fire occurs, there are fire hazards or in the early stage of a fire (such as smoldering or slow combustion), and a combustion process (temperature rising stage) is required for early fire warning. Therefore, the time period composed of the alarm time of the temperature sensor to the Nth monitoring time is used as the early warning time period. In this embodiment, the value of N is 30. That is to say, within this early warning time period, the fire has started to occur, but the early fire does not reach the alarm threshold of the temperature sensor. Therefore, this early warning time period should also be focused on. Thus, the importance levels of each monitoring time can be determined in combination with the early warning time period. If the frequency of occurrence of all early warning time periods at any monitoring time is high, it indicates that this monitoring time is more important, and the importance level of this monitoring time is higher.

[0027] In one embodiment, calculating the importance levels of each monitoring time includes: for any monitoring time, counting the frequency of coincidence between the historical early warning time period and this monitoring time, and taking the normalized frequency as the importance level of this monitoring time.

[0028] It should be noted that for each alarm moment of the temperature sensor, a warning time period with a fixed length is used to calculate the importance degree of each monitoring moment. However, each time the temperature sensor alarms, the time interval between the fire ignition and the temperature sensor alarm is also different, because the actual time interval from the fire ignition to the temperature sensor alarm varies for different events. For example, some fires develop rapidly and may trigger an alarm within a few minutes; while some fires develop slowly and may take a longer time. Therefore, using a warning time period with a fixed length cannot adapt to the time characteristics of different fire events, resulting in some monitoring moments being wrongly included in the calculation. Thus, calculating the importance degree of these monitoring moments is redundant and will also affect the accuracy of calculating the importance degree of other monitoring moments. Herein, another embodiment is provided to adaptively obtain the warning time period corresponding to each alarm moment.

[0029] In another embodiment, a method for calculating the warning time period corresponding to each alarm moment of the temperature sensor is provided, which specifically includes: for any alarm moment, obtaining the sequences composed of the monitoring values collected by various sensors at all monitoring moments before this alarm moment on the same day, so as to obtain the temperature sequence, the smoke concentration sequence, and the carbon monoxide sequence before this alarm moment. And respectively obtaining the change curves of the temperature along the time sequence in the temperature sequence, the change curves of the smoke concentration along the time sequence in the smoke concentration sequence, and the change curves of the carbon monoxide concentration along the time sequence in the carbon monoxide sequence, and obtaining the inflection points (points where the curve slope changes) in the above three change curves. According to the inflection points, the warning time period corresponding to each alarm moment is determined. The reason is that the development of a fire is usually accompanied by a sudden change in the parameter change rate (such as a sudden increase in temperature and a sudden increase in smoke concentration). The inflection point can reflect the "starting point of the abnormal trend". Moving the warning period forward to the fire budding stage, compared with a fixed time window (N moments before the alarm), the dynamic time period based on the inflection point is more in line with the fire evolution law, and can discover the monitoring moments of "trend abnormal" in advance, thereby improving the accuracy of calculating the importance degree of each monitoring moment.

[0030] In one embodiment, determining the warning time period corresponding to each alarm moment according to the inflection points includes: selecting the inflection point closest to this alarm moment from all the inflection points in the three change curves as the target inflection point, and using the time period between the monitoring moment corresponding to the target inflection point and this alarm moment as the warning time period of this alarm moment. Thus, the warning time periods of all historical alarms can be obtained, and the importance degree of each monitoring moment can be obtained according to the warning time periods of all historical alarms.

[0031] In another embodiment, determining the early warning time periods corresponding to each alarm time according to the inflection points includes: respectively screening out the inflection points closest to the alarm time from the three change curves, where one type of data corresponds to one inflection point, obtaining three final inflection points, and the three final inflection points include the temperature inflection point determined from the change curve of the temperature along the time sequence in the temperature sequence, the smoke inflection point determined from the change curve of the smoke concentration along the time sequence in the smoke concentration sequence, and the carbon monoxide inflection point determined from the change curve of the carbon monoxide concentration along the time sequence in the carbon monoxide sequence. Construct the time periods between each final inflection point and the alarm time as initial time periods, calculate the mean of the three target time periods to obtain the target time period, and obtain the time period with the same length as the target time period before the alarm time as the early warning time period of the alarm time. By comprehensively considering the closest inflection points of the three types of parameters and taking the mean of the time periods to delimit the early warning interval, the interference of single-parameter fluctuations is avoided, and the critical time periods of multi-parameter collaborative monitoring are accurately locked.

[0032] S104: Calculate the risk level at the current monitoring time and determine whether to give an early warning.

[0033] In one embodiment, a single sensor can be used alone to calculate the risk level at the current monitoring time. Taking the temperature sensor as an example, specifically, according to the credibility of each temperature sensor calculated in the above steps, calculate the difference between the monitoring value of the temperature sensor at the current monitoring time and the alarm threshold, where the alarm threshold can be set manually, and weight the difference using the credibility of the temperature sensor and the importance level of the current monitoring time to obtain the risk level at the current monitoring time. The calculation formula is: . In the formula, represents the risk level at the current monitoring time, represents the monitoring value of the temperature sensor at the current monitoring time, represents the alarm threshold of the temperature sensor, represents the importance level of the current monitoring time, represents the credibility of the temperature sensor, represents the hyperbolic tangent function, where The smaller the value of , the closer the temperature at the current moment is to the alarm threshold, and the higher the risk level at the current moment; the larger , the higher the credibility of the sensor, and its measured value can more accurately reflect the actual temperature situation, and a greater weight should be given when calculating the risk level. It should be noted that the effect of this formula is to give an early warning based on the monitoring value at the current moment. When

[0034] In another embodiment, to avoid false alarms caused by single sensor failures or environmental interferences (for example, when the temperature sensor fluctuates briefly due to equipment heating, if its credibility is low, the weight is reduced), and at the same time, through multi-dimensional data cross-verification, the comprehensiveness of the calculation of the danger level is improved. Specifically, the data of multiple sensors can be combined to give an early warning of warehouse fires, and each of the multiple sensors has its own alarm threshold, which can be set manually. According to the above steps, the credibility of each sensor and the importance level at each monitoring moment can be calculated, and then the difference between the monitoring value of each sensor at the current monitoring moment and the alarm threshold is calculated, and the difference is weighted by the credibility of each sensor and the importance level at the current monitoring moment to obtain the danger level at the current monitoring moment. Specifically, a calculation formula for calculating the danger level at the current monitoring moment is provided, and the calculation formula is: . In the formula, represents the danger level at the current monitoring moment, represents the monitoring value of the th sensor at the current monitoring moment, represents the alarm threshold of the th sensor, represents the importance level at the current monitoring moment, represents the credibility of the th sensor, represents the number of sensors, represents the hyperbolic tangent function.

[0035] In one embodiment, when the danger level at the current monitoring moment is calculated, if the danger level at the current monitoring moment is greater than the danger threshold, it indicates that there is an abnormality in the warehouse, and an early warning is given to remind the staff. If the danger level at the current monitoring moment is less than or equal to the danger threshold, the warehouse can be continuously monitored. The empirical value of the danger threshold is 0.7. In other embodiments, the empirical value of the danger threshold can be 0.8 or 0.76, etc., and can be adjusted according to the actual situation.

[0036] The above embodiments only represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. An Internet of Things data early warning method for multi-sensor collaborative monitoring, characterized in that, Including: Collect data in the warehouse using various sensors, where the data includes temperature, CO concentration, and smoke concentration, and construct each type of data collected daily into a set of sequences; Calculate the sum of the differences between the monitoring values of each sensor at the current monitoring moment and the alarm thresholds, and weight the differences using the credibility of each sensor and the importance level of the current monitoring moment to obtain the risk level at the current monitoring moment, and issue a warning based on the magnitude of the risk level; Among them, the credibility of each sensor reflects the degree of fluctuation of the data collected by the sensor; Calculate the importance level of each monitoring moment, including: counting the frequency of occurrence of the early warning time periods corresponding to all alarm moments at this monitoring moment, and the importance level is positively correlated with the frequency, where the alarm moment refers to the alarm moment of the temperature sensor.

2. The method for warning of Internet of Things data in multi-sensor collaborative monitoring according to claim 1, characterized in that, Calculate the importance level of the monitoring moment, including: Determine the early warning time periods corresponding to each alarm moment; Count the frequency of occurrence of the early warning time periods corresponding to the alarm moment at this monitoring moment, and use the normalized frequency as the importance level of this monitoring moment.

3. The method for warning of Internet of Things data by multi-sensor collaborative monitoring according to claim 2, characterized in that, Determine the early warning time period, including: Take the time period between the alarm moment and the Nth monitoring moment as the early warning time period corresponding to this alarm moment, where N takes a natural constant value.

4. The method for warning of Internet of Things data in multi-sensor collaborative monitoring according to claim 2, characterized in that, Determine the early warning time period, including: Determine the early warning time periods corresponding to each alarm moment according to the data change trend of each alarm moment.

5. The Internet of Things data early warning method for multi-sensor collaborative monitoring according to claim 4, characterized in that Determine the early warning time periods corresponding to each alarm moment, including: For any alarm moment, obtain the temperature sequence, smoke concentration sequence, and carbon monoxide sequence constructed from the data before this alarm moment; Based on the change curve of temperature over time in the temperature sequence, the change curve of smoke concentration over time in the smoke concentration sequence, and the change curve of carbon monoxide concentration over time in the carbon monoxide sequence, obtain the inflection points in the three change curves, where the inflection point is the point where the curve slope changes; Determine the early warning time periods corresponding to each alarm moment according to the inflection points.

6. The Internet of Things data early warning method for multi-sensor collaborative monitoring according to claim 5, characterized in that Determine the early warning time periods corresponding to each alarm moment according to the inflection points, including: Screen out the inflection point closest to this alarm moment among all the inflection points of the three change curves as the target inflection point, and take the time period between the monitoring moment corresponding to the target inflection point and this alarm moment as the early warning time period of this alarm moment.

7. The method for warning of Internet of Things data in multi-sensor collaborative monitoring according to claim 5, characterized in that, Determine the early warning time periods corresponding to each alarm moment according to the inflection points, including: Respectively screen out the inflection point closest to this alarm moment from the three change curves, where one type of data corresponds to one inflection point, to obtain three final inflection points, and the three final inflection points include a temperature inflection point, a smoke inflection point, and a carbon monoxide inflection point; Construct the initial time periods between each final inflection point and the alarm moment, and calculate the average value of the three target time periods to obtain the target time period; Obtain the time period with the same length as the target time period before this alarm moment as the early warning time period of this alarm moment.

8. The method for warning of Internet of Things data in multi-sensor collaborative monitoring according to claim 1, wherein Calculate the credibility of the temperature sensor, including: Taking the historical temperature sequence when a fire has not occurred as the target sequence, obtaining multiple target sequences before the current day, and calculating the credibility of the temperature sensor according to the variance magnitude of the multiple target sequences. The credibility of the temperature sensor is inversely correlated with the difference between the variance of the multiple target sequences and the overall variance, where the variance can reflect the degree of fluctuation of the data collected by the sensor.

9. The method for warning of Internet of Things data in multi-sensor collaborative monitoring according to claim 1, characterized in that, Determining the historical temperature sequence when a fire has not occurred, including: Manually assigning temperature sequence labels to each historical day, where the labels include whether a fire has occurred or not.

10. The method for warning of Internet of Things data by multi-sensor collaborative monitoring according to any one of claims 1-9, characterized in that, Issuing a warning based on the magnitude of the danger level, including: If the danger level is greater than the alarm threshold, warning relevant personnel.

Images