Sensor data detection system and detection method of monitoring system

Abstract

The invention provides a sensor data detection system and detection method of a monitoring system. The detection system comprises a server and a plurality of sensors. The server is provided with a data processing module, a numerical value classification operation module and a numerical value comparison module. The data processing module performs a data cleaning procedure on the sensing data of the plurality of sensors to generate a plurality of cleaned data. The numerical value classification operation module calculates a difference value combination of the cleaned data of a plurality of pens of every two sensors according to a specified sequence direction. The numerical value comparison module subtracts the measurement precision from the difference values of the difference value combinations to obtain a plurality of second difference value combinations, calculates a first characteristic value of each second difference value combination, and calculates a p value corresponding to the first characteristic value of any second difference value combination when the p value corresponding to the first characteristic value of any second difference value combination is smaller than a preset value. A marker is recorded for the two sensors associated with this second difference combination.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to Taiwan Patent Application No. 112123457 filed on June 21, 2023, which is incorporated herein by reference for all purposes as if fully set forth herein. Technical Field

[0003] The present invention relates to a detection system, and in particular to a detection system and a detection method for detecting sensor data. Background Art

[0004] In recent years, in order to achieve the goal of net zero emissions, many places have begun to conduct carbon emission analysis and carbon reduction strategy planning. In order to conduct effective analysis and planning, it is necessary to collect data over a period of time through sensors installed in areas or buildings. Therefore, the accuracy of sensor data is very important for the long-term goal of net zero emissions.

[0005] Traditional buildings will build a central monitoring system for the purpose of centralized control and management of electromechanical systems. Under the framework of the central monitoring system, the building will be equipped with many sensors at the same time to achieve precise control operation results. However, if the traditional central monitoring system wants to ensure the accuracy of the sensing data of these sensors, it can only define simple upper and lower limits or perform sensor calibration regularly to ensure the accuracy of sensor data. However, the above approach is difficult to avoid the risk of system control inaccuracy due to deviation of certain sensor data during daily operation.

[0006] Therefore, how to detect abnormal sensors based on the analysis and comparison of sensor data, and thereby remind users to go and eliminate or repair them, is an issue that urgently needs to be studied in this technical field. Summary of the invention

[0007] The main purpose of the present invention is to provide a sensor data detection system and detection method for a monitoring system, which can mark at least two sensors with significant differences in sensing data, thereby discovering abnormal sensors in real time.

[0008] In order to achieve the above-mentioned purpose, the detection system of the present invention comprises:

[0009] A plurality of sensors correspond to the same specific category and respectively generate a plurality of sensing data;

[0010] A plurality of output-input modules are respectively connected to the plurality of sensors and continuously collect the plurality of sensing data of the plurality of sensors;

[0011] A server is connected to the plurality of input and output modules through a network communication device to receive the plurality of sensing data from the plurality of sensors, and comprises:

[0012] a data processing module configured to perform a data cleaning process on the plurality of sensing data of the plurality of sensors to respectively generate a plurality of cleaned data of the plurality of sensors;

[0013] a numerical classification operation module configured to calculate a plurality of differences of the plurality of cleaned data of any two of the plurality of sensors according to a specified order direction to establish a plurality of difference combinations; and

[0014] A numerical comparison module is configured to obtain a measurement accuracy of the specific category, subtract the measurement accuracy from the multiple differences in each of the difference combinations to obtain multiple second difference combinations, normalize each of the second difference combinations, and calculate a first eigenvalue of each of the second difference combinations, obtain a p value corresponding to each of the second difference combinations based on each of the second difference combinations and the corresponding first eigenvalue, and record a mark for the two sensors associated with the second difference combination when the p value corresponding to the first eigenvalue of one of the multiple second difference combinations is less than a first preset value.

[0015] As described above, the data processing module is configured to execute the following procedures to perform the data cleaning procedure on the plurality of sensing data of the plurality of sensors:

[0016] Setting a standard working condition to initially filter the plurality of sensing data;

[0017] Deleting blank value data rows in the plurality of sensing data;

[0018] Deleting abnormal value data rows in the plurality of sensing data; and

[0019] A data feature value of the plurality of sensing data in a specific historical time interval is calculated, and outlier data rows in the plurality of sensing data are deleted based on the data feature value.

[0020] As mentioned above, the data characteristic value is the average value or geometric mean value of the plurality of sensing data in the specific historical time interval.

[0021] As described above, the numerical classification operation module is configured to determine whether the plurality of differences in each difference combination are all positive numbers after calculating the plurality of difference combinations, and to perform a normalization process on the plurality of difference combinations when any of the plurality of difference values ​​are negative numbers.

[0022] As described above, the normalization process includes obtaining a minimum difference among the plurality of difference values ​​of the plurality of difference value combinations, and subtracting the minimum difference from all the difference values ​​in the plurality of difference value combinations.

[0023] As described above, the numerical comparison module is configured to first calculate a second characteristic value of the corresponding difference combination and obtain the corresponding p value based on the difference combination and the second characteristic value when the p value corresponding to the first characteristic value of one of the multiple second difference combinations is less than the first preset value, and then record the mark for the two sensors associated with the second difference combination when the p value corresponding to the second characteristic value is less than a second preset value.

[0024] As described above, the numerical comparison module is configured to first calculate a variance of the difference combination and obtain a corresponding p-value based on the difference combination and the variance when the p-value of the first characteristic value of one of the multiple second difference combinations is less than the first preset value and the p-value corresponding to the second characteristic value is less than the second preset value, and to record the mark for the two sensors associated with the second difference combination when the p-value corresponding to the variance is less than a third preset value.

[0025] As described above, the first eigenvalue is the expected value, average value or geometric mean of the multiple differences in the second difference combination, and the second eigenvalue is the expected value, average value or geometric mean of the multiple differences in the difference combination.

[0026] As described above, the server also includes an error counting module, which is configured to record a detection failure mark for each of the two sensors having the mark, record an error count for any sensor when the detection failure mark of any sensor in this detection cycle accumulates to a first preset number, and issue an alarm when the error count of the sensor accumulates to a second preset number.

[0027] In order to achieve the above object, the detection method of the present invention is applied to the above detection system and comprises the following steps:

[0028] a) continuously collecting, by the server, a plurality of sensing data from the plurality of sensors;

[0029] b) classifying the plurality of sensing data by the server to summarize the plurality of sensing data of the plurality of sensors belonging to a specific category;

[0030] c) performing a data cleaning procedure on the plurality of sensing data of the plurality of sensors by the server to respectively generate a plurality of cleaned data of each of the sensors;

[0031] d) calculating, by the server, a plurality of differences of the plurality of cleaned data of any two of the plurality of sensors according to a specified sequential direction to establish a plurality of difference combinations;

[0032] e) obtaining a measurement accuracy of the specific category from the server;

[0033] f) the server subtracts the measurement accuracy from the plurality of differences in each of the difference combinations to obtain a plurality of second difference combinations;

[0034] g) The server normalizes each of the second difference combinations, calculates a first eigenvalue of each of the second difference combinations, and obtains a p value corresponding to each of the second difference combinations based on each of the second difference combinations and the corresponding first eigenvalue, and records a mark for the two sensors associated with the second difference combination when the p value corresponding to the first eigenvalue of one of the plurality of second difference combinations is less than a first preset value.

[0035] As mentioned above, wherein the step c) comprises:

[0036] c1) setting a standard working condition to initially filter the plurality of sensing data;

[0037] c2) deleting blank value data rows in the plurality of sensing data;

[0038] c3) deleting abnormal value data rows in the plurality of sensing data;

[0039] c4) calculating a data characteristic value of the plurality of sensing data in a specific historical time interval; and

[0040] c5) deleting outlier data rows in the plurality of sensing data based on the data feature values ​​to generate the plurality of cleaned data.

[0041] As mentioned above, the data characteristic value is the average value or geometric mean value of the plurality of sensing data in the specific historical time interval.

[0042] As mentioned above, wherein the step d) comprises:

[0043] d1) subtracting the plurality of post-cleaning data of any two of the plurality of sensors according to the specified sequential direction to generate the plurality of difference values ​​and establish the plurality of difference value combinations;

[0044] d2) determining whether the plurality of difference values ​​in the plurality of difference value combinations are all positive numbers; and

[0045] d3) When any of the differences is a negative number, a normalization process is performed on the plurality of difference combinations.

[0046] As described above, step d3) includes obtaining a minimum difference of the plurality of difference values ​​in the plurality of difference value combinations, and subtracting the minimum difference from all the difference values ​​in all the difference value combinations to complete the normalization process.

[0047] As mentioned above, wherein the step g) comprises:

[0048] g1) determining whether the p value corresponding to the first characteristic value of each second difference combination is less than the first preset value;

[0049] g2) when the p value corresponding to the first eigenvalue of one of the plurality of second difference combinations is less than the first preset value, calculating a second eigenvalue of the corresponding difference combination and obtaining the corresponding p value according to the difference combination and the second eigenvalue;

[0050] g3) determining whether the p value corresponding to the second eigenvalue is less than a second preset value; and

[0051] g4) when the p value corresponding to the second characteristic value is less than the second preset value, recording the mark for the two sensors associated with the second difference value combination.

[0052] As mentioned above, wherein the step g4) comprises:

[0053] g41) when the p-value corresponding to the second eigenvalue is less than the second preset value, calculating a variance of the corresponding difference combination and obtaining the corresponding p-value according to the difference combination and the variance;

[0054] g42) determining whether the p-value corresponding to the variance is less than a third preset value; and

[0055] g43) When the p value corresponding to the variance is less than the third preset value, the mark is recorded for the two sensors associated with the second difference combination.

[0056] As described above, the first eigenvalue is the expected value, average value or geometric mean of the multiple differences in each of the second difference combinations, and the second eigenvalue is the expected value, average value or geometric mean of the multiple differences in each of the difference combinations.

[0057] As mentioned above, this also includes:

[0058] h) the server records a detection failure mark for each of the two sensors having the mark;

[0059] i) the server determines whether any sensor among the multiple sensors has a detection failure mark that accumulates to a first preset number in this detection cycle;

[0060] j) when the detection failure marks of any sensor in this detection cycle accumulate to the first preset number, record an error count for the sensor; and

[0061] k) The server issues a warning when the cumulative number of errors of the sensor reaches a second preset number.

[0062] The present invention obtains sensing data of multiple sensors of the same category, and analyzes and compares the sensing data according to their interdependence. When a significant difference is detected between the sensing data of two sensors, an abnormality warning can be issued in real time to notify the user to perform relevant detection actions on the sensors, thereby eliminating the impact of the sensing data of the abnormal sensor on the system.

[0063] In order to further understand the techniques, means and effects adopted by the present invention to achieve the predetermined objectives, please refer to the following detailed description and drawings of the present invention. It is believed that the objectives, features and characteristics of the present invention can be deeply and specifically understood thereby. However, the drawings are only provided for reference and illustration and are not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is an embodiment of a block diagram of a detection system of the present invention;

[0065] Figure 2 is an embodiment of a flow chart of the detection method of the present invention;

[0066] Figure 3 An embodiment of a data cleaning flow chart of the present invention;

[0067] Figure 4 An embodiment of the numerical calculation flow chart of the present invention;

[0068] Figure 5 is an embodiment of a numerical comparison flow chart of the present invention;

[0069] Figure 6 An embodiment of a block diagram of a server of the present invention;

[0070] Figure 7 is an embodiment of the error warning flow chart of the present invention;

[0071] Figure 8 Schematic diagram of an embodiment of the significant differences of the present invention.

[0072] Wherein, the reference numerals are:

[0073] 1-Detection system;

[0074] 2- Server;

[0075] 21-data processing module;

[0076] 22-Numerical operation module;

[0077] 23-Numerical comparison module;

[0078] 24- Error counting module;

[0079] 3- Sensor;

[0080] 31- output and input module;

[0081] 4- Network communication equipment;

[0082] S21-S28-detection steps;

[0083] S31-S35-cleaning steps;

[0084] S41~S44-operation steps;

[0085] S51-S56-comparison step;

[0086] S71~S75-warning steps. DETAILED DESCRIPTION

[0087] In this specification, many specific details are provided to provide a thorough understanding of the specific embodiments of the present invention; however, those skilled in the art should know that the present invention can still be practiced without one or more of these specific details; in other cases, well-known details are not shown or described to avoid obscuring the main technical features of the present invention. The technical content and detailed description of the present invention are described as follows with the help of the drawings:

[0088] To plan energy conservation and emission reduction for an environment (such as a building), it is necessary to set up a large number of sensors in the environment and perform statistics and analysis on the sensor data generated by these sensors so as to plan strategies. However, when using these sensor data, it is necessary to first detect these sensor data and eliminate sensor data that are significantly different from other data, so as to avoid establishing wrong strategies due to insufficient accuracy of sensor data.

[0089] First Read Figure 1 , which is an embodiment of a block diagram of a detection system of the present invention. The present invention discloses a sensor data detection system (hereinafter referred to as detection system 1), such as Figure 1 As shown, the detection system 1 mainly includes a server 2 and a plurality of sensors 3, wherein each sensor 3 has a corresponding input / output module (hereinafter referred to as an IO module) 31. The server 2 is connected to a network communication device 4 via a network protocol, and is connected to the IO module 31 of each sensor 3 via the network communication device 4.

[0090] The detection system 1 of the present invention can be mainly set up for a specific environment (such as a building), and the building is provided with multiple sensors 3 of many categories, such as thermometers, hygrometers, air quality detectors, etc., without limitation. One of the technical features of the present invention is that the server 2 continuously receives the sensing data of all sensors 3 through the network communication device 4, and compares the sensing data of multiple sensors 3 of the same category, so as to detect whether there is an abnormal sensor among the multiple sensors 3 of the same category.

[0091] As described above, the server 2 in the present invention compares the sensing data of multiple sensors 3 of the same category. Therefore, if the detection system 1 includes n types of sensors 3, the server 2 will perform n detection actions to respectively detect whether there are abnormal sensors among the multiple sensors 3 of various categories.

[0092] For ease of understanding, the following description will be given by taking an example where the server 2 is connected to a plurality of sensors 3 of a single category via the network communication device 4 .

[0093] The multiple sensors 3 are respectively arranged in multiple same / different areas in the same environment (e.g., a building) to sense the areas and generate corresponding sensing data. In one embodiment, each sensor 3 senses according to a set sensing cycle (e.g., every 100ms, 10ms, or 1s as a cycle) to generate corresponding sensing data. Specifically, each sensor 3 generates multiple sensing data in one sensing cycle, not limited to one sensing data.

[0094] Each sensor 3 has at least one IO module 31, and each IO module 31 obtains sensing data generated by the sensor 3 when the sensor 3 is in operation. In one embodiment, the IO module 31 is a wireless transmission interface, and the sensor 3 is wirelessly connected to the network communication device 4 through the IO module 31. In another embodiment, the IO module 31 is a wired transmission interface, and the sensor 3 is connected to the network communication device 4 through the IO module 31 through a transmission line.

[0095] The server 2 is connected to the network communication device 4 through the network, and is connected to each sensor 3 through the network communication device 4. In the present invention, the server 2 can continuously receive sensing data from multiple sensors 3 through the network communication device 4 and multiple IO modules 31.

[0096] exist Figure 1 In the embodiment, the server 2 is divided into a data processing module 21, a numerical calculation module 22 and a numerical comparison module 23 according to the functions to be implemented. It is worth mentioning that the detection system 1 of the present invention can also have a database (not shown in the figure), and the server 2 can store the collected sensing data in the database.

[0097] In one embodiment, the data processing module 21, the numerical calculation module 22 and the numerical comparison module 23 are hardware modules implemented by physical components, for example, they can be implemented by a microcontroller unit (MCU), a central processing unit (CPU), a system on chip (SoC) or a programmable logic controller (PLC). In another embodiment, the server 2 records a computer executable program code. When the server 2 is started and executes the computer executable program code, the data processing module 21, the numerical calculation module 22 and the numerical comparison module 23 can be created and run according to the required functions. That is, the modules 21-23 can be software modules, but are not limited thereto.

[0098] The data processing module 21 is used to perform a data cleaning procedure on the sensing data of the multiple sensors 3 before performing the analysis and comparison procedure to generate a plurality of cleaned data. If each sensor 3 generates a plurality of pen sensing data, the data processing module 21 generates a plurality of pen cleaned data for each sensor 3 through the data cleaning procedure.

[0099] In one embodiment, the server 2 receives the sensing data of all types of sensors 3 through the network communication device 4. Before performing the analysis and comparison procedure, the server 2 first classifies all the sensing data through the data processing module 21 to aggregate the sensing data of the same type into the same data group. For example, the data processing module 21 can aggregate the sensing data of multiple flow meters into a first data group, aggregate the sensing data of multiple thermometers into a second data group, and so on. After completing the classification, the data processing module 21 performs a data cleaning procedure for multiple sensing data of the same type.

[0100] In the data cleaning process, the data processing module 21 filters out blank data, abnormal data and outlier data in a data group, thereby reducing the deviation caused by these noises when performing the analysis and comparison process.

[0101] The numerical classification operation module 22 is used to obtain a plurality of cleaned data processed by the data processing module 21, and compare the plurality of cleaned data to generate corresponding comparison results. The detection system 1 of the present invention mainly detects one or more abnormal sensors 3 based on the comparison results.

[0102] Specifically, the numerical classification operation module 22 mainly calculates the difference between every two cleaned data in a data group according to the specified order direction, and generates a difference combination. That is, if the data group contains the sensing data of n sensors 3 of the same category, the numerical classification operation module 22 will calculate a difference combination. For example, if the data group contains the sensing data of four sensors 3 of the same category, the numerical classification operation module 22 will calculate six difference combinations (i.e.,).

[0103] More specifically, each difference value combination may include multiple difference values ​​based on the number of sensing data pieces measured by the sensors 3 in each sensing cycle. For example, if the sensors 3 generate 100 sensing data pieces in each sensing cycle, each difference value combination may include 100 difference values.

[0104] When calculating the difference, the numerical classification operation module 22 first selects any two sensors 3 and obtains all the post-cleaning data of the two sensors 3 (the number of records should be the same). Then, the numerical classification operation module 22 selects one post-cleaning data from each of the two sensors 3 (for example, based on the sensing time) and subtracts them to generate a difference. When all the post-cleaning data of the two sensors 3 are subtracted two by two, the numerical classification operation module 22 can generate a plurality of difference values, which can form a difference combination associated with the two sensors 3.

[0105] One of the technical features of the present invention is that the server 2 can detect whether the sensing data groups between any two sensors 3 have significant differences by performing an analysis and comparison procedure based on multiple difference value combinations. In this way, the server 2 can obtain multiple sensors 3 that are abnormal or suspected of being abnormal based on whether the sensing data group has significant abnormalities.

[0106] The value comparison module 23 is used to obtain the measurement accuracy of a specific type (such as a flow meter, a thermometer or a hygrometer, etc.), and detect abnormal data based on the measurement accuracy.

[0107] Specifically, the measurement accuracy is information that has been determined after the sensor 3 of this type is manufactured, and can be provided by the manufacturer of the sensor 3 and imported into the detection system 1. It is worth mentioning that all sensors 3 of the same type have the same measurement accuracy.

[0108] In the present invention, the numerical comparison module 23 can mainly subtract the measurement accuracy corresponding to this category from each difference in each difference combination obtained by the numerical classification operation module 22 to obtain multiple second difference combinations. Then, the numerical comparison module 23 calculates the first eigenvalue of each second difference combination, and obtains the corresponding p-value based on each second difference combination and its first eigenvalue. In one embodiment, the first eigenvalue can be, for example, an expected value, an average value, or a geometric mean, etc., without limitation. The present invention can observe whether multiple sensing data of this category (i.e., multiple sensor data after cleaning) have significant differences by detecting the expected value, average value, or geometric mean of each second difference combination.

[0109] The above-mentioned p value is common knowledge in the relevant technical field and will not be described in detail here. If the p value corresponding to the first characteristic value of a second difference combination is less than the first preset value, it means that the sensing data groups of the two sensors 3 corresponding to the second difference combination have significant differences; if the p value corresponding to the first characteristic value of a second difference combination is not less than the first preset value, it means that the sensing data groups of the two sensors 3 corresponding to the second difference combination do not have significant differences.

[0110] Specifically, the present invention can calculate multiple difference values ​​(forming a difference combination) and a characteristic value based on multiple sensing data of two sensors, and finally obtain a corresponding p value, thereby, this p value can determine whether there are significant differences in the multiple sensing data of the two sensors. For example, the first thermometer of the present invention generates n sensing data in a sensing cycle, and the second thermometer also generates n sensing data in the same sensing cycle. If the p value obtained based on these sensing data is too small (for example, less than the first preset value), it means that there are significant differences between the n sensing data of the first thermometer and the n sensing data of the second thermometer. However, it is still impossible to determine whether the problem is with the first thermometer or the second thermometer, so both thermometers need to be marked first, and then judged in the subsequent procedure (to be described in detail later).

[0111] Please also see Figure 8 ,in Figure 8 Schematic diagram of the significant differences of the present invention. Figure 8 In the embodiment of , the two thermometers were tested before October 2017 and officially operated from November 2017. Figure 8As can be seen in the figure, starting from April 2018 (in the box), one of the thermometers (e.g., the first thermometer) became abnormal, causing a significant difference between the data of the first thermometer and the data of the other thermometer (e.g., the second thermometer). After June 2018, the first thermometer was replaced, so the data returned to normal, and there was no significant difference between the data of the first thermometer and the data of the second thermometer.

[0112] One of the technical features of the present invention is that if the value comparison module 23 determines after detection that the p value corresponding to the first characteristic value of a difference combination is less than the first preset value, it means that there is a significant difference between the sensing data groups of the two sensors 3 corresponding to the difference combination. At this time, the value comparison module 23 records a mark for the two sensors 3. Through this mark, the detection system 1 can detect abnormal sensors from multiple sensors 3 (to be described in detail later).

[0113] Please also refer to Figure 2 , is an embodiment of a flow chart of the detection method of the present invention. Figure 2 The specific implementation steps of the sensor data detection method of the present invention (hereinafter referred to as the detection method) are disclosed, and the detection method is mainly applied to Figure 1 The detection system 1 is shown, but is not limited thereto.

[0114] like Figure 2 As shown, after the detection system 1 is started, the server 2 continuously collects the sensing data of the connected multiple sensors 3 (step S21), and the server 2 classifies the multiple sensing data (step S22), thereby classifying the sensing data of the multiple sensors 3 of the same category into the same data group.

[0115] Before performing the analysis and comparison procedures for the multiple sensing data of each category, the server 2 performs a data cleaning procedure on the multiple sensing data of the same category to generate multiple cleaned data (step S23). For ease of understanding, the following is an example of grouping multiple cleaned data of the same category into one data group and performing the analysis and comparison procedure on multiple cleaned data in a single data group.

[0116] After step S23, the server 2 calculates a plurality of difference combinations of the plurality of post-cleaning data of each two sensors in a data group of a specific category according to the specified sequential direction (step S24), wherein each difference combination includes a plurality of difference values ​​of the plurality of post-cleaning data of the two sensors. For example, if the data group includes one hundred first post-cleaning data generated by the first sensor in the first sensing cycle, one hundred second post-cleaning data generated by the second sensor in the first sensing cycle, and one hundred third post-cleaning data generated by the third sensor in the first sensing cycle, then the plurality of difference combinations include a difference combination A recording one hundred difference values ​​of the first post-cleaning data of the first sensor and the second post-cleaning data of the second sensor, a difference combination B recording one hundred difference values ​​of the second post-cleaning data of the second sensor and the third post-cleaning data of the third sensor, and a difference combination C recording one hundred difference values ​​of the first post-cleaning data of the first sensor and the third post-cleaning data of the third sensor.

[0117] Next, the server 2 obtains the measurement accuracy of this specific category (step S25), and subtracts the measurement accuracy obtained in step S25 from each difference in the plurality of difference combinations obtained in step S24 to obtain a plurality of second difference combinations (step S26).

[0118] Since each difference may be negative after subtracting the measurement accuracy, after step S26, the server 2 first normalizes the multiple differences in the multiple second difference combinations, and then respectively calculates the first eigenvalue of each second difference combination (step S27). In one embodiment, the server 2 calculates the expected value of the multiple differences in each second difference combination in step S27. In one embodiment, the server 2 calculates the average value of the multiple differences in each second difference combination in step S27. In one embodiment, the server 2 calculates the geometric mean of the multiple differences in each second difference combination in step S27. However, the above are only some specific implementation examples of the present invention, but are not limited to this.

[0119] After step S27, the server 2 obtains the p value corresponding to each second difference combination based on the multiple differences in each second difference combination and the first characteristic value, and determines whether the p value is less than the first preset value. In one embodiment, the first preset value may be, for example, 0.05, 0.03 or 0.01, etc., without limitation. If the p value corresponding to the first characteristic value of a second difference combination is less than the first preset value, the server 2 determines that the sensing data groups of the two sensors 3 associated with the second difference combination have significant differences. At this time, the server 2 records a mark for the two sensors 3 whose sensing data groups have significant differences (step S28).

[0120] Through the mark recorded in step S28, the server 2 can learn about the sensor 3 that is suspected to be abnormal, and then determine whether it is necessary to issue an alarm for the abnormal sensor (to be described in detail later).

[0121] See also Figure 3 , which is an embodiment of the data cleaning flow chart of the present invention. Figure 3 To use for Figure 2 The data cleaning procedure in step S23 is further explained.

[0122] like Figure 3 As shown, when the server 2 executes the data cleaning program through the data processing module 21, a standard working condition can be set manually by the user or automatically by the server 2 to initially filter multiple sensing data in a data group (step S31). Specifically, in step S31, the data processing module 21 filters out sensing data of multiple sensors 3 with the same working condition from the data group.

[0123] For example, the standard operating condition can be set as being powered on, operating, and not undergoing maintenance within a certain time period (e.g., working hours from Monday to Friday). In this way, the server 2 can pre-eliminate one or more sensors 3 that are not powered on, powered on but not operating, or undergoing maintenance during this time period. Because the sensing data of the sensors 3 that do not meet the standard operating conditions are not used, misjudgment by the server 2 can be avoided.

[0124] Next, the data processing module 21 deletes the blank value data rows in the plurality of sensing data from the data group (step S32), and deletes the abnormal value data rows in the plurality of sensing data (step S33). In one embodiment, the abnormal value is, for example, data recorded as N / A, garbled code, special symbol or noise in the data field, but is not limited thereto.

[0125] After step S32 and step S33, the data processing module 21 calculates the data characteristic values ​​of the plurality of sensing data in the set specific historical time interval (step S34). In one embodiment, the data characteristic value may be, for example, the average value or geometric mean of the plurality of sensing data in the specific historical time interval, but is not limited thereto.

[0126] For example, the user can manually or automatically set the specific historical time interval to the fifth month through the server 2, and accordingly, the data processing module 21 will calculate the data characteristic value based on the sensing data of the first four months and the sensing data of the fifth month. In addition, the data processing module 21 deletes the outlier data rows from the multiple sensing data based on the data characteristic value (step S35), thereby generating multiple cleaned data. For example, the data processing module 21 can delete one or more sensing data that exceeds three standard deviations (Standard Deviation, SD) of the data characteristic value in step S35, but is not limited to this. In the present invention, the server 2 performs subsequent analysis and comparison procedures based on multiple cleaned data.

[0127] See also Figure 4 , which is an embodiment of the numerical calculation flow chart of the present invention. Figure 4 To use for Figure 2 Step S24 is described in further detail.

[0128] like Figure 4 As shown, the server 2 obtains the multiple cleaned data processed by the data processing module 21 through the numerical calculation module 22, and subtracts the multiple cleaned data belonging to the two sensors 3 to generate multiple difference values ​​and form a difference combination (step S41). For example, if the two sensors 3 each have one hundred cleaned data, the numerical calculation module 22 subtracts the multiple cleaned data one hundred times in step S41 to generate one hundred difference values ​​and form a difference combination. In addition, the numerical calculation module 22 determines whether all the difference combinations have been calculated and established (step S42), that is, whether the sensing data groups of the two sensors 3 have been subtracted according to the specified sequential direction and the corresponding difference combinations have been generated. If the difference combination has not been calculated and established, the numerical calculation module 22 executes step S41 again.

[0129] For example, the server 2 is connected to four sensors 3 of the same type, including a first sensor S1, a second sensor S2, a third sensor S3, and a fourth sensor S4. In step S41, the numerical calculation module 22 sequentially calculates a difference combination D12 (i.e., S2-S1), a difference combination D13 (i.e., S3-S1), a difference combination D14 (i.e., S4-S1), a difference combination D23 (i.e., S3-S2), a difference combination D24 (i.e., S4-S2), and a difference combination D34 (i.e., S4-S3). That is, if the number of sensors 3 of the same type is n, the numerical calculation module 22 will calculate difference combinations, each of which includes a plurality of differences between the sensing data groups of two sensors 3.

[0130] After all difference combinations are calculated and established, the numerical calculation module 22 further determines whether all the differences in these difference combinations are positive numbers (step S43). If all the differences are positive numbers (including zero), the numerical calculation module 22 does not perform any action, and the server 2 then executes Figure 2 If at least one difference is negative, the numerical calculation module 22 normalizes all the differences in the plurality of difference combinations (step S44). Furthermore, the server 2 executes the following after step S44: Figure 2 Steps S25 to S28.

[0131] In one embodiment, the numerical calculation module 22 mainly obtains the minimum difference in the plurality of difference combinations in step S44, and then subtracts the minimum difference from all the difference values ​​in the plurality of difference combinations to complete the normalization process. Specifically, if the numerical calculation module 22 determines in step S43 that it is no, the minimum difference in the plurality of difference combinations must be a negative number. Accordingly, after all the difference values ​​in all the difference combinations are subtracted from the minimum difference, the minimum difference will be corrected to zero, and the remaining difference values ​​will be corrected to positive numbers greater than zero. In this way, the numerical calculation module 22 can complete the normalization process of the plurality of difference values.

[0132] For example, the difference combination D12 includes differences such as 2, 1, -1, -2, 4, and 0, and the minimum difference is -2. In this embodiment, the numerical calculation module 22 can subtract -2 from all the differences in all the difference combinations. The difference values ​​in the difference combination D12 after calculation include 2-(-2)=4, 1-(-2)=3, -1-(-2)=1, -2-(-2)=0, 4-(-2)=6, and 0-(-2)=2. As described above, the numerical calculation module 22 can make all the differences after normalization processing positive numbers.

[0133] It is worth mentioning that if all the differences in a difference combination (such as D12 or D23, etc.) are positive numbers, then this difference combination does not need to be normalized.

[0134] In the present invention, the server 2 performs the subsequent analysis and comparison procedure based on the normalized difference combination. By excluding the difference using negative numbers, the complexity of the analysis and comparison procedure can be reduced.

[0135] See also Figure 5 , which is an embodiment of the numerical comparison flow chart of the present invention. Figure 5 To use for Figure 2 Step S28 is described in more detail.

[0136] exist Figure 2In the embodiment of the present invention, the server 2 calculates a plurality of second difference combinations based on the plurality of difference combinations and the measurement accuracy of the target category, and then respectively calculates the first characteristic values ​​of these second difference combinations. Furthermore, the server 2 generates a p value corresponding to each second difference combination based on each second difference combination and its first characteristic value, and when the p value of any second difference combination is less than the first preset value, directly records a corresponding mark for the two sensors 3 associated with this second difference combination. Figure 2 The above procedure can be regarded as the first stage of the analysis and comparison procedure performed by the server 2 through the numerical comparison module 23. In other words, Figure 2 In the embodiment, the value comparison module 23 directly marks two sensors 3 whose sensing data groups have significant differences after only the first stage of analysis and comparison procedures.

[0137] exist Figure 5 In the embodiment of the present invention, the value comparison module 23 will mark two sensors 3 whose sensing data groups have significant differences after a multi-stage analysis and comparison procedure.

[0138] Specifically, in Figure 5 In the embodiment of the invention, after calculating the first characteristic value of each second difference combination, the numerical comparison module 23 will generate the p value corresponding to each second difference combination based on each second difference combination and its first characteristic value, and first determine whether the p value corresponding to each first characteristic value is less than the first preset value (step S51). If the p value corresponding to the first characteristic value of a second difference combination is not less than the first preset value, it means that there is no significant difference between the sensing data groups of the two sensors 3 associated with this second difference combination, so the numerical comparison module 23 does not mark the sensor 3.

[0139] exist Figure 5 In the embodiment, even if it is determined in step S51 that the p value corresponding to the first characteristic value of any second difference combination is less than the first preset value, the value comparison module 23 still does not perform the marking action temporarily, but first continues to perform the second stage analysis and comparison procedure.

[0140] Specifically, the numerical comparison module 23 calculates the second characteristic value of each difference combination in the second stage analysis and comparison procedure (step S52), wherein the second characteristic value can be, for example, the expected value, average value or geometric mean of multiple differences in each difference combination, but is not limited thereto.

[0141] After calculating the second eigenvalue of each difference combination, the numerical comparison module 23 obtains the p value corresponding to each difference combination based on each difference combination and its second eigenvalue, and determines whether the p value corresponding to each second eigenvalue is less than the second preset value (step S53). In one embodiment, the second preset value may be equal to the first preset value. In another embodiment, the second preset value is any value between 0.01 and 0.05, but is not limited thereto.

[0142] In one embodiment, the numerical comparison module 23 can determine that the sensing data groups of the two sensors 3 associated with the difference combination have significant differences when the p value corresponding to the second characteristic value of any difference combination is less than the second preset value. At this time, the numerical comparison module 23 records a mark for the two sensors 3 with significant differences in the sensing data groups. In another embodiment, when the numerical comparison module 23 determines that the p value corresponding to the second characteristic value of any difference combination is less than the second preset value, it does not perform the marking action temporarily, but continues to perform the third stage of the analysis and comparison procedure.

[0143] In the third phase of the analysis and comparison program, the numerical comparison module 23 calculates the variance of the multiple differences of each difference combination (step S54). After calculating the variance of the multiple differences of each difference combination, the numerical comparison module 23 obtains the p value corresponding to each difference combination based on each difference combination and its variance, and determines whether the p value corresponding to each variance is less than a third preset value (step S55). In one embodiment, the third preset value is less than the first preset value and the second preset value. In another embodiment, the third preset value is 0.01, but is not limited thereto.

[0144] In this embodiment, when the p-value corresponding to the variance of any difference combination is less than the third preset value, the numerical comparison module 23 determines that the sensing data groups of the two sensors 3 associated with the second difference combination (or the corresponding difference combination) have significant differences. At this time, the numerical comparison module 23 records a mark for the two sensors 3 (step S56).

[0145] The present invention sequentially determines the p-value of the first characteristic value of each second difference combination, the p-value of the second characteristic value of each difference combination, and the p-value of the variance of each difference combination, and the determination of the p-value can understand the distribution of the data difference between each two sensors 3. In this way, the server 2 can confirm whether there is a significant difference between the sensing data groups of each sensor 3.

[0146] If the server 2 determines that the sensing data groups of any two sensors 3 have significant differences in the first phase of the analysis and comparison procedure, the server 2 can then execute the second phase of the analysis and comparison procedure; if the server 2 determines that the sensing data groups of the two sensors 3 have significant differences in the second phase of the analysis and comparison procedure, the server 2 can then execute the third phase of the analysis and comparison procedure. The present invention detects multiple sensing data groups of multiple sensors 3 through multiple analysis and comparison procedures like a sieve, which can avoid misjudgment due to overly rough detection conditions.

[0147] In one embodiment, the server 2 may perform the above analysis and comparison procedure based on a preset detection cycle, for example, once a month, but not limited thereto. Taking the detection cycle as one month as an example, the marks recorded by the server 2 may be as shown in Table 1 below:

[0148]

[0149] Table 1

[0150] In the embodiment of Table 1 above, the server 2 performs the above analysis and comparison procedure with each month as the detection cycle, wherein in January, the difference combination D12, D13 and D14 is judged to have significant differences and is marked (i.e., the multiple sensors 3 associated with the difference combination D12, D13 and D14 are marked), in February, the difference combination D24 is judged to have significant differences and is marked (i.e., the multiple sensors 3 associated with the difference combination D24 are marked), in March, the difference combination D12, D13 and D14 is judged to have significant differences and is marked. In April, the difference combination D12, D13 and D14 is judged to have significant differences and is marked (that is, the multiple sensors 3 associated with the difference combination D12, D13 and D14 are marked), and in June, the difference combination D12, D13 and D14 is judged to have significant differences and is marked (that is, the multiple sensors 3 associated with the difference combination D12, D13 and D14 are marked).

[0151] Through the above marking, the server 2 can detect one or more sensors 3 that are suspected to be abnormal. For example, if the difference combination D12 is marked, it means that the first sensor S1 and the second sensor S2 may be abnormal; if the difference combination D24 is marked, it means that the second sensor S2 and the fourth sensor S4 may be abnormal.

[0152] Please also refer to Figure 6 and Figure 7 ,in Figure 6 An embodiment of a block diagram of a server of the present invention is shown in FIG. Figure 7 This is an embodiment of the error warning flow chart of the present invention.

[0153] exist Figure 6 In the embodiment of the present invention, the server 2 further has an error counting module 24. As described above, after executing one or more stages of analysis and comparison procedures, the value comparison module 23 will mark multiple sensors 3 with significant abnormalities in the sensing data group (for example, as shown in the above table, the sensors 3 associated with the difference combinations D12, D13, D14 and D24 are marked). A difference combination is associated with two sensors 3 of the same category. In the present invention, the error counting module 24 records the marks for the corresponding two sensors 3 based on the marks of a difference combination.

[0154] like Figure 7 As shown, in the present invention, the server 2 records a detection failure mark for each of the two sensors 3 corresponding to the difference value combination with a mark through the error counting module 24 (step S71). In one embodiment, the two sensors 3 corresponding to the difference value combination without a mark may also be recorded with a detection failure mark, but the detection failure mark is 0.

[0155] Specifically, the detection failure mark may be as shown in Table 2 below:

[0156]

[0157] Table 2

[0158] As shown in Table 2 above, server 2 marked the difference combinations D12, D13 and D14 in the analysis and comparison program in January. Therefore, in the aforementioned step S71, the error counting module 24 records three detection failure marks for the first sensor S1 (corresponding to the difference combinations D12, D13 and D14), records one detection failure mark for the second sensor S2 (corresponding to the difference combination D12), records one detection failure mark for the third sensor S3 (corresponding to the difference combination D13) and records one detection failure mark for the fourth sensor S4 (corresponding to the difference combination D14).

[0159] For another example, the server 2 marks the difference combination D24 in the analysis and comparison program in February, so in the aforementioned step S71, the error counting module 24 records a detection failure mark for the second sensor S2 and a detection failure mark for the fourth sensor S4.

[0160] Back to Figure 7After the marking is completed (including the marking of the difference combination and the marking of the sensor 3), the server 2 determines whether the detection failure mark of any sensor 3 has accumulated to a first preset number in this detection cycle (step S72). In one embodiment, the first preset number can be, for example, three, but is not limited to this. When the detection failure mark of any sensor 3 in this detection cycle accumulates to a first preset number, the error counting module 24 records the number of errors for this sensor 3 (step S73).

[0161] If the first preset number is three, in the embodiment presented in Table 2 above, the first sensor S1 has a detection failure mark accumulated three times in the detection cycles of January, March, April, and June. In this embodiment, the error counting module 24 will record the number of errors for the first sensor S1 once in the analysis and comparison procedures of January, March, April, and June.

[0162] Specifically, the number of errors may be shown in Table 3 below:

[0163]

[0164] Table 3

[0165] It can be seen from Table 3 above that the error counting module 24 in the present invention calculates the number of errors in units of each detection cycle. If the accumulated detection failure marks of a sensor (such as the first sensor S1) in this detection cycle reach the first preset number, the error counting module 24 will record the number of errors for this sensor 3 once; if the accumulated detection failure marks of a sensor (such as the second sensor S2, the third sensor S3 and the fourth sensor S4) in this detection cycle do not reach the first preset number, the error counting module 24 will not record the number of errors.

[0166] In the present invention, when entering the next detection cycle (ie, when the analysis and comparison procedures are performed next time), the detection failure marks of each sensor 3 will not be accumulated, but the number of errors will be accumulated.

[0167] like Figure 7 As shown, after each detection cycle, the error counting module 24 determines whether the error times of any sensor 3 among the multiple sensors 3 have reached a second preset number (step S74). In one embodiment, the second preset number is, for example, three times, but not limited thereto.

[0168] In one embodiment, the counting of the number of errors does not need to be continuous, but can be accumulated continuously before the corresponding sensor 3 is repaired or replaced.

[0169] When it is determined in step S74 that the accumulated error count of any sensor 3 reaches the second preset number, the server 2 may issue an alert for the sensor 3 (step S75). In this way, the user can know that the sensor 3 is abnormal or suspected to be abnormal, and then repair or replace the sensor 3. After the sensor 3 is repaired or replaced, the error counting module 24 will reset the error count of the sensor 3 to zero and re-accumulate it.

[0170] The present invention collects sensing data of multiple sensors of the same category under the same working conditions through a detection system, and after data cleaning, uses a multi-stage analysis and comparison procedure from large to small like a sieve to perform detection based on the data dependency. If some sensors are found to have problems during a detection procedure, the system can mark the sensors and accumulate them to the next detection procedure. When the cumulative number exceeds the preset number, the system will actively issue an abnormal warning to notify the user to perform relevant detection operations on the sensors.

[0171] Compared to traditional systems that require regular maintenance of sensors or the definition of upper and lower limits, the present invention can detect abnormal sensors in real time through data analysis and comparison. In this way, the present invention can ensure that the analysis and planning based on the sensing data will not cause serious distortion or errors due to abnormal or uncalibrated sensors.

[0172] However, the above is only a preferred embodiment of the present invention, and it cannot limit the scope of the present invention. That is, all equivalent changes and modifications made according to the present invention should still fall within the scope of the patent coverage of the present invention. The present invention can also have many other embodiments. Without departing from the spirit and essence of the present invention, those familiar with the art can make various corresponding changes and deformations according to the present invention, but these corresponding changes and deformations should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A sensor data detection system for a monitoring system, characterized in that: include: A plurality of sensors correspond to the same specific category and respectively generate a plurality of sensing data; A plurality of output-input modules are respectively connected to the plurality of sensors and continuously collect the plurality of sensing data of the plurality of sensors; A server is connected to the plurality of input and output modules through a network communication device to receive the plurality of sensing data from the plurality of sensors, and comprises: a data processing module configured to perform a data cleaning process on the plurality of sensing data of the plurality of sensors to respectively generate a plurality of cleaned data of the plurality of sensors; a numerical classification operation module configured to calculate a plurality of differences of the plurality of cleaned data of any two of the plurality of sensors according to a specified order direction to establish a plurality of difference combinations; and A numerical comparison module is configured to obtain a measurement accuracy of the specific category, subtract the measurement accuracy from the multiple differences in each of the difference combinations to obtain multiple second difference combinations, normalize each of the second difference combinations, and calculate a first eigenvalue of each of the second difference combinations, obtain a p value corresponding to each of the second difference combinations based on each of the second difference combinations and the corresponding first eigenvalue, and record a mark for the two sensors associated with the second difference combination when the p value corresponding to the first eigenvalue of one of the multiple second difference combinations is less than a first preset value.

2. The sensor data detection system of the monitoring system according to claim 1, characterized in that: The data processing module is configured to execute the following procedures to perform the data cleaning procedure on the plurality of sensing data of the plurality of sensors: Setting a standard working condition to initially filter the plurality of sensing data; Deleting blank value data rows in the plurality of sensing data; Deleting abnormal value data rows in the plurality of sensing data; and A data feature value of the plurality of sensing data in a specific historical time interval is calculated, and outlier data rows in the plurality of sensing data are deleted based on the data feature value.

3. The sensor data detection system of the monitoring system according to claim 2, characterized in that: The data characteristic value is an average value or a geometric mean value of the plurality of sensing data in the specific historical time interval.

4. The sensor data detection system of the monitoring system according to claim 1, characterized in that: The numerical classification operation module is configured to determine whether the plurality of differences in each difference combination are all positive numbers after calculating the plurality of difference combinations, and to perform a normalization process on the plurality of difference combinations when any of the difference values ​​is a negative number.

5. The sensor data detection system of the monitoring system according to claim 4, characterized in that: The normalization process includes obtaining a minimum difference among the plurality of difference values ​​of the plurality of difference value combinations, and subtracting the minimum difference from all the difference values ​​in the plurality of difference value combinations.

6. The sensor data detection system of the monitoring system according to claim 1, characterized in that: The numerical comparison module is configured to first calculate a second characteristic value of the corresponding difference combination and obtain a corresponding p value based on the difference combination and the second characteristic value when the p value corresponding to the first characteristic value of one of the multiple second difference combinations is less than the first preset value, and to record the mark for the two sensors associated with the second difference combination when the p value corresponding to the second characteristic value is less than a second preset value.

7. The sensor data detection system of the monitoring system according to claim 6, characterized in that: The numerical comparison module is configured to first calculate a variance of the difference combination and obtain a corresponding p-value based on the difference combination and the variance when the p-value corresponding to the first characteristic value of one of the multiple second difference combinations is less than the first preset value and the p-value corresponding to the second characteristic value is less than the second preset value, and to record the mark for the two sensors associated with the second difference combination when the p-value corresponding to the variance is less than a third preset value.

8. The sensor data detection system of the monitoring system according to claim 7, characterized in that: The first eigenvalue is the expected value, average value or geometric mean of the multiple differences in the second difference combination, and the second eigenvalue is the expected value, average value or geometric mean of the multiple differences in the difference combination.

9. The sensor data detection system of the monitoring system according to claim 1, characterized in that: The server also includes an error counting module, which is configured to record a detection failure mark for each of the two sensors having the mark, record an error count for one of the sensors when the detection failure mark in the current detection cycle accumulates to a first preset number, and issue a warning when the error count of the sensor accumulates to a second preset number.

10. A sensor data detection method for a monitoring system, applied to a detection system, the detection system having a plurality of sensors and a server connected to the plurality of sensors via a network communication device, characterized in that: The test method includes: a) continuously collecting, by the server, a plurality of sensing data from the plurality of sensors; b) classifying the plurality of sensing data by the server to summarize the plurality of sensing data of the plurality of sensors belonging to a specific category; c) performing a data cleaning procedure on the plurality of sensing data of the plurality of sensors by the server to respectively generate a plurality of cleaned data of each of the sensors; d) calculating, by the server, a plurality of differences of the plurality of cleaned data of any two of the plurality of sensors according to a specified sequential direction to establish a plurality of difference combinations; e) obtaining a measurement accuracy of the specific category from the server; f) the server subtracts the measurement accuracy from the plurality of differences in each of the difference combinations to obtain a plurality of second difference combinations; g) The server normalizes each of the second difference combinations, calculates a first eigenvalue of each of the second difference combinations, and obtains a p value corresponding to each of the second difference combinations based on each of the second difference combinations and the corresponding first eigenvalue, and records a mark for the two sensors associated with the second difference combination when the p value corresponding to the first eigenvalue of one of the plurality of second difference combinations is less than a first preset value.

11. The sensor data detection method of the monitoring system according to claim 10, characterized in that: This step c) comprises: c1) setting a standard working condition to initially filter the plurality of sensing data; c2) deleting blank value data rows in the plurality of sensing data; c3) deleting abnormal value data rows in the plurality of sensing data; c4) calculating a data characteristic value of the plurality of sensing data in a specific historical time interval; and c5) deleting outlier data rows in the plurality of sensing data based on the data feature values ​​to generate the plurality of cleaned data.

12. The sensor data detection method of the monitoring system according to claim 11, characterized in that: The data characteristic value is an average value or a geometric mean value of the plurality of sensing data in the specific historical time interval.

13. The sensor data detection method of the monitoring system according to claim 10, characterized in that: The step d) comprises: d1) subtracting the plurality of post-cleaning data of any two of the plurality of sensors according to the specified sequential direction to generate the plurality of difference values ​​and establish the plurality of difference value combinations; d2) determining whether the plurality of difference values ​​in the plurality of difference value combinations are all positive numbers; and d3) When any difference is a negative number, a normalization process is performed on the plurality of difference combinations.

14. The sensor data detection method of the monitoring system according to claim 13, characterized in that: Step d3) includes obtaining a minimum difference value of the plurality of difference values ​​in the plurality of difference value combinations, and subtracting the minimum difference value from all the difference values ​​in all the difference value combinations to complete the normalization process.

15. The sensor data detection method of the monitoring system according to claim 10, characterized in that: The step g) comprises: g1) determining whether the p value corresponding to the first characteristic value of each second difference combination is less than the first preset value; g2) when the p value corresponding to the first eigenvalue of one of the plurality of second difference combinations is less than the first preset value, calculating a second eigenvalue of the corresponding difference combination and obtaining the corresponding p value according to the difference combination and the second eigenvalue; g3) determining whether the p value corresponding to the second eigenvalue is less than a second preset value; and g4) when the p value corresponding to the second characteristic value is less than the second preset value, recording the mark for the two sensors associated with the second difference value combination.

16. The sensor data detection method of the monitoring system according to claim 15, characterized in that: The step g4) comprises: g41) when the p-value corresponding to the second eigenvalue is less than the second preset value, calculating a variance of the corresponding difference combination and obtaining the corresponding p-value according to the difference combination and the variance; g42) determining whether the p-value corresponding to the variance is less than a third preset value; and g43) When the p value corresponding to the variance is less than the third preset value, the mark is recorded for the two sensors associated with the second difference combination.

17. The sensor data detection method of the monitoring system according to claim 16, characterized in that: The first eigenvalue is the expected value, average value or geometric mean of the multiple differences in each of the second difference combinations, and the second eigenvalue is the expected value, average value or geometric mean of the multiple differences in each of the difference combinations.

18. The sensor data detection method of the monitoring system according to claim 10, characterized in that: Also includes: h) the server records a detection failure mark for each of the two sensors having the mark; i) the server determines whether any sensor among the multiple sensors has a detection failure mark that accumulates to a first preset number in this detection cycle; j) when the detection failure marks of any sensor in this detection cycle accumulate to the first preset number, record an error count for the sensor; and k) The server issues a warning when the cumulative number of errors of the sensor reaches a second preset number.

Images