A sensor data analysis system for underwater explosion detection

Abstract

The present invention relates to the technical field of sensor data analysis, and in particular to a sensor data analysis system for underwater explosion detection, comprising an information acquisition module, an underwater model construction module, a feature association analysis module, a real-time data analysis module and an abnormality confirmation module. The system determines the normal overall association value and the phenomenon overall association value of the sensor by analyzing the data collected by the sensor in a normal state and the data collected by the sensor after the underwater explosion respectively, calculates the real-time phenomenon association value and the real-time normal association value of the real-time collected sensor data, and determines the device trust value of the sensor based on the real-time phenomenon association value and the real-time normal association value. The abnormal sensor is then determined based on the device trust value. The accuracy of sensor data collection for underwater explosion detection is improved through data fusion and feature analysis in multi-source devices.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor data analysis, and in particular to a sensor data analysis system for underwater explosion detection. Background Art

[0002] Underwater explosion is a common marine disaster that has a serious impact on the marine environment and human activities. Therefore, timely and accurate detection of underwater explosions is crucial to prevent and mitigate their impact.

[0003] Prior art CN112557500A discloses an underwater elastic wave full-wavefield nondestructive testing system and method, the system comprising: an excitation component for applying an impact force to the surface of an underwater structure to be tested to excite an elastic wave field; a force sensor for measuring the impact force information of the excitation component; a detection component for sensing the elastic wave full-wavefield information of the surface of the structure to be tested in three directions, namely, X, Y and Z, and the pressure wavefield information of the water on the surface of the structure to be tested; a data analysis module for performing inversion analysis on the impact force information, the elastic wave full-wavefield information and the pressure wavefield information to obtain a damage model corresponding to the structure to be tested.

[0004] When detecting underwater explosions, since the sensors are in an underwater environment, they are usually operated and maintained within a fixed periodic time. However, in the early stage of a sensor failure, the data offset is low and the system does not detect the sensor abnormality in time. At this time, the system will analyze the data collected by the abnormal sensor as normal data, which will cause errors in the data analysis results, thereby reducing the accuracy of the system analysis results. Summary of the invention

[0005] The purpose of the present invention is to solve the problems in the background technology and to propose a sensor data analysis system for underwater explosion detection.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A sensor data analysis system for underwater explosion detection, comprising:

[0008] An information collection module is used to collect underwater information of the target detection area and transmit it to the underwater model building module. The underwater information includes sensor information and water area information;

[0009] An underwater model construction module is used to construct an underwater model, wherein the underwater model includes a three-dimensional coordinate system and a position data packet, obtain the sensor position in the sensor information, mark the sensor position in the three-dimensional coordinate system, determine the position coordinates of the sensor, and then determine the position data packet between adjacent sensors based on the position coordinates, wherein the position data packet includes a position distance and a position angle;

[0010] The feature association analysis module is used to calculate the data features of the sensor. The data features include the normal overall association value and the phenomenon overall association value. The normal overall association value refers to obtaining the target sensor and its adjacent combination, and calculating the association value of two normal data in the adjacent combination, and then comprehensively processing the association value of the adjacent combination of the target sensor, and marking the processing result as the normal overall association value;

[0011] The phenomenon overall correlation value includes a forward overall correlation value and a reverse overall correlation value. A transmission direction is set between adjacent sensors, and the transmission direction includes forward transmission and reverse transmission. According to the energy transmission direction, the phenomenon data collected by the sensor is divided into forward data and reverse data. The target sensor and its adjacent combination are selected, and the forward data and reverse data in the adjacent combination are respectively obtained and calculated to obtain the forward correlation value and the reverse correlation value. Then, the forward correlation value and the reverse correlation value in the adjacent combination of the target sensor are respectively processed with the position data packet to obtain the forward overall correlation value and the reverse overall correlation value.

[0012] The real-time data analysis module is used to divide the real-time collected data into real-time normal data and real-time phenomenon data, and then determine the transmission direction of the sensor based on the real-time phenomenon data, and calculate the real-time phenomenon association value, and then obtain the real-time normal data and calculate the real-time normal association value, and comprehensively calculate the real-time phenomenon association value and the real-time normal association value to obtain the device trust value of the sensor;

[0013] The abnormal confirmation module is used to determine the abnormal sensor according to the device trust value of the sensor, and generate an abnormal signal, which is transmitted to the terminal device of the staff, and the staff confirms the fault of the abnormal sensor.

[0014] As a further solution of the present invention, a method for calculating a location data packet includes:

[0015] S1: Select any position in the target detection area as the reference position, and set the three-dimensional coordinate system with the reference position as the origin. Then, based on the position where each sensor is placed, mark the position coordinates of the sensor as Gi (xi, yi, zi), where i represents the number corresponding to different sensors;

[0016] S2: Randomly select two sensors a and b, and obtain the position coordinates Ga and Gb of the two sensors, where a and b both belong to i. Then use the distance formula to obtain the position distance Dab between sensor a and sensor b. ;

[0017] S3: Based on the formula and Get the position angle respectively and , the position angle and They are respectively bound with the position distance Dab and integrated to obtain the position data packets Wab (Wa, Wb) of sensor a and sensor b, where: , ;

[0018] in, It means to describe the position of sensor b with sensor a as the reference point. It means that the position of sensor a is described with sensor b as the reference point.

[0019] As a further solution of the present invention, a method for calculating the normal overall correlation value includes:

[0020] SS1: Obtain sensor data in the target detection area under normal conditions and mark them as normal data NU. Based on the underwater model, arbitrarily select a sensor as the target sensor, and combine the target sensor with adjacent sensors respectively and mark them as adjacent combinations.

[0021] SS2: Obtain the normal data in each interproximal combination, and use the Pearson correlation coefficient algorithm to calculate the correlation value of each interproximal combination, and mark the calculation result as Rj, where j represents different interproximal combinations;

[0022] SS3: Get the position distance Dj of the two sensors in the adjacent combination, and then use the formula The normal overall correlation value Tr of the target sensor is obtained, where J represents the total number of neighboring combinations of target sensors, is the influence coefficient.

[0023] As a further solution of the present invention, a method for calculating the positive correlation value includes:

[0024] SS4: Obtain sensor data in the target detection area after the underwater explosion occurs and mark it as phenomenon data. Then set the transmission direction between adjacent sensors, including forward transmission and reverse transmission.

[0025] SS5: According to the location and transmission direction of the underwater explosion, the phenomenon data is divided into forward data and reverse data. The forward data refers to the phenomenon data collected during forward transmission, and the reverse data refers to the phenomenon data collected during reverse transmission.

[0026] SS6: A sensor is randomly selected as the target sensor again, and the neighbor combination of the target sensor is obtained. The positive data in the neighbor combination is extracted, and the correlation value between the positive data in the neighbor combination is calculated using the Pearson correlation coefficient algorithm, and marked as the positive correlation value ZRm, where m represents the number of different neighbor combinations in the target sensor, and m∈[1,M], where M represents the total number of neighbor combinations of the target sensor.

[0027] As a further solution of the present invention, a method for calculating the positive overall association value includes:

[0028] SS7: Get the location data packets corresponding to the two sensors in the adjacent combination, and get the corresponding location angles based on the transmission direction;

[0029] SS8: Exploiting the formula Get the positive overall correlation value Zr of each target sensor, where , , Dm is the position distance between two sensors in adjacent combination m, It represents the position angle of the adjacent combination during forward transmission. represents the lateral influence coefficient during forward transmission, represents the longitudinal influence coefficient during forward transmission, Represents the overall correction factor during forward transmission.

[0030] As a further solution of the present invention, a method for calculating the reverse overall correlation value includes:

[0031] According to the processing method in steps SS6 to SS8 above, the reverse data in the adjacent combination of the target sensor is obtained, the correlation value of the reverse data in each adjacent combination is calculated, and the calculation result is marked as the reverse correlation value NRm, and then the corresponding position angle is obtained based on the transmission direction. , combining the inverse correlation value with the position angle, using the formula Determine the inverse overall correlation value Nr of the target sensor, represents the lateral influence coefficient during reverse transmission, represents the longitudinal influence coefficient during reverse transmission, Indicates the overall correction factor during reverse transmission.

[0032] As a further solution of the present invention, a method for calculating a device trust value includes:

[0033] ST1: Compare the real-time collected data with the threshold data, and divide the real-time collected data into real-time normal data and real-time phenomenon data according to the comparison result. If the real-time collected data is real-time normal data, the data is stored. If the real-time collected data is real-time phenomenon data, the explosion location of the underwater explosion is determined.

[0034] ST2: According to the location of the underwater explosion, determine the energy transmission direction between sensors, and then determine the phenomenon overall correlation value Xr of each sensor according to the transmission direction, wherein the phenomenon overall correlation value Xr includes a positive overall correlation value Zr and a reverse overall correlation value Nr;

[0035] ST3: arbitrarily select a sensor as the target sensor, and obtain the neighboring combination of the target sensor. Then, based on the real-time phenomenon data in each sensor, the real-time phenomenon data is calculated according to the method of calculating the overall phenomenon association value, and the calculation result is marked as the real-time phenomenon association value GX of the target sensor;

[0036] ST4: Taking the time of the determined real-time phenomenon data as the node time, obtain the real-time normal data collected by the target sensor and the sensors in its neighboring combination before the node time, and mark the calculation result as the real-time normal correlation value GS of the target sensor according to the calculation method of the normal overall correlation value;

[0037] ST5: Then use the formula Get the device trust value RE of the target sensor, c1∈(0,1), c2∈(0,1).

[0038] As a further solution of the present invention, a method for determining an abnormal sensor includes:

[0039] Compare the device trust value with the standard trust threshold. If the device trust value is less than the standard trust threshold, the corresponding sensor is marked as an abnormal sensor. Otherwise, if the device trust value is greater than or equal to the standard trust threshold, the corresponding sensor is marked as a normal sensor.

[0040] When an abnormal sensor is detected, an abnormal signal is generated and the position of the abnormal sensor is obtained at the same time. The abnormal signal and the position of the abnormal sensor are simultaneously transmitted to the terminal device of the staff, and the staff confirms the fault of the abnormal sensor.

[0041] Compared with the prior art, the advantages of the present invention are:

[0042] The present invention analyzes the data collected by the sensor in the normal state and the data collected by the sensor after the underwater explosion occurs, and calculates the overall correlation value of the sensor in the normal state and the overall correlation value of the data between the adjacent sensors after the underwater explosion occurs according to the data association between adjacent sensors, determines the normal overall correlation value and the phenomenon overall correlation value, then calculates the real-time phenomenon correlation value and the real-time normal correlation value of the real-time collected sensor data, and determines the device trust value of the sensor based on the real-time phenomenon correlation value and the real-time normal correlation value, and then determines the abnormal sensor according to the device trust value, improves the accuracy of sensor data collection for underwater explosion detection through data fusion and feature analysis in multi-source devices, and also further improves the efficiency of equipment fault diagnosis response. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0045] Reference Figure 1 , a sensor data analysis system for underwater explosion detection, including an information acquisition module, an underwater model building module, a feature association analysis module, a real-time data analysis module, and an anomaly confirmation module;

[0046] The information collection module is used to collect underwater information of the target detection area, wherein the underwater information includes sensor information and water area information. Furthermore, the sensor information includes the number, position, model and corresponding equipment accuracy of the sensors placed in the target detection area. The water area information includes the depth of the target detection area, the pressure at each position and the flow rate, etc. The target detection water area refers to the area where the sensor is used to detect underwater explosions. The information collection module and the underwater model construction module are connected in a one-way communication manner.

[0047] The underwater model building module is used to obtain sensor information, and based on the sensor placement position in the sensor information, determine the relative position between sensors, and then build an underwater model of the target detection area according to the relative position relationship between all sensors. The specific underwater model building method includes:

[0048] S1: Select any position in the target detection area as the reference position, and set the three-dimensional coordinate system with the reference position as the origin. Then, based on the position where each sensor is placed, mark the position coordinates of the sensor as Gi (xi, yi, zi), where i represents the number corresponding to different sensors;

[0049] S2: arbitrarily select two sensors a and b, and obtain the position coordinates Ga and Gb of the two sensors, where a and b both belong to i, and then use the distance formula to obtain the position distance Dab between sensor a and sensor b. Specifically, ;

[0050] S3: Based on the formula and Get the position angle respectively and , and then the position angle and They are respectively bound with the position distance Dab and integrated to obtain the position data packets Wab (Wa, Wb) of sensor a and sensor b, where: , ;

[0051] It needs to be further explained that It means to describe the position of sensor b with sensor a as the reference point. It means that the position of sensor a is described with sensor b as the reference point;

[0052] S4: According to the processing methods in the above steps S2 and S3, the position data packets between all adjacent sensors are processed, and then the three-dimensional coordinate system and the position data packets are merged and marked as an underwater model;

[0053] Afterwards, there is a two-way communication connection between the underwater model building module and the feature association analysis module;

[0054] The feature association analysis module is used to determine the data features of the sensor based on the underwater model, wherein the data features include the normal overall association value and the phenomenon overall association value. Specifically, the method for determining the data features of the sensor includes:

[0055] For normal overall correlation values:

[0056] SS1: Get the sensor data of the target detection area in the normal state and mark it as normal data NU. Based on the underwater model, arbitrarily select a sensor as the target sensor, and combine the target sensor with the adjacent sensors respectively, and mark them as adjacent combinations. For example, select sensor 1 as the target sensor, and the adjacent sensors above, below, left, and right of sensor 1 are 11, 12, 13, and 14 respectively. At this time, there are 4 groups of adjacent combinations of sensor 1, namely (1, 11), (1, 12), (1, 13), and (1, 14).

[0057] SS2: Obtain the normal data in each inter-proximal combination, and use the Pearson correlation coefficient algorithm to calculate the correlation value of each inter-proximal combination, and mark the calculation result as Rj, where j represents different inter-proximal combinations. The Pearson correlation coefficient algorithm is an existing technology and will not be described in detail here;

[0058] SS3: Get the position distance Dj of the two sensors in the adjacent combination, and then use the formula The normal overall correlation value Tr of the target sensor is obtained, where J represents the total number of neighboring combinations of target sensors, is the influence coefficient, The specific values ​​are obtained by technicians in this field after big data calculations;

[0059] Then, all sensors are taken as target sensors in turn and processed according to the above steps SS1 to SS3 to obtain the normal overall correlation value of each sensor;

[0060] For the overall correlation value of the phenomenon:

[0061] The overall correlation value of the phenomenon includes the positive overall correlation value and the reverse overall correlation value. The specific calculation method includes:

[0062] SS4: Obtain sensor data of the target detection area after the underwater explosion occurs and mark it as phenomenon data. Then set the transmission direction between adjacent sensors. The transmission direction includes forward transmission and reverse transmission. For example, for sensor a and sensor b, the direction of energy transmission from sensor a to sensor b is marked as forward transmission, and the reverse direction is reverse transmission. When the underwater explosion occurs, if the energy is transmitted from sensor a to sensor b, the transmission direction is forward transmission. If the energy is transmitted from sensor b to sensor a, the transmission direction is reverse transmission.

[0063] SS5: Then, according to the location and transmission direction of the underwater explosion, the phenomenon data is divided into forward data and reverse data;

[0064] Further, forward data refers to phenomenon data collected during forward transmission, and reverse data refers to phenomenon data collected during reverse transmission;

[0065] SS6: again randomly select a sensor as the target sensor, obtain the neighboring combination of the target sensor, extract the positive data in the neighboring combination, calculate the correlation value between the positive data in the neighboring combination using the Pearson correlation coefficient algorithm, and mark it as the positive correlation value ZRm, where m represents the number of the different neighboring combinations in the target sensor, and m∈[1,M], where M represents the total number of neighboring combinations of the target sensor;

[0066] SS7: Get the location data packets corresponding to the two sensors in the adjacent combination, and get the corresponding location angles based on the transmission direction;

[0067] It is necessary to further explain the corresponding relationship between the transmission direction and the position angle. For example, for sensors a and b in the adjacent combination, if the energy is transmitted from the position of sensor a to the position of sensor b as forward transmission, and the energy is transmitted from the position of sensor b to the position of sensor a as reverse transmission, then, if the forward transmission is in progress, the position angle is selected as On the contrary, if the transmission is in reverse, the position angle selection ;

[0068] SS8: Then use the formula Get the positive overall correlation value Zr of each target sensor, where , , Dm is the position distance between two sensors in adjacent combination m, It represents the position angle of the adjacent combination during forward transmission. represents the lateral influence coefficient during forward transmission, represents the longitudinal influence coefficient during forward transmission, represents the overall correction factor during forward transmission, , and The specific values ​​of are obtained by technicians in this field after big data calculation;

[0069] Then, according to the processing method in the above steps SS6 to SS8, the reverse data in the adjacent combination of the target sensor is obtained, the correlation value of the reverse data in each adjacent combination is calculated, and the calculation result is marked as the reverse correlation value NRm, and then the corresponding position angle is obtained based on the transmission direction. , combining the inverse correlation value with the position angle, using the formula Determine the inverse overall correlation value Nr of the target sensor, represents the lateral influence coefficient during reverse transmission, represents the longitudinal influence coefficient during reverse transmission, represents the overall correction factor for reverse transmission, , and The specific values ​​of are obtained by technicians in this field after big data calculation;

[0070] According to the processing method in steps SS6 to SS8 above, all sensors are taken as target sensors in turn to obtain the positive overall correlation value Zr and the negative overall correlation value Nr of each sensor;

[0071] Then there is a one-way communication connection between the feature association analysis module and the real-time data analysis module;

[0072] The real-time data analysis module is used to analyze the data collected by the sensors in the target detection area in real time and determine the data trust value of each sensor. The specific method for determining the device trust value includes:

[0073] ST1: Compare the real-time collected data with the threshold data, and divide the real-time collected data into real-time normal data and real-time phenomenon data according to the comparison result. If the real-time collected data is real-time normal data, the data is stored. If the real-time collected data is real-time phenomenon data, the explosion location of the underwater explosion is determined.

[0074] Among them, the threshold data is set by those skilled in the art based on their experience in big data. It should be further explained that the real-time normal data refers to the sensor data collected in real time under the normal state of the target detection area, and the real-time phenomenon data refers to the sensor data collected in real time after an underwater explosion occurs in the target detection area;

[0075] ST2: According to the location of the underwater explosion, determine the energy transmission direction between the sensors, and then determine the phenomenon overall correlation value Xr of each sensor according to the transmission direction. For example, if the energy is transmitted in the forward direction between two adjacent sensors, at this time, the phenomenon overall correlation values ​​of the two sensors both select the forward overall correlation value. Conversely, if the energy is transmitted in the reverse direction between the two adjacent sensors, at this time, the phenomenon overall correlation values ​​of the two sensors both select the reverse overall correlation value.

[0076] ST3: arbitrarily select a sensor as the target sensor, and obtain the neighboring combination of the target sensor. Then, based on the real-time phenomenon data in each sensor, the real-time phenomenon data is calculated according to the method of calculating the overall phenomenon association value, and the calculation result is marked as the real-time phenomenon association value GX of the target sensor;

[0077] ST4: Taking the time of the determined real-time phenomenon data as the node time, obtain the real-time normal data collected by the target sensor and the sensors in its neighboring combination before the node time, and mark the calculation result as the real-time normal correlation value GS of the target sensor according to the calculation method of the normal overall correlation value;

[0078] ST5: Then use the formula Obtain the device trust value RE of the target sensor, c1∈(0,1), c2∈(0,1), and the specific values ​​of c1 and c2 are obtained by technicians in this field after big data calculation;

[0079] It should be further explained that when the device trust value is larger, the accuracy of the data collected by the target sensor as valid data is higher. On the contrary, if the device trust value is smaller, the accuracy of the data collected by the target sensor as valid data is lower, and the probability of abnormal failure of the target sensor is greater.

[0080] Afterwards, the real-time data analysis module transmits the device trust value of the sensor to the abnormality confirmation module;

[0081] The abnormal confirmation module is used to receive the device trust value of the sensor and compare the device trust value with the standard trust threshold. If the device trust value is less than the standard trust threshold, the corresponding sensor is marked as an abnormal sensor. Otherwise, if the device trust value is greater than or equal to the standard trust threshold, the corresponding sensor is marked as a normal sensor. The standard trust threshold is obtained by a person skilled in the art after big data calculation.

[0082] When an abnormal sensor is detected, an abnormal signal is generated and the position of the abnormal sensor is obtained at the same time. The abnormal signal and the position of the abnormal sensor are simultaneously transmitted to the terminal device of the staff, and the staff confirms the fault of the abnormal sensor.

[0083] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A sensor data analysis system for underwater explosion detection, characterized in that: include: An information collection module is used to collect underwater information of the target detection area and transmit it to the underwater model building module. The underwater information includes sensor information and water area information; The underwater model construction module is used to construct an underwater model, wherein the underwater model includes a three-dimensional coordinate system and a position data packet, obtains the sensor position in the sensor information, marks the sensor position in the three-dimensional coordinate system, determines the position coordinates of the sensor, and then determines the position data packet between adjacent sensors based on the position coordinates. The position data packet includes a position distance and a position angle. The specific method for calculating the position data packet includes: S1: Select any position in the target detection area as the reference position, and set the three-dimensional coordinate system with the reference position as the origin. Then, based on the position where each sensor is placed, mark the position coordinates of the sensor as Gi (xi, yi, zi), where i represents the number corresponding to different sensors; S2: Randomly select two sensors a and b, and obtain the position coordinates Ga and Gb of the two sensors, where a and b both belong to i. Then use the distance formula to obtain the position distance Dab between sensor a and sensor b. ; S3: Based on the formula and Get the position angle respectively and , the position angle and They are respectively bound with the position distance Dab and integrated to obtain the position data packets Wab (Wa, Wb) of sensor a and sensor b, where Wa=( , Dab), Wb=( , Dab); in, It means to describe the position of sensor b with sensor a as the reference point. It means that the position of sensor a is described with sensor b as the reference point; The feature association analysis module is used to calculate the data features of the sensor. The data features include the normal overall association value and the phenomenon overall association value. The normal overall association value refers to obtaining the target sensor and its adjacent combination, and calculating the association value of two normal data in the adjacent combination, and then comprehensively processing the association value of the adjacent combination of the target sensor, and marking the processing result as the normal overall association value. The calculation method of the normal overall association value includes: SS1: Obtain sensor data in the target detection area under normal conditions and mark them as normal data NU. Based on the underwater model, arbitrarily select a sensor as the target sensor, and combine the target sensor with adjacent sensors respectively, and mark them as adjacent combinations. SS2: Obtain the normal data in each interproximal combination, and use the Pearson correlation coefficient algorithm to calculate the correlation value of each interproximal combination, and mark the calculation result as Rj, where j represents different interproximal combinations; SS3: Get the position distance Dj of the two sensors in the adjacent combination, and then use the formula The normal overall correlation value Tr of the target sensor is obtained, where J represents the total number of neighboring combinations of target sensors, is the influence coefficient; The phenomenon overall correlation value includes a forward overall correlation value and a reverse overall correlation value. A transmission direction is set between adjacent sensors, and the transmission direction includes forward transmission and reverse transmission. According to the energy transmission direction, the phenomenon data collected by the sensor is divided into forward data and reverse data. The target sensor and its adjacent combination are selected, and the forward data and reverse data in the adjacent combination are respectively obtained and calculated to obtain the forward correlation value and the reverse correlation value. Then, the forward correlation value and the reverse correlation value in the adjacent combination of the target sensor are respectively processed with the position data packet to obtain the forward overall correlation value and the reverse overall correlation value. The calculation methods of the positive overall correlation value and the reverse overall correlation value include: SS4: Obtain sensor data in the target detection area after the underwater explosion occurs and mark it as phenomenon data. Then set the transmission direction between adjacent sensors, including forward transmission and reverse transmission. SS5: According to the location and transmission direction of the underwater explosion, the phenomenon data is divided into forward data and reverse data. The forward data refers to the phenomenon data collected during forward transmission, and the reverse data refers to the phenomenon data collected during reverse transmission. SS6: again randomly select a sensor as the target sensor, obtain the neighboring combination of the target sensor, extract the positive data in the neighboring combination, calculate the correlation value between the positive data in the neighboring combination using the Pearson correlation coefficient algorithm, and mark it as the positive correlation value ZRm, where m represents the number of the different neighboring combinations in the target sensor, and m∈[1,M], where M represents the total number of neighboring combinations of the target sensor; SS7: Get the location data packets corresponding to the two sensors in the adjacent combination, and get the corresponding location angles based on the transmission direction; SS8: Exploiting the formula Get the positive overall correlation value Zr of each target sensor, where , , Dm is the position distance between two sensors in adjacent combination m, It represents the position angle of the adjacent combination during forward transmission. represents the lateral influence coefficient during forward transmission, represents the longitudinal influence coefficient in forward transmission, Indicates the overall correction factor during forward transmission; According to the processing method in steps SS6 to SS8 above, the reverse data in the adjacent combination of the target sensor is obtained, the correlation value of the reverse data in each adjacent combination is calculated, and the calculation result is marked as the reverse correlation value NRm, and then the corresponding position angle is obtained based on the transmission direction. , combining the inverse correlation value with the position angle, using the formula Determine the inverse overall correlation value Nr of the target sensor, represents the lateral influence coefficient during reverse transmission, represents the longitudinal influence coefficient during reverse transmission, Indicates the overall correction factor during reverse transmission; The real-time data analysis module is used to divide the real-time collected data into real-time normal data and real-time phenomenon data, and then determine the transmission direction of the sensor based on the real-time phenomenon data, and calculate the real-time phenomenon association value, then obtain the real-time normal data, and calculate the real-time normal association value, and comprehensively calculate the real-time phenomenon association value and the real-time normal association value to obtain the device trust value of the sensor. Specifically, the calculation method of the device trust value includes: ST1: Compare the real-time collected data with the threshold data, and divide the real-time collected data into real-time normal data and real-time phenomenon data according to the comparison result. If the real-time collected data is real-time normal data, the data is stored. If the real-time collected data is real-time phenomenon data, the explosion location of the underwater explosion is determined. ST2: According to the location of the underwater explosion, determine the energy transmission direction between sensors, and then determine the phenomenon overall correlation value Xr of each sensor according to the transmission direction, wherein the phenomenon overall correlation value Xr includes a positive overall correlation value Zr and a reverse overall correlation value Nr; ST3: arbitrarily select a sensor as the target sensor, and obtain the neighboring combination of the target sensor. Then, based on the real-time phenomenon data in each sensor, the real-time phenomenon data is calculated according to the method of calculating the overall phenomenon association value, and the calculation result is marked as the real-time phenomenon association value GX of the target sensor; ST4: Taking the time of the determined real-time phenomenon data as the node time, obtain the real-time normal data collected by the target sensor and the sensors in its neighboring combination before the node time, and mark the calculation result as the real-time normal correlation value GS of the target sensor according to the calculation method of the normal overall correlation value; ST5: Then use the formula Get the device trust value RE of the target sensor, c1∈(0,1), c2∈(0,1); The abnormal confirmation module is used to determine the abnormal sensor according to the device trust value of the sensor, and generate an abnormal signal, which is transmitted to the terminal device of the staff, and the staff confirms the fault of the abnormal sensor.

2. A sensor data analysis system for underwater explosion detection according to claim 1, characterized in that: Methods for determining abnormal sensors include: Compare the device trust value with the standard trust threshold. If the device trust value is less than the standard trust threshold, the corresponding sensor is marked as an abnormal sensor. Otherwise, if the device trust value is greater than or equal to the standard trust threshold, the corresponding sensor is marked as a normal sensor. When an abnormal sensor is detected, an abnormal signal is generated and the position of the abnormal sensor is obtained at the same time. The abnormal signal and the position of the abnormal sensor are simultaneously transmitted to the terminal device of the staff, and the staff confirms the fault of the abnormal sensor.

Images