Remote Monitoring System of Geological Exploration in Lava Area Based on GPS

The architecture of System Hardware

Geological exploration in lava areas requires high equipment requirements. All remote monitoring equipment selection has been certified by the security department (^{Riordan et al., 2017}). The three core function modules of a remote monitoring system for geological exploration in lava area based on GPS are the data acquisition module of geological exploration in lava area, the data analysis and management module of geological exploration in lava area, and the location module of geological anomaly based on GPS (^{He & Yan, 2017}).

The data acquisition module of geological exploration in the lava area is the monitoring end of the system. It transmits the collected geological data of the lava area to the data analysis and management module of geological exploration in the lava area using a GPS network. After filtering and analyzing the acquired data, if there are abnormal values in exploration data, it transmits positioning instructions to the geology based on GPS. The anomaly location module carries out positioning feedback to users (^{Liu et al., 2017}). The overall structure of the remote monitoring system for geological exploration in the lava area based on GPS is shown in Figure 1.

Figure 1 Structural diagram of the remote monitoring system for geological exploration in the lava area based on GPS 

(1) Data acquisition module for geological exploration in the lava area The main task of the data acquisition module of geological exploration in lava area is to combine the old three pieces (geological hammer, compass, magnifier) and the new three pieces (notebook, digital camera, GPS) to achieve multi-S integration and to collect formation information, rock information and structural information of lava area (^{Geng et al., 2017}). The structure of the data acquisition module for geological exploration in the lava area is shown in Figure 2.

Figure 2 Structure diagram of geological data acquisition module in the lava area 

(2) Data analysis and management module of geological exploration in the lava area

The data analysis and management module of geological exploration in the lava area is responsible for receiving, analyzing, and managing the geological data of the lava area collected by the data acquisition module of geological exploration in the lava area. After receiving the geological data collected by the data acquisition module of geological exploration in the lava area, the module filters the geological data in the lava area using the method of monitoring data processing based on unascertained filtering. According to the data, it judges whether there are abnormal values in geological exploration in the lava area. If there are abnormal values, it transmits positioning instructions to the geological abnormal situation positioning module based on GPS (^{Kostopoulou et al., 2017}). The structure of the data analysis and management module for geological exploration in the lava area is shown in Figure 3.

Figure 3 Structure diagram of data analysis and management module for geological exploration in the lava area 

(3) Geological anomaly location module based on GPS

A geological anomaly location module based on GPS consists of a space part, a ground control part, and a user part (^{Alison et al., 2017}). Details are shown in Figure 4.

Figure 4 GPS - based geological anomaly location module structure 

GPS - based geological anomaly location module

a. Spatial part

The space part of GPS is composed of 24 GPS working satellites, which together constitute the satellite constellation. Among them, 21 are localizable satellites and 3 are standby microsatellites. The 24 satellites are evenly distributed in six orbital planes about 20183 km from the ground. The angle of intersection between each orbital plane and the equatorial plane is 55°, and the distance between the orbital planes is 60° (i.e. the ascending point of the orbits is 60° different from each other). There are four satellites in each orbit. Each GPS working satellite sends out signals for locating abnormal materials in the geology of lava areas.

b. Ground control part

The ground control part is composed of a monitoring system consisting of several tracking stations distributed around the world. According to their different functions, these tracking stations are divided into main control stations, monitoring stations, and injection stations.

c. User equipment section

The user equipment is composed of GPS receiver, data processing software, and corresponding user equipment. Its function is to receive GPS satellite signals to obtain necessary positioning information.

Monitoring Data Processing Method based on Unascertained Filtering

During the operation of the remote monitoring system for geological exploration in the lava area based on GPS, there are more or fewer errors in monitoring data due to the influence of monitoring instrument accuracy and observation conditions. When analyzing the monitoring data, it will be found that there will be jumps in the monitoring data and local jumps in the time-displacement curve drawn. On this basis, the analysis and survey of geological stability in the lava area will encounter difficulties in data screening. Therefore, before analyzing the monitoring information, it is better to do filtering to minimize the systematic error, random error, and gross error in the monitoring data, so that the monitoring data can better reflect the geological conditions of lava areas.

In the monitoring of geological exploration in the lava area, abnormal data processing is very important. There are generally two reasons for abnormal monitoring data: errors caused by monitoring, called gross errors; abnormal monitoring data caused by geological abnormalities, called abnormal values. In data processing, gross errors should be eliminated, abnormal values should be processed and analyzed, and effective measures should be taken to deal with the geological abnormal part of the lava area (^{Hu et al., 2017}).

Based on the modern error theory, this paper uses the unascertained filtering method to judge the gross errors and outliers of monitoring data, to effectively eliminate gross errors and identify abnormal values (^{Song et al., 2017}).

There are essential differences between gross errors and abnormal values. The greatest difference is that gross errors are contingent, and the frequency of occurrence is not high. Compared with the monitoring data of this group, gross errors are usually single and have a certain contingency. The abnormal values are continuous and accumulative after the catastrophic data are generated, which is essentially due to the geological anomalies in lava areas. In this paper, we use unknown rational numbers to distinguish gross errors from abnormal values (^{Amiri-Simkooei et al., 2017}).

Taking the monitoring data sequence V ₁ ,V ₂ ,V ₃ ,….V _n as an example, an unascertained rational number is defined.

Where, 1 ≤ V ₁ ,≤ n, f (x), f (x) is the distribution density function of the reliability of remote monitoring data for geological exploration in the lava area. The main purpose of processing monitoring data is to identify and screen anomalous data, eliminate the gross errors caused by various factors, and reduce their impact on data analysis. Through the analysis of gross error and abnormal value of monitoring data, it is found that if the value of gross error appears in the monitoring sequence, the data must exist in isolation, and there is only one [a, b] in the remote monitoring data interval of geological exploration in lava area; if the value is abnormal value, the adjacent monitoring sequence has continuity. Over time, this data appears more and more frequently. Therefore, the combination of monitoring data and its reliability density function f (x) can effectively eliminate gross errors, identify outliers, and reduce its impact on monitoring data analysis.

If the trend of monitoring data sequence changes too much, it will affect L. number of neighborhoods of monitoring sequence, thus affecting the judgment of gross errors. In order to reduce the impact of sequence trend change on V, n detection values of monitoring data can be divided into several data segments, each data segment has k + 1 detection value, k< n, then:

When x = V _i ,

When x ≠ V _¡ ,

Where, x represents the monitoring value in remote monitoring data interval [a, b] of geological exploration in the lava area, and ζ. represents the number of (i ≠ j) in V _i field.

The size of the credibility density function determines the probability of V _i as the likelihood of gross error, the greater the reliability of V _i is, the smaller the probability of gross error is; and the smaller the reliability of V is, the greater the probability of gross error is. Then the gross error is determined according to the value of f(x). After determining the gross error in the monitoring sequence, the gross error needs to be eliminated. In order to eliminate the gross error, the unascertained expectation of V _i and its monitored values in the neighborhood is used as the final result of the monitoring sequence. It is not known that the expected E (A) calculation method is:

Where, l means that in the original monitoring data, the value of monitoring data starts from l, and the length of the data segment is k+1. l value varies with the location of V in the selected monitoring data. Let m be a definite even number. The selection of monitoring data is as follows:

When i = 1, the first and second monitoring data are taken as the search area of ζ in the data sequence segment where V is located. When formula (2) and formula (3) are used to calculate, k of confidence density function f (x) is taken as 1, and l = 1 of unascertained expectation is calculated by the formula (4).

When , the values of in the monitoring data are taken as the sequence search area, and then

When , the values from 2i - n to n-th in the monitoring data are taken as the sequence search area, where

When i = n, the n-1 and n values in the monitoring data are taken as the search area of the sequence, and then k = 1 and l = n - 1 are taken.

Gross error recognition depends on the reliability of monitoring data. The reliability depends on the number of monitoring data in the V _i neighborhood, and the number of monitoring data depends on the size of λ in the neighborhood radius. The value of λ is determined according to the actual situation in the field. When the average difference between two adjacent monitoring values is large, the value of λ is larger. When the average difference between two adjacent monitoring values is small, the value of λ is smaller. In practical engineering, if the value of λ is too large, the ability to identify gross errors will decrease; if the value of λ is too small, the normal data may be misidentified as gross errors, resulting in increased workload and inaccurate calculation. Therefore, the reasonableness of λ value will directly affect the effectiveness of data processing by filtering method (^{Yu & Lu, 2017}).

For the value of λ, it is usually determined by the method of calculating the difference variance of the monitoring data sequence. With the help of the criterion of double or triple median error in gross error identification, the difference variance of double or triple is obtained. Considering that λ is bigger in the monitoring data sequence with gross errors, λ is taken as double-difference variance in the processing of remote monitoring data of geological exploration in the lava area.

The upper and lower limits of the neighborhood of the selected monitoring sequence are determined according to the neighborhood radius calculated by the formula.

Where λ is the threshold parameter of gross error recognition. In the process of gross error identification, gross errors need to be eliminated and the remaining monitoring data need to re-calculate λ until there is no gross error (^{Li et al., 2017}).

Because the value of k depends on the selection of m, m generally chooses all even numbers from 2 to n/2 and calculates the value of r iteratively. The value that makes λ converge sufficiently and can identify gross errors is taken as the selection value of m. In concrete calculation, if m is too small and there are continuous gross errors, there will be zero gross errors, and the total reliability of the data segment is zero. At this time, E (A) cannot replace gross errors. We should select a larger data segment to obtain the unascertained expected value, or select more data to expand the neighborhood radius.

For the monitoring data sequence with a small change range, m usually chooses a larger value to get a better result. If the data segment is wider, the ability of gross error recognition is stronger; for the monitoring data sequence with the large change, if m is selected, it is difficult to obtain satisfactory results. The reason is that the detection data sequence has volatility, which will result in the occlusion of gross error recognition, thus affecting the accuracy of data processing results (^{Wang et al., 2017}).

Test Settings

The strata in the test area are mainly limestone of the Tianziling Formation in the Devonian. In the detailed exploration stage of engineering geology in this area, besides ascertaining the stratum structure and geophysical and mechanical properties (^{Gao et al., 2013}; ^{Liu et al., 2014}), the emphasis is on ascertaining the distribution, shape, size, scale of lava and the associated soil caves and ground subsidence in the overlying strata. This system is used to monitor the process of engineering geological exploration in this area.

(1) Location function test

In the lava monitoring area used in the experiment, 12 small caves are set up. The system is used for remote monitoring of exploration, and the positioning function of the system is tested (^{Hendrikx & Gestel 2017}).

(2) Reliability testing of monitoring data

Combined with the distribution of small soil caves in the experimental area, the monitoring data of 23 m and 3.2 m geological survey points in the lava area are extracted as the analysis objects.

23 m: A total of 22 monitoring data from May 2019 to July 2019 are selected for analysis. When the serial number of the monitoring data is greater than 3, the monitoring data is taken as the center and three values are taken forward and backward as their neighborhoods to calculate the reliability of the monitoring data in their neighborhoods (^{Liu et al., 2018}).

3.2 m: Through collating and analyzing the monitoring data of this system, 22 monitoring data from August 29, 2019, to August 12, 2019, are selected for analysis.

Test Results of Location Function

In this paper, the location function test results of 12 small soil caves in the geology of lava area are shown in Table 1.

Table 1 The test results of the system positioning function are presented in this paper 

The statistical results of the positioning errors of the system in this paper are shown in Table 2.

Table 2 Systematic positioning errors in this paper 

Note: X and Y are coordinate values of soil cave and H is height value. ∆X, ∆Y and ∆H are positioning differences.

According to Table 1 and Table 2, we can see that the maximum error is 0.3 when the system locates 12 small soil caves, and the error is small.

The Reliability Test Results of Monitoring Data of the Systemin This Paper

(1) Reliability test results of monitoring data at 23m survey point

The calculation process is shown in Table 3, where _i represents the number of monitoring data in the neighborhood. In this paper, the difference variance of R is λ times. The reliability of the monitoring data of the system is tested several times by setting different monitoring dates.

Table 3 Reliability calculation table of monitoring data at 23m survey point 

According to Table 3, the reliability of monitoring data before May 28, 2019 is 1, and the reliability of monitoring data is high. During the period from May 28, 2019 to June 2, 2019, the reliability of monitoring data fluctuates to a certain extent, but it remains above 0.8, which can be considered that the monitoring data is basically in the normal range. From June 8, 2019 to June 23, the reliability of monitoring data is maintained above 0.8, and the reliability is basically stable. The reliability of monitoring data from June 28, 2019 is 0.81, and that of monitoring data from June 30, 2019 to July 2, 2019 is 0.83, which basically meet the requirements of the project. After July 2, 2019, the reliability of monitoring data has remained basically stable.

According to the reliability analysis of the monitoring data of the system in this paper, we can see that most of the monitoring data have high reliability, which is basically maintained above 0.8. The reliability of the monitoring data in June-4, 2019 and June-28, 2019 are both 0.81, which indicates that the monitoring data in these two periods are encouraging, and the credibility is the lowest compared with other monitoring data. On the whole, the reliability of the monitoring data in this system is better, which meets the remote monitoring standard of geological exploration in lava area.

(2) Reliability test results of monitoring data at 3.2m survey point Combined with the distribution of small soil caves in the experimental area, the monitoring data of 3.2m geological survey points in the lava area are extracted as the analysis object. Through collating and analyzing the monitoring data of this system, 22 monitoring data are selected from August 29, 2019 to July 2, 2019. Details are shown in Table 4.

Table 4 Reliability calculation table of monitoring data at 3.2 m survey point 

According to Table 4, the reliability curve of monitoring data of 3.2m survey point is drawn, as shown in Figure 5.

Figure 5 Reliability curve of 3.2m monitoring data at survey point 

Among them, the abscissa represents the monitoring day, the ordinate represents the credibility data, and the maximum value is 1. During the period of August 29, 2018 to Dec. 29, 2018, the credibility of monitoring data is 1, and the credibility of monitoring data is high in this period; during the period of January 1, 2019 to February 25, 2019, the credibility of monitoring data continues to decrease, to a minimum of 0.81; during the period of March 14, 2019 to May 7, 2019, the credibility of monitoring data continues to rise, to a maximum of 1; after May 7, 2019, the credibility of monitoring data remains stable, at this time, the credibility of monitoring data is high.