1. Introduction
Natural gas hydrate is an important green (low carbon) energy source [
1], of which the solid phase only exists in low temperature and high pressure [
2]. It is easy to burn and is commonly called “combustible ice” [
2,
3,
4,
5]. Submarine gas hydrates bring new energy prospects to humans, while it also presents a hazard for the environment and human beings. If the conditions change, the hydrate releases methane gas and the physical properties of the seafloor sediment are also altered. As a result, the mechanical properties of seabed sediments are greatly reduced and the seabed is softened, causing large-scale submarine landslides, collapses, and the destruction of subsea engineering facilities [
6,
7,
8]. Therefore, in-situ and long-term monitoring of subsidence during the exploitation of submarine gas hydrates is of great significance for the study of the formation mechanism of submarine landslides, hydrate environmental impact assessment, and early warning.
At present, the means of seafloor subsidence monitoring are mainly through satellite remote sensing and single-beam and multi-beam devices, among which acoustic measurements, such as multi-beam, are the most common method. These acoustic devices need to be mounted on a measuring platform, such as a surveying ship, a remotely controlled vehicle (ROV) or an autonomous underwater vehicle (AUV). These monitoring methods cannot achieve in-situ, long-term monitoring due to high costs [
9,
10,
11,
12].
In Japan, pressure sensors and three-component servo-accelerometers were used for subsidence monitoring, but only for a single-point measurement to monitor the settlement of a certain point [
8,
13,
14] In recent decades, the technology of microelectromechanical systems (MEMS) has developed rapidly. MEMS inclinometers and accelerometers have been widely used for land subsidence and landslide detection on land. The Measurand Company of Canada produces a MEMS accelerometer array called the Shape Accel Array (SAA). The SAA measures the shapes of the path in the borehole and structure and monitors the deformation of the land-based structure and slope [
8,
15]. However, there are few articles on submarine in-situ surface mapping and multiple-point subsidence monitoring at home and aboard.
Besides, there are many researches on data acquisition in multi-node and distributed systems, as well as time synchronization. Most of these systems are in the form of a single bus, and few articles use the combination of inter-integrated circuit (IIC) bus (short distance, high speed) and controller area network (CAN) bus (long distance, multi-node) to achieve high-speed and long-distance data transmission [
11,
16,
17]. Meanwhile, underwater sensor networks have attracted more and more attention, and are widely used in marine environmental monitoring, resource exploration, target detection, tracking, and positioning [
18,
19].
It is meaningful to match the data collected by the distributed sensor nodes with the time information. Data with same time collected is also the basis for implementing technologies such as network cooperative work. Generally, time synchronization is mainly based on the use of independent and high-precision clock reference sources, such as a global positioning system (GPS) or Beidou time-scaler, which are commonly used on land. The method makes it easy to achieve time synchronization of the system, where the requirements of the clock source are very demanding. However, the seawater shields electromagnetic waves, light waves, and other signal transmissions. Therefore, many underwater time synchronizations must obtain a synchronization signal by setting a receiving base station on the sea level through an optical fiber or a cable. Li et al. used the satellite signal of the GPS/Beidou dual reference source received by the precision time protocol (PTP) master time (base station) as the time source. The decoded output network time protocol (NTP) synchronization signal and PTP synchronization signal were transmitted through the fiber optic cable to obtain an accurate time [
20,
21]. Moreover, there is also a time synchronization technology based on IRIG-B code, which is a time standard code developed to realize information exchange between the fields. The IRIG-B code combines the advantages of pulse timing and serial message timing, of which encoding has high synchronization precision and can reach microsecond level. However, the generation and reception of IRIG-B code is very complicated, requiring special equipment and high cost [
22,
23]. In addition, few researches have been done on time synchronization in the deep sea where clock sources cannot be used.
In this article, the topography of 30 m × 30 m around the natural gas hydrate mining area was taken as the research object. The multi-node data acquisition technology and system time synchronization method were taken as the main research contents. The remainder of the article is organized as follows.
Section 2 describes the design of subsidence monitoring system based on MEMS sensor.
Section 3 gives a description of two-stage data acquisition system based on IIC bus and CAN bus.
Section 4 presents the design of relative time synchronization method for data synchronous acquisition.
Section 5 presents a series of tests including synchronous testing and error evaluation, submarine subsidence simulation, and sensor array pressure test, which were carried out to demonstrate the performance of the data synchronous acquisition method. At last,
Section 6 presents the conclusions and further research.
2. Design of Subsidence Monitoring System Based on MEMS Sensor for Gas Hydrate Mining Area
In-situ and long-term monitoring are two major difficulties of the system. In this article, multiple observation points were set, each of which can perform high-accuracy measurements, and network communication is required among observation points.
The displacement measurement is an effective method to measure the subsidence of seafloor topography. In this article, the MEMS attitude sensor (JY901 Weite Intelligent Co., Ltd. Shenzhen, China) was used to collect the acceleration information of each node, and the displacement of each point was obtained by filtering and integration algorithm to construct a three-dimensional seafloor subsidence model. The shape between the sensors is fitted with a short straight line, and all the sensors on the array need to maintain a non-twisted state during operation. If the array can twist freely, the directions between subsidence and uplift cannot be distinguished when the sensor is reversed. Therefore, the sensors were arranged in a strip-shaped rigid base (polyamide) with segments, where the joints were connected by flexible joints (composite rubber). MEMS sensors, microcontroller unit (MCU) (STM32F103ZE) slave stations, and the data acquisition cable are on the array. The structure of MEMS sensor array is shown in
Figure 1.
Four MEMS sensor arrays were placed in the shape of ‘X’ in the gas hydrate mining area, as shown in
Figure 2a. There were 21 MEMS attitude sensors on each array. The controller was set in the center of the entire system and was responsible for the communication, storage, and power supply of the whole system. The collected data was finally sent to the controller to reconstruct the three-dimensional model of the seafloor subsidence. The data synchronization acquisition system diagram is shown in
Figure 2b.
3. Two-Stage Data Acquisition System Based on IIC (Inter-Integrated Circuit) Bus and CAN (Controller Area Network) Bus
The data collection in this system has two main challenges. First, the number of sensor nodes is large. If each node is independently connected to the controller, the stability and reliability of the system would be reduced. Second, long-distance data collection needs to ensure real-time data collection.
In order to solve the difficulty of data collection, the set one MCU slave station for every three sensors. The data synchronous acquisition and transmission system adopted the master-slave structure, the master station (controller) was responsible for collecting, storing, and pre-processing, while the slave was responsible for the acquisition of the acceleration sensor data. The data transmission between the sensors and the slave used the IIC bus (the first-stage data acquisition and transmission system). The data of the slaves and the masters were transmitted through the CAN bus (the second-stage data acquisition and transmission system). Two-stage data acquisition and transmission system are shown in
Figure 3.
3.1. The First-Stage Data Acquisition and Transmission System Based on IIC Bus
The slaves need to control the time of data collection, frequency, and transmission sequence of the sensor node. After collecting the data of the corresponding node, timestamp and identification characters were added at the end of the data. “#” was used as the start character and “
$” was used as the terminator. The data was preprocessed, including filter and data frame packing. It was converted as a string, facilitating the extraction and processing of the data in the three-dimensional dynamic reconstruction of the subsidence. The data processing program of the slave is as shown in the
Figure 4.
3.2. Figures, Tables, and Schemes
The MCU controller adopts the form of master-slave stations. At first, the master sends a verification code to the slaves via the CAN bus, which is used to designate a slave to send data to the master. Once the slave sends the data to the CAN bus, the master will receive the data via the CAN interrupt and determine the integrity of the data. If the data is complete, it will be stored for processing and analyzing. In addition, the slave notifies the master at the end of transmission, and the master also informs the slaves when the data is received. The closed-loop feedback can be formed to ensure efficient and orderly transmission. The flow diagrams of the master and the slave are as shown in the
Figure 5a,b.
4. Design of Relative Time Synchronization System
The reconstruction of the three-dimensional subsidence dynamic model requires high simultaneity of data. If the time of the data collected by each sensor is inconsistent, the reconstructed three-dimensional subsidence dynamics would be inconsistent with the actual seabed as well. Therefore, time synchronization of all the accelerometer data on the four arrays is required, which means time synchronization of all the slaves needs to achieve a better synchronization accuracy. The time of the electronic component is generally provided by a quartz crystal oscillator. The quartz crystal oscillator generates a fundamental frequency, which is converted into a local time signal of the node by frequency division and frequency multiplication. However, there are slight differences in the production of crystal oscillators [
24]. Meanwhile, environmental conditions such as temperature can also cause changes in the oscillation frequency of the crystal oscillator [
25]. Realizing the synchronous acquisition of the data of subsidence in the seabed environment without adding any extra hardware devices is very meaningful.
4.1. The First-Stage Data Acquisition and Transmission System Based on IIC Bus
STM32F103ZET6 was selected as the MCU of the slave and the time drift of the slave was tested by comparing the real-time clock (RTC) of the MCU with the coordinated universal time (UTC). The MCU continuously runs for 60 days without interruption and automatically saves RTC information and UTC information at 9:00 every day. As
Figure 6a shows, three slaves were tested. The results of the test are as shown in the
Figure 6b. Within the first 20 days, the time drift of the slave was small, with a maximum of 14.3 ms. However, the drift increased rapidly over time and 60 days later, the drift reached 820.28 ms. Moreover, the drift rates of the three slaves are inconsistent.
4.2. The Second-Stage Data Acquisition and Transmission System Based on CAN Bus
Considering the particularity of the system submarine working environment, it is unable to communicate with the maritime base station under the deep sea nor is it suitable to add more hardware devices to achieve time synchronization. A relative time synchronization method was designed to ensure the relative synchronization of the slaves and master in the system. When the slaves collect the sensor data, a timestamp is added to each group of data. It is beneficial to implement time synchronization and data extraction in the subsequent three-dimensional reconstruction. The time source for synchronization of each slave is periodically synchronized with the time of the master. The CAN bus is a serial bus, which means the information on the CAN bus is transmitted sequentially in order. In other words, the information cannot be transmitted to each sensor node at the same time. On the other hand, the signal on the input/output (I/O) can be synchronously transmitted [
26]. Therefore, the relative synchronization was performed as follows: The master clock information was transmitted through the CAN bus in advance, and the timing signal was simultaneously sent from the I/O to the slaves. The system relative time synchronization schematic is shown in
Figure 7.
When the master reaches a certain moment A, the moment B (B is later than A) will be transmitted to each slave through the CAN bus in advance and stored as a variable. The interval between B–A is to ensure that the master clock information can be passed to each slave during this time. For example, TIME_2. TIME_1 is a variable of the local time of the slave. Then, when the time of the master reaches the moment B, a trigger signal is sent to each slave through the I/O. After that, the variable TIME_1 is covered by the variable TIME_2. The schematic is shown in the
Figure 8.
The system relative time synchronization method involves using each slave to periodically synchronize with the master. The master sequentially sends the time information through the CAN bus in advance, then sends the trigger signal through the I/O synchronization. In this way, each slave is synchronized with the clock source of the master, so that the system can ensure the relative synchronization of the internal time.
After the initialization is completed, the master will send the advance time to the slaves through the CAN bus when time reaches to a moment set. When the time of the master reaches the advanced time of the transmission, the master will send the synchronous trigger signal through I/O. Then, the master enters the next wait and loops. The slave receives the advance time sent by the master through the CAN interruption. Identically, it is received in order and stored in a certain variable. The slave will periodically check this variable to determine whether the value of the variable has changed. If there is no change, it continues the task of sending or receiving data. Otherwise, it will stop the data receiving and sending, scanning the I/O continuously until the arrival of trigger signal is received. Once this trigger signal is received, the defined variable is overwritten. Time synchronization master and slaver flow diagrams are shown in
Figure 9a,b.
6. Summary and Outlook
The article proposed a high-speed synchronous acquisition method for multi-node, distributed systems, which has the following characteristics:
(1) The data acquisition of MEMS sensor was carried out by combining the two stages of IIC bus and CAN bus, where pre-processing, such as feedback control and time stamping, can effectively realize data collection and transmission, which greatly facilitates post-processing of data.
(2) According to the particularity of the submarine working environment, a system-time synchronization scheme was proposed, which can realize the time synchronization between different sensor nodes. This method can also be applied to the case where it is impossible to use a universal independent high-precision time reference source, such as the GPS/Beidou time scale.
(3) Additional hardware devices are not necessary for achieving synchronous acquisition, which improves the stability of the system.
(4) The accuracy of the synchronization can be adjusted by the synchronization interval.
Compared to the methods of acoustic measurement, such as multi-beam method, the data synchronization acquisition system and monitoring method have advantages of strong anti-interference and anti-deformation ability, independent measurement, etc. In addition, the system has good expandability and can be used for monitoring of various deformed surfaces such as engineering and medicine in the future. The system will be tested in the South China Sea gas hydrate mining area next year and we will continue to report it.