The flexible measurement array for deep-seated landslide displacement consists of individual measurement probes as the basic units, connected to each other through cables and high-pressure rubber hoses (as shown in Fig. 2). The high-pressure rubber hoses ensure a significant degree of flexibility and bending capability between the probes, allowing for a better match to the deformation of the borehole. The instrument cables of the nodes pass through hollow rubber hoses, maintaining a single cable connection from start to finish. The nodes can be connected in series as needed to meet different measurement requirements at various depths. From an external perspective, it appears as a cable line connected to a long string of probes. The four-core cable consists of power supply lines and a 485 bus. The internal probes are all connected to the 485 bus for data transmission. Each probe is equipped with a three-axis accelerometer module and a geomagnetic module, connected to an STM32 series microcontroller through the SPI bus. The data acquisition host controls the STM32 microcontroller via the 485 bus to simultaneously collect incline and azimuth angle data.

This instrument serves as a flexible inclinometer array capable of obtaining full-hole incline data from the bottom to the top of the borehole. It can be customized according to the depth of the borehole to achieve a perfect match. With acceleration and magnetic field sensors installed in each probe, the incline angle and direction of the borehole can be measured, allowing users to gain a more comprehensive understanding of the deformation of deep-seated landslide displacement. At present, the instrument can carry up to 32 measurement probes. Under deep-hole conditions, the full-hole inclinometer can be adapted to deep-hole conditions by modifying the program, thereby modifying the acquisition logic and communication protocol structure. However, if only the number of measurement probes is increased, the overall measurement cost will become very high. Therefore, in engineering applications, the method of increasing the length of flexible joints will dilute the probe density in fixed hole depths to reduce system application costs.

The main technical parameters of the flexible measurement array for deep-seated landslide displacement are shown in Table 1.

The duration of each data acquisition and network transmission of the instrument is about 2 min, and the working current is about 200 mA. The rest of the time is in a dormant state. Working every day consumes approximately 0.33 A h of electricity. The power supply of the full-hole inclinometer comes from a lead–acid battery located on the ground of the orifice. Considering the extreme situation where lead–acid batteries cannot be recharged after a continuous rainy month, using a 20 A h lead–acid battery can ensure the redundant power consumption design of the system.

During each acquisition process, the full-hole inclinometer can perform edge calculation and compare the current acquisition value with the previous acquisition value. If the collected values do not change or fluctuate within a certain range, the full-hole inclinometer can continue to collect and sleep at the predetermined collection frequency. If the collected value exceeds the range of variation, the full-hole inclinometer will immediately carry out repeated collection. If it is determined that the collected value exceeds the limit, the full-hole inclinometer will not sleep and will continue to collect and transmit until three consecutive data points are stable before entering sleep. As a result, the system has implemented adaptive encryption collection logic.

Based on the structural characteristics of the flexible measurement array described in Sect. 2, the measurement principle of deep-seated displacement is illustrated in Fig. 3, as mentioned in the previous section.

As shown in Fig. 3, the full-hole inclinometer measures the angle θi between the inclinometer casing and the vertical line at each section along the PVC pipe using the inclinometer spacing L. The horizontal displacement di at different elevations is then calculated using Eq. (1). 1di=Lsin⁡θi By incrementally summing from the bottom measurement point of the inclinometer pipe, the actual horizontal displacement, denoted as bi, at any elevation can be obtained, as shown in Eq. (2). 2bi=∑j=1idj The accumulated horizontal displacement at the mouth of the inclinometer pipe is denoted as B. 3B=∑i=1ndi Based on the angle and direction data obtained from each probe, the variation of horizontal displacement at different depths along the axis of the inclinometer pipe can be determined. In the above equation, di represents the horizontal displacement of the measurement section; L denotes the segmented length of the measurement point, which is the instrument spacing; θi represents the angle between the inclinometer pipe and the vertical line for the measurement section; bi represents the horizontal displacement of the ith measurement point starting from the bottom of the hole; B represents the horizontal displacement at the mouth of the current measurement pipe; and n represents the number of segments for the monitoring hole, where n=H/L and H is the depth of the hole.

The flexible inclinometer sensor adopts a triaxial accelerometer and a triaxial magnetic sensor, forming a dual triaxial (x,y,z) system. The axes of the accelerometer and magnetic sensor are parallel, with the z axis pointing downward along the instrument axis, while the x and y axes lie in the plane perpendicular to the instrument axis. The coordinate system for top-foot measurement is illustrated in Fig. 4.

By using the components of gravity acceleration g sensed by the accelerometer and the components of the magnetic field sensed by the magnetic sensor, the values of borehole inclination and azimuth angles at the instrument's location can be calculated according to Eqs. (3) and (4) (Liu, 2016). 4α=arctan⁡gαyHx-αxHyHzαx2+αy2+αzαxHx+αyHy5θ=arctan⁡ax2+ay2az In the equations, ax is the output value of the x-axis accelerometer, ay is the output value of the y-axis accelerometer, az is the output value of the z-axis accelerometer, Hx is the magnetic flux gate (or magnetic resistance) output value along the x axis, Hy is the magnetic flux gate (or magnetic resistance) output value along the y axis, Hz is the magnetic flux gate (or magnetic resistance) output value along the z axis, and g is the value of gravitational acceleration g=ax2+ay2+az2.

Prepare the installation equipment, including tools, sensors, sensor mounting plates, flexible connectors, pressure-resistant outer pipes, sealing rings, connecting wires, and non-magnetic screws, required for sensor installation (see Fig. 5).

The sensors can be assembled into the configuration shown in Fig. 7 for sensor calibration. As shown in Fig. 8, the sensors are then clamped onto the HZT-60 inclinometer calibration table (Li et al., 2015), and calibration is performed for zenith angles and azimuth angles at 16 different positions.

After calibrating each sensor, the sensors are sequentially connected using high-strength flexible connectors to form a complete instrument assembly (as shown in Fig. 9).

A demonstration site for monitoring deep-seated landslide displacement was established in Baiyan 3 Group, town of Lujiang, city of Baoshan, Yunnan Province, using the flexible measurement array for deep-seated displacement. The main objectives of this field experiment were as follows.

Testing the reliability of flexible joint connections.

In the field experiment, the flexible measurement array was installed in a borehole with a depth of 25.4 m, and a ∅ 50 PVC pipe was used for fixation inside the borehole. The instrument, consisting of 14 sensors, was connected through flexible high-strength joints. The first sensor was placed at a depth of 4 m inside the borehole, while the 14th sensor was positioned at 11 m. After the instrument was properly set up, the PVC pipe was filled with fine sand. The experiment confirmed the normal functioning of the instrument, demonstrating the feasibility of this connection method.

In this field experiment, the installation schematic of the flexible measurement array is shown in Fig. 10. The flexible measurement array is fixed at the predetermined depth position using wellhead steel-wire ropes and then filled with drilling mud to achieve coupling between the measurement array and the rock wall. The measurement data are transmitted through cables to the wellhead data acquisition device, which packages and organizes the data before sending them to the server. The wellhead is equipped with a solar charger to provide power for the underground measurement array and the wellhead data acquisition device.

After the installation and commissioning of the equipment, the project team initiated the automated monitoring and early warning work. The average value of the data transmitted by the equipment on the first day was taken as the initial data, and subsequent data points were extracted at typical time intervals based on the deformation conditions for the purpose of constructing deformation curves.

As shown in Fig. 11, analysis of the data curves reveals that from March to May 2021, there was minimal deformation in the deep-seated displacement of the landslide. However, during the flood season, a deformation of approximately 10 mm occurred between depths of 5.5 and 8 m on the slope. This was followed by an accumulated deformation of 23 mm until 11 September. Subsequently, the deformation gradually converged. The increasing displacement observed between depths of 5.5 and 8 m indicates movement along the main sliding direction of the landslide. Conversely, a decrease in displacement was observed between depths of 4 and 5.5 m due to the shifting movement of the sliding zone at 5.5 m. Below 5.5 m, the sensors recorded movement along the main sliding direction, while above 5.5 m, the sensors detected a reverse inclination caused by the movement of the sliding zone, leading to a decrease in measured displacement by the upper sensors. This method of analyzing the deep-seated displacement curve serves as a primary means to identify the position of the sliding zone.

Based on the deformation curve, it can be concluded that there is a distinct sliding zone between depths of 5.5 and 8 m in the landslide. Additionally, notable displacement variations occurred during the flood season. This variation aligns with the concentrated rainfall in the region where the landslide is located, confirming that rainfall is the main triggering factor for the landslide. The deformation curve further demonstrates the accurate monitoring capability of the flexible measurement array for deep-seated landslide displacement, surpassing traditional devices with a limited number of sensors. It provides a more precise and comprehensive measurement of displacement variations throughout the entire borehole depth.