Edge intelligent multi-node landslide monitoring and alarming system and method
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG UNIV
- Filing Date
- 2026-03-10
- Publication Date
- 2026-06-09
Smart Images
Figure CN122176867A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent monitoring of slopes and landslides, and particularly relates to an edge intelligent multi-node landslide monitoring and alarm system and method. Background Technology
[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.
[0003] Landslides are a common and serious geological hazard in mountainous highways, railways, and mining projects. Existing landslide monitoring devices, such as inclinometers, accelerometers, strain gauges, and wire displacement gauges, upload monitoring data to data acquisition instruments or cloud platforms via wired or wireless means for unified analysis and early warning. Traditional monitoring nodes mostly only perform periodic sampling and simple threshold judgments, making it difficult to identify complex vibration patterns. They can only upload large amounts of raw data or low-level processing results to host computers or the cloud, resulting in high communication bandwidth consumption and power consumption. At the same time, single threshold criteria are either too rigid to provide early warning of landslides or are easily affected by vehicle vibrations, wind loads, construction, etc., leading to a high false alarm rate. When a slope slips or landslides occur, it is often a large-scale disturbance, and multiple nodes may detect anomalies and report them simultaneously within a short period of time. For wireless networks using ALOHA-type random access such as LoRa, if the frequency, spreading factor, and backoff strategy for simultaneous reporting by multiple nodes are not specifically designed, a large number of data packet collisions can easily occur, leading to the loss of alarm information. Summary of the Invention
[0004] To address at least one of the technical problems mentioned above, this invention provides an edge-intelligent multi-node landslide monitoring and alarm system and method, which can improve the accuracy, real-time performance, and robustness of landslide monitoring and early warning.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of the present invention provides an edge-intelligent multi-node landslide monitoring and alarm system.
[0006] An edge-intelligent multi-node landslide monitoring and alarm system includes: edge-intelligent landslide monitoring nodes, edge-intelligent gateways, and a cloud management platform; Multiple edge intelligent landslide monitoring nodes are deployed on the monitored slope; each edge intelligent landslide monitoring node includes an inertial sensor module and a node microcontroller unit; the inertial sensor module has a built-in machine learning core; the machine learning core is configured to identify events based on the acceleration and angular velocity characteristics output by the inertial sensor module, and match corresponding data reporting strategies based on the event identification results; At least one of the edge smart gateways communicates with multiple edge smart landslide monitoring nodes. The edge smart gateway is configured to: receive and aggregate data reported by all edge smart landslide monitoring nodes, parse the reported data, update the timestamp, and send it to each edge smart landslide monitoring node, and then assess and issue an alarm for the overall stability of the slope. The cloud management platform is configured to receive evaluation results and alarm information uploaded by the edge intelligent gateway for visualization and to assist human decision-making.
[0007] As one implementation, the event identification results include events and non-events, where events include suspected slippage and typical landslides, and non-events include normal micro-movements and environmental disturbances.
[0008] As one implementation method, when the machine learning core determines that it is an event, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up from the low-power sleep state, record the local timestamp of the event trigger time, switch to setting a high sampling rate to record inertial data within a preset time window, and report the event characteristics and timestamp to the edge intelligent gateway according to the predetermined reporting mechanism.
[0009] As one implementation method, when the machine learning core determines that it is a non-event, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up at a preset period, collect and calculate the tilt angle and acceleration statistical features, and send periodic status reports to the edge intelligent gateway in a set low data volume manner.
[0010] As one implementation, the edge smart gateway has a built-in real-time clock; after obtaining a unified time base, the edge smart gateway stores it in a local real-time clock or system clock variable.
[0011] In one implementation, the edge smart gateway is configured to: acquire a high-precision unified time reference and maintain the current absolute time of the gateway locally; receive a time synchronization request frame actively sent by a monitoring node to the edge smart gateway, immediately acquire the current absolute time of the edge smart gateway as a sending timestamp, write it into a response data frame and send it immediately; The edge intelligent landslide monitoring node is configured to: after receiving a response data frame, parse the transmission timestamp, calculate the flight time of the wireless signal in the air based on the current communication spreading factor and bandwidth parameters, superimpose the transmission timestamp and the flight time to calculate the current real absolute time, and update the node's local real-time clock accordingly.
[0012] As one implementation method, when the edge intelligent landslide monitoring node reports a landslide event, it directly carries an absolute timestamp generated by the synchronized local real-time clock to achieve accurate alignment of data from multiple nodes on a unified time axis.
[0013] As one implementation method, based on the distance between the edge intelligent landslide monitoring node and the edge intelligent gateway, the line-of-sight conditions, or the spatial location in the slope, multiple edge intelligent landslide monitoring nodes are divided into several groups, and different uplink frequencies and spreading factors are configured for different groups, so that the frequencies and spreading factors are spatially dispersed.
[0014] As one implementation method, when an event is triggered by an edge intelligent landslide monitoring node, it does not report immediately. Instead, it generates a random delay based on the edge intelligent landslide monitoring node ID within a preset backoff window before reporting. It can then report several times in the subsequent time, with different random backoffs added between each report to reduce the probability of collisions when multiple nodes report simultaneously.
[0015] As one implementation method, the calculation process for random delay is as follows:
[0016] in, For random delay; For edge-based intelligent landslide monitoring node ID; For the event ID; Count the events; To set the upper limit of the window for backoff.
[0017] In one implementation, the edge smart gateway includes a multi-channel LoRa concentrator module, a gateway processor module, a time reference acquisition unit, and a communication module; The multi-channel LoRa concentrator module is used to receive and demodulate uplink data packets under multiple frequency points and multiple spreading factors in parallel. The time reference acquisition unit is used to acquire a unified time, periodically send time synchronization frames to each edge intelligent landslide monitoring node, or return a time synchronization response when it receives periodic reports from the edge intelligent landslide monitoring nodes, so that each edge intelligent landslide monitoring node adjusts its local real-time clock to the same time system as the gateway. The gateway processor module embeds an edge AI risk assessment model to assess and issue warnings on the overall stability of the slope based on the data reported by the edge intelligent landslide monitoring nodes. The communication module is used for data interaction between the edge smart gateway and the cloud management platform.
[0018] As one implementation, the edge AI risk assessment model is trained offline and then converted by a tool into an inference library adapted to the embedded platform and deployed in the gateway processor module.
[0019] A second aspect of the present invention provides an edge-intelligent multi-node landslide monitoring and alarm method.
[0020] An edge-intelligent multi-node landslide monitoring and alarm method includes: The machine learning core performs event recognition on the acceleration and angular velocity features output by each inertial sensor module, and matches the data reporting strategy of the edge intelligent landslide monitoring node based on the event recognition results; The edge intelligent gateway receives and aggregates data reported by all edge intelligent landslide monitoring nodes, parses the reported data, updates the timestamp, and sends it to each edge intelligent landslide monitoring node. Then, it assesses the overall stability of the slope and issues an alarm. The cloud management platform receives evaluation results and alarm information uploaded by the edge smart gateway for visualization and to assist human decision-making.
[0021] The beneficial effects of this invention are: (1) The present invention integrates a machine learning core into the inertial sensor module. The machine learning core performs event recognition based on the acceleration and angular velocity characteristics output by the inertial sensor module. It matches the corresponding data reporting strategy based on the event recognition results. By introducing a machine learning module at the edge intelligent landslide monitoring node, local pattern recognition is achieved, which further reduces false alarms and power consumption. (2) The present invention triggers high-frequency sampling and alarm reporting only when the machine learning core determines that it is an event, which significantly reduces the communication and energy burden under normal working conditions, and reduces the false alarm rate of the traditional simple threshold method through multi-class pattern recognition.
[0022] (3) This invention designs a time synchronization method that combines periodic time synchronization and event timestamp reporting for low-power LoRa nodes, so that the local clock error of each node is controlled within the range of seconds or even sub-seconds. By carrying the local timestamp in the event reporting and reconstructing the unified time axis on the gateway side, the timing relationship of multi-point response during landslide evolution can be accurately restored, solving the problem that existing systems are difficult to perform multi-point timing analysis.
[0023] (4) This invention addresses the multi-frequency points, multi-spreading factors, and random backoff reporting mechanism in disaster scenarios to improve communication reliability under multi-node concurrency. This invention groups and configures uplink frequency points and spreading factors according to node spatial distribution and link conditions, and applies random backoff and repeated reporting strategies in event reporting to significantly reduce the probability of LoRa packet collision when multiple nodes alarm simultaneously. Combined with the multi-channel and multi-demodulator resources of the edge smart gateway, the system can still reliably receive most node alarm data in a short time under extreme conditions such as landslides, thus improving the system robustness.
[0024] (5) This invention embeds an edge AI risk assessment model within the gateway processor module. Based on the data reported by the edge intelligent landslide monitoring nodes, it assesses and alerts on the overall stability of the slope, realizing local trend analysis and risk assessment. This improves the real-time performance and offline capability of early warning. This invention deploys a lightweight edge AI risk assessment model at the gateway end, comprehensively analyzes the slow variables and event characteristics reported periodically by multiple nodes, and outputs the risk index or risk level of the node-level, region-level, and overall slope, realizing local assessment and early warning of landslide risk. Compared with solutions that rely entirely on cloud analysis, this invention can still provide early warning functions when the network is unstable or the cloud is unavailable, improving the real-time performance and autonomy of the system.
[0025] (6) The present invention forms a multi-layer collaborative architecture of “node edge intelligence – gateway edge intelligence – cloud platform”, which makes full use of the sensitivity of node MLC to local high-frequency information and the comprehensive analysis capability of gateway to multi-node and multi-time scale data, and moves some of the computation to the edge side, which reduces the communication pressure on the cloud and improves the accuracy and reliability of the entire system, and has good engineering promotion value.
[0026] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0027] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0028] Figure 1 This is a structural block diagram of the edge intelligent multi-node landslide monitoring and alarm system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the hardware structure of a landslide monitoring node according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the hardware structure of the edge smart gateway according to an embodiment of the present invention; Figure 4(a) is a schematic diagram of the periodic reporting of monitoring nodes according to an embodiment of the present invention; Figure 4(b) is a timing diagram of time synchronization according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the gateway edge risk assessment process according to an embodiment of the present invention. Detailed Implementation
[0029] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0030] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0031] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0032] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.
[0033] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.
[0034] Existing technologies rely excessively on cloud-based computing for alerts, exhibiting inherent latency and heavy network dependence. Current systems often dedicate complex analysis and model calculations to cloud servers, with gateways merely handling data forwarding. Consequently, timely risk assessments and warnings are impossible if the network is unstable or the cloud server malfunctions. Furthermore, most low-power wireless nodes lack sophisticated time synchronization mechanisms, resulting in local clock drift and typically only uploading relative timestamps or indices of the "collection moment." This makes it difficult for the cloud to accurately align multi-point data on a unified timeline, hindering precise analysis of the temporal evolution of landslides. Therefore, it is necessary to propose a multi-node landslide prediction and early warning system that integrates node edge AI and gateway edge AI, enabling real-time monitoring and local early warning of slope risks while maintaining low power consumption and reliable communication.
[0035] according to Figure 1 This invention provides an edge intelligent multi-node landslide monitoring and alarm system, comprising: edge intelligent landslide monitoring nodes, edge intelligent gateways, and a cloud management platform; Multiple edge intelligent landslide monitoring nodes are deployed on the monitored slope; each edge intelligent landslide monitoring node includes an inertial sensor module and a node microcontroller unit; the inertial sensor module has a built-in machine learning core (MLC); the machine learning core is configured to identify events based on the acceleration and angular velocity characteristics output by the inertial sensor module, and match the corresponding data reporting strategy based on the event identification results.
[0036] like Figure 2 As shown, the inertial sensor module is used to collect information such as acceleration and angular velocity of the node. The inertial sensor module is a six-axis inertial measurement unit with a machine learning core (MLC). The node microcontroller unit is used to control the inertial sensor data acquisition, MLC configuration and result reading, reporting logic control, and low power management. The node microcontroller unit has LoRa radio frequency function, which can communicate wirelessly with the edge smart gateway via LoRa and supports the configuration of different frequencies, bandwidths and spreading factors. The power management module is used to store solar energy and power the above modules. The local event monitoring and reporting strategy runs on the MLC of the inertial sensor module and the node microcontroller unit.
[0037] The inertial sensor module is implemented using a six-axis inertial sensor (e.g., LSM6DSV16X), which has a built-in MLC; a low-power MCU (e.g., STM32WLE series) connects to the inertial sensor via I2C or SPI bus. A LoRa wireless communication module (which can be implemented by integrating the MCU's RF core or using an external LoRa chip); and a power management and battery module are also included. Before system deployment, multiple sets of inertial data are collected for the target slope or similar conditions under normal, slight disturbance, slip, and landslide states. Feature extraction and MLC model training are performed on the data on the host computer, and the trained MLC configuration parameters are then downloaded to the node inertial sensors.
[0038] The core machine learning configuration involves event recognition based on the acceleration and angular velocity features output by the inertial sensor module. Data acquisition and filtering: When the chip is running, it first acquires raw data from the inertial sensor. This raw data is grouped into sets of time windows, such as continuously acquiring data for 0.2 seconds as a set. Then, it is filtered to remove noise interference. Feature calculation: The mean, variance, energy, and bimodal values are calculated from the processed raw data. The decision tree classification and output process calculates feature values and inputs them into a pre-trained decision tree. The decision tree performs classification layer by layer until it reaches the bottom layer and outputs a classification result such as "typical landslide" or "environmental disturbance". When multiple sets of data show the same classification result, for example, multiple sets of data all classify "typical landslide", that result is output. Finally, this result is notified to the point microcontroller unit, which then makes the next control decision.
[0039] In this embodiment of the invention, the event identification results include events and non-events. Events include suspected slippage and typical landslides, while non-events include normal micro-movements and environmental disturbances.
[0040] When the machine learning core determines that it is an event, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up from the low-power sleep state, record the local timestamp of the event trigger time, switch to setting a high sampling rate to record inertial data within a preset time window, and report the event characteristics and timestamp to the edge intelligent gateway according to the predetermined reporting mechanism.
[0041] When the machine learning core determines that it is a non-event, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up at a preset period, collect and calculate the tilt angle and acceleration statistical features, and send periodic status reports to the edge intelligent gateway in a low data volume mode.
[0042] The inertial sensor has a minimum sampling rate of 12.5Hz and a maximum of 833Hz. During normal monitoring, the sampling rate is set to a low rate of 12.5Hz to save power. Once an event occurs, such as detecting a "landslide precursor," the sampling rate is immediately increased to obtain more accurate results. When the machine learning core determines that the landslide is suspected or a typical landslide pattern, the edge intelligent landslide monitoring node wakes up from the low-power sleep state, records the local timestamp of the event trigger, switches to a high sampling rate to record inertial data within a preset time window, and reports the event characteristics and timestamp to the edge intelligent gateway through the wireless communication module according to the predetermined reporting mechanism. In non-event states, the node wakes up at a preset period, collects and calculates tilt angle and acceleration statistical characteristics, and sends periodic status reports to the edge intelligent gateway in a low-data-volume manner (uploading once every hour at a frequency of 0.5Hz; once an event occurs, uploading starts immediately and the frequency is increased to 1Hz).
[0043] To reduce the probability of LoRa packet collisions caused by simultaneous reporting from multiple nodes in sudden situations such as landslides, this invention further provides the following reporting mechanism: Multiple monitoring nodes are divided into several groups based on the distance between the node and the gateway, line-of-sight conditions, or spatial location within the slope. Different uplink frequencies and spreading factors are configured for different groups, thus dispersing the frequencies and spatial distribution of the data streams. When a node triggers an event, it does not report immediately. Instead, it reports after a random delay is generated based on the node ID within a preset backoff window (e.g., 0 to several seconds). This backoff can be repeated several times in subsequent periods, with different random backoffs added between each report to reduce the probability of collisions from simultaneous reporting by multiple nodes. Based on the LoRa packet time-on-air (TOA) and the SX1302's at least sixteen-channel concurrent demodulation capability, the frequency and spreading factor allocation and backoff window design ensure that, under extreme conditions such as landslides, the system can successfully receive alarm data from most nodes within a few-second time window.
[0044] As shown in Figures 4(a) and 4(b), the periodic time synchronization process includes: When a node reports during its execution cycle, it writes a local timestamp into the reported data frame. This timestamp represents the current value read by the node from the RTC. The gateway records the reception time when it receives this reported data frame. The gateway constructs a time synchronization downlink frame within a short period of time. The downlink frame includes the gateway's current time. and optional node time echo field After receiving a time synchronization frame within a preset receiving window, the node reads the data contained within it. And adjust the local RTC time directly or through a linear mapping method to In a simplified implementation, nodes can directly set their local RTC to... Since the receiving window time (tens to hundreds of milliseconds) is much smaller than the node RTC drift and sampling period, this error is negligible. Through the above process, the deviation between the local RTC of each node and the gateway time is kept within 1 second or even 0.5 seconds. The periodic time synchronization interval is determined according to the crystal oscillator accuracy and application requirements, and can be 2 hours, 4 hours, or 24 hours, etc.
[0045] When a node detects an event trigger, it immediately reads the current local RTC time. The gateway maintains a record of the most recent synchronization time for each node, including its local time. This timestamp is also included in the event reporting data frame. The corresponding gateway time When the gateway receives a message containing... After the event reporting frame is completed, the occurrence time of the event on the unified timeline is estimated using the following formula. :
[0046] When RTC drift is small and synchronization intervals are short, the above linear relationship is sufficient to provide second-level accuracy. The gateway can handle multiple nodes... By sorting and analyzing, the spatial-temporal evolution process of the landslide event is obtained.
[0047] To improve communication reliability during concurrent reporting by multiple nodes in emergency scenarios such as landslides, the reporting mechanism in this embodiment is designed as follows.
[0048] 1. Multi-frequency and multi-SF configuration; In the 470 MHz band, configure eight 125 kHz uplink channels, for example: Channel CH0: Frequency 470.1 MHz, bandwidth 125 kHz; Channel CH1: Frequency 470.3 MHz, bandwidth 125 kHz; Channel CH2: Frequency 470.5 MHz, bandwidth 125 kHz; Channel CH3: Frequency 470.7 MHz, bandwidth 125 kHz; Channel CH4: Frequency 470.9 MHz, bandwidth 125 kHz; Channel CH5: Frequency 471.1 MHz, bandwidth 125 kHz; Channel CH6: Frequency 471.3 MHz, bandwidth 125 kHz; Channel CH7: Frequency 471.5 MHz, bandwidth 125 kHz.
[0049] Based on the distance and obstruction between the monitoring nodes and the gateway, the nodes are divided into several groups: Nodes with shorter distances and better link conditions use SF7 to SF8, primarily using CH0 to CH2; medium- to long-distance nodes use SF9 to SF10, primarily using CH3 to CH4; extremely distant nodes or nodes with significant environmental interference can use larger SF values and reserved channels CH5 to CH6. By distributing the frequency and SF space, the reported events from a large number of nodes are "distributed" across different channels, reducing the probability of collisions on a single channel.
[0050] SF is the spreading factor, an important parameter in the linear spread spectrum frequency modulation communication technology used by LORA. Simply put, a small spreading factor, such as SF7, transmits data at a high rate but is easily interfered with, making it suitable for short-distance transmission. A large spreading factor, such as the maximum SF12, transmits data at a slower rate but has strong anti-interference capabilities, making it suitable for long-distance transmission. Here, it is reasonable to assign a small spreading factor to nodes closer to the gateway and a large spreading factor to nodes farther from the gateway.
[0051] Based on the distance between the edge intelligent landslide monitoring nodes and the edge intelligent gateway, the line-of-sight conditions, or the spatial location in the slope, multiple edge intelligent landslide monitoring nodes are divided into several groups, and different uplink frequencies and spreading factors are configured for different groups, so that the frequencies and spreading factors are spatially dispersed.
[0052] When an event is triggered at the edge intelligent landslide monitoring node, it does not report immediately. Instead, it generates a random delay based on the edge intelligent landslide monitoring node ID within a preset backoff window before reporting. It can then report several times in the subsequent time, with different random backoffs added between each report to reduce the probability of collisions when multiple nodes report simultaneously.
[0053] The calculation process for random delay is as follows: ; in, For random delay; For edge-based intelligent landslide monitoring node ID; For the event ID; Count the events; To set the upper limit of the fallback window; Indicates the remainder.
[0054] In other embodiments, the MCU pseudo-random number function is directly used in [0, Delay is generated within the specified range. Nodes within... Afterwards, LoRa transmission is initiated, sending the event reporting frame to the gateway. To further improve reliability, the node can repeat the reporting 1 to 2 times within tens of seconds after the event occurs, regenerating a random backoff value each time, thereby reducing overlap with other nodes in the same time slice.
[0055] At least one edge smart gateway communicates with multiple edge smart landslide monitoring nodes. The edge smart gateway is configured to: receive and aggregate data reported by all edge smart landslide monitoring nodes, parse the reported data, update the timestamp, and send it to each edge smart landslide monitoring node, and then assess and issue an alarm for the overall stability of the slope.
[0056] The edge smart gateway includes a multi-channel LoRa concentrator module, a gateway processor module, a time reference acquisition unit, and a communication module. Multi-channel LoRa concentrator modules are used to receive and demodulate uplink data packets at multiple frequency points and multiple spreading factors in parallel; for example, multi-channel LoRa concentrator modules use multi-channel LoRa concentrator chips such as SX1302+SX1250. The time reference acquisition unit is used to acquire a unified time and periodically send time synchronization frames to each edge intelligent landslide monitoring node, or return a time synchronization response when it receives periodic reports from the edge intelligent landslide monitoring nodes, so that each edge intelligent landslide monitoring node adjusts its local real-time clock to the same time system as the gateway. The gateway processor module embeds an edge AI risk assessment model to assess and issue warnings on the overall stability of the slope based on the data reported by the edge intelligent landslide monitoring nodes. The communication module is used for data interaction between the edge smart gateway and the cloud management platform, and can be a 4G cellular communication module, wired Ethernet or Wi-Fi module.
[0057] The edge AI risk assessment model is trained offline and then converted into an inference library adapted to the embedded platform and deployed in the gateway processor module.
[0058] The edge smart gateway has a built-in real-time clock; after obtaining a unified time base, the edge smart gateway stores it in a local real-time clock or system clock variable.
[0059] The edge intelligent gateway obtains a unified time base via NTP, GPS, or other methods and maintains the gateway's current time in its local RTC. When a monitoring node reports its periodic status, it writes its current local time (first time information) into the reporting data frame. After receiving this reporting data, the gateway reads the node's local time and the gateway's receiving time, and sends a time synchronization response data frame containing the gateway's time (second time information) to enable the node to update its local RTC based on the first and second time information. When receiving event reports from each node, the gateway calculates the actual time of the event on the gateway's unified timeline by combining the stored gateway time of the node's most recent synchronization and the node's local time with the event's local timestamp carried in the event report. This enables cross-node time alignment and reconstructing of the landslide evolution process.
[0060] The LoRa concentrator module of the edge smart gateway uses the SX1302 + front-end RF chip, which can simultaneously receive LoRa signals from multiple SFs in the above 8 uplink channels and has at least 16 concurrent demodulation capabilities.
[0061] With the combined effect of multi-frequency point, multi-SF configuration and random backoff strategy, even under extreme conditions such as landslides where dozens or even hundreds of nodes are triggered almost simultaneously, this embodiment can still reliably receive alarm data from most nodes within a few seconds, ensuring the communication reliability of the system during disasters.
[0062] The cloud management platform is configured to receive evaluation results and alarm information uploaded by the edge smart gateway for visualization and to assist human decision-making.
[0063] In one or more embodiments, an edge-intelligent multi-node landslide monitoring and alarm method includes: The machine learning core performs event recognition on the acceleration and angular velocity features output by each inertial sensor module, and matches the data reporting strategy of the edge intelligent landslide monitoring node based on the event recognition results; The edge intelligent gateway receives and aggregates data reported by all edge intelligent landslide monitoring nodes, parses the reported data, updates the timestamp, and sends it to each edge intelligent landslide monitoring node. Then, it assesses the overall stability of the slope and issues an alarm. The cloud management platform receives evaluation results and alarm information uploaded by the edge smart gateway for visualization and to assist human decision-making.
[0064] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. An edge-intelligent multi-node landslide monitoring and alarm system, characterized in that, include: Edge-based intelligent landslide monitoring nodes, edge-based intelligent gateways, and cloud management platforms; Multiple edge intelligent landslide monitoring nodes are deployed on the monitored slope; each edge intelligent landslide monitoring node includes an inertial sensor module and a node microcontroller unit; the inertial sensor module has a built-in machine learning core; the machine learning core is configured to identify events based on the acceleration and angular velocity characteristics output by the inertial sensor module, and match corresponding data reporting strategies based on the event identification results; At least one of the edge smart gateways communicates with multiple edge smart landslide monitoring nodes. The edge smart gateway is configured to: receive and aggregate data reported by all edge smart landslide monitoring nodes, parse the reported data, update the timestamp, and send it to each edge smart landslide monitoring node, and then assess and issue an alarm for the overall stability of the slope. The cloud management platform is configured to receive evaluation results and alarm information uploaded by the edge intelligent gateway for visualization and to assist human decision-making.
2. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, The event identification results include events and non-events. Events include suspected slippage and typical landslides, while non-events include normal micro-movements and environmental disturbances.
3. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1 or 2, characterized in that, When the machine learning core determines that an event has occurred, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up from a low-power sleep state, records the local timestamp of the event trigger time, switches to a high sampling rate to record inertial data within a preset time window, and reports the event features and timestamp to the edge intelligent gateway according to a predetermined reporting mechanism.
4. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1 or 2, characterized in that, When the machine learning core determines that it is a non-event, the node microcontroller unit controls the edge intelligent landslide monitoring node to wake up at a preset period, collect and calculate the tilt angle and acceleration statistical features, and send periodic status reports to the edge intelligent gateway in a low data volume setting.
5. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, The edge smart gateway has a built-in real-time clock; after obtaining a unified time base, the edge smart gateway stores it in a local real-time clock or system clock variable.
6. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, The edge smart gateway is configured to: acquire a high-precision unified time reference and maintain the current absolute time of the gateway locally; receive a time synchronization request frame actively sent by the monitoring node to the edge smart gateway, immediately acquire the current absolute time of the edge smart gateway as the sending timestamp, write it into the response data frame and send it immediately. The edge intelligent landslide monitoring node is configured to: after receiving a response data frame, parse the transmission timestamp, calculate the flight time of the wireless signal in the air based on the current communication spreading factor and bandwidth parameters, superimpose the transmission timestamp and the flight time to calculate the current real absolute time, and update the node's local real-time clock accordingly.
7. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, When the edge intelligent landslide monitoring node reports a landslide event, it directly carries an absolute timestamp generated by the synchronized local real-time clock to achieve accurate alignment of data from multiple nodes on a unified timeline.
8. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, Based on the distance between the edge intelligent landslide monitoring nodes and the edge intelligent gateway, the line-of-sight conditions, or the spatial location in the slope, multiple edge intelligent landslide monitoring nodes are divided into several groups, and different uplink frequencies and spreading factors are configured for different groups, so that the frequencies and spreading factors are spatially dispersed.
9. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, When an event is triggered at the edge intelligent landslide monitoring node, it does not report immediately. Instead, it generates a random delay based on the edge intelligent landslide monitoring node ID within a preset backoff window before reporting. It can then report several times in the subsequent time, with different random backoffs added between each report to reduce the probability of collisions when multiple nodes report simultaneously.
10. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 9, characterized in that, The calculation process for random delay is as follows: ； in, For random delay; For edge-based intelligent landslide monitoring node ID; For the event ID; Count the events; To set the upper limit of the fallback window; Indicates the remainder.
11. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 1, characterized in that, The edge smart gateway includes a multi-channel LoRa concentrator module, a gateway processor module, a time reference acquisition unit, and a communication module; The multi-channel LoRa concentrator module is used to receive and demodulate uplink data packets under multiple frequency points and multiple spreading factors in parallel. The time reference acquisition unit is used to acquire a unified time, periodically send time synchronization frames to each edge intelligent landslide monitoring node, or return a time synchronization response when it receives periodic reports from the edge intelligent landslide monitoring nodes, so that each edge intelligent landslide monitoring node adjusts its local real-time clock to the same time system as the gateway. The gateway processor module embeds an edge AI risk assessment model to assess and issue warnings on the overall stability of the slope based on the data reported by the edge intelligent landslide monitoring nodes. The communication module is used for data interaction between the edge smart gateway and the cloud management platform.
12. The edge intelligent multi-node landslide monitoring and alarm system as described in claim 11, characterized in that, The edge AI risk assessment model is trained offline and then converted into an inference library adapted to the embedded platform by a tool and deployed in the gateway processor module.
13. A monitoring and alarm method based on the edge intelligent multi-node landslide monitoring and alarm system as described in any one of claims 1-12, characterized in that, include: The machine learning core performs event recognition on the acceleration and angular velocity features output by each inertial sensor module, and matches the data reporting strategy of the edge intelligent landslide monitoring node based on the event recognition results; The edge intelligent gateway receives and aggregates data reported by all edge intelligent landslide monitoring nodes, parses the reported data, updates the timestamp, and sends it to each edge intelligent landslide monitoring node. Then, it assesses the overall stability of the slope and issues an alarm. The cloud management platform receives evaluation results and alarm information uploaded by the edge smart gateway for visualization and to assist human decision-making.