An online monitoring device and early warning system for geological disasters of aviation kerosene pipelines

Abstract

This invention relates to a multi-parameter online monitoring device and early warning system for geological hazards in aviation kerosene pipelines. The device includes a main control unit, tilt sensors, vibration sensors, moisture content sensors, a 4G communication module, a power management module, and a waterproof housing. The tilt, vibration, and moisture content sensors are connected to the main control unit to collect multi-parameter data. The 4G communication module connects to the main control unit for remote data transmission, and the power management module supplies power to all components. The system adopts an "edge-cloud" architecture, with multiple monitoring devices accessing a data acquisition platform and cloud server via a 4G network. The cloud-based system incorporates a fusion analysis model based on LightGBM or deep learning for hazard identification and probability prediction. The visualization platform integrates a GIS map and supports multi-level alarm linkage. This invention achieves all-weather online monitoring, intelligent early warning, and a closed-loop emergency response, significantly improving the accuracy and timeliness of geological hazard identification, reducing operation and maintenance costs, and is suitable for geological hazard safety assurance in various long-distance pipelines, including aviation kerosene pipelines.

Description

Technical Field

[0001] This invention relates to the field of long-distance pipeline safety monitoring technology, and in particular to a multi-parameter online monitoring device and early warning system for geological disasters applicable to pipelines transporting high-risk media such as aviation kerosene. Specifically, it is applied to the real-time monitoring, intelligent identification, and emergency early warning of geological disasters such as landslides, debris flows, surface subsidence, and flood erosion along pipelines under complex terrain conditions. It can realize the fully automated management of the entire process from data collection and analysis to risk early warning, and provide intelligent technical support for the safe operation and maintenance of aviation kerosene pipelines. Background Technology

[0002] Traditional manual inspection methods for the operation and maintenance of aviation kerosene pipelines have significant limitations: First, the inspection cycle is long (usually 1-2 weeks / time), making it difficult to capture sudden changes in geological hazards; second, they are greatly affected by severe weather such as typhoons, heavy rains, and dense fog, making it impossible to carry out operations under extreme weather conditions; third, the inspection coverage rate is less than 30% in areas such as uninhabited areas and steep mountains. Meanwhile, existing monitoring technologies have obvious shortcomings: the accuracy rate of single-parameter monitoring instruments in identifying complex geological hazards is less than 50%. For example, landslide precursors are often accompanied by changes in surface dip angle, a sudden increase in soil moisture content, and high-frequency vibration signals; single-parameter monitoring is prone to missed or false alarms due to environmental interference.

[0003] In addition, existing systems generally lack standardized interface capabilities with local geological disaster monitoring platforms, and incompatible data formats lead to serious information silos; moreover, they are difficult to meet the stringent requirements of aviation kerosene pipelines in terms of power supply reliability and adaptability to extreme environments. Summary of the Invention

[0004] The purpose of this invention is to provide an online monitoring device and system with multi-parameter integration, remote communication, and high adaptability to solve problems such as delayed early warning, low identification accuracy, and insufficient linkage in the prior art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: An online monitoring device for geological hazards in aviation kerosene pipelines includes: The main control unit is used to realize data acquisition, processing and control; An inclination sensor, connected to the main control unit, is used to monitor changes in the inclination angle of the pipeline and send the inclination angle data to the main control unit. A vibration sensor, connected to the main control unit, is used to monitor the vibration of the pipeline and send the vibration data to the main control unit. A moisture content sensor, connected to the main control unit, is used to monitor the moisture content of the soil around the pipeline and send the moisture content data to the main control unit. A 4G communication module, connected to the main control unit, is used to remotely transmit the data processed by the main control unit and receive instructions; The power management module is connected to the main control unit, tilt sensor, vibration sensor, moisture content sensor and 4G communication module, and is used to provide working power to each component; The housing is used to house and protect the main control unit, tilt sensor, vibration sensor, moisture content sensor, 4G communication module and power management module. The housing has an IP68 waterproof design and a buffer and shock absorption layer on its inner side.

[0006] Beneficial effects This invention significantly improves the accuracy and timeliness of geological disaster early warning through multi-parameter fusion and edge-cloud collaborative analysis, raising the early warning accuracy rate to over 92%, reducing the false alarm rate to below 5%, and shortening the alarm response time to the second level. The time from data anomaly to alarm response is ≤30 seconds, more than 20 times faster than traditional manual judgment. Solar power supply and low-power design enable 24 / 7 online monitoring, reducing the frequency of manual inspections from once a week to once a month, reducing labor costs by more than 50%, and eliminating inspection blind spots under extreme weather conditions. IP68 protection, a wide temperature range of -40℃ to 70℃, and an impact-resistant structure ensure stable and reliable operation of the device under extreme conditions such as typhoons and rainstorms. It supports seamless integration with emergency systems, allowing remote valve closure within 5 minutes of an alarm, forming a closed loop of "monitoring-early warning-response," with actual measurements showing a reduction of pipeline disaster losses by more than 60%. The initial investment for a single unit is 40% lower than similar imported products, and the total life-cycle (10-year) maintenance cost is reduced by 35%. Compared to traditional threshold methods, the multi-parameter AI model extends the lead time for landslide event identification from 15 minutes to 40 minutes, and reduces the false alarm rate from 25% to below 5%. Through local inference at the edge, data transmission volume is reduced by 60%, and alarm response latency is shortened from 30 seconds to 10 seconds. The adaptive threshold adjustment function ensures that the system's identification accuracy deviation is ≤3% under different geological conditions such as cohesive soil and gravel. The system conforms to national standard interfaces, is easy to interface with external platforms, and can be flexibly expanded for application in various long-distance pipelines. Attached Figure Description

[0007] Figure 1 This is a connection diagram of an online monitoring device for geological hazards along a jet fuel pipeline. Figure 2 This is a connection diagram of an online early warning system for geological disasters along aviation kerosene pipelines. Figure 3 This is a photograph of a landslide caused by Typhoon Haikui in 2023, as described in an embodiment of the present invention. Figure 4 This is a physical image of the multi-parameter tilt angle monitoring device in an embodiment of the present invention. Figure 5 This is a flowchart illustrating the linkage logic of the early warning system in an embodiment of the present invention. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0009] An online monitoring device for geological hazards in aviation kerosene pipelines, characterized in that it comprises: The main control unit is used to realize data acquisition, processing and control; An inclination sensor, connected to the main control unit, is used to monitor changes in the inclination angle of the pipeline and send the inclination angle data to the main control unit. A vibration sensor, connected to the main control unit, is used to monitor the vibration of the pipeline and send the vibration data to the main control unit. A moisture content sensor, connected to the main control unit, is used to monitor the moisture content of the soil around the pipeline and send the moisture content data to the main control unit. A 4G communication module, connected to the main control unit, is used to remotely transmit the data processed by the main control unit and receive instructions; The power management module is connected to the main control unit, tilt sensor, vibration sensor, moisture content sensor and 4G communication module, and is used to provide working power to each component; The housing is used to house and protect the main control unit, tilt sensor, vibration sensor, moisture content sensor, 4G communication module and power management module. The housing has an IP68 waterproof design and a buffer and shock absorption layer on its inner side.

[0010] The main control unit uses a 32-bit ARM Cortex-M4 core processor with a main frequency of 168MHz; it has an RS485 interface, and the communication rate can be adaptively adjusted within the range of 1200bps to 115200bps, supporting the Modbus-RTU protocol; the main control unit is also connected to a TF card storage module, which supports a maximum of 64GB expansion, and adopts a cyclic overwrite storage mechanism. The stored data includes the original monitoring values, timestamps, and device status information.

[0011] The main control unit integrates an edge computing module that supports a lightweight AI inference engine (such as TensorFlowLite) for local preprocessing of real-time monitoring data (including outlier removal and feature extraction) and preliminary anomaly determination. Only the determination results or suspected anomaly data are uploaded through the 4G communication module, reducing transmission bandwidth and improving response speed.

[0012] The tilt sensor is a three-axis MEMS gyroscope with a measurement range of ±180° on the X-axis, ±90° on the Y-axis, and ±180° on the Z-axis, and an accuracy of ±0.1°. It has a built-in temperature compensation circuit to eliminate the influence of ambient temperature.

[0013] The vibration sensor is a vibration sensor supporting an acceleration range of ±2g, with a sampling frequency of up to 1kHz. It employs a differential signal output method to suppress electromagnetic interference, and its measurement axis direction can be configured via software to adapt to different installation postures. The sensor uses a differential signal output method, which can effectively suppress industrial electromagnetic interference around the pipeline; its measurement axis direction can be configured via software to be horizontal, vertical, or a custom angle to adapt to different installation posture requirements.

[0014] The 4G communication module supports TD-LTE / FDD-LTE dual-mode standards, covering the mainstream frequency bands of China Mobile, China Unicom, and China Telecom; it has a built-in TCP / IP protocol stack, supports NTP network timing function, and integrates a hardware encryption chip to encrypt transmitted data using AES-128 to ensure data security.

[0015] The power management module includes a solar panel, a battery, and an MPPT controller, capable of supporting full-load operation for 30 consecutive cloudy / rainy days. The device features a timed sleep mode and an alarm sleep mode, with extremely low power consumption during sleep. Currently, the power management module uses a 20W monocrystalline silicon solar panel and a 12V / 100Ah lithium iron phosphate battery pack, equipped with an MPPT maximum power point tracking controller, achieving a charging efficiency of ≥92%. The battery pack supports deep discharge to 20% capacity, supporting full-load operation for 30 consecutive cloudy / rainy days. The module also includes overcharge, over-discharge, and short-circuit protection circuits, with an operating temperature range of -30℃ to 55℃. The timed mode allows setting sampling intervals from 1 minute to 60 minutes (in 1-minute steps), with power consumption ≤5mA during sleep. The alarm mode supports up to 8 custom wake-up times, automatically performing a full-parameter acquisition and uploading data upon wake-up, suitable for monitoring sudden geological activity scenarios.

[0016] This invention also provides an online early warning system for geological disasters along aviation kerosene pipelines, comprising multiple monitoring devices, a data acquisition platform, a remote cloud server, and a visualization analysis platform. The monitoring devices establish a communication link with the data acquisition platform via a 4G wireless network. The data acquisition platform synchronizes data in real time with the remote cloud server via a dedicated fiber optic line. The visualization analysis platform accesses the cloud server to obtain monitoring information via a browser or client.

[0017] Generally, the monitoring device is deployed in high-risk geological disaster areas (including landslide fronts, fault fracture zones, and gully catchment sections) along the aviation kerosene pipeline. The deployment spacing is dynamically adjusted according to the risk level, and real-time data is uploaded to the cloud server via 4G. The data upload frequency is linked to the device sampling frequency and is adjustable.

[0018] The cloud server has built-in disaster identification algorithms, such as a landslide probability prediction model based on LightGBM or a multi-parameter fusion analysis model. The multi-parameter fusion analysis model takes parameters such as the rate of change of tilt angle, vibration spectrum characteristics, and water content gradient as inputs. It supports dynamic threshold adjustment of multiple parameters and can automatically calibrate the threshold according to environmental parameters such as pipeline burial depth, soil type, and surrounding terrain. It also supports manual remote configuration of threshold parameters.

[0019] The visualization analysis platform supports displaying real-time monitoring curves and alarm status, and integrates a GIS map module, which can intuitively display the device deployment location, operating status, and alarm level. The system supports multi-level threshold settings (such as yellow, orange, and red levels) and alarm triggering mechanisms, for example, yellow warning probability ≥0.6, orange ≥0.8, and red ≥0.9; it supports SLA / LTA algorithms or set level thresholds to trigger alarms, where the SLA time window can be set to 1-10 seconds, and the LTA time window can be set to 100-1000 seconds; after the alarm is triggered, it automatically executes multi-level responses, including pushing SMS / APP alarms to maintenance personnel within 15 seconds, platform audible and visual alarms within 30 seconds, and linkage of pipeline emergency shut-off valves within 1 minute, realizing closed-loop management of "monitoring-early warning-response".

[0020] This invention supports long-term historical data recording and disaster trend retrospective analysis; the cloud server adopts a distributed storage architecture, with a single device's historical data storage capacity of ≥3 years (based on a sampling frequency of 10 minutes / time), and supports multi-dimensional queries by time interval, parameter type, device number, etc.; the trend analysis module can generate daily / weekly / monthly statistical reports, predict the development trend of geological disasters through algorithms such as linear regression and moving average, and output trend curves and risk level assessment results.

[0021] This invention is also compatible with local geological disaster monitoring platforms, conforming to GB / T 40702-2021 "Geological Disaster Monitoring Data Exchange Format" and DZ / T 0460-2023 "General Technical Conditions for Geological Disaster Monitoring Instruments" standards; the interface adopts the WebService protocol, supporting the synchronization of three types of data: monitoring data, equipment status, and alarm information, with a synchronization cycle that can be set to 1-60 minutes.

[0022] The outer shell of the device is made of 6061-T6 aluminum alloy (thickness ≥3mm), and is treated with anodizing for corrosion protection (film thickness ≥15μm). It has impact resistance (complies with GB / T 2423.5-1995 standard, no plastic deformation after 10J impact); the surface of the outer shell is coated with a wear-resistant coating, which is suitable for complex environments such as landslides (soil compression), floods (water erosion), and collapses (rockfall impact), with a service life of more than 5 years.

[0023] This invention can also be extended to various long-distance pipelines such as natural gas, refined oil, and crude oil. For different media pipelines, only the measurement range of the water content sensor (which can be replaced with a methane concentration sensor for natural gas pipelines) and the range of the vibration sensor (which can be extended to ±5g for crude oil pipelines) need to be adapted. The main control unit and communication module remain universal, and the installation dimensions and system architecture of the device do not need to be changed.

[0024] The cloud server has a built-in deep learning-based abnormal event recognition subsystem, including a data preprocessing module, a feature fusion module, a disaster classification model, and a threshold adaptive adjustment module. The disaster classification model adopts an LSTM-CNN hybrid network structure. The input is time-series multi-parameter monitoring data (including tilt change rate, vibration spectrum features, and water content gradient), and the output is the abnormal event type (landslide / settlement / debris flow / disturbance) and risk probability value.

[0025] The invention will now be described in detail with reference to specific examples, including the complete process of model training and optimization.

[0026] Training data preparation Step 1: Historical Data Collection. Collect monitoring data along the target pipeline for more than 3 years, including: normal operating condition data (continuous sampling during disaster-free periods, with a cumulative total of ≥100,000 data points); disaster event data (complete sequences of events such as landslides and settlements, manually labeled, with ≥50 cases of each type, including data from 12 hours before the event to its end); and disturbance data (such as construction vibration and rainfall disturbances, labeled as "non-disaster anomalies").

[0027] Step 2: Data Augmentation and Labeling. The sample size was increased by flipping the time series data and adding noise (simulating sensor noise); a dual labeling method of "event-non-event" binary classification + "multi-hazard type" multi-class classification was adopted, and the labeling standard referred to the "Classification and Grading Standard for Geological Hazards" (DZ / T 0286-2015).

[0028] Model training and optimization Step 1: Model structure initialization. The LSTM layer has 3 hidden layers (64 / 32 / 16 neurons), and the CNN layer has 2 convolutional blocks (3×3 kernels, stride 1). The Adam optimizer (learning rate 0.001) is used, and the loss function is cross-entropy loss.

[0029] Step 2: Phased training. First, pre-train on a public dataset (such as the landslide monitoring dataset from China University of Geosciences), then fine-tune using local data from the target pipeline (freeze the bottom-level parameters and train only the top fully connected layer) to avoid overfitting.

[0030] Step 3: Threshold calibration. Determine the probability thresholds for different warning levels through ROC curve analysis (e.g., a landslide probability ≥ 0.9 for a red warning), and dynamically adjust them in conjunction with the importance of the pipeline (e.g., lower the threshold by 10% for tunnel crossings).

[0031] Deployment and Iteration Step 1: Lightweight deployment at the edge. The trained model is compressed to ≤5MB using a model compression algorithm (such as knowledge distillation) and deployed on the main control unit (ARM Cortex-M4 processor) of the monitoring device, supporting 10 inferences per second locally.

[0032] Step 2: Cloud-based collaborative reasoning. Suspected abnormal data (probability ≥ 0.6) is uploaded from the edge device to the cloud. The cloud then calls the full model for secondary verification and corrects the results by combining surrounding environmental data (such as rainfall and terrain slope).

[0033] Step 3: Online Iteration. False positives / false negatives are collected monthly from manually reviewed cases. The model is updated using federated learning (only gradients are uploaded to protect data privacy), ensuring the model's annual accuracy remains above 90%.

[0034] The invention will be described below with reference to specific examples. Example 1: Device Installation Case In the section of the pipeline from Fuzhou Changle Airport to Binhai, the system provided by this invention is used for geological hazard monitoring. Based on the geological survey report, a monitoring device is deployed every 500 meters in high-risk areas along the pipeline (such as the leading edge of landslides and fault fracture zones). The device is installed on a prefabricated base 30cm from the bottom of the pipe using clamps, or buried 1.5m deep using boreholes, ensuring good coupling between the sensor and the soil. All deployments are equipped with corrosion-resistant grounding devices.

[0035] Example 2: System Operation Instance The device collects tilt angle, vibration, and moisture content data at set intervals (e.g., 10 minutes). The main control unit acquires sensor data via an internal bus, performs preliminary processing by the edge computing module, and then encrypts and uploads the data to the data acquisition platform via a 4G communication module, ultimately synchronizing it to the cloud server. The cloud server's built-in LSTM-CNN hybrid network model analyzes the multi-parameter time series in real time. When the landslide probability exceeds a threshold, it triggers an alarm of the corresponding level and executes an emergency response according to the linkage logic (e.g., ...). Figure 3 (As shown).

[0036] Example 3: Model Optimization Case To maintain high recognition accuracy, the system adopts a federated learning mechanism, collecting false positive / false negative cases that are manually reviewed every month and updating the model in the cloud to ensure that the model's annual recognition accuracy remains above 90%.

[0037] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online monitoring device for geological hazards in aviation kerosene pipelines, characterized in that, include: The main control unit is used to realize data acquisition, processing and control; An inclination sensor, connected to the main control unit, is used to monitor changes in the inclination angle of the pipeline and send the inclination angle data to the main control unit. A vibration sensor, connected to the main control unit, is used to monitor the vibration of the pipeline and send the vibration data to the main control unit. A moisture content sensor, connected to the main control unit, is used to monitor the moisture content of the soil around the pipeline and send the moisture content data to the main control unit. A 4G communication module, connected to the main control unit, is used to remotely transmit the data processed by the main control unit and receive instructions; The power management module is connected to the main control unit, tilt sensor, vibration sensor, moisture content sensor and 4G communication module, and is used to provide working power to each component; The housing is used to house and protect the main control unit, tilt sensor, vibration sensor, moisture content sensor, 4G communication module and power management module. The housing has an IP68 waterproof design and a buffer and shock absorption layer on its inner side.

2. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The main control unit uses a 32-bit or higher core processor, has an RS485 interface and supports the Modbus-RTU protocol; the main control unit is also connected to a TF card storage module, which supports a cyclic overwrite storage mechanism.

3. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The main control unit integrates an edge computing module and supports a lightweight AI inference engine for local preprocessing and preliminary anomaly detection of real-time monitoring data.

4. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The tilt sensor is a three-axis MEMS gyroscope with a measurement range of ±180° on the X-axis, ±90° on the Y-axis, and ±180° on the Z-axis, an accuracy of ±0.1°, and a built-in temperature compensation circuit.

5. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The vibration sensor is a vibration sensor that supports an acceleration range of ±2g, adopts a differential signal output method, and its measurement axis direction can be configured by software.

6. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The 4G communication module supports TD-LTE / FDD-LTE dual-mode, has a built-in TCP / IP protocol stack, supports NTP network timing function, and integrates a hardware encryption chip.

7. The online monitoring device for geological hazards in aviation kerosene pipelines according to claim 1, characterized in that: The power management module includes a solar panel, a battery, and an MPPT controller; the device has a timed sleep mode and an alarm clock sleep mode to reduce power consumption.

8. An online early warning system for geological disasters in aviation kerosene pipelines, characterized in that, Includes multiple monitoring devices, data acquisition platforms, remote cloud servers, and visualization analysis platforms as described in any one of claims 1-7; The monitoring device and the data acquisition platform establish a communication link through a 4G wireless network. The data acquisition platform synchronizes data with a remote cloud server in real time via a dedicated fiber optic line. The visualization analysis platform accesses the cloud server to obtain monitoring information through a browser or client.

9. The online early warning system for geological disasters in aviation kerosene pipelines according to claim 8, characterized in that: The cloud server has a built-in disaster identification algorithm, which is a landslide probability prediction model or a multi-parameter fusion analysis model based on LightGBM, and supports dynamic threshold adjustment of multiple parameters.

10. The online early warning system for geological disasters in aviation kerosene pipelines according to claim 8, characterized in that: The visualization analysis platform supports displaying real-time monitoring curves and alarm status, and integrates a GIS map module; the system supports multi-level threshold settings and alarm triggering mechanisms, and automatically executes multi-level responses after an alarm is triggered, including pushing SMS / APP alarms, platform audible and visual alarms, and linkage with pipeline emergency shut-off valves.

Images