Mountain displacement monitoring device and monitoring method for power transmission line

Abstract

The invention discloses a power transmission line mountain displacement monitoring device and a power transmission line mountain displacement monitoring method, and relates to the technical field of power system power transmission line maintenance monitoring. The power transmission line mountain displacement monitoring device obtains multi-source monitoring data of a monitoring point through a sensing module, and an edge calculation module directly completes feature anomaly recognition locally at the monitoring point; the method does not need to wait for full-amount data to be transmitted to a far-end platform for analysis, the time from data collection to abnormal feature capture is greatly shortened, reliable uploading of key data is guaranteed through dual-channel transmission, meanwhile, the platform carries out space-time coupling prediction based on received feature classification data to obtain a predicted displacement value and a predicted displacement trend curve, and the prediction accuracy is improved. The method can accurately excavate the correlation between the time evolution rule and the spatial feature of the displacement, improves the displacement monitoring and trend pre-judgment precision, carries out the grading early warning decision through combining with the prediction result, and effectively achieves the consideration of the real-time performance and the high precision of the mountain displacement monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system transmission line maintenance monitoring, and in particular to a transmission line mountain displacement monitoring device and a monitoring method. BACKGROUND

[0002] As the core channel of power supply of the power system, the safe and stable operation of the transmission line is directly related to the reliability of energy transmission. The transmission line is widely distributed in complex terrain regions such as mountainous areas and hilly areas. Such regions are prone to geological disasters such as landslides and slope instability induced by heavy rain, earthquakes and other natural factors. Among them, the small displacement of the mountain has the characteristics of strong concealment and significant cumulative effect. If it is not monitored and warned in time, the displacement amount will accumulate over time and easily cause major accidents such as tilting of transmission line towers, tower collapse, and conductor fracture, which not only causes huge economic losses, but also may lead to regional power outages, seriously threatening the safety of the power system and the power supply guarantee of social production and life. Therefore, accurate and timely monitoring of the displacement of the mountains along the transmission line has become a key link to ensure the safe operation of the transmission line.

[0003] At present, for the monitoring of the displacement of the mountains along the transmission line, the industry mainly relies on two types of technical solutions: one is manual regular inspection, which measures the terrain coordinates and inclination angles of the preset monitoring points by professional equipment such as total station and level instrument at fixed cycles, and compares the historical data to judge the displacement risk in combination with the apparent characteristics such as surface cracks and vegetation inclination; the second is a traditional simple monitoring device, which is a single GPS positioning device or an inclination sensor arranged at the monitoring point, and records the position coordinates or inclination data through a local storage module, and some devices are supplemented with simple wireless transmission function to realize data reporting. The above two types of solutions are the mainstream application methods for monitoring the displacement of the mountains along the transmission line in the mountainous area, but their technical characteristics determine that there are obvious limitations in actual application. Among them, the manual regular inspection relies on fixed cycle operation and cannot capture sudden displacement such as short-time sliding after heavy rain, and the real-time performance is seriously insufficient. The traditional simple monitoring device can realize fixed point monitoring, but the single GPS module is easily affected by the shielding of the mountain and the vegetation, resulting in positioning error and being unable to identify millimeter-level small displacement. In addition, the device has poor adaptability and is difficult to install in complex terrain such as rock face and steep slope. At the same time, the endurance is weak and the data transmission is lagging, which further aggravates the problems of early risk omission and incomplete monitoring coverage of the mountain displacement, and cannot meet the safety requirements of early discovery, early warning and early disposal of the mountain displacement along the transmission line. SUMMARY

[0004] The present application provides a transmission line mountain displacement monitoring device and a monitoring method, which solves the technical problem that the prior art cannot meet the requirements of real-time and high-precision monitoring.

[0005] The present application provides a transmission line mountain displacement monitoring device in the first aspect, which comprises:

[0006] a perception module configured to acquire multi-source monitoring data of a power transmission line mountain monitoring point;

[0007] an edge computing module configured to perform feature anomaly identification on the multi-source monitoring data to obtain feature classification data;

[0008] a transmission module configured to perform double-channel transmission on the feature classification data;

[0009] a platform configured to perform spatiotemporal coupling prediction on the received feature classification data after transmission to obtain a predicted displacement value and a predicted displacement trend curve, and perform hierarchical early warning decision-making.

[0010] Optionally, the feature classification data includes a plurality of normal feature data and a plurality of abnormal feature data, and the edge computing module includes:

[0011] a preprocessing submodule configured to perform edge preprocessing on the multi-source monitoring data to obtain preprocessed data;

[0012] a cleaning and alignment submodule configured to perform cleaning and alignment on the preprocessed data to obtain a plurality of multi-source fusion data frames;

[0013] a noise suppression submodule configured to perform noise suppression on each of the multi-source fusion data frames to obtain a plurality of denoising fusion data frames;

[0014] a feature fusion submodule configured to perform feature extraction on each of the denoising fusion data frames to obtain a plurality of extracted feature data;

[0015] an anomaly discrimination submodule configured to respectively determine whether each of the extracted feature data satisfies a preset anomaly determination condition;

[0016] If yes, the extracted feature data is taken as the abnormal feature data.

[0017] If no, a key feature is extracted from the extracted feature data for mean operation to obtain the normal feature data.

[0018] Optionally, the preset anomaly determination condition specifically refers to that any extracted feature in the extracted feature data is greater than an associated out-of-limit threshold.

[0019] Optionally, the transmission module includes:

[0020] a signal strength submodule configured to acquire signal strength data of a monitoring device in real time;

[0021] a classification encryption packaging submodule configured to perform classification encryption packaging on the feature classification data to obtain an encrypted data message containing an identifier;

[0022] The dual-channel transmission submodule is configured to select a corresponding transmission channel to transmit the encrypted data packet to the platform according to the signal strength data obtained at the current time.

[0023] Optionally, the dual-channel transmission submodule comprises:

[0024] The main channel execution submodule is configured to transmit the encrypted data packet to the platform through a main channel when the signal strength data is greater than or equal to a preset signal strength threshold.

[0025] The standby channel execution submodule is configured to fragment the encrypted data packet into a plurality of fragmented packets, and sequentially transmit the plurality of fragmented packets in the fragmented packet stream to the platform through a standby channel at a preset time interval when the signal strength data is less than the preset signal strength threshold.

[0026] Optionally, the platform comprises:

[0027] The data aggregation submodule is configured to aggregate the encrypted data packet and the fragmented packet stream to obtain feature classification data after transmission.

[0028] The time series convolution submodule is configured to input the feature classification data after transmission into a pre-trained time series convolution network to perform displacement prediction and obtain a predicted displacement value.

[0029] The space-time feature fusion submodule is configured to input the feature classification data after transmission into a pre-trained space-time feature fusion prediction network to perform trend prediction and obtain a predicted displacement trend curve.

[0030] The index calculation submodule is configured to determine a real-time displacement rate index and an acceleration index according to the predicted displacement value and the predicted displacement trend curve.

[0031] The early warning grading submodule is configured to perform grading early warning decision-making using the real-time displacement rate index and the acceleration index, and generate an early warning decision-making instruction.

[0032] Optionally, the index calculation submodule comprises:

[0033] The first calculation unit is configured to calculate the real-time displacement rate index using a sliding window method based on the predicted displacement value.

[0034] The second calculation unit is configured to fit a quadratic curve using a least squares method based on the predicted displacement trend curve, and extract an acceleration term of the quadratic curve as the acceleration index.

[0035] Optionally, the early warning grading submodule comprises:

[0036] The first early warning unit is used to generate a first-level early warning command when the real-time displacement rate index is in the first displacement rate threshold range and the acceleration index is in the first acceleration threshold range.

[0037] The second early warning unit is used to generate a secondary early warning command when the real-time displacement rate index is in the second displacement rate threshold range or the acceleration index is in the second acceleration threshold range.

[0038] The third early warning unit is used to generate a three-level early warning command when the real-time displacement rate index is in the third displacement rate threshold range or the acceleration index is in the third acceleration threshold range.

[0039] Optionally, the multi-source monitoring data includes BeiDou positioning data, nine-axis attitude data, and environmental data.

[0040] A second aspect of the present invention provides a monitoring method for a power transmission line mountain displacement monitoring device, comprising:

[0041] Acquire multi-source monitoring data from mountain monitoring points along power transmission lines;

[0042] Feature anomaly identification is performed on multi-source monitoring data to obtain feature classification data;

[0043] Feature classification data is transmitted via dual channels;

[0044] The received transmitted feature classification data is used to perform spatiotemporal coupling prediction to obtain the predicted displacement value and the predicted displacement trend curve, and then a graded early warning decision is made.

[0045] As can be seen from the above technical solutions, the present invention has the following advantages:

[0046] The transmission line mountain displacement monitoring device of this invention acquires multi-source monitoring data from monitoring points through a sensing module, providing comprehensive data support for subsequent analysis. The edge computing module directly completes feature anomaly identification locally at the monitoring point, eliminating the need to wait for the full data transmission to the remote platform for analysis, significantly shortening the time from data acquisition to anomaly feature capture. The transmission module adopts a dual-channel transmission method, enabling real-time reporting even in mountainous areas with weak signals, significantly improving the real-time performance of data processing and ensuring reliable uploading of critical data through dual-channel transmission. Simultaneously, the platform performs spatiotemporal coupling prediction based on the received feature classification data to obtain predicted displacement values ​​and predicted displacement trend curves, accurately uncovering the temporal evolution law and spatial feature correlation of displacement, improving the accuracy of displacement monitoring and trend prediction. Furthermore, it combines the prediction results to make hierarchical early warning decisions, effectively achieving a balance between real-time performance and high precision in mountain displacement monitoring, providing reliable support for the prevention and control of transmission line mountain displacement risks. Attached Figure Description

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0048] Figure 1 A module block diagram of the power transmission line mountain displacement monitoring device of the embodiment of the present application;

[0049] Figure 2 A double-channel transmission structure diagram of the power transmission line mountain displacement monitoring device of the embodiment of the present application;

[0050] Figure 3 A step flow chart of a monitoring method applied to the power transmission line mountain displacement monitoring device of the embodiment of the present application. DETAILED DESCRIPTION

[0051] The embodiment of the present application provides a power transmission line mountain displacement monitoring device and a monitoring method, which are used to solve the technical problem that the prior art cannot simultaneously meet the real-time and high-precision monitoring requirements.

[0052] In order to make the application purpose, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the accompanying drawings of the embodiments of the present application. Obviously, the following described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0053] At present, the power transmission line is the core channel of power supply, and often faces the threat of geological disasters such as mountain landslide in mountainous and complex terrain areas. If the small displacement of the mountain is not monitored in time, it may gradually accumulate and cause serious accidents such as tower collapse and wire fracture. At present, the monitoring of such mountain displacement still mainly relies on artificial periodic patrol and traditional simple monitoring devices (such as single sensor and ordinary positioning equipment). These two ways have many limitations due to the limitation of technical principles and application scenarios:

[0054] The inherent defects of artificial patrol are:

[0055] (1) Insufficient real-time: fixed cycle patrol cannot capture sudden displacement (such as short-time sliding after heavy rain), and it is difficult to timely warn;

[0056] (2) Limited accuracy: the error of manual operation measurement tool is large, and it is difficult to identify millimeter-level small displacement, and it is easy to miss the early risk;

[0057] (3) Efficiency and coverage short board: complex terrain in mountainous areas, time-consuming and laborious patrol, limited coverage range in a single day, remote areas prone to monitoring blind spots;

[0058] (4) Safety hazards: when patrolling in landslide-prone areas, staff face direct safety threats such as collapse and rockfall. Technical limitations of traditional monitoring devices:

[0059] (1) Poor positioning accuracy: single GPS is easily affected by mountain and tree shading, unstable signal, and positioning error often exceeds 1 meter;

[0060] (2) Weak endurance: relying on ordinary batteries or external power supply, field endurance is usually less than 3 months, and maintenance is frequent;

[0061] (3) Data transmission lag: some require manual data recovery, and wireless transmission is prone to loss or delay due to weak signal;

[0062] (4) Insufficient early warning capability: only records data without intelligent analysis, requires manual judgment of risks, and response lag.

[0063] The current detection method adopts fixed cycle site monitoring, and professional equipment such as total station and level is used to monitor the mountains along the transmission line; equipment is erected at the preset monitoring points, and terrain coordinates, inclination angle and other data are collected and recorded; combined with the apparent features of the site terrain (such as ground cracks, vegetation inclination), the risk of mountain displacement is judged by comparing historical data. However, the following shortcomings exist:

[0064] (1) Lack of real-time performance: relying on fixed cycle monitoring, it is difficult to capture sudden displacement (such as short-time mountain sliding after heavy rain and earthquake), and it is difficult to achieve early warning;

[0065] (2) Limited accuracy: measurement accuracy is affected by terrain shading, weather (such as heavy fog and dust) and equipment erection error, and it is difficult to identify millimeter-level micro-displacement, and early risks are prone to be missed;

[0066] (3) Low coverage efficiency: single monitoring needs to set up equipment between each point, and only 10-15 monitoring points can be covered in a single day in mountainous areas, resulting in low overall monitoring efficiency;

[0067] (4) There are monitoring blind spots: remote and high-risk (such as steep slopes and rockfall areas) areas are difficult to reach on a regular basis, and have been in an unmonitored state for a long time.

[0068] The current detection method also uses traditional fixed monitoring devices for monitoring, specifically: single sensors or simple positioning equipment are arranged on the mountains along the transmission line, and the equipment is fixedly installed at the preset monitoring points; the sensor collects inclination angle or position coordinate data and stores it in the local storage module. However, the following shortcomings exist:

[0069] (1) Positioning accuracy is insufficient: single GPS module is easily blocked by mountains and trees, signal stability is poor, positioning error is often more than 1 meter, and millimeter-level micro displacement cannot be identified;

[0070] (2) Weak endurance: mostly dependent on ordinary lithium battery power supply, no energy supply design, usually less than 3 months of endurance in the wild, and the battery needs to be frequently replaced on site;

[0071] (3) Data transmission lag: the device has no real-time transmission function and needs manual data recovery; wireless transmission is limited by signal coverage and data loss often occurs in remote areas;

[0072] (4) Limited adaptability: the device structure is fixed and can only be installed on flat ground, and cannot adapt to complex terrains such as steep slopes and rock surfaces, and the monitoring range is limited.

[0073] The present application provides a power transmission line mountain displacement monitoring device, which takes "high-precision real-time monitoring, long-term operation in the wild, intelligent automatic early warning" as the core target, combines Beidou positioning, attitude sensing, edge computing, new energy power supply and remote communication technology, and builds a "perception-processing-transmission-analysis-early warning" whole-process closed-loop system, which can adapt to complex terrains and extreme climate environments in mountainous areas, and realize accurate monitoring and risk warning of mountain displacement.

[0074] Please refer to Figure 1 and Figure 2 , the present application provides a power transmission line mountain displacement monitoring device, which comprises:

[0075] a perception module for acquiring multi-source monitoring data of a power transmission line mountain monitoring point;

[0076] an edge computing module for identifying feature anomalies of the multi-source monitoring data to obtain feature classification data;

[0077] a transmission module for transmitting the feature classification data through two channels;

[0078] a platform for performing spatiotemporal coupling prediction on the received transmitted feature classification data to obtain predicted displacement values and predicted displacement trend curves, and making graded early warning decisions.

[0079] The perception module refers to a set of hardware components deployed at the power transmission line mountain monitoring point, which is used to collect various types of original data related to mountain displacement of the monitoring point.

[0080] The multi-source monitoring data refers to a comprehensive monitoring data set formed by the Beidou positioning module, the nine-axis attitude sensor, the temperature and humidity sensor and the rainfall sensor in the perception module.

[0081] Edge computing module refers to the computing unit integrated locally in the monitoring device, responsible for real-time processing of multi-source monitoring data such as denoising, feature extraction, and identifying abnormal features in the data through preset rules, and finally outputting classified feature data to realize "localization of data processing" to reduce transmission delay.

[0082] Feature anomaly recognition refers to the core operation performed by the edge computing module, which involves extracting key features such as displacement rate and inclination rate from multi-source monitoring data, comparing them with preset thresholds to determine whether the features are within the normal range, and distinguishing between "normal features" and "abnormal features".

[0083] Feature classification data refers to the structured data obtained after processing by the edge computing module, including normal feature data (such as displacement trend values during stable periods) and abnormal feature data (such as full-dimensional displacement, inclination, and environmental parameters at the time of exceeding the limit), both with timestamps to ensure time sequence.

[0084] Transmission module refers to the hardware component responsible for data transmission, which uses two different communication methods (such as 4G / 5G private network and Beidou short message) to build transmission channels, ensuring that feature classification data can be transmitted stably and timely from monitoring points to the platform.

[0085] Dual-channel transmission refers to the transmission strategy adopted by the transmission module, which involves using two communication channels (such as ground wireless communication and satellite communication) simultaneously or alternatively to transmit data, in order to cope with unstable signal scenarios in mountainous areas and ensure the continuity and reliability of data transmission.

[0086] Platform refers to the remote data processing and decision-making system that receives feature classification data sent by the transmission module, analyzes, predicts, and makes decisions through algorithm models, and is the core hub for implementing displacement prediction and early warning.

[0087] Spacetime coupling prediction refers to the data analysis method performed by the platform, which involves combining the time series changes and spatial feature associations of feature classification data to generate prediction results through algorithms such as time series convolution networks and spatiotemporal fusion models.

[0088] Predicted displacement value refers to one of the outputs of spacetime coupling prediction, which is the specific numerical value of mountain displacement at a future time (such as 1 hour or 24 hours) calculated by the model, reflecting short-term accurate displacement changes.

[0089] Predicted displacement trend curve refers to the second output of spacetime coupling prediction, which is the continuous change trajectory of mountain displacement over a future period (such as 7 days) fitted by the model, reflecting the long-term displacement evolution trend.

[0090] Hierarchical early warning decision-making refers to a process in which the platform determines the mountain displacement risk level (such as first, second, and third levels) based on the predicted displacement value and trend curve, in combination with preset risk level rules (such as displacement rate and acceleration threshold), and generates corresponding early warning instructions.

[0091] Multi-source monitoring data includes Beidou positioning data, nine-axis attitude data, and environmental data. The perception module includes a Beidou positioning module, a nine-axis attitude sensor, and an environmental sensor.

[0092] Beidou positioning module: The core component integrated with a Beidou satellite navigation system receiver, used to obtain the three-dimensional coordinates (longitude, latitude, and height) of the monitoring point, achieving high-precision positioning monitoring.

[0093] Attitude sensor: A combination sensor containing an accelerometer and a gyroscope, used to capture subtle mountain displacement (surface displacement, inclination angle, etc.) and changes in the device's own attitude.

[0094] Carrying Beidou-3 dual-frequency carrier phase difference (B1I / B2I frequency band), static positioning accuracy ≤3mm in plane and ≤5mm in height, dynamic sampling response ≤1 second, through anti-multipath algorithm to improve signal stability in complex terrain, fixed on the surface of the landslide of the monitoring point of the power transmission line mountain, metal / concrete observation pile preset, height ≥0.5m, avoiding ground deformation interference, real-time collection of monitoring point longitude, latitude, and height three-dimensional coordinates, positioning accuracy up to millimeter level, reflecting the spatial displacement of the mountain at this point.

[0095] Nine-axis attitude sensor integrates an accelerometer (±16g range), a gyroscope (±2000° / s range), and a magnetometer (±4800μT range), which can synchronously capture the micro-attitude changes of the mountain, with an output rate of 1-100Hz adjustable. It is integrated with the Beidou module on the same bracket, closely attached to the same metal base with a spacing of s10cm, collecting the attitude data of the monitoring point, including three-axis inclination angle (measurement range ±45°, accuracy ±0.01°) and vibration state.

[0096] It should be noted that the Beidou positioning module and the attitude sensor are integrated in one body to ensure physical position coincidence.

[0097] Environmental sensors include temperature and humidity sensors and rainfall sensors.

[0098] Temperature and humidity sensors are fixed in the lower part of the monitoring point bracket, 1.2~1.5m from the ground, collecting the temperature and humidity of the air around the mountain monitoring point (measurement range -40℃~85℃, accuracy ±0.5℃; humidity 0~100%RH, accuracy ±3%RH).

[0099] The rainfall sensor is arranged in an open and unobstructed area near the mountain monitoring point, is higher than the surrounding vegetation by 0.5 m or more, avoids being covered by fallen leaves and sundries, is fixed by an independent support, and collects rainfall in a preset area where the mountain monitoring point is located (resolution 0.2 mm, error ≤±4%).

[0100] The perception module takes the Beidou module PPS second pulse as a reference, triggers each sensor through hardware synchronization, ensures that the collection time deviation is ≤50us, all collected data are attached with millisecond-level UTC time stamps and device IDs, and after cache verification, the multi-source data are accurately associated in the spatial position and time dimensions.

[0101] It is worth mentioning that the Beidou positioning module and the nine-axis attitude sensor are installed at the same point, ensuring that the "spatial coordinates and attitude deformation" data of the same displacement monitoring point of the mountain are coupled; the temperature and humidity and rainfall are arranged around the area where the monitoring point is located, realizing the association of "mountain deformation and induced environmental factors".

[0102] The power transmission line mountain displacement monitoring device provided by the application has the characteristics of the characteristic classification data including a plurality of normal characteristic data and a plurality of abnormal characteristic data, and the edge calculation module including:

[0103] The edge calculation module is equipped with a low-power ARM Cortex-M7 embedded processor, a 512 kilobyte random access memory and a 4 megabyte flash memory, and supports local real-time data processing.

[0104] The preprocessing submodule is used for edge preprocessing of the multi-source monitoring data to obtain preprocessing data.

[0105] In the embodiment of the application, the monitoring data of the Beidou positioning module (three-dimensional coordinates + UTC time stamp), the attitude sensor (inclination angle + vibration acceleration), the temperature and humidity sensor and the rainfall sensor are received through a serial port, temporarily stored according to the device type, filtered for format error data, and classified preprocessing data are obtained, the preprocessing data contain independent time stamps, and at this time, the state is unsynchronized.

[0106] The cleaning and alignment submodule is used for cleaning and aligning the preprocessing data to obtain a plurality of multi-source fusion data frames.

[0107] In the embodiment of the application, the Beidou positioning module second pulse synchronization time stamp signal is taken as a reference, the mean value of a plurality of data of the nine-axis sensor within 1 second is aggregated, the temperature and humidity / rainfall data are interpolated according to time to complete, and a multi-source fusion data frame of 1 second is constructed.

[0108] It is worth mentioning that, based on the Beidou positioning module second pulse synchronization timestamp for data cleaning alignment, on the one hand, through the unified time reference, the time drift of multi-source devices such as nine-axis sensor, temperature and humidity / rainfall sensor is eliminated, ensuring the strict alignment of different types of monitoring data in the time dimension, avoiding the misjudgment of characteristics caused by time misalignment; on the other hand, the mean value of multiple data of the nine-axis sensor within 1 second is aggregated to smooth high-frequency noise and retain the core deformation characteristics, and the temperature and humidity / rainfall data are interpolated according to time to complete the sampling gap and ensure data continuity. Finally, each second of the multi-source fusion data frame is constructed, which realizes the spatio-temporal consistency integration of multi-source data and improves the data integrity and stability.

[0109] The noise suppression submodule is used for noise suppression of each multi-source fusion data frame to obtain a plurality of denoising fusion data frames.

[0110] In the embodiment of the application, the Beidou coordinates are filtered by 4-order multi-Bessel wavelet 5-layer decomposition to remove drift noise, the nine-axis data are filtered by Kalman filter to remove jitter interference, and the environmental data are corrected by three-sigma criterion to remove jump, thereby obtaining the denoising fusion data frame after denoising.

[0111] It is worth mentioning that, for the noise characteristics of different types of data, different noise suppression methods are adopted: the drift noise of Beidou coordinates is filtered by 4-order multi-Bessel wavelet 5-layer decomposition, which can effectively retain the millimeter-level micro-displacement signal and avoid the interference of long-term drift on positioning accuracy; the jitter interference of nine-axis data is removed by Kalman filter, which can smooth the high-frequency noise caused by instantaneous attitude motion and accurately extract stable tilt change characteristics; the jump of environmental data is corrected by three-sigma criterion, which can eliminate abnormal values caused by sudden interference and ensure the continuity of temperature, humidity and rainfall parameters. The denoising fusion data frame obtained thereby not only eliminates the specific noise of various data, but also retains the key features of mountain deformation and the real trend of environmental parameters.

[0112] The feature fusion submodule is used for feature extraction of each denoising fusion data frame to obtain a plurality of extracted feature data.

[0113] In the embodiment of the application, the displacement rate, cumulative displacement, tilt change rate, vibration amplitude and 24-hour cumulative rainfall are calculated based on the denoising fusion data frame, and are aggregated into an hourly extracted feature data set containing [displacement rate, cumulative displacement, tilt change rate, maximum amplitude and 24h rainfall].

[0114] It is worth mentioning that based on the denoising fusion data frame, the core features such as displacement rate and inclination change rate are calculated and aggregated into an hourly extracted feature dataset. On the one hand, by refining the key dynamic indicators of mountain deformation (such as displacement rate reflecting the speed of deformation and inclination change rate reflecting the posture trend) and environmental correlation parameters (24-hour cumulative rainfall), the original sensor data is converted into structured features that can be directly used for analysis, focusing on monitoring core information. On the other hand, the hourly aggregation processing not only simplifies the data size while preserving the time sequence trend, but also avoids the redundant transmission and processing burden of the original high-frequency data, and provides a representative input for the feature anomaly recognition of the edge computing module, making the anomaly discrimination accurately capture the deformation feature changes at the hourly scale.

[0115] The anomaly discrimination sub-module is used to judge whether each extracted feature data meets the preset anomaly judgment condition respectively.

[0116] If it meets, the extracted feature data is taken as the abnormal feature data.

[0117] If it does not meet, the key features are extracted from the extracted feature data for mean operation to obtain normal feature data.

[0118] In the embodiment of the present application, whether each extracted feature data meets the preset anomaly judgment condition is judged respectively, if it meets, the extracted feature data is taken as the abnormal feature data, the abnormal feature data contains [displacement rate, cumulative displacement, inclination change rate, maximum amplitude, 24h rainfall] and has a time stamp, if it does not meet, the key features are extracted from the extracted feature data, the key features are displacement rate and inclination change rate, the average rate within one hour and the average change rate within 1 hour are calculated to obtain normal feature data, the normal feature data contains [displacement rate average, inclination change rate average] and has a time stamp.

[0119] The preset anomaly judgment condition of the mountain displacement monitoring device for power transmission line provided by the present application is that any extracted feature in the extracted feature data is greater than the associated threshold value.

[0120] It should be noted that the threshold value is calculated based on the feature value (i.e. historical normal feature data) without obvious deformation for 3 months after the installation of the mountain displacement monitoring device for power transmission line.

[0121] S1, obtain displacement rate threshold and inclination change slope threshold;

[0122] In specific implementation, displacement rate threshold = historical mean + 3 times standard deviation, and inclination change slope threshold is the same.

[0123] S2, use environmental data to correct displacement rate threshold and inclination change slope threshold to generate displacement rate threshold and inclination change slope threshold.

[0124] In practice, if the 24-hour rainfall exceeds 50 mm, it is considered heavy rainfall. Since rainwater can exacerbate the risk of landslides, the threshold is lowered by 30%. The threshold for exceeding the displacement rate limit is 0.7 × displacement rate threshold, and the threshold for exceeding the slope change limit is 0.7 × slope change threshold.

[0125] In this embodiment of the invention, if the displacement rate does not exceed the corrected threshold, the tilt change rate exceeds the corrected threshold, and the maximum amplitude does not exceed the preset maximum amplitude over-limit threshold, then the extracted feature data is determined to meet the preset anomaly determination conditions.

[0126] If the displacement rate exceeds the corrected threshold, the tilt change rate does not exceed the corrected threshold, and the maximum amplitude does not exceed the preset maximum amplitude limit threshold, then the extracted feature data is determined to meet the preset anomaly judgment conditions.

[0127] If the displacement rate does not exceed the corrected threshold, the tilt change rate does not exceed the corrected threshold, and the maximum amplitude exceeds the preset maximum amplitude limit threshold, then the extracted feature data is determined to meet the preset anomaly judgment conditions.

[0128] If the displacement rate does not exceed the corrected threshold, the tilt change rate does not exceed the corrected threshold, and the maximum amplitude does not exceed the preset maximum amplitude limit threshold, then the extracted feature data is determined not to meet the preset anomaly judgment conditions.

[0129] It is worth mentioning that by simultaneously correcting the displacement rate threshold and the tilt change slope threshold using 24-hour rainfall data, rather than adjusting only a single threshold, the coupling characteristics of geological risks can be accurately matched. The risk of landslides in mountainous areas is not solely determined by displacement or tilt, but rather by the coupling of "displacement / tilt changes + environmental triggers." During heavy rainfall, rainwater seeps into the mountainside, reducing the strength of the soil and rock mass, which may lead to an increased displacement rate or a sudden change in the tilt slope. Simultaneously correcting both types of thresholds avoids misjudgments such as "correcting only the displacement threshold while overlooking tilt risks," ensuring that the anomaly judgment logic is consistent with the actual physical laws of mountain instability, and significantly reducing the risk. The probability of missing early risks due to incomplete threshold adjustment can be reduced. At the same time, it is necessary to balance the sensitivity of early warning and the false alarm rate. If only the displacement rate threshold is adjusted while the tilt change slope threshold remains unchanged, there may be a lag problem during heavy rainfall where "the tilt has exceeded the stable threshold but is judged as normal because it has not been corrected". Or, to avoid missing reports, blindly lowering a single threshold will lead to an increase in false alarms. Therefore, adjusting the two types of core thresholds simultaneously can not only improve the overall sensitivity of anomaly identification when environmental risks increase, but also avoid false alarms caused by excessive adjustment of a certain threshold based on the basic threshold logic of "historical mean + 3 times standard deviation", thus achieving a balance between "high sensitivity early warning" and "low false alarm rate".

[0130] The environmental data acquired by the perception module provides a basis for threshold correction. The edge computing module makes judgments based on the corrected dual thresholds, ensuring that the identification of abnormal feature data is more in line with the deformation law of mountains in complex mountain environments. It not only dynamically adjusts the judgment criteria through real-time environmental data to avoid the judgment deviation of fixed thresholds when the environment changes, but also improves the accuracy of anomaly identification through synchronous correction of dual thresholds, thereby reducing monitoring errors caused by unreasonable threshold design from the source.

[0131] Please see Figure 2 The present invention provides a power transmission line mountain displacement monitoring device, the transmission module comprising:

[0132] The signal strength submodule is used to acquire signal strength data from the monitoring device in real time.

[0133] In this embodiment of the invention, the signal strength data of the power transmission line mountain displacement monitoring device is acquired in real time.

[0134] The classification encryption and encapsulation submodule is used to classify and encrypt feature classification data to obtain encrypted data packets containing identifiers.

[0135] In this embodiment of the invention, based on the feature classification data output by the edge computing module, abnormal feature data is encrypted using IPSec VPN (Internet Protocol Security Virtual Private Network), and "excessive data" is marked and encapsulated in real time (delay ≤ 10 seconds). Normal feature data is encapsulated in batches every hour to generate encrypted data packets with data type markings.

[0136] It is worth mentioning that IPSec VPN encryption effectively ensures the security and integrity of feature classification data transmission, preventing data leakage or tampering. This provides data quality support for the platform to conduct accurate spatiotemporal analysis of displacement, generate reliable predicted displacement values ​​and trend curves, and solidify the foundation for high-precision monitoring. At the same time, classification and encapsulation based on data type differences, with real-time encapsulation of abnormal feature data with a low latency of ≤10 seconds, ensures the rapid uploading of abnormal mountain displacement signals, meeting the real-time monitoring requirements. Normal feature data is encapsulated in batches every hour, which reduces the bandwidth consumption of invalid transmissions, further optimizing the efficiency of dual-channel transmission. Moreover, data type marking allows the platform to quickly identify data categories and process them accordingly.

[0137] The dual-channel transmission submodule is used to select the corresponding transmission channel to transmit encrypted data packets to the platform based on the signal strength data acquired at the current moment.

[0138] In this embodiment of the invention, the signal strength data acquired at the current moment is compared with a preset signal strength threshold, and the corresponding transmission channel is selected based on the comparison result to transmit the encrypted data packet to the platform.

[0139] The present invention provides a power transmission line mountain displacement monitoring device, the dual-channel transmission submodule comprising:

[0140] The main channel execution submodule is used to transmit encrypted data packets to the platform through the main channel when the signal strength data is greater than or equal to a preset signal strength threshold.

[0141] The backup channel execution submodule is used to divide the encrypted data packet into multiple fragment packets when the signal strength data is less than the preset signal strength threshold, and transmit the multiple fragment packets in the fragment packet stream to the platform in sequence through the backup channel according to the preset time interval.

[0142] It should be noted that, as Figure 2 As shown, the main channel uses a 4G full-network compatible module (supporting LTE Cat.1 Category 1 Long Term Evolution technology, with a transmission rate of 10Mbps), employing APN private network encryption for transmission and complying with the GB / T36572-2018 security standard. The backup channel uses BeiDou short message communication (100 bytes per message, configurable transmission interval, minimum 10 minutes), automatically switching in public network signal dead zones (such as deep mountain valleys) to ensure uninterrupted data transmission. The entire link employs a triple security strategy of encryption, verification, and access control, combined with intelligent switching between satellite and public networks, to construct a reliable communication system of "stable transmission + emergency backup."

[0143] In this embodiment of the invention, based on the detected 4G signal strength, if it is ≥-110dBm, an encrypted marker message is sent through the 4G APN private network. Based on the identifier, abnormal feature data is uploaded in real time and normal feature data is pushed periodically. If the 4G signal strength is <-110dBm, the BeiDou short message channel (i.e., the backup channel) is triggered. The encrypted marker message is fragmented into 100-byte segments, and a sequence number and CRC32 check code are added. The segments are sent at intervals of ≥10 minutes. The fragment integrity is ensured through a retransmission mechanism, and a BeiDou fragmented message stream is generated.

[0144] This invention provides a power transmission line mountain displacement monitoring device, the platform comprising:

[0145] The data aggregation submodule is used to receive encrypted data packets and fragmented packet streams, aggregate them, and obtain the transmitted feature classification data.

[0146] In this embodiment of the invention, the data aggregation submodule receives encrypted data packets and fragmented packet streams, parses and restores them into abnormal / normal feature data, and connects to the IoT platform data middleware via the MQTT protocol (Message Queuing Telemetry Transport). It uses the InfluxDB (InfluxDB Time Series Database) time series database adapted for high-frequency time series data, and stores the data in partitions according to "device ID + date". At the same time, it uses Redis (Remote Dictionary Server) to cache the data of the most recent 24 hours, supporting millisecond-level queries, forming a structured multi-source time series database. Both normal and abnormal feature data are continuously stored in the platform's structured multi-source time series database according to timestamps.

[0147] The temporal convolutional submodule is used to input the transmitted feature classification data into a pre-trained temporal convolutional network to predict displacement and obtain the predicted displacement value.

[0148] The pre-trained temporal convolutional network employs Bayesian optimization-temporal convolutional network, a hybrid model that combines the temporal modeling capability of Temporal Convolutional Network (TCN) with the hyperparameter optimization capability of Bayesian optimization (BO). Its core principle can be broken down into the synergistic effect of two parts:

[0149] Temporal Convolutional Networks (TCNs) are deep learning models designed for time series data. Their core uses causal convolution (ensuring that only historical data is used and no future information is introduced, which conforms to the logic of time series prediction) and dilated convolution (by expanding the receptive field of the convolution kernel, capturing long-distance temporal dependencies, such as the cumulative impact of mountain displacement over several consecutive days) to extract dynamic features in time series. They can effectively handle multi-dimensional time series data (such as the coordinated changes of displacement, tilt, and rainfall) and output accurate short-term prediction results (such as displacement values ​​in the next 24 hours).

[0150] Bayesian optimization (BO) is an efficient hyperparameter optimization method. It constructs a probabilistic model of the objective function based on Bayes' theorem (usually using a Gaussian process), and iteratively selects the hyperparameter combination that is "most likely to improve model performance" to replace the blindness of traditional grid search. It can quickly find the optimal hyperparameter configuration under limited computing resources.

[0151] When the two are combined, Bayesian optimization provides adaptive hyperparameter tuning for TCN, solving the problems of low efficiency and reliance on experience in manual parameter tuning of TCN; while TCN, with the help of the optimized network structure, can more accurately capture hidden patterns in time series data.

[0152] In this embodiment of the invention, feature classification data after transmission in the most recent 72 hours, including displacement, tilt angle and rainfall, are extracted from the platform's structured multi-source time series database. The "spatiotemporal coupling relationship of displacement, tilt and rainfall" is learned through a temporal convolutional network to predict the displacement value per hour in the next 24 hours, thereby achieving short-term fine prediction.

[0153] The spatiotemporal feature fusion submodule is used to input the transmitted feature classification data into a pre-trained spatiotemporal feature fusion prediction network to perform trend prediction and obtain the predicted displacement trend curve.

[0154] The pre-trained spatiotemporal feature fusion prediction network employs an ensemble empirical mode decomposition-convolutional neural network-long short-term memory network. Leveraging the temporal preprocessing capabilities of ensemble empirical mode decomposition (EEMD), the spatial feature extraction capabilities of convolutional neural networks (CNN), and the long temporal dependency capture capabilities of long short-term memory networks (LSTM), it completes the processing and fusion of spatiotemporal data in stages. Based on historical data, it pre-trains and optimizes parameters to ultimately achieve high-precision prediction. Its fundamental principle can be broken down into a logical chain of "temporal deconstruction - spatial encoding - spatiotemporal fusion prediction":

[0155] First, an ensemble Empirical Mode Decomposition (EEMD) module is used as a preprocessing module for non-stationary, multi-scale spatiotemporal time series data such as mountain displacement (e.g., displacement sequences at different monitoring points and accompanying environmental time series data). Through iterative operations of "adding Gaussian white noise multiple times + Empirical Mode Decomposition (EEMD)," the original complex time series data is decomposed into several Intrinsic Mode Functions (IMFs) and one residual component. The IMFs correspond to stationary subsequences of different frequencies in the data, and the residual component corresponds to the overall trend of the data. This solves the problem of interference in subsequent model feature extraction caused by the non-stationarity of the original time series data, laying a "clean" time series foundation for subsequent spatiotemporal feature fusion.

[0156] Secondly, the Convolutional Neural Network (CNN) is responsible for spatial feature extraction: For multi-source monitoring data (such as displacement and tilt data of multiple mountain monitoring points, or multi-dimensional sensing data of the same monitoring point), CNN captures the spatial correlation between different monitoring points (such as deformation coordination of adjacent monitoring points, deformation differences in different regions) and the spatial coupling relationship of multi-dimensional data of the same monitoring point (such as spatial linkage between horizontal and vertical displacement) through the local perception and parameter sharing mechanism of convolutional kernels. It encodes discrete spatial data into structured spatial feature vectors, making up for the lack of spatial dimension analysis in single time series models.

[0157] Finally, the Long Short-Term Memory (LSTM) network undertakes the task of spatiotemporal feature fusion and prediction: On the one hand, LSTM utilizes gating mechanisms (input gate, forget gate, output gate) to effectively capture the long-term temporal dependencies of each stationary subsequence after EEMD decomposition (such as the cumulative change of a certain IMF component over time, and the temporal correlation between different IMFs); on the other hand, it integrates the spatial feature vectors extracted by CNN into the temporal modeling process, achieving a deep fusion of "dynamic laws of temporal subsequences + spatial correlation features"; at the same time, the "pre-trained" characteristic of the network means that the model has completed parameter optimization (such as the noise intensity of EEMD, the kernel size / number of layers of CNN, the number of hidden units of LSTM, etc.) through a large amount of historical spatiotemporal data (such as the spatiotemporal sequence of past mountain displacement, and the corresponding environmental data), ensuring that the optimized parameters can be quickly called up during actual prediction, and accurately output prediction results containing spatiotemporal correlation information (such as the displacement values ​​and displacement trends of multiple monitoring points in a future period).

[0158] The logical closed loop of the entire network is as follows: EEMD first "deconstructs" the complex temporal sequence to reduce interference, CNN then "encodes" spatial correlation to supplement the dimension, and LSTM finally "fusions" the spatiotemporal features and combines them with pre-trained parameters to achieve prediction. The three work together to solve the problem that a single model is difficult to process spatiotemporal data and capture multi-scale patterns at the same time. It is especially suitable for prediction scenarios such as mountain displacement of transmission lines, which require combining the spatial correlation of multiple monitoring points and long-term deformation patterns.

[0159] In this embodiment of the invention, the feature classification data after transmission in the most recent 72 hours, including the BeiDou coordinate sequence, is extracted from the platform's structured multi-source time-series database. It should be noted that the abnormal feature data and normal feature data transmitted by the transmission module, such as cumulative displacement and displacement rate, are essentially derived results of BeiDou coordinates. BeiDou coordinates exist in the form of derived features such as "displacement rate and cumulative displacement", which not only meets the requirements of "deformation monitoring" but also achieves efficient data transmission. The value of the original coordinates has been released in advance through "feature calculation" and there is no need for redundant transmission. Therefore, the cumulative displacement output by the edge computing module is directly used as the instantaneous value of Δx input to the algorithm, and the displacement rate is restored to the change in Δx through reverse calculation, and finally integrated into the BeiDou coordinate sequence required by the algorithm.

[0160] In the specific implementation, the BeiDou coordinate sequence is decomposed into 10 intrinsic mode functions through EEMD decomposition. The top 3 low-frequency IMFs are selected to reflect the long-term trend. CNN extracts the spatial distribution features of the low-frequency IMFs, and LSTM captures the time dependence to predict the displacement trend curve for the next 7 days.

[0161] The index calculation submodule is used to determine the real-time displacement rate index and acceleration index based on the predicted displacement value and the predicted displacement trend curve.

[0162] The indicator calculation submodule includes:

[0163] The first calculation unit is used to calculate the real-time displacement rate index based on the predicted displacement value using the sliding window method.

[0164] In this embodiment of the invention, for the predicted displacement value, a sliding window of 1 hour is used to calculate the displacement change Δs within the window. Combined with the time interval Δt seconds, the real-time displacement rate v=Δs / Δt within the prediction period is derived.

[0165] The second calculation unit is used to fit a quadratic curve using the least squares method based on the predicted displacement trend curve, and extract the acceleration term of the quadratic curve as an acceleration index.

[0166] In this embodiment of the invention, the predicted displacement trend curve is fitted with a quadratic curve s(t)=at2+bt+c by the least squares method, and the acceleration term a is extracted as the acceleration index.

[0167] The early warning classification submodule is used to make classified early warning decisions using real-time displacement rate and acceleration indicators, and to generate early warning decision instructions.

[0168] It should be noted that the upper and lower limits of the first, second, and third acceleration threshold intervals can be corrected by the 24-hour rainfall. The specific correction method is the same as step S2 above, and will not be repeated here.

[0169] The early warning classification submodule includes:

[0170] The first early warning unit is used to generate a first-level early warning command when the real-time displacement rate index is in the first displacement rate threshold range and the acceleration index is in the first acceleration threshold range.

[0171] The second early warning unit is used to generate a secondary early warning command when the real-time displacement rate index is in the second displacement rate threshold range or the acceleration index is in the second acceleration threshold range.

[0172] The third early warning unit is used to generate a level 3 early warning command when the real-time displacement rate index is in the third displacement rate threshold range or the acceleration index is in the third acceleration threshold range.

[0173] The warning level table is shown below:

[0174]

[0175] Level 1 Warning Instruction (Attention Level): The displacement rate is low and stable, the environment is not significantly abnormal, and the platform is marked as low risk only. Trend reports are generated regularly.

[0176] Level 2 Warning Instruction (Warning Level): The rate of rainfall is moderate or shows an accelerating trend, accompanied by adverse environmental conditions such as rainfall. The warning is sent to the person in charge, requiring on-site verification within 48 hours.

[0177] Level 3 Early Warning Order (Emergency Level): If the rate is high or a significant increase is predicted in the short term, the highest level of response will be triggered immediately, and the information will be simultaneously sent to the emergency command center to generate a response plan.

[0178] It is worth mentioning that the power transmission line mountain displacement monitoring device of the present invention adopts a dual power supply mode of solar energy + lithium battery, optimizes charging efficiency through MPPT technology, supports automatic switching between three modes of sleep, low power consumption and full power consumption, and adapts to different operating states by relying on task scheduling algorithm to ensure long-term stable operation.

[0179] The monitoring platform presents the status of monitoring points in the form of heat maps, displacement vector maps, etc., and supports 3D terrain modeling; the mobile APP can push early warning information in real time, provide functions such as historical data query and electronic fence setting, and is suitable for use in offline field scenarios.

[0180] The power transmission line mountain displacement monitoring device of this invention adopts a modular structural design, with an outer shell reaching IP65 protection level, making it adaptable to extreme working conditions such as high humidity in coastal areas and low temperature in high-altitude areas. Positioning, sensing, and communication modules can all be independently disassembled, supporting rapid replacement and remote firmware upgrades. Simultaneously, electromagnetic compatibility design (double-layer metal shielding shell and lightning protection grounding) enhances anti-interference capabilities, while CRC cyclic redundancy check and digital signature technology ensure data integrity, preventing information loss or tampering during transmission and ensuring stable operation in complex environments.

[0181] This invention has the following advantages:

[0182] (1) High-precision positioning technology: High-precision real-time positioning is achieved by relying on satellite systems such as Beidou. The key lies in signal stability under complex terrain and error reduction through multi-system fusion.

[0183] (2) Attitude and displacement sensing technology: using high-sensitivity sensors to measure minute displacements and tilts, the core is to ensure reliability in extreme environments and reasonable sensor placement.

[0184] (3) Data fusion and analysis algorithm: The algorithm integrates data from multiple sensors, processes interference, analyzes displacement trends and anomalies, and optimizes the algorithm to improve efficiency and early warning accuracy.

[0185] (4) Energy self-sufficiency system: It adopts solar energy + lithium battery power supply and low power consumption management to ensure long-term unattended operation in remote areas.

[0186] (5) Remote communication and data transmission: Integrates 4G (APN card) and other modules to ensure that monitoring data is transmitted to the monitoring center in real time and stably.

[0187] (6) Ease of installation and ease of maintenance: Considering the ease of actual deployment and subsequent maintenance, the device is designed to be easy to install and maintain, while also providing remote diagnostic and update functions.

[0188] Please see Figure 3 The present invention provides a monitoring method for a power transmission line mountain displacement monitoring device, comprising:

[0189] Step 101: Obtain multi-source monitoring data from mountain monitoring points along the transmission line;

[0190] Step 102: Perform feature anomaly identification on the multi-source monitoring data to obtain feature classification data;

[0191] Step 103: Transmit the feature classification data through dual channels;

[0192] Step 104: Perform spatiotemporal coupling prediction on the received transmitted feature classification data to obtain the predicted displacement value and the predicted displacement trend curve, and make hierarchical early warning decisions.

[0193] In this embodiment of the invention, a sensing module collects multi-source monitoring data in real time at pre-set monitoring points along the power transmission line (such as landslide hazard zones around towers and steep slope areas). This multi-source monitoring data includes millimeter-level three-dimensional coordinate data output by the BeiDou positioning module, tilt angle and vibration data collected by a nine-axis attitude sensor, and environmental parameter data obtained by temperature and humidity sensors and rainfall sensors. All data carries a timestamp corresponding to the collection time, ensuring the spatiotemporal correlation of the data. The edge computing module performs anomaly identification of the multi-source monitoring data based on a pre-set dynamic threshold algorithm, first converting the multi-source data into core features such as displacement rate, tilt change rate, and 24-hour cumulative rainfall. Then, by combining real-time environmental parameters, the threshold is dynamically adjusted (e.g., the displacement rate threshold is lowered by 30% when rainfall exceeds 50mm), thereby dividing the feature data into normal feature data (e.g., the average displacement trend per hour) and abnormal feature data exceeding the limit, forming feature classification data. This enables local real-time identification of abnormal data, avoiding delays caused by data transmission to the platform for analysis and ensuring real-time monitoring. Dual-channel transmission is completed by the transmission module, specifically using a dual transmission channel composed of a 4G APN private network and Beidou short message service. If the monitoring point is located in a mountainous area with weak 4G signal, it automatically switches to the Beidou short message service to ensure no data loss. If it is located in an area with normal signal, the normal feature data is processed separately. Data and anomaly feature data are transmitted hourly via a dedicated 4GAPN network, ensuring both real-time performance and transmission efficiency. The platform collaboratively performs spatiotemporal coupling prediction of the received, transmitted feature classification data using a pre-trained dual-network model. Specifically, the BO-TCN network model performs short-term spatiotemporal coupling calculations on multi-dimensional feature data (time series of displacement, tilt, and environmental parameters), outputting predicted displacement values ​​for the next 24 hours. Simultaneously, the EEMD-CNN-LSTM network model performs long-term spatiotemporal feature mining on continuous time series data with a single displacement dimension. After noise filtering via EEMD decomposition, the CNN extracts spatial distribution features, and the LSTM... The TM captures time-dependent relationships and generates a predicted displacement trend curve for the next 7 days. Subsequently, the platform combines the predicted displacement values ​​to calculate the real-time displacement rate, fits a quadratic equation based on the predicted displacement trend curve to extract deformation acceleration, and compares it with dynamically corrected thresholds (Level 1: displacement rate ≤ low threshold and acceleration ≤ low threshold; Level 2: displacement rate between low threshold and medium threshold or acceleration between low threshold and medium threshold; Level 3: displacement rate > medium threshold or acceleration > medium threshold) to make graded early warning decisions. Finally, it generates an early warning decision instruction that includes the early warning level, predicted data, and related environmental parameters, and pushes it to the transmission line operation and maintenance management terminal to provide operation and maintenance personnel with accurate handling basis.

[0194] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0195] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0196] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power line mountain displacement monitoring device, characterized by, The method comprises the following steps: a perception module is used to obtain multi-source monitoring data of a power transmission line mountain monitoring point; an edge computing module is used to perform feature anomaly identification on the multi-source monitoring data to obtain feature classification data; a transmission module is used to perform double-channel transmission on the feature classification data; a platform is used to perform space-time coupling prediction on the received feature classification data after transmission to obtain predicted displacement values and predicted displacement trend curves, and to perform hierarchical early warning decision-making.

2. The power line mountain displacement monitoring apparatus according to claim 1, characterized by, The feature classification data includes multiple normal feature data and multiple abnormal feature data, and the edge computing module includes: a preprocessing submodule is used to perform edge preprocessing on the multi-source monitoring data to obtain preprocessed data; a cleaning and alignment submodule is used to clean and align the preprocessed data to obtain multiple multi-source fusion data frames; a noise suppression submodule is used to suppress noise of each multi-source fusion data frame to obtain multiple denoising fusion data frames; a feature fusion submodule is used to extract features from each denoising fusion data frame to obtain multiple extracted feature data; an anomaly discrimination submodule is used to determine whether each extracted feature data meets a preset anomaly judgment condition; if yes, the extracted feature data is taken as the abnormal feature data; if no, key features are extracted from the extracted feature data for mean value operation to obtain the normal feature data.

3. The power line mountain displacement monitoring apparatus according to claim 2, wherein The preset anomaly judgment condition is that any extracted feature in the extracted feature data is greater than an associated limit threshold.

4. The power line mountain displacement monitoring apparatus according to claim 1, wherein The transmission module includes: a signal strength submodule is used to obtain signal strength data of a monitoring device in real time; a classification encryption packaging submodule is used to classify and encrypt the feature classification data to obtain encrypted data messages containing identifiers; a double-channel transmission submodule is used to select a corresponding transmission channel according to the signal strength data obtained at the current time to transmit the encrypted data messages to the platform.

5. The power line mountain displacement monitoring apparatus according to claim 4, wherein The double-channel transmission submodule includes: a main channel execution submodule is used to transmit the encrypted data messages to the platform through a main channel when the signal strength data is greater than or equal to a preset signal strength threshold; a backup channel execution submodule is used to fragment the encrypted data messages into multiple fragmented messages when the signal strength data is less than the preset signal strength threshold, and to sequentially transmit the multiple fragmented messages in the fragmented message stream to the platform through a backup channel at a preset time interval.

6. The power line mountain displacement monitoring apparatus according to claim 5, wherein The platform includes: a data aggregation submodule is used to aggregate the encrypted data messages and the fragmented message stream to obtain the feature classification data after transmission; a time series convolution submodule is used to input the feature classification data after transmission into a pre-trained time series convolution network to perform displacement prediction and obtain predicted displacement values; a space-time feature fusion submodule is used to input the feature classification data after transmission into a pre-trained space-time feature fusion prediction network to perform trend prediction and obtain a predicted displacement trend curve; an index calculation submodule is used to determine a real-time displacement rate index and an acceleration index according to the predicted displacement values and the predicted displacement trend curve, The early warning grading submodule is configured to make grading early warning decisions by using the real-time displacement rate index and the acceleration index, and generate early warning decision instructions.

7. The power line mountain displacement monitoring apparatus according to claim 6, wherein The index calculation submodule includes: A first calculation unit configured to calculate the real-time displacement rate index by using a sliding window method based on the predicted displacement value; A second calculation unit configured to extract an acceleration term of a quadratic curve as the acceleration index by fitting the quadratic curve by using a least square method based on the predicted displacement trend curve.

8. The power line mountain displacement monitoring apparatus according to claim 6, wherein The early warning grading submodule includes: A first early warning unit configured to generate a first-level early warning instruction when the real-time displacement rate index is in a first displacement rate threshold interval and the acceleration index is in a first acceleration threshold interval; A second early warning unit configured to generate a second-level early warning instruction when the real-time displacement rate index is in a second displacement rate threshold interval or the acceleration index is in a second acceleration threshold interval; A third early warning unit configured to generate a third-level early warning instruction when the real-time displacement rate index is in a third displacement rate threshold interval or the acceleration index is in a third acceleration threshold interval.

9. The power line mountain displacement monitoring apparatus according to any one of claims 1 to 8, characterized by, The multi-source monitoring data includes Beidou positioning data, nine-axis attitude data, and environmental data.

10. A monitoring method applied to the power transmission line mountain displacement monitoring device according to any one of claims 1-9, characterized in that, The method includes: Obtaining multi-source monitoring data of a power transmission line mountain monitoring point; Performing feature anomaly recognition on the multi-source monitoring data to obtain feature classification data; Performing double-channel transmission on the feature classification data; Performing spatiotemporal coupling prediction on the received transmitted feature classification data to obtain a predicted displacement value and a predicted displacement trend curve, and making grading early warning decisions.

Images