Cableway operation safety assessment method, system and equipment based on multi-sensor fusion and medium
By deploying multiple types of sensors on the cableway system and combining them with a random forest model for risk assessment, the problems of single monitoring dimensions and insufficient real-time performance in the cableway safety monitoring system have been solved. This has enabled multi-dimensional real-time monitoring and efficient risk assessment of the cableway, reducing safety risks and improving assessment accuracy and operational efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG LANGCHAO SMART CULTURAL TOURISM IND DEV CO LTD
- Filing Date
- 2025-12-19
- Publication Date
- 2026-05-01
AI Technical Summary
Existing cableway safety monitoring systems suffer from problems such as limited monitoring dimensions, insufficient real-time performance, limited sensor types, insufficient deployment density, inability to capture subtle anomalies throughout the entire cableway chain, long manual inspection and offline testing cycles, inability to cope with instantaneous failures, communication technology is susceptible to interference in mountainous areas, resulting in data delays or loss, and data analysis lacks multi-source fusion capabilities, leading to high false alarm and false negative rates.
By employing a multi-sensor fusion approach, wind speed sensors, triaxial inertial measurement units, rope tension sensors, rope wheel vibration sensors, motor current sensors, and rope offset sensors are deployed on the cableway system. Multi-dimensional sensor arrays are acquired through a preset time window, single-sensor risk score data is calculated, and risk assessment is performed using preset logical rules and a random forest model to achieve real-time high-risk early warning and safety risk assessment.
It enables multi-dimensional real-time monitoring of the cableway, accurately capturing subtle anomalies such as wire rope wear and structural deformation, reducing safety risks, improving the accuracy and real-time nature of assessments, reducing false alarm rates, and optimizing operational efficiency.
Smart Images
Figure CN121961209A_ABST
Abstract
Description
A method, system, equipment, and medium for cableway operation safety assessment based on multi-sensor fusion. Technical Field
[0001] This application relates to the field of safety assessment technology, and in particular to a method, system, equipment and medium for cableway operation safety assessment based on multi-sensor fusion. Background Technology
[0002] As an important mode of passenger transport for mountainous areas and tourism, the safety of cableways is directly related to the safety of people's lives and property. At present, cableway safety monitoring mainly relies on a small number of sensors such as anemometers, door switches, braking circuits, and tension sensors, as well as regular manual inspections and offline detection methods such as magnetic resonance imaging (MRT).
[0003] The core problems of the current cableway safety monitoring system lie in its limited monitoring dimensions and insufficient real-time performance. On the one hand, it relies on a limited range of sensor types (e.g., a lack of continuous monitoring of key parameters such as vibration and temperature), and the sensor density is insufficient to capture subtle anomalies throughout the entire cableway chain. On the other hand, manual inspections and offline testing have long cycles (e.g., daily or weekly), making it difficult to cope with instantaneous failures; while technologies such as magnetic induction can detect wire rope wear, they require shutdown operations, impacting operational efficiency. The limitations of communication technology further exacerbate the risks: wireless transmission is susceptible to interference in mountainous areas, and the high cost of fiber optic deployment leads to data delays or loss. Furthermore, existing systems rely heavily on simple rules such as threshold alarms for data analysis, lacking the ability to fuse and analyze multi-source data, resulting in high false alarm and false negative rates. Summary of the Invention
[0004] This application provides a method, system, device, and medium for cableway operation safety assessment based on multi-sensor fusion, in order to solve the problems of single monitoring dimensions and insufficient real-time performance in existing solutions.
[0005] Firstly, this application provides a method for safety assessment of cableway operation based on multi-sensor fusion. The method includes: deploying several types of sensors on the cableway system; acquiring a multi-dimensional sensor array collected by several types of sensors according to a preset time window; calculating single-sensor risk score data for the multi-dimensional sensor array within the preset time window; determining whether a preset high risk exists based on preset logical rules and the single-sensor risk score data; directly outputting a high-risk warning when a preset high risk exists; and inputting the multi-dimensional sensor array into a trained random forest model when no preset high risk exists to obtain a safety risk assessment result.
[0006] In one implementation of this application, several types of sensors are deployed on the cableway system, specifically including: a wind speed sensor installed at a preset location on the line to collect ambient wind speed; a three-axis inertial measurement unit installed in the carriage to acquire three-axis acceleration and angular velocity and calculate the carriage's sway amplitude and attitude; a rope tension sensor installed on the tensioning device to collect the tension force of the traction cable; a rope sheave vibration sensor installed on the key tower sheave group or the drive-side sheave bracket to collect rope sheave vibration signals; a motor current sensor installed in the drive motor circuit to collect motor current data; and a rope offset sensor arranged at a preset location on the support / pressure sheave to monitor the offset of the traction cable relative to the sheave groove; and all sensors collect data synchronously according to a unified time reference.
[0007] In one implementation of this application, a multidimensional sensor array is obtained from several types of sensors according to a preset time window. Specifically, this includes: reading all sensor data uploaded by several types of sensors once according to the preset time window; and combining the sensor data collected by each sensor at the same time stamp into a multidimensional sensor array according to the collection timestamp in the sensor data.
[0008] In one implementation of this application, a single sensor includes at least: a wind speed sensor, a three-axis inertial measurement unit, a rope tension sensor, a rope pulley vibration sensor, a motor current sensor, and a rope offset sensor. The risk score for a single sensor is calculated for a multi-dimensional sensor array within a preset time window, specifically including: calculating the average wind speed, maximum wind speed, and wind speed change rate based on the ambient wind speed in the multi-dimensional sensor array within the preset time window; obtaining the risk score data for the wind speed sensor based on the scoring intervals into which the average wind speed, maximum wind speed, and wind speed change rate fall; calculating the peak yaw angle, yaw RMS, and main yaw frequency based on the carriage sway amplitude and attitude in the multi-dimensional sensor array within the preset time window; obtaining the risk score data for the three-axis inertial measurement unit based on the scoring intervals into which the peak yaw angle, yaw RMS, and main yaw frequency fall; and calculating the risk score for the traction cable tension based on the tension in the multi-dimensional sensor array within the preset time window. Calculate the average tension, tension deviation, and tension change rate; obtain the risk score data for the rope tension sensor based on the scoring intervals into which the average tension, tension deviation, and tension change rate fall; calculate the vibration RMS, peak value, kurtosis, and frequency band energy based on the rope sheave vibration signal in the multi-dimensional sensor array within a preset time window; obtain the risk score data for the rope sheave vibration sensor based on the scoring intervals into which the vibration RMS, peak value, kurtosis, and frequency band energy fall; calculate the current RMS, fluctuation amplitude, and surge amount based on the motor current data in the multi-dimensional sensor array within a preset time window; obtain the risk score data for the motor current sensor based on the scoring intervals into which the current RMS, fluctuation amplitude, and surge amount fall; calculate the deviation excess ratio based on the wheel groove offset in the multi-dimensional sensor array within a preset time window; determine the risk score data for the rope deviation sensor based on the scoring interval into which the deviation excess ratio falls.
[0009] In one implementation of this application, the existence of a preset high risk is determined by using preset logical rules and single-sensor risk scoring data. Specifically, this includes: determining the existence of a preset high risk when the average wind speed score in the single-sensor risk scoring data is greater than a preset wind speed threshold, the maximum wind speed score is greater than a preset maximum wind speed threshold, and the wind speed change rate score is greater than a preset wind speed change rate threshold; determining the existence of a preset high risk when the peak yaw angle score in the single-sensor risk scoring data is greater than a preset yaw angle threshold, the yaw RMS score is greater than a preset yaw threshold, and the main yaw frequency score is greater than a preset main yaw frequency threshold; and determining the existence of a preset high risk when the average tension score in the single-sensor risk scoring data is greater than a preset average tension threshold, the tension deviation score is greater than a preset tension deviation threshold, and the tension change rate score is greater than a preset threshold. When a tension change rate threshold is set, a preset high risk is identified. A preset high risk is identified when the vibration RMS score, peak value, kurtosis, and frequency band energy scores in the single-sensor risk scoring data all exceed the preset vibration RMS threshold, the peak value exceeds the preset peak value threshold, the kurtosis score exceeds the preset kurtosis threshold, and the frequency band energy score exceeds the preset energy threshold. A preset high risk is identified when the current RMS score, fluctuation amplitude, and surge value in the single-sensor risk scoring data all exceed the preset current RMS threshold, the fluctuation amplitude score exceeds the preset fluctuation threshold, and the surge value exceeds the preset surge threshold. A preset high risk is identified when the deviation exceeding the limit ratio score in the single-sensor risk scoring data exceeds the preset deviation threshold. The average of the average wind speed score, the maximum wind speed score, and the wind speed change rate score is taken as the wind speed risk score. The average of the peak yaw angle score, the RMS yaw score, and the main yaw frequency score is taken as the sway risk score. The average of the average tension score, tension deviation score, and tension change rate score is taken as the tension risk score. The average of the vibration RMS score, peak value, and kurtosis score is taken as the rope pulley vibration risk score. The average of the current RMS score, fluctuation range score, and surge value score is taken as the motor current risk score. The percentage of deviation exceeding the limit is taken as the rope deviation risk score. ; Through the formula: Calculate the overall score ;in, , , , , , When the overall score If the score exceeds the preset comprehensive score threshold, a preset high risk is identified.
[0010] In one implementation of this application, after determining that a preset high risk exists, the method further includes: determining whether the trigger for the preset high risk is a preset comprehensive score threshold; when it is not a preset comprehensive score threshold, sending the output high risk warning to the corresponding single sensor maintenance terminal; when it is a preset comprehensive score threshold, sending the output high risk warning to a preset central maintenance terminal.
[0011] In one implementation of this application, before inputting the multidimensional sensor array into a trained random forest model to obtain the security risk assessment result, the method further includes: collecting multidimensional sensor arrays generated within a preset historical time period to obtain state labels; constructing a training sample set based on the multidimensional sensor array and state labels; and training the random forest model using the training sample set so that the trained random forest model can predict the probability of the multidimensional sensor array falling into various state labels; wherein the state labels include at least: normal, general abnormal, and severe abnormal.
[0012] Secondly, this application provides a cableway operation safety assessment system based on multi-sensor fusion. The system includes: an acquisition module, used to deploy several types of sensors on the cableway system and acquire a multi-dimensional sensor array collected by several types of sensors according to a preset time window; an output module, used to calculate single-sensor risk score data on the multi-dimensional sensor array within the preset time window, and determine whether a preset high risk exists through preset logical rules and single-sensor risk score data; when a preset high risk exists, a high-risk warning is directly output; and an acquisition module, used to input the multi-dimensional sensor array into a trained random forest model when no preset high risk exists, to obtain the safety risk assessment result.
[0013] Thirdly, this application provides a cableway operation safety assessment device based on multi-sensor fusion. The device includes: a processor; and a memory storing executable code thereon. When the executable code is executed, the processor executes a cableway operation safety assessment method based on multi-sensor fusion as described above.
[0014] Fourthly, this application provides a non-volatile computer storage medium storing computer instructions, which, when executed, implement a cableway operation safety assessment method based on multi-sensor fusion as described above.
[0015] As can be seen from the above technical solutions, this application has the following advantages: Enhanced multi-dimensional monitoring capabilities: By deploying multiple types of sensors, such as vibration and temperature sensors, and forming a high-density monitoring network, this technical solution completely solves the problems of single sensor type and sparse arrangement in traditional cableway monitoring. Continuous acquisition of multi-dimensional sensor arrays enables real-time monitoring of the entire cableway chain, accurately capturing subtle anomalies such as wire rope wear and structural deformation, avoiding the limitations of traditional magnetic induction technology that requires system shutdown for inspection. This multi-parameter fusion monitoring method allows the system to comprehensively assess the cableway's health status from multiple dimensions such as mechanical stress and thermodynamic state, reducing safety risks caused by monitoring blind spots.
[0016] Real-time Risk Assessment and Two-Level Judgment Mechanism: The innovative technical solution adopts a two-level assessment architecture of "preset logical rules + random forest model." It achieves instantaneous high-risk early warning by calculating single-sensor risk score data in real time (response time down to milliseconds), and utilizes machine learning models to fuse and analyze complex multi-source data. This design overcomes the shortcomings of traditional threshold alarm systems, such as high false alarm rates and limited analytical dimensions, improving assessment accuracy while ensuring real-time performance. In particular, the introduction of the random forest model enables dynamic learning of cableway operation characteristics and adaptive optimization of assessment rules, giving the system continuous optimization capabilities.
[0017] Synergistic optimization of operational efficiency and safety: Through a dynamic early warning output mechanism, the system achieves risk-level response: high-risk events are immediately triggered to avoid delays in manual inspections; low-risk states are continuously evaluated through models to reduce unnecessary downtime for maintenance. Attached Figure Description
[0018] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 is a flowchart of a cableway operation safety assessment method based on multi-sensor fusion provided in an embodiment of this application.
[0020] Figure 2 is a schematic diagram of the internal structure of a cableway operation safety assessment system based on multi-sensor fusion provided in an embodiment of this application.
[0021] Figure 3 is a schematic diagram of the internal structure of a cableway operation safety assessment device based on multi-sensor fusion provided in an embodiment of this application. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0023] Those skilled in the art should understand that the embodiments described below are merely preferred embodiments of this disclosure and do not imply that this disclosure can only be implemented through these preferred embodiments. These preferred embodiments are merely used to explain the technical principles of this disclosure and are not intended to limit the scope of protection of this disclosure. Based on the preferred embodiments provided by this disclosure, all other embodiments obtained by those skilled in the art without creative effort should still fall within the scope of protection of this disclosure.
[0024] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0025] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.
[0026] The embodiment provides a method for assessing the safety of cableway operation based on multi-sensor fusion. As shown in Figure 1, the method provided by this embodiment mainly includes the following steps: Step 110: Deploy several types of sensors on the cableway system and acquire a multi-dimensional sensor array collected by several types of sensors according to a preset time window.
[0027] In some embodiments, several types of sensors are deployed on the cableway system, specifically including: wind speed sensors installed at preset locations along the line to collect ambient wind speed; three-axis inertial measurement units installed in the carriages to acquire three-axis acceleration and angular velocity and calculate the carriage's sway amplitude and attitude; rope tension sensors installed on the tensioning device to collect the tension force of the traction cable; rope sheave vibration sensors installed on key tower sheave groups or drive-side sheave brackets to collect sheave vibration signals; motor current sensors installed in the drive motor circuit to collect motor current data; and rope offset sensors arranged at preset locations on the support / pressure sheaves to monitor the offset of the traction cable relative to the sheave groove; and all sensors collect data synchronously according to a unified time reference.
[0028] To elaborate further, ultrasonic or mechanical anemometers are installed at typical wind-prone locations along the cableway route to collect real-time external wind speed data. The anemometers collect instantaneous wind speed values at a sampling frequency of 1–5 Hz and automatically add an internal timestamp. The collected data is transmitted to the data acquisition unit via RS485 or CAN bus for subsequent wind-induced risk analysis.
[0029] A three-axis IMU (Inertial Measurement Unit), including a three-axis accelerometer and a three-axis gyroscope, is fixedly installed in the center of the floor or near the center of gravity of a representative carriage. The IMU collects three-axis acceleration and angular velocity signals at a sampling frequency of 20–100 Hz. The acquisition module calculates attitude parameters such as carriage yaw angle, pitch angle, and sway amplitude in real time, adds a high-precision clock timestamp, and transmits them to the safety assessment unit.
[0030] Tension sensors or hydraulic pressure sensors are installed at the tensioning device of the traction cable to obtain the real-time tension value of the traction cable through force-pressure conversion. The tension sensor collects tension data, including parameters such as instantaneous tension and rate of change of tension, at a sampling frequency of 5–20 Hz. The data is transmitted to the acquisition unit via a wired network to identify signs of slack, abnormal load, or slippage in the traction cable.
[0031] Triaxial accelerometer vibration sensors are installed on the roller assemblies or drive-side sheave supports of key tower sections to monitor the vibration status of sheave bearings, axles, and structural components. The vibration sensors acquire vibration signals using a high-frequency sampling method of 200–1000 Hz to obtain mechanical health indicators such as RMS, peak value, kurtosis, and specific frequency band energy. The raw vibration waveforms are cached by the edge acquisition module and then packaged and uploaded to the evaluation unit.
[0032] Hall effect current sensors or isolated current transmitters are installed in the power supply circuit of the drive motor to monitor the motor's operating current in real time, reflecting the load status of the drive system. The current sensor collects drive current values at a sampling frequency of 10–50 Hz, including instantaneous current, RMS, current fluctuation amplitude, current surge, and other indicators, and correlates them with the speed data of the drive spindle encoder to identify risks such as drive jamming, abnormal load, or cable gripper slippage.
[0033] Proximity switches, photoelectric sensors, or laser rangefinders are installed on both sides of the cable-supporting or pressure-bearing reels on each tower to detect the lateral offset of the traction cable. The offset sensor outputs the offset distance or over-limit status signal in real time, and generates an alarm flag when the traction cable deviates from the center of the reel groove beyond a set threshold. The data acquisition frequency is typically 5–20 Hz, and the data is directly fed into the safety assessment unit as an important input for identifying the risk of cable derailment.
[0034] All sensor acquisition modules are synchronized using a unified time base. The data acquisition unit uses a uniform timestamp format for each data record, ensuring that data from different sensors can be aligned, segmented, and fused for analysis on the same timeline. The acquisition unit uploads six types of data—wind speed, IMU, tension, vibration, current, and offset—to the cableway server at fixed intervals, and retains cached data for the most recent 5 to 30 minutes locally for data transfer in case of transmission failure or fault tracing.
[0035] Specifically, according to a preset time window, a multi-dimensional sensor array is obtained from several types of sensors. This can be achieved by: reading all sensor data uploaded by several types of sensors once according to the preset time window; and combining the sensor data collected by each sensor at the same time point into a multi-dimensional sensor array based on the collection timestamp in the sensor data.
[0036] Step 120: Calculate single-sensor risk score data for the multi-dimensional sensor array within the preset time window. Determine whether there is a preset high risk based on preset logic rules and single-sensor risk score data. If there is a preset high risk, output a high-risk warning directly.
[0037] It should be noted that a single sensor includes at least: a wind speed sensor, a three-axis inertial measurement unit, a rope tension sensor, a rope pulley vibration sensor, a motor current sensor, and a rope offset sensor.
[0038] In this step, the risk score of a single sensor is calculated for the multi-dimensional sensor array within a preset time window. Specifically, this can be done as follows: Based on the ambient wind speed in the multi-dimensional sensor array within the preset time window, the average wind speed, maximum wind speed, and wind speed change rate are calculated; based on the scoring intervals into which the average wind speed, maximum wind speed, and wind speed change rate fall, the risk score data of the wind speed sensor is obtained; based on the carriage sway amplitude and attitude in the multi-dimensional sensor array within the preset time window, the peak yaw angle, yaw RMS, and main yaw frequency are calculated; based on the scoring intervals into which the peak yaw angle, yaw RMS, and main yaw frequency fall, the risk score data of the three-axis inertial measurement unit is obtained; based on the traction cable tension in the multi-dimensional sensor array within the preset time window, the average tension, tension deviation, and tension change rate are calculated. The risk score data for the rope tension sensor is obtained based on the scoring intervals into which the average tension, tension deviation, and tension change rate fall. The vibration RMS, peak value, kurtosis, and frequency band energy are calculated based on the rope sheave vibration signal in the multi-dimensional sensor array within a preset time window. The risk score data for the rope sheave vibration sensor is obtained based on the scoring intervals into which the vibration RMS, peak value, kurtosis, and frequency band energy fall. The current RMS, fluctuation amplitude, and surge amount are calculated based on the motor current data in the multi-dimensional sensor array within a preset time window. The risk score data for the motor current sensor is obtained based on the scoring intervals into which the current RMS, fluctuation amplitude, and surge amount fall. The deviation excess ratio (the proportion of time within 5 seconds when the deviation exceeds the threshold) is calculated based on the wheel groove deviation in the multi-dimensional sensor array within a preset time window. The risk score data for the rope deviation sensor is determined based on the scoring interval into which the deviation excess ratio falls.
[0039] It should be noted that the specific range of the scoring interval can be determined by those skilled in the art based on the actual situation.
[0040] Specifically, the existence of a preset high risk is determined by using preset logical rules and single-sensor risk scoring data. This can be achieved as follows: a preset high risk is identified when the average wind speed score in the single-sensor risk scoring data is greater than a preset wind speed threshold, the maximum wind speed score is greater than a preset maximum wind speed threshold, and the wind speed change rate score is greater than a preset wind speed change rate threshold; a preset high risk is identified when the peak yaw angle score in the single-sensor risk scoring data is greater than a preset yaw angle threshold, the yaw RMS score is greater than a preset yaw threshold, and the main yaw frequency score is greater than a preset main yaw frequency threshold; and a preset high risk is identified when the average tension score in the single-sensor risk scoring data is greater than a preset average tension threshold, the tension deviation score is greater than a preset tension deviation threshold, and the tension change rate score is greater than a preset tension change rate threshold. When the rate threshold is reached, a preset high risk is determined; when the vibration RMS score in the single sensor risk scoring data is greater than the preset vibration RMS threshold, the peak score is greater than the preset peak threshold, the kurtosis score is greater than the preset kurtosis threshold, and the frequency band energy score is greater than the preset energy threshold, a preset high risk is determined; when the current RMS score in the single sensor risk scoring data is greater than the preset current RMS threshold, the fluctuation amplitude score is greater than the preset fluctuation threshold, and the surge score is greater than the preset surge threshold, a preset high risk is determined; when the offset excess ratio score in the single sensor risk scoring data is greater than the preset offset threshold, a preset high risk is determined; the average of the average wind speed score, the maximum wind speed score, and the wind speed change rate score is taken as the wind speed risk score. The average of the peak yaw angle score, the RMS yaw score, and the main yaw frequency score is taken as the sway risk score. The average of the average tension score, tension deviation score, and tension change rate score is taken as the tension risk score. The average of the vibration RMS score, peak value, and kurtosis score is taken as the rope pulley vibration risk score. The average of the current RMS score, fluctuation range score, and surge value score is taken as the motor current risk score. The percentage of deviation exceeding the limit is taken as the rope deviation risk score. ; Through the formula: Calculate the overall score ;in, , , , , , When the overall score If the score exceeds the preset comprehensive score threshold, a preset high risk is identified.
[0041] For example, security: < Warning: <= < Preset high risk: >= .
[0042] After determining that a preset high risk exists, the method further includes: determining whether the trigger for the preset high risk is a preset comprehensive score threshold; when it is not a preset comprehensive score threshold, sending the output high risk warning to the corresponding single sensor maintenance terminal; when it is a preset comprehensive score threshold, sending the output high risk warning to the preset central maintenance terminal.
[0043] It should be noted that a monitoring network covering key parameters of cableway operation was constructed through multi-dimensional data acquisition from six types of sensors, including wind speed sensors and triaxial inertial measurement units. Each sensor calculates 12 core indicators, such as average wind speed / tension, vibration RMS, and deviation exceeding limits, and generates a quantitative risk score based on preset scoring ranges. This enables independent assessment of risk sources such as mechanical stress, environmental interference, and electrical anomalies. This multi-parameter collaborative analysis mechanism not only overcomes the limitations of traditional single-sensor monitoring but also accurately locates the specific stage of risk occurrence (such as abnormal sheave vibration or motor current fluctuations) through six types of specialized scores, including wind speed risk scores and sway risk scores, providing clear direction for subsequent maintenance.
[0044] Furthermore, a dual-layer early warning architecture of "single sensor threshold judgment + comprehensive scoring decision" is adopted: when any sensor's average wind speed, vibration peak value, or other key indicators exceed a preset threshold, a high-risk warning is immediately triggered (response time down to milliseconds); when the comprehensive score of multiple parameters exceeds the threshold, a system-level warning is initiated. This design ensures instantaneous response to extreme conditions (such as safety risks caused by sudden wind speed changes) and comprehensively assesses the overall risk status through a weighted formula, avoiding the false alarm problem of traditional single-threshold alarms. A dynamically adjustable risk assessment framework is constructed through preset scoring intervals and logical rules. The scoring intervals can be flexibly configured according to parameters such as cableway model and environmental characteristics (e.g., the wind speed threshold can be increased in high-altitude areas), enabling the system to adapt to different application scenarios. The scoring data of the six types of sensors can be analyzed independently (e.g., assessing rope tension risk separately) or fused together through a weighted formula. This modular design facilitates rapid integration when adding new sensor types (e.g., adding corrosion monitoring sensors).
[0045] Step 130: When there is no preset high risk, input the multidimensional sensor array into the trained random forest model to obtain the safety risk assessment result.
[0046] In some embodiments, before inputting the multidimensional sensor array into the trained random forest model to obtain the safety risk assessment results, the method further includes: collecting multidimensional sensor arrays generated within a preset historical time period to obtain state labels; constructing a training sample set based on the multidimensional sensor array and state labels (for example, during the cableway's historical operation phase, continuous multi-sensor data is recorded, and a multidimensional feature vector for each 5-second time window is obtained, while the following information is used to label each time window: operation logs (speed limit, emergency stop, fault records), maintenance personnel's inspection results, safety event records (wind-induced swaying, offset alarm, cable grip malfunction), and equipment alarm records (drive overload, vibration alarm, etc.). Based on historical records, samples are labeled into three categories: Normal: stable operation, no alarms; Minor Fault: slight swaying, slight tension deviation, short-term vibration increase, etc.; Major Fault: strong wind swaying, offset approaching the limit, bearing damage, significant current surge, etc.); finally forming a training sample set. , , The feature vector corresponding to the multidimensional sensor array. (For safety level labels); using the training sample set, a random forest model is trained so that the trained random forest model can predict the probability of a multidimensional sensor array falling into various state labels; where the state labels include at least: normal, general abnormal and severe abnormal.
[0047] It should be noted that the training process of random forests is as follows: Automatic execution: Bootstrap sampling: randomly sampling from all training samples to construct a subset of the dataset; Feature subset sampling: each decision tree uses only a subset of features. This randomness gives the model excellent robustness, enabling it to handle sensor noise and complex nonlinear relationships.
[0048] For each tree: randomly select a subset of features; randomly select a subset of training samples; train the decision tree using the CART (Classification and Regression Tree) algorithm; use information gain or Gini coefficient as the splitting criterion. The number of trees is typically 50–200 (for optimal engineering performance).
[0049] Random forest outputs the probability that each sample belongs to one of the three states: , , The category is determined by majority vote or the highest probability. To facilitate subsequent fusion, this invention defines a machine learning risk index: In other words, the higher the probability that the model considers a "serious anomaly", the higher the risk.
[0050] After training, the following methods are used to verify model performance: K-fold cross-validation (typically K=5), confusion matrix, F1 score, recall (especially critical in the security domain), and ROC-AUC (discriminative power). If the model has a low recognition rate for a certain type of anomaly, adjustments can be made to: the number of decision trees, the maximum depth of each tree, class weights (giving higher weights to anomalous samples), and feature engineering (removing noisy features or adding combined features). Finally, a stable and reliable random forest security assessment model is obtained.
[0051] After training is complete, the process can be configured as follows: Every 5 seconds, the latest multi-sensor feature vectors are automatically generated. These feature vectors are then input into the trained random forest model to obtain the outputs Pnormal, Pminor, and Pmajor. The probability of severe anomaly is directly taken as the risk index for subsequent fusion decisions. The output state is determined based on the maximum probability: if Pnormal is the highest → Normal; if Pminor is the highest → Moderate anomaly; if Pmajor is the highest → Severe anomaly.
[0052] In addition, to facilitate model processing, all feature multidimensional sensor arrays are normalized using z-score mapping to a uniform dimension range (0 to 1 or standard normal distribution) and combined in a fixed order to form a multidimensional feature vector, which will be used as the input to the random forest model.
[0053] In addition, Figure 2 of this application illustrates a cableway operation safety assessment system based on multi-sensor fusion provided in an embodiment of this application. As shown in Figure 2, the system provided in this embodiment mainly includes: an acquisition module 210, used to deploy several types of sensors on the cableway system and acquire a multi-dimensional sensor array collected by several types of sensors according to a preset time window; an output module 220, used to calculate single-sensor risk score data for the multi-dimensional sensor array within the preset time window, and determine whether there is a preset high risk through preset logical rules and single-sensor risk score data; when there is a preset high risk, a high-risk warning is directly output; and an acquisition module 230, used to input the multi-dimensional sensor array into a trained random forest model when there is no preset high risk to obtain a safety risk assessment result.
[0054] The above are method embodiments of this application. Based on the same inventive concept, this application also provides a cableway operation safety assessment device based on multi-sensor fusion. As shown in FIG3, the device includes: a processor; and a memory storing executable code thereon. When the executable code is executed, the processor performs a cableway operation safety assessment method based on multi-sensor fusion as described in the above embodiments.
[0055] Specifically, the server deploys several types of sensors on the cableway system and acquires a multi-dimensional sensor array collected by these sensors according to a preset time window. It then calculates single-sensor risk score data for the multi-dimensional sensor array within the preset time window and determines whether a preset high risk exists based on preset logical rules and the single-sensor risk score data. When a preset high risk exists, a high-risk warning is directly output. When no preset high risk exists, the multi-dimensional sensor array is input into a trained random forest model to obtain a safety risk assessment result.
[0056] In addition, this application embodiment also provides a non-volatile computer storage medium storing executable instructions, which, when executed, implement the cableway operation safety assessment method based on multi-sensor fusion as described above.
[0057] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for assessing the safety of cableway operation based on multi-sensor fusion, characterized in that, The method includes: deploying several types of sensors on the cableway system; acquiring a multi-dimensional sensor array collected by the sensors according to a preset time window; calculating single-sensor risk score data for the multi-dimensional sensor array within the preset time window; determining whether a preset high risk exists based on preset logical rules and the single-sensor risk score data; directly outputting a high-risk warning when a preset high risk exists; and inputting the multi-dimensional sensor array into a trained random forest model when no preset high risk exists to obtain a safety risk assessment result.
2. The cableway operation safety assessment method based on multi-sensor fusion according to claim 1, characterized in that, Several types of sensors are deployed on the cableway system, including: wind speed sensors installed at preset locations along the line to collect ambient wind speed; three-axis inertial measurement units installed in the carriages to acquire three-axis acceleration and angular velocity and calculate the carriage's sway amplitude and attitude; rope tension sensors installed on the tensioning device to collect the tension force of the traction cable; rope sheave vibration sensors installed on key tower sheave groups or drive-side sheave brackets to collect sheave vibration signals; motor current sensors installed in the drive motor circuit to collect motor current data; and rope offset sensors arranged at preset locations on the support / pressure sheaves to monitor the offset of the traction cable relative to the sheave groove. All sensors collect data synchronously according to a unified time reference.
3. The cableway operation safety assessment method based on multi-sensor fusion according to claim 1, characterized in that, Based on a preset time window, a multidimensional sensor array is obtained from several types of sensors. Specifically, this includes: reading all sensor data uploaded by several types of sensors at once according to the preset time window; and combining the sensor data collected by each sensor at the same time point into a multidimensional sensor array based on the collection timestamp in the sensor data.
4. The cableway operation safety assessment method based on multi-sensor fusion according to claim 1, characterized in that, A single sensor includes at least: a wind speed sensor, a three-axis inertial measurement unit (IMU), a rope tension sensor, a rope pulley vibration sensor, a motor current sensor, and a rope offset sensor. The risk score for each single sensor is calculated within a preset time window using a multi-dimensional sensor array. Specifically, this includes: calculating the average wind speed, maximum wind speed, and wind speed change rate based on the ambient wind speed in the multi-dimensional sensor array within the preset time window; obtaining the risk score data for the wind speed sensor based on the scoring intervals into which the average wind speed, maximum wind speed, and wind speed change rate fall; calculating the peak yaw angle, yaw RMS, and main yaw frequency based on the carriage sway amplitude and attitude in the multi-dimensional sensor array within the preset time window; obtaining the risk score data for the three-axis IMU based on the scoring intervals into which the peak yaw angle, yaw RMS, and main yaw frequency fall; and calculating the average tension, average yaw angle, and main yaw frequency based on the traction cable tension in the multi-dimensional sensor array within the preset time window. The risk score data for the rope tension sensor is obtained based on the scoring intervals into which the average tension, tension deviation, and tension change rate fall. The risk score data for the rope sheave vibration sensor is obtained based on the vibration signals from the multi-dimensional sensor array within a preset time window, including the RMS, peak value, kurtosis, and frequency band energy. The risk score data for the rope sheave vibration sensor is also obtained based on the scoring intervals into which these parameters fall. Furthermore, the risk score data for the motor current sensor is obtained based on the motor current data from the multi-dimensional sensor array within a preset time window, including the RMS, fluctuation amplitude, and surge amount. Finally, the risk score data for the rope offset sensor is determined based on the wheel groove offset from the multi-dimensional sensor array within a preset time window and the scoring interval into which the offset offset falls.
5. The cableway operation safety assessment method based on multi-sensor fusion according to claim 1, characterized in that, Based on preset logical rules and single-sensor risk scoring data, the existence of preset high risks is determined. Specifically, this includes: determining the existence of a preset high risk when the average wind speed score, maximum wind speed score, and wind speed change rate score in the single-sensor risk scoring data are all greater than preset wind speed thresholds; determining the existence of a preset high risk when the peak yaw angle score, yaw RMS score, and main yaw frequency score in the single-sensor risk scoring data are all greater than preset yaw angle thresholds; and determining the existence of a preset high risk when the average tension score, tension deviation score, and tension change rate score in the single-sensor risk scoring data are all greater than preset tension change rate thresholds. When the value is greater than a preset high risk threshold, the following conditions are met: The vibration RMS score, peak value, kurtosis score, and frequency band energy score in the single-sensor risk scoring data are all greater than preset high risks. Similarly, the current RMS score, fluctuation amplitude score, and surge score in the single-sensor risk scoring data are all greater than preset high risks. The offset exceeding the limit score in the single-sensor risk scoring data is also greater than a preset offset threshold. The average of the average wind speed score, maximum wind speed score, and wind speed change rate score is taken as the wind speed risk score. The average of the peak yaw angle score, the RMS yaw score, and the main yaw frequency score is taken as the sway risk score. The average of the average tension score, tension deviation score, and tension change rate score is taken as the tension risk score. The average of the vibration RMS score, peak value, and kurtosis score is taken as the rope pulley vibration risk score. The average of the current RMS score, fluctuation range score, and surge value score is taken as the motor current risk score. The percentage of deviation exceeding the limit is taken as the rope deviation risk score. ; By formula: Calculate the overall score ;in, 、 、 、 、 、 When the overall score If the score exceeds the preset comprehensive score threshold, a preset high risk is identified.
6. The cableway operation safety assessment method based on multi-sensor fusion according to claim 5, characterized in that, After determining that a preset high risk exists, the method further includes: determining whether the trigger for the preset high risk is a preset comprehensive score threshold; when it is not a preset comprehensive score threshold, sending the output high risk warning to the corresponding single sensor maintenance terminal; when it is a preset comprehensive score threshold, sending the output high risk warning to a preset central maintenance terminal.
7. The cableway operation safety assessment method based on multi-sensor fusion according to claim 1, characterized in that, Before inputting the multidimensional sensor array into the trained random forest model to obtain the security risk assessment result, the method further includes: collecting multidimensional sensor arrays generated within a preset historical time period to obtain state labels; constructing a training sample set based on the multidimensional sensor array and state labels; and training the random forest model using the training sample set so that the trained random forest model can predict the probability of the multidimensional sensor array falling into various state labels; wherein, the state labels include at least: normal, general abnormal, and severe abnormal.
8. A cableway operation safety assessment system based on multi-sensor fusion, characterized in that, The system includes: an acquisition module, used to deploy several types of sensors on the cableway system and acquire a multi-dimensional sensor array collected by the several types of sensors according to a preset time window; an output module, used to calculate single-sensor risk score data for the multi-dimensional sensor array within the preset time window, and determine whether there is a preset high risk based on preset logical rules and single-sensor risk score data; when there is a preset high risk, a high-risk warning is directly output; and an acquisition module, used to input the multi-dimensional sensor array into a trained random forest model when there is no preset high risk to obtain a safety risk assessment result.
9. A cableway operation safety assessment device based on multi-sensor fusion, characterized in that, The device includes: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform a cableway operation safety assessment method based on multi-sensor fusion as described in any one of claims 1-7.
10. A non-volatile computer storage medium, characterized in that, It stores computer instructions, which, when executed, implement a cableway operation safety assessment method based on multi-sensor fusion as described in any one of claims 1-7.