Cableway carriage operation deflection attitude monitoring device and control method

Abstract

The invention relates to the field of cableway transportation safety control, in particular to a cableway lift carriage operation deflection attitude monitoring device and a control method, and the method comprises the steps: collecting the position, speed, attitude angle and local micro deflection disturbance information of a lift carriage in real time by constructing a multi-mode sensing network; the frequency of the sensor is dynamically adjusted by adopting an adaptive sampling mechanism, and space-time alignment and fusion of heterogeneous data are completed in combination with dynamic time warping and minimum mean square estimation; strengthening the key disturbance dimension based on a disturbance-driven attention mechanism, and predicting a future deflection angle sequence; a decoupling torque is generated through spatial deflection vector field modeling, regional coupling tensor analysis and a dynamic direction alignment matrix, and power consumption constraint and a dynamic environment parameter optimization control instruction are fused; and finally, driving the quadrilateral moment gyroscope to execute correction, and constructing a multi-dimensional evaluation index based on the attitude response residual error. According to the method, adaptive closed-loop adjustment of the control strategy is realized, and the robustness under wind speed disturbance and load variation is improved.

Description

Technical Field

[0001] This invention relates to the field of cableway transportation safety control technology, specifically to a cableway gondola swaying attitude monitoring device and control method. Background Technology

[0002] Existing cableway gondolas are susceptible to strong wind disturbances, load imbalances, and cableway vibrations during operation, which can cause the gondolas to sway beyond the limit. Traditional solutions rely on fixed threshold alarms and emergency braking mechanisms, which suffer from passive response delays and a lack of predictive correction capabilities.

[0003] Chinese invention patent application CN120194686A discloses a three-dimensional attitude monitoring method for a suspended traction chair based on multiple intelligent sensors, including: acquiring and processing raw sensor data to obtain sensor data; calculating attitude data using the sensor data; generating attitude data prediction values ​​by constructing and training a multivariate autoregressive integral moving average model; calculating dynamic attitude data using the sensor data; fusing the attitude data prediction values ​​and the data in the dynamic attitude data to obtain a comprehensive pitch angle, a comprehensive roll angle, and a comprehensive yaw angle; further, obtaining a comprehensive attitude value; and determining whether to trigger an early warning mechanism based on the comprehensive attitude value, initiating an automatic adjustment operation or initiating an emergency shutdown procedure.

[0004] As cableway systems evolve towards higher speeds, larger capacities, and more complex environments, precise control of gondola operation has become a key technological requirement for ensuring safety and comfort. Currently, the field of cableway control is undergoing a technological evolution from passive response to active prediction, with multimodal sensor fusion, nonlinear state prediction, and dynamic decoupling optimization becoming research hotspots. The swaying problem of gondolas under strong wind disturbances, sudden load changes, and cableway vibrations urgently requires a systematic solution that integrates high-precision spatiotemporal perception, advanced risk prediction, and adaptive closed-loop correction to meet the engineering requirements of modern cableway systems for real-time performance, energy efficiency, and long-term operational robustness. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background technology by proposing a monitoring device and control method for the swaying attitude of a cable car.

[0006] The technical solution of the present invention: a control method for a cableway gondola yaw attitude monitoring device, comprising the following specific implementation steps: S1. Collect information on the position, velocity, attitude angle and local micro-sway disturbance of the gondola through a multi-source sensing network, adaptively adjust the sampling frequency, normalize the data and then use dynamic time warping and least mean square estimation to fuse the asynchronous data, and output a fused state vector sequence. S2. Define sensor channel perturbation energy index through local perturbation tensor analysis, screen the set of significant interference dominant factors, and construct a perturbation attention weighting mechanism to generate enhanced state vectors by weighting and strengthening key perturbation dimensions; input the D-GRU network to predict future yaw angle sequences, and trigger an early warning and correction mechanism when the absolute value of the maximum yaw angle of the predicted sequence exceeds the safety threshold. S3. Construct a spatial vector field based on the predicted yaw sequence, analyze the torque interference between corners through regional coupling tensor analysis, combine dynamic direction alignment matrix projection decoupling, and optimize the torque command vector by integrating power consumption constraints and dynamic environmental parameters. It is also encapsulated as a standard control frame to synchronously drive the four corner torque gyroscopes to perform corrections. S4. Construct a comprehensive scoring function based on the attitude residual sequence to classify the control risk level, and combine the abnormal event graph attribution mechanism to perform short-period parameter fine-tuning and long-period model evolution update.

[0007] Preferably, the multi-source sensing network includes: GPS / RTK unit: provides the gondola's position P(t), velocity V(t), and heading angle θ(t); High-precision six-axis IMU unit: provides pitch angle α(t), roll angle β(t), yaw angle γ(t), and angular velocity; Distributed neuron pose array: deployed at the four corners of the gondola, providing local micro-sway perturbation information δ i (t) represents the local perturbation sway angle detected by the i-th neuron attitude array sensing node at time t.

[0008] Preferably, the process for generating the enhanced state vector is as follows: By calculating the average rate of change of disturbance for each sensor channel within the disturbance observation window as a disturbance energy index, channels exceeding a set threshold are selected to be included in the current set of dominant disturbance factors, and significant sway driving trends are identified. The attention weights of each channel are calculated based on the set of interference-dominant factors. An exponential amplification mechanism is used to strengthen the key perturbation dimensions, and the state vector is weighted and reconstructed to generate a perturbation-enhanced state vector.

[0009] Preferably, the prediction process of the D-GRU network for predicting future yaw angle sequences is as follows: ; in, The perturbation-gated variable represents the time t. This represents the hidden state at the previous time step; Indicates the candidate hidden state; and These represent the weight matrices corresponding to the current input and the previous state in the perturbation gating, respectively; and Let W and U represent the input and state weight matrices of the reset gate, respectively; W and U represent the input and weight matrices of the candidate state update function, respectively. This indicates element-wise multiplication; This represents the perturbation-enhanced state vector after perturbation attention weighting; Indicates the current hidden state; Output the predicted sequence of the current state. : ; ; in, This represents the prediction sequence consisting of all predicted yaw angles from time 1 to time T in the future; This represents the predicted yaw angle value at the k-th time step in the future; This represents the D-GRU hidden state at prediction time t+k, i.e., the intrinsic memory of the system state; This represents the weight matrix of the output mapping layer, used to map the hidden states. The value is mapped to a specific angle prediction.

[0010] Preferably, the optimization process for the torque command vector is as follows: Based on the predicted yaw sequence, a spatial vector field is constructed to map the trend torque at the four corners. By combining the coupling coefficient with the position vector, moment of inertia and directional angle, a regional coupled torque tensor is constructed to analyze the interference between corners. Based on the regional torque distribution and coupling tensor, the yaw trend gradient is calculated to obtain the main perturbation direction. A dynamic direction alignment matrix is ​​constructed to map the yaw direction to the gyroscope installation direction. The decoupling torque control vector is obtained, and the decoupling calculation is completed. By integrating physical boundaries, energy constraints, and historical inertial response characteristics through a dynamic torque optimization function, and combining dynamic parameters and maximum power limits, the decoupled torque command is optimized to generate a stable and executable modified torque vector.

[0011] Preferably, the process for constructing the region-coupled moment tensor is as follows: Based on the predicted yaw angle sequence, it is transformed into a spatial yaw vector field through a nonlinear mapping function, which maps the yaw trend torque vectors at the four corners of the bottom of the gondola to form a vector array; The four corners of the bottom of the gondola include: front left FL, front right FR, rear left BL, and rear right BR; Constructing the region coupling tensor: ; in, This represents the coupling moment tensor term between the i-th and j-th angles of the gondola; This represents the coupling coefficient between the i-th and j-th control angles in the gondola structure; and These represent the yaw tendency torque vectors at the i-th and j-th control angles, respectively; and These represent the position vectors of control angles i and j relative to the center of mass of the gondola; I i and I j These represent the local moments of inertia about the center of mass; This represents the angle between the principal directions of the sway at points i and j.

[0012] Preferably, the dynamic torque optimization function is: ; ; Where Tv represents the output torque vector of the four torque gyroscopes to be optimized at the current moment; This represents the system power consumption estimation function required for the current output torque; This indicates the set weighting factor; This represents the adaptive control penalty factor for the i-th control point; Indicates the basic penalty item; This represents the weighting coefficient corresponding to each dynamic indicator; This indicates the set wind speed reference value; Indicates the upper limit of the reference value for the angular velocity of the gondola; This indicates the acceptable upper limit of vertical vibration; This represents the maximum tolerable safety delay threshold of the control system; Q(t) represents the theoretically optimal control torque vector after decoupling. This represents the historical output sequence of torque. This represents the current set of dynamic parameters for the gondola's operation, including wind speed. angular velocity of the hoisting car Vertical vibration acceleration of the hoisting car Control delay ; This indicates the maximum output power of the torque gyroscope; This represents the dynamic torque optimization function.

[0013] Preferably, the comprehensive scoring function consists of three indicators: stability, overshoot, and response delay, and its form is as follows: ; ; ; ; in, This indicates a correction to the stability index; This represents the weighted residual contribution value at time point t+k; This indicates the correction of the overshoot indicator; This indicates the maximum acceptable safe sway angle threshold for the gondola's operation; This indicates the response latency metric; This represents the tolerance threshold for the yaw angle residual. This represents the overall score of the control scoring function, i.e., the overall scoring function; , and This represents the weighting coefficient.

[0014] Preferably, the disturbance energy index is calculated using a sliding observation window method, with the window length being a dynamically adjustable parameter that is adaptively increased or decreased according to the disturbance intensity.

[0015] The technical solution of the present invention: a cable car yaw attitude monitoring device, which is applicable to the above-mentioned cable car yaw attitude monitoring device control method, comprising: The integrated operating attitude perception and preprocessing module is used to synchronously collect multi-dimensional status information of the gondola during operation and perform data preprocessing operations. The integrated operating attitude perception and preprocessing module includes a GPS position and motion perception unit, a three-axis gyroscope unit, and a main control processing unit (MCU). The main control processing unit (MCU) integrates data-driven attitude recognition and yaw prediction algorithms, and executes the entire process from data preprocessing, state modeling, trend prediction to correction strategy generation. GPS location and motion sensing unit is used to collect the location and motion information of the gondola in real time during operation; The three-axis gyroscope unit collects the angular velocity and attitude changes of the gondola in three degrees of freedom, capturing micro yaw trends and spatial disturbance responses. The disturbance-driven yaw trend prediction and modeling module constructs a future yaw angle prediction sequence in the time domain and establishes a disturbance-gated recursive unit network (D-GRU) for gondola attitude changes to predict the future yaw trend angle sequence. The multi-axis coupled torque decoupling and correction strategy generation module is used to construct a yaw vector field based on the predicted sequence, perform coupled tensor modeling, direction alignment matrix generation and dynamic power consumption optimization, and generate control torque commands for the gyroscope to execute corrections. The control closed-loop performance evaluation and adaptive strategy adjustment module is used to score the correction effect, attribute anomalies, and update strategies based on attitude residuals, thereby completing the system's self-evolutionary regulation.

[0016] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: This invention designs a cableway gondola yaw attitude monitoring device and control method. By constructing a multimodal sensor network (GPS / RTK unit, high-precision IMU unit, distributed neuron attitude array) to collect gondola position, speed, attitude angle and local micro-yaw disturbance information in real time, and combining an adaptive sampling mechanism (frequency mapping scheduling) and a cross-modal synchronous transformation function (DTW dynamic time warping + least mean square estimation) to achieve spatiotemporal alignment and high-precision fusion of heterogeneous data, eliminate multi-source asynchronous errors, and lay a reliable control data foundation; Based on the perturbation attention mechanism, the dominant perturbation dimension is selected and the enhanced state vector is reconstructed. The input is a self-developed perturbation-gated recursive unit (D-GRU) network to predict future sway sequences. The newly added perturbation gating term dynamically regulates the hidden state update, which significantly improves the accuracy of nonlinear trend prediction and realizes advanced risk warning. Based on the prediction results, the coupling torque interference weight is calculated by modeling the spatial yaw vector field and performing regional coupling tensor analysis. Combined with the dynamic direction alignment matrix, torque decoupling and spatial projection optimization are achieved, generating feasible control commands that take into account physical constraints (power limit, dynamic delay, mechanical inertia) to drive the four corner torque gyroscopes for collaborative correction. The closed-loop control system constructs a multi-dimensional scoring mechanism (stability, overshoot, and response delay) based on real-time residual sequences. Combined with abnormal event graph attribution and dynamic strategy adjustment (short-cycle parameter fine-tuning and long-cycle model evolution), it significantly improves the system's robustness in dealing with wind speed disturbances, load changes, and long-term operation. Ultimately, it achieves comprehensive benefits such as rapid and stable convergence of the gondola's attitude, reduced emergency braking rate, suppression of ineffective power consumption (dynamic power constraint optimization to reduce gyroscope losses), and mitigation of mechanical fatigue. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for controlling the swaying attitude monitoring device of a cableway gondola, as proposed in this invention. Detailed Implementation

[0018] Example 1, as Figure 1 As shown, the present invention proposes a control method for a cableway gondola yaw attitude monitoring device, which includes the following specific implementation steps: S1. By constructing a multimodal sensor network, an adaptive sampling mechanism, and a cross-modal synchronous transformation function, spatial fusion and temporal alignment of heterogeneous data are achieved, laying a high-quality data foundation for subsequent state evolution prediction and yaw control. The specific implementation process is as follows: S11. Construct a multi-source sensing network based on an edge computing architecture, where sensors include, but are not limited to: GPS / RTK unit: provides the gondola's position P(t), velocity V(t), and heading angle θ(t); High-precision six-axis IMU unit: provides attitude angles (three-axis angles: pitch angle α(t), roll angle β(t), and yaw angle γ(t) and angular velocity); Distributed neuron pose array: deployed at the four corners of the gondola, providing local micro-sway perturbation information δ i (t), which is the local perturbation sway angle detected by the i-th neuron attitude array sensing node at time t, reflecting the instantaneous attitude fluctuation of a certain corner of the gondola; S12. Set up a frequency mapping scheduling mechanism to adaptively adjust the sampling period according to the disturbance change rate, so that different modal data can form a synchronization cluster under event-driven conditions: ; in, This represents the adaptive sampling frequency of the i-th type of sensor at time t; Indicates the preset base sampling frequency; This represents the disturbance measured by the i-th type of sensor at time t; The value represents the rate of change of the disturbance; k represents the disturbance sensitivity coefficient (empirical setting). S13. Normalize the data from each sensor to obtain normalized data, and then construct the state fusion vector S(t): ; in, This represents the normalized vector of the position triple P(t); and These represent the normalized results for velocity and heading angle, respectively. This represents the normalized value of the yaw angle perturbation at the i-th attitude node; S14. An asynchronous interpolation fusion mechanism based on Dynamic Time Warping (DTW) and least mean square estimation is introduced to interpolate the optimal state estimate under a unified time index. Specifically: Establish the target timeline This timeline serves as a unified time base for state modeling; All data is synchronized using a combination of DTW (Dynamic Time Warping) and linear interpolation. ; Interpolation also introduces cross-modal similarity constraints: ; in, This represents the estimated value of the i-th dimension data calculated by interpolation at the target alignment time t; and These represent the data of dimension i at time point t. k+1 and t k The original value on; t k and t k+1These represent the previous and next sampling time points closest to the target time t (which are less than and greater than t, respectively); The temporal alignment error cost function is used to measure the interpolation consistency of all modes at the target time. The smaller the value, the more successful the interpolation alignment between different data streams. This represents the similarity weight between the i-th and j-th modes at time t, i.e., the weight of their physical, logical, or functional coupling relationship (e.g., position and velocity influence each other, so the weight is large; yaw angle and height have no strong correlation, so the weight is small). S15, Output the fused state vector sequence {S(t)} i )}.

[0019] S2. The high-dimensional temporal state vector is transformed into the yaw attitude evolution trajectory in the future time domain, and the nonlinear evolution trend of the gondola attitude is modeled and predicted. The specific implementation process is as follows: S21. Construct a perturbation energy spectrum model based on local perturbation tensor analysis, and apply this model to the input fused state vector sequence. Perform a disturbance significance analysis to determine whether there is a significant sway driving trend in the current system state, specifically: Define the perturbation energy index: ; in, This represents the disturbance energy of the i-th sensor channel; This represents the yaw angle data of the i-th node; Indicates the length of the disturbance observation window; If a certain dimension Exceeding the threshold Then the channel is included in the current set of dominant interference factors D. t ; S22. Based on the above-selected set of dominant disturbance channels D t We construct a perturbation-driven attention distribution mechanism to weight and strengthen key perturbation dimensions in high-dimensional state sequences; The following perturbation attention weight function is used: ; in, This represents the perturbation attention weight of the i-th perturbation-dominant channel at time t. The larger the value, the greater the influence of this channel on the overall sway trend. This represents the perturbation energy amplification factor (perturbation sensitivity factor), which controls the degree of influence of perturbation energy on attention distribution. The value is generally adjusted between 2 and 5. This represents the natural exponential function, used to map perturbation energy values ​​to nonlinear attention values; D represents the current set of dominant interference factors. t The disturbance energy of the j-th sensor channel; At each time t, a perturbation-enhanced state vector is generated by perturbation-attention-weighted reconstruction of the state vector S(t). : ; in, This represents the perturbation-enhanced state vector after perturbation attention weighting; The normalized state vector component of the i-th channel at time t is the sub-dimensional vector extracted from S(t); S23. The weighted state sequence The input is fed into a disturbance-gated recurrent unit (D-GRU) network for nonlinear state modeling. The D-GRU adds a disturbance gating term to the traditional GRU. Used to dynamically control the magnitude and direction of memory state updates: ; in, The disturbance gating variable at time t determines whether (and how) the current state information is updated due to the disturbance, and its range is [0,1]. The hidden state (memory vector) at the current moment is the main state representation inside the D-GRU unit, carrying historical and current attitude change information; It represents the hidden state of the previous time step, and is used to control whether the current state continues the previous trend; This represents the candidate hidden state, which is the intermediate result of the current input state update (with perturbation control), used for fusion with the previous state; and These represent the weight matrices corresponding to the current input and the previous state in the perturbation gating, respectively; and Let W and U represent the input and state weight matrices of the reset gate, respectively; W and U represent the input and weight matrices of the candidate state update function, respectively. This indicates element-wise multiplication; S24. Finally, output the predicted sequence of the current state. : ; ; in, This represents a prediction sequence consisting of all predicted yaw angles over a future period (times 1 to T). This represents the predicted yaw angle value at the k-th time step in the future; This represents the D-GRU hidden state at prediction time t+k, i.e., the intrinsic memory of the system state; This represents the weight matrix of the output mapping layer, used to map the hidden states. Mapped to a specific angle prediction value; S25. After predicting the output, for the entire To determine the trend: like If this occurs, the early yaw correction mechanism will be triggered, and an alarm signal will be generated. in, This indicates the safety sway threshold (set as the upper limit angle of the safe sway of the gondola). Exceeding this value is considered a potentially dangerous situation. This represents the absolute value of the maximum sway angle in the predicted sequence, used to determine whether early correction is needed.

[0020] S3. To address the predicted future yaw trend of the gondola, a dynamic adaptive spatial correction strategy generation mechanism is constructed, combining the gondola's spatial structural characteristics and torque interference laws. This mechanism not only considers the interference and coordination relationships between multi-point couplings, but also generates real-time executable control commands through decoupling modeling, orientation registration, and power optimization to drive four high-precision torque gyroscope modules to work collaboratively. The specific implementation process is as follows: S31, Based on Future Deflection Angle Sequence The overall yaw trend is transformed into a spatial yaw vector field using a set of nonlinear mapping functions: ; in, These represent the sway trend vectors (derived from the sway prediction values) experienced by the four bottom corners of the gondola (front left FL, front right FR, rear left BL, rear right BR) at the current moment. This represents a vector array consisting of the four corner yaw vectors; Constructing the region coupling tensor: ; in, This represents the coupling moment tensor term between the i-th and j-th angles of the gondola; This represents the coupling coefficient between the i-th and j-th control angles in the gondola structure, used to measure the disturbance transmission strength between the two angles; and These represent the yaw tendency torque vectors at the i-th and j-th control angles (e.g., front left and rear right); and These represent the position vectors of control angles i and j relative to the center of mass of the gondola; I i and I j These represent the local rotational inertia about the center of mass, reflecting the sensitivity of this corner point to overall attitude disturbances. This represents the angle between the principal directions of the sway at points i and j; S32. Based on the obtained regional moment distribution and coupling tensor, a dynamic direction alignment matrix is ​​constructed. The spatial projection direction of the control command is automatically adjusted according to the current dominant yaw direction. A dynamic projection mechanism between the dominant disturbance direction and the spatial control normal is adopted to achieve highly robust decoupling calculations. Specifically: Calculate the principal direction of the perturbation based on the gradient of the yaw trend function: ; Construct an orientation alignment matrix A(t) to describe the spatial mapping relationship between the yaw direction and the gyroscope mounting direction: ; Based on this, the decoupling torque control vector is obtained: ; in, The vector representing the dominant direction of the yaw trend; Indicates skewing prediction sequence The gradient; A(t) represents the unit vector of direction of the gyroscopes installed at the four corners of the gondola; A(t) represents the dynamic direction alignment matrix; Q(t) represents the theoretically optimal control torque vector after decoupling. S33. Combining physical boundaries, energy constraints, and historical inertial response characteristics, the decoupled torque command is dynamically optimized to ensure that the output command is both executable and meets system stability requirements. Specifically, a control optimization function is introduced: ; in, This represents the optimized true torque command vector, which is the physical torque value that will ultimately be used to drive the correction actions of each gyroscope. This represents the historical output sequence of torque, used to smooth the difference between the current output and the historical output; This represents the current set of dynamic parameters for the gondola's operation, including but not limited to wind speed. angular velocity of the hoisting car Vertical vibration acceleration of the hoisting car Control delay This is used to adjust the strength of the control strategy and the response delay; This indicates the maximum output power of the torque gyroscope, used to constrain the maximum amplitude of the current control torque and prevent damage to the device due to excessive power. Represents the dynamic torque optimization function; Specifically, the projection optimization function used in this embodiment Taking into account the consistency between the current correction target and the theoretical expectation, the inertia of maintaining the historical correction magnitude, the soft constraint penalty on the power consumption upper limit, and the adjustment penalty term consistent with the current dynamic environment, a physically feasible and dynamically stable correction moment vector is automatically generated by minimizing the overall cost function projection. ; ; Where Tv represents the output torque vector of the four torque gyroscopes to be optimized at the current moment; This represents the system power consumption estimation function required for the current output torque; This indicates the set weighting factor; This represents the adaptive control penalty factor for the i-th control point (torque gyroscope); This represents the basic penalty term, used to maintain the minimum suppression under low dynamic disturbance conditions and prevent zero penalty from leading to unstable corrections. This represents the weighting coefficient corresponding to each dynamic indicator; This indicates the set wind speed reference value; Indicates the upper limit of the reference value for the angular velocity of the gondola; This indicates the acceptable upper limit of vertical vibration; This indicates the maximum safe delay threshold that the control system can tolerate; S34. After obtaining the practically feasible correction torque, this step is responsible for converting it into standardized control commands and issuing them synchronously to the four designated torque gyroscopes, and then entering closed-loop correction control. The operation process includes: use Determine the direction code; Constructing encapsulated control instruction frames ; The data is transmitted concurrently to four gyroscope control chips, and the real-time monitoring feedback (attitude response and angular velocity residual) is used as the feedback input for the next cycle. in, The direction code (encoding indicates the direction of action, such as clockwise / counterclockwise or axial number) represents the direction executed by the i-th gyroscope and is used to precisely control the direction of rotation of the gyroscope. This indicates the duration of the torque action, which is determined by the persistence of the yaw trend and the dynamic response window setting, affecting the duration and intensity of the corrective action. Represents a standardized control command frame; This indicates an instruction encapsulation function that combines control parameters into a communication frame format recognizable by the device, facilitating distribution by the MCU and identification by the communication module.

[0021] S4. Construct a dynamic adaptive closed-loop control system that integrates real-time response evaluation, feedback anomaly identification, and control strategy reconfiguration to continuously improve the response efficiency and robustness of the gondola yaw correction system, ensuring that the gondola still has stable correction capabilities after wind speed disturbances, load changes, or long-term operation. The specific implementation process is as follows: S41. Based on the actual attitude change results of the gondola after the torque correction command is issued, construct the residual sequence between control execution and the target, specifically: Real-time acquisition of the actual attitude state of the gondola after the application of torque, and construction of a feedback state sequence: ; Predicted sway trend sequence To make a comparison, define the yaw correction residual vector: ; in, This represents the actual observed sequence of yaw angles of the gondola, i.e., the true yaw response state after the torque correction is executed; This represents the yaw correction residual vector, used to evaluate the attitude correction error between the prediction and the actual situation; This represents the time decay weighting coefficient; T represents the control evaluation window length, i.e., the length of the future time series to be evaluated. S42. After obtaining the complete residual vector, construct a multi-dimensional evaluation index to score the effectiveness of the current control strategy and identify the risk level of potential control failure or overcompensation. This scoring mechanism uses a set of custom multi-dimensional evaluation index functions: Corrected stability index (correcting whether it quickly tends to stabilize): ; Adjust the overshoot indicator (is there an overcorrection trend?): ; Response delay metric (actual correction start lag time): ; Based on the above indicators, a comprehensive scoring function is constructed: ; in, This indicates the corrected stability index, which is the relative accuracy of the average residual. It is used to assess whether the corrected state of the gondola smoothly approaches the target state. This represents the weighted residual contribution value at time point t+k, which measures the degree of influence of the yaw error on the overall correction effect at that moment. This indicates the overshoot correction index, which is the ratio between the maximum value of the actual attitude deviation after correction and the safety limit, reflecting whether overcorrection has occurred. This indicates the maximum acceptable safe sway angle threshold for the gondola's operation; This represents the response delay metric, which is the number of time steps required for the actual yaw angle to first approach the predicted angle. The smaller the value, the faster the response. This represents the tolerance threshold for the yaw angle residual. This represents the overall score of the control scoring function, which is used to evaluate the overall performance of the current control strategy by combining three weighted indicators. , and Indicates the weighting coefficient; Therefore: Risk levels are determined based on the comprehensive score from the control scoring function. Grade A (Excellent Control) > The corrections are precise, the response is rapid, and there is no significant overshoot; the current strategy remains unchanged without adjustment. Grade B (suboptimal control), There may be slight response delay or yaw rebound; triggering slight adjustments to strategy parameters, such as the magnitude or duration of control torque. Grade C (imbalance or overcompensation), ≤ This manifests as correction failure, severe overcompensation, or a continuous increase in yaw; triggering the adaptive control strategy update process and reconstructing the torque decoupling and execution strategy. in, and The set scoring threshold is dynamically adjusted according to different cableway systems; S43. Based on the residual distribution and policy execution history, attribute abnormal correction behaviors, identify the sources of yaw anomalies, determine whether the control policy needs to be adjusted, and construct a correction anomaly event map G. e : ; Among them, G e This represents the anomaly response event graph; V represents the set of vertices in the graph, containing key variables related to the anomaly events, including residuals. Torque command and the direction of disturbance ; S44. Based on the evaluation results, the core model of the yaw correction strategy (such as the yaw decoupling matrix and control command mapping function) is locally fine-tuned or updated through long-term learning to form a control system with evolutionary capabilities: Short-cycle fine-tuning (strategy layer) includes, but is not limited to: adjusting the weights of the torque distribution matrix Q(t); adjusting the duration of control commands. and direction threshold parameters; Long-term model self-evolution (model layer) includes, but is not limited to: accumulating historical response-instruction data pairs within a sliding window; and using lightweight RNNs or autoregressive models to model the control effect through "prediction-inversion-correction".

[0022] Example 2: The present invention proposes a cable car yaw attitude monitoring device, which is applicable to the control method of the cable car yaw attitude monitoring device proposed in Example 1. It includes: an integrated operation attitude perception and preprocessing module, a disturbance-driven yaw trend prediction and modeling module, a multi-axis coupling torque decoupling and correction strategy generation module, and a control closed-loop performance evaluation and adaptive strategy adjustment module.

[0023] The integrated operation attitude perception and preprocessing module synchronously collects multi-dimensional state information of the gondola during operation, performs data preprocessing operations, and provides high-quality input for subsequent modeling; The integrated operation attitude perception and preprocessing module includes a GPS position and motion perception unit, a three-axis gyroscope unit, and a main control processing unit (MCU). The main control processing unit (MCU), as the core of the system, integrates data-driven attitude recognition and yaw prediction algorithms, and executes the entire process from data preprocessing, state modeling, trend prediction to correction strategy generation. The GPS location and motion sensing unit is used to collect information such as the speed, geographical location, height change and direction of travel of the gondola in real time, as a sensing entry point for the macroscopic operating status. The three-axis gyroscope unit collects the angular velocity and attitude changes of the gondola in three degrees of freedom to capture micro yaw trends and spatial disturbance responses. The disturbance-driven yaw trend prediction and modeling module constructs a yaw angle prediction sequence in the future time domain, establishes a time-direction coupled trend model of the gondola's attitude change, and outputs yaw vectors at multiple future moments. The multi-axis coupled torque decoupling and correction strategy generation module constructs a yaw vector field based on the prediction results, executes regional coupled tensor modeling, direction alignment matrix solving and dynamic power constraint optimization strategies, generates the minimum disturbance correction vector for controlling the four corner torque gyroscopes and sends it down to generate directional counter-torque to maintain the stability of the gondola; The control closed-loop performance evaluation and adaptive strategy adjustment module collects the gondola attitude correction response and prediction residuals, constructs a multi-index control effect evaluation system, determines the current correction effect level, combines the residual map attribution mechanism to determine the cause of yaw anomalies, and adjusts the decoupling matrix, control commands and parameter models online to form a systematic dynamic adaptive correction closed loop.

[0024] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A control method for a cableway gondola yaw attitude monitoring device, characterized in that, The specific implementation steps include the following: S1. Collect information on the position, velocity, attitude angle and local micro-sway disturbance of the gondola through a multi-source sensing network, adaptively adjust the sampling frequency, normalize the data and then use dynamic time warping and least mean square estimation to fuse the asynchronous data, and output a fused state vector sequence. S2. Define sensor channel perturbation energy index through local perturbation tensor analysis, screen the set of significant interference dominant factors, and construct a perturbation attention weighting mechanism to generate enhanced state vectors by weighting and strengthening key perturbation dimensions; input the D-GRU network to predict future yaw angle sequences, and trigger an early warning and correction mechanism when the absolute value of the maximum yaw angle of the predicted sequence exceeds the safety threshold. S3. Construct a spatial vector field based on the predicted yaw sequence, analyze the torque interference between corners through regional coupling tensor analysis, combine dynamic direction alignment matrix projection decoupling, and optimize the torque command vector by integrating power consumption constraints and dynamic environmental parameters. It is also encapsulated as a standard control frame to synchronously drive the four corner torque gyroscopes to perform corrections. S4. Construct a comprehensive scoring function based on the attitude residual sequence to classify the control risk level, and combine the abnormal event graph attribution mechanism to perform short-period parameter fine-tuning and long-period model evolution update.

2. The control method for a cableway gondola yaw attitude monitoring device according to claim 1, characterized in that, Multi-source sensing networks include: GPS / RTK unit: provides the gondola's position P(t), velocity V(t), and heading angle θ(t); High-precision six-axis IMU unit: provides pitch angle α(t), roll angle β(t), yaw angle γ(t), and angular velocity; Distributed neuron pose array: deployed at the four corners of the gondola, providing local micro-sway perturbation information δ i (t) represents the local perturbation sway angle detected by the i-th neuron attitude array sensing node at time t.

3. The control method for a cableway gondola yaw attitude monitoring device according to claim 2, characterized in that, The process of generating the enhanced state vector is as follows: By calculating the average rate of change of disturbance for each sensor channel within the disturbance observation window as a disturbance energy index, channels exceeding a set threshold are selected to be included in the current set of dominant disturbance factors, and significant sway driving trends are identified. The attention weights of each channel are calculated based on the set of interference-dominant factors. An exponential amplification mechanism is used to strengthen the key perturbation dimensions, and the state vector is weighted and reconstructed to generate a perturbation-enhanced state vector.

4. The control method for a cableway gondola yaw attitude monitoring device according to claim 3, characterized in that, The prediction process of the D-GRU network for predicting future yaw angle sequences is as follows: ； in, The perturbation-gated variable represents the time t. This represents the hidden state at the previous time step; Indicates the candidate hidden state; and These represent the weight matrices corresponding to the current input and the previous state in the perturbation gating, respectively; and Let W and U represent the input and state weight matrices of the reset gate, respectively; W and U represent the input and weight matrices of the candidate state update function, respectively. This indicates element-wise multiplication; This represents the perturbation-enhanced state vector after perturbation attention weighting; Indicates the current hidden state; Output the predicted sequence of the current state. : ； ； in, This represents the prediction sequence consisting of all predicted yaw angles from time 1 to time T in the future; This represents the predicted yaw angle value at the k-th time step in the future; This represents the D-GRU hidden state at prediction time t+k, i.e., the intrinsic memory of the system state; This represents the weight matrix of the output mapping layer, used to map the hidden states. The value is mapped to a specific angle prediction.

5. The control method for a cableway gondola yaw attitude monitoring device according to claim 4, characterized in that, The optimization process for the torque command vector is as follows: Based on the predicted yaw sequence, a spatial vector field is constructed to map the trend torque at the four corners. By combining the coupling coefficient with the position vector, moment of inertia and directional angle, a regional coupled torque tensor is constructed to analyze the interference between corners. Based on the regional torque distribution and coupling tensor, the yaw trend gradient is calculated to obtain the main perturbation direction. A dynamic direction alignment matrix is ​​constructed to map the yaw direction to the gyroscope installation direction. The decoupling torque control vector is obtained, and the decoupling calculation is completed. By integrating physical boundaries, energy constraints, and historical inertial response characteristics through a dynamic torque optimization function, and combining dynamic parameters and maximum power limits, the decoupled torque command is optimized to generate a stable and executable modified torque vector.

6. The control method for a cableway gondola yaw attitude monitoring device according to claim 5, characterized in that, The process of constructing the region-coupled moment tensor is as follows: Based on the predicted yaw angle sequence, it is transformed into a spatial yaw vector field through a nonlinear mapping function, which maps the yaw trend torque vectors at the four corners of the bottom of the gondola to form a vector array; The four corners of the bottom of the gondola include: front left FL, front right FR, rear left BL, and rear right BR; Constructing the region coupling tensor: ； in, This represents the coupling moment tensor term between the i-th and j-th angles of the gondola; This represents the coupling coefficient between the i-th and j-th control angles in the gondola structure; and These represent the yaw tendency torque vectors at the i-th and j-th control angles, respectively; and These represent the position vectors of control angles i and j relative to the center of mass of the gondola; I i and I j These represent the local moments of inertia about the center of mass; This represents the angle between the principal directions of the sway at points i and j.

7. The control method for a cableway gondola yaw attitude monitoring device according to claim 6, characterized in that, The dynamic torque optimization function is: ； ； Where Tv represents the output torque vector of the four torque gyroscopes to be optimized at the current moment; This represents the system power consumption estimation function required for the current output torque; This indicates the set weighting factor; This represents the adaptive control penalty factor for the i-th control point; Indicates the basic penalty item; This represents the weighting coefficient corresponding to each dynamic indicator; This indicates the set wind speed reference value; Indicates the upper limit of the reference value for the angular velocity of the gondola; This indicates the acceptable upper limit of vertical vibration; This represents the maximum tolerable safety delay threshold of the control system; Q(t) represents the theoretically optimal control torque vector after decoupling. This represents the historical output sequence of torque. This represents the current set of dynamic parameters for the gondola's operation, including wind speed. angular velocity of the hoisting car Vertical vibration acceleration of the hoisting car Control delay ; This indicates the maximum output power of the torque gyroscope; This represents the dynamic torque optimization function.

8. The control method for a cableway gondola yaw attitude monitoring device according to claim 7, characterized in that, The comprehensive scoring function consists of three indicators: stability, overshoot, and response delay, and its form is as follows: ； ； ； ； in, This indicates a correction to the stability index; This represents the weighted residual contribution value at time point t+k; This indicates the correction of the overshoot indicator; This indicates the maximum acceptable safe sway angle threshold for the gondola's operation; Indicates response latency; This represents the tolerance threshold for the yaw angle residual. This represents the overall score of the control scoring function, i.e., the overall scoring function; , and This represents the weighting coefficient.

9. The control method for a cableway gondola yaw attitude monitoring device according to claim 3, characterized in that, The disturbance energy index is calculated using a sliding observation window method, with the window length being a dynamically adjustable parameter that adaptively increases or decreases according to the disturbance intensity.

10. A cable car gondola yaw attitude monitoring device, applicable to the control method for a cable car gondola yaw attitude monitoring device as described in any one of claims 1 to 9, characterized in that, include: The integrated operation attitude perception and preprocessing module is used to synchronously collect multi-dimensional status information of the gondola during operation and perform data preprocessing operations. The integrated operation attitude perception and preprocessing module includes a GPS position and motion perception unit, a three-axis gyroscope unit, and a main control processing unit (MCU). The main control processing unit (MCU) integrates data-driven attitude recognition and yaw prediction algorithms, and executes the entire process from data preprocessing, state modeling, trend prediction to correction strategy generation. GPS location and motion sensing unit is used to collect the location and motion information of the gondola in real time during operation; The three-axis gyroscope unit collects the angular velocity and attitude changes of the gondola in three degrees of freedom, capturing micro yaw trends and spatial disturbance responses. The disturbance-driven yaw trend prediction and modeling module constructs a future yaw angle prediction sequence in the time domain and establishes a disturbance-gated recursive unit network (D-GRU) for gondola attitude changes to predict the future yaw trend angle sequence. The multi-axis coupled torque decoupling and correction strategy generation module is used to construct a yaw vector field based on the predicted sequence, perform coupled tensor modeling, direction alignment matrix generation and dynamic power consumption optimization, and generate control torque commands for the gyroscope to execute corrections. The control closed-loop performance evaluation and adaptive strategy adjustment module is used to score the correction effect, attribute anomalies, and update strategies based on attitude residuals, thereby completing the system's self-evolutionary regulation.

Images