CN-121800021-B - Inclined shaft skip posture abnormality detection and control method based on multi-source data fusion

CN121800021BCN 121800021 BCN121800021 BCN 121800021BCN-121800021-B

Abstract

The invention relates to the technical field of industrial automation control, in particular to a method for detecting and controlling abnormal postures of a skip of an inclined shaft based on multi-source data fusion. The method comprises the steps of acquiring multi-source operation data, and performing time lag compensation on suspension tension data based on a cross-correlation principle to realize space-time synchronization of the multi-source data. And constructing a longitudinal dynamics observation model and identifying the current load quality in real time by using a recursive least square method of the self-adaptive forgetting factor. And inhibiting track impact interference and estimating the attitude inclination in real time by utilizing an unscented Kalman filtering algorithm of an inertial attitude data combined robust estimation theory. And constructing a dynamic safety envelope based on the current load quality and the running speed, calculating a dynamic gesture safety threshold, and comparing the gesture inclination angle with the dynamic gesture safety threshold to realize running control. The scheme of the invention can identify the load in real time, filter interference, dynamically adjust the safety bottom line and realize accurate operation control under complex working conditions.

Inventors

YANG JIAN
LIU GANG
WU WEIXING
ZHOU BINGLEI
YANG ZI
DONG GUANJUN

Assignees

洛阳点晶智控科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (8)

1. The inclined shaft skip posture abnormality detection and control method based on multi-source data fusion is characterized by comprising the following steps: Acquiring inertial attitude data, driving condition data and suspension tension data of a skip of an inclined shaft, performing time lag compensation on the suspension tension data based on a cross-correlation principle, and realizing time-space synchronization of multi-source data by taking a time stamp of the inertial attitude data as a reference; establishing an observation equation based on Newton's second law, wherein the equation describes a linear combination relation between suspension tension and skip gravity along a shaft component, track friction force and inertia force; setting the real-time total mass of the skip as a time-varying parameter to be identified, taking a physical quantity combination comprising the gravity acceleration, the shaft inclination angle, the friction coefficient and the actually measured longitudinal acceleration as an observation input, and taking the synchronous suspension tension as an observation value, so that the dynamics problem is converted into a parameter estimation problem; the method for identifying the current load quality of the skip in real time by using the data subjected to space-time synchronization through a recursive least square method introducing the self-adaptive forgetting factor comprises the steps of dynamically calculating the forgetting factor at the current moment according to the energy of a prediction error, wherein the calculation relational expression is as follows: , as a forgetting factor, Is a preset lower limit of the forgetting factor, In order for the prediction error to be a priori, As a variance of the background noise, In order to observe the input scalar quantity, A covariance scalar for the last time; the inertial attitude data are utilized to inhibit track impact interference by introducing an unscented Kalman filtering algorithm of an robust estimation theory, and the attitude dip angle of the skip is estimated in real time; Based on the current load mass and the running speed in the driving working condition data, a dynamic safety envelope of load and speed coupling is constructed, a dynamic gesture safety threshold is calculated, and the gesture inclination angle is compared with the dynamic gesture safety threshold to realize the running control of the skip.
2. The method for detecting and controlling the abnormal posture of the skip in the inclined shaft based on the multi-source data fusion according to claim 1, wherein the real-time estimation of the posture inclination angle of the skip by the unscented kalman filter algorithm introducing the robust estimation theory comprises the steps of dynamically adjusting the observed noise variance according to the innovation residual error, wherein the adjustment relation is as follows: ; In the formula, In order to observe the variance of the noise, For the sensor base noise variance, In order to penalize the gain factor, For the purpose of the innovation residual error, As a theoretical covariance of the innovation, Is a constant value of the stability of the numerical value, For the abnormality discrimination threshold value, Representing the trace of the matrix.
3. The method for detecting and controlling the abnormal posture of the inclined shaft skip bucket based on the multi-source data fusion according to claim 1, wherein the calculation of the dynamic posture safety threshold is as follows: ; In the formula, For the dynamic gesture safety threshold at the current time, As a safety angle of the reference structure, For a rated full-load mass of the material, In order to identify the resulting current load mass, For a smooth bias of the quality, As a speed-decay factor, For the purpose of real-time operation of the speed, In order to design the maximum speed, In order for the speed compensation constant to be a function of, Is the minimum safety bottom line.
4. The method for detecting and controlling the abnormal posture of the inclined shaft skip bucket based on the multi-source data fusion according to claim 1, wherein the time-lag compensation is performed on the suspension tension data based on a cross-correlation principle, and the method specifically comprises the following steps: Intercepting suspension tension data and a change rate sequence thereof through a sliding window, and calculating the cross correlation degree of the change rate sequence and an acceleration speed sequence of inertial attitude data; traversing different time displacement amounts, searching the displacement amount which enables the cross correlation degree to reach the maximum value, and determining the displacement amount as the total lag time of system transmission and mechanical transmission; and carrying out time axis translation correction on the suspension tension data by utilizing the total lag time to ensure that the change phase of the suspension tension data is strictly aligned with the inertia attitude data.
5. The method for detecting and controlling the abnormal posture of the inclined shaft skip based on the multi-source data fusion, which is disclosed by claim 1, is characterized in that the prior prediction error is calculated by subtracting a theoretical tension value obtained based on the product of a mass estimated value at the last moment and a current observation input scalar from synchronous tension data at the current moment, wherein the observation input scalar is formed by the sum of the component of gravity acceleration along the shaft direction, the product of an orbit friction coefficient and the component of gravity acceleration vertical to the shaft direction and the actual measurement longitudinal acceleration.
6. The method for detecting and controlling the abnormal posture of the inclined shaft skip bucket based on the multi-source data fusion according to claim 2, wherein the unscented Kalman filtering algorithm predicts the state by using the angular velocity in the inertial posture data and updates the measurement by using the inclination angle calculated by the accelerometer, and the logic for dynamically adjusting the observed noise variance is as follows: Constructing a standardized statistic representing the degree of deviation of the current observed value from the theoretical prediction range; setting a statistical discrimination threshold for discriminating normal noise from abnormal impact; when the standardized statistics exceeds the judging threshold, judging that the current observation is interfered by non-Gaussian impact, and rapidly amplifying the observation noise variance through nonlinear mapping to reduce the trust weight of the filter on the observation data at the moment, so that the smoothness of the attitude estimation is maintained.
7. The method for detecting and controlling the abnormal posture of a skip in a deviated well based on multi-source data fusion according to claim 3, wherein the logic for constructing the dynamic security envelope comprises: Establishing a negative correlation mapping mechanism of an attitude safety threshold and a skip real-time momentum state; in the mass dimension, as the recognized load mass increases, the gesture safety threshold is compressed through the inverse proportion relation so as to adapt to the inertial risk under large mass; In the speed dimension, as the running speed increases, the attitude safety threshold is further compressed through a quadratic relation to adapt to centrifugal instability at high speed; through double correction, the gesture safety threshold automatically contracts to a preset minimum safety base line under a heavy-load and high-speed limit working condition.
8. The method for detecting and controlling the abnormal posture of the inclined shaft skip bucket based on the multi-source data fusion according to claim 1, wherein the implementation of the operation control of the skip bucket specifically comprises the following steps: The attitude dip angle of the skip bucket is monitored in real time, and compared with the calculated dynamic attitude safety threshold value; If the attitude inclination angle is smaller than or equal to the dynamic attitude safety threshold value, maintaining the current operation strategy of the skip; if the attitude inclination angle is larger than the dynamic attitude safety threshold, judging that the skip is in an abnormal attitude dangerous state, immediately sending a scram instruction to the driving system, and triggering an audible and visual alarm.

Description

Inclined shaft skip posture abnormality detection and control method based on multi-source data fusion Technical Field The invention relates to the technical field of industrial automation control. More particularly, the invention relates to a method for detecting and controlling abnormal postures of a inclined shaft skip based on multi-source data fusion, which is particularly suitable for safety monitoring of a mine inclined shaft lifting system. Background Inclined shaft skip is a core transport container in inclined shaft of mine, and safety in operation is important. In actual production working conditions, the inclined shaft track environment is extremely complex, and joint, dislocation or local deformation phenomena are commonly caused. The skip can generate severe instantaneous impact vibration when passing through the uneven area at high speed. Meanwhile, the load parameters of the skip have extremely strong time variability, and the skip can undergo the processes of no-load lowering, loading, full-load lifting, unloading and the like in a single operation period, and the total mass of the skip can be changed drastically from a few tons to tens of tons. Furthermore, modern mine monitoring systems involve multi-source heterogeneous data, such as drive condition data from wellhead hoist room control systems and attitude data from downhole follow-up sensors. In order to ensure operation safety and cope with the complex working conditions, various improved schemes are developed in the existing monitoring and control technology. For impact vibrations caused by track irregularities, conventional monitoring algorithms typically build data processing models based on gaussian noise assumptions, in an attempt to thereby filter out environmental disturbances and identify pose information. In the face of severe changes of skip quality, a control strategy for setting a fixed gesture alarm threshold is generally adopted in the industry, and is used as a unified safety protection boundary. In the utilization of multi-source heterogeneous data, the existing system generally directly aggregates the collected surface driving operation data with the attitude and mechanical data acquired by the underground sensor, and synchronously inputs the collected surface driving operation data and the attitude and mechanical data into a system dynamics model for fusion calculation, so that the real-time evaluation and safety control of the skip operation state are expected to be realized. However, the improved technology still has great defects in practical application. The instantaneous impact signals generated by the track joints are obvious non-Gaussian noise in statistical characteristics, the amplitude of the instantaneous impact signals is often far beyond the normal vibration range, and the algorithm based on Gaussian assumption is extremely easy to misjudge the instantaneous impact signals as abnormal postures such as derailment or tipping, so that frequent false alarms and system scram are caused, and the mine lifting efficiency is seriously reduced. Meanwhile, the change of the quality directly affects the rotational inertia and stability boundary of the system, an obvious logic loophole exists in a fixed threshold strategy, if the threshold is set according to a full load standard, frequent false alarm is caused when no load occurs, if the threshold is set according to the no load standard, false alarm is caused when the full load is severely rocked, and the real safety protection effect is difficult to play. More seriously, the physical distance-induced wireless transmission delay and the mechanical transmission delay caused by the elasticity of the steel wire rope lead to natural phase differences of various heterogeneous data on a time axis, and if accurate space-time synchronization processing is not carried out, delayed mechanical data and real-time motion data are directly fused and calculated, so that the calculation of a dynamics model is inevitably invalid and an erroneous state estimation result is generated. Disclosure of Invention The invention aims to provide a multi-source data fusion-based inclined shaft skip posture abnormality detection and control method, which is used for solving the problems of high false alarm rate caused by non-Gaussian impact interference of a track, failure of a fixed threshold value caused by severe change of a load parameter, large model calculation error caused by asynchronous multi-source data and the like in the prior art. The invention provides a multi-source data fusion-based inclined shaft skip posture abnormality detection and control method, which comprises the following steps: The method comprises the steps of acquiring inertial attitude data, driving working condition data and hanging tension data of a skip of an inclined shaft, performing time lag compensation on the hanging tension data based on a cross-correlation principle, achieving time-space synchronization o