A dynamic self-adaptive control method and system for a non-coal mine track inspection robot

Abstract

A non-coal mine track inspection robot dynamic self-adaptive control method and system, step S1: track abnormal working condition real-time sensing, step S2: drive system adaptive control, step S3: catastrophic response decision, according to the trigger condition, execute corresponding action: when B8=1, trigger electromagnetic brake and sound and light alarm, electromagnetic brake response time is less than or equal to 70ms; Through multi-dimensional technical innovation, it provides an efficient, safe and intelligent technical solution for non-coal mine track inspection, which has important significance for promoting the technical progress of non-coal mine automatic inspection field.

Description

Technical Field

[0001] This invention relates to the field of intelligent mining equipment technology, and in particular to a dynamic adaptive control method and system for a non-coal mine track inspection robot. Background Technology

[0002] In the daily operation of non-coal mines, equipment condition monitoring and work environment inspection are core aspects of ensuring production safety. Traditional manual inspection methods have significant limitations: on the one hand, due to the complex terrain and work area of ​​the mine, manual inspections cannot achieve full coverage, and a single inspection can take 4-8 hours, requiring a large investment of manpower for high-frequency monitoring; on the other hand, non-coal mines commonly have safety hazards such as roof falls, wall spalling, and toxic gas leaks. According to industry statistics, environmental risks account for 32% of safety accidents caused by manual inspections.

[0003] Existing inspection robot technology also has significant shortcomings: motion control mostly adopts a fixed speed mode, lacking the ability to dynamically adapt to on-site working conditions. When abnormal situations such as track icing, breakage, or oil contamination occur, the operating strategy cannot be adjusted in real time, leading to frequent malfunctions such as robot derailment and power failure, seriously affecting the continuity of inspections. Especially in winter operation scenarios in open-pit mines, robot slippage accidents caused by track icing account for as much as 45%, urgently requiring solutions to these problems through intelligent control methods. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a dynamic adaptive control method and system for a non-coal mine track inspection robot. Through multi-dimensional technological innovation, it provides an efficient, safe and intelligent technical solution for non-coal mine track inspection, which is of great significance to promoting technological progress in the field of automated inspection in non-coal mines.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a dynamic adaptive control method for a non-coal mine track inspection robot, comprising the following steps:

[0006] Step S1: Real-time detection of abnormal track conditions, specifically including:

[0007] S11: Contact-type track temperature monitoring. A spring-loaded metal probe contacts the top surface of the track with a contact force >5N. A PT1000 temperature sensor measures the temperature difference of the track through conductive heat collection, and a thermal inertia compensation algorithm is used to calculate the compensated temperature difference.

[0008]

[0009] Where K is the thermal conductivity coefficient, dynamically optimized using a BP neural network, and the temperature difference after compensation is... When this occurs, the freezing flag TG=1 is triggered;

[0010] S12: Environmental temperature and humidity monitoring, collecting ambient temperature and humidity data. and humidity The high humidity icing risk indicator TH=1 is triggered.

[0011] S13: Track friction coefficient estimation, using Kalman filtering to fuse drive motor current. With accelerometer signal Based on data fusion model

[0012] Calculate the measured friction coefficient, when the measured friction coefficient Theoretical value When this occurs, the friction anomaly flag C=1 is triggered;

[0013] S14: Dynamic determination of risk index, calculated based on an adaptive weighting model.

[0014] Where the weights α, β, and γ are adaptive weights, generated through training on historical accident data, and are calculated using the following formula:

[0015]

[0016]

[0017]

[0018] in, , and These represent the standard deviations of the data sources corresponding to the icing indicator, the high humidity icing risk indicator, and the abnormal friction indicator, respectively; when the test data Greater than the set threshold When this happens, the system will trigger a track anomaly alarm;

[0019] S15: Track breakage detection, which identifies track breakage features by improving the YOLO algorithm and triggers the breakage flag B8=1;

[0020] Step S2: Adaptive control of the drive system, specifically including:

[0021] S21: Construct a three-dimensional track model with turning radius R and slope angle θ based on the finite element method, and generate graded track constraint speeds;

[0022] S22: Acquire raw vibration signals using a vibration sensor. Calculate the vibration correction velocity

[0023] in The initial velocity, This is the vibration attenuation coefficient;

[0024] S23: Online weight optimization using particle swarm optimization algorithm , Through formula

[0025]

[0026] The final velocity is calculated using the following weight update formula:

[0027]

[0028] c1 is the cognitive factor, and c2 is the social factor;

[0029] S24: Dynamic correction based on vibration signal-to-noise ratio :

[0030] ,

[0031] When the track is slippery hour, ;

[0032] Step S3: Disaster response decision, execute the corresponding action according to the triggering condition: when B8=1, trigger electromagnetic brake and audible and visual alarm, electromagnetic brake response time ≤70ms;

[0033] when Trigger audible and visual alarms and remote alarms;

[0034] when Trigger a full response at the time;

[0035] Furthermore, step 14 sets a threshold. The threshold value is determined based on the 95% confidence interval of historical accident data over the past 3 years, and supports a quarterly dynamic adjustment of ±15%, with the adjustment based on the quarterly mine operating condition risk level.

[0036] Furthermore, in step S15, the improved YOLO algorithm adds an attention mechanism module to the feature extraction network, assigning 1.5-2.0 times weight to track seams and fracture areas, achieving a detection frame rate of ≥25fps and a fracture recognition accuracy of ≥98%.

[0037] Furthermore, in step 23, the cognitive factor c1 of the particle swarm algorithm has a value range of 0.8-1.2, and the social factor c2 has a value range of 1.5-2.0. When the iteration error is >5%, c1=0.8 and c2=2.0 to accelerate convergence; when the iteration error is ≤5%, c1=1.2 and c2=1.5 to optimize accuracy.

[0038] Furthermore, in step S3, the intensity of the audible and visual alarm is 30±5 times / minute for red light and ≥85dB for sound intensity; the remote alarm includes robot GPS positioning information with an accuracy of ±3m and an abnormal working condition code.

[0039] A dynamic adaptive control system for a non-coal mine track inspection robot includes:

[0040] Track anomaly sensing module: includes contact temperature measurement unit, environmental monitoring unit, friction estimation unit, risk assessment unit, and fracture detection unit;

[0041] The contact temperature measurement unit includes a spring-loaded probe, a PT1000 sensor, and a BP neural network module; the environmental monitoring unit includes a temperature and humidity sensor; the friction estimation unit includes a current sensor, an accelerometer, and a Kalman filter module; the risk assessment unit includes an adaptive weight calculation module; and the fracture detection unit includes a high-definition camera and an improved YOLO algorithm module.

[0042] Driven adaptive control module: contains finite element modeling unit, vibration correction unit, particle swarm optimization unit, and weight correction unit;

[0043] The vibration correction unit includes a vibration sensor and a correction calculation module; the particle swarm optimization unit optimizes the weights λ1 and λ2; the weight correction unit dynamically adjusts λ1 based on the signal-to-noise ratio.

[0044] Disaster response module: includes braking actuator and alarm unit; braking actuator includes dual-coil electromagnetic locking mechanism, alarm unit includes audible and visual alarm and remote communication module;

[0045] The modules work together to implement the control method described above.

[0046] The beneficial effects of this invention are: significantly improving the intelligence and automation level of inspection operations; realizing real-time intelligent decision-making and dynamic control based on on-site working conditions by constructing a multi-level dynamic control architecture; the real-time track abnormality perception system relies on multi-sensor fusion detection algorithms to accurately identify abnormal states such as track icing and breakage; the drive system adaptive control system, through multi-source data fusion decision-making and particle swarm optimization, can dynamically adjust the robot's running speed and output torque, making the inspection process accurately adapt to the complex and ever-changing environmental characteristics of the mine, effectively solving the technical problems of insufficient intelligence and lagging environmental response in the prior art.

[0047] Significantly enhances the safety and reliability of robot operation; the disaster response decision mechanism pre-sets clear trigger conditions and corresponding response actions for different scenarios such as track breakage, risk index exceeding threshold, and compound disasters, and makes strict specifications for braking performance parameters; when encountering emergencies such as track breakage, the braking and alarm procedures can be quickly activated to minimize the probability of robot driving failure and safety accidents, avoid inspection interruption or equipment damage caused by improper handling of abnormal conditions, and ensure the continuous and stable operation of inspection work.

[0048] It possesses excellent environmental adaptability and working condition adaptability; the track constraint adaptive algorithm in the drive system can dynamically correct by combining parameters such as turning radius, slope angle and real-time vibration sensor signals, and can accurately adapt to the complex working conditions of different roadways; the application of technologies such as thermal inertia compensation algorithm and friction coefficient estimation ensures that the robot can maintain a stable working state in outdoor environments with significant temperature differences, effectively overcoming the adverse effects of complex environments in non-coal mines on inspection equipment, and ensuring the continuity and stability of inspection operations. Attached Figure Description

[0049] The present invention will be further described below with reference to the accompanying drawings:

[0050] Figure 1 Here is a schematic diagram of a multi-level dynamic control architecture;

[0051] Figure 2 Here is a flowchart of the real-time sensing system for abnormal track conditions.

[0052] Figure 3 Here is a flowchart of the adaptive control system for the drive system.

[0053] Figure 4 Here is a diagram illustrating the adjustment of cognitive and social factors.

[0054] Figure 5 Here is the overall flowchart of the multi-level dynamic control architecture. Detailed Implementation

[0055] The present invention will be further described in detail below with reference to embodiments and specific implementation methods:

[0056] Example 1

[0057] In a certain open-pit iron mine, where winter temperatures range from -5℃ to -2℃, the tracks are prone to icing, causing robots to slip. After adopting the method of this invention:

[0058] Orbital anomaly detection:

[0059] Contact temperature measurement: probe contact force 6N, =4℃, dT / dt=0.6℃ / s, BP neural network optimization K=2.5, =4+2.5×0.6=5.5℃>5℃, TG=1;

[0060] Environmental monitoring: =-3℃, =93%, TH=1;

[0061] Friction estimation: =6A, =0.2g, =0.08< =0.35, C=1;

[0062] Risk Index: =0.3、 =0.25、 =0.35, α=0.33, β=0.28, γ=0.39, =0.33×1+0.28×1+0.39×1=1.0> =0.65, triggering an alarm.

[0063] Drive adaptive control:

[0064] Track model: straight track, =1m / s;

[0065] Vibration correction: =0.8g, =1 - 0.2 × 0.8 = 0.84 m / s;

[0066] Particle swarm optimization: c1=0.8, c2=2.0, initial λ1=0.5, λ2=0.5. =0.5×1+0.5×0.84=0.92m / s;

[0067] Weight adjustment: because =0.08<0.35, δ'(t)=1.3, λ1 is corrected to 0.5×1.3=0.65. =0.65×1+0.35×0.84=0.65+0.294=0.944m / s, strengthening track constraints and reducing the risk of slippage.

[0068] Disaster response: No track breakage was detected, only an audible and visual alarm was triggered, with red light flashing 32 times per minute and sound intensity at 88 dB.

[0069] Example 2

[0070] The track of an underground copper mine was contaminated by oil leakage, resulting in a reduced coefficient of friction. After applying the method of this invention:

[0071] Orbital anomaly detection:

[0072] Contact temperature measurement: probe contact force 6N, =1℃, =1.18℃<5℃, TG=0;

[0073] Environmental monitoring: =18℃, =75%, TH=0;

[0074] Friction estimation: =5A, =0.3g, =0.12< =0.35, C=1;

[0075] Risk Index: =0.1、 =0.1、 =0.5, α=0.15, β=0.15, γ=0.7, =0.7> =0.65, triggering an alarm.

[0076] Drive adaptive control:

[0077] Track model: straight track, =1m / s;

[0078] Vibration correction: =0.6g, =1 - 0.15 × 0.6 = 0.91 m / s;

[0079] Particle swarm optimization: λ1=0.55, λ2=0.45, =0.55×1+0.45×0.91=0.96m / s;

[0080] Weight adjustment: because <0.35, δ'(t)=1.2, λ1 is corrected to 0.55×1.2=0.66. =0.66×1+0.34×0.91=0.66+0.31=0.97m / s, strengthening track constraints and reducing the risk of slippage.

[0081] Example 3

[0082] A limestone mine track developed a 10cm crack due to corrosion. The response process of this invention is as follows:

[0083] Fracture detection: High-definition camera captures images, and the improved YOLO algorithm identifies fracture features within 25ms with a confidence level of 99% and B8=1.

[0084] Risk Index: =1.2> =0.65, triggering a compound disaster.

[0085] Disaster Response:

[0086] Electromagnetic locking brake is triggered within 65ms, and the brake disc generates a braking torque of 80N·m.

[0087] Activate the audible and visual alarm, with red light flashing 31 times per minute and sound intensity 87dB, and remote alarm with positioning accuracy of ±2m;

[0088] The robot came to a complete stop within a distance of 1.2m without derailing.

[0089] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic adaptive control method for a non-coal mine track inspection robot, characterized in that, Includes the following steps: Step S1: Real-time detection of abnormal track conditions, specifically including: S11: Contact-type track temperature monitoring. A spring-loaded metal probe contacts the top surface of the track with a contact force >5N. A PT1000 temperature sensor measures the temperature difference of the track through conductive heat collection, and a thermal inertia compensation algorithm is used to calculate the compensated temperature difference. ， Where K is the thermal conductivity coefficient, dynamically optimized using a BP neural network, and the temperature difference after compensation... When this occurs, the freezing flag TG=1 is triggered; S12: Environmental temperature and humidity monitoring, collecting ambient temperature and humidity data. and humidity The high humidity icing risk indicator TH=1 is triggered. S13: Track friction coefficient estimation, using Kalman filtering to fuse drive motor current. With accelerometer signal Based on data fusion model ， Calculate the measured friction coefficient, when the measured friction coefficient Theoretical value When this occurs, the friction anomaly flag C=1 is triggered; S14: Dynamic determination of risk index, calculated based on an adaptive weighting model. ， Where the weights α, β, and γ are adaptive weights, generated through training on historical accident data, and are calculated using the following formula: ， ， ， in, , and These represent the standard deviations of the data sources corresponding to the icing indicator, the high humidity icing risk indicator, and the abnormal friction indicator, respectively; when the test data Greater than the set threshold When this happens, the system will trigger a track anomaly alarm; S15: Track breakage detection, which identifies track breakage features by improving the YOLO algorithm and triggers the breakage flag B8=1; Step S2: Adaptive control of the drive system, specifically including: S21: Construct a three-dimensional track model with turning radius R and slope angle θ based on the finite element method, and generate graded track constraint speeds; S22: Acquire raw vibration signals using a vibration sensor. Calculate the vibration correction velocity ， in The initial velocity, This is the vibration attenuation coefficient; S23: Online weight optimization using particle swarm optimization algorithm , Through formula ， The final velocity is calculated using the following weight update formula: ， c1 is the cognitive factor, and c2 is the social factor; S24: Dynamic correction based on vibration signal-to-noise ratio : , ， When the track is slippery hour, ; Step S3: Disaster response decision, execute the corresponding action according to the triggering condition: when B8=1, trigger electromagnetic brake and audible and visual alarm, electromagnetic brake response time ≤70ms; when Trigger audible and visual alarms and remote alarms; when Trigger a full response at the time; Step 14 sets the threshold. The threshold value is determined based on the 95% confidence interval of historical accident data over the past 3 years, and supports a quarterly dynamic adjustment of ±15%, with the adjustment based on the quarterly mine operating condition risk level. In step S15, the improved YOLO algorithm adds an attention mechanism module to the feature extraction network, assigning 1.5-2.0 times weight to track seams and fracture areas, achieving a detection frame rate of ≥25fps and a fracture recognition accuracy of ≥98%. In step 23, the cognitive factor c1 of the particle swarm algorithm has a value range of 0.8-1.2, and the social factor c2 has a value range of 1.5-2.

0. When the iteration error is >5%, c1=0.8 and c2=2.0 to accelerate convergence; when the iteration error is ≤5%, c1=1.2 and c2=1.5 to optimize accuracy.

2. The method according to claim 1, characterized in that, In step S3, the intensity of the audible and visual alarm is 30±5 times / minute for red light and ≥85dB for sound intensity; the remote alarm includes robot GPS positioning information with an accuracy of ±3m and abnormal working condition codes.

3. A dynamic adaptive control system for a non-coal mine track inspection robot, characterized in that, include: Track anomaly sensing module: includes contact temperature measurement unit, environmental monitoring unit, friction estimation unit, risk assessment unit, and fracture detection unit; The contact temperature measurement unit includes a spring-loaded probe, a PT1000 sensor, and a BP neural network module. The environmental monitoring unit includes temperature and humidity sensors; the friction estimation unit includes a current sensor, an accelerometer, and a Kalman filter module; the risk assessment unit includes an adaptive weight calculation module; and the fracture detection unit includes a high-definition camera and an improved YOLO algorithm module. Driven adaptive control module: contains finite element modeling unit, vibration correction unit, particle swarm optimization unit, and weight correction unit; The vibration correction unit includes a vibration sensor and a correction calculation module; the particle swarm optimization unit adjusts the weights. , Optimization; Weight correction unit based on signal-to-noise ratio Dynamic adjustment; Disaster response module: includes a braking execution unit and an alarm unit; the braking execution unit includes a dual-coil electromagnetic locking mechanism, and the alarm unit includes an audible and visual alarm and a remote communication module; the modules work together to implement the control method described in any one of claims 1-2.

Images