Intelligent rock drilling robot angle correction method and system based on artificial intelligence

Abstract

The invention discloses an intelligent rock drilling robot angle correction method and system based on artificial intelligence. Comprising the following steps: acquiring distance data between a drill rod and a target rock surface by using a laser ranging module, calculating a current deflection angle and generating a deviation correction signal; inputting the correction signal into a pre-trained BP neural network to output a dynamic correction threshold value; generating a control instruction if the current deflection angle absolute value is greater than or equal to a threshold; the hydraulic servo mechanism executes an instruction to adjust the angle of the drill rod, and the inertial measurement unit performs feedback to form closed-loop control. The system comprises a data acquisition module, a decision control module, an execution mechanism and a feedback module. The pulse type laser ranging technology is adopted, the angle is calculated by combining multi-position data fitting, and the measurement precision reaches + / -1 mm. The BP neural network is trained based on different operation data and can output a reasonable correction threshold value. Besides, the method is optimized, such as echo intensity analysis and acquisition frequency adjustment, so that the problem of drill rod angle control can be effectively solved, and the drilling efficiency and the drilling precision are improved.

Description

Technical Field

[0001] The present invention is based on the technical field of rock drilling robot angle correction, and specifically provides an artificial intelligence-based intelligent rock drilling robot angle correction method and system. Background Art

[0002] Rock drilling robots are increasingly being used. However, drilling operations rely on manual labor, which is labor-intensive, difficult to guarantee accuracy, and prone to accidents. While unmanned, intelligent rock drilling robots are becoming increasingly common with the advancement of automation technology, controlling the drill rod angle remains a major challenge in actual operations. Complex operating environments, uneven rock texture, and cracks can lead to uneven force on the drill rod during drilling, causing it to deflect. Excessive deviation in the drill rod angle not only reduces drilling efficiency and wastes energy, but can also cause the hole to deviate from its intended location. Summary of the Invention

[0003] The purpose of the present invention is to provide an angle correction method and system for an intelligent rock drilling robot based on artificial intelligence to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solution: an angle correction method for an intelligent rock drilling robot based on artificial intelligence, comprising the following steps:

[0005] Step 1: Use the laser ranging module to collect real-time distance data between the drill rod of the rock drilling robot and the target rock surface, calculate the current deflection angle θc of the drill rod based on the distance data, and generate a correction signal containing the angle deviation value;

[0006] Step 2: Input the correction signal into a pre-trained BP neural network model, perform nonlinear mapping on the input signal through the hidden layer of the BP neural network, and output a dynamic correction threshold Δθ, wherein the BP neural network is trained by the mapping relationship between the deflection angle and the optimal correction threshold in the historical operation data;

[0007] Step 3: Compare the current deflection angle θc with the dynamic correction threshold Δθ. If |θc| ≥ Δθ, generate a control instruction containing the correction direction and force. The force of the control instruction is directly proportional to |θc - Δθ|.

[0008] Step 4: The control command is executed by the hydraulic servo mechanism to drive the drill rod to adjust in the correction direction at a preset angular velocity ω. At the same time, the adjusted deflection angle θ'c is fed back in real time through the inertial measurement unit until |θ'c| < Δθ is satisfied, forming a closed-loop control.

[0009] Furthermore, the laser ranging module adopts pulse laser ranging technology to calculate the distance between the drill rod and the target rock surface by measuring the time difference between the emission and reception of the laser pulse, and the measurement accuracy can reach ±1mm.

[0010] Furthermore, when calculating the current deflection angle θc of the drill pipe, a fitting calculation is performed by combining the distance data of at least three different positions collected by the laser ranging module through trigonometric function relationships to improve the accuracy of the angle calculation.

[0011] Furthermore, the BP neural network model includes an input layer, at least two hidden layers and an output layer. The number of input layer nodes is determined according to the distance data characteristics, the number of hidden layer nodes is determined by trial and error and empirical formulas, and the number of output layer nodes is 1, which is used to output the dynamic correction threshold Δθ.

[0012] Furthermore, during the training process of the BP neural network model, historical operation data includes the mapping relationship between the deflection angle and the optimal correction threshold under different geological conditions, different drill rod lengths and different drilling depths, and the training data is normalized to accelerate the training convergence speed.

[0013] Furthermore, when comparing the current deflection angle θc with the dynamic correction threshold Δθ, if |θc| ≥ Δθ, the correction direction is determined by judging the positive and negative relationship between θc and Δθ. When θc is positive and greater than Δθ, the correction direction is negative adjustment, otherwise it is positive adjustment.

[0014] Furthermore, the hydraulic servo mechanism includes a hydraulic pump, a hydraulic cylinder and an electromagnetic reversing valve. The electromagnetic reversing valve switches the direction of the oil circuit according to the control instruction, so that the hydraulic cylinder drives the drill rod to adjust in the correction direction at a preset angular velocity ω. The preset angular velocity ω is determined according to the material, length and working environment of the drill rod, and ranges from 0.1 to 1 rad / s.

[0015] An artificial intelligence-based intelligent rock drilling robot angle correction system, comprising:

[0016] Data acquisition module: including a laser ranging unit and an inertial measurement unit. The laser ranging unit is used to obtain the distance data between the drill rod and the rock surface in real time, and the inertial measurement unit is used to detect the spatial attitude angle of the drill rod;

[0017] Decision control module: includes an embedded processor and a BP neural network algorithm stored in the processor, wherein the embedded processor is configured to execute:

[0018] Calculate the current deflection angle θc based on the distance data;

[0019] Output the dynamic correction threshold Δθ through BP neural network;

[0020] Generate control instructions proportional to |θc-Δθ|;

[0021] Actuator module: includes a hydraulic servo driver and a correction actuator connected to the driver, wherein the hydraulic servo driver drives the correction actuator to adjust the angle according to the control instructions;

[0022] Feedback module: includes a gyroscope installed on the drill pipe base, which is used to monitor the adjusted deflection angle θ'c in real time and feed it back to the decision control module to form a closed-loop control system.

[0023] Furthermore, the laser transmitter used in the laser ranging unit emits a laser beam with a wavelength of 905 nm, and the laser receiver uses an avalanche photodiode to improve the sensitivity and anti-interference ability of distance measurement.

[0024] Furthermore, the embedded processor is an ARM architecture microprocessor, which integrates a high-speed computing unit and a large-capacity storage unit, can quickly process data and run the BP neural network algorithm, and the storage unit is used to store historical operation data and trained BP neural network model parameters.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] Using pulsed laser ranging technology, the measurement accuracy can reach ±1mm. By collecting distance data from at least three different locations and combining it with trigonometric function fitting to calculate the current drill pipe deflection angle, the accuracy of angle measurement is greatly improved, laying a solid foundation for subsequent precise correction. In complex operating environments, the actual state of the drill pipe can be more accurately obtained, reducing correction errors caused by angle measurement errors. The BP neural network model outputs a dynamic correction threshold based on the input correction signal through nonlinear mapping of the hidden layer. Because historical operation data covers different geological conditions, drill pipe lengths, and drilling depths, reasonable correction thresholds can be given for different operating scenarios to avoid problems such as excessive or insufficient correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Schematic diagram of the method flow of the present invention;

[0028] Figure 2 Schematic diagram of the nervous system control of the present invention. DETAILED DESCRIPTION

[0029] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0030] See also Figure 1 —2. The present invention provides a technical solution: an angle correction method for an intelligent rock drilling robot based on artificial intelligence, comprising the following steps:

[0031] Step 1: Use the laser ranging module to collect real-time distance data between the drill rod of the rock drilling robot and the target rock surface, calculate the current deflection angle θc of the drill rod based on the distance data, and generate a correction signal containing the angle deviation value;

[0032] When collecting the distance data between the drill rod and the target rock surface through the laser ranging module, in addition to using pulsed laser ranging technology, an echo intensity analysis function is added. Rocks of different textures have different reflection intensities for lasers. By analyzing the echo intensity and combining it with the distance data, the rock characteristics can be preliminarily judged to provide more information for subsequent decision-making. When encountering hard rock, the drill rod is more likely to deflect, and the system can adjust the correction strategy in advance accordingly. In addition, to further improve the accuracy of angle calculation, when collecting distance data, the collection frequency is increased from 10 times per second to 50 times per second. The large amount of collected data is processed in real time using a sliding window algorithm to remove outliers and perform smoothing, so that the calculated current deflection angle more accurately reflects the actual status of the drill rod;

[0033] Step 2: Input the correction signal into the pre-trained BP neural network model. The hidden layer of the BP neural network performs nonlinear mapping on the input signal and outputs the dynamic correction threshold Δθ. The BP neural network is trained based on the mapping relationship between the deflection angle and the optimal correction threshold in historical operation data.

[0034] BP neural network model can be referenced Figure 2 In the above figure, θd1 and θd2 are the expected values of the joints, θ1 and θ2 are the actual values; e1 and e2 are errors, ec1 and ec2 are the error change rates; t1 and t2 are the torques acting on the joints. The neural network replaces the original fuzzy control part as the controller of the system.

[0035] In addition to the correction signal containing the angular deviation value, the BP neural network model's input layer also includes collected rock property data (such as hardness and density) and real-time drill pipe vibration data as additional input features. This information helps the model gain a more comprehensive understanding of the operating environment and drill pipe status, thereby outputting dynamic correction thresholds that better meet actual needs.

[0036] When training BP neural networks, the mini-batch gradient descent method is used instead of the traditional gradient descent method. Mini-batch gradient descent uses a small batch of data each time to calculate the gradient and update the weights, which not only reduces the amount of calculation but also avoids falling into local optimal solutions. At the same time, an early stopping mechanism is introduced. During the training process, a portion of the data is used as a validation set. When the model error on the validation set no longer decreases, training is stopped to prevent overfitting and improve the model's generalization ability.

[0037] Step 3: Compare the current deflection angle θc with the dynamic correction threshold Δθ. If |θc| ≥ Δθ, generate a control instruction containing the correction direction and force. The force of the control instruction is directly proportional to |θc - Δθ|.

[0038] Step 4: The hydraulic servo mechanism executes the control command to drive the drill rod to adjust in the correction direction at the preset angular velocity ω. At the same time, the inertial measurement unit provides real-time feedback on the adjusted deflection angle θ'c until |θ'c| < Δθ is satisfied, forming a closed-loop control.

[0039] When the drill rod is driven in the correcting direction at a preset angular velocity ω, the preset angular velocity is dynamically adjusted based on the drill rod's current deflection angle and the deviation from the target angle. When the deviation is large, the angular velocity is appropriately increased to speed up the correction. When approaching the target angle, the angular velocity is reduced to avoid excessive correction due to inertia, achieving more precise correction control.

[0040] When generating control commands, the elastic deformation of the drill pipe during the correction process is taken into account. By establishing a drill pipe elasticity model and combining parameters such as drill pipe material and length, the elastic deformation of the drill pipe under different correction forces is calculated. This allows the control commands to be modified, bringing the actual angle after correction closer to the desired target and improving correction accuracy.

[0041] The laser ranging module uses pulsed laser ranging technology to calculate the distance between the drill rod and the target rock surface by measuring the time difference between the emission and reception of the laser pulse. The measurement accuracy can reach ±1mm.

[0042] When calculating the current deflection angle θc of the drill pipe, a fitting calculation is performed through trigonometric function relationships combined with distance data of at least three different positions collected by the laser ranging module to improve the accuracy of the angle calculation.

[0043] The BP neural network model consists of an input layer, at least two hidden layers and an output layer. The number of input layer nodes is determined according to the distance data characteristics, the number of hidden layer nodes is determined by trial and error and empirical formulas, and the number of output layer nodes is 1, which is used to output the dynamic correction threshold Δθ.

[0044] During the training process of the BP neural network model, historical operation data includes the mapping relationship between the deflection angle and the optimal correction threshold under different geological conditions, different drill rod lengths, and different drilling depths, and the training data is normalized to accelerate the training convergence speed.

[0045] When comparing the current deflection angle θc with the dynamic correction threshold Δθ, if |θc| ≥ Δθ, the correction direction is determined by judging the positive or negative relationship between θc and Δθ. When θc is positive and greater than Δθ, the correction direction is negative adjustment, otherwise it is positive adjustment.

[0046] The hydraulic servo mechanism includes a hydraulic pump, a hydraulic cylinder and an electromagnetic reversing valve. The electromagnetic reversing valve switches the direction of the oil circuit according to the control instruction, so that the hydraulic cylinder drives the drill rod to adjust in the correction direction at a preset angular velocity ω. The preset angular velocity ω is determined according to the material, length and working environment of the drill rod, and the range is 0.1-1rad / s.

[0047] An artificial intelligence-based intelligent rock drilling robot angle correction system, comprising:

[0048] Data acquisition module: including laser ranging unit and inertial measurement unit. The laser ranging unit is used to obtain the distance data between the drill rod and the rock surface in real time, and the inertial measurement unit is used to detect the spatial attitude angle of the drill rod.

[0049] Decision control module: includes an embedded processor and a BP neural network algorithm stored in the processor, and the embedded processor is configured to execute:

[0050] Calculate the current deflection angle θc based on the distance data;

[0051] Output the dynamic correction threshold Δθ through BP neural network;

[0052] Generate control instructions proportional to |θc-Δθ|;

[0053] Actuator module: includes a hydraulic servo drive and a correction actuator connected to the drive. The hydraulic servo drive drives the correction actuator to adjust the angle according to the control instructions;

[0054] Feedback module: includes a gyroscope installed on the drill pipe base, which is used to monitor the adjusted deflection angle θ′c in real time and feed it back to the decision control module to form a closed-loop control system.

[0055] During the closed-loop control process, increase the monitoring and adjustment of the system response time. Record in real time the time from issuing the control command to the feedback of the adjusted deflection angle by the inertial measurement unit. If the response time is too long, analyze whether it is caused by data transmission delay, slow response of the actuator, or other reasons, and take corresponding measures. For example, optimize the data transmission protocol to reduce delays, perform maintenance on the actuator to improve the response speed, and ensure the timeliness and effectiveness of closed-loop control. At the same time, establish a historical correction data database to record detailed information on each correction, including the initial angle deviation, various parameters during the correction process, and the final correction results. Through the analysis of historical data, summarize the best correction strategies under different operating conditions, provide reference for subsequent operations, and continuously optimize the correction methods.

[0056] The laser transmitter used in the laser ranging unit emits a laser beam with a wavelength of 905nm, and the laser receiver uses an avalanche photodiode to improve the sensitivity and anti-interference ability of distance measurement.

[0057] The embedded processor is an ARM architecture microprocessor, which integrates a high-speed computing unit and a large-capacity storage unit. It can quickly process data and run the BP neural network algorithm. The storage unit is used to store historical operation data and trained BP neural network model parameters.

Claims

1. An angle correction method for an intelligent rock drilling robot based on artificial intelligence, characterized in that: The following steps are involved: Step 1: Use the laser ranging module to collect the distance data between the drill rod of the rock drilling robot and the target rock surface in real time, and calculate the current deflection angle θ of the drill rod based on the distance data c , and generate a correction signal including the angle deviation value; Step 2: Input the correction signal into a pre-trained BP neural network model, perform nonlinear mapping on the input signal through the hidden layer of the BP neural network, and output a dynamic correction threshold Δθ, wherein the BP neural network is trained by the mapping relationship between the deflection angle and the optimal correction threshold in the historical operation data; Step 3: Compare the current deflection angle θ c and the dynamic correction threshold Δθ, if |θ c |≥Δθ, a control instruction including the direction and strength of the correction is generated. The strength of the control instruction is related to |θ c -Δθ| is in direct proportion; Step 4: The hydraulic servo mechanism executes the control command to drive the drill rod to adjust in the correction direction at a preset angular velocity ω, and the inertial measurement unit provides real-time feedback on the adjusted deflection angle θ. c , until |θ' is satisfied c |<Δθ, forming a closed-loop control.

2. An artificial intelligence-based intelligent rock drilling robot angle correction system, used to implement the method of claim 1, characterized in that: include: Data acquisition module: including a laser ranging unit and an inertial measurement unit. The laser ranging unit is used to obtain the distance data between the drill rod and the rock surface in real time, and the inertial measurement unit is used to detect the spatial attitude angle of the drill rod; Decision control module: includes an embedded processor and a BP neural network algorithm stored in the processor, wherein the embedded processor is configured to execute: Calculate the current deflection angle θ based on the distance data c ; Output the dynamic correction threshold Δθ through BP neural network; Generate and |θ c -Δθ| proportional control command; Actuator module: includes a hydraulic servo driver and a correction actuator connected to the driver, wherein the hydraulic servo driver drives the correction actuator to adjust the angle according to the control instructions; Feedback module: includes a gyroscope installed on the drill pipe base, which is used to monitor the adjusted deflection angle θ' in real time c And feed back to the decision control module to form a closed-loop control system.

3. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: The laser ranging module adopts pulse laser ranging technology to calculate the distance between the drill rod and the target rock surface by measuring the time difference between the emission and reception of the laser pulse, with a measurement accuracy of ±1mm.

4. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: When calculating the current deflection angle θ of the drill pipe c When calculating the angle, the trigonometric function relationship is used to combine the distance data of at least three different positions collected by the laser ranging module to perform fitting calculation to improve the accuracy of the angle calculation.

5. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: The BP neural network model includes an input layer, at least two hidden layers and an output layer. The number of input layer nodes is determined according to the distance data characteristics, the number of hidden layer nodes is determined by trial and error and empirical formulas, and the number of output layer nodes is 1, which is used to output the dynamic correction threshold Δθ.

6. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: During the training process of the BP neural network model, historical operation data includes the mapping relationship between the deflection angle and the optimal correction threshold under different geological conditions, different drill rod lengths and different drilling depths, and the training data is normalized to accelerate the training convergence speed.

7. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: Comparing the current deflection angle θ c When the dynamic correction threshold Δθ is used, if |θ c |≥Δθ, by judging θ c The positive and negative relationship with Δθ determines the direction of correction. c When it is positive and greater than Δθ, the correction direction is negative adjustment, otherwise it is positive adjustment.

8. The angle correction method of an intelligent rock drilling robot based on artificial intelligence according to claim 1, characterized in that: The hydraulic servo mechanism includes a hydraulic pump, a hydraulic cylinder and an electromagnetic reversing valve. The electromagnetic reversing valve switches the direction of the oil circuit according to the control instruction, so that the hydraulic cylinder drives the drill rod to adjust in the correction direction at a preset angular velocity ω. The preset angular velocity ω is determined according to the material, length and working environment of the drill rod and ranges from 0.1 to 1 rad / s.

9. The angle correction system for an intelligent rock drilling robot based on artificial intelligence according to claim 2, characterized in that: The laser transmitter used in the laser distance measurement unit emits a laser beam with a wavelength of 905nm, and the laser receiver uses an avalanche photodiode to improve the sensitivity and anti-interference ability of distance measurement.

10. The angle correction system for an intelligent rock drilling robot based on artificial intelligence according to claim 2, characterized in that: The embedded processor is an ARM architecture microprocessor, which integrates a high-speed computing unit and a large-capacity storage unit. It can quickly process data and run the BP neural network algorithm. The storage unit is used to store historical operation data and trained BP neural network model parameters.

Images