Furnace front robot drill rod angle automatic compensation control method

Abstract

The invention provides a stokehole robot drill rod angle automatic compensation control method, and relates to the technical field of stokehole robots, and the method comprises the steps: calculating a theoretical lifting angle of a drill rod according to a target position; gradually extending the drill rod to the target position, and acquiring vibration data of the drill rod in the process of gradually extending the drill rod; predicting the real-time rod length and deformation quantity of the drill rod based on the vibration data and a pre-trained prediction model M; the deformation quantity of the drill rod is converted into the required angle compensation quantity; the actual required angle of the drill rod is obtained based on the angle compensation amount and the theoretical lifting angle; and the drill rod reaches the actually required angle. A mapping relation is established between vibration data and deflection deformation, limitation of a traditional rigid body model is broken through, a traditional artificial experience compensation mode is replaced, the requirement for experience of an operator is reduced, and it can be ensured that drilling or furnace hole plugging is conducted at the target position at the optimal operation time.

Description

Technical Field

[0001] The invention relates to the technical field of furnace-front robots, and in particular to an automatic compensation control method for a drill rod angle of a furnace-front robot. Background Art

[0002] A furnace-front robot primarily consists of a robot body, a main arm, a drill rod ejection mechanism, a drill rod, and a rotating mechanism. The drill rod, driven by the main arm drill rod ejection mechanism, retracts and retracts from the main arm. Once extended, it is used to seal furnace holes or drill holes. When the drill rod is extended to drill a hole at the target location, the weight of the drill bit is concentrated at the distal end, and the drill rod is relatively long, causing deflection.

[0003] Currently, when calculating its forward path, the furnace-front robot treats the drill rod as a rigid body (no deflection or deformation occurs) and only performs calculations along the rod's length, without considering the impact of its deflection or deformation. Therefore, experienced operators are required to operate the furnace-front robot. However, experience is acquired in actual combat and is not inherited. New operators need a lot of practice to summarize their own experience, which often results in missing the best operating opportunity (or damaging the drill rod, or failing to align with the target position). Summary of the Invention

[0004] In response to the above situation, the present invention provides an automatic compensation control method for the drill rod angle of a furnace-front robot, aiming to solve the technical problem that the current furnace-front robot regards the drill rod as a rigid body when calculating its forward path, and only calculates along the length direction of the rod without considering the impact of its deformation.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a method for automatically compensating the drill rod angle of a furnace-front robot, comprising:

[0007] Step S1: Calculate the theoretical lifting angle of the drill rod according to the target position ;

[0008] Step S2: gradually extending the drill rod to the target position, and obtaining vibration data of the drill rod during the process of gradually extending the drill rod; based on the vibration data and a pre-trained prediction model M, predicting the real-time rod length l_pred and deformation δ_pred of the drill rod;

[0009] Step S3, converting the deformation amount δ_pred of the drill rod into the required angle compensation amount Δθ;

[0010] Step S4: Based on the angle compensation Δθ and the theoretical lifting angle , get the actual required angle θ_actual of the drill rod;

[0011] Step S5, making the drill rod reach the actual required angle θ_actual;

[0012] Step S6: When the drill rod is extended to reach the target position, repeat steps S2 to S5 multiple times until the positioning is completed.

[0013] In some embodiments of the present invention, in step S2, the method for obtaining the prediction model M includes:

[0014] Step S21, collecting vibration data corresponding to the drill rod at multiple extension lengths;

[0015] Step S22: performing FFT transformation on the vibration data;

[0016] Step S23: Build and train the prediction model M.

[0017] In some embodiments of the present invention, step S21 includes:

[0018] Step S211, controlling the drill rod to gradually extend at intervals of 10 cm;

[0019] Step S212: Every time the drill rod is extended by 10 cm, the original vibration signal of the drill rod at the corresponding rod length is collected.

[0020] In some embodiments of the present invention, step S22 includes:

[0021] Step S221: Perform FFT transformation on the original vibration signal to obtain a vibration spectrum, and extract the first five resonance frequencies based on the vibration spectrum. 、 、 、 、 , calculate the amplitude characteristic RMS value A_rms and peak-to-peak value A_pp;

[0022] Step S222: construct input vector X=[ , , , A_rms, A_pp];

[0023] Step S223: Construct an output vector Y=[l_pred, δ_pred].

[0024] In some embodiments of the present invention, the prediction model M in step S23 adopts a Stacking architecture.

[0025] In some embodiments of the present invention, the training strategy of the prediction model M is:

[0026] Data augmentation: Add ±5% Gaussian noise to the original vibration features to generate a 3x expanded dataset;

[0027] Dataset split: split the 3-fold expanded dataset into training and validation sets, training set: validation set = 8:2;

[0028] Early stopping mechanism: terminate training when the validation set loss does not decrease for 10 consecutive rounds;

[0029] Loss function: L = 0.7 MAE(l_pred) + 0.3 MAE(δ_pred).

[0030] In some embodiments of the application, in step S3, the calculation formula of the angle compensation amount Δθ is: Δθ = arctan( (δ_pred + k·v) / l_pred ) ;

[0031] Wherein, v is the real-time advancing speed of the drill rod; k is the dynamic compensation coefficient.

[0032] In some embodiments of the application, in step S5, the actual angle of the drill rod is made to reach the θ_actual±0.5° tolerance range.

[0033] In some embodiments of the application, in step S4, the calculation formula of the actual demand angle θ_actual is: θ_actual = + Δθ.

[0034] In some embodiments of the application, in step S6, when repeating steps S2-S5 multiple times, a plurality of time-sequenced deformation amounts δ_pred are obtained, and when a plurality of continuous deformation amounts δ_pred < a preset threshold value δ_th, it is determined that the angle compensation of the drill rod is completed.

[0035] The embodiments of the application have at least the following advantages or beneficial effects:

[0036] The application establishes a mapping relationship between vibration data and deflection deformation, breaks through the limitation of the traditional rigid body model, replaces the traditional artificial experience compensation method, reduces the experience requirement of the operator, and can ensure drilling or plugging the furnace eye at the target position at the best operation opportunity.

[0037] Other features and advantages of the application will be described in the following description. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0039] Figure 1 A flow chart of the furnace robot drill rod angle automatic compensation control method is shown in FIG. 1.

[0040] Figure 2 An architecture schematic diagram of the prediction model M is shown in FIG. 2. DETAILED DESCRIPTION

[0041] In the following, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the embodiments of the present application.

[0042] The embodiments of the present application are described in detail below.

[0043] Embodiment 1

[0044] Referring to FIG. 1, Figure 1-Figure 2 the present embodiment provides a furnace robot drill rod angle automatic compensation control method, comprising the following steps:

[0045] Step S1, calculating a theoretical lifting angle of the drill rod according to a target position (x_t, y_t) , x_t is the horizontal coordinate of the target position, and y_t is the vertical coordinate;

[0046] Step S2, gradually extending the drill rod to the target position, and acquiring vibration data of the drill rod in the process of gradually extending the drill rod; based on the vibration data and a pre-trained prediction model M, predicting a real-time rod length l_pred and a deformation amount δ_pred of the drill rod;

[0047] Step S3, converting the deformation amount δ_pred of the drill rod into a required angle compensation amount Δθ;

[0048] Step S4, obtaining an actual required angle θ_actual of the drill rod based on the angle compensation amount Δθ and the theoretical lifting angle

[0049] Step S5, making the drill rod reach the actual required angle θ_actual;

[0050] Step S6, repeatedly performing steps S2-S5 multiple times in the process of extending the drill rod to the target position until positioning is completed (the drill rod reaches the target position).

[0051] ​In practical application, the corresponding deformation and the length of the drill rod can be roughly inferred from the vibration data of the drill rod, and then the lifting angle of the main arm is calculated from the deformation and the length of the drill rod, so as to accurately position and facilitate drilling or sealing the furnace eye. The application establishes a mapping relationship between vibration data and deflection deformation, breaks through the limitation of the traditional rigid body model (the drill rod is regarded as a rigid body when the furnace robot calculates its forward path), and replaces the traditional artificial experience compensation method, thereby reducing the experience requirement of the operator and ensuring drilling or sealing the furnace eye at the target position at the best operation time.

[0052] In step S2, when acquiring the vibration data of the drill rod, a three-axis vibration sensor for collecting vibration data of the drill rod is installed at the base of the drill rod spitting device or other parts, and the sampling frequency of the three-axis vibration sensor is set to 8 kHz.

[0053] In step S2, the method for acquiring the prediction model M includes:

[0054] Step S21, collect vibration data corresponding to a plurality of extension lengths of the drill rod;

[0055] Step S22, perform FFT transformation (feature engineering processing) on the vibration data;

[0056] Step S23, construct and train the prediction model M.

[0057] Step S21 includes:

[0058] Step S211, extend the length of the drill rod from 0.5 m to 5.0 m finally, and control the drill rod to be gradually extended at an interval of 10 cm;

[0059] Step S212, collect the vibration original signal of the drill rod under the corresponding rod length every 10 cm of the drill rod extension, so as to obtain the vibration data.

[0060] Step S22 includes:

[0061] Step S221, perform FFT transformation on the vibration original signal to obtain a vibration frequency spectrum, and based on the vibration frequency spectrum, extract the first five order resonance frequencies 、 、 、 、 , calculate the amplitude characteristic RMS value A_rms and the peak-to-peak value A_pp;

[0062] Step S222, construct an input vector X=[ , , , A_rms, A_pp];

[0063] Step S223, construct output vector Y=[l_pred, δ_pred] (output rod length and deformation variable).

[0064] The prediction model M in step S23 adopts a Stacking architecture, which includes a base model layer and a meta model layer, wherein:

[0065] The base model layer includes: Random Forest: (n_estimators=200, max_depth=8), XGBoost: (learning_rate=0.1, max_depth=6); 1D-CNN (3 layers of convolution kernel, kernel size=5);

[0066] The meta model layer includes: Ridge Regression (alpha=0.5).

[0067] In step S23, the prediction model M is trained offline by historical experimental data, and the training strategy of the prediction model M is:

[0068] Data augmentation: add ±5% Gaussian noise to the original vibration features (resonant frequency and amplitude) to generate a 3-fold expanded data set to improve the robustness of the industrial noise environment;

[0069] Data set division: divide the 3-fold expanded data set into a training set and a validation set, training set: validation set=8:2;

[0070] Early stopping mechanism: terminate training when the validation set loss does not decrease for 10 consecutive rounds (patience=10 epochs); through the early stopping mechanism, overfitting is avoided;

[0071] Loss function: L = 0.7 MAE(l_pred) + 0.3 MAE(δ_pred), wherein l_pred is the rod length and δ_pred is the deformation variable. Through the loss function, the prediction of the rod length and the deformation variable is jointly optimized.

[0072] When using the above prediction model M for online prediction, the vibration data of the drill rod is acquired in real time → FFT transformation → construct X vector → input the prediction model M → output the prediction result (l_pred, δ_pred).

[0073] In step S3, the calculation formula of the angle compensation amount Δθ is: Δθ =arctan( (δ_pred + k·v) / l_pred ), wherein δ_pred is the deformation variable; l_pred is the rod length; v is the real-time advancing (extending) speed of the drill rod (unit: mm / s); k is the dynamic compensation coefficient, which is calibrated through experiments, k=0.15.

[0074] The calculation formula of the above angle compensation Δθ introduces the velocity term k·v, which effectively compensates for the hysteresis error caused by motion inertia.

[0075] In step S4, the actual required angle θ_actual is calculated as follows: θ_actual = + Δθ, where is the theoretical lifting angle of the drill rod.

[0076] In step S5 , the drill rod is lifted by a hydraulic cylinder for driving the main arm so that the actual angle of the drill rod reaches a tolerance range of θ_actual±0.5°.

[0077] In step S6, when steps S2 to S5 are repeated multiple times, multiple deformation variables δ_pred arranged in time sequence are obtained. When multiple (3) consecutive deformation variables δ_pred are less than the preset threshold δ_th, it is determined that the angle compensation of the drill rod is completed (the deformation variable is small and no further compensation is required).

[0078] Example 2

[0079] This embodiment is improved on the basis of embodiment 1.

[0080] See also Figure 1-Figure 2 In step S2, a displacement sensor is installed between the main arm drill ejection device and the drill rod to directly measure the drill rod extension length l_meas. If |l_pred - l_meas| > the allowable error, an alarm is triggered and the operation is suspended (suspending subsequent steps S3-S6). The allowable error is a predetermined value. This enables this embodiment to have both a prediction mechanism and a safety verification mechanism (prediction + verification), ensuring system robustness in the event of a triaxial vibration sensor failure. In other words, it ensures that the system can still be safely interrupted in the event of a triaxial vibration sensor failure.

[0081] Finally, it should be noted that the above are merely preferred embodiments of the present application and are not intended to limit the present application. Persons skilled in the art will readily appreciate that the present application is susceptible to various modifications and variations. The embodiments and features of the embodiments may be combined arbitrarily without conflict. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for automatically compensating the drill rod angle of a furnace robot, characterized in that: include: Step S1: Calculate the theoretical lifting angle of the drill rod according to the target position ; Step S2: gradually extending the drill rod to the target position, and obtaining vibration data of the drill rod during the process of gradually extending the drill rod; based on the vibration data and a pre-trained prediction model M, predicting the real-time rod length l_pred and deformation δ_pred of the drill rod; Step S3, converting the deformation amount δ_pred of the drill rod into the required angle compensation amount Δθ; Step S4: Based on the angle compensation Δθ and the theoretical lifting angle , get the actual required angle θ_actual of the drill rod; Step S5, making the drill rod reach the actual required angle θ_actual; Step S6: When the drill rod is extended to reach the target position, repeat steps S2 to S5 multiple times until the positioning is completed.

2. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 1 is characterized in that: In step S2, the method for obtaining the prediction model M includes: Step S21, collecting vibration data corresponding to the drill rod at multiple extension lengths; Step S22: performing FFT transformation on the vibration data; Step S23: Build and train the prediction model M.

3. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 2 is characterized in that: Step S21 includes: Step S211, controlling the drill rod to gradually extend at intervals of 10 cm; Step S212: Every time the drill rod is extended by 10 cm, the original vibration signal of the drill rod at the corresponding rod length is collected.

4. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 3 is characterized in that: Step S22 includes: Step S221: Perform FFT transformation on the original vibration signal to obtain a vibration spectrum, and extract the first five resonance frequencies based on the vibration spectrum. 、 、 、 、 , calculate the amplitude characteristic RMS value A_rms and peak-to-peak value A_pp; Step S222: construct input vector X=[ , , , A_rms, A_pp]; Step S223: Construct an output vector Y=[l_pred, δ_pred].

5. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 2 is characterized in that: The prediction model M in step S23 adopts a Stacking architecture.

6. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 5 is characterized in that: Training strategy for prediction model M: Data augmentation: Add ±5% Gaussian noise to the original vibration features to generate a 3x expanded dataset; Dataset division: The 3-fold expanded dataset is divided into a training set and a validation set, with a training set: validation set ratio of 8:

2. Early stopping mechanism: training is terminated when the validation set loss does not decrease for 10 consecutive rounds; Loss function: L = 0.7 MAE(l_pred) + 0.3 MAE(δ_pred).

7. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 1 is characterized in that: In step S3, the angle compensation amount Δθ is calculated as follows: Δθ = arctan( (δ_pred + k·v) / l_pred ); Among them, v is the real-time advancement speed of the drill rod; k is the dynamic compensation coefficient.

8. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 7, characterized in that: In step S4, the actual required angle θ_actual is calculated as follows: θ_actual = + Δθ.

9. The automatic compensation control method for the drill rod angle of the furnace robot according to claim 8, characterized in that: In step S5 , the actual angle of the drill rod is made to reach a tolerance range of θ_actual±0.5°.

10. The automatic compensation control method for the drill rod angle of a furnace robot according to any one of claims 1 to 9, characterized in that: In step S6, when steps S2 to S5 are repeated multiple times, a plurality of deformation variables δ_pred arranged in time sequence are obtained. When a plurality of continuous deformation variables δ_pred are less than a preset threshold value δ_th, it is determined that the angle compensation of the drill rod is completed.

Images