Reliability diagnosis method for heading machine cutting arm pose detection system

Abstract

The invention belongs to the technical field of automatic tunneling, and provides a reliability diagnosis method for a position and posture detection system of a cutting arm of a tunneling machine in order to solve the problem that the position and posture detection precision and reliability of the cutting arm of the tunneling machine under different working conditions are difficult to meet actual working requirements. Processing the pose data by using a deep learning algorithm to obtain fused pose data; an initial fault data set is established, multi-level optimization of alarm thresholds is carried out based on the running state of the heading machine, the initial fault data set and the fusion pose data, and the optimized fault data set is stored; and associating the fused pose data with the fault data set, determining pose data and fault data which are consistent in use time, then establishing a cutting arm pose data analysis model based on the pose data and the use time, and performing reliability evaluation on the heading machine cutting arm pose detection system based on the data analysis model. The multi-stage fault alarm during the pose detection is realized, and the pose detection accuracy and reliability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of automated tunneling, and particularly relates to a reliability diagnosis method for a pose detection system of a roadheader cutting arm. Background Art

[0002] In the construction of modern intelligent coal mines, it is a major challenge to achieve automatic cutting by a roadheader. The pose detection technology of the roadheader cutting arm is the key to realizing automatic cutting. Due to the complex and changeable environment of the tunneling face, the roadheader needs to adjust the pose of the cutting arm in real time during operation to adapt to different geological conditions. Traditional pose detection methods often rely on a single sensor or measurement means, and it is difficult to meet the requirements of high reliability. Summary of the Invention

[0003] In order to solve at least one of the above technical problems existing in the prior art, the present invention provides a reliability diagnosis method for a pose detection system of a roadheader cutting arm.

[0004] The present invention is implemented by adopting the following technical solutions: A reliability diagnosis method for a pose detection system of a roadheader cutting arm includes the following steps:

[0005] S1: Obtain the pose data of the cutting arm, including at least position data, angle data, and speed data;

[0006] S2: Process the pose data by using a deep learning algorithm to obtain fused pose data;

[0007] S3: Establish an initial fault data set, and perform multi-level optimization of the alarm threshold based on the operating state of the roadheader, the initial fault data set, and the fused pose data, and save the optimized fault data set;

[0008] S4: Correlate the fused pose data with the fault data set to determine the pose data with the same usage time, and then establish a cutting arm pose data analysis model based on the pose data and the usage time, and perform reliability evaluation on the pose detection system of the roadheader cutting arm based on the data analysis model.

[0009] Preferably, step S3 specifically includes:

[0010] Set an initial fault alarm threshold based on the initial fault data set;

[0011] Perform the first-level optimization of the alarm threshold based on the fused pose data and determine the first-level alarm threshold;

[0012] Dynamically adjust the alarm threshold according to the operating state of the roadheader, perform the second-level optimization, and determine the second-level alarm threshold.

[0013] Preferably, it further includes:

[0014] Respectively conduct simulation experiments on the alarm thresholds after the first-level optimization and the second-level optimization, determine that the optimized alarm thresholds meet the operation requirements of the roadheader. When the operation requirements of the roadheader are not met, perform a third-level optimization on the corresponding alarm thresholds and determine the third-level alarm thresholds.

[0015] Preferably, step S2 specifically includes:

[0016] Divide the pose data into a training set and a test set;

[0017] Use a deep learning algorithm to construct a pose detection model. At the same time, select a recurrent neural network for model training. Input the training set into the pose detection model, and adjust the model parameters of the pose detection model by the backpropagation method to obtain a trained pose detection model;

[0018] Input the test set into the trained pose detection model, evaluate the performance of the pose detection model through a preset evaluation index, adjust the model hyperparameters according to the evaluation results to obtain a final pose detection model. Finally, input the pose data into the final pose detection model to obtain fused pose data.

[0019] Preferably, it further includes:

[0020] Use the Kalman filter algorithm to remove noise and determine the optimal model parameters.

[0021] Preferably, step S4 specifically includes:

[0022] Establish a cutting arm pose analysis model with the usage time as the independent variable and the pose data as the dependent variable;

[0023] Perform calculations based on the data analysis model to determine the reliability evaluation index data.

[0024] Preferably, it further includes:

[0025] Judge whether the corresponding alarm threshold is reached according to the reliability evaluation index data. When the alarm threshold is reached, send an alarm signal; otherwise, the roadheader cutting arm pose detection system is reliable.

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] The present invention utilizes deep learning algorithms to accurately detect the pose parameters of the cutting arm of a roadheader. Meanwhile, an initial fault dataset is created and multi-level optimization is performed on the initial fault dataset to improve the accuracy and reliability of fault alarms for the pose detection of the cutting arm of the roadheader. Moreover, based on the multi-level alarm thresholds corresponding to the multi-level optimization, the degree of fault of the cutting arm of the roadheader can be judged, and the corresponding staff can be reminded in a timely manner so as to take repair measures in a timely manner, reduce manual intervention, and lower operation risks, which is of great significance for promoting the process of coal mine intelligent construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0029] Figure 1 It is a schematic flowchart of a method for diagnosing the reliability of a pose detection system for the cutting arm of a roadheader provided by an embodiment of the present invention.

[0030] Figure 2 It is a schematic flowchart of a data preprocessing method for a pose detection system for the cutting arm of a roadheader provided by an embodiment of the present invention.

[0031] Figure 3 It is a schematic flowchart of a data analysis process for a pose detection system for the cutting arm of a roadheader provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] Combined with the drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other implementation manners obtained by those of ordinary skill in the art without creative efforts belong to the scope protected by the present invention.

[0033] It should be noted that the structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limited conditions under which the present invention can be implemented. Therefore, they do not have any technical essence. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should fall within the scope covered by the technical content disclosed in the present invention. It should be noted that in this specification, relational terms such as first and second are only used to distinguish one entity from several other entities, and do not necessarily require or imply any actual relationship or order between these entities.

[0034] The present invention provides an embodiment:

[0035] As Figure 1 shown, a schematic flow diagram of a reliability diagnosis method for a pose detection system of a roadheader cutting arm is provided, including the following steps:

[0036] S1: Obtain the pose data of the cutting arm, including at least position data, angle data, and speed data.

[0037] In this embodiment, the pose data of the roadheader cutting arm is continuously recorded by sensors, including key parameters such as position data, angle data, and speed data, and the obtained pose data is preprocessed. The method is as Figure 2 shown. Record the sensor data every 1 ms within 10 ms, compare the timestamp returned by the sensor data with the system time, and retain the data with the same time. Use the quicksort method to sort the 10 data, remove the maximum and minimum values, calculate the weighted average of the remaining data, and use it as the data recorded by the sensor within 10 ms. Repeat this process for data preprocessing to facilitate subsequent reliability evaluation based on the pose data.

[0038] S2: Use a deep learning algorithm to process the pose data to obtain fused pose data.

[0039] Optionally, divide the pose data into a training set and a test set, and construct a pose detection model through a deep learning algorithm. Select a recurrent neural network for model training, input the training set into the pose detection model, and adjust the parameters of the pose detection model through the backpropagation method to obtain a trained model; input the test set into the trained model, evaluate the performance of the model through a preset evaluation index, and adjust the model hyperparameters according to the evaluation results to obtain the final pose detection model; finally, input the pose data into the final pose detection model to obtain the fused pose data. Optionally, use the Kalman filter algorithm to remove noise and determine the optimal model parameters.

[0040] In this embodiment, according to the type and use of the pose data, the position data, angle data, and speed data are divided into a training set and a test set. The training set is used for training the pose detection model, and the test set is used for evaluating the performance of the pose detection model.

[0041] In this embodiment, according to the characteristics of the pose detection of the roadheader cutting arm, a recurrent neural network is selected for training. The training set is input into the pose detection model, and the model parameters of the pose detection model are adjusted through the backpropagation algorithm so that the pose detection model can accurately predict the pose data of the cutting arm.

[0042] In this embodiment, in the trained pose detection model, a test set is imported, and the performance of the pose detection model is evaluated through preset evaluation metrics such as accuracy and recall rate. According to the evaluation results, the hyperparameters of the pose detection model are adjusted. The hyperparameters at least include learning rate, batch size, number of network layers, etc., so as to improve the overall performance of the pose detection model.

[0043] In this embodiment, training is performed based on the training set and the parameters of the Kalman filter algorithm. Then, the parameters of the trained Kalman filter algorithm are passed to the pose detection model to remove noise and determine the model parameters of the optimal pose detection model. Finally, more accurate fused pose data is calculated through the pose detection model.

[0044] S3: Establish an initial fault data set, and perform multi-level optimization of the alarm threshold based on the running state of the roadheader, the initial fault data set, and the fused pose data, and save the optimized fault data set.

[0045] Optionally, set an initial fault alarm threshold based on the initial fault data set; perform the first-level optimization of the alarm threshold on the initial fault alarm threshold based on the fused pose data, and determine the first-level alarm threshold; dynamically adjust the alarm threshold according to the running state of the roadheader, and perform the second-level optimization, and determine the second-level alarm threshold.

[0046] Optionally, it further includes: performing simulation experiments on the alarm thresholds after the first-level optimization and the second-level optimization respectively, determining that the optimized alarm thresholds meet the running requirements of the roadheader. When they do not meet the running requirements of the roadheader, perform the third-level optimization on the corresponding alarm thresholds, and determine the third-level alarm thresholds.

[0047] In this embodiment, by simulating data loss or distortion caused by faults, etc., a corresponding initial fault data set is made. On the premise of ensuring the accuracy of pose detection, the initial fault data set is trained to obtain an initial fault alarm threshold; then, based on the fused pose data and the initial fault alarm threshold, a first-level optimization of the alarm threshold is performed, and a first-level alarm threshold is determined. According to the operating state of the roadheader during actual operation, the alarm threshold is dynamically adjusted for a second-level optimization, and at the same time, a second-level alarm threshold is determined to meet the requirements under different working conditions. The first-level alarm threshold and the second-level alarm threshold are transmitted to the pose detection system, and it is ensured that the pose detection system can accurately identify the corresponding alarm thresholds and can perform hierarchical early warning of faults. Through actual tests or simulation experiments, it is verified whether the optimized first-level alarm threshold and second-level alarm threshold meet the actual operating requirements of the roadheader. When they do not meet the operating requirements of the roadheader, according to the results of actual tests or simulation experiments, a third-level optimization of the first-level alarm threshold or the second-level alarm threshold is performed to determine a third-level alarm threshold. In practical applications, the setting of the alarm threshold should be able to accurately reflect the abnormal changes in the pose of the roadheader cutting arm, and at the same time avoid false alarms and missed alarms. For example, for different requirements of the cutting arm pose due to factors such as the hardness of coal and rock and the shape of the roadway, multi-level optimization of the alarm threshold is required.

[0048] S4: Associate the fused pose data with the fault data set to determine pose data and fault data with consistent usage times, and then establish a regression analysis model based on the pose data and the usage time, and perform reliability evaluation on the roadheader cutting arm pose detection system based on the regression analysis model.

[0049] Optionally, taking the usage time as the independent variable and the pose data as the dependent variable, establish a cutting arm pose analysis model; perform calculations based on the cutting arm pose analysis model to determine reliability evaluation index data.

[0050] Optionally, it further includes: judging whether the corresponding alarm threshold is reached according to the reliability evaluation index data. When the alarm threshold is reached, an alarm signal is issued; otherwise, the roadheader cutting arm pose detection system is reliable.

[0051] In this embodiment, the fused pose data is associated with time. Taking time as the independent variable and pose data as the dependent variable, a cutting arm pose analysis model is established. Through this cutting arm pose analysis model, the change rate of the cutting arm pose data is determined. In this embodiment, by analyzing the angle data collected by the cutting arm lifting inclination sensor, in the first step, it is judged whether the inclination angle is within the mechanical angle range of the cutting arm lifting. If it exceeds the angle threshold range, it is judged that the sensor data fails; in the second step, it is judged whether the angle change direction is consistent with the switching direction of the corresponding lifting solenoid valve. If they are inconsistent, the collected angle data fails.

[0052] In this embodiment, the lifting speed of the cutting arm of the roadheader is limited by the telescopic speed of the lifting hydraulic cylinder. Therefore, the calculation and analysis of the angle change rate of the cutting arm lifting inclination sensor are carried out. The angle change rate can be calculated using the following formula:

[0053]

[0054] Where:

[0055] -δ is the angle change rate of the cutting arm lifting

[0056] -α1 is the lifting angle of the cutting arm at time t1

[0057] -α0 is the lifting angle of the cutting arm at time t0

[0058] In this embodiment, the data analysis process for the angle change rate is as follows Figure 3 , successively judge whether the lifting angle value of the cutting arm is valid from the mechanical angle and the angle change direction. When neither can be judged, obtain the angles α0 and α1 calculated by the cutting arm inclination sensor at times t0 and t1, and at the same time determine the angle change rate δ of the roadheader cutting arm lifting. By comparing the angle change rate δ with the maximum angle change rate δ max , so as to determine whether the calculated angle value of the cutting arm lifting inclination sensor is valid.

[0059] By collecting and preprocessing the pose data during the production process of the roadheader and using a deep learning network for parameter model training and optimization, the present invention can significantly improve the accuracy of pose detection. At the same time, by making a dataset for training when the sensor is missing or distorted, the warning and alarm thresholds are optimized, and a perfect data fusion and reliability evaluation mechanism is established. When a large data offset or system anomaly is detected, a reliability alarm can be quickly triggered, which helps the operator take timely measures to avoid the adverse impact of the failure on the operation of the roadheader.

[0060] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A reliability diagnosis method for a cutting arm posture detection system of a tunnel boring machine, characterized in that: The steps include: S1: Acquire the position data of the cutting arm, including at least position data, angle data, and speed data; S2: Processing the posture data using a deep learning algorithm to obtain fused posture data; S3: establishing an initial fault data set, and performing multi-level optimization of the alarm threshold based on the operation status of the tunnel boring machine, the initial fault data set and the fused posture data, and saving the optimized fault data set; S4: Associating the fused posture data with the fault data set to determine the posture data with consistent usage time, then establishing a cutting arm posture data analysis model based on the posture data and the usage time, and performing reliability evaluation of the cutting arm posture detection system of the tunnel boring machine based on the data analysis model.

2. The reliability diagnosis method of the cutting arm posture detection system of a tunnel boring machine according to claim 1 is characterized in that: Step S3 specifically includes: Setting an initial fault alarm threshold based on the initial fault data set; Performing a first-level optimization of the alarm threshold on the initial fault alarm threshold based on the fused posture data, and determining a first-level alarm threshold; According to the operating status of the tunnel boring machine, the alarm threshold is dynamically adjusted to perform the second-level optimization and determine the second-level alarm threshold.

3. The reliability diagnosis method of the cutting arm posture detection system of the tunnel boring machine according to claim 2 is characterized in that: Also includes: Simulation experiments are carried out on the alarm thresholds after the first and second level optimizations to determine whether the optimized alarm thresholds meet the operation requirements of the tunnel boring machine. When the operation requirements of the tunnel boring machine are not met, the corresponding alarm thresholds are subjected to the third level optimization and the third level alarm thresholds are determined.

4. The reliability diagnosis method of the cutting arm posture detection system of a tunnel boring machine according to claim 1 is characterized in that: Step S2 specifically includes: Dividing the posture data into a training set and a test set; A posture detection model is constructed by using a deep learning algorithm, and a recurrent neural network is selected for model training. The training set is input into the posture detection model, and the model parameters of the posture detection model are adjusted by a back propagation method to obtain a trained posture detection model. The test set is input into the trained pose detection model, the performance of the pose detection model is evaluated by preset evaluation indicators, the model hyperparameters are adjusted according to the evaluation results to obtain the final pose detection model, and finally the pose data is input into the final pose detection model to obtain fused pose data.

5. The reliability diagnosis method of the cutting arm posture detection system of a tunnel boring machine according to claim 4 is characterized in that: Also includes: The Kalman filter algorithm is used to remove noise and determine the optimal model parameters.

6. The reliability diagnosis method of the cutting arm posture detection system of a tunnel boring machine according to claim 1, characterized in that: Step S4 specifically includes: Taking the usage time as the independent variable and the posture data as the dependent variable, a cutting arm posture analysis model is established; Calculation is performed based on the cutting arm posture analysis model to determine reliability evaluation index data.

7. The reliability diagnosis method of the cutting arm posture detection system of a tunnel boring machine according to claim 6, characterized in that: Also includes: It is determined whether the corresponding alarm threshold is reached according to the reliability evaluation index data. When the alarm threshold is reached, an alarm signal is issued. Otherwise, the cutting arm posture detection system of the tunnel boring machine is reliable.

Images