Deep hole drilling status monitoring and analysis

Deep hole drilling is carried out in a closed state, and the cutting condition of the tool cannot be directly observed. At present, we can only rely on experience to judge whether the drilling process is normal by listening to the sound, looking at the chips, observing the pressure gauge and touching the vibration of the drill rod. In addition to normal wear, the deep hole drill bit is prone to vibration and the chips are prone to blockage due to the large aspect ratio of the drilled hole, poor rigidity of the process system, and the deep hole drill is often damaged randomly. Since the deep hole drilling process is very unstable, a deep hole drilling monitoring system is very necessary to monitor the state and related information of the drilling process in real time, perform signal processing in real time, identify abnormal states in the processing process, and take measures such as changing the drilling amount or replacing the tool in time to eliminate the abnormal state. If necessary, an alarm will be displayed to avoid accidents and huge economic losses, thereby greatly improving the tool durability and ensuring the normal progress of deep hole drilling.

Due to the complexity of deep hole processing conditions and the lack of in-depth understanding of the physical sources that cause these state changes, there is currently no more complete method and system for monitoring the deep hole drilling process. Traditional single-factor monitoring and single-factor model analysis methods have great limitations. The influence of random factors is too great, and it is difficult to correctly reflect the real situation of the system.

This paper mainly studies the monitoring scheme and signal analysis of deep hole drilling process. A multi-sensor deep hole drilling force and drilling vibration signal acquisition device is designed. The multi-information fusion technology is used to analyze and process the various signals collected during the deep hole drilling process to obtain accurate judgment samples, providing a criterion for the monitoring and control of the drilling process. The cutting force of

the monitoring scheme and device

contains a lot of information in the cutting process. When almost all cutting faults occur, the cutting force signal changes accordingly. However, the characteristic signals shown by different cutting processing methods are also different. Judging tool wear or breakage based on this single criterion alone will cause certain false alarms. Therefore, according to the characteristics of the deep hole drilling system, this study designs a multi-sensor force and vibration measurement device for deep hole drilling, which has the advantages of strong versatility, fast dynamic response, and easy installation and use.

Figure 1 Schematic diagram of multi-sensor force and vibration measurement device

The principle diagram of the multi-sensor force and vibration measuring device is shown in Figure 1: the sensor device is equipped with a strain sensor and an accelerometer, which is connected to the tailstock of the deep hole drilling machine; the sensor device is universal and suitable for all drill rods of different diameters (that is, suitable for all holes of different diameters); the hoop is a clamp fastened to the drill rod, and its function is to transmit the axial force and torque on the drill rod to the sensor device. The hoop is not universal and must be matched with the drill rod. The deep hole drilling force and vibration signals collected by the sensor device are transmitted to the computer through the data acquisition device for analysis and processing to obtain correct monitoring results, which can be used for real-time control of the drilling process.

Experimental research

This experimental study collected two working condition signals of normal and failure during deep hole processing. Drill bits with different degrees of wear and cutting parameters are used to obtain signals of different experimental states and the degree of influence on the signals. The data acquisition device uses 4 channels to collect force, torque, vertical and horizontal vibration signals respectively.

Data acquisition under normal drilling conditions:

the change law of tool status signal when the tool is initially worn and normally worn;
the change law of tool status signal when the speed is changed;
the change law of tool status signal when the feed rate is changed.

Data acquisition under simulated fault conditions :

the change law of tool status signal when the tool is rapidly worn and damaged;
the change law of tool status signal when the chip is blocked;
the change law of tool status signal when the guide block is broken.

Signal analysis:

Since the measurement signal not only contains useful information reflecting the working state of the object, but also contains a lot of useless background noise, the useful information reflecting the working state is often submerged in the useless background noise, and it is generally difficult to directly discover and extract it. In the time domain waveform analysis of the four signals of force, torque, vertical vibration and horizontal vibration, it is found that the original signal is not very different in the time domain and cannot be used for online monitoring of the drill bit. For this reason, the original signal should be processed.

When the frequency domain signal analysis of the four signals is performed, it is found that the force and torque signals have no obvious spectral peaks in the frequency domain, and there is no obvious change law. Only when the tool is damaged does it have a characteristic frequency, so time series analysis is applied to the force and torque signals. The AR model is used to estimate its parameters and obtain the residual equation of the force time series model.

The vibration signal changes significantly in the frequency domain signal. The spectrum analysis of the vibration signal in the horizontal and vertical directions is shown in Figure 2.

Figure 2 Power spectrum analysis of low-frequency vibration signal

The following conclusions can be drawn from the above analysis:

The characteristic of the frequency spectrum change of the power spectrum density of the vibration signal in the low-frequency band is that as the tool wear increases, the main peak amplitude increases rapidly and then tends to be gentle, while the main peak frequency position moves from high frequency to low frequency. When the tool is rapidly worn and close to damage, the main peak frequency in the horizontal direction is concentrated at about 400Hz, and many new spectrum peaks appear; the main peak frequency in the vertical direction is concentrated at about 210Hz, and many new spectrum peaks also appear. This change reflects the generation and development of tool wear, which stimulates the harmonic components of the tool-workpiece-machine tool processing system through cutting force, resulting in changes in the vibration modal parameters of the system. The appearance of multiple spectrum peaks disperses the signal energy, so the spectrum peak amplitude increases slowly, and the increase in cutting damping caused by the deterioration of the contact area between the tool and the workpiece reduces the main peak frequency position of the vertical vibration. The

change law of the power spectrum in the high-frequency band is basically the same as that in the low-frequency band, but the law is not as prominent as that in the low-frequency band. The frequency of the main peak amplitude appearing in the high-frequency region is very small, and there are few spectrum peaks. This is because the forced vibration source of the cutting processing system is generally within 1,000Hz. Therefore, the high frequency spectrum can effectively isolate or weaken the harmonic components of the processing system. Whether

the pattern recognition

state classification can be successful depends to a large extent on the feature analysis and the selection of feature quantities. The dynamic system of deep hole processing is a random process, which is difficult to analyze with a certain deterministic time function. The purpose of feature analysis is to transform the original signal into a feature quantity and find out its relationship with the working condition. There are many such feature quantities, but to reflect the regularity, sensitivity and clustering type of the working condition state in the pattern space, the separability is not the same. It is necessary to select feature quantities with good regularity and strong sensitivity as pattern vectors based on feature analysis to make them have better separability.

This experimental study analyzed the characteristics of vibration, axial force and torque signals in the time domain and frequency domain respectively. On the basis of using time series models, power spectra and autocorrelation spectra, the characteristic spectrum peaks of horizontal vibration and the residual variance of the time series model of force were selected as pattern vectors, and the perceptron algorithm was used to obtain the weighted vector w=(-4,540, 131,090, 300)T. Therefore, the classification function is obtained:

Through the above classification function, we determined the machining status of the tool, as shown in the test samples in Table 1. After many tests, the accuracy rate reached more than 90%.

Conclusion

Experiments have proved that the designed multi-sensor force and vibration measuring device for deep hole drilling is simple to install, reliable in operation, and the combined signals are complementary, which improves the accuracy of monitoring and is suitable for monitoring the status of drilling tools during deep hole drilling.

The vibration signal has significant characteristics in the frequency domain that reflect the wear state of the tool. Under different drilling conditions and drilling amounts, its power spectrum changes differently, but the trend of change is consistent. The

amplitude of the main peak of horizontal vibration and the residual variance of the force are selected as feature vectors. The mode vector is constructed. It can fully reflect the operation status of the deep hole processing system. (end)