CN107756139A - Tool state detection system and method - Google Patents

Abstract

The present invention provides a tool state detection system, which includes: a sensor to capture a sound signal of a tool cutting process; and a controller connected to the sensor signal and receiving the sound signal, the controller determines a processing interval according to the sound signal, determines a tool sound source signal in the processing interval with a frequency distribution, and analyzes the tool sound source signal and a tool sound source time waveform into a tool sound pressure value, wherein the tool sound pressure value is compared with a detection reference value, and the processing interval is compared with a reference interval to determine whether the tool sound pressure value and the processing interval meet the detection reference value and the reference interval.

Description

Tool state detection system and its method

technical field

The present invention relates to a detection system and its method, especially to a tool state detection system and its method.

Background technique

In the metal cutting process, the condition of the cutting tool is the main key that affects product quality and production cost. When the cutting tool is in poor condition (such as wear and tear, the cutting edge is broken), the cutting tool needs to be replaced. However, replacing the cutting tool too early will result in waste of cost, and replacing the cutting tool when the state of the cutting tool is seriously bad will result in defective products.

Furthermore, Taiwan Patent No. I419761 discloses a method and device for detecting the tool state of a processing machine. It prepares a detection device for a microphone array group, a signal processing group, and a monitoring group, and the microphone array group is installed in a The processing signal of the tool is detected on the processing machine platform. The signal processing group is electrically connected to the microphone array group and a computer is provided. The computer is provided with a spatial filter module, a signal conversion module, and a feature selection module. A set and an identification module, the computer processes and presents the detected sound signal, and outputs the identified characteristic value signal, so that the user can know the status of the tool in real time.

However, the microphone array must be installed on the processing machine platform, and the cutting direction of the tool must be found first, and processing signals unrelated to cutting must be eliminated. Moreover, a capture card needs to be installed between the microphone array module and the computer connection. Only then can the processing signal of the cutting tool captured by each microphone of the microphone array group be transmitted to the computer. The procedure for capturing the processing signal is too complicated, and it is necessary to set up multiple microphones and capture cards, and the installation cost is too high.

Furthermore, the computer needs to use the WINER filter to perform Beamforming (summation) of the machining signal of the cutting tool to find out the cutting direction, and use the delay rate wave summing purpose to improve the signal to noise ratio of the cutting tool (signal to Nosieratio) and convert the machining After the signal is boosted and amplified, wavelet transform, Fourier transform, and fast Fourier transform must be used again. Feature extraction is performed according to the frequency distribution, and then, the state of the tool can be identified through artificial intelligence calculation methods (such as neural network, fuzzy).

In other words, the computer needs to process the processing signal through multiple calculations, filtering, amplification and other calculation steps to obtain the characteristics of the tool to identify the status of the tool. The calculation process and judgment process are relatively complicated and time-consuming, and if the microphone array is placed If it is not good, or if there is too much interference in the processing environment, it will easily affect the processing signal and identification results. In the tool detection, it is easy to be restrained and affect the accuracy of the detection results.

Therefore, how to improve the above-mentioned shortcomings and know the state of the cutting tool in a simpler and more accurate way to ensure the processing quality and processing yield has yet to be resolved by the relevant industry.

Contents of the invention

In order to solve the above-mentioned problems, the present invention provides a tool state detection system and its method. By capturing the sound signal of the tool cutting process, the sound signal is used to establish the judgment index of the tool. When the judgment index of the tool is abnormal, it can be judged The status of the tool, so that the judgment and calculation process can be omitted, and the status of the tool can be accurately grasped.

One embodiment of the present invention provides a tool state detection system, which includes: a sensor, which is used to pick up the sound signal of the cutting process of the tool; and a controller, which is connected with the sensor signal and receives For the sound signal, the controller has an analysis module, a conversion module, a feature module and a judgment module. The analysis module judges a processing interval according to the sound signal, and judges a frequency distribution of the sound signal in the processing interval. The tool sound source signal, the conversion module converts the tool sound source signal into a tool sound source time waveform, the feature module analyzes the tool sound source signal and the tool sound source time waveform into a tool sound pressure value, and the judgment module will The tool sound pressure value is compared with a detection reference value, and the processing interval is compared with a reference interval to determine whether the tool sound pressure value and the processing interval meet the detection reference value and the reference interval.

In one embodiment of the present invention, the frequency distribution is that the processing interval is divided into multiple processing time points, each processing time point has a plurality of frequency intervals, and each of the frequency intervals with the same and consecutive occurrences is extracted from the processing interval .

In one embodiment of the present invention, the conversion module analyzes the tool sound source signal based on the frequency-domain harmonic wavelet transform to obtain the time waveform of the tool sound source.

In one embodiment of the present invention, the feature module divides the tool sound source time waveform into complex conversion times, and the feature module uses the root mean square to obtain the tool sound pressure value according to the tool sound source signal and each conversion time.

In one embodiment of the present invention, the processing interval is the time difference between the appearance time and disappearance time of the sound signal.

Based on the above, the effect that the embodiment of the present invention can achieve is that the sensor captures the sound signal of the tool cutting process, and the controller analyzes the sound signal to directly establish the tool status index, and when the index changes, it can directly judge Tool status, thereby saving too much signal analysis and calculation process, and more effectively and quickly and accurately judging the tool status.

Furthermore, the present invention can acquire the required signal by using one sensor, which is simpler than the known detection device, does not occupy the processing space, and has relatively low cost.

Another embodiment of the present invention provides a tool state detection method, which includes the following steps: extracting a sound signal generated by the cutting process of the tool; judging a processing interval according to the sound signal, and judging the sound signal in the processing interval according to a frequency distribution Generate a tool sound source signal; analyze a tool sound source time waveform according to the tool sound source signal; analyze a tool sound pressure value based on the tool sound source signal and the tool sound source time waveform; and the tool sound pressure value and a detection reference value Comparison, comparing the processing interval with a reference interval to judge whether the tool sound pressure value and processing interval meet the detection reference value and reference interval. When the tool sound pressure value and processing interval do not meet the detection reference value and reference interval, an abnormal signal is sent .

In one embodiment of the present invention, the detection reference value and the reference interval are established by analyzing the sound signal of the initial cutting process of the tool.

In one embodiment of the present invention, the cutting tool is initially cut multiple times to obtain the sound signal of each time, the processing interval of each time is judged according to the sound signal of each time, and the reference interval is obtained by averaging the processing intervals of each time, and Convert and analyze each sound signal into the sound pressure value of the tool, and establish it as the detection reference value.

In an embodiment of the present invention, the tool sound source signal is analyzed based on frequency-domain harmonic wavelet transform to obtain the time waveform of the tool sound source.

In an embodiment of the present invention, the tool sound source time waveform is divided into complex conversion times, and the tool sound source signal and each conversion time use the root mean square to obtain the tool sound pressure value.

Based on the above, the effect that the present invention can achieve is to directly establish the tool status indicator by the sound signal of the tool cutting process, and when the indicator changes, the tool status can be directly judged, thereby saving too much signal analysis Calculation process, more effective and fast and accurate judgment of tool status.

Furthermore, the present invention can detect the state of the tool in the cutting process in real time without stopping the production line or adjusting parameters according to the cutting tools of different machines. If an abnormality occurs, an abnormal signal can be sent to remind the processing personnel.

Description of drawings

Fig. 1 is a schematic diagram of the system architecture of the present invention.

Figure 2 is a schematic diagram of the erection of the embodiment of the present invention.

Fig. 3 is a schematic diagram of the waveform of the sound signal of the present invention.

Fig. 4 is a comparison schematic diagram (1) of the present invention, showing that the tool sound pressure value and the processing interval meet the detection reference value and the reference interval.

Fig. 5 is a comparison schematic diagram (2) of the present invention, showing that the tool sound pressure value and the processing interval do not meet the detection reference value and reference interval.

List of reference numbers

knife 1

Tool status detection system 100

sensor 10

controller 20

Analysis Module 21

Conversion Module 22

Feature Mods 23

Judgment Module 24

Tool sound pressure value A

Detection reference value B

Processing interval T

Conversion time T1

Benchmark interval T2

Retrieve step S1

Judgment step S2

Confirm sound source step S3

Conversion step S4

Feature step S5

Establishing a Baseline Step S6

Compare step S7

Detailed ways

In order to illustrate the central idea of the present invention expressed in the column of the above-mentioned summary of the invention, it is expressed in specific embodiments. Various objects in the embodiments are drawn in proportions, sizes, deformations or displacements suitable for illustration, rather than drawn in proportion to actual components.

Referring to Fig. 1 to Fig. 5, the present invention provides a tool state detection system 100, which includes:

A sensor 10, which is used to pick up the sound signal of the tool 1 of the processing machine during the cutting process. In the embodiment of the present invention, the sensor 10 is a sound pressure microphone, and the number of the sensor 10 is one. It is erected facing the direction of the tool 1, wherein the sampling frequency of the sensor 10 is at least 25,000 Hz, which means that the sensor 10 samples at least 25,000 times per second, and the sampling frequency can be adjusted according to the processing type or demand. Furthermore, the sound signal is a voiceprint, and the extracted sound signal is composed of multiple sound sources. The voiceprint of each sound source is different, and the frequency intensity of each sound source is also different.

A controller 20, which is connected to the sensor 10 in a signal connection mode is a wired signal connection or a wireless signal connection. The controller 20 is used to receive the sound signal. The controller 20 has an analysis module 21. The analysis module 21 judges a processing interval T according to the sound signal. The processing interval T is the time difference between the appearance time and the disappearance time of the sound signal. During the cutting process, there is not always contact with the workpiece. Therefore, the time when the tool 1 is in contact with the workpiece is the appearance time of the sound signal, and the time when the tool 1 is not in contact with the workpiece is the time when the sound signal disappears.

In the embodiment of the present invention, the analysis module 21 is provided with a threshold range. When the sound signal is always within the threshold range, and at a certain time point, the sound signal exceeds the threshold range, it indicates the appearance time of the sound signal, and when the sound signal When the signal is within the threshold range again at a certain time point, it indicates the disappearance time of the sound signal. Therefore, the processing interval T can be obtained by subtracting the appearance time from the disappearance time of the sound signal.

Moreover, the analysis module 21 judges a tool sound source signal from the sound signal in a frequency distribution within the processing interval T, wherein the frequency distribution divides the processing interval T into a plurality of processing time points, and each processing time point has For complex frequency intervals, the analysis module 21 extracts the frequency intervals with the same and continuous appearance from the processing interval T, for example: the processing interval T is from 0 seconds to 10 seconds, and the processing time point is every 0.1 second, wherein the processing time point There are 100, 300 and 500HZ frequencies in the frequency range that occurs at the 0.1 second, 200, 400 and 800HZ at the processing time point that occurs at the 0.3 second, and the frequency range that occurs at the 0.4 second at the processing time point. The frequency of 100, 300 and 500HZ, the processing time point is the frequency range of 200, 400 and 800HZ in the first second, then the 0.1 second and 0.4 second are regarded as the sound from the same sound source, the 0.3 second and the second 1 second is regarded as the sound from the same source. Since the sound signal composed of various sound sources will always exist during the cutting process of the tool 1, it is used to determine whether the same frequency range occurs at the same time to judge the sound source signal of the tool, thereby avoiding interference from environmental sounds Influence, can omit the calculation method of filtering and amplification, so as to quickly and accurately determine the sound source as the tool 1.

The controller 20 has a conversion module 22 connected with the analysis module 21 signal. The conversion module 22 analyzes the tool sound source time waveform based on the frequency domain harmonic wavelet conversion of the tool sound source signal. Therefore, the present invention utilizes the frequency domain Wavelet transform is used for time-frequency analysis (Time-frequency transform), which can adjust the resolution of the tool sound source time waveform according to the needs.

The controller 20 has a feature module 23 connected with the signal conversion module 22. The feature module 23 analyzes the tool sound source signal and the tool sound source time waveform into a tool sound pressure value A, wherein the feature module 23 converts the tool The time waveform of the sound source is divided into complex conversion times T1, and the feature module 23 uses the root mean square to obtain the sound pressure value A of the tool according to the sound source signal of the tool and each conversion time T1, wherein, in the embodiment of the present invention, each conversion time T1 is 0.1 second, the minimum value of the sound pressure value A of the tool is calculated every 0.1 second, as the characteristic state of the tool 1 in the cutting process, as shown in FIG. 4 .

The controller 20 has a judging module 24 connected with the signal module 23, and the judging module 24 is used to compare the tool sound pressure value A with a detection reference value B, and compare the processing interval T with a reference interval T2 For comparison, judge whether the tool sound pressure value A and the processing interval T meet the detection reference value B and the reference interval T2, as shown in Figure 4, the tool sound pressure value A and the processing interval T both meet the detection reference value B and the reference interval T2 , the judgment result is normal; as shown in Figure 5, if the tool sound pressure value A is less than the detection reference value B, or the processing interval T is greater than or smaller than the reference interval T2, an abnormal signal will be sent to remind the processing personnel to operate the tool 1 The replacement action to ensure the processing accuracy and yield of the workpiece. For example: when the tool 1 is in the broken state, it cannot contact the workpiece correctly, so the sound pressure value A of the tool will be smaller than the detection reference value B; in addition, when the tool 1 is in the broken state, it cannot be correctly contacted with the workpiece. Therefore, the processing time of the tool 1 will be prolonged, and the processing interval T will be greater than the reference interval T2.

The detection reference value B uses the sound signal captured by the sensor 10 during the initial cutting process of the tool 1, and the sound pressure value A of the tool is analyzed and obtained by the analysis module 21, the conversion module 22 and the characteristic module 23 of the controller 20. , established as the detection reference value B, and stored in the judgment module 24, and then the sound pressure value A of the tool calculated during each cutting process of the tool 1 can be compared with the detection reference value for judgment.

Furthermore, in the embodiment of the present invention, the reference interval T2 is to initially cut the tool 1 multiple times, wherein, the initial cutting times of the tool 1 are 5 to 15 times, and the cutting times can be adjusted according to needs; 10 captures the sound signal of each cutting process of the tool 1, and the analysis module 21 judges the processing interval T of each time according to the sound signal of each time, and averages the processing interval T of each time to obtain the reference interval T2.

Since the tools 1 used by each processing machine are different, the length of each cutting time and the cutting state of each tool 1 are also different. Therefore, the initial cutting process of each tool 1 is used to establish a dedicated tool according to the characteristics of the tool 1. The detection reference value B and the reference interval T2 are based on the detection reference value B and the reference interval T2 specific to each tool 1 as the judgment standard for abnormal problems. Compared with directly setting the comparison value, it is more accurate and effectively finds out the abnormality place.

Using the aforementioned tool state detection system 100, another embodiment of the present invention provides a tool state detection method, which includes the following steps:

Extraction step S1: Acquire the sound signal generated by the cutting process of the tool 1. In the embodiment of the present invention, the sensor 10 is erected to face the direction of the tool 1, and the sensor 10 captures the sound signal generated by the cutting process of the tool 1. The sound signal connects the sensor 10 with the controller 20 , and the sensor 10 transmits the sound signal to the controller 20 .

Judgment step S2: judge the processing interval T according to the sound signal. In the embodiment of the present invention, the analysis module 21 of the controller 20 has a threshold range. When the sound signal is always within the threshold range, and at a certain point in time, the sound If the signal exceeds the threshold range, it indicates the appearance time of the sound signal, and when the sound signal is within the threshold range again at a certain point in time, it indicates the disappearance time of the sound signal. Time and disappearance time difference to obtain the processing interval T.

Confirm the sound source step S3: judge the sound source signal of the tool by the frequency distribution of the sound signal in the processing interval T, wherein, in the embodiment of the present invention, the frequency distribution divides the processing interval T into multiple processing time points, each processing The time point has a plurality of frequency intervals, and the analysis module 21 extracts the same frequency intervals that appear consecutively from the processing interval T. Therefore, the tool sound source signal that matches the sound source of the tool 1 can be found from the processing interval T.

Transformation step S4: According to the tool sound source signal based on the frequency domain harmonic wavelet transformation to analyze the tool sound source time waveform, wherein, in the embodiment of the present invention, the conversion module 22 of the controller 20 converts the tool sound source signal based on the frequency domain Harmonic wavelet transform analyzes the time waveform of the tool sound source. Therefore, the present invention utilizes the frequency domain wavelet transform for time-frequency analysis (Time-frequency transform), and can adjust the resolution of the tool sound source time waveform as required.

Characteristic step S5: analyze the tool sound pressure value A according to the tool sound source signal and the tool sound source time waveform. The module 23 obtains the tool sound pressure value A by using the root mean square according to the tool sound source signal and each conversion time T1, wherein, in the embodiment of the present invention, each conversion time T1 is 0.1 second, and the tool sound is calculated every 0.1 second The minimum value of the pressure value A is used as a characteristic aspect of the cutting process of the tool 1.

Establishing a benchmark step S6: using the sound signal analysis of the initial cutting process of the tool 1 to establish a detection reference value B and a reference interval T2, wherein the initial cutting process of the tool 1 refers to confirming whether the tool 1 is installed every time a new tool 1 is installed It is a normal test cutting process. Therefore, in the embodiment of the present invention, the tool 1 is initially cut multiple times. After the acquisition step S1, the sensor 10 is used to capture each sound signal and sent to the controller 20. Then, Through the judgment step S2, the analysis module 21 of the controller 20 judges the processing interval T each time according to each sound signal, and averages the processing interval T each time to obtain the reference interval T2, wherein the initial cutting times of the tool 1 is 5 to 15 times, the initial number of cuts can be adjusted as needed.

In addition, by confirming the sound source step S3, converting step S4 and characteristic step S5, each sound signal is converted and analyzed into a tool sound pressure value A, and the tool sound pressure value A obtained each time is established using a normal distribution method To detect the reference value B, and store the detection reference value B in the judging module 24 as a detection standard for each cutting process of the tool 1 after the initial cutting. In this way, based on the initial cutting process of each tool 1, an exclusive detection reference value B and reference interval T2 are established according to the characteristics of the tool 1, and the detection reference value B and reference interval T2 specific to each tool 1 are used as the judgment of abnormal problems The standard is more accurate than directly setting the comparison value, and it can effectively find out the abnormality.

Comparison step S7: compare the tool sound pressure value A with the detection reference value B, and compare the processing interval T with the reference interval T2, and judge whether the tool sound pressure value A and the processing interval T meet the detection reference value T and the reference interval T2, when the sound pressure value A of the tool and the processing interval T do not meet the detection reference value A and the reference interval T2, an abnormal signal is sent. Among them, in the embodiment of the present invention, the judging module 24 of the controller 20 compares the tool sound pressure value A with the detection reference value B, and compares the processing interval T with the reference interval T2, if the tool sound pressure value A is less than the detection reference value B, indicating that the shape of the cutting tool 1 has changed due to wear, and the workpiece has not been completely cut, resulting in a smaller vibration of the workpiece, so the sound pressure value A of the tool detected by the sound signal is also reduced; In addition, the processing interval T is greater than or smaller than the reference interval T2, for example: when the processing interval T is greater than the reference interval T2, it means that the shape of the cutting tool 1 is changed due to wear, and the workpiece is not completely cut, so the cutting process of the tool 1 is elongated, so that The detected processing interval T is larger than the reference interval T2. Therefore, if the detection reference value A and the reference interval T2 are not met, an abnormal signal will be sent to remind the processing personnel to replace the tool 1 to ensure the processing accuracy and yield of the workpiece.

Through the above, the effect that the present invention can achieve is to directly establish the status index of the tool 1 through the sound signal of the cutting process of the tool 1, and when the indicator changes, the status of the tool 1 can be directly judged, thereby saving Eliminate excessive signal analysis and calculation processes, and judge the status of tool 1 more effectively and quickly and accurately.

Furthermore, the present invention uses one sensor 10 to capture the required signal, thereby, with a simple detection system, the state of the tool 1 can be quickly and accurately known without occupying a processing space.

In addition, the tool state detection system 100 and its method of the present invention can detect the state of the cutting tool 1 in real time, without stopping the production line, and without adjusting the parameters of the cutting tool 1 according to different processing machines. It can send out abnormal signals to remind processing personnel from time to time to ensure processing quality and yield.

The above-mentioned embodiments are only used to illustrate the present invention, and are not intended to limit the scope of the present invention. All modifications or changes that do not violate the spirit of the present invention belong to the intended protection category of the present invention.

Claims (10)

1. a kind of cutting tool state detecting system, it is included：

One sensor, it is capturing one of tool cutting process sound signal；And

One controller, it links with the sensor signal and receives the sound signal, and the controller has an analysis module, one Conversion module, a feature module and one judge module, and the analysis module judges a processing interval according to the sound signal, in The sound signal is judged into a cutter sound source signal with a frequency distribution in the processing interval, the module of changing is by institute State cutter sound source signal and be converted to a cutter sound source time waveform, the feature module is according to the cutter sound source signal and described The analysis of cutter sound source time waveform is a cutter sound pressure level, wherein, the judgement module detects the cutter sound pressure level and one A reference value is compared, and the processing interval and a benchmark section are compared, and judges the cutter sound pressure level and the processing interval Whether the detection a reference value and the benchmark section are met.

2. cutting tool state detecting system according to claim 1, wherein, the frequency distribution system divides the processing interval For plural process time point, each process time point has Complex frequency section, by being captured in the processing interval with identical And each frequency separation of continuous appearance.

3. cutting tool state detecting system according to claim 1, wherein, the conversion module interrogates the cutter sound source Number, the cutter sound source time waveform is analyzed based on frequency domain harmonic wavelet conversion.

4. cutting tool state detecting system according to claim 1, wherein, the feature module is by the cutter sound source time Waveform is divided into the complex conversion time, and the feature module is according to the cutter sound source signal and each conversion time, using equal Root obtains the cutter sound pressure level.

5. cutting tool state detecting system according to claim 1, wherein, the processing interval occurs for the sound signal Time and the time difference of extinction time.

6. a kind of cutting tool state detection method, is comprised the steps of：

Capture one of tool cutting process generation sound signal；

One processing interval is judged according to the sound signal, in the processing interval by the sound signal with a frequency point Cloth judges a cutter sound source signal；

One cutter sound source time waveform is analyzed according to the cutter sound source signal；

One cutter sound pressure level is analyzed according to the cutter sound source signal and the cutter sound source time waveform；And

The cutter sound pressure level compares with a detection a reference value, and the processing interval compares with a benchmark section, judges the knife Whether tool sound pressure level and the processing interval meet the detection a reference value and benchmark section, when the cutter sound pressure level and The processing interval does not meet the detection a reference value and the benchmark section, sends an abnormal signal.

7. cutting tool state detection method according to claim 6, wherein, utilize the sound of cutter initial cuts process The detection a reference value and the benchmark section are established in signal analysis.

8. cutting tool state detection method according to claim 7, wherein, it is every to obtain by cutter initial cuts plural number Take second place the sound signal, and according to often taking second place, the sound signal judges the processing interval that often takes second place, and will often take second place described Processing interval averagely obtains the benchmark section, and it is the cutter sound that the sound signal that will often take second place, which carries out transformational analysis, Pressure value, to be established as the detection a reference value.

9. cutting tool state detection method according to claim 6, wherein, it is small that the cutter sound source signal is based on frequency domain harmonic Ripple transformational analysis goes out the cutter sound source time waveform.

10. cutting tool state detection method according to claim 6, wherein, the cutter sound source time waveform is divided into plural number Conversion time, the cutter sound source signal and each conversion time obtain the cutter sound pressure level using root mean square.