KR20140072331A - Method for preliminary surveillance of failure diagnosis - Google Patents

Abstract

A method of prior monitoring and diagnosing repair or replacement of an abnormal component by confirming abnormality with abnormal vibration before occurrence of abnormal vibration and component damage of a production facility such as a robot is disclosed.

Description

[0001] The present invention relates to a method for preliminary diagnosis,

The present invention relates to a pre-diagnosis diagnostic method, and in particular, before abnormal vibration and parts damage of a production facility of a robot or the like occurs, it is checked by abnormality with an abnormal vibration to proactively monitor and diagnose repair or replacement of an abnormal component ≪ / RTI >

The vibration monitoring and diagnosis system monitors and diagnoses the abnormal vibration related to various faults that may occur due to an unpredictable cause during operation of the machine, thereby finding problems in advance.

As machines become larger and more sophisticated, their importance is increasing. For example, in the case of LCD display production plant robots that require very precise motion control, if one of the robots in the entire production line has a problem, the whole production line is stopped and the abnormal part is replaced. And the cause has been identified. As a result, the production rate and the defect rate were affected by the decrease of the utilization rate and the stability, resulting in an economic loss due to a decrease in productivity.

Therefore, in order to minimize the occurrence of problems in the entire production process, it is ideal to establish measures for open repair and normal maintenance by selecting a period of less influence on the entire production system once the abnormal signs are grasped at all times through regular monitoring and diagnosis can do.

For this purpose, the abnormal vibration monitoring and fault diagnosis system establishes a database of equipment vibration accumulated over a long period of time, and judges and responds to abnormal vibration and malfunction. By storing data on such abnormal vibration, the computer can perform all processing, thereby enabling automatic diagnosis of production equipment. Therefore, by using various vibration analysis methods, a production and transportation robot that is being operated for the display glass transport should check the abnormality before abnormal vibration and parts damage and build up a pre-detection system to repair or replace the abnormal parts It is possible to shorten production schedule and minimize unnecessary time / human waste.

As industry develops and technology develops, awareness of maintenance, repair and diagnosis of various systems as well as importance of product production is increasing, and it is already forming a big economic area. Condition monitoring is a system that continuously or regularly monitors the operation status of a mechanical system in order to reduce enormous economic and human losses caused by unexpected failure of a mechanical system or failure of parts, To detect and detect material loss by vibration, noise measurement and abrasion. In addition, it can be applied to various kinds of materials.

In particular, since a large-sized mechanical device includes many parts that rotate or move while rubbing against peripheral parts, if the maintenance and maintenance are neglected, the entire mechanical device can not be driven due to a failure in a specific part A phenomenon may occur. In the meantime, in a system known as a system for predicting or diagnosing a failure of an industrial facility, vibration or noise generated from a mechanical device is monitored and converted into a digital signal, and then a digital signal in a steady state is compared with a digital signal in an abnormal state The method of judging has been mainly applied.

So far, signal processing techniques used in fault diagnosis of machinery have been used in the spectrum analysis in the frequency domain such as RMS, peak-to-peak, and crest factor of time domain. to be. In this way, when the vibration or noise signals generated in the machinery and the like are analyzed in the frequency domain, it can be suitably applied to detect the overall characteristics and errors of the signals.

Fast Fourier Transform (FFT) is generally used for analyzing the spectrum of vibration or noise to a mechanical device in the frequency domain. Such a conversion method is disadvantageous in that the operation speed is too slow when the amount of data is large However, when the computed data is used as the input value of the inference system, another operation is required.

In 2001, the Tribology Research Center conducted research on the development of a remote monitoring system using accelerometers and distributed object technology through research projects on integrated mechanical condition monitoring technology. In the Department of Mechanical Engineering, Sungkyunkwan University, We have developed a system to monitor remotely using on - line monitoring method by developing new technology.

Many attempts have been made with foreign institutions for a more systematic and scientific approach to state diagnosis technology. The International Organization for Standard (ISO) provides standard guidance for all areas, ranging from conclusions about status diagnosis to risk criteria such as measurement location, vibration, wear and tear, VDI, American Petroleum Association (API), NAS (National Aeronautics and Space Administration), and others have regulations on wear and vibration limits.

Among the measurement information, vibration information is used as a very important data for judging the soundness of the rotating machine. Recently, there has been remarkable growth in the field of data acquisition technology and vibration analysis, and the direct relationship between the pattern of vibration signals and the soundness of the machine is why vibration signals are important in the field of rotary machines.

In the United States, wear monitoring, which is a typical technique for diagnosing mechanical conditions along with vibration information, measures and analyzes the amount of wear occurring in a mechanical system to determine whether the system is damaged or not. Is often compared to the role and characteristics of blood flowing within the body. It is possible to extract important information that can not be obtained from the vibration information and it is possible to observe and detect the degree of damage of the mechanical element parts without disassembling the machine in the operating state. In addition, state diagnostic techniques using noise and temperature are being used.

The IRD Mechanization Research Team of the United States published devices and reports for machine maintenance management systems, such as diagnosis and monitoring of safety status of general rotating machinery. .

It is an object of the present invention to provide an abnormal diagnosis advance monitoring method applicable to various facilities.

It is another object of the present invention to provide an abnormality diagnosis advance monitoring method capable of pre-diagnosing an abnormality symptom by analyzing vibration characteristics of a steady-state facility.

Another object of the present invention is to provide an abnormality diagnosis advance monitoring method capable of identifying the facility and each part of the inspection / replacement cycle and reducing maintenance cost.

The above object is achieved by a method comprising: registering a component part of a facility; Measuring a natural frequency of each part; Automatically analyzing the measured data to form a database of the vibration characteristics of the facility; Performing vibration monitoring to measure vibration of each part and analyze the frequency; Performing pre-diagnosis to detect an abnormal condition; And diagnosing the failure and generating and reporting an alarm to the failed part.

According to the above-described structure, it is possible to diagnose a specialized production facility by developing a vibration measuring device specialized in a production facility, find a singularity with a DB accumulated through an abnormal signal identification algorithm, Accident handling is possible.

In addition, it is possible to apply the pre-predictive check technology by directly applying the vibration-based production facility diagnosis technology to the LCD production facility.

Also, it is possible to provide a prognostic tool according to the identification of abnormal operation characteristics by developing a vibration analysis pattern algorithm of each production facility.

In addition, after the completion of the technology development, it can be expanded to the proactive diagnosis system of the industrial robot, not limited to the LCD production robot.

In addition, due to the proactive diagnosis system of the LCD production robot, for example, the LCD manufacturer can perform repair or replacement of the robot in advance in the event of an abnormality due to the proactive diagnosis of the LCD production robot, It is possible to detect the failure in advance, thereby increasing the production rate of the LCD and increasing the profit.

FIG. 1 shows the concept of the anomaly diagnosis advance monitoring method according to an embodiment of the present invention.
2 is a flowchart illustrating an abnormality diagnosis advance monitoring method according to an embodiment of the present invention.
Figure 3 shows a sample matching technique algorithm.
Figure 4 shows an axis data separation algorithm.
Figure 5 shows a part analysis algorithm.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 shows the concept of the anomaly diagnosis advance monitoring method according to an embodiment of the present invention.

First, the constituent parts of the facility, such as a stage, a motor, a cylinder, and a pump, are registered (step S1).

Then, the natural frequency of each part is measured (step S2). That is, the frequency of each part is measured during operation and stop of the facility.

Next, system identification is performed (step S3). That is, the measured data is automatically analyzed to make the vibration characteristics of the facility database.

Then, vibration monitoring is performed (step S4). That is, the vibration of each part is measured and the frequency is analyzed.

Next, the pre-diagnosis is performed (step S5). That is, it detects an abnormal state.

Finally, a fault is diagnosed (step S6), an alarm is generated and notified to the faulty part, and maintenance and replacement are performed.

2 is a flowchart illustrating an abnormality diagnosis advance monitoring method according to an embodiment of the present invention. In the following description, an LCD robot will be described as an example of production facilities, but the present invention is not limited thereto.

Data is collected using a 3-axis acceleration sensor. The three-axis vibration acceleration sensor is built-in or portable and operates the X, Y and Z axes and the bottom Z axis on four channels, the center-of-gravity axis of the target robot.

Specifically, in order to measure the vibration data when the motors of the target robot move, the jig of the cube is fixedly mounted on the center axis of the target robot and the acceleration sensors (ch1, ch2 , ch3), and an acceleration sensor (ch4) is installed in the Z axis direction on the ground surface of 500 mm of the target robot in order to measure a reference value (bottom vibration) with respect to the ground, and data measured by each acceleration sensor is transmitted to a wireless data logger To digital data and wirelessly transmits the data to the analysis PC.

The vibration of the robot is measured during the operation of the target robot, and the operating conditions of the target robot are measured for the output standard of 20%, 50%, and 100%.

The vibration response analysis analyzes the acceleration response of the time history acquired from the acceleration sensor. The vibration level measurement is performed by using the power spectral density (PSD), the root mean square (RMS), and the fast Fourier transform Fourier transform).

Separates the trigger signal from the received data, identifies the operation pattern, and stores the trigger data.

After the number of data set for a predetermined time is secured in this way, a matching score of the trigger data is discriminated and repeated data of the same operation pattern is extracted.

Figure 3 shows a sample matching technique algorithm.

After loading the stored trigger data and comparing them with each other (Cross correlation coefficient, Coherence (Peak & RMS, Mean), Weighting Function Option of each value) and Matching Score Threshold value, Extracts representative data from each other, classifies the same operation pattern, and performs reference registration.

That is, if the number is equal to or larger than the threshold value, it is stored in the pattern database to be grouped, and if so, it is determined whether or not to register in the new pattern and registered in the new pattern database.

Apply the axis data separation algorithm and the part analysis algorithm in the expert system from the temporal pattern database.

Figure 4 shows an axis data separation algorithm.

That is, axial motion data is extracted from the same motion pattern, and the minutiae points are as follows.

- Axis Basis DOF: Axis (+, -), Axis (+, -), Upper arm (+, -), Lower arm ) -> Analysis of vibration characteristics of each free sagging basis

- Estimate the number of degrees of freedom for each motion (Define the number of possible combinations)

- Extraction of vibration data of axis motion of each operation pattern (mixed data matching technique)

- Save axis data DB

Figure 5 shows a part analysis algorithm.

Axis motion data is extracted from the same motion pattern, and the minutiae points are as follows.

- Date of axis data (in month / week) Matching Score management (abnormal vibration detection)

- Comparison of RMS value, Crest Factor, VDV value

- Frequency analysis (comparison of existing accumulated axis data history with current axis data and comparison algorithm)

· Spectrum & Cepstrum / Overall data analysis (frequency analysis of rotating machine)

· Hankel matrix-based SVD (Stability Chart): signal separation, frequency analysis

· Wavelet Transform: Signal Separation, Frequency Analysis

- Axis data error signal detection / Parts error detection

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. Accordingly, the scope of the present invention should not be construed as being limited to the embodiments described above, but should be construed in accordance with the following claims.

Claims (1)

Registering a component part of the facility;
Measuring a natural frequency of each part;
Automatically analyzing the measured data to form a database of the vibration characteristics of the facility;
Performing vibration monitoring to measure vibration of each part and analyze the frequency;
Performing pre-diagnosis to detect an abnormal condition; And
And diagnosing the fault and generating and reporting an alarm to the faulty part.

Images