KR101729446B1 - Method for detecting fault of blade, computer program and recording media for the same - Google Patents

Abstract

The present invention relates to a method for detecting defects of blades of an axial-flow fan and a turbine. The purpose of the present invention is to provide: the method for detecting the defects of the blade capable of accurately monitoring the defects of the blade and parts where the defects are generated by a data study per section corresponding to the blade, a data comparative analysis per section corresponding to the blade, an order analysis result study, and an order analysis result comparative analysis; a computer program for the same; and a recording medium. To achieve this, the method for detecting the defects of the blades includes a normal data study step of receiving a pressure value sensed by a wind pressure sensor installed in a front side or a rear side of the blades of the axial-flow turbine or the axial-flow fan while the blades have no defects and setting a normal range and a defect detecting step of receiving a pressure value sensed by the wind pressure sensor installed on the front side or the rear side of the blades at a detection timing and determining the defects by analyzing whether the sensed pressure value is within the normal range obtained in the normal data study step or not. In the defect detecting step, data about one rotation of the fan or the turbine is divided and re-sampled into sections identical to the number of the blades, and the defects of the blades of a number corresponding to each section.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a blade defect detection method, a computer program for the same,

The present invention relates to a method of detecting a defect in a rotating blade, and more particularly, to a method of detecting a defect in a blade by using data learning and order analysis, And a method for detecting a blade defect. The present invention also relates to a computer program and a recording medium for performing the defect detection method.

The blower or the compressor is a device that rotates the fan using power and supplies the fluid accelerated or pressurized by the fan to the rear end. The centrifugal fan discharges the fluid in the radial direction of the fan and the fluid is discharged in the direction parallel to the rotary shaft And an axial flow fan. The axial flow fan has a plurality of blades arranged along the circumferential direction, and each blade is disposed in an inclined direction with respect to the rotational direction to provide a transfer force to the fluid. Typically, the axial flow fan is equipped with tens to hundreds of blades.

The turbine engine is a device that rotates the turbine using the flow of fluid and provides the rotational power of the turbine to the connected device, which operates roughly in the opposite direction to the blower or compressor. The turbine is also divided into a centrifugal turbine that discharges the fluid in the radial direction of the turbine and an axial turbine that discharges the fluid in a direction parallel to the rotational axis of the turbine. The axial turbine has a plurality of blades arranged along the circumferential direction and inclined with respect to the direction of rotation. Typically, the axial turbine is equipped with hundreds to thousands of blades.

As described above, the axial fan or the axial turbine includes a plurality of blades. Since the blades rotate at a high speed while receiving a large force, defects such as breakage or deformation of the blades can be caused by a very small factor. The failure of some of the blades, even a single blade, can significantly degrade the performance of the device of the fan or turbine, and the defective blades can cause the rotor to become unbalanced and cause fatal damage to the entire device have.

Accordingly, it is important to quickly and accurately detect the presence / absence and position of a defect in the blade. A conventional visual detection method using a human force or an apparatus is difficult to confirm the defect because the blade rotates at high speed during operation of the apparatus, It is possible to perform an accurate inspection only when it is stopped. Furthermore, in the case of a turbine in which a large number of blades are installed in a large number of rows in a duct, it is very difficult to confirm whether the blades are defective or not even in a state where operation is stopped. This causes large disadvantages and inefficiencies in terms of maintenance of the apparatus.

In order to solve such a problem, Korean Patent No. 10-0954157 discloses a turbomachine blade damage monitoring unit. The monitoring unit includes a pressure sensor disposed near the rotating body having the blades rotated at a constant rotation speed to measure the value of the voltage of the blades, and a pressure sensor for determining whether the measured voltage value is included in the predetermined reference voltage value range And a control unit. With such a configuration, the monitoring unit monitors the voltage force behind each of the blades to check the damage of the blades in real time.

However, in the above-mentioned patent, it is possible to confirm that a defect has occurred in an unspecified one of the blades, but it is possible to provide specific information such as a position of the blade where the defect occurs, a defect in the part of the blade none.

In addition, since it depends only on the comparison and analysis of the reference voltage value, reliability and accuracy can not be sufficiently secured in determining whether or not a blade is defective.

Korean Patent No. 10-0954157

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a method and apparatus for accurately monitoring the presence or absence of defects generated in a blade by performing learning and comparison analysis of data corresponding to each blade, And to provide a computer program and a recording medium therefor.

According to an aspect of the present invention, there is provided a method for detecting a blade defect, the method comprising: receiving a pressure value sensed by a wind pressure sensor installed in front of or behind a blade of an axial flow fan or an axial flow turbine in a state where there is no defect of the blade, A normal data learning step; And a defect detection step of receiving a pressure value detected from a wind pressure sensor installed at the front or rear of the blade at the time of detection and determining whether a defect is generated by analyzing whether or not the pressure value belongs to a normal range obtained in the normal data learning step, The data for one revolution of the fan or turbine is resampled to the same number of sections as the number of blades to determine whether the blades corresponding to the respective regions have defects.

Preferably, the normal data learning step comprises analyzing the order of the sensed pressure value data to set a normal range and storing the set normal range as order analysis learning data, and the defect detection step is a step of learning the current state And comparing the pressure value data with the order analysis learning data to determine whether or not a defect has occurred in the blade in accordance with the presence or absence of data values deviating from a set normal range.

The normal data learning step and the defect detecting step may detect a defect in another part of the blade by receiving a sensed pressure value from a plurality of wind pressure sensors sensing pressure values at different positions of the blades.

The normal data learning step receives the pressure value sensed by the wind pressure sensor and sets an upper limit value and a lower limit value of the normal range, and the defect detection step detects the pressure value detected by the wind pressure sensor out of the range between the upper limit value and the lower limit value It can be determined that a defect of the face blade has occurred.

Also, the root mean square of the sensed pressure values may be obtained for each divided section according to the number of blades, and the root mean square of each section may be compared to determine that a defect has occurred in the blade corresponding to the section in which the deviation occurs.

The defect detection step may determine whether a defect has occurred according to the frequency of the detection data that deviates from the normal range at the sampling time.

Further, the present invention provides a computer program stored in a recording medium and a computer readable recording medium having recorded thereon a computer program for executing the method for detecting a blade defect, in order to execute the above-described method for detecting a blade defect.

According to the present invention configured as described above, the data for one revolution of the fan or the turbine is divided and resampled into the same number of sections as the number of blades, thereby determining whether the blades corresponding to the respective regions have defects. So that it is possible to detect the position quickly and accurately.

Further, the reliability of defect detection can be further improved by comparing the order analysis data and determining whether or not a defect has occurred in the blade in accordance with the presence or absence of data values deviating from the normal range.

FIG. 1 is a graph showing wind pressure sensed data sensed by a wind pressure sensor in real time in the blade detection method of the present invention.
FIG. 2 is a graph showing wind pressure sensed data by dividing and resampling the wind pressure sensing data into a section corresponding to the number of blades per revolution of the turbine.
3 is a graph showing pressure range learning results obtained by performing the normal data learning step.
4 is a graph showing an order analysis learning result obtained by performing the normal data learning step.
FIG. 5 is a graph showing a result of comparison analysis of order analysis obtained by performing the defect presence / absence detection step.
6 is a graph showing a comparison result of the pressure value obtained by performing the defect position detecting step.
FIG. 7 is a graph showing a result of the defect position judgment obtained by performing the defect position detecting step. FIG.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

It is to be understood that the following specific structure or functional description is illustrative only for the purpose of describing an embodiment in accordance with the inventive concept, and that the embodiments according to the concept of the present invention may be embodied in various forms, It should not be construed as being limited to examples.

Since the embodiments according to the concept of the present invention can make various changes and have various forms, specific embodiments are illustrated in the drawings and described in detail herein. However, it should be understood that the embodiments according to the concept of the present invention are not intended to limit the present invention to specific modes of operation, but include all modifications, equivalents and alternatives falling within the spirit and scope of the present invention.

The terms first and / or second etc. may be used to describe various components, but the components are not limited to these terms. The terms may be named for the purpose of distinguishing one element from another, for example, without departing from the scope of the right according to the concept of the present invention, the first element being referred to as the second element, The second component may also be referred to as a first component.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when it is mentioned that an element is "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between. Other expressions for describing the relationship between components, such as "between" and "between" or "adjacent to" and "directly adjacent to" should also be interpreted.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. It will be understood that the terms "comprises", "having", and the like in the specification are intended to specify the presence of stated features, integers, steps, operations, elements, parts or combinations thereof, But do not preclude the presence or addition of steps, operations, elements, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the meaning of the context in the relevant art and, unless explicitly defined herein, are to be interpreted as ideal or overly formal Do not.

Hereinafter, the present invention will be described in detail with reference to the preferred embodiments of the present invention with reference to the accompanying drawings.

A method for detecting a blade defect according to the present invention is a method for detecting a blade defect by installing a wind pressure sensor on the back or front of a blade and receiving real time data of a pressure value sensed by the wind pressure sensor, And the position of the defect is detected.

In the present embodiment, the application of the present invention to a turbine having twelve blades will be exemplarily described. In addition, three wind pressure sensors for detecting the pressure at the corresponding positions are provided so as to individually detect defects at the upper, middle, and lower ends of the blade.

The blade defect detection method of the present invention includes a normal data learning step and a defect detection step, and the defect detection step includes a defect presence detection step and a defect position detection step.

In the normal data learning step, the pressure value sensed by the wind pressure sensor at the time of rotation of the turbine in a state in which all the blades are confirmed to be in a normal state without damage or deformation is inputted in real time for a predetermined time and stored as data, Set the upper and lower limits of the normal category. At this time, the detected pressure may be a positive pressure or a dynamic pressure.

The pressure values sensed by the three wind pressure sensors are displayed in real time with three data values (vertical axis) with respect to time (abscissa) as shown in Fig. In FIG. 1, the three data values indicate the upper, middle, and lower side pressure values of the blade, the white color represents the upper end side pressure value, the red color represents the intermediate portion side pressure value, and the green color represents the lower end side pressure value, respectively .

The pressure values measured as described above can be individually displayed as a sensed data graph by wind pressure sensors installed at the top, middle, and bottom of the blade as shown in FIG. Fig. 2 is a graph showing the relationship between the pressure value of the turbine and the pressure of the turbine, Is resampled according to the number of blades, and divided into 12 sections. In FIG. 2, it can be seen that the pressure value repeatedly appears in waveforms of a similar pattern for each of the 12 divided sections, and the numbers of the sections are matched with the numbers of the blades, .

As described above, when the input and the storage of the pressure value data are continued for a predetermined time, the upper limit value and the lower limit value of the normal category can be learned and set. This is stored in the pressure range learning data file and can be displayed as a pressure range learning result graph as shown in Fig. In FIG. 3, the red dotted line at the top represents the upper limit value and the red dotted line at the lower side represents the lower limit value.

In addition, order analysis of the sensed pressure value data can be performed to learn and set the order analysis result in the steady state. This is stored as an order analysis learning data file, and can be displayed as a graph of order analysis learning results as shown in FIG. The order analysis graph of FIG. 4 shows sensor data logically displayed, and shows results obtained by applying weights to small measurement values suitable for calculation or display. The formula applied to the illustrated result is 20 x log (PSD / 0.00001), which is the criterion for 100dB when PSD is 1. FIG. 4 shows values in a 12th order multiple (harmony component) corresponding to 12, which is the number of blades, and a red dotted line indicates a curve (reference) value of the order analysis normal range.

In general, order analysis is an algorithm for analyzing vibration noise signals for a machine that performs rotational or repetitive motion, the result of which is independent of the rotational speed of the object and is influenced by the blade number of blades. FIG. 4 is a graph showing the data of the sensed pressure value with respect to the order (abscissa). When there is no deformation of the blade, a value is generated in integer multiples of the number of wings. In this embodiment, since the number of the blades is 12, it can be confirmed that the values are generated in the 12th, 24th, 36th, 48th, and the like.

After the normal data learning is performed as described above, the defect presence / absence detection step is performed at a required point in time. The defect presence / absence detection step more reliably discriminates whether a defect such as breakage or deformation has occurred in a plurality of blades through comparison analysis of the order analysis data.

In the state where the turbine is in operation, pressure value data of the current state detected by the wind pressure sensor is input, and the data is subjected to order analysis and compared with the order analysis learning data. The comparison result can be displayed as a graph of the order analysis comparison result as shown in FIG.

When the blade is deformed, the wind intensity is not constant, so the value is generated even outside the integer number of wings. Fig. 5 shows data obtained when the stop portion of the blade is deformed. In the middle graph, it can be seen that a value has occurred even outside an integer multiple of 12.

When the graph of FIG. 5 is compared with the order analysis learning result graph of FIG. 4, it is indicated that a large number of data values (red dot indication portions) out of the order analysis normal range are generated in the middle order analysis graph, It indicates that there is a blade which is not specific to which blade but has breakage or damage in the middle portion. This makes it possible to accurately discriminate the presence or absence of a defect in the blade separately from the defect position detecting step described later.

0.025영역에 최대값을 기준으로 학습하고 세로축은 15% 가중치를 설정하여 학습 기준선을 생성한 것을 예시하였다. 상기한 최대값과 가중치는 사용자가 설정 가능하다.The red dashed line is a reference line for distinguishing positive judgment, and is generated by learning normal data in which no deformation has occurred. The red dotted horizontal axis indicates the

0.025 for the maximum value and 15% for the vertical axis. The maximum value and the weight value can be set by the user.

As described above, when the wind pressure generated in the deformed blade is measured, the result of the order analysis is not constant. Thus, it is possible to check whether the blade is deformed in real time.

The defective position detection step is performed by comparing the pressure value data of the present state sensed by the wind pressure sensor with the above-described pressure range learning data. That is, whether the upper limit value and the lower limit value of the normal category are set based on the pressure range learning data and whether the real time pressure value of the current state falls within the range between the upper limit value and the lower limit value is determined to determine whether the normal state or the defect has occurred. The result can be displayed as a graph of the pressure comparison analysis result of FIG. In the graph of FIG. 6, it can be seen that a pressure value deviating from the normal range is inputted in the middle of the No. 3 blade, indicating that breakage or deformation has occurred in the blade and the corresponding blade.

In addition, the root mean square (RMS) of each section is divided into twelve sections to determine the similarity of the waveforms for the same sampling time. The root mean square is obtained by averaging the squares of the data and is represented by the following equation.

As a result of calculating the RMS values for each of the 12 sections, the RMS values of the blades 1, 2 and 4 to 12 are almost the same, but the deviation of the blade RMS of the No. 3 blade occurs, which means that the middle portion of the blade 3 has a defect To improve the reliability. The range of the RMS deviation to determine whether a coupling has occurred can be optionally set by the user.

When the occurrence of defects of the specific blade and the part is confirmed through the detection step as described above, the resultant data is stored and displayed to the user. For example, a defect position discrimination graph as shown in Fig.

7, the upper part (Top) of the blade No. 3 and the middle part of the blade No. 2 outside the middle part of the blade No. 3 determined to have a defect (red square display area) A pressure value appears as input. This result can not be relied upon to see that the blades and parts in question are actually broken or deformed. For example, the momentary deformation over a very short period of time can result in temporary pressure input.

In order to prevent the detection error due to the temporary input value, the frequency of occurrence of defects using Log Odds is reflected in the result, and it is determined that a defect has occurred at the position only when the frequency is high. Also, the defect occurrence probability at the previous sampling time is stored and reflected in the next sampling time. Log Odds Ratio is obtained by dividing the probability of occurrence (p) by the probability of not occurring (1-p) and then taking the log. If p = 1/2, Log Odds becomes 0 and p> 1 / 2, Log Odds becomes positive, and if p <1/2, Log Odds becomes negative.

As described above, if the number of occurrences of defects obtained by the currently input real-time data is equal to or greater than a predetermined ratio using the log-likes, it is determined that a defect has occurred at the corresponding position. Otherwise, it is determined that no defect occurs. In the graph of FIG. 7, the upper part (Top) of the third blade and the middle part (Middle) of the second blade show the result that the occurrence frequency is low and the defect is not caused.

As described above, the blade defect detection method of the present invention learns steady state data for multiple channels and can collect and store data in real time. Based on the learned steady state data, The position can be detected.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the inventions. It will be apparent to those of ordinary skill in the art.

Claims (8)

A normal data learning step of receiving a pressure value sensed by a wind pressure sensor installed in front of or behind a blade of the axial flow fan or the axial flow turbine in a state where the blade is not defective and setting a normal range; And
And a defect detection step of receiving a pressure value detected from a wind pressure sensor installed at the front or rear of the blade at the time of detection and analyzing whether the pressure value belongs to the normal range obtained in the normal data learning step,
Wherein the defect detection step divides the data for one revolution of the fan or the turbine by the same number of sections as the number of the blades to determine whether the blades corresponding to the respective areas have defects,
Wherein the normal data learning step includes analyzing the order of the sensed pressure value data to set a normal range and store the set normal range as order analysis learning data,
Wherein the defect detection step includes a step of analyzing the pressure value data of the current state sensed by the wind pressure sensor and comparing the pressure value data with the order analysis learning data to determine whether or not a defect has occurred in the blade, And a detecting step,
The normal data learning step and the defect detecting step may detect a defect in another part of the blade by receiving a pressure value sensed from a plurality of wind pressure sensors sensing the pressure values of the upper,
The normal data learning step sets the upper and lower limits of the normal range by receiving the pressure value sensed by the wind pressure sensor disposed at the upper, middle, and lower ends of the blade, and the defect detecting step detects the pressure value sensed by the wind pressure sensor Determines that a defect of the blade has occurred if the difference between the upper limit value and the lower limit value is out of the range between the upper limit value and the lower limit value
Blade defect detection method.
delete delete delete The method according to claim 1,
The root mean square of the sensed pressure values is obtained for each divided section according to the number of blades, and the root mean square of each section is compared to determine that a defect has occurred in the blade corresponding to the section in which the deviation occurs.
Blade defect detection method.
The method according to claim 1,
Wherein the defect detection step determines whether or not a defect is generated according to the frequency of the detection data that deviates from the normal range at the sampling time
Blade defect detection method.
delete delete

Images