JPS6239372B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a method and apparatus for diagnosing vibrations of a rotating machine. In particular, a method for diagnosing abnormal vibrations caused by unbalance that occurs in a rotating shaft by detecting the rotational speed when the rotor of a rotating machine rotates and vibrations of bearings at both ends of the rotor and determining the position of the unbalance. Regarding equipment. This type of technology can be used for maintenance of rotating equipment such as turbines. Particularly recently, as thermal and nuclear power plants have increased in capacity, there has been a demand for further improvements in the reliability of their main equipment, such as turbine generators.
In order to supply power smoothly, it is essential to quickly perform maintenance after an abnormality occurs in equipment.
Therefore, vibration diagnosis of this type of rotating machine has become extremely important. There are various causes for abnormalities in rotating machines, and there are also various abnormal phenomena, but there are especially many cases of abnormal shaft vibration, and this is a concern that may lead to serious failures. It is. Furthermore, since the rotating shaft systems used in power generation plants such as turbine generators are long and large-scale systems, even if it is known that an abnormality has occurred, it takes a great deal of time to find the location of the abnormality. be. Against this background, as a means of improving the reliability of rotating machines, it is possible to detect abnormalities in the rotating shaft early from the vibrations generated in the rotating shaft, and to diagnose the type of abnormality and estimate the location of occurrence in a short time. There is a strong need for the development of vibration diagnostic methods and devices. In particular, many of the abnormal vibrations that occur in rotating shafts include vibrations that are synchronized with the rotational speed, that is, unbalanced vibrations.Therefore, there is an urgent need to develop technology that can diagnose the location of unbalance occurrence, which is the source of such unbalanced vibrations. . In other words, in maintenance systems for turbines, etc., functions such as abnormality determination, phenomenon determination, and position evaluation are required, but it is especially important to evaluate the location where unbalanced vibrations occur.
And this is a technical problem that is difficult to solve. However, conventionally, in order to deal with such unbalanced vibrations of rotating machines, the main focus has been on developing vibration reduction technology that reduces unbalanced vibrations. In other words, we have been able to reduce the vibration of the rotating shaft by calculating an equivalent balancing weight that cancels out the unbalance that exists in the rotating shaft, and by adding this balancing weight to the rotating shaft. . However, since this method limits the weight application position to the axial direction, it does not directly consider the diagnosis of unbalanced positions that exist at arbitrary positions in the axial direction. Further, even if the balancing weight can be calculated, it is difficult to correctly diagnose the unbalanced position that exists on the rotating shaft. In short, the above method cannot be used for diagnosing an unbalanced position, and even if there are similarities, it is a different technology from the unbalanced position diagnosing technique. Therefore, this technology can also be used for large-scale rotating shaft systems.
The development of diagnostic technology that can determine the position of unbalance that occurs in a rotating shaft from journal (bearing) vibration remains an issue. The following two techniques can be considered as techniques for diagnosing such an unbalanced position. The first method is to diagnose the unbalanced position by taking into account changes in the vibration mode of the rotor. The second method is to diagnose the unbalanced position by focusing only on the rotational speed at which abnormal vibration occurs. In order to apply the former first technique, vibration monitoring in the operating speed range is required as a condition.
Moreover, it is necessary to include a plurality of vibration modes within the rated rotation speed or below the abnormal vibration generation rotation speed. Therefore, in this first technique, when abnormal vibration occurs near the rated rotational speed, the unbalance position cannot be diagnosed unless the rotational speed is lowered to a low speed. Moreover, diagnosis is limited only when there is a mode change. Thus, with the first technique, it is not always possible to diagnose the position of unbalance for all abnormal vibrations caused by unbalance. On the other hand, the latter second technique does not have the above problem and can diagnose the unbalanced position based on the vibration of the rotational speed at which the abnormality occurs, so the diagnosis can be made at the same time as the abnormality occurs without the need to change the rotational speed. be able to. Therefore, this second technique is effective, and the present inventors have worked diligently to establish a diagnostic method and apparatus using this technique. As mentioned above, the purpose of the present invention is to determine abnormal vibrations occurring in a rotating machine such as a large turbine,
It is an object of the present invention to provide a vibration method and apparatus for diagnosing an unbalanced position, which is a source of unbalanced vibration, in the rotating machine, thereby enabling smooth operation of a rotating machine. In order to achieve this objective, in the present invention,
When diagnosing the vibration of a rotor by detecting the rotation speed and vibration of the bearings at both ends of the rotor when the rotor of a rotating machine rotates, unbalanced vibrations that are synchronized with the rotation speed are extracted from the detected vibration signals, and the unbalanced vibrations that are synchronized with the rotation speed are extracted. The balanced vibration is compared with a reference signal indicating allowable vibration to determine whether the unbalanced vibration is abnormal or not. The method is to calculate the abnormal vibration component by vector subtracting the vibration of the bearing during rotation, and to diagnose the unbalance state by comparing the calculated abnormal vibration component with the pre-stored vibration component at the time of unbalance. Take. In this case, in the case of a multi-span rotor in which a plurality of single rotors are connected and combined in the longitudinal direction, vibration is detected for a pair of bearings supporting both ends of each single rotor. Therefore, for a multi-span rotor, vibration detection will be performed for three or more bearings. When this method is implemented in a diagnostic device, a rotation speed detector is provided to detect the rotation speed of the rotor, and this rotation speed detector also serves as a phase reference signal to suppress bearing vibration. A vibration detector is provided for detection, a vibration analysis is provided to extract unbalanced vibration synchronized with the rotational speed from this vibration signal and the phase reference signal, and it is determined whether or not this unbalanced vibration is abnormal. A judge is provided to judge,
And when it is abnormal, a vector calculator is provided to extract only the abnormal vibration from the abnormal unbalanced vibration and the normal vibration of the bearing in normal conditions stored in advance in the memory device, and further normalizes this abnormal vibration component. A computing unit is also provided for estimating the unbalance position by comparing the abnormal vibration pattern from this computing unit with an unbalanced vibration pattern previously stored in a pattern storage device. A display can be provided and configured to display the results. The above-mentioned vibration pattern refers to the amplitude distribution state at a plurality of locations, such as the six measurement points B ₁ to B ₆ in FIGS. 4a and 4b, which will be described in detail later. Hereinafter, an example of implementation of the present invention will be described with reference to the drawings. In this embodiment, the present invention is applied to vibration diagnosis in a power generation plant using a turbine. FIG. 1 shows this embodiment in a block diagram. This example illustrates a multi-span rotor that combines a total of three single rotors: turbine rotors 1 and 2 and a generator rotor 3, and a total of six bearings 41 to 4.
6 supports both ends of each single rotor. The present invention provides rotors 1, 2, 3 as shown in the drawings.
This is a technique for diagnosing rotor vibration by detecting the rotational speed of the rotor and vibrations of the rotor bearings 41 to 46, but a rotational speed detector 5 is provided to detect the rotational speed of the rotor. . This rotation speed detector 5 can detect the rotation speed of the rotor when the rotation speed increases or decreases, and in the illustrated example, it is arranged near one bearing 41 of the turbine rotor and detects the rotation speed. I'm starting to do that. The signal from the rotation speed detector 5 is also used as a phase reference signal a used to analyze the vibration signal and extract unbalanced vibrations from it. A vibration detector 6 is provided to detect journal vibration of each bearing 41 to 46. In this example, each bearing 41 to 46 is equipped with a sensor 6 for detecting vibration.
1 are arranged, and the detection signal b from each sensor 61 is inputted to the vibration detector 6. The vibration signal c from the vibration detector 6 is sent to a vibration analyzer 7, where it is analyzed and an unbalanced vibration u synchronized with the rotational speed is extracted. That is, since the state of vibration differs depending on the number of rotations, the measured vibration signal c is standardized by analyzing it based on the number of rotations, and then sent to the next determination unit 8. Such analysis is performed using both the phase reference signal a sent from the rotation speed detector 5 and the vibration signal c sent from the vibration detector 6. In this way, the vibration analyzer 7
The unbalanced vibration u analyzed in is sent to the determiner 8, where it is determined whether or not the vibration u is abnormal. This vibration u has already been compared and analyzed with the reference signal a indicating permissible vibration, so if it is permissible, it will not be output from the determiner 8, but it will be passed through the determiner 8 only if it is abnormal. It is configured in advance. Therefore, when the unbalanced vibration component u is determined to be abnormal by the determiner 8,
The abnormal unbalanced vibration u' passes through the determiner 8 and enters the vector calculator 10. That is, when the vibration component u is abnormal, u' = u, and when it is not abnormal,
u′=0. The vector calculator 10 is for extracting only the abnormal vibration component v from the abnormal unbalanced vibration u' and the normal vibration. That is, since normal vibrations are also included in the abnormal unbalanced vibrations u', the abnormal vibrations v caused only by the abnormalities are extracted from comparison with the normal vibrations. For this purpose, a signal U representing normal vibration is required, but this is done by storing the normal vibration of each bearing 41 to 46 in the normal state in advance in the memory 9, and calculating the signal U of normal vibration from there by vector calculation. so that it is sent to container 10. In this way, only the abnormal vibration v due to unbalance occurring in any of the rotors 1, 2, and 3 is extracted, and this vibration v is further sent to the normalization calculator 11, where the abnormal vibration v is normalized by the maximum amplitude. In the present invention, normalization refers to arithmetic processing in which the maximum amplitude among a plurality of vibrations is set to 1, and the amplitudes of other vibrations are expressed as a decimal (or %). This normalization is required in order to next compare various unbalanced vibration patterns with this abnormal amplitude component, and its principle meaning will be described later. Arithmetic unit 1
The signal normalized by 1, that is, the abnormal vibration pattern V
is sent to the pattern recognizer 13, where various unbalanced vibration patterns P are stored in advance.
compared to This storage is performed by the pattern storage device 12, but specifically, in this example, the rotor 1,
An influence coefficient representing the vibration of each bearing journal when a unit unbalance is added to any one point in the axial direction of 2 and 3 is stored. In this case, the amplitude of the influence coefficient is normalized to the maximum amplitude. In the present invention, "adding unit imbalance"
This refers to the operation of attaching an unbalanced weight of an appropriate size and dividing it by the vector mr (unit: grams mm). Therefore, the abnormal signal pattern V in the normalized state and various unbalanced vibration patterns P in the same normalized state are compared in the pattern recognizer 13, and as a result, the unbalanced position occurring in the rotor can be determined. It is estimated. In other words, since the pattern storage device 12 stores patterns P when imbalances occur at various positions,
By sequentially comparing the signal pattern V which normalized the actual abnormal vibration and the various unbalanced vibration patterns P, and selecting the closest pattern P, it is possible to compare the signal pattern V which normalized the actual abnormal vibration and the various unbalanced vibration patterns P, and then select the pattern P that is the closest to the unbalanced vibration pattern P. The corresponding position can be estimated to be the position where the abnormal imbalance actually occurred. The recognition result d obtained in this way is displayed on the display 14.
As the display 14, a CRT, a line printer, etc. can be used. The configuration of this embodiment is shown in a flow diagram in FIG.
Each block in FIG. 2 is given a reference numeral corresponding to that in FIG. Next, the principle of unbalance position diagnosis in the diagnosis technique of this embodiment will be described. To simplify the explanation below, the following symbols are defined. k: Rotation speed number (k = 1, 2, ..., K) i: Bearing position number (i = 1, 2, ..., I) j = Unbalance position number (j = 1, 2, ...,
J) U _i ^(k) : Unbalanced vibration of bearing i at the k-th rotational speed when it is normal U _i ^(k) : Unbalanced vibration of bearing i when the k-th rotational speed is abnormal V _i ^(k) : Bearing Unbalanced vibration due only to unbalance when the kth rotation speed at position i is abnormal V _i ^(k) : Normalized u _i ^(k) (abnormal vibration pattern) P _ij ^(k) : Unit per unbalance position j Normalized vibration 11 of bearing i position at k-th rotation speed when unbalance is added: Vibration amplitude (Note that "normalization" is used in the explanation of V _i ^(k) and P _ij ^(k) above)
means that the maximum amplitude is 1). This will be explained below. First, assume that abnormal vibration occurs at the i-position of the bearing at the k-th rotation speed. At this time, abnormal vibrations are detected by the vibration detector 6 via the sensor 61 and input to the vibration analyzer 7. In this vibration analyzer 7, as mentioned above, the vibration signal c from the vibration detector 6 is analyzed with the phase reference signal a from the rotation speed detector 5, and the unbalanced vibration u _i ^(k)
is extracted, and since it is abnormal, it passes through the determiner 8 and enters the Bertol calculator 10, where v _i ^(k) = u _i ^(k) − U _i ^(k) ...(1) ( i=1, 2, . . . , I) is calculated, and only the abnormal vibration component v _i ^(k) is extracted. In short, vibration u due to unbalance
By cutting out the normal vibration U from the equation, only the abnormal vibration v can be obtained. Abnormal vibration v obtained here
_i ^(k) is a vector quantity including phase and amplitude information. Next, the arithmetic unit 11 for normalization calculates v _i ⁽
Find the ^maximum value of |v|=|v _i ^(k) |max ...(2), and further calculate V _i ^(k) =v _i ^(k) /|v| ...(3) (i=1 , 2, ..., I). As a result, V _i ^(k) is a vector quantity having phase and amplitude information, and |V _i ^(k) |max=
1 (k: const,) is shown.
Therefore, normalization is performed here. An example of such a pattern is shown in the graph of FIG. This is data when the rotational speed is set to k rpm, and the horizontal axis of the figure is the measurement position i, and B ₁ , B ₂ , ..., BI are the respective bearings 41, This corresponds to the data in 42, . The vertical axis in the figure is the normalized vibration amplitude of vibrations caused only by anomalies, and for example, as shown by _B2 , it takes 1 at most. On the other hand, the pattern memory 12 stores various unbalance positions j (j=1, 2, ...,
Influence coefficient P when unit imbalance is added to J)
_ij ^(k) is memorized. This influence coefficient P _ij ^(k) is also normalized and is a vector quantity having information on phase and amplitude, |P _ij ^(k) |max=|(i,
k: const, ) indicates the pattern of unbalanced position and vibration. An example of this pattern is shown graphically in FIG. Figures a and b are both when the rotational speed is k rpm, but a is when the unbalance occurs at the position j = 1, that is, the position illustrated as j1 in Figure 1. Bearings 41, 42,......,
46 is a graph showing that the unbalanced vibration amplitudes of B ₁ , B ₂ , . . . , B ₆ were present. For b, j
When an imbalance occurs at the position =2, that is, the position indicated by j2 in Fig. 1, the bearings 41, 42,...,
46 shows the unbalanced vibration amplitude indicated by numeral 46. The horizontal axis of the figure indicates the measurement position as in Figure 3, and as mentioned above, B ₁ , B ₂ , ..., B ₆ correspond to bearings 41, 42, ..., 46 in Figure 1, respectively. do. The vertical axis shows the unbalanced vibration amplitude, which is also normalized so that the maximum amplitude is 1. The pattern recognizer 13 performs pattern recognition using the pre-stored unbalanced vibration amplitude pattern |P _ij ^(k) | and the abnormal vibration pattern, that is, the abnormal vibration amplitude pattern obtained by performing various operations after the above-mentioned actual measurement. This is what we do. As mentioned above, various patterns |P _ij ^(k) |, which are normalized signals obtained when unit imbalances are given to various specific positions j, are stored in the pattern memory 12 in advance, so this and actual inspection measurements are stored in the pattern memory 12. The abnormal vibration patterns derived from |V _i ^(k) | are compared and the most similar one is identified.
Note that in this operation example, only the amplitude is considered when comparing patterns, and phase information is ignored (an example in which phase information is also considered will be described later). The pattern comparison operation in this example is performed as follows. First, we focus on the amplitude pattern of abnormal vibration |V _i ^(k) |, and in descending order of the values, |V _i1 ^(k) |, |V _i2 ^(k) |, ..., |V _it ^(k) | Arrange like this. Next, from the stored unbalanced vibration amplitude pattern |P _ij ^(k) |, i=i1.
Find the combination of 1 and j. Furthermore, for j where i=i1, a combination of i2 and j' where i=i2 is found. By repeating this operation, it can finally be estimated that the unbalanced vibration amplitude pattern having the combination of i and j that has been repeated many times is close to the abnormal vibration amplitude pattern that has occurred in the rotor. In other words, from among various stored unbalanced vibration amplitude patterns |P _ij ^(k) |, select a pattern in which the bearing position exhibiting the maximum amplitude is i1, and then select a pattern that exhibits the second largest amplitude from the selected pattern. By selecting the pattern in which the bearing position is i2, and selecting the third, fourth, etc. in this way, it can be estimated that the last remaining pattern is closest to the abnormal amplitude pattern that occurred in the rotor. Therefore, the unbalance position that is most similar among the stored patterns indicates the unbalance position that actually occurs in the loader. Next, the abnormal vibration amplitude |V _i ^(k) | and the unbalanced vibration amplitude for each unbalanced position |P _ij ^(k)
The following calculation is performed between |. R _j =√ _i ...(5) (j=1, 2, ..., J) The calculation of equations (4) and (5) calculates the actual abnormal vibration amplitude |V _i ^(k) | This represents the degree of deviation of the stored and given unbalanced vibration amplitude pattern |P _ij ^(k) | with respect to . Therefore _, among the various positions j related to the various memory patterns | _{P ij} ^(k) It will show the balance position. Therefore, it is possible to diagnose the unbalanced position that has occurred in the rotor from the unbalanced position found by the above two operations, that is, the combination of i and j, and the unbalanced position that takes the minimum value of equations (4) and (5). Can be done. The following table shows an example of this pattern recognition.

[Table] Among the judgment descriptions in this table, 〇 indicates that an unbalanced position has been determined, △ indicates that there is a possibility of an unbalanced position, and × indicates that there is no such possibility. In other words, with the data in this table, each bearing 41,
It is shown that among the abnormal vibration amplitudes |V _i ^(k) | detected from ₄₂ , . On the other hand, when a unit unbalance is added to the unbalance positions j1, j2..., the maximum amplitude of the unbalance vibration amplitude pattern |P _ij ^(k) It can be seen that the objects are B ₁ , B ₂ , B ₃ , B _{4 .} . . for each position j1, j2 . Therefore, there are only two locations where the two match and the patterns are approximate, j1 and j2, and it is presumed that an imbalance actually occurs at one of these locations. On the other hand, if the values of S or R indicating the deviation are taken in descending order, the position j
It can be seen that the data at 2 is the minimum. Therefore, there is an extremely high probability that an unbalance occurs at the position where the results of this operation match, that is, position j2. However, since the deviation at position j3 is also the second smallest and the amplitude peaks also match, it can be determined that there is a possibility. As a result, by diagnosing the unbalance position through the above operation, it becomes possible to quickly find out the source of the rotor unbalance abnormality that occurs during operation.
Moreover, according to the above, it is also possible to automate the diagnosis of unbalanced positions, which can contribute to the automation of abnormality discovery. Moreover, it goes without saying that maintenance and inspection of a device using a rotating machine becomes easier. Note that in the above embodiment, the phase information is ignored,
Although the diagnosis was made based only on the vibration amplitude pattern, an effective diagnosis can also be made by considering phase information. However, in order to use such a diagnostic technique, it is necessary not only to normalize the amplitude pattern but also to standardize the entire axial position using the maximum amplitude as a reference (=0) for the phase. With respect to this standardized phase θ, |θ|90° is considered to be in phase, and 90°<θ<270° is considered to be out of phase. As a result, the pattern that appeared as shown in FIG. 3 in the above embodiment becomes as shown in FIG. 5. In addition, by using V _i ^{(k) for} |V _i (k) | and P _ij ^(k) for |P _ij ( ^k) |, we can examine not only the amplitude pattern but also the relationship between phase and amplitude. Comparisons are made using vector quantities that have both types of information. As detailed above, the vibration diagnosis technology for a rotating machine of the present invention diagnoses the vibration of a rotor of a rotating machine by detecting the rotational speed and the vibration of the rotor both end bearings when the rotor rotates. , extracts unbalanced vibration synchronized with the rotational speed from the detected vibration signal, compares the extracted unbalanced vibration with a reference signal indicating allowable vibration, and determines whether or not the unbalanced vibration is abnormal. If an abnormality is determined, the vibration caused only by the abnormality is determined by vector subtraction from the unbalanced vibration and the vibration of the bearing during normal rotation stored in advance, and the abnormal vibration pattern is normalized and determined. This system is characterized by diagnosing the unbalanced position of the rotating shaft by comparing the abnormal vibration pattern with a pre-stored unbalanced vibration pattern, so abnormal vibrations occurring in the rotating machine can be quickly determined. In addition, the unbalance position, which is the source of unbalanced vibration, can be quickly diagnosed.
Therefore, it can be effectively applied to large-sized rotating machines, and has the effect of achieving smooth and stable operation of various types of rotating machines. Note that, although the above-described embodiments have various other effects depending on their specific configurations, the present invention is not limited only to these embodiments.

[Brief explanation of the drawing]

1 to 4 show an example of the implementation of the present invention, FIG. 1 is a block diagram thereof, FIG. 2 is a flowchart thereof, and FIG. 3 is an example of a normalized vibration pattern of an abnormal component. FIG. 4 is a graph showing an example of an unbalanced vibration pattern stored in advance. FIG. 5 is a graph showing an example of a normalized vibration pattern of an abnormal component when a modification of the above example is used. 1, 2, 3...Rotor, 41-46...Bearing,
5... Rotation speed detector, 6... Vibration detector, 7...
Vibration analyzer, 8... Judgment device, 9... Storage device, 1
0... Vector calculator, 11... Normalization calculator, 12... Pattern memory, 13... Pattern recognizer, 14... Display, k... Number of rotations (number), a... Phase Reference signal, c... Vibration signal, u
...Unbalanced vibration, u'...Unbalanced vibration (determined to be abnormal), U...Normal (unbalanced) vibration, v...Abnormal (only) vibration, V...
...normalized abnormal vibration pattern, P... (previously memorized) unbalanced vibration pattern, j,
j1, j2 ~ jJ... Unbalanced position.

Claims (1)

[Scope of Claims] 1. A vibration diagnosis method for diagnosing vibration of a rotor by detecting the rotational speed of a rotor of a rotating machine and vibrations of a plurality of rotor bearings supporting the rotor, comprising: a. If it is normal, measure and memorize the unbalanced vibration of each bearing when the rotor is rotated at a constant rotation speed, b. Measure in advance the unbalance vibration when rotating at the constant rotation speed with unit unbalance added, normalize and store this, c. Rotate the rotor to be diagnosed at the constant rotation speed. d. Extract unbalanced vibration synchronized with the rotational speed from the detected vibration signal, and compare the extracted unbalanced vibration with a reference signal indicating allowable vibration to determine that the unbalanced vibration is abnormal. e) If it is determined to be abnormal, calculate and normalize the abnormal vibration component by subtracting the normal vibration from the detected vibration signal as a vector; A method for diagnosing vibrations in a rotating machine, comprising diagnosing the axial position of unbalance by comparing each bearing with the vibration at the time of unbalance. 2. Claim 1, wherein the calculated abnormal vibration component and the stored unbalanced vibration are compared based on the magnitude of both vibrations.
Vibration diagnosis method for rotating machines as described in . 3 Comparison of the above-mentioned calculated abnormal vibration component and the memorized unbalance vibration is made by comparing the sum of squares or the square root of the average value of the sum of squares of the difference between the abnormal vibration component of both vibrations and the unbalance vibration amplitude. A vibration diagnosis method for a rotating machine according to claim 1, characterized in that: 4. A vibration diagnosis device for diagnosing vibration of a rotor of a rotating machine by detecting the number of rotations when the rotor rotates and the vibrations of a plurality of bearings supporting the rotor, which includes: a) detecting the number of rotations of the rotor; a rotation speed detector that generates a phase reference signal; b. a vibration detector that detects vibrations of each of the bearings; c. a vibration analyzer that extracts synchronized unbalanced vibration; d The amplitude of the unbalanced vibration extracted by the vibration analyzer is compared with a reference signal indicating allowable vibration, and based on the magnitude relationship, it is determined that the unbalanced vibration is abnormal. a determiner that determines whether or not; a vector subtractor that extracts only the component; f a computing unit that normalizes the abnormal vibration component extracted by the vector subtractor; g a pattern of abnormal vibration normalized by the computing unit; It comprises: an arithmetic unit having a function of estimating an unbalance position by comparing unbalanced vibrations stored in a pattern memory; and (h) a display device that displays the results estimated by the arithmetic unit. A vibration diagnosis device for a rotating machine characterized by the following.