Method for detecting abnormal vibration of rolling mill, abnormality detection device, rolling method, and method for manufacturing metal strip

Abstract

False detection caused by noise and the like generated from peripheral equipment of a rolling mill is prevented, and abnormal vibration is extracted and evaluated with high accuracy. The abnormal vibration detection method of a rolling mill includes a collection step of collecting vibration data of the rolling mill; a frequency analysis step of performing frequency analysis of the vibration data to generate first analysis data indicating vibration intensity of each frequency; a data conversion step of converting the first analysis data into second analysis data indicating vibration intensity of each pitch on the basis of a rolling speed; and a map generation step of generating a vibration map obtained by arranging a plurality of the second analysis data along a time series.

Description

TECHNICAL FIELD

[0001] The present application relates to a method of detecting vibration generated in a rolling mill that makes a steel sheet into a prescribed sheet thickness, and particularly, to a method of detecting abnormal vibration of a rolling mill that generates defects on a surface of a steel sheet, an abnormality detection device, a rolling method, and a manufacturing method of a metal strip. BACKGROUND

[0002] Generally, a steel sheet used for automobiles, beverage cans, and the like is subjected to continuous casting, hot rolling, and cold rolling, and after an annealing process and a plating process, is processed in accordance with the respective uses. The cold rolling process is the final process that determines the sheet thickness of the steel sheet as a product. Since the surface of the steel sheet before plating determines the surface of the final product after plating, a function of preventing surface defects is required in the cold rolling process.

[0003] As one of the surface defects generated in the cold rolling process, a chatter mark can be cited. The chatter mark is a pattern in which a linear mark extending in the width direction of the metal strip periodically appears in the length direction of the metal strip, and is considered to be generated mainly due to vibration (chatter) of the rolling mill. A very slight chatter mark is not found in visual inspection, sheet thickness measurement, and the like after rolling, and is found after the plating process, and thus becomes a factor that greatly hinders productivity. In addition, it is known that in thin materials such as a can steel sheet, an electromagnetic steel sheet, and the like, a phenomenon such as sheet breakage occurs due to a sharp change in sheet thickness, tension, and the like caused by chatter, and hinders production.

[0004] Conventionally, from the viewpoint of hindrance to productivity and prevention of surface defects, various methods of detecting chatter and methods of preventing chatter have been developed (for example, refer to Patent Documents 1 to 3). In Patent Document 1, it is disclosed that a vibration detector is installed to a rolling mill, and frequency analysis is performed on vibration obtained by the vibration detector and rolling parameters. It is described that the following method of detecting chatter is performed: the fundamental frequency that can be generated for each vibration generation factor is calculated at the same time, and in a case where vibration energy at a frequency that is an integral multiple of the fundamental frequency that can be generated for each generation factor in the frequency analysis result exceeds a set value, it is determined that chatter has occurred.

[0005] In Patent Documents 2 and 3, the following method of detecting and method of preventing chatter are described: in the method of detecting, a vibration detector is disposed not only to the main body of the rolling mill but also to a roll (small diameter roll) disposed between each stand and the entry and exit side of the cold rolling mill and around which the metal sheet is wound at an angle of 90 degrees or more, frequency analysis is performed on the obtained vibration value, and in a case where the vibration value exceeds a threshold value at a frequency that coincides with the chordal vibration frequency of the steel sheet, it is determined that chatter has occurred, and in the method of preventing chatter, the chordal vibration frequency is controlled so as not to coincide with the fundamental frequency of the rolling mill by controlling the tension.

[0006] PRIOR ART DOCUMENTS

[0007] PATENT DOCUMENTS

[0008] Patent Literature 1: Japanese Patent Application Laid-Open No. 08-108205

[0009] Patent Literature 2: Japanese Patent Application Laid-Open No. 2016-153138

[0010] Patent Literature 3: Japanese Patent Application Laid-Open No. 2016-2582 SUMMARY

[0011] PROBLEMS TO BE SOLVED BY THE INVENTION

[0012] However, in the case of Patent Literature 1, noise generated from the peripheral equipment of the rolling mill and vibration generated from the vibration source provided to the main body of the rolling mill are also detected at the same time, and many false detections are generated. In addition, in the cases of Patent Literatures 2 and 3, although generation of vibration caused by string vibration can be suppressed, it is difficult to detect vibration other than this as a vibration source. Furthermore, it is difficult to determine the frequency at which chatter is generated in advance, and often the frequency of chatter is recognized after the vibration in a certain frequency band becomes large. Thus, even if a certain frequency is focused on in advance and a threshold value corresponding to the amplitude and the like corresponding to the frequency is set, it is difficult to detect chatter with high precision. In particular, in a continuous cold rolling mill (tandem rolling mill), the conveyance speed (rolling speed) of the metal strip differs for each stand. Due to this, the rotational speed of the work roll differs for each stand, and vibration of a plurality of frequencies overlaps, and detection of chatter becomes difficult. That is, in a method in which the frequency of chatter is determined in advance and the intensity of vibration in the frequency band is detected as in the conventional technique, there is a problem that generation of chatter marks caused by slight vibration cannot necessarily be prevented.

[0013] The present application has been achieved in view of the above problems, and aims to provide an abnormal vibration detection method of a rolling mill, an abnormal detection device, a rolling method, and a manufacturing method of a metal strip, which detect abnormal vibration that causes chatter marks with high precision.

[0014] MEANS FOR SOLVING THE PROBLEMS

[0015] [1] An abnormal vibration detection method of a rolling mill having one pair of work rolls and a plurality of backup rolls that support the work rolls, comprising: a collection step of collecting vibration data of the rolling mill; a frequency analysis step of performing frequency analysis of the vibration data to generate first analysis data that indicates the intensity of vibration for each frequency; a data conversion step of converting the first analysis data into second analysis data that indicates the intensity of vibration for each interval on the basis of a rolling speed; and a map generation step of generating a vibration map in which a plurality of the second analysis data are arranged along a time series.

[0016] [2] The abnormal vibration detection method of a rolling mill according to [1], further comprising a principal component analysis step of performing principal component analysis on the second analysis data using reference data representing a normal state, determining a deviation component of each pitch calculated as a residual error of a projection of the second analysis data with respect to the reference data, the mapping generation step further generating a deviation component map obtained by arranging a plurality of the deviation components of each pitch extracted by the principal component analysis step in time series.

[0017] [3] The abnormal vibration detection method of a rolling mill according to [2], in the principal component analysis step, a plurality of principal components used as the reference data are set in a manner that a cumulative value of a contribution rate of a principal component when principal component analysis is performed on normal analysis data acquired when rolling is performed using the normal rolling mill becomes a reference contribution rate or more.

[0018] [4] The abnormal vibration detection method of a rolling mill according to any one of [1] to [3], the rolling mill being a cold rolling mill.

[0019] [5] An abnormality detection device of a rolling mill, which is an abnormality detection device of a rolling mill having one pair of work rolls and a plurality of backup rolls that support the work rolls, wherein a data collection section that collects vibration data of the rolling mill, a frequency analysis section that performs frequency analysis of the vibration data to generate first analysis data representing a vibration intensity of each frequency, a data conversion section that converts the first analysis data into second analysis data representing a vibration intensity of each pitch based on a rolling speed, and a mapping generation section that generates a vibration map obtained by arranging a plurality of the second analysis data in time series are provided.

[0020] [6] The abnormality detection device of a rolling mill according to [5], further comprising a principal component analysis section that performs principal component analysis on the second analysis data using reference data representing a normal state, determines a deviation component of each pitch calculated as a residual error of a projection of the second analysis data with respect to the reference data, the mapping generation section further generates a deviation component map obtained by arranging a plurality of the deviation components of each pitch extracted by the principal component analysis section in time series.

[0021] [7] A rolling method comprising a backup roll replacement step of using the abnormal vibration detection method of a rolling mill according to any one of [1] to [4], previously setting a monitoring pitch corresponding to the rolling mill, and replacing a backup roll of the rolling mill in a case where a vibration intensity at the monitoring pitch of a vibration map or a deviation component map generated in the mapping generation step exceeds a previously set boundary vibration intensity.

[0022] [8] A method of manufacturing a metal strip, comprising a step of manufacturing a metal strip using the rolling method described in [7] above.

[0023] Effects of Invention

[0024] According to the present application, a vibration map obtained by arranging a plurality of second analysis data, which is to be converted into vibration intensity for each pitch, along a time series is created. Thus, false detection caused by noise and the like generated from the peripheral equipment of the rolling mill can be prevented, and abnormal vibration can be extracted and evaluated with high precision. As a result, operation of the rolling mill, in which abnormal vibration is prevented or suppressed, can be performed, defects generated on the surface of the metal strip due to abnormal vibration can be prevented or suppressed, and a metal strip with excellent appearance can be manufactured. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a schematic view showing an example of a rolling apparatus of an abnormality detection device of a rolling mill to which the present application is applied.

[0026] Figure 2 is a functional block diagram showing a preferred embodiment of the abnormality detection device of the rolling mill of the present application.

[0027] Figure 3 is a graph showing an example of a deviation component map in Example 1.

[0028] Figure 4 is a graph showing an example of a deviation component map in Example 2.

[0029] Figure 5 is a graph showing an example of a deviation component map in Example 2 in which the rolling speed is 800 mpm or more and 850 mpm or less.

[0030] Figure 6 is a functional block diagram showing another preferred embodiment of the abnormality detection device of the rolling mill of the present application.

[0031] Figure 7 is an example of time-series vibration data collected by any one of a plurality of vibration meters in the collection step.

[0032] Figure 8 is an example of vibration intensity for each frequency generated in the frequency analysis step.

[0033] Figure 9 is an example of second analysis data of vibration intensity for each pitch after conversion in the data conversion step.

[0034] Figure 10 is an example of a vibration map generated in the map generation step.

[0035] Figure 11is a graph showing an example of a vibration map with respect to the second analysis data in Embodiment 3.

[0036] Figure 12 is a graph showing an example of a vibration map with respect to the first analysis data in Comparative Example. DETAILED DESCRIPTION

[0037] Hereinafter, a method for detecting abnormal vibration of a rolling mill, an abnormality detection device, a rolling method, and a method for manufacturing a metal strip relating to an embodiment of the present application will be described with reference to the accompanying drawings. Figure 1 is a schematic view showing an example of a rolling apparatus of an abnormality detection device of a rolling mill to which the present application is applied. Figure 1 The rolling apparatus 1 of is a cold rolling apparatus that cold-rolls a steel strip S as a metal strip S, for example, and four rolling mills 2A, 2B, 2C, 2D (4 stands) are arranged along a rolling direction. Each of the rolling mills 2A, 2B, 2C, 2D has the same structure, and includes a housing 3, a pair of work rolls 4 housed in the housing 3 and rolling the metal strip S, a plurality of backup rolls 5 supporting the work rolls 4, and a driving device 6 that rotationally drives the work rolls 4. In addition, a small-diameter roll 7 on which the metal strip S to be rolled is hung is provided on the downstream side of each of the rolling mills 2A, 2B, 2C, 2D in the rolling direction of the metal strip S.

[0038] A vibration meter 8A, 8B, 8C, 8D is attached to the housing 3 of each of the rolling mills 2A, 2B, 2C, 2D. The vibration meters 8A, 8B, 8C, 8D measure the vibration generated in the rolling mills 2A, 2B, 2C, 2D, and are constituted by, for example, an acceleration sensor. Note that the vibration meters 8A, 8B, 8C, 8D are not limited to the housing 3 as long as they are provided at a position at which the vibration of the rolling mills 2A, 2B, 2C, 2D can be detected, and can be provided at, for example, each roll chock or the small-diameter roll 7 on which the metal strip S to be rolled is hung.

[0039] Specifically, in a case where the small-diameter rollers 7 are provided with the vibration meters 8A, 8B, 8C, 8D, the vibration data acquired by the vibration meters 8A, 8B, 8C, 8D can be considered to correspond to the vibration of the rolling mills 2A, 2B, 2C, 2D disposed on the upstream side of the small-diameter rollers 7 provided with the vibration meters 8A, 8B, 8C, 8D in the rolling direction of the metal strip S. The rolling speed in the present embodiment refers to the circumferential speed of the work roll 4 in the rolling mill 2A, 2B, 2C, 2D or the conveyance speed (outlet side speed) of the metal strip S on the outlet side of the rolling mill 2A, 2B, 2C, 2D. The rolling speed is determined for each of the rolling mills 2A, 2B, 2C, 2D provided with the vibration meters 8A, 8B, 8C, 8D (in the following description, the portion provided with the vibration meters 8A, 8B, 8C, 8D is sometimes referred to as a stand). In addition, in a case where the small-diameter rollers 7 are provided with the vibration meters 8A, 8B, 8C, 8D, the vibration data acquired by the vibration meters 8A, 8B, 8C, 8D corresponds to the rolling speed of the rolling mills 2A, 2B, 2C, 2D disposed on the upstream side thereof. In addition, the standard rolling speed in the present embodiment is an arbitrary rolling speed set for each of the rolling mills 2A, 2B, 2C, 2D. The standard rolling speed can be a rolling speed empirically known as a rolling speed at which chatter is likely to occur. For example, as the standard rolling speed of the final stand 2D, 900 m / min can be selected from the speed range of 800 m / min or more and 1300 m / min or less at which chatter is likely to occur. In this case, the standard rolling speeds of the rolling mills 2A, 2B, 2C on the upstream side of the final stand 2D can be set respectively in accordance with the pass schedule set as a standard on the basis of the standard rolling speed set for the final stand 2D.

[0040] Figure 2 is a functional block diagram showing a preferred embodiment of the abnormality detection device of the rolling mill of the present application. Note that, Figure 2 The structure of the abnormality detection device 10 of the rolling mill of the present application is constructed by hardware resources such as a computer. The abnormality detection device 10 of the rolling mill detects abnormal vibration of the rolling mill 2A, 2B, 2C, 2D that causes chatter marks, and is provided with a data collection section 11, a frequency analysis section 12, a data conversion section 13, and a map generation section 14. In addition, the abnormality detection device 10 can be provided with a principal component analysis section 15 described later.

[0041] The data collection section 11 collects vibration data detected by each vibration meter 8A, 8B, 8C, 8D. In the case where the vibration meters 8A, 8B, 8C, 8D are acceleration sensors, acceleration data of vibration is transmitted from the vibration meters 8A, 8B, 8C, 8D to the data collection section 11. The data collection section 11 continuously acquires the acceleration data. Also, the data collection section 11 time-integrates acceleration data measured within a data sampling time (for example, a period of 0.2 seconds) set in advance among the acquired acceleration data to convert it into velocity data, and collects it as vibration data at each time, that is, at each data sampling time. As a result, the vibration data becomes vibration velocity arranged in time series.

[0042] In addition, the data collection section 11 performs measurement and calculation of vibration data for a data sampling time of, for example, 0.2 seconds at a data acquisition cycle set in advance (for example, every 1 second). The data sampling time in the continuous cold rolling mill is preferably set to 0.1 seconds or more and 1 second or less, and the data acquisition cycle is preferably set to 1 second or more and 5 seconds or less. This is because, in the case where the data sampling time is shorter than 0.1 second, it can not be possible to obtain data to the extent that the degree of vibration of the rolling mill can be determined, and if it exceeds 1 second, the calculation load of frequency analysis and the like can become high, so these cases are avoided. In addition, this is because, in the case where the data acquisition cycle is shorter than 1 second, the calculation load of frequency analysis and the like becomes high, and if it exceeds 5 seconds, it can become difficult to perform detection of abnormal vibration early, so these cases are avoided. Note that, in the example shown here, an example is illustrated regarding the case where the data collection section 11 collects vibration data from each vibration meter 8A, 8B, 8C, 8D, but it can be configured to be able to collect vibration data from any one of the vibration meters 8A, 8B, 8C, 8D. This is because, based on vibration data collected by any one of the vibration meters 8A, 8B, 8C, 8D, it is possible to reliably detect chatter in the rolling mill (stand) 2A, 2B, 2C, 2D in which the vibration meter 8A, 8B, 8C, 8D is provided. Note that, for the vibration meters 8A, 8B, 8C, 8D, not only acceleration sensors, but also position sensors, velocity sensors that can measure vibration can be used. This is because acceleration, velocity, and displacement (displacement amount) data can be converted to each other by time integration and time differentiation.

[0043] The frequency analysis section 12 performs frequency analysis on the vibration data collected by the data collection section 11 in the data sampling time, and generates analysis data (hereinafter, sometimes referred to as first analysis data) composed of the vibration intensity of each frequency every data acquisition period. The frequency analysis section 12 extracts the amplitude and phase of the vibration velocity of each frequency, for example, by Fourier transform, and extracts the absolute value of the amplitude of the vibration velocity at each frequency as the vibration intensity. Note that the frequency after Fourier transform of digital data becomes a discrete value by the number of data of Fourier transform and the sampling frequency.

[0044] In the present embodiment, the frequency at which the frequency analysis section 12 performs frequency analysis is set to a plurality of frequencies, which are referred to as reference frequencies. The reference frequencies can be arbitrarily selected from a frequency band of 1 / 2 or less of the sampling frequency of the vibration meter 8A, 8B, 8C, 8D, with the sampling frequency as a reference. The sampling frequency refers to the number of times of measurement of vibration (e.g., acceleration) per second by the vibration meter, and differs depending on the specifications of the vibration meter used. In the present embodiment, the sampling frequency of the lowest vibration meter among the sampling frequencies of the plurality of vibration meters 8A, 8B, 8C, 8D can be used as a representative value. The reference frequencies are preferably selected from a frequency band of 1 / 2 or less of the sampling frequency, and the number of frequencies is preferably 20 or more and 1600 or less. This is because, if the number of reference frequencies is less than 20, it can be difficult to detect the occurrence of chattering, and if the number of reference frequencies exceeds 1600, it is necessary to set the data acquisition period long in a manner that avoids an excessive increase in the computational load of the frequency analysis section 12, and it can be difficult to detect the occurrence of chattering early, and thus these situations are avoided. The number of reference frequencies is more preferably 200 or more and 800 or less. For example, the frequency analysis section 12 sets the sampling frequency of the vibration meter 8A, 8B, 8C, 8D to 5120 Hz, sets the reference frequencies at intervals of 5 Hz in a range of 5 Hz or more and 2000 Hz or less (a total of 400), and analyzes the vibration intensity for each reference frequency. Note that the frequency analysis section 12 is not limited to Fourier transform, and can use a known frequency analysis method such as wavelet transform, windowed Fourier transform, and the like, as long as it can analyze the vibration data into the vibration intensity of each frequency. In this case, the reference frequencies can be set using the same method as described above.

[0045] The data conversion section 13 converts the analysis data of the vibration intensity for each reference frequency, that is, the first analysis data, into the vibration intensity for each pitch based on the rolling speed. The data conversion section 13 converts the first analysis data indicating the vibration intensity corresponding to the reference frequency into the second analysis data indicating the vibration intensity for each pitch for each of the rolling mills 2A, 2B, 2C, 2D on which the vibration meters 8A, 8B, 8C, 8D are provided (the rolling mills 2A, 2B, 2C, 2D corresponding to the vibration meters 8A, 8B, 8C, 8D). Here, the pitch in the present embodiment is an index corresponding to the distance in the length direction of the metal strip S or the circumferential distance of the work roll 4 of the rolling mill 2A, 2B, 2C, 2D in relation to the frequency of the vibration. That is, the pitch means the interval of the vibration peaks that become adjacent to each other in the length direction of the metal strip S or the circumferential direction of the work roll 4 as a result of the above-described data conversion in the data conversion section 13. Specifically, the pitch P (mm) is related to the rolling speed V (m / min) and the frequency f (Hz) of the vibration by the following equation.

[0046] P = (1000 x V) / (f x 60) (1)

[0047] Note that the above-described "based on the rolling speed" means that the rolling speed V is used as indicated by the equation (1) to convert the first analysis data into the vibration intensity for each pitch.

[0048] In the data conversion section 13, a standard pitch is stored as a pitch corresponding to the standard rolling speed. The standard pitch is a pitch calculated from the reference frequency f of the frequency analysis performed by the frequency analysis section 12 and the standard rolling speed V according to the above (1). The standard pitch thus set is a plurality of discrete numerical strings corresponding to the reference frequency. The reason for using the standard pitch in the present embodiment is as follows. That is, the rolling speed in the case where the metal strip S is rolled by the rolling mills 2A, 2B, 2C, 2D is not necessarily constant, and the rolling speed also varies within the metal strip S when one metal strip S is rolled. Thus, even if vibrations are generated at the same pitch, if the rolling speed is different, the vibrations are measured as vibrations of different frequencies. In this case, if vibrations of a plurality of frequency bands overlap, it is not possible to definitely grasp whether the causes of the vibrations are the same in the case where the rolling speed has changed. Therefore, in order to evaluate vibration phenomena generated from the same vibration source and observed at different frequencies according to the rolling speed using a unified index, the standard pitch is set. That is, with respect to a vibration source that generates vibrations at a constant pitch, vibration behavior observed as vibrations of different frequencies due to a difference in the rolling speed is converted into vibration behavior corresponding to the standard rolling speed, and is expressed as the vibration intensity of each pitch. Thus, it is possible to evaluate the vibration intensity at an arbitrary rolling speed actually obtained in operation using a constant index of the vibration intensity corresponding to the standard pitch.

[0049] Also, the data conversion section 13 converts the vibration intensity of each reference frequency (first analysis data) into the vibration intensity of each standard pitch (second analysis data) by performing data interpolation such as interpolation or extrapolation. At this time, the interpolation can use linear interpolation, and with respect to the direct current component whose frequency component is "0", interpolation is performed as "0". In addition, with respect to the frequency that becomes extrapolation, all are set to "0". Thus, even if the rolling speed differs for each metal strip, it is possible to evaluate the frequency at which an abnormality has occurred using a constant index of the standard pitch. Note that in the following description, from the viewpoint of determining an abnormality of the vibration corresponding to a specific pitch, in the case where it is said that "pitch", it is used as the meaning of "standard pitch" established by the correspondence of the reference frequency and the standard rolling speed. That is, unless specifically described, "pitch" is synonymous with "standard pitch".

[0050] Here, the vibrations measured by the vibration meters 8A, 8B, 8C, 8D provided to the rolling mills 2A, 2B, 2C, 2D are described. In the vibration meters 8A, 8B, 8C, 8D, vibrations caused by rotation of the work rolls 4 and the like and vibrations of the natural period of the rolling mills 2A, 2B, 2C, 2D are measured superimposedly. The former vibrations vary depending on the rolling speed, and the latter vibrations are measured as vibrations independent of the rolling speed. Thus, if the rolling speed varies, the frequency of the vibrations measured by the vibration meters 8A, 8B, 8C, 8D with respect to the vibrations caused by rotation of the work rolls 4 and the like varies. On the other hand, with respect to the vibration intensity corresponding to the vibrations of the natural period of the rolling mills 2A, 2B, 2C, 2D, although there is no large variation in the frequency of the vibrations, the magnitude (amplitude) of the vibration intensity often varies. According to such characteristics of the vibrations of the rolling mills, in a method of focusing on a specific frequency and detecting abnormal vibrations of the rolling mills based on the vibration intensity at the frequency, even if abnormalities corresponding to the vibrations of the natural period of the rolling mills 2A, 2B, 2C, 2D can be detected, abnormalities related to rotating bodies such as the work rolls 4, the backup rolls 5, and bearing portions thereof in the rolling mills 2A, 2B, 2C, 2D are difficult to detect. In contrast to this, in the present embodiment, the vibration intensity converted to each standard pitch is calculated in the case where the rolling speed is different, and thus abnormalities of the vibrations caused by rotation occurring at a specific pitch are easily detected.

[0051] The map generation section 14 creates a vibration map obtained by arranging along the time series the plurality of second analysis data after the conversion of the transformation generated by the data conversion section 13 into the vibration intensity of each pitch (refer to the vibration map described later Figure 3 ). By generating and displaying such a vibration map, the occurrence or the precursor of abnormal vibrations caused by rotating bodies of the rolling mills 2A, 2B, 2C, 2D that become the cause of the scabs can be detected. In particular, in the case where abnormal vibrations occur or progress, by referring to the vibration map, the behavior in which the vibration intensity corresponding to a specific pitch increases over time can be visually captured. Thus, compared to a detection method based only on the amplitude information at the current time point measured by the vibration meters 8A, 8B, 8C, 8D, the occurrence of abnormal vibrations can be clearly recognized. Note that the vibration map is preferably generated by arranging along the time series the vibration intensity of each pitch generated every data acquisition period. However, instead of arranging all the vibration intensities every data acquisition period, the vibration intensities every certain period can be excluded and displayed.

[0052] The reason for generating the vibration map by the map generation section 14 in the present embodiment is based on the following. That is, in the chatter (vibration of the rolling mills 2A, 2B, 2C, 2D) that causes the chatter marks, the chatter that is generated at a constant pitch due to the rotational motion of the equipment that constitutes the rolling mills 2A, 2B, 2C, 2D is more. For example, in the case where a defect is generated in the speed reducer that drives the rolling mills 2A, 2B, 2C, 2D, the frequency of the vibration varies depending on the rolling speed, but even if the rolling speed varies, the pitch is constant. In addition, in the case where an uneven shape is formed in the circumferential direction of the backup roll 5, specifically, for example, in the case where the backup roll 5 is worn or is deformed into a polygonal shape, the frequency of the vibration varies depending on the rolling speed, but the pitch of the chatter marks on the surface of the backup roll 5 does not vary depending on the rolling speed. Therefore, if the pitch of the chatter marks is grasped and the vibration intensity is continuously monitored, the generation of abnormal vibration and the like can be detected. However, in reality, the uneven shape (fine marks and the like) on the surface of the backup roll 5 is not visible or does not occur on the surface of the backup roll 5 at the point in time before the backup roll 5 is loaded into the rolling mills 2A, 2B, 2C, 2D, and the pitch of the chatter marks cannot be predicted in advance. In addition, although it is abnormal vibration caused by the rotational motion, in the case where a defect is generated on the surface of the rotating body and in the case where the rotating body is deformed into a polygonal shape, the pitch that is observed as abnormal vibration is different, and thus the pitch of the chatter marks cannot be predicted in advance.

[0053] Therefore, in the present application, the vibration data is frequency-analyzed and data-converted at a constant time interval (data acquisition period), and in the map generation section 14, a vibration map in which the relationship between the pitch and the vibration intensity is arranged in time series is generated. As a result, it is possible to visually grasp that the vibration intensity of the chatter marks gradually increases over time from the vibration map. That is, even if the pitch of the chatter marks cannot be predicted in advance, it is possible to grasp and detect the generation of abnormal vibration by visually grasping the change in the vibration intensity on the vibration map. Such a vibration map is generated for each of the vibration meters 8A, 8B, 8C, 8D provided to the rolling mills 2A, 2B, 2C, 2D. In addition, the vibration intensity depends on the wavelength (pitch) and the time, and thus each vibration map becomes a three-dimensional display. The map generation section 14 can also distinguish the values of the vibration intensity, and assign a color to each distinction, and generate a vibration map in which the pitch is the vertical axis and the time is the horizontal axis. The vibration map generated by the map generation section 14 is displayed in an operation room or the like that manages the operation of the rolling mills 2 by the display device 20. By referring to the vibration map, it is possible to determine whether the vibration intensity corresponding to a specific pitch is large, and thus it is possible to detect abnormal vibration at an early stage.

[0054] Note that, in the rolling mills 2A, 2B, 2C, 2D, there are sometimes rolling speeds at which vibrations are easily generated. For example, there are cases in which resonance occurs due to vibrations caused by the rotational movement of the rotating bodies of the rolling mills 2A, 2B, 2C, 2D and vibrations caused by the natural period of the rolling mills 2A, 2B, 2C, 2D, and the like. In such cases, the content of the vibration map display is preferably selectable according to the rolling speed. By these means, it is possible to determine that the imprint on the backup roll 5 is progressing with respect to a vibration peak that is initially unclear as a vibration peak and that increases with the passage of time of several hours to several days at a specific pitch on the vibration map.

[0055] Further, the map generation section 14 can have a function of performing principal component analysis on the second analysis data of the vibration intensities corresponding to the standard pitches generated by the data transformation section 13 and generating a deviation component map based on the result of the principal component analysis, in addition to the above-described function of generating a vibration map.

[0056] The abnormality detection device 10 for a rolling mill can further include a principal component analysis section 15 that performs principal component analysis on the second analysis data of the vibration intensities of each standard pitch after transformation by the data transformation section 13 using reference data indicating a normal state, and determines a deviation component for each pitch calculated as a residual error of a projection (evaluation data) of the second analysis data with respect to the reference data. The evaluation data refers to data obtained by projecting observation data (in the present embodiment, the second analysis data) toward a space composed of principal component vectors. That is, the evaluation data is determined by scalars obtained by projecting the observation data toward the direction of each of the plurality of principal component vectors, and is composed of information of the same number of scalars as the number of principal component vectors. The principal component vectors (reference data) to which the principal component analysis is applied are described later. Note that, "principal component analysis" is sometimes used in the meaning of both analysis of synthesizing variables called principal components that best represent the deviation of the whole from a small number of variables having no correlation from a large number of variables having correlation, and calculation of a projection of observation data with respect to a space composed of principal component vectors set in advance, but the principal component analysis performed by the principal component analysis section 15 of the present embodiment is used in the latter meaning. That is, the principal component analysis section 15 in the present embodiment has a function of calculating a projection (evaluation data) of the second analysis data with respect to a space composed of principal component vectors (reference data) set in advance indicating a normal state, and determines a deviation component as the difference between the second analysis data and the projection (evaluation data) of the second analysis data.

[0057] The first principal component to the i-th principal component (reference data) set as the principal component vectors used in the principal component analysis performed in the principal component analysis section 15 are set based on the vibration intensity of each standard pitch (reference vibration data) obtained at a normal time when no abnormal vibration is generated in the rolling mills 2A, 2B, 2C, 2D. With respect to the reference vibration data, the reference data is generated by performing the principal component analysis in the principal component derivation section 16 described later. Note that the principal component analysis performed by the principal component derivation section 16 means analysis of principal component vectors that synthesize many variables having correlation to express a deviation of the whole with a small number of variables having no correlation best. The normal time when no abnormal vibration is generated in the rolling mills 2A, 2B, 2C, 2D means a state in which no abnormal vibration is generated in all of the rolling mills 2A, 2B, 2C, 2D at a standard rolling speed. Note that the abnormal vibration is described later. The reference vibration data is obtained, for example, by performing the frequency analysis described above on vibration data measured at a rolling time within 12 hours from when the backup roll 5 is replaced with a new one, and further converting the data after the frequency analysis into the vibration intensity of each pitch. The reference vibration data is sometimes referred to as normal analysis data, as data obtained by analyzing a normal vibration behavior in which no abnormal vibration is generated. Further, the reference vibration data can be obtained by analyzing vibration data measured at a rolling time within 24 hours from when the backup roll 5 is replaced with a new one. This is because it is empirically known that at least 2 days or more are required before the backup roll 5 is worn into a polygonal shape, and the abnormal vibration is not generated for about 2 days from when the backup roll 5 is replaced with a new one. The data sampling time at the time of obtaining the reference vibration data is preferably set to be the same as the data sampling time in the case of performing the abnormality detection in the operation (after 24 hours elapses from when the backup roll 5 is replaced with a new one). With respect to the data acquisition period, it is also possible to set different periods in the case of obtaining the reference vibration data and in the case of obtaining the vibration data in the operation.

[0058] The reference vibration data is generated by taking the vibration intensity of each standard interval at the data sampling time as a data set every data taking period at the normal time, and thus contains a plurality of data sets. The number of data sets contained in the reference vibration data is preferably 30,000 or more and 200,000 or less. Using the reference vibration data thus taken, a principal component vector is derived by principal component analysis with the standard interval as a variable, and is referred to as reference data. Specifically, by the principal component analysis by the principal component derivation section 16 described later, a principal component that represents the deviation of the entire reference vibration data with a small number of components without correlation is derived from a plurality of reference vibration data with correlation, and the cumulative value of the contribution rate is calculated from the principal component with a high contribution rate to the characteristic quantity representing the reference vibration data in order of accumulation, and i principal components selected before the cumulative value (cumulative contribution rate) of the contribution rate calculated reaches a value set in advance are selected as the reference data. Here, the cumulative contribution rate set in advance is referred to as a reference contribution rate or a set contribution rate. The reference contribution rate in the present embodiment can be arbitrarily set from a value of 1 (100%) or less. In a general tandem mill, the reference contribution rate is preferably set to 0.4 (40%) or more and 0.7 (70%) or less, and more preferably 0.6 (60%) or more and 0.7 (70%) or less. Here, the reference contribution rate is an index that affects the degree of reproduction (reproducibility) of the vibration behavior of the reference vibration data on the principal component space. If the reference contribution rate is too large, although the vibration behavior of the reference vibration data can be reproduced on the principal component space with high accuracy, measurement noise and the like contained in the reference vibration data can also be reproduced on the principal component space. On the other hand, if the reference contribution rate is too small, although the influence of the measurement noise contained in the reference vibration data can be excluded, there is a tendency that the characteristics related to the vibration behavior of the reference vibration data are lost in the principal component space. The appropriate range of the reference contribution rate depends on the mill used and the rolling conditions of the steel sheet, but for the purpose of detecting abnormal vibration of the tandem mill, it is preferably set to the above range.

[0059] At the time of derivation of the reference data, as Figure 6As shown, the principal component deriving section 16 can be provided which derives principal components using the reference vibration data (normal analysis data) generated by the data transforming section 13 of the abnormality detection device 10 of the rolling mill. The principal component deriving section 16 performs analysis for determining a principal component vector which best represents the entire deviation with a small number of no correlation from a plurality of reference vibration data having correlation. The first to i-th principal components (reference data) obtained by the principal component deriving section 16 can be temporarily stored in a storage section not shown, and transmitted to the principal component analysis section 15 at the time of the later operation, and the projection (evaluation data) of the second analysis data obtained in the operation to the first to i-th principal components is calculated by the principal component analysis section 15. In addition, in the case where the pitch at which the chatter mark is likely to occur in the rolling mill 2A, 2B, 2C, 2D is known in advance, a plurality of standard pitches of the same degree as the pitch can be selected in advance at the time of derivation of the principal components in the principal component deriving section 16, and the number of variables used in the principal component analysis in the principal component analysis section 15 can be reduced.

[0060] The principal component analysis section 15 performs principal component analysis for calculating the evaluation data using the first to i-th principal components (reference data) derived by the principal component deriving section 16, on the second analysis data obtained in the operation which represents the vibration intensity of each standard pitch. Specifically, the principal component analysis section 15 uses the second analysis data obtained in the operation which represents the vibration intensity of each standard pitch, decomposes into the projection to the first to i-th principal components as the reference data and the residual portion after subtracting the projection of the reference data to the principal components from the second analysis data, and determines the residual portion as the deviation component. The deviation component is sometimes referred to as the deviation degree, Q statistic. The deviation component calculated by the principal component analysis section 15 becomes an index representing the deviation from the vibration behavior at the normal time, and thus, by monitoring the deviation component, the abnormal vibration of the rolling mill 2A, 2B, 2C, 2D can be easily detected.

[0061] Further, the map generating section 14 can have a function of generating a deviation component map based on the deviation component of each pitch calculated by the principal component analysis section 15. That is, the map generating section 14 generates a vibration map obtained by arranging the deviation component of each pitch calculated by the principal component analysis section 15 along the time series. In the present embodiment, the vibration map thus generated is referred to as the deviation component map. The deviation component map is a map obtained by arranging the deviation component obtained by the principal component analysis section 15 along the time series, but the deviation component is sometimes calculated as a negative value, and thus such a deviation component is preferably displayed as "0 (zero)". This is because, in the case where the deviation component is negative, it means that the vibration in the operation is small compared to the normal time, and does not indicate the abnormal vibration. By the deviation component map, it is easy to visually recognize that the abnormal vibration has occurred.

[0062] The map generation section 14 can also distinguish the values of the deviation component and generate a deviation component map having the distance as the vertical axis and the time as the horizontal axis by assigning a color to each of the distinctions (refer to the following Figure 4 and Figure 5 ). The deviation component map generated by the map generation section 14 is displayed in an operation room or the like that manages the operation of the rolling mill 2 by the display device 20. By referring to the deviation component map, it is possible to determine whether the deviation component is large, and thus it is possible to detect abnormal vibration at an early stage. Alternatively, the map generation section 14 can also generate a three-dimensional deviation component map having the x-axis as the time, the y-axis as the distance, and the z-axis as the deviation component (refer to the following Figure 3 ). Thus, it is possible to easily grasp the tendency that the abnormal vibration gradually becomes large.

[0063] The effects of the embodiments of the present application will be described with reference to Figure 1 and Figure 2 . First, the vibrations of the rolling mills 2A, 2B, 2C, 2D at the time of cold rolling (at the time of operation) are measured by the vibration meters 8A, 8B, 8C, 8D, and the vibration data is collected in the data collection section 11 (collection step). Figure 7 is an example of the time-series vibration data collected by any one of the vibration meters 8A, 8B, 8C, 8D in the collection step. This is an example in which the acceleration obtained from the vibration meters 8A, 8B, 8C, 8D during the data sampling time 0.2 Sec is converted into a vibration velocity. Thereafter, frequency analysis of the vibration data is performed by the frequency analysis section 12, and first analysis data indicating the vibration intensity of each frequency is generated (frequency analysis step). Figure 8 is an example of the vibration intensity of each frequency generated in the frequency analysis step. Further, in the data conversion section 13, the first analysis data is converted into second analysis data indicating the vibration intensity of each distance (data conversion step).

[0064] Figure 9 is an example of the second analysis data indicating the vibration intensity of each distance after conversion in the data conversion step. The data conversion from the first analysis data to the second analysis data in the data conversion step is performed every data acquisition period. Thereafter, a vibration map obtained by arranging a plurality of vibration data (second analysis data) after conversion into the vibration intensity of each distance along the time series is generated, and this vibration map is updated in real time (map generation step). Figure 10 is an example of the vibration map generated in the map generation step. This is obtained by arranging the second analysis data indicating the vibration intensity of each distance along the time series every predetermined time interval.

[0065] According to the above-described embodiment, the abnormal vibration of the rolling mill 2A, 2B, 2C, 2D that causes a scratch can be detected with high accuracy. In addition, the second analysis data of the vibration intensity of each pitch generated in the data transformation step is subjected to principal component analysis using the reference data representing a normal state (principal component analysis step). Thereby, the deviation component of each pitch is calculated as a residual of the projection of the second analysis data with respect to the reference data. In this way, by determining the principal component vector that is a characteristic amount of the reference vibration data representing a normal time as a feature of the characteristics of the device, such as the vibration component naturally generated due to the meshing of the gears of the rolling mill 2A, 2B, 2C, 2D, and the vibration characteristics of the bearings of the rolling mill 2A, 2B, 2C, 2D, it is possible to achieve analysis that highlights only the vibration that is abnormal.

[0066] Specifically, in the abnormal vibration of the rolling mill 2A, 2B, 2C, 2D, the natural vibration of the rolling mill 2A, 2B, 2C, 2D, and the vibration caused by the rotation of the device due to bearing failure, meshing of the gears, poor coupling, or looseness, etc. are many. Thus, the detection of the abnormal vibration in the past is performed based on whether the amplitude of a specific frequency exceeds a certain threshold. On the other hand, in the case where a scratch is generated, from the time point before the scratch is generated, a slight vibration is generated at a frequency corresponding to the pitch of the scratch, and the vibration gradually becomes larger as time passes, and in conjunction therewith, the defect at the surface of the metal strip S caused by the vibration gradually becomes larger and grows. That is, first, a slight vibration caused by the device is generated, and then a scratch is generated on the surface of the metal strip S. However, in actual operation, the rolling speed varies in the length direction of one metal strip S, and the set rolling speed also varies for each different metal strip S, and thus, if only a specific frequency is focused on, it is difficult to detect the slight vibration at the time point before the abnormal vibration. In contrast, in the present embodiment, the first analysis data representing the vibration intensity of each frequency is transformed into the second analysis data representing the vibration intensity of each standard pitch, and the vibration map or the deviation component map is generated in the map generation step based on the second analysis data. Thus, it is possible to visually recognize early on the condition in which the vibration corresponding to a specific pitch gradually becomes larger.

[0067] The abnormal vibration detection method of the rolling mill 2A, 2B, 2C, 2D described above can be used, and a standard pitch (hereinafter, referred to as a monitoring pitch) that should be monitored is set in advance for each of the rolling mills 2A, 2B, 2C, 2D, and in a case where the vibration intensity at the set monitoring pitch exceeds a pre-set limit vibration intensity, the rotating body (support roll replacement step) of the rolling mill 2A, 2B, 2C, 2D, which is the cause of the abnormal vibration of the rolling mill, is replaced. The monitoring pitch is a pitch at which chatter marks are easily generated on the surface of the metal strip S in a case where the metal strip S is rolled, and this can be empirically obtained or obtained through experiments. Specifically, the chatter marks are periodic pattern-like defects generated on the surface of the metal strip S, and thus the pitch of the chatter marks can be determined in an inspection process of the metal strip S. Therefore, the pitch of the chatter marks determined in the inspection process can be set as the monitoring pitch. The monitoring pitch can be set as a specific value, or can be set as a numerical range of the pitch at which the chatter marks are generated. For example, in a case where the pitch at which the chatter marks are easily generated is 30 mm, 27 mm or more and 33 mm or less can be set as the monitoring pitch as a numerical range of ±10%. The range of the monitoring pitch can be determined by taking into account the deviation of the pitch of the chatter marks grasped empirically from the operation performance of the rolling mill 2A, 2B, 2C, 2D.

[0068] The limit vibration intensity described above means a vibration intensity at which defects generated on the surface of the metal strip S due to the vibration of the rolling mill 2A, 2B, 2C, 2D are likely to become a quality problem as a product of the metal strip S. That is, the limit vibration intensity means an upper limit value of the vibration intensity that can be tolerated as the vibration generated in the rolling mill 2A, 2B, 2C, 2D. Specifically, if excessive vibration is generated at a specific pitch, chatter marks are generated on the metal strip S, and are likely to become a cosmetic defect of the metal strip S. Therefore, the performance data of the vibration intensity at each pitch is obtained in advance, and based on the shipment criteria as a product of the metal strip S and the performance data of the vibration intensity, the upper limit value of the vibration intensity at which the product of the metal strip S does not have a quality problem is set as the limit vibration intensity. Note that the vibration that is likely to have a quality problem as a product of the metal strip S, that is, the vibration exceeding the limit vibration intensity corresponds to the abnormal vibration of the rolling mill 2A, 2B, 2C, 2D in the embodiment of the present application. In addition, a state in which no vibration is generated and a state in which vibration is generated but does not reach the abnormal vibration correspond to the normal time and the normal state of the rolling mill 2A, 2B, 2C, 2D in the embodiment of the present application.

[0069] Then, based on the monitoring intervals and the limit vibration intensity set as described above, the abnormal vibration detection method of the rolling mill 2A, 2B, 2C, 2D described above is used to generate a vibration map or a deviation component map. In the case where the vibration intensity of the corresponding interval exceeds the limit vibration intensity in these vibration maps or deviation component maps, the operation of the rolling mill 2A, 2B, 2C, 2D is temporarily stopped, and the rotating body that is the cause of the abnormal vibration of the rolling mill 2A, 2B, 2C, 2D that has generated the abnormal vibration is replaced. In particular, the cause of the abnormal vibration of the rolling mill is in many cases the backup roll 5 of the rolling mill 2A, 2B, 2C, 2D, and therefore the backup roll 5 of the stand that has generated the abnormal vibration is replaced. Thus, even in the case where a plurality of metal strips S are rolled over a long period of time, it is possible to realize the operation of the rolling mill 2A, 2B, 2C, 2D that has been prevented from generating the abnormal vibration at a specific interval. In addition, by such rolling, it is possible to manufacture a metal strip S that has an excellent appearance in which no chatter marks are generated on the surface of the metal strip S.

[0070] Furthermore, the vibration source of the abnormal vibration that is the cause of the chatter marks is often a fine mark at the same interval as the chatter marks generated on the surface of the backup roll 5 above or below. In this case, if the vibration caused by the fine mark on the backup roll 5 and the vibration of the rolling mill 2A, 2B, 2C, 2D resonate at a predetermined rolling speed, the vibration of the rolling mill 2A, 2B, 2C, 2D gradually becomes large while the fine mark gradually becomes clear. Therefore, in the embodiment of the present application, the vibration data is frequency-analyzed at constant time intervals (every other data acquisition period), and the relationship between the frequency and the vibration intensity at the constant time intervals is calculated. Then, the frequency is converted to a standard interval based on the rolling speed, and the relationship between the standard interval and the vibration intensity is generated and displayed as a vibration map, thereby making it possible to monitor over time.

[0071] In addition, in the vibration data, vibrations of many other factors that generate vibrations at constant intervals, such as the meshing frequency of bearings and gears, are also superimposed, and a clear vibration peak of the chatter marks is not obtained from the beginning. Therefore, a deviation component map that strictly distinguishes the vibration peak of the chatter marks from other factors can also be generated using a principal component analysis method.

[0072] Example 1

[0073] An example of the present application will be shown below. In Example 1, a tandem rolling mill composed of 5 rolling mills (5 stands) was used, and a vibration meter composed of a piezoelectric element was installed on the upper part of the housing on the operator side and on the upper part of the housing on the motor side of each rolling mill. The test material was various from very low carbon steel to high tensile steel, and a plurality of coils having an entry side thickness of 2 mm or more and 5 mm or less, an exit side thickness of 0.6 mm or more and 2.4 mm or less, and a steel sheet width of 850 mm or more and 1880 mm or less were used. In addition, the recognition of abnormal vibration causing a chatter was performed based on the vibration data measured at the housing of the rolling mill of the final stand (the fifth stand which is the most downstream in the rolling direction of the steel sheet). Specifically, the data sampling time was set to 0.2 Sec, the data acquisition period was set to 1 Sec, and the vibration data was collected by the data collection section 11. Then, the vibration data after Fourier transform by the frequency analysis section 12 (first analysis data) was converted into the vibration intensity at the standard pitch (second analysis data) in the data conversion section 13.

[0074] Note that, in Example 1, the abnormality detection was performed using the vibration meter installed on the upper part of the housing on the operator side among the 2 vibration meters installed on the final stand. In this example, for the vibration intensity at each standard pitch generated by the data conversion section 13, the deviation component was calculated in the principal component analysis section 15, and the deviation component map was generated in the map generation section 14. In addition, with respect to the sampling frequency of the vibration meter of 2000 Hz, the frequencies were selected every 5 Hz in the frequency band of 0 Hz to 1000 Hz, and these frequencies were used as the reference frequencies. The standard rolling speed was set to 600 m / min. Thus, 201 pitches were set as the standard pitch. Furthermore, the reference vibration data was collected within 2 days after the backup roll of the tandem rolling mill used was replaced with a new one, 22 principal components (reference data) were derived in the principal component derivation section 16, and stored in the principal component analysis section 15. After the reference vibration data was collected, the vibration data in the operation of the tandem rolling mill was collected, the deviation component was calculated in the principal component analysis section 15 at all times, the deviation component map was updated in the map generation section 14 at all times, and the deviation component map was displayed in the operation room of the tandem rolling mill by the display device 20.

[0075] Figure 3 is a graph of the deviation component map generated based on the vibration data of the final stand and the rolling speed in the operation, which is shown as an example of the deviation component map in Example 1. Note that, in Figure 3 , from the completion of the collection of the reference vibration data, the principal component analysis and the extraction of the deviation component were performed with respect to the vibration data for about 2 weeks, i.e., for about 9 days of the rolling period and for a rolling length of about 5600 km, and the data interval in the time axis direction was excluded to be every 100 Sec. In Figure 3In the middle, data when the weld joint after joining the front and rear coiled materials is passed is also included, so that a large vibration occurring throughout the entire pitch can be confirmed occasionally. In addition, it is known that under a specific pitch (standard pitch. In Figure 3 In the middle, data when the weld joint after joining the front and rear coiled materials is passed is also included, so that a large vibration occurring throughout the entire pitch can be confirmed occasionally. In addition, it is known that under a specific pitch (standard pitch. In

[0076] Example 2

[0077] The rolling mill used in Example 2 is a continuous rolling mill composed of 4 stands, and a vibration meter composed of a piezoelectric element is installed on the upper part of the housing on the operator side and the upper part of the housing on the motor side of each stand. The steel grade, the steel sheet thickness, and the sheet width are set to the same conditions as in Example 1, and the rolling amount is the same degree as in Example 1. The identification of the abnormal vibration that causes the chatter mark is based on the data of the vibration meter provided on the upper part of the housing on the operator side in the housing of the third stand (third stand) counted from the upstream side in the rolling direction of the steel sheet. The data sampling time, the data acquisition period, and the reference frequency are set to the same conditions as in Example 1. However, vibrations above the frequency that is considered to be unable to detect vibrations due to the characteristics of the housing are ignored.

[0078] Figure 4 is a graph showing an example of the deviation component map in Example 2, and specifically, is an example of the deviation component map obtained from the operation data including all rolling speeds, without limiting the conditions of the rolling speed in the operation. In Figure 4In the deviation component map, the period in which the chatter did not occur at all and the period in which the chatter occurred are shown in the graph. In the principal component analysis, the data of 1 day from the replacement of the backup roll in the test material is taken as the reference vibration data. At this time, the vibration data of the normal time is divided by the rolling speed every 50 mpm (m / min), the reference frequency is converted to the standard pitch using the rolling speed of each division, and the standard pitch is set for each division of the rolling speed. Thus, the principal components (reference data) are derived for each division of the rolling speed. At this time, the contribution rate is set to "0.5" for each rolling speed, and a plurality of principal components to be extracted as the reference data are selected for each division of the rolling speed. Thus, the principal components corresponding to the division of the rolling speed are stored in the principal component analysis section 15. After that, the vibration data of the rolling mill in operation (second analysis data) is taken as the object, and the principal component analysis section 15 calculates the deviation components corresponding to each standard pitch. Note that, regarding the calculation of the deviation components by the principal component analysis section 15, the principal components corresponding to the rolling speed in operation are selected, and the data at the time of operation for each rolling speed is calculated at any time using the selected principal components. Here, the deviation degree of the standard pitch sometimes takes a negative value, but they are shown as "0" in the deviation component map. Figure 4

[0079] In the deviation component map, the period in which the chatter did not occur at all and the period in which the chatter occurred are shown in the graph. In the principal component analysis, the data of 1 day from the replacement of the backup roll in the test material is taken as the reference vibration data. At this time, the vibration data of the normal time is divided by the rolling speed every 50 mpm (m / min), the reference frequency is converted to the standard pitch using the rolling speed of each division, and the standard pitch is set for each division of the rolling speed. Thus, the principal components (reference data) are derived for each division of the rolling speed. At this time, the contribution rate is set to "0.5" for each rolling speed, and a plurality of principal components to be extracted as the reference data are selected for each division of the rolling speed. Thus, the principal components corresponding to the division of the rolling speed are stored in the principal component analysis section 15. After that, the vibration data of the rolling mill in operation (second analysis data) is taken as the object, and the principal component analysis section 15 calculates the deviation components corresponding to each standard pitch. Note that, regarding the calculation of the deviation components by the principal component analysis section 15, the principal components corresponding to the rolling speed in operation are selected, and the data at the time of operation for each rolling speed is calculated at any time using the selected principal components. Here, the deviation degree of the standard pitch sometimes takes a negative value, but they are shown as "0" in the deviation component map. Figure 4

[0080] Figure 5 is a graph showing the deviation component map generated from the vibration data at the time of operation obtained under the condition defined in the rolling speed of 800 mpm or more and 850 mpm or less in Example 2. By making and displaying the deviation component map in which the rolling speed is defined as in Figure 5 such, it is possible to determine that the corresponding vibration is abnormal before the occurrence of the chatter, and it is possible to take measures before the occurrence of the chatter.

[0081] The embodiments of the present application are not limited to the above-described embodiments, and various modifications can be made. For example, in the above-described embodiments, the rolling speed is divided every 50 mpm (m / min), but the division of the rolling speed is not limited to this. For example, the rolling speed can be divided every 100 mpm (m / min) or every 200 mpm (m / min). Figure 3 ​​In the present embodiment, the case where the three-dimensional display is performed using the depth of the color corresponding to the deviation component is exemplified, but is not limited thereto. The display using the color specified for each vibration intensity, or the display using the depth of the color corresponding to the vibration intensity, or the display using both of the color specified for each vibration intensity and the depth of the color corresponding to the vibration intensity, or the like can be adopted. By these methods, with respect to the vibration peak that is initially unclear at the pitch and becomes larger over time of several hours to several days, it is possible to determine that the print of the backup roll is progressing, or the like. In the present embodiment, the case where the metal strip S is a cold-rolled steel sheet is exemplified, but the metal strip S can be a stainless steel material, or can be a hot-rolled steel sheet. In addition, the rolling mills 2A, 2B, 2C, and 2D can not be the same structure, and for example, a four-stage rolling mill and a six-stage rolling mill can be mixed as the form of the rolling mill.

[0082] Example 3

[0083] As Example 3 of the present embodiment, the detection of the abnormal vibration of the rolling mill was performed using the tandem rolling mill used in Example 1, under the same conditions as in Example 1. Note that Example 3 is different from Examples 1 and 2 described above, and is an example in which the vibration map was generated using the mapping generation section 14 based on the analysis data of the vibration intensity at each pitch generated by the data conversion section 13, without using the principal component analysis section 15.

[0084] In Example 3 as well, the data of the vibration meter provided on the upper part of the housing on the operator side of the final stand was acquired by the data collection section 11. In the collection step performed by the data collection section 11, the data sampling time was set to 0.2 sec, and the vibration data after the acceleration acquired from the vibration meter was converted to the vibration velocity was acquired. In the frequency analysis section 12, the first analysis data composed of the vibration intensity at each frequency was acquired by performing Fourier transform on the time-series vibration data. In the frequency analysis step performed by the frequency analysis section 12, with respect to the sampling frequency of the vibration meter of 2000 Hz, the frequencies were selected every 5 Hz in the frequency band of 0 Hz to 1000 Hz, and these frequencies were used as the reference frequencies. In the data conversion step performed by the data conversion section 13, the standard rolling speed was set to 600 m / min, 201 standard pitches were set, and the data (second analysis data) related to the vibration intensity at each pitch was acquired every data acquisition period. In the vibration map generation step performed by the mapping generation section 14, the vibration map with respect to the second analysis data in which the magnitude of the vibration intensity acquired in the data conversion step was expressed as the depth in the gray scale was generated. The vibration map thus generated is shown in FIG. 6. In FIG. 6, the horizontal axis represents the pitch, and the vertical axis represents the vibration intensity. The vibration map is expressed as the depth in the gray scale, and the darker the color, the larger the vibration intensity. In the vibration map shown in FIG. 6, the vibration intensity at each pitch is expressed as the depth in the gray scale, and the vibration map is generated. In the vibration map, the vibration intensity at each pitch is expressed as the depth in the gray scale, and the vibration map is generated. Figure 11 Figure 11 ​In this case, the time point at which the backup roll of the continuous rolling mill is replaced is set as the origin of time shown by the horizontal axis, and the vibration intensity of each standard pitch is arranged in time series.

[0085] In Figure 11 In the example of the vibration map shown in FIG. 6, it is known that from the time point at which the measurement of the vibration data is started (0 seconds) to 6,000 seconds, relatively large vibrations are intermittently generated in the vicinity of the standard pitch 33 mm, but do not reach a significant vibration, i.e., an abnormal vibration. However, it is known that if more than 10,000 seconds elapses from the start time point, the vibration corresponding to the standard pitch 33 mm increases. Also, it is confirmed that if more than 15,000 seconds elapses from the start time point, the vibration in the vicinity of the standard pitch 33 mm significantly occurs. Also, according to the observation of the surface of the metal strip S rolled at this time point (15,000 seconds), it is confirmed that a chatter mark is generated on the surface of the metal strip S at the pitch corresponding to the standard pitch 33 mm.

[0086] Next, a comparative example performed for verifying the abnormal vibration detection method of the present embodiment 3 is described. In the comparative example, the same data as the vibration data obtained in the above-described present embodiment 3 is used, and the vibration intensity of each frequency obtained by the frequency analysis section 12 is obtained. Figure 12 is a view showing an example of a vibration map with respect to the first analysis data, which is made by arranging the vibration intensity of each frequency along the time series. Also, in Figure 12 , the magnitude of the vibration intensity of each frequency of the vibration is displayed using the darkness of the color. If referring to Figure 12 , the frequency band corresponding to the pitch 33 mm at which the chatter mark is detected in the embodiment 3 changes depending on the rolling speed, but is approximately 50 Hz or more and 100 Hz or less. Also, according to the map of Figure 12 , from the start time point (0 seconds) to 6,000 seconds, a tendency of high vibration intensity can be seen in the vicinity of the frequencies 50 Hz and 250 Hz. However, if more than 15,000 seconds elapses from the start time point, the frequency of high vibration intensity becomes a frequency band of 100 Hz or more and 150 Hz or less. Also, if more than 20,000 seconds elapses from the start time point, a tendency of an increase in the vibration intensity is present in the frequency band of 50 Hz or more and 100 Hz or less corresponding to the pitch 33 mm at which the chatter mark is detected in the embodiment 3, but the vibration intensity in the frequency band of 150 Hz or more and 200 Hz or less also increases. Also, in the frequency band of 150 Hz or more and 200 Hz or less, a tendency of high vibration intensity can be seen, but in this frequency band, the variation in the vibration intensity accompanying the passage of time is large, and it is difficult to determine a specific frequency band in advance to discriminate the abnormal vibration.

[0087] According to the above results, in the method of determining the frequency and band of the chatter generation in advance and detecting the vibration intensity in the frequency and band, it is difficult to grasp the generation of the chatter mark caused by the slight vibration at an early stage. In contrast, if, as in Embodiments 1 to 3, the vibration intensity of each frequency is converted into the vibration intensity of each standard pitch on the basis of the rolling speed, and the vibration map obtained by arranging the same along the time series is generated, even if the rolling conditions differ for each metal strip S, the pattern in which the abnormal vibration gradually becomes clear can be visually captured. Thus, it is known that the abnormal vibration generated in the rolling mill can be reliably detected.

[0088] Explanation of Reference Signs

[0089] 1 Rolling apparatus

[0090] 2A, 2B, 2C, 2D Rolling mill

[0091] 3 Housing

[0092] 4 Work roll

[0093] 5 Backup roll

[0094] 6 Drive device

[0095] 7 Small-diameter roll

[0096] 8A, 8B, 8C, 8D Vibration meter

[0097] 10 Abnormality detection device of rolling mill

[0098] 11 Data collection section

[0099] 12 Frequency analysis section

[0100] 13 Data conversion section

[0101] 14 Map generation section

[0102] 15 Principal component analysis section

[0103] 16 Principal component derivation section

[0104] 20 Display device

[0105] S Metal strip

Claims

1. An abnormal vibration detection method of a rolling mill, which is an abnormal vibration detection method of a rolling mill having a pair of work rolls and a plurality of backup rolls that support the work rolls, wherein Comprising: a collecting step of collecting vibration data of the rolling mill; a frequency resolution step of performing frequency resolution of the vibration data to generate first resolution data indicating vibration intensity of each frequency; a data transformation step of transforming the first resolution data into second resolution data indicating vibration intensity of each pitch based on a rolling speed; and a map generation step of generating a vibration map obtained by arranging a plurality of the second resolution data along a time series, the pitch P (mm) is an interval of vibration peaks that become adjacent to each other in a length direction of a metal strip and a circumferential direction of a work roll as a result of data transformation, and is related by the following equation using a rolling speed V (m / min) and a frequency f (Hz) of vibration, P = (1000 x V) / (f x 60).

2. The abnormal vibration detection method of a rolling mill according to claim 1, further comprising a principal component analysis step of performing principal component analysis of the second resolution data using reference data indicating a normal state to determine a deviation component of each pitch calculated as a residual of a projection of the second resolution data with respect to the reference data, the map generation step further generates a deviation component map obtained by arranging a plurality of the deviation components of each pitch extracted by the principal component analysis step along a time series.

3. The abnormal vibration detection method of a rolling mill according to claim 2, in the principal component analysis step, a plurality of principal components used as the reference data are set in a manner that an accumulated value of a contribution rate of a principal component when principal component analysis is performed on normal resolution data acquired when rolling is performed using the normal rolling mill becomes a reference contribution rate or more.

4. The abnormal vibration detection method of a rolling mill according to any one of claims 1 to 3, the rolling mill is a cold rolling mill.

5. An abnormality detection device of a rolling mill, which is an abnormality detection device of a rolling mill having a pair of work rolls and a plurality of backup rolls that support the work rolls, wherein Comprising: a data collection unit that collects vibration data of the rolling mill; a frequency resolution unit that performs frequency resolution of the vibration data to generate first resolution data indicating vibration intensity of each frequency; a data transformation unit that transforms the first resolution data into second resolution data indicating vibration intensity of each pitch based on a rolling speed; and a map generation unit that generates a vibration map obtained by arranging a plurality of the second resolution data along a time series, the pitch P (mm) is an interval of vibration peaks that become adjacent to each other in a length direction of a metal strip and a circumferential direction of a work roll as a result of data transformation, and is related by the following equation using a rolling speed V (m / min) and a frequency f (Hz) of vibration, P = (1000 x V) / (f x 60).

6. The abnormal detection device according to claim 5, further comprising a principal component analysis unit that performs principal component analysis of the second resolution data using reference data indicating a normal state to determine a deviation component of each pitch calculated as a residual of a projection of the second resolution data with respect to the reference data, the map generation unit further generates a deviation component map obtained by arranging a plurality of the deviation components of each pitch extracted by the principal component analysis unit along a time series. ​ 7. A rolling method comprising the following backup roll replacement step: The abnormal vibration detection method using the rolling mill according to any one of claims 1 to 4, the monitoring interval corresponding to the rolling mill is set in advance, and in a case where the vibration intensity at the monitoring interval of the vibration map or the deviation component map generated in the map generation step exceeds the pre-set limit vibration intensity, the backup roll of the rolling mill is replaced.

8. A metal strip manufacturing method comprising the step of manufacturing a metal strip using the rolling method according to claim 7.

Images