Roller state monitoring device

Abstract

The roll condition monitoring device is provided with: a rolling load detecting mechanism configured to detect a rolling load signal of a monitoring object roll selected from among an upper roll set including at least one roll and a lower roll set including at least one roll, in a case where a roll material is rolled between the upper roll set and the lower roll set; a signal extracting mechanism configured to extract a rolling load high frequency signal having a frequency of a predetermined prescribed frequency or more from the rolling load signal; and a roll condition determining mechanism configured to determine a condition of the monitoring object roll based on a test result of a statistical test method for a plurality of rolling load values included in the rolling load high frequency signal.

Description

[0001] This application is a divisional application of the Chinese Patent Application No. 201980005772.5, with the title of "Roller Condition Monitoring Device", filed on August 28, 2019. TECHNICAL FIELD

[0002] The present application relates to a roller condition monitoring device. BACKGROUND

[0003] In the past, as described in, for example, Japanese Patent Application Publication No. H63-040608, a device that detects eccentricity of a roller used in a rolling device and corrects it is known. The device described in, for example, the relevant claim 1 in the patent publication has a mechanism that supplies an electrical pressure signal to a relatively narrow band filter having a frequency band characteristic that allows passage of a signal indicating a frequency of eccentricity of a roller; a mechanism that receives a signal to which the filter is applied and generates an electrical display signal based on the signal; and a mechanism that applies the electrical display signal to a display that is visually recognizable for the purpose of checking by an operator to indicate the magnitude of eccentricity of the roller. In the eccentricity alarm device described in the relevant patent publication, the display (reference numeral 50) outputs an audible and / or visual alarm to the operator when the eccentricity exceeds a prescribed value.

[0004] Further, in the past, as described in, for example, Japanese Patent No. 5637637, a sheet thickness control device configured to discriminate a roller eccentricity amount is known. The technology of discriminating a roller eccentricity amount is described in, for example, paragraph 0016 and paragraph 0117 of the patent publication. For example, in paragraph 0016, it is described that the roller eccentricity amount of each of the upper and lower backup rollers is discriminated, and a work roll gap command value between the upper work roll and the lower work roll is calculated based on the discriminated roller eccentricity amount.

[0005] Prior Art Documents

[0006] Patent Documents

[0007] Patent Document 1: Japanese Patent Application Publication No. H63-040608

[0008] Patent Document 2: Japanese Patent No. 5637637 SUMMARY

[0009] Problems to be Solved by the Invention

[0010] Generally, the rolling load of the roll is obtained from the output signal of the rolling load sensor. Sometimes, an abnormal sensor output signal is transmitted with noise or the like. In a technique of calculating a discrimination value or making a roll condition determination based on one rolling load detection value, the precision is greatly affected when an abnormal value is mixed in. As a result, by the mixing in of the abnormal value, the discrimination precision of the roll eccentricity amount is greatly decreased or the determination precision of the roll condition is greatly decreased.

[0011] For example, Japanese Patent Application Publication No. H63-040608, described above, teaches two determination techniques for the roll eccentricity amount. The first technique is a determination method in which an operator visually recognizes the display of a display and checks the size of the roll eccentricity. The second technique is an eccentricity alarm device that issues an alarm when the eccentricity exceeds a prescribed value. In the case of using these techniques, it is possible to determine that the roll condition is abnormal based on an abnormal value of the noise signal indicating a high eccentricity. In this case, an incorrect alarm is issued.

[0012] Further, Japanese Patent No. 5637637, described above, at most only discloses the roll eccentricity amount discrimination technique described in paragraph 0016 and paragraph 0117 and the like. That is, in this publication, the problem of the decrease in the discrimination precision of the roll eccentricity amount due to the mixing in of the abnormal value is not recognized. As above, the conventional techniques have yet to be improved in terms of improving the determination precision of the roll condition.

[0013] The present application is made to solve the above-described problem, and the object is to provide a roll condition monitoring device that improves the discrimination precision or determination precision of the roll condition.

[0014] Means for solving the problem

[0015] The first roll condition monitoring device related to the present application is provided with a rolling load detection mechanism, a load variation value extraction mechanism, and a discrimination mechanism. The rolling load detection mechanism is configured to detect the rolling load of a monitoring target roll selected from among an upper roll set including at least one roll and a lower roll set including at least one roll in a case where a roll material is rolled between the upper roll set and the lower roll set. The load variation value extraction mechanism is configured to extract a rolling load variation value based on the rolling load at each rotational position of the monitoring target roll. The discrimination mechanism is configured to discriminate the roll eccentricity amount of the monitoring target roll by separately accumulating the value of one of the rolling load variation value and a roll gap equivalent value calculated based on the rolling load variation value at each rotational position of a plurality of rotational positions of the monitoring target roll, obtaining a plurality of accumulated values, and dividing the plurality of accumulated values by a correction coefficient corresponding to the number of roll rotations, which is the number of rotations of the monitoring target roll during the accumulation of the plurality of accumulated values.

[0016] The correction coefficient is preferably a variable value that is set to be larger the more the number of rotations of the monitoring target roll is monitored during the accumulation period of the accumulated value. The correction coefficient can be, for example, the same value as the number of rotations of the monitoring target roll, or can be set to be less than or more than the number of rotations of the monitoring target roll. The division correction based on the correction coefficient converts the accumulated value that is accumulated for a certain period of time to a value corresponding to the rotational speed of the monitoring target roll.

[0017] In the above-described first roll condition monitoring device, the discrimination mechanism can also be configured to convert the rolling load variation value into the roll gap equivalent value using a load roll gap conversion formula that includes a plasticity coefficient of the rolled material. If the reason is explained, since there are harder rolled materials and softer rolled materials depending on the steel grade, it is preferable to distinguish the difference in hardness. By setting the plasticity coefficient corresponding to the rolled material respectively using the conversion formula including the plasticity coefficient, the roll eccentricity amount can be discriminated with good accuracy, and thus is preferable.

[0018] The monitoring target roll can have a first side end portion and a second side end portion on the opposite side of the first side end portion. The first side can be, for example, an operator side (OS). The second side can be, for example, a drive side (DS). The rolling load detection mechanism can be configured to detect a first side rolling load at the first side end portion and to detect a second side rolling load at the second side end portion. The load variation value extraction mechanism can be configured to extract a first side rolling load variation value that is a value of the first side rolling load at each of the rotational positions of the monitoring target roll, and a second side rolling load variation value that is a value of the second side rolling load at each of the rotational positions of the monitoring target roll, respectively. The discrimination mechanism can be configured to calculate the roll eccentricity amount of each of the first side end portion and the second side end portion based on the first side rolling load variation value and the second side rolling load variation value, and to calculate the plurality of accumulated values corresponding to the plurality of rotational positions for each of the first side end portion and the second side end portion, respectively.

[0019] In the above-described first roll condition monitoring device, the discrimination mechanism that discriminates the respective roll eccentricity amounts for the first side end portion and the second side end portion can be specifically configured as follows. The discrimination mechanism can calculate a plurality of first side cumulative values that are the plurality of cumulative values with respect to the first side end portion corresponding to the plurality of rotational positions of the monitoring object roll, by separately accumulating the value of one of the first side rolling load variation value and the first side roll gap equivalent value calculated based on the first side rolling load variation value, for each of the plurality of rotational positions of the monitoring object roll. The discrimination mechanism can calculate a plurality of second side cumulative values that are the plurality of cumulative values with respect to the second side end portion corresponding to the plurality of rotational positions of the monitoring object roll, by separately accumulating the value of one of the second side rolling load variation value and the second side roll gap equivalent value calculated based on the second side rolling load variation value, for each of the plurality of rotational positions of the monitoring object roll. The discrimination mechanism can discriminate the roll eccentricity amounts for the first side end portion and the second side end portion, respectively, by dividing the first side cumulative values and the second side cumulative values by correction coefficients corresponding to the number of rotations of the monitoring object roll.

[0020] The above-described first roll condition monitoring device can further include a roll condition determination mechanism. The roll condition determination mechanism can determine the condition of the monitoring object roll during the second rolling period, by comparing the roll eccentricity amount calculated by the discrimination mechanism with a determination reference. The determination reference can be a predetermined prescribed reference value. The prescribed reference value can be a fixed value or a variable set value. The determination reference can be a "normal roll eccentricity representative value" generated by applying the technology of the second roll condition monitoring device described later. The determination reference can be updated at any timing.

[0021] The second roll condition monitoring device according to the present application is provided with a rolling load detecting mechanism, a load variation value extracting mechanism, a discriminating mechanism, a recording mechanism, and a roll condition judging mechanism. The rolling load detecting mechanism is configured to detect a rolling load of a monitoring target roll selected from among an upper roll set including at least one roll and a lower roll set including at least one roll, in a case where a roll material is rolled between the upper roll set and the lower roll set. The load variation value extracting mechanism is configured to extract a rolling load variation value that is a value of the rolling load at each rotational position of the monitoring target roll. The discriminating mechanism is configured to discriminate a roll eccentricity amount on the basis of the rolling load variation value. The recording mechanism records a plurality of roll eccentricity amounts calculated by the discriminating mechanism according to a plurality of rotational positions of the monitoring target roll in a first rolling period set in advance. The roll condition judging mechanism judges a condition of the monitoring target roll in a second rolling period performed after the first rolling period, on the basis of a normal roll eccentricity amount representative value that is a representative value calculated from the plurality of roll eccentricity amounts calculated by the discriminating mechanism in the first rolling period, and the roll eccentricity amount calculated by the discriminating mechanism in the second rolling period.

[0022] In the second roll condition monitoring device described above, the "representative value" can also be a well-known value called summary statistics. As the summary statistics, there are, for example, an average value, a standard deviation, a median, a range, and a mode. The normal roll eccentricity amount representative value can also be set to be any one of a normal roll eccentricity amount peak interval value, a normal roll eccentricity amount maximum average value, and a normal roll eccentricity amount minimum average value.

[0023] The normal roll eccentricity amount peak interval value is a difference between a maximum value and a minimum value among a plurality of roll eccentricity amounts calculated in a prescribed rolling period set in advance. It is also called a "range" that is one of the summary statistics. A waveform in which a plurality of roll eccentricity amounts obtained in the "prescribed rolling period" are arranged in a time series can also be called an "eccentricity amount data waveform". The normal roll eccentricity amount maximum average value can also be an average value with respect to a plurality of positive eccentricity amount peaks included in the eccentricity amount data waveform. The normal roll eccentricity amount minimum average value can also be an average value with respect to a plurality of negative eccentricity amount peaks included in the eccentricity amount data waveform. The prescribed rolling period can also be a period in which a number of roll materials set in advance are rolled. Further, the prescribed rolling period can also be a period from the start of a rolling process to the elapse of a prescribed time set in advance.

[0024] The first rolling period can be the time required to roll one of the rolling materials or the time required to roll a predetermined number of the rolling materials. The first rolling period can be set to a predetermined time regardless of the number of rolling materials. The second rolling period can be the same length as the first rolling period or can be set to a longer or shorter period than the first rolling period.

[0025] In the second roll condition monitoring device, the roll condition determination mechanism can be configured to determine the condition of the monitoring target roll by comparing a predetermined multiple of the normal roll eccentricity representative value with another representative value of the roll eccentricity obtained during the second rolling period. The other representative value is the same kind of value as the representative value calculated from a plurality of the roll eccentricities calculated by the discrimination mechanism during the second rolling period.

[0026] In the second roll condition monitoring device, the roll condition determination mechanism can be configured to determine the condition of the monitoring target roll based on a test result of a statistical test method for a plurality of the roll eccentricities. The statistical test method can use various known test methods. The statistical test method can be chi-square test as an example. The roll condition determination mechanism can determine the condition of the monitoring target roll based on a plurality of the roll eccentricities according to an outlier detection method based on Hotelling.

[0027] The third roll condition monitoring device according to the present application includes a rolling load detection mechanism, a signal extraction mechanism, and a roll condition determination mechanism. The rolling load detection mechanism is configured to detect a rolling load signal of a monitoring target roll selected from among an upper roll set including at least one roll and a lower roll set including at least one roll, in a case where a rolling material is rolled between the upper roll set and the lower roll set. The signal extraction mechanism extracts a rolling load high-frequency signal having a frequency of a predetermined prescribed frequency or more from the rolling load signal. The roll condition determination mechanism is configured to determine the condition of the monitoring target roll based on a test result of a statistical test method for a plurality of rolling load values included in the rolling load high-frequency signal.

[0028] In the aforementioned third roll condition monitoring device, the roll condition determination mechanism can also calculate the probability density distribution of rolling load values ​​based on the plurality of rolling load values. Furthermore, the roll condition determination mechanism can also be configured to determine the condition of the monitored roll based on a comparison of the probability density distribution of the rolling load values ​​with a pre-set reference distribution. Furthermore, in the aforementioned third roll condition monitoring device, the roll condition determination mechanism can also include a normal distribution roll condition determination mechanism, a Rayleigh distribution roll condition determination mechanism, or be configured to include at least one of these mechanisms. The normal distribution roll condition determination mechanism can calculate the probability density distribution of the plurality of rolling load values ​​as the probability density distribution of the rolling load values, or use the normal distribution as the reference distribution. The Rayleigh distribution roll condition determination mechanism can also calculate the maximum and minimum probability density distribution of the probability density distributions of the plurality of rolling load maxima and minimum values ​​contained in the high-frequency rolling load signal, as the probability density distribution of the rolling load values. The Rayleigh distribution roll condition determination mechanism can also use the Rayleigh distribution as the reference distribution. In the case where the roller condition determination mechanism includes both the normal distribution roller condition determination mechanism and the Rayleigh distribution roller condition determination mechanism, the monitored roller can also be determined to be abnormal if the determination result of at least one of them is abnormal.

[0029] In the aforementioned third roll condition monitoring device, as an example, the standard deviation σ of multiple rolling load values ​​can also be calculated. The probability density distribution of positive and negative kσ (after multiplying the standard deviation σ by a predetermined coefficient k) can also be compared with a normal distribution. The value of the difference between this probability density distribution and the normal distribution can also be used as the aforementioned test result. Alternatively, the value of the difference between the aforementioned maximum and minimum probability density distribution and the Rayleigh distribution can also be used as the aforementioned test result. The value of the difference between the calculated probability density distributions can also be a value selected from a group consisting of the KL distance, the sum of squared errors, and the sum of absolute error values.

[0030] In the above third roll condition monitoring device, the monitoring object roll can also have a first side end portion and a second side end portion opposite the first side end portion. The rolling load detection mechanism can also be configured to detect a first side rolling load signal from a first rolling load sensor provided at the first side end portion and to detect a second side rolling load signal from a second rolling load sensor provided at the second side end portion. The signal extraction mechanism can also extract a rolling load high frequency signal having a frequency of the above prescribed frequency or more from each of the first side rolling load signal and the second side rolling load signal. The roll condition determination mechanism can also be configured to determine the condition of each of the first side end portion and the second side end portion of the monitoring object roll based on the results of the statistical test method with respect to the rolling load high frequency signals extracted by the signal extraction mechanism.

[0031] In the above third roll condition monitoring device, the roll condition can also be determined based on "each stand test result" which is the result of the test method with respect to the statistics of each of a plurality of stands. In this case, in the above third roll condition monitoring device, the upper roll set can also include a plurality of upper roll sets which constitute the plurality of stands. The lower roll set can also include a plurality of lower roll sets which constitute the plurality of stands together with the plurality of upper roll sets, respectively. The rolling load detection mechanism can also acquire a plurality of rolling load signals from rolling load sensors of each of the plurality of stands. The signal extraction mechanism can also extract a plurality of rolling load high frequency signals having a frequency of the above prescribed frequency or more from the plurality of rolling load signals, respectively. The roll condition determination mechanism can also be configured to acquire a test result of each of the plurality of stands corresponding to the plurality of stands, respectively, as the result of the statistical test method with respect to a plurality of rolling load values included in each of the plurality of rolling load high frequency signals, and determine the condition of the monitoring object roll based on the test result of each of the plurality of stands.

[0032] In the above first to third roll condition monitoring devices, the "monitoring object roll" can also include at least one of an upper monitoring object roll and a lower monitoring object roll. The "upper monitoring object roll" is one roll selected from among the "upper roll set". The "lower monitoring object roll" is one roll selected from among the "lower roll set".

[0033] The upper side roll set includes an upper side work roll. In addition to this, the upper side roll set can also include an upper side backup roll, and can also include an upper side intermediate roll. In a case where the upper side roll set is composed only of the upper side work roll, the upper side monitoring target roll is the upper side work roll. In a case where the upper side roll set is composed of the upper side work roll and the upper side backup roll, at least one of the upper side work roll and the upper side backup roll is selected as the upper side monitoring target roll. In a case where the upper side roll set is composed of the upper side work roll, the upper side backup roll, and the upper side intermediate roll, at least one of the upper side work roll, the upper side backup roll, and the upper side intermediate roll is selected as the upper side monitoring target roll.

[0034] The lower side roll set includes a lower side work roll. In addition to this, the lower side roll set can also include a lower side backup roll, and can also include a lower side intermediate roll. In a case where the lower side roll set is composed only of the lower side work roll, the lower side monitoring target roll is the lower side work roll. In a case where the lower side roll set is composed of the lower side work roll and the lower side backup roll, at least one of the lower side work roll and the lower side backup roll is selected as the lower side monitoring target roll. In a case where the lower side roll set is composed of the lower side work roll, the lower side backup roll, and the lower side intermediate roll, at least one of the lower side work roll, the lower side backup roll, and the lower side intermediate roll is selected as the lower side monitoring target roll.

[0035] In the first to third roll state monitoring devices described above, the monitoring target roll can include both the upper side monitoring target roll and the lower side monitoring target roll. In this case, roll state determination of the upper side monitoring target roll and roll state determination of the lower side monitoring target roll can also be performed respectively.

[0036] In the case of the first roll state monitoring device and the second roll state monitoring device, the rolling load detection mechanism can also detect the upper rolling load with respect to the upper side monitoring target roll and the lower rolling load with respect to the lower side monitoring target roll respectively by distributing the output signal of the rolling load sensor at a preset ratio. The preset ratio can be 1:1, or a ratio other than this. Further, in this case, the load variation value extraction mechanism can also perform extraction of the upper rolling load variation value which is the value of the upper rolling load at each rotational position of the upper side monitoring target roll, and independently of this, perform extraction of the lower rolling load variation value which is the value of the lower rolling load at each rotational position of the lower side monitoring target roll.

[0037] Effects of the Invention

[0038] According to the first roll state monitoring device of the present application, the cumulative value which cumulates the rolling load or the roll gap equivalent value is found for each roll rotational position. By correcting the cumulative value with a correction coefficient corresponding to the number of roll rotations respectively, the roll eccentricity amount can be calculated for each roll rotational position. Thus, compared to a case where one rolling load detection value and one discrimination value are calculated in a one-to-one relationship, precision degradation caused by abnormal values due to noise and the like can be suppressed, so there is an advantage that discrimination with high precision can be performed.

[0039] In the second roll condition monitoring device of the present application, the normal roll eccentricity representative value is a value representative of a plurality of roll eccentricities calculated by the discrimination mechanism when the condition of the monitoring object roll is normal. The normal roll eccentricity representative value is used as a criterion for determining the condition of the roll. The normal roll eccentricity representative value is generated based on actual discrimination data obtained when the monitoring object roll was normal during a past rolling period. By using the normal roll eccentricity representative value based on a plurality of roll eccentricities, an appropriate roll condition determination criterion for each rolling plant can be established while suppressing the influence of abnormal values. Thus, there is an advantage of improving the determination accuracy of the roll eccentricity.

[0040] According to the third roll condition monitoring device of the present application, it is possible to statistically determine whether a plurality of rolling load values included in the rolling load high frequency signal are included within a normal value. The roll condition determination based on the statistical determination can determine the presence or absence of roll eccentricity abnormality with good accuracy based on the overall trend, as compared with the roll condition determination based on the detection result of a single or small number of data. Thus, it is possible to monitor the roll eccentricity abnormality with good accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a diagram illustrating an example of a rolling mill to which the roll condition monitoring device according to Embodiment 1 is applied.

[0042] Figure 2 is a diagram for explaining the structure of the roll condition monitoring device, the upper roll set, and the lower roll set according to Embodiment 1.

[0043] Figure 3 is a diagram for explaining the relationship between the division of the backup rolls and the work rolls according to Embodiment 1.

[0044] Figure 4 is a diagram illustrating the condition of the variation of the rolling load according to Embodiment 1.

[0045] Figure 5 is a diagram for specifically explaining the method of extracting the variation of the rolling load and the roll eccentricity discrimination according to Embodiment 1, and the structure of the device that implements it.

[0046] Figure 6 is a flowchart for explaining the first roll condition determination technique according to Embodiment 1.

[0047] Figure 7 is a flowchart for explaining the second roll condition determination technique according to Embodiment 1.

[0048] Figure 8 is a flowchart for explaining the second roll condition determination technique according to Embodiment 1.

[0049] Figure 9 FIG. 4 is a graph showing the progress of the actual roll eccentricity amount relating to Embodiment 1.

[0050] Figure 10 FIG. 5 is a graph showing the structure of the roll condition monitoring device relating to the first modification of Embodiment 1.

[0051] Figure 11 FIG. 6 is a graph for specifically showing the method of extracting the rolling load variation and the roll eccentricity amount discrimination relating to the fifth modification of Embodiment 1, and the structure of the device that implements it.

[0052] Figure 12 FIG. 7 is a graph showing an example of a rolling mill to which the roll condition monitoring device relating to Embodiment 2 is applied.

[0053] Figure 13 FIG. 8 is a graph showing the structure of the roll condition monitoring device, the upper roll set, and the lower roll set relating to Embodiment 2.

[0054] Figure 14 FIG. 9 is a graph for showing the roll condition determination technique relating to Embodiment 2.

[0055] Figure 15 FIG. 10 is a graph showing the probability density distribution relating to Embodiment 2.

[0056] Figure 16 FIG. 11 is a graph showing the probability density distribution relating to the first modification of Embodiment 2.

[0057] Figure 17 FIG. 12 is a graph showing the probability density distribution relating to the first modification of Embodiment 2.

[0058] Figure 18 FIG. 13 is a graph showing the minimum value and the maximum value relating to the first modification of Embodiment 2.

[0059] Figure 19 FIG. 14 is a graph showing the KL (Kullback-Leibler) distance relating to Embodiment 2.

[0060] Figure 20 FIG. 15 is a graph showing an example of the hardware structure of the roll condition monitoring device relating to Embodiments 1 and 2. DETAILED DESCRIPTION

[0061] Embodiment 1.

[0062] Figure 1 FIG. 16 is a graph showing an example of a rolling mill 50 to which the roll condition monitoring device 20 relating to Embodiment 1 is applied. Figure 1The rolling mill 50 shown includes a heating furnace 52 for heating slab 51, a roughing mill 53, a bar heater 54 for heating bar 55, a finishing mill 57, an inlet thermometer 56 disposed on the inlet side of the finishing mill 57, a plate thickness and width gauge 58 for measuring plate thickness and width, an outlet thermometer 59 disposed on the outlet side of the finishing mill 57, an output roller table 63, a thermometer 60, a coiler 61, and a roll condition monitoring device 20.

[0063] A thermometer 60 is positioned on the inlet side of the winding machine 61. The winding machine 61 winds up the product roll 62. Figure 1 The diagram illustrates the rolling direction RD, the operator side OS, and the drive side DS. The roll condition monitoring device 20 of Embodiment 1 is provided as a function included in the control device for controlling the rolling mill 50 that rolls the rolled material 1.

[0064] In this embodiment, a rolling mill 50 in a hot sheet rolling process will be used as a specific example for explanation. In Embodiment 1, a rolling mill 50 including a 2-stage roughing mill 53 and a 7-stage finishing mill 57 is illustrated as an example, but this is only one example.

[0065] Typically, rolling mills thin out blocks of steel or non-ferrous materials such as aluminum and copper by rolling them, making them easier to process into automotive or electrical products. There are various types of rolling mills. These include hot plate mills for rolling sheet metal, cold rolling mills, mills for rolling bars and wire rods, mills for rolling H-beams, and 12-stage and 20-stage mills for rolling harder materials such as stainless steel. The rolls used in each rolling process are also diverse. In these various types of rolling mills, the roll condition monitoring device 20 described in Embodiment 1 can be used. This is because, although there are minor differences in specifications, the structures of the various types of rolling mills that have been put into practical use are often similar.

[0066] exist Figure 1 The rolling mill 50 shown includes a two-stage roughing mill 53 and a seven-stage finishing mill 57. Furthermore, although not shown, it is equipped with a high-capacity electric motor for driving the rolling rolls up and down. Although not shown, it is also possible to provide shafts or similar components connecting the rolls to the electric motor.

[0067] Figure 1 In the case where the roughing mill 53 has only one work roll 3a and one work roll 3b, it can also be composed of four rolls in total: work rolls 3a and 3b and support rolls 4a and 4b with a diameter larger than theirs. On the other hand, Figure 1 The finishing mill 57 has the first rolling stand #1 to the seventh rolling stand #7.

[0068] Each roll stand of the finishing mill 57 is composed of four rolls arranged in upper and lower rows. That is, it is composed of work rolls 3a, 3b and backup rolls 4a, 4b. One or more intermediate rolls can be provided between the work rolls 3a, 3b and the backup rolls 4a, 4b, in which case one roll stand can be composed of six or more rolls arranged in upper and lower rows.

[0069] The roll condition monitoring device 20 relating to Embodiment 1 monitors the roll condition of the finishing mill 57. However, as a modification, the roll condition monitoring device 20 can monitor the roll condition of the roughing mill 53, and the roll condition monitoring device 20 can monitor the roll condition of both the roughing mill 53 and the finishing mill 57.

[0070] The roll condition monitoring device 20 relating to Embodiment 1 is configured to detect an abnormality of a roll and notify the abnormality in advance by monitoring the condition of the roll for rolling. The roll condition monitoring device 20 can accurately identify the roll eccentricity amount, and determine an abnormality by comparing the identified roll eccentricity amount with the roll eccentricity amount in a normal condition. The roll condition monitoring device 20 can also have various types of reporting mechanisms such as a display or an alarm signal that present the determination result of the roll condition to an operator or the like.

[0071] Figure 2 is a view for explaining the structure of the roll condition monitoring device 20 relating to Embodiment 1, the upper roll set, and the lower roll set. Figure 2 indicates one roll stand in the finishing mill 57 relating to Embodiment 1 and the roll condition monitoring device 20 connected thereto.

[0072] Figure 1 The first roll stand #1 to the seventh roll stand #7 included in the finishing mill 57 of Figure 2 have the structure shown in Figure 2 As shown in

[0073] As shown in Figure 2 The work rolls 3a, 3b are composed of an upper work roll 3a and a lower work roll 3b. The backup rolls 4a, 4b are composed of an upper backup roll 4a and a lower backup roll 4b. An oil bearing can be used in the bearing for rotating the backup rolls 4a, 4b. The screwdown mechanism 5 is a screwdown device that applies a rolling load to the material 1. The rolling load detection mechanism 6 is a device that detects the rolling load.

[0074] The roll rotational speed detector 7 detects the roll rotational speed. The roll rotational speed referred to here means the number of rotations of the roll. The roll rotational speed detector 7 can also be a counter that increments by 1 each time the roll rotates by 1 revolution. Alternatively, in the case where the roll rotational speed detector 7 is a sensor that measures the roll rotational speed (i.e., the number of rotations of the roll per unit time), the number of rotations of the roll in a certain time can also be calculated by multiplying the roll rotational speed by the time.

[0075] The roll reference position detector 8 detects a prescribed reference position each time the backup roll 4a, 4b rotates by 1 revolution. The roll gap detector 9 detects the gap between the work rolls 3a, 3b, i.e., the roll gap.

[0076] The upper roll set is formed by the upper work roll 3a and the upper backup roll 4a. On the other hand, the lower roll set is formed by the lower work roll 3b and the lower backup roll 4b.

[0077] In Embodiment 1, the case of a 4Hi rolling mill is described as an example. The 4Hi rolling mill is formed by 4 rolls of 2 work rolls 3a, 3b and 2 backup rolls 4a, 4b. However, the structure is not limited to this, and it can also be a so-called 2Hi rolling mill. The 2Hi rolling mill is formed by only 2 work rolls. Alternatively, it can also be a so-called 6Hi rolling mill. The 6Hi rolling mill is formed by 6 rolls of 2 work rolls, 2 intermediate rolls, and 2 backup rolls. Alternatively, it can also be a rolling mill formed by a number of rolls greater than this number.

[0078] The work rolls 3a, 3b, which have been appropriately adjusted in terms of roll gap and speed, roll the material 1 to become the desired plate thickness on the exit side. The upper work roll 3a is supported from above by the upper backup roll 4a. The lower work roll 3b is supported from below by the lower backup roll 4b. As a result, the deflection in the roll width direction is reduced. Furthermore, the backup rolls 4a, 4b are supported so as to be able to rotate freely with respect to the rolling mill housing 2. The backup rolls 4a, 4b are formed in a structure that is able to sufficiently withstand the rolling load acting on the material 1.

[0079] The screwdown mechanism 5 adjusts the gap between the work rolls 3a, 3b, i.e., the roll gap. In the screwdown mechanism 5, an electric screwdown device based on motor control or an oil pressure screwdown device based on oil pressure control is used. Since the oil pressure screwdown has the advantage of being easier to obtain a high-speed response, the screwdown mechanism 5 can also be an oil pressure screwdown device.

[0080] In control performed in correspondence with a wave component of a short period such as interference caused by roll eccentricity, it is generally preferable to use an oil pressure screwdown that is able to respond at high speed. However, as a modification, the screwdown mechanism 5 can also be an electric screwdown device. In the case of monitoring the roll state, since the high-speed nature of the screwdown mechanism is not a concern, the roll state monitoring device 20 can also be applied to a rolling stand that does not have an oil pressure screwdown.

[0081] The rolling load detection mechanism 6, for example, detects the rolling load. One example of a rolling load detection method is to directly measure the rolling load using a load cell embedded between the mill housing 2 and the reduction mechanism 5. Another example of a rolling load detection method is to calculate the rolling load based on the pressure detected by the hydraulic reduction mechanism. The rolling load detection mechanism 6 can also be, for example, a load sensor or a pressure sensor; specifically, it can be a strain gauge, a load cell, or a hydraulic sensor.

[0082] Roller speed detector 7 detects the rotational speed of work rolls 3a, 3b, etc. Roller speed detector 7 can also be installed on work rolls 3a, 3b. Alternatively, roller speed detector 7 can be installed on the shaft (not shown) of the motor that drives work rolls 3a, 3b.

[0083] The roller speed detector 7 may, for example, include a pulse output mechanism that outputs pulses corresponding to the rotation angles of the work rollers 3a and 3b, and an angle calculation mechanism that detects the pulses output from the pulse output mechanism and calculates the rotation angles of the work rollers 3a and 3b. The roller speed detector 7 may also be composed of a pulse output mechanism and an angle calculation mechanism, so as to enable precise detection of the roller speed and rotation angle of the work rollers 3a and 3b.

[0084] Furthermore, when the ratio of the diameters of the work rolls 3a and 3b to the support rolls 4a and 4b is known, the rotational speed and rotation angle of the support rolls 4a and 4b can also be calculated. Specifically, based on the rotational speed and rotation angle of the work rolls 3a and 3b detected by the roll speed detector 7, the rotational speed and rotation angle of the support rolls 4a and 4b can be calculated under the condition that there is no slippage between the work rolls 3a and 3b and the support rolls 4a and 4b.

[0085] The roller reference position detector 8 detects the reference position by using sensors such as the support rollers 4a and 4b, which detect objects mounted on the support rollers 4a and 4b, for example, every time the support rollers 4a and 4b rotate once, or by using proximity switches. Alternatively, the roller reference position detector 8 can detect the reference position by using, for example, a pulse generator to extract pulses that depend on the rotation angle of the support rollers 4a and 4b, and by detecting the rotation angle of the support rollers 4a and 4b.

[0086] In addition, Figure 2 The diagram illustrates the case where the roller reference position detector 8 is installed only on the upper support roller 4a. However, as a variation, the roller reference position detector 8 can also be installed on both support rollers 4a and 4b, so that the reference positions of each support roller 4a and 4b can be detected independently.

[0087] The roll gap detector 9 is provided between the backup roll 4a and the screwdown mechanism 5 as an example. The roll gap detector 9 indirectly detects the roll gap formed between the work rolls 3a, 3b.

[0088] As shown in Figure 2 , the roll condition monitoring device 20 of Embodiment 1 is provided with a rolling load up-down distribution section 10, a rolling load variation extraction section 11, a roll eccentricity amount discrimination section 12, a roll eccentricity amount recording section 13, and a roll condition determination section 14. The roll condition monitoring device 20 determines the condition of the monitoring target roll. In Embodiment 1, the backup rolls 4a, 4b are taken as the monitoring target rolls as an example, respectively.

[0089] The rolling load detection mechanism 6 detects the rolling load with respect to a plurality of rotational positions of the work rolls 3a, 3b and the backup rolls 4a, 4b as described later in Figure 3 and Figure 4 . The rolling load up-down distribution section 10 up-down distributes the rolling load detected by the rolling load detection mechanism 6 based on the ratio of the upper side rolling load to the lower side rolling load. The distribution ratio is set in advance. The upper side rolling load is the load received by the upper side work roll 3a and the upper side backup roll 4a as the upper side roll set from the rolled material 1. The lower side rolling load is the load received by the lower side work roll 3b and the lower side backup roll 4b as the lower side roll set from the rolled material 1. In addition, the upper side rolling load and the lower side rolling load can be distributed, for example, at a ratio of 1:1. However, the actual lower side rolling load also bears the weight of the upper side work roll and the upper side backup roll. As a result, the lower side rolling load is slightly larger than the upper side rolling load as the actual load. The weight of the roll is 30 to 40 tons when the work roll and the backup roll are added together, and the rolling load is several hundred tons to two or three thousand tons with respect thereto. Thus, when the roll weight is taken into account, the lower side rolling load is slightly larger than the upper side rolling load as a ratio.

[0090] The rolling load variation extraction section 11 extracts the upper side rolling load variation value ΔP Tj and the lower side rolling load variation value ΔP Bj based on the rolling load of the upper side roll set and the lower side roll set up-down distributed by the rolling load up-down distribution section 10. The subscript j is j = 0, 1, 2,... n-1. The upper side rolling load variation value ΔP Tj and the lower side rolling load variation value ΔP Bj are variation values that occur in association with the rotational position of the upper side roll set and the lower side roll set.

[0091] The roll eccentricity amount discrimination section 12 transforms each variation component ΔP of the rolling load extracted by the rolling load variation extraction section 11 up and down into a roll gap equivalent value ΔS. The roll eccentricity amount discrimination section 12 converts the roll gap equivalent value ΔS obtained by the transformation into a roll eccentricity amount ΔE Figure 5The values outputted from the roll eccentricity amount identification section 12 are added in the plurality of adders 121d, 122d described later. The reason for performing the conversion into the roll gap equivalent value ΔS is because it is made so that an unnecessary deviation does not occur in the roll load variation value due to a difference in the characteristics of the rolled material (for example, the hardness of the rolled material). For example, because there is a tendency for the roll load variation to also become large in a harder material.

[0092] In addition, in the rolling mill 50, by actually adjusting the roll gap using the roll gap equivalent value ΔS, it is possible to reduce the plate thickness variation of the rolled material 1. However, in Embodiment 1, the roll state monitoring device 20 does not have a function of moving the roll gap to reduce the influence of the roll eccentricity on the plate thickness variation. Therefore, in Embodiment 1, the data is continuously added in the adders 121d, 122d during rolling, and the value in the adders 121d, 122d continuously increases in correspondence with the rotational speed of the roll. Therefore, in Embodiment 1, in order to obtain the roll eccentricity amount, the output value of the adders 121d, 122d is divided by a correction coefficient corresponding to the rotational speed of the roll to perform correction.

[0093] The roll eccentricity amount recording section 13 records the plurality of output values y Tj , y Bj The subscript j is j = 0, 1, 2,... n - 1. The output values y Tj , y Bj are the identification values of the roll eccentricity amount.

[0094] From the data recorded in the roll eccentricity amount recording section 13, it is possible to calculate the roll eccentricity amount peak interval value Δy peak . The roll eccentricity amount peak interval value Δy peak is the difference between the maximum value and the minimum value of the roll eccentricity amount identified by the roll eccentricity amount identification section 12.

[0095] The roll eccentricity amount recording section 13 records the roll eccentricity amount peak interval value Δy peak identified by the roll eccentricity amount identification section 12 during a predetermined rolling period set in advance as the "normal roll eccentricity amount peak interval value Δy nor_peak ". The normal roll eccentricity amount peak interval value Δy nor_peak is a determination value indicating the roll eccentricity amount peak interval value Δy peak when the monitoring object roll is in a normal state.

[0096] In addition, the "predetermined rolling period set in advance" described above can be either a period from when the roll is just replaced until a predetermined time elapses, or a period from when the roll is just replaced until a predetermined number of rolled materials 1 are rolled. The roll eccentricity amount peak interval value Δy peak of each rolled material 1 is obtained each time the rolling of the rolled material 1 ends. The obtained roll eccentricity amount peak interval value Δy peakRoll eccentricity peak value Δy peak Recorded.

[0097] In addition, as a modification of the roll eccentricity recording section 13, instead of the roll eccentricity peak value Δy peak the roll eccentricity maximum value y max (or the peak value on the positive side) or the roll eccentricity minimum value y min (or the peak value on the negative side) can be recorded. In this modification, the roll eccentricity recording section 13 can record the roll eccentricity maximum value y max or the roll eccentricity minimum value y min respectively. At this time, the roll eccentricity recording section 13 records the roll eccentricity maximum value y max or the roll eccentricity minimum value y min identified by the roll eccentricity identifying section 12 within a predetermined rolling period set in advance. The roll eccentricity maximum value y max or the roll eccentricity minimum value y min in the normal state of the roll is recorded. The roll eccentricity maximum value y max in the normal state of the roll is also called "normal roll eccentricity maximum value y nor_max ". The roll eccentricity minimum value y min in the normal state of the roll is also called "normal roll eccentricity minimum value y nor_min ".

[0098] In addition, the period from when the roll is newly replaced until a certain time or until a certain number of rolled materials are rolled is set as the period required for rolling of the "predetermined number". The predetermined number is preferably set to a number such as 5 or 10, which is a certain degree more than a number. The value of 5 or 10 is described. The replacement cycle of the work roll is the time when 100 or so of the rolled materials 1 are rolled. If the predetermined number described above is set to 40 to 50, the number of the rolled materials 1 that are the objects of determination of normality or abnormality becomes very small, which is not practical. Therefore, the predetermined number described above is preferably, for example, 10 or so, which is within 10% of 100. If supplemented, the replacement cycle of the backup roll is several days to 10 days or so. The number of the rolled materials 1 rolled in this period reaches several thousand. Therefore, when the backup roll is the monitoring object, the predetermined number can be set to be more than 5 to 10. The work roll is easily worn in the vicinity of the central portion in the width direction because it directly contacts the rolled material, and therefore the roll needs to be frequently replaced and ground. Therefore, the work roll is set to the replacement cycle described above. On the other hand, the backup roll can have a longer replacement cycle because it does not directly contact the rolled material. In addition, the roll can be normal after being ground. This is because, when the roll enters the human eye in the grinding process, abnormalities can be easily found if there are any.

[0099] The roll condition determination section 14 determines the condition of each of the backup rolls 4a, 4b as a monitoring target roll using the data recorded in the roll eccentricity amount recording section 13.

[0100] In Embodiment 1, the roll condition determination section 14 can also perform a comparison determination based on data within a set time after roll replacement, for example. This comparison determination is implemented by the routine of Figure 6 described later. Further, the roll condition determination section 14 of the modification example can determine the normality and abnormality of the roll condition not based on data within a set time after roll replacement, but based on a fixed value or a statistical value determined from data obtained in the past. This modification example is implemented by the routine of Figure 7 described later. The specific method of determination by the roll condition determination section 14 will be described later using Figure 6 and Figure 7 .

[0101] Next, the operation of the roll condition monitoring device 20 relating to Embodiment 1 will be described in detail with reference to Figures 3-8 .

[0102] First, the structures and operations of the rolling load up-down distribution section 10 and the rolling load variation extraction section 11 will be described in detail with reference to Figure 3 and Figure 4 . Figure 3 is a diagram for explaining the relationship between the backup rolls 4a, 4b and the work rolls 3a, 3b relating to Embodiment 1. Figure 3 indicates the positional relationship of the work rolls 3a, 3b and the backup rolls 4a, 4b. In addition, the backup roll is sometimes referred to simply as "BUR", and the work roll is sometimes referred to simply as "WR".

[0103] As shown in Figure 3 , the backup rolls 4a, 4b are provided with a position scale 15 for rotational position detection. Further, a reference position 4c that is set in advance in a portion of the backup rolls 4a, 4b and rotates in conjunction with the rotation of the backup rolls 4a, 4b is indicated. The position scale 15 is provided, for example, on the outermost side of the backup rolls 4a, 4b in a manner of surrounding the periphery of the backup rolls 4a, 4b. The scale is provided in a manner of equally dividing the entire circumference n of the backup rolls 4a, 4b. That is, the scale is provided at every predetermined angle (360 / n degrees) with the rotation axis of the backup rolls 4a, 4b as the center. Further, the reference position 15a (fixed reference position) of the position scale 15 is set to 0, and numbers are assigned up to the (n-1)th. In addition, the above n is set to a value of, for example, n = 30 to 90 or so. Here, the above position scale 15 is provided for the purpose of explanation of the rolling load variation extraction section 11 and the like, and the scale itself can not be provided on the actual device.

[0104] Here, θ WT0θ is the rotation angle of the work roller 3 when the reference position 4c of the support rollers 4a and 4b coincides with the fixed reference position 15a. WT The support rollers 4a and 4b rotated by θ. BT The rotation angle of the work roller 3. Here, θ represents the angle, W represents the work roller 3, B represents the support roller 4, T represents the upper roller, and B represents the lower roller.

[0105] Furthermore, the rotation angles of the support rollers 4a and 4b described below represent the angles by which the reference position 4c of the support rollers 4a and 4b moves from the fixed reference position 15a in conjunction with the rotation of the support rollers 4a and 4b. For example, a rotation angle of 90 degrees for the support rollers 4a and 4b indicates that the reference position 4c of the support rollers 4a and 4b is at a position that has rotated 90 degrees from the fixed reference position 15a in the rotation direction of the support rollers 4a and 4b. Moreover, the state where the rotation angle of the support rollers 4a and 4b is at the closest scale of position scale 15 (e.g., the j-th scale of position scale 15) is defined as the rotation angle number of the support rollers 4a and 4b as j.

[0106] Alternatively, a roller reference position detector 8 can be constructed by embedding a proximity sensor or similar sensor at the reference position 4c of the support rollers 4a and 4b and at the fixed reference position 15a, along with a detection object detected by the sensor. In this case, for example, the proximity sensor located at the reference position 4c of the support rollers 4a and 4b rotates together with the support rollers 4 to reach the fixed reference position 15a, and the detection object embedded at the reference position 15a is detected by the proximity sensor. That is, it is identified that the reference position 4c of the support rollers 4a and 4b has passed through the fixed reference position 15a. Furthermore, the roller reference position detector 8 is not essential for Embodiment 1.

[0107] The division positions from the fixed reference position 0 to n-1 will be set as described later. Figure 5 Recording area of ​​rolling load in Figure 5 P0 to P n－1 The division of the regions is equal, and the rolling loads at these division locations are saved in the recording area. Typically, a value of n = 30 to 90 is used. To increase n, it is preferable that the controller's processing power is sufficiently high; therefore, it is preferable to consider the inverse relationship between control fineness and processing power.

[0108] Hereinafter, the so-called support roll rotation angle indicates the angle by which the support roll reference position moves from the fixed reference position with the rotation of the support rolls 4a, 4b. For example, the so-called support roll rotation angle is 90 degrees, indicating the position at which the support roll reference position moves from the fixed reference position to a position at 90 degrees in the direction of rotation of the support rolls 4a, 4b. Further, it is assumed that when the support roll rotation angle is at the nearest position of the above-mentioned position scale (for example, the i-th position scale), the support roll rotation angle number is i.

[0109] Figure 4 is a graph illustrating the condition of the variation of the rolling load relating to Embodiment 1. Based on the above-mentioned rolling load P Figure 4 , the method of extracting the variation component of the roll eccentricity caused by the rolling load is described.

[0110] Figure 4 indicates the variation of the rolling load accompanying the change of the rotation angle of the support roll. In the above-mentioned graph, the rolling load P Figure 4 is shown as P 10 when the reference position 4c of the support roll 4 is at the reference position 15a, that is, when the rotation angle number of the support roll 4 is 0. As the rotation angle number of the support roll 4 increases by 1, 2, 3,..., the rolling load changes to P 11 , P 12 , P 13 ,.... Further, the support roll 4 rotates by 1 revolution, and the rotation angle number becomes 0 again from (n-1).

[0111] In the case where the rolling loads P 20 at the time when the rolling load P 10 and P 20 are connected by a straight line 103, the straight line 103 can also be regarded as the rolling load other than the variation of the rolling load by the roll eccentricity. Thus, the variation of the rolling load by the roll eccentricity can also be found from the difference between the rolling loads P 11 , P 12 , P 13 ... P 20 measured at each rotation angle number and the above-mentioned straight line 103.

[0112] Further, in the value (actual value) of the actually measured rolling load P ij , the noise component is often included in addition to the variation of the rolling load by the temperature variation / plate thickness variation / tension variation, etc. and the variation of the rolling load by the roll eccentricity. Therefore, the actual value of the actual rolling load P ij is not distributed on the smooth curve shown in Figure 4 , but there are cases where it is difficult to determine the rolling load P i0 at the start point and the rolling load P(i+1)0 the case.

[0113] Therefore, the average value-based calculation described below can also be performed. First, assume that the rolling load P i0 and the rolling load P (i+1)0 vary little. If so, the measured respective rolling loads P i0 , P i1 , P i2 , P i3 ... P (i+1)0 may also be regarded as the variation components of the rolling load resulting from the roll eccentricity, with respect to the difference ΔP AVE_n of the average value ΔP ij of each. The average value ΔP AVE_n is the n average values of the rolling loads P i0 , P i1 , P i2 , P i3 ... P i(n－1) .

[0114] The average value-based calculation method has the advantage that the actual values of the rolling loads can be reduced up to the (n-1)th zone, and is also strong against variations in the rolling loads caused by noise and the like. In addition, it is also effective to reduce the noise components by applying a filter process to the actual values of the rolling loads.

[0115] Figure 5 is a diagram for specifically explaining the method of extracting the rolling load variation and the roll eccentricity amount discrimination relating to Embodiment 1, and the specific device structure for realizing it. Based on Figure 5 , the specific structure and operation of the above-described rolling load variation extraction section 11 and the roll eccentricity amount discrimination section 12 will be described. As Figure 5 shown in the drawing, the rolling load variation extraction section 11 is provided with an upper-side load variation extraction section 111 and a lower-side load variation extraction section 112.

[0116] The upper-side load variation extraction section 111 extracts the upper-side rolling load variation value ΔP T based on the rolling loads P T distributed by the rolling load up-down distribution section 10. The upper-side rolling load variation value ΔP T is a value in which the variation components of the rolling loads P Tj at a plurality of rotational positions of the upper-side backup roll 4a resulting from the roll eccentricity are extracted. For the plurality of rotational positions of the upper-side backup roll 4a, a plurality of upper-side rolling load variation values ΔP T0 , ΔP T1 ,... ΔP Tn－1 are respectively calculated.

[0117] The lower side load variation extracting section 112 extracts a lower side rolling load variation value ΔP B based on the rolling load P B assigned by the rolling load up and down assigning section 10. The lower side rolling load variation value ΔP B is a value of a variation component due to eccentricity of the roll, which is extracted at a plurality of rotational positions of the lower side backup roll 4b. Bj For the plurality of rotational positions of the lower side backup roll 4b, a plurality of lower side rolling load variation values ΔP B0 , ΔP B1 ,... ΔP Bn－1 are respectively calculated.

[0118] Further, the upper side load variation extracting section 111 has a rolling load recording section 111a, an average value operation mechanism 111b, and a deviation operation mechanism 111c. Similarly, the lower side load variation extracting section 112 also has a rolling load recording section 112a, an average value operation mechanism 112b, and a deviation operation mechanism 112c.

[0119] The rolling load recording sections 111a, 112a are n rolling load recording sections provided corresponding to each rotational angle number of the backup rolls 4a, 4b, respectively. In each rolling load recording section 111a, 112a, the rolling load P Tj , P Bj at which the backup roll 4a, 4b reached the corresponding rotational angle number during a prescribed period is recorded.

[0120] The average value operation mechanism 111b operates an average value ΔP Tj based on the rolling loads P AVE_Tn recorded in each rolling load recording section 111a. The average value ΔP AVE_Tn is an average value of the n rolling loads P Tj (j = 0 ~ (n - 1)) detected during one revolution of the upper side backup roll 4a.

[0121] The average value operation mechanism 112b operates an average value ΔP Bj based on the rolling loads P AVE_Bn recorded in each rolling load recording section 112a. The average value ΔP AVE_Bn is an average value of the n rolling loads P Bj (j = 0 ~ (n - 1)) detected during one revolution of the lower side backup roll 4b.

[0122] The plurality of deviation operation mechanisms 111c are provided in a one-to-one correspondence with the plurality of rolling load recording sections 111a, respectively. The deviation operation mechanism 111c operates a plurality of deviations ΔP Tj and outputs them every time the backup roll 4a rotates one revolution. The plurality of deviations ΔP Tj are the rolling loads PTj the average value ΔP AVE_Tn . The rolling load P Tj is recorded in the corresponding rolling load recording section 111a. The deviation calculation mechanism 112c of the lower side load variation extraction section 112 also outputs the deviation ΔP Bj by performing the same operation processing.

[0123] The roll eccentricity amount discrimination section 12 has an upper side addition mechanism 121 and a lower side addition mechanism 122.

[0124] The upper side addition mechanism 121 has a conversion block 121a, a limiter 121b, a switch 121c, an adder 121d, and a rotational speed correction block 121e. The upper side addition mechanism 121 converts the variation component of the rolling load P Tj output from the upper side load variation extraction section 111 due to the roll eccentricity into a roll gap equivalent value ΔS Tj using the conversion block 121a. The converted roll gap equivalent value ΔS Tj passes through the limiter 121b and the switch 121c, and is added by the plurality of adders 121d for each rotational angle number, respectively.

[0125] The lower side addition mechanism 122 has a conversion block 122a, a limiter 122b, a switch 122c, an adder 122d, and a rotational speed correction block 122e. The lower side addition mechanism 122 converts the variation component of the rolling load P Bj output from the lower side load variation extraction section 112 due to the roll eccentricity into a roll gap equivalent value ΔS Bj using the conversion block 122a. The converted roll gap equivalent value ΔS Bj passes through the limiter 122b and the switch 122c, and is added by the plurality of adders 122d for each rotational angle number, respectively.

[0126] In addition, in Figure 5 , in order to distinguish, the roll gap equivalent value input to the limiter 121b is particularly noted as ΔS Tj LM , and the roll gap equivalent value output from the limiter 121b is noted as ΔS Tj . Similarly, the roll gap equivalent value input to the limiter 122b is particularly noted as ΔS Bj LM , and the roll gap equivalent value output from the limiter 122b is noted as ΔS Bj . However, as a modification example of Embodiment 1, the limiters 121b, 122b can be omitted, and in the case where such a structure is omitted, the distinction of the roll gap equivalent values before and after the limiter is not required.

[0127] Further, the upper side adding mechanism 121 and the lower side adding mechanism 122 have the same structure. Thus, the following describes the operation of the upper side adding mechanism 121, and the description of the lower side adding mechanism 122 is omitted or simplified as needed.

[0128] In the upper side adding mechanism 121, first, the conversion block 121a corresponding to the j-th rotational position converts the load variation value ΔP Tj to the roll gap equivalent value ΔS Tj . The operation processing of the conversion block 121a can be implemented based on the following equation (3). Assume that the load variation value ΔP and the roll gap equivalent value ΔS of the equation (3) are ΔP Tj and ΔS Tj , respectively. In the equation (3), M is a rolling mill constant, and Q is a plasticity coefficient of the rolled material. These parameters are calculated through a setting calculation performed before the passage of each rolled material.

[0129]

[0130] The following describes the reason for converting the rolling load variation value ΔP to the roll gap equivalent value ΔS using the above-described equation (3). If the steel type is different, it is possible that the rolling load variation value is also different. For example, ΔP of a harder steel type is larger, and on the other hand, ΔP of a softer steel type is smaller. Suppose that in the case where the normal roll eccentricity peak value interval Δy nor_peak is calculated by rolling a softer steel type after a roll change, then ΔP is detected to be larger by rolling a harder material. In this case, according to the setting of the threshold value, it is possible that the roll is determined to be abnormal when rolling the harder material is performed.

[0131] To this point, if the above-described equation (3) is used, since the roll gap equivalent value is used, a substantially constant value is calculated regardless of whether it is a softer material or a harder material, as long as the roll state is normal. Thus, it is possible to determine whether the roll state is normal with good accuracy. Further, the conversion block 122a of the lower side adding mechanism 122 also calculates ΔS B in the same way as the conversion block 121a by performing operation processing according to the equation (3).

[0132] The limiter 121b of the upper side adding mechanism 121 checks the upper and lower limits of each of the plurality of roll gap equivalent values ΔS Tj (j = 0, 1,..., n - 1) input from the plurality of deviation operation mechanisms 111c. The limiter 122b of the lower side adding mechanism 122 also checks the upper and lower limits of each of the plurality of roll gap equivalent values ΔS Bj (j = 0, 1,..., n - 1) in the same way as the limiter 121b. Through the limiter 121b and the limiter 122b, the roll gap equivalent values ΔS Tj , ΔS BjThe value is limited to a preset range. Furthermore, the purpose of limiters 121b and 122b is to detect roller abnormalities. If the upper and lower limit values ​​of each of the limiters 121b and 122b are set too narrow, abnormalities may not be detected. Preferably, the upper and lower limit values ​​of each of the limiters 121b and 122b are not set too narrow. These limiters 121b and 122b are set to avoid the influence of sudden and large noise. Here, for convenience, the upper and lower limit values ​​of each of the limiters 121b and 122b are also referred to as "limiter amplitude". An example of a method for setting the limiter amplitude is described below. (Further details will follow.) Figure 6 The decision process in step S1403 of the flowchart uses a coefficient m. The coefficient m is... Figure 6 The coefficient used for anomaly determination in step S1403. The limiter amplitude can also be set according to the relationship with this coefficient m. The value of multiplying the normal roll eccentricity, the maximum roll eccentricity, or the minimum roll eccentricity by m is used as the comparison determination value for anomaly determination. Since m is set to 2 as an example of whether it is an anomaly, it is meaningless to set a value smaller than 2 in the limiter. Here, m = 2 means m times the normal roll eccentricity or the maximum or minimum roll eccentricity. Therefore, it is preferable to also measure or assume the normal roll eccentricity in advance as the limiter, and set the upper and lower limit amplitude values ​​to be (2m) times or more of this value. As a result, it is possible to prevent the limiter amplitude from being set too narrow.

[0133] Switch 121c includes n unit switches SW corresponding to each rotation angle number of the upper support roller 4a. TI For each revolution of the upper support roller 4a (i.e., whenever the average value calculation in the average value calculation mechanism 111b ends), n unit switches, including switch 121c, are activated in sequence according to their rotation angle numbers. Switch 121c controls the roller gap equivalent ΔS after passing through the limiter 121b. T0 , …ΔS Tn－1 Output to the adder 121d in the later stage.

[0134] In addition, the switch 122c of the lower addition mechanism 122 also includes n unit switches SW corresponding to each rotation angle number of the lower support roller 4b. BI Switch 122c, through the same action as switch 121c, sets the roller gap equivalent value ΔS. B0 , …ΔS Bn－1 Output to the adder 122d in the later stage.

[0135] Adder 121d includes n unit adders Σ arranged according to each rotation angle number of the upper support roller 4a. T0 , Σ T1 ,…Σ Tj ,…Σ Tn－1n unit adders Σ T0 n unit adders Σ T1 n unit adders Σ Tn－1 n unit adders Σ T0 n unit adders Σ Tn－1 n unit adders Σ ATj (j = 0, 1,..., n-1) are calculated by individually adding the roll gap equivalent values ΔS

[0136] As an example, when the upper side support roll 4a has rotated ten times, the cumulative value ΔS T0 calculated by the unit adder Σ AT0 is a cumulative value that adds up 10 roll gap equivalent values ΔS T0 . Similarly, in the adder 122d of the lower side addition mechanism 122, n unit adders Σ B0 , Σ B1 ,..., Σ Bj ,..., Σ Bn－1 also individually add the roll gap equivalent values ΔS B0 ,..., ΔS Bn－1 , and calculate n cumulative values ΔS ABj (j = 0, 1,..., n-1).

[0137] In addition, it is also possible to clear the adders 121d, 122d once the rolling of 1 piece of the rolled material is completed.

[0138] The rotation speed correction block 121e is a function that corrects the case where the roll eccentricity amount is continuously accumulated. In Embodiment 1, since the roll eccentricity of the actual device is not suppressed by the reduction control operation based on the roll eccentricity amount and the like, the rotation speed correction block 121e specifically divides the output from the adder 121d by the roll rotation speed. The rotation speed correction block 121e outputs the calculation result to the n roll division numbers.

[0139] The correction operation of the rotation speed correction block 121e is an operation that corrects the output from the adder 121d by a correction coefficient corresponding to the roll rotation speed. The correction coefficient is preferably a variable value that is set to be larger as the number of rotations of the monitoring object roll during the accumulation period of the n cumulative values ΔS ABj (j = 0, 1,..., n-1) is larger. In Embodiment 1, the correction coefficient is set to the same value as the number of rotations of the monitoring object roll, but it is also possible to be a correction coefficient other than this. As another example, the correction coefficient can also be set to be less than or more than the number of rotations of the monitoring object roll. For example, the correction coefficient can also be a value that subtracts or adds a value set in advance from or to the number of rotations of the monitoring object roll. As another example, it is also possible to calculate the correction coefficient as a variable value that is proportional to the number of rotations of the monitoring object roll by multiplying a proportion coefficient set in advance by the number of rotations of the monitoring object roll.

[0140] Further, the rotation speed correction block 122e of the lower addition mechanism 122 also performs the same correction operation as the rotation speed correction block 121e. The output value y T0 ,... y Tn－1 of the rotation speed correction block 121e and the output value y B0 ,... y Bn－1 of the rotation speed correction block 122e are the roller eccentricity amounts identified by the roller eccentricity amount identification section 12.

[0141] Through the above-described mechanism, Figure 5 the upper addition mechanism 121 of the upper side outputs the roller eccentricity amount y T0 ,... y Tn－1 of the upper support roller 4a that is the monitoring target roller in the upper roller set. Figure 5 the lower addition mechanism 122 of the lower side outputs the roller eccentricity amount y B0 ,... y Bn－1 of the lower support roller 4b that is the monitoring target roller in the lower roller set.

[0142] (Details of the specific processing related to the roller state determination)

[0143] Next, the operation of the roller eccentricity amount recording section 13 and the roller state determination section 14 will be described using Figures 6-8 As shown in Figure 2 , the roller eccentricity amount recording section 13 stores the roller eccentricity amount y Tj of the upper monitoring target roller (i.e., the upper support roller 4a) and the roller eccentricity amount y Bj of the lower monitoring target roller (i.e., the lower support roller 4b) that are transmitted from the roller eccentricity amount identification section 12. The roller state determination section 14 determines the roller state based on the data taken out from the roller eccentricity amount recording section 13 according to one of the routines of Figure 6 and Figure 7 and Figure 8 .

[0144] Figure 6 is a flowchart for explaining the first roller state determination technique of Embodiment 1. Figure 6 The routine of Figure 6 is executed by the roller eccentricity amount recording section 13 and the roller state determination section 14. In Figure 5 , the method of determining the abnormality of the roller state by the roller eccentricity amount recording section 13 and the roller state determination section 14 after the roller eccentricity amount of the rolled material is identified in is shown.

[0145] In Embodiment 1, the first determination method, the second determination method, and the third determination method are provided as the first roller state determination technique. The first determination method is a method of comparing the normal roller eccentricity amount peak interval value Ay nor_peak with the roller eccentricity amount peak interval value Ay peakThe comparison method. The second judgment method is to use the maximum value of the normal roller eccentricity y. nor_max The maximum value of the roll eccentricity y for each rolled material max The comparison method. The third judgment method is to use the minimum value of the normal roller eccentricity y. nor_min Minimum eccentricity y of the rolls relative to each rolled material min The method of comparison.

[0146] One of the first, second, and third determination methods can be used. Alternatively, two of these methods can be combined, or all three can be used. The peak value of the roll eccentricity Δy peak Maximum value of roller eccentricity y max Minimum value of roller eccentricity y min These three are based on roller eccentricity y Tj y Bj The calculated representative values ​​can also be considered as having equal judgment functions.

[0147] exist Figure 6 In the routine, firstly, the roller eccentricity y is calculated. Tj y Bj The record (step S1301). Each time a piece of rolled material 1 is completed, a record is made. Figure 5 The roller eccentricity y identified by the roller eccentricity identification unit 12 in the middle T0 y T1 , ..., y Tn－1 and roller eccentricity y B0 y B1 , ...y Bn－1 The recorded data is saved to the recording medium inside the roller eccentricity recording unit 13 (step S1302).

[0148] Next, it is determined whether a preset time has elapsed, or whether a preset number of rolled pieces 1 have been rolled (step S1303). Alternatively, either the elapsed time or the predetermined number of rolled pieces can be used as the condition for step S1303. Alternatively, at least one of the conditions of the elapsed time and the predetermined number of rolled pieces can be met as the condition for step S1303. Alternatively, both the elapsed time and the predetermined number of rolled pieces can be met as the condition for step S1303.

[0149] The process in step S1303 is a determination process used to determine the passage of the "first rolling period". According to Embodiment 1, the appropriateness of the roll eccentricity in the second rolling period, which is later than the first rolling period, is evaluated using the identification value of the roll eccentricity obtained in the first rolling period.

[0150] Next, the data for each rolled material record is read in (step S1401). In this step, the data changes are read according to the determination process described later.

[0151] Next, based on the data read in step S1401 above, the following calculations (a1) to (a3) ​​are performed (step S1402).

[0152] (a1) Calculate the peak value Δy of the roller eccentricity peak The average value is calculated and used as the peak value Δy of the normal roll eccentricity. nor_peak .

[0153] (a2) Calculate the maximum value of roller eccentricity y max The average value is calculated and used as the maximum value of the normal roll eccentricity y. nor_max .

[0154] (a3) Calculate the minimum value of roller eccentricity y min The average value is calculated and used as the minimum value of normal roll eccentricity y. nor_min .

[0155] Furthermore, when there are multiple monitored rollers, it is preferable to perform the data processing of (a1) to (a3) ​​separately for each monitored roller. In Embodiment 1, in step S1402, based on the roller eccentricity y T0 y T1 , ..., y Tn－1 Calculate the representative value Δy of the roller eccentricity with respect to the upper support roller 4a. Tnor_peak y Tnor_max y Tnor_min On the other hand, in step S1402, based on the roller eccentricity y... B0 y B1 , ...y Bn－1 Calculate the representative value Δy of the roller eccentricity with respect to the lower support roller 4b. Bnor_peak y Bnor_max y Bnor_min .

[0156] Next, based on whether at least one of the following conditions (b1) to (b3) is met, it is determined whether the support rollers 4a and 4b, which are the monitored rollers, are abnormal (step S1403). Alternatively, as an example, the coefficient m can be set to 2.

[0157] (b1) Interpeak value of roller eccentricity Δy peak The peak value of the eccentricity of the normal roller Δy nor_peak The value is m times larger.

[0158] (b2) the maximum value y of the roll eccentricity max a value larger than the maximum value y of the normal roll eccentricity nor_max by m times.

[0159] (b3) the minimum value y of the roll eccentricity min a value smaller than the minimum value y of the normal roll eccentricity nor_min by m times.

[0160] In addition, in the case where there are a plurality of monitored roll, it is preferable to implement the roll state determination based on the above-described plurality of conditions (b1) to (b3) for each of the monitored rolls separately. In Embodiment 1, using the plurality of representative values Ay Tnor_peak , y Tnor_max , y Tnor_min calculated in step S1402, the roll state of the upper support roll 4a is determined. On the other hand, using the plurality of representative values Ay Bnor_peak , y Bnor_max , y Bnor_min calculated in step S1402, the roll state of the lower support roll 4b is determined.

[0161] In addition, as a modification, it is also possible to determine that the monitored roll is abnormal in the case where two or more of the plurality of conditions (b1) to (b3) are satisfied. Further, it is also possible to determine that the monitored roll is abnormal in the case where all of the plurality of conditions (b1) to (b3) are satisfied.

[0162] Figure 7 and Figure 8 is a flowchart for explaining the second roll state determination technique concerning the modification of Embodiment 1. In the second roll state determination technique of Figure 7 and Figure 8 , the roll eccentricity recording section 13 and the roll state determination section 14 perform abnormal determination of the roll state in a different method from the first roll state determination technique of Figure 6

[0163] The second roll state determination technique that is the basis of the routine of Figure 7 and Figure 8 is a roll state determination based on "statistical test method". In Embodiment 1, as an example of the second roll state determination technique, H(x) is calculated in accordance with the following formula (1).

[0164]

[0165] The parameter included on the right side of formula (1) is explained. Here, as an example, the statistical test method is implemented for the roll eccentricity peak interval Ay peak . The parameter x is substituted with the roll eccentricity peak interval Ay peak obtained in this rolling process. In the parameter x​N_AVE In the middle, substitute the previously obtained peak values ​​of normal roller eccentricity Δy nor_peak The average value is calculated using the average parameter σ. N In the middle, substitute the peak value Δy of the roller eccentricity. peak The standard deviation of x. Used to calculate these parameters x. N_AVE and σ N The data was obtained by performing multiple rolling processes of rolled material 1 under the condition that the monitored object rolls were the same.

[0166] H(x) expressed in equation (1) follows a chi-square (χ²) with 1 degree of freedom. 2 The distribution is known as Hotelling's theory. That is, the probability of occurrence is determined by substituting H(x) into a chi-square distribution with 1 degree of freedom.

[0167] Since the value of the chi-square distribution is usually a numerical table, it can be obtained from the numerical table or calculated by the following formula (2).

[0168]

[0169] Here, k = 1, y = H(x). The gamma function G is G(1 / 2) = √π.

[0170] Additionally, when given a data column X = {x1, x2, ..., x...} n When X is a data series, the standard deviation σ can be calculated as follows. Where X... AVE It is the average value of data column X.

[0171]

[0172] In the above example, when H(x) = 5.7, the chi-square distribution with 1 degree of freedom has a value of 0.0097. The probability of obtaining x as H(x) = 5.7 is 0.97%, or less than 1%. A larger H(x) corresponds to a situation where x is significantly different from its past average. In such a case, since an anomalous state with a very low probability of occurrence has occurred, the roller state can be considered anomaly.

[0173] Typically, a 5% significance level or a 1% significance level is used. Thus, an abnormality is determined with a 5% hazard rate or a 1% hazard rate.

[0174] Next, the explanation Figure 7 and Figure 8 The specific content of control. Figure 7 and Figure 8 The routine is executed by the roller eccentricity recording unit 13 and the roller state determination unit 14.

[0175] In addition, Figure 7 Steps S1414 and Figure 8 Steps S1415, S1416 realize the second roll condition determination technique based on the above-described formula (1) or the like. However, on the other hand, in Figure 7 and Figure 8 , the third roll condition determination technique (steps S1412, S1413) is also included. The third roll condition determination technique determines whether the roll condition is normal based on a comparison determination using a fixed value set in accordance with data obtained in the past.

[0176] In the routine of Figure 8 , first, the roll eccentricity amount identified by the roll eccentricity amount identification section 12 is recorded with the roll eccentricity amount recording section 13 (step S1311). In this step, the roll eccentricity amount recording section 13 records the roll eccentricity amounts y T0 , y T1 ,... y Tn－1 , and the roll eccentricity amounts y B0 , y B1 ,... y Bn－1 for each of the rolling of the one piece of rolled material 1, respectively. The recorded data is saved to a recording medium inside the roll eccentricity amount recording section 13 (step S1312).

[0177] Next, it is determined whether a fixed threshold value set in advance is set as a determination reference (step S1411). Whether the fixed threshold value is used in step S1411 is decided in accordance with the state of a determination method flag prepared in advance. If the determination method flag is 1, the determination result of step S1411 is affirmative (Yes). If the determination method flag is 0, the determination result of step S1411 is negative (No). The determination method flag is set in advance and can be changed afterward.

[0178] In the case where the determination result of step S1411 is affirmative (Yes), the processing proceeds to steps S1412 and Figure 8 S1413, and the above-described third roll condition determination technique is implemented.

[0179] First, in step S1412, three kinds of threshold values shown in (c1) to (c3) below are read out from the recorded data of the roll eccentricity amount recording section 13. These threshold values are fixed values set in advance by using rolling data obtained in the past or simulation or the like. The three kinds of threshold values can be set for the upper-side monitoring target roll and the lower-side monitoring target roll, respectively, or can be set to common values for both of the upper and lower monitoring target rolls.

[0180] (c1) A first threshold value Y peak for determination set to the roll eccentricity amount peak interval value Δy peak_th

[0181] (c2) is set to the maximum value of roller eccentricity y. max The second threshold Y used for the determination max_th

[0182] (c3) is set to the minimum value of roller eccentricity y. min The third threshold Y used for the determination min_th

[0183] Next, in Figure 8 In step S1413, based on whether at least one of the following conditions (d1) to (d3) is met, it is determined whether the support rollers 4a and 4b of the monitored object are abnormal.

[0184] (d1) Peak value of roller eccentricity Δy peak Compared to the first threshold Y peak_th big.

[0185] (d2) Maximum value of roller eccentricity y max Compared to the second threshold Y max_th big.

[0186] (d3) Minimum value of roller eccentricity y min Compared to the third threshold Y min_th Small.

[0187] In addition, when there are multiple monitored rollers, it is preferable to perform roller state determination based on the above-mentioned multiple conditions (d1) to (d3) separately for each monitored roller.

[0188] Furthermore, as a variation, the monitored roller can be determined to be abnormal if two of the aforementioned conditions (d1) to (d3) are met. Additionally, the monitored roller can be determined to be abnormal if all of the aforementioned conditions (d1) to (d3) are met.

[0189] If the determination result in step S1411 is negative (no), then proceed to step S1414 and... Figure 6 Steps S1415 and S1416 proceed. Thus, the aforementioned second roller state determination technique is implemented.

[0190] First, in step S1414, the calculations of various parameters described in (e1) to (e3) below are performed.

[0191] (e1) Regarding the peak value Δy of roller eccentricity peak average x N_AVE and standard deviation σ N

[0192] (e2) Regarding the maximum value of roller eccentricity y max average x N_AVE and standard deviation σN

[0193] (e3) Regarding the minimum value of roller eccentricity y min average x N_AVE and standard deviation σ N

[0194] Next, in Figure 7 In step S1415, it is determined whether the support rollers 4a and 4b of the monitored objects are abnormal based on whether at least one of the following conditions (f1) to (f3) is met. Additionally, the threshold H1 can be preset. For example, to perform a test at the 1% significance level, H1 can be set to 5.7.

[0195] (f1)H(x=Δy peak It is larger than the threshold H1.

[0196] (f2)H(x=y max It is larger than the threshold H1.

[0197] (f3)H(x=y min It is larger than the threshold H1.

[0198] However, under the conditions (f1) to (f3) above, H(x=Δy) peak ) is the value of the peak value Δy of the roller eccentricity. peak average x N_AVE and standard deviation σ N Substitute the value into equation (1). H(x=y max ) is about the maximum value of the roller eccentricity y max average x N_AVE and standard deviation σ N Substitute the value into equation (1). H(x=y min ) is the minimum value of roller eccentricity y min average x N_AVE and standard deviation σ N The value is substituted into equation (1).

[0199] Furthermore, when there are multiple monitored rollers, the calculation of parameters (e1) to (e3) and the roller state determination based on multiple conditions (f1) to (f3) can be performed separately for each monitored roller. In Embodiment 1, these processes are performed on the upper support roller 4a and the lower support roller 4b respectively.

[0200] That is, in implementation 1, a method based on roller eccentricity y is used. T0 y T1 , ..., y Tn－1The plurality of parameters calculated in step S1414 are used to determine the roll state of the upper backup roll 4a in step S1415. On the other hand, using the roll eccentricity amount y B0 , y B1 ,... y Bn－1 The plurality of parameters calculated in step S1414 are used to determine the roll state of the lower backup roll 4b in step S1415.

[0201] In addition, as a modification, it is also possible to determine that the monitored roll is abnormal in a case where two or more of the plurality of conditions (fl) to (f3) are satisfied. Further, it is also possible to determine that the monitored roll is abnormal in a case where all of the plurality of conditions (fl) to (f3) are satisfied.

[0202] In step S1416, depending on which of normal and abnormal is determined as a result of the roll state determination, the calculation data of step S1414 is saved in the recording medium of the roll eccentricity amount recording section 13 with the identifier of normal / abnormal attached. In a case where there are a plurality of monitored rolls, it is also possible to perform the data saving process with the identifier of step S1416 for each of the monitored rolls. In Embodiment 1, the plurality of parameters (el) to (e3) calculated in step S1414 for the upper backup roll 4a and the lower backup roll 4b are saved in a state where the identifier of normal and abnormal is attached to one of the normal and abnormal.

[0203] In addition, in a case where the Hotelling theory is implemented by the routine of the above-described Figure 8 , the number of data of the normal state is about 5 to 10, which is slightly small as the number of data for determination. On the other hand, in a case where the routine of Figure 7 and Figure 8 , by accumulating the past data in a larger number by the roll eccentricity amount recording section 13, it is possible to sufficiently secure the data to be compared. Therefore, in a case where the routine of Figure 9 and Figure 9 , there is an advantage that it is easy to apply abnormality determination based on the Hotelling theory.

[0204] Figure 9 is a graph illustrating the progress of the actual roll eccentricity amount with respect to Embodiment 1. In Embodiment 1, as an example, the roll state determination section 14 has a function of displaying the roll eccentricity amount peak interval value Ay peak . In Figure 9 , as an example, a plurality of roll eccentricity amount peak interval values Ay peak from the rolled material 1 for which rolling was completed most recently to the past are displayed. The roll eccentricity amount peak interval value Ay peak is the difference between the maximum value and the minimum value of the roll eccentricity amount that is output from the roll eccentricity amount discrimination section 12.

[0205] Figure 9The horizontal axis represents the number of rolled sections. Figure 9 In the first and second rollers, the roller condition is normal. Figure 9 It is speculated that the roller breakage may have started near the third or fourth roller. Figure 3 In the example, on the 10th roll, the operator noticed an anomaly and stopped mill 50. After pulling out and inspecting the roll, a section of damage was found on the upper support roll on the drive side (DS). Figure 4 In the middle, the increase in the eccentricity of the upper support roller 4a is consistent with the phenomenon of partial damage to the roller.

[0206] (First variation of Implementation Method 1)

[0207] The first variation of the embodiment is described. Regarding the roller condition monitoring device 20 of Embodiment 1... Figure 5 , Figure 5 and Figure 10 The support rollers 4a and 4b are used as the monitoring rollers, but this is not a limitation. The work rollers 3a and 3b can also be used as the monitoring rollers. The monitoring rollers can be arbitrarily selected from multiple rollers included in the upper and lower roller groups.

[0208] Alternatively, both support rollers 4a and 4b and work rollers 3a and 3b can be used as monitoring rollers. In this case, two... Figure 10 The roller condition monitoring device 20 is shown. This is because, since the rotational speeds of the support rollers 4a and 4b and the work rollers 3a and 3b are different, it is preferable to perform roller condition determination by different roller condition monitoring devices 20.

[0209] (Second variation of Implementation Method 1)

[0210] Figure 5 This is a diagram illustrating the structure of the roller condition monitoring device 20 in a modified example of Embodiment 1. Additionally, in Figure 3 For convenience, in the middle, Figure 4 Blocks 10, 11, 12, 111, 112, 121, and 122 are simplified and recorded.

[0211] In the roller condition monitoring device 20 of embodiment 1, Figure 5 , Figure 1 and Figure 10 In this case, support rolls 4a and 4b are used as the monitored rolls, and one rolling load value is used for each stand. However, in the rolling mill 50, the rolling load at two points at the ends of the rolls in the width direction can also be measured for stands #1 to #7 respectively.

[0212] The two ends in the width direction of the roll are a drive side (DS) and an operator side (OS). These are also illustrated in Figure 5 the second modification. In the second modification, as shown in Figure 10 the drive side roll load detection mechanism 6ds and the operator side roll load detection mechanism 6os are provided at the two ends in the width direction of the roll.

[0213] In the second modification, the two roll state monitoring devices 20 are respectively assigned to the DS roll load and the OS roll load. The roll state monitoring device 20 for the DS roll load mainly monitors the roll state on the drive side based on the output signal of the drive side roll load detection mechanism 6ds. The roll state monitoring device 20 for the OS roll load mainly monitors the roll state on the operator side based on the output signal of the operator side roll load detection mechanism 6os.

[0214] In addition, an abnormality occurring in the central portion in the width direction of the roll is detected by both the drive side and the operator side. Therefore, a first case in which an abnormality is detected only on the drive side, a second case in which an abnormality is detected only on the operator side, and a third case in which an abnormality is detected on both the drive side and the operator side can occur. According to the second modification, it is also possible to generally determine which of the drive side, the operator side, and the central portion in the width direction of the roll at which an abnormality has occurred by distinguishing the first case, the second case, and the third case. In addition, compared with Figure 10 the case in Figure 10 which the processing amount is about twice, it is preferable to confirm the calculation ability in advance.

[0215] (Third Modification of Embodiment 1)

[0216] The roll state monitoring device 20 described above regarding the second modification takes the backup rolls 4a, 4b as the monitoring target rolls, but in the third modification, the work rolls 3a, 3b are taken as the monitoring target rolls. In addition, in the case where the backup rolls 4a, 4b and the work rolls 3a, 3b are respectively taken as the monitoring target rolls, the roll state monitoring device 20 shown in Figure 11 is sufficient as long as four roll state monitoring devices 20 are provided in total.

[0217] (Fourth Modification of Embodiment 1)

[0218] The fourth modification is a modification including the roll state monitoring device 20 described above regarding the second modification and the third modification. That is, the backup rolls 4a, 4b and the work rolls 3a, 3b are taken as the objects, and the roll state monitoring function is provided in the DS and the OS respectively. Since Figure 11The upper and lower pairs shown are also required for the work rolls, so that four roll condition monitoring devices 20 are provided in total. Therefore, the processing amount of the computer becomes about four times compared with the structure of Embodiment 1. In this way, the number of roll condition monitoring devices 20 can be increased in correspondence with the increase in the number of monitoring target rolls.

[0219] (Fifth Modification of Embodiment 1)

[0220] Figure 5 is a view for specifically explaining the method of extracting the roll load variation and the roll eccentricity amount discrimination, and the structure of the device that implements it, concerning the modification of Embodiment 1. In Figure 12 the modification of Embodiment 1, the transformation blocks 121a, 122a are omitted from Figure 13 the structure of Embodiment 1. In this case, the transformation of the roll gap equivalent values ΔS Tj , ΔS Bj to the roll load variation values ΔP Tj , ΔP Bj is not performed, and the roll load variation values ΔP Tj , ΔP Bj are delivered to the limiters 121b, 122b. In the adders 121d, 122d, the roll load variation values ΔP corresponding to a plurality of roll rotational positions are also accumulated.

[0221] As described above, there is a preferable feature that the deviation of the calculation result based on the difference in the characteristics (for example, the hardness of the rolled material) of the rolled material 1 that the rolling mill 50 is targeted for can be suppressed by the transformation of the roll gap equivalent values ΔS Tj , ΔS Bj to the roll load variation values ΔP Tj , ΔP Bj by the transformation blocks 121a, 122b. However, such a preferable feature is not necessarily required, and the transformation blocks 121a, 122b can be omitted. Thereby, the computational load in the roll eccentricity amount discrimination section 12 can be reduced.

[0222] Embodiment 2.

[0223] Figure 13 is a view that explains an example of the rolling mill 250 to which the roll condition monitoring device 220 concerning Embodiment 2 is applied. Figure 14 is a view for explaining the structure of the roll condition monitoring device 220, the upper side roll pair, and the lower side roll pair concerning Embodiment 2.

[0224] Embodiment 1 differs from Embodiment 2 in that the roll condition monitoring device 20 is replaced with the roll condition monitoring device 220. As Figures 14-20 shown, the roll condition monitoring device 220 is provided with a roll load signal processing section 210, a load data processing section 211, and a roll condition determination section 212. Hereinafter, the structure common to Embodiment 1 is given the same reference numerals and the explanation is omitted, and the differences between Embodiment 1 and Embodiment 2 are explained as the center.

[0225] Figure 13 is a diagram for explaining the roll condition judging technique relating to Embodiment 2. In Embodiment 2, as well as in Embodiment 1, the rolling load detection mechanism 6 detects the rolling load received by the rolling mill 250 from the rolled material 1. The load detection signal detected by the rolling load detection mechanism 6 is also referred to as a raw signal.

[0226] In Embodiment 2, based on the raw signal detected by the rolling load detection mechanism 6, the signal processing and judging processing described later are implemented. Figure 14 The monitoring target roll of Embodiment 2 is a roll that receives the rolling load of the load detection signal to which these signal processing and judging processing are applied.

[0227] In Embodiment 2 as well as in Embodiment 1, the monitoring target roll can be arbitrarily selected. In Embodiment 1, the rolling load is distributed to the upper and lower rolls by the rolling load up-down distribution section 10, but at least one of the upper and lower rolls can be selected as the monitoring target roll. Figure 14 In Embodiment 1, the rolling load up-down distribution section 10 is omitted, but in a case where the rolling load is distributed to the upper and lower rolls by the rolling load up-down distribution section 10, at least one of the upper and lower rolls can be selected as the monitoring target roll. The rolling load detection mechanism 6 can be configured to detect the rolling load in the DS and the OS, respectively, as in the fourth modification example of the above-described Embodiment 1.

[0228] In the upper part of Figure 14 , the low-frequency component and the high-frequency component included in the raw signal are schematically illustrated. Here, it is assumed that the raw signal is a signal indicating the absolute value of the rolling load. The detected raw signal generally includes a low-frequency component (dotted line in the upper part of Figure 14 ) indicating a gentle vibration and a high-frequency component (thin solid line in the upper part of Figure 14 ) like noise.

[0229] The rolling load signal processing section 210 applies an HPF (high-pass filter) to the raw signal. By thus removing the low-frequency component of the rolling load signal with the high-pass filter or the like and extracting the high-frequency component, the high-frequency component can be taken out as the rolling load high-frequency signal S HF . In the lower part of Figure 14 , an example of the rolling load high-frequency signal S HF extracted by the HPF is schematically illustrated. The graph in the lower part of Figure 14 is only a schematic diagram, and the waveform of the actual rolling load high-frequency signal S HF may be different from this.

[0230] The load data processing section 211 calculates the standard deviation σ of the rolling load high-frequency signal S HF . The load data processing section 211 calculates the difference d of the probability density distribution of ±kσ from the normal distribution. k is a value of, for example, 2 to 5.

[0231] In the load data processing unit 211, a high-frequency signal S that fully includes the rolling load is set. HF The vertical axis range D of the amplitude. For example... Figure 15 As shown, the vertical axis range D is divided into n pre-defined intervals D. n The load data processing unit 211 processes the high-frequency rolling load signal S... HF As a collection of data, each interval D of the vertical axis range D is processed. n The number of data contained in it is counted.

[0232] The load data processing unit 211 calculates the probability of each interval by dividing the number of data belonging to each interval by the total number of data. This is done by processing multiple intervals D1, D2, D3, ... D... n The entire application of such calculations yields... Figure 15 The probability density distribution is shown on the right side of the lower paragraph.

[0233] To fully incorporate the high-frequency signal S of the rolling load HF The amplitude of the amplitude can be set, and the vertical axis range D can also be set to approximately 4σ, which is four times the standard deviation σ. This allows almost all the data to be included within the vertical axis range. Specifically, the data ranges covered by the vertical axis range D are 2σ = 95.4%, 3σ = 99.7%, and 4σ = 99.994%, etc., depending on σ.

[0234] Figure 15 It is a graph illustrating the probability density distribution of implementation method 2. Figure 9 This is an example of a real probability density distribution. In Figure 15 In the diagram, the probability density distribution of the actual data is plotted using a solid line, and is used in conjunction with... Figure 15 The same data is used in the curve graph. Figure 9 The solid line data is based on the rolling load data on the drive side of the damaged rolling stand. Figure 16 The solid line data is obtained through analysis of... Figure 16 The high-frequency signal S of the rolling load is obtained by applying a high-pass filter to the data of the first rolling process. HF The probability density distribution.

[0235] Figure 15 It is a graph illustrating the probability density distribution of implementation method 2. Figure 9 solid line data and Figure 15 The difference is illustrated in the diagram. Figure 16 The high-frequency rolling load signal S extracted from the rolling load signal of the 10th rolling process. HF The probability density distribution. Figure 5 and Figure 15 Take the horizontal axis Figure 16±4σ of the signal of the 10th in the center as a common scale range.

[0236] In Figure 15 and Figure 15 , a normal distribution for a comparison target is illustrated by a broken line. In Figure 16 , a broken line graph showing a normal distribution coincides with a solid line graph showing actual data. In the case where the roll is normal, as in Figure 16 , the probability density distribution calculated from the rolling load high frequency signal S HF is in agreement with the normal distribution. In contrast to this, if an abnormality occurs in the roll condition, as in Figure 19 , the probability density distribution is clearly different from the normal distribution. By such a difference, it is possible to determine whether or not there is an abnormality in the roll condition.

[0237] The roll condition determination section 212 can also present the graph of Figure 19 directly to an operator or the like via a device such as a display. Thereby, it is also possible to clearly recognize an abnormality by visual observation by a human. However, it is also possible to express the difference in the distribution shape by a numerical value, and it is also possible for the roll condition determination section 212 to automatically output an abnormality determination signal based on the numerical value. Thereby, it is also possible to objectively and automatically warn that an abnormality has occurred.

[0238] In the calculation of the difference d of the probability density distribution from the normal distribution, as an example, each of the following numerical indexes shown in the following equations (4) to (6) can be used. Equation (4) is an equation for calculating the value D KL of the KL distance (Kullback-Leivler Divergence). Equation (5) is an equation for calculating the value D SQ of the error sum of squares. Equation (6) is an equation for calculating the value D ABS of the error sum of absolute values.

[0239] The roll condition determination section 212 can also calculate the difference d of the probability density distribution from the normal distribution based on at least one of the three examples shown in equations (4) to (6). That is, the difference d can be one of the value D KL , the value D SQ , and the value D ABS . It is also possible to determine that the roll condition is abnormal in the case where this difference d is equal to or greater than a predetermined determination value set in advance.

[0240]

[0241]

[0242]

[0243] In the above equations, P A(x) is the actual probability density of the data x. In Implementation 2, the data x is the high-frequency signal S of the rolling load. HF The value of P. N (x) follows a normal distribution. Generally, high-frequency signals can be roughly considered as noise. Noise is white noise and can be considered to follow a normal distribution. However, when the rolling load signal contains noise due to some anomaly, the high-frequency rolling load signal S... HF The probability density distribution is clearly different from the normal distribution. Therefore, based on the comparison between the probability density distribution and the normal distribution, anomalies in the roller condition can be determined.

[0244] Figure 9 This is a diagram illustrating the KL distance in Implementation Method 2. Figure 19 Indicates according to Figure 16 The results were obtained from data acquired during the 10th rolling process. The high-frequency rolling load signals S on both the drive and operator sides of multiple rolling stands were analyzed. HF After determining the probability density distribution, the KL distance D, representing an example of the difference d between the probability density distribution and the normal distribution, was plotted. KL .

[0245] If the value of the KL distance is D KL If the value D is large, then the difference between the two distributions being compared is also large. Therefore, for example, it could also be if the value D... KL It is a pre-set judgment value D KL_th If the above conditions are met, the roller condition is determined to be abnormal. Similarly, it could also be determined if the value D... SQ or value D ABS It is a pre-set determination value D SQ_th Or D ABS_th If the above conditions are met, the roller condition is determined to be abnormal.

[0246] The above D KL_th D SQ_th and D ABS_th Also referred to as the specified judgment value d th The specified judgment value d th It is the comparison judgment value used in the evaluation of difference d. The judgment value d is specified. th It can be a pre-set fixed value or a variable value that is updated sequentially. For example, specify the judgment value d. th The value of the difference d, obtained from at least one past rolling operation where the roll condition was normal, can be set as a fixed value or updated sequentially. For example, suppose n differences d are obtained from n past rolling operations (p1, p2, p3…pn) where the roll condition was originally normal. p1 d p2 d p3 …d pnFor example, it can also be based on d. p1 ~d pn average value d p_ave To set the specified judgment value d th For example, specifying the judgment value d. th It can also be the average value d p_ave Multiplied by a pre-defined coefficient k d The value after (k) d ×d p_ave ).

[0247] exist Figure 17 In the middle, the first result of the project number is based on the high-frequency signal S of the rolling load on the drive side of the first stand #1. HF The second result for project number is based on the high-frequency signal S of the rolling load on the operator side of the first stand #1. HF The third result for project number is based on the high-frequency signal S of the rolling load on the drive side of the second stand #2. HF The item numbers are assigned according to this rule up to the tenth item.

[0248] The result of the tenth item is based on the high-frequency signal S of the rolling load on the drive side of the broken upper support roll 4a. HF The result of the tenth one is the same as... Figure 17 The anomaly occurred as shown in the curve. The tenth result is due to the KL distance value D compared to other item numbers. KL The value is significantly large, indicating that it is a probability density distribution that can be distinguished from the normal distribution.

[0249] (First variation of implementation method 2)

[0250] Figure 17 This is a graph illustrating the probability density distribution of the first related variation example of Embodiment 2. If the high-frequency signal S of the rolling load in Embodiment 2 is used... HF If the maximum and minimum values ​​are plotted as two probability density distributions on a curve, then this is taken as an example and becomes... Figure 18 That way.

[0251] exist Figure 18 The diagram illustrates the probability density distributions of the maxima, minima, and Rayleigh distributions. In signals where the roller condition is normal, the probability density distributions of the maxima and minima closely approximate Rayleigh distributions. Conversely, if the roller condition includes anomalies, the probability density distributions of the maxima and minima deviate from the Rayleigh distributions.

[0252] Figure 15 This is a graph illustrating the minimum and maximum values ​​of the first variation of embodiment 2. (As shown in...) Figure 16As is apparent from the graphically represented middle, each time the high frequency signal waveform switches from decrease to increase, a minimum value and a maximum value are obtained, so the high frequency signal S HF contains a plurality of minimum values and a plurality of maximum values.

[0253] (Second Modification of Embodiment 2)

[0254] As the second modification of Embodiment 2, roll condition determination based on comparison of the results of each stand inspection can also be performed. The "result of each stand inspection" can also be the difference d obtained for each of the stands #1 to #7. Specifically, in this second modification, the difference d can also be calculated for each of the plurality of stands #1 to #7 in the finishing mill 57, and these plurality of differences d can also be compared with each other. The difference d in this second modification can be a difference with respect to the normal distribution explained in Figure 17 and Figure 18 . The difference d in this second modification can also be a difference with respect to the Rayleigh distribution explained in Figure 13 and Figures 14-19 .

[0255] That is, as shown in Figure 20 , the plurality of stands #1 to #7 each include the rolling load detecting mechanism 6, and thus the rolling load signal processing section 210 can individually extract the rolling load high frequency signal S HF for each of the plurality of stands #1 to #7. In the second modification, the load data processing section 211 can also individually calculate the difference di to d7 with respect to each of the stands #1 to #7 based on the rolling load high frequency signal S HF . This difference d is the result of each stand inspection performed on the rolling load signal output from the rolling load detecting mechanism 6 of each stand by the statistical test method described in Figure 20 . The difference d is the result of each stand inspection performed on the rolling load signal output from the rolling load detecting mechanism 6 of each stand by the statistical test method described in

[0256] In the second modification, in a case where i is an arbitrary integer, the roll condition determination section 212 can also compare the difference d i of the i-th stand with the difference d j of the j-th stand (where j ≠ i). Here, an arbitrary value different from i is substituted into j, and the j-th stand inclusively represents all stands other than the i-th stand. The roll condition determination section 212 can also be, for example, if the "representative value of a plurality of d j differs from d i by a prescribed multiple or more, the i-th stand is determined to be abnormal. The prescribed multiple can also be set in advance to a value such as 3 or the like. The representative value of a plurality of d j can also be the average value of a plurality of d j . Since j = 2 to 7 when, for example, i = 1, a plurality of d jThe representative value can also be an average value of d2, d3,... d7.

[0257] ​ Fig. 1 is a diagram showing an example of a hardware structure of the roll condition monitoring device 20, 220 relating to Embodiments 1, 2. The various control actions, calculation processes, and determination processes explained in Embodiments 1, 2 can also be executed by the hardware structure explained below.

[0258] The functions of the roll condition monitoring device 20, 220 are realized by a processing circuit. The processing circuit can also be a dedicated hardware 350. Alternatively, the processing circuit can also have a processor 351 and a memory 352. The processing circuit can also be partly formed as the dedicated hardware 350, and have the processor 351 and the memory 352. ​ Fig. 1 is a diagram showing an example of a hardware structure of the roll condition monitoring device 20, 220 relating to Embodiments 1, 2. The various control actions, calculation processes, and determination processes explained in Embodiments 1, 2 can also be executed by the hardware structure explained below.

[0259] In the case where at least a part of the processing circuit is at least one dedicated hardware 350, the processing circuit corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a structure combining them.

[0260] In the case where the processing circuit has at least one processor 351 and at least one memory 352, each function of the roll condition monitoring device 20, 220 is realized by software, firmware, or a combination of software and firmware. The software and firmware are described as programs, and are stored in the memory 352. The processor 351 realizes the functions of each part by reading out and executing the programs stored in the memory 352. The processor 351 is also called a CPU (Central Processing Unit), a central processing device, a processing device, an arithmetic device, a microprocessor, a microcomputer, a DSP. The memory 352 corresponds to, for example, a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, an EEPROM, and the like.

[0261] In this way, the processing circuit realizes each function of the roll condition monitoring device 20, 220 by hardware, software, firmware, or a combination of these.

[0262] Explanation of Reference Signs

[0263] 1 rolled material; 2 rolling mill housing; 3a work roll (upper work roll); 3b work roll (lower work roll); 4a backup roll (upper backup roll); 4b backup roll (lower backup roll); 4c reference position; 5 screwdown mechanism; 6 rolling load detecting mechanism; 6ds drive side rolling load detecting mechanism; 6os operator side rolling load detecting mechanism; 7 roll rotational speed detector; 8 roll reference position detector; 9 roll gap detector; 10 rolling load up-down distribution section; 11 rolling load variation extracting section; 12 roll eccentricity amount discriminating section; 13 roll eccentricity amount recording section; 14 roll condition judging section; 14a reference position; 15 position scale; 15a reference position; 20, 220 roll condition monitoring device; 50, 250 rolling mill; 51 slab; 52 heating furnace; 53 rough rolling mill; 54 bar heater; 55 bar; 56 entry side thermometer; 57 finish rolling mill; 58 plate thickness and width gauge; 59 exit side thermometer; 60 thermometer; 61 coiler; 62 product coil; 63 output roller table; 111 upper side load variation extracting section; 112 lower side load variation extracting section; 111a, 112a rolling load recording section; 111b, 112b average value operation mechanism; 111c; 112c deviation operation mechanism; 121 upper side addition mechanism; 122 lower side addition mechanism; 121a, 122a conversion block; 121b, 122b limiter; 121c, 122c switch; 121d, 122d adder; 121e, 122e rotational speed correction block; 210 rolling load signal processing section; 211 load data processing section; 212 roll condition judging section; 350 dedicated hardware; 351 processor; 352 memory; OS operator side; DS drive side; RD rolling direction; n roll division number; P rolling load; y Tj T0 T1 Tn－1 Bj B0 B1 Bn－1 Roll eccentricity amount; ΔP rolling load variation value; ΔS, ΔS Tj Bj Roll gap equivalent value; Δy peak Roll eccentricity amount peak-to-peak value; Δy nor_peak Normal roll eccentricity amount peak-to-peak value; S HF Rolling load high frequency signal; D longitudinal axis range; D n Interval.​​​​​​​​

Claims

1. A roller condition monitoring device, characterized in that, have: A rolling load detection mechanism is configured to detect the rolling load signal of a monitoring target roll selected from the upper roll group and the lower roll group when rolling a rolled material between an upper roll group including at least one roll and a lower roll group including at least one roll. The signal extraction mechanism extracts a high-frequency rolling load signal with a frequency above a predetermined frequency from the aforementioned rolling load signal. as well as The roll condition determination mechanism determines the condition of the monitored roll based on the test results of a statistical test method for multiple rolling load values ​​contained in the high-frequency rolling load signal.

2. The roller condition monitoring device as described in claim 1, characterized in that, The aforementioned roll condition determination mechanism is constructed by calculating the probability density distribution of the rolling load values ​​based on the aforementioned multiple rolling load values, and determining the condition of the monitored roll by comparing the probability density distribution of the rolling load values ​​with a pre-set reference distribution.

3. The roller condition monitoring device as described in claim 2, characterized in that, The aforementioned roller state determination mechanism includes a normally distributed roller state determination mechanism; The aforementioned normal distribution roll state determination mechanism is constructed such that it calculates the probability density distribution of the aforementioned multiple rolling load values ​​as the probability density distribution of the aforementioned rolling load values, and uses a normal distribution as the aforementioned reference distribution.

4. The roller condition monitoring device as described in claim 2, characterized in that, The aforementioned roller condition determination mechanism includes a Raleigh distribution roller condition determination mechanism; The aforementioned Rayleigh distribution roller state determination mechanism is constructed as follows: As the probability density distribution of the aforementioned rolling load values, the maximum and minimum probability density distributions of the probability density distributions of the multiple rolling load maxima and minimum values ​​contained in the aforementioned high-frequency rolling load signal are calculated. The Rayleigh distribution is used as the baseline distribution mentioned above.

5. The roller condition monitoring device as described in claim 1, characterized in that, The aforementioned monitored roller has a first side end and a second side end opposite to the first side end; The rolling load detection mechanism described above is configured to detect a first-side rolling load signal from a first rolling load sensor located at the first end of the first side, and to detect a second-side rolling load signal from a second rolling load sensor located at the second end of the second side. The aforementioned signal extraction mechanism extracts high-frequency rolling load signals with frequencies above the specified frequency from both the first-side rolling load signal and the second-side rolling load signal. The aforementioned roll state determination mechanism is configured to determine the state of the first side end and the second side end of the monitored roll based on the test results of the statistical test method for the high-frequency rolling load signal extracted by the aforementioned signal extraction mechanism.

6. The roller condition monitoring device as described in claim 1, characterized in that, The aforementioned upper roll group includes multiple upper roll groups that constitute multiple rolling frames; The aforementioned lower roll group includes multiple lower roll groups that together with the aforementioned multiple upper roll groups constitute the aforementioned multiple rolling frames; The aforementioned rolling load detection mechanism obtains multiple rolling load signals from the rolling load sensors of each of the aforementioned multiple rolling stands; The aforementioned signal extraction mechanism extracts multiple high-frequency rolling load signals with frequencies above the specified frequency from the aforementioned multiple rolling load signals; The aforementioned roll condition determination mechanism is constructed to obtain the inspection results of each of the multiple rolling stands corresponding to the multiple rolling stands, and use them as the inspection results of the statistical inspection method for the multiple rolling load values ​​contained in the multiple high-frequency rolling load signals, and determine the condition of the monitored roll based on the inspection results of each of the multiple rolling stands.

Images