JPH0527486B2 - - Google Patents

Description

[Detailed description of the invention]

(a) Industrial Application Field The present invention relates to a method for comprehensively monitoring and diagnosing the deterioration status of rolling mill equipment. (b) Prior Art A rolling mill rotates while pressing a rolled material with upper and lower rolls or left and right rolls, etc. to obtain a product of a desired size. Maintenance of the equipment requires a lot of experience, labor, and expense (materials such as consumables). In particular, universal joints such as slipper metal and gear couplings are used in the work roll rotation drive system to transmit the enormous power (the largest in the rolling mill) and to constantly change the position of the rolls. ing. This universal joint always has a constant clearance (gap). However, this clearance wears out due to the huge transmission force mentioned above.
If it becomes large, the quality of the rolled material will be affected. Therefore, it is necessary to periodically replace the metals used in the universal joints for driving work rolls. However, conventionally, there is no suitable method for measuring the amount of wear (measuring the relative movement by applying force when the machine is stopped), and if the replacement time is early, the material cost will increase, and conversely, if the replacement time is late, the product Losses were incurred due to quality deterioration, and determining the optimal replacement timing was a maintenance issue. The following are typical examples of conventional methods for determining the optimal replacement time. The equipment deterioration status evaluation method (Japanese Unexamined Patent Publication No. 59-77514) particularly relates to a method for detecting abnormalities in a rotational force transmission system. There is also a method (Japanese Unexamined Patent Publication No. 77367/1983) that uses this method to evaluate the deterioration state of the motor system. Furthermore, a roll drive monitoring device (Japanese Patent Application Laid-Open No. 61-88776) uses a mathematical model of a rolling mill roll drive system and utilizes it to control a transmission system. When controlling a rolling mill by modeling the roll drive system using a mathematical model, the wear of the joints will greatly affect the control system, making the above-mentioned gap management even more important. After all, conventional equipment monitoring/diagnosis methods detect the state of specific, limited parts of the equipment using various sensors, perform some kind of processing on the detected state, and perform diagnosis. This can be said to be equipment monitoring within a limited range. For example, when torque or housing vibration is used as a parameter, these quantities change depending on operating conditions, so it is not possible to clearly determine whether the change is due to equipment deterioration or the influence of operation based on changes in this quantity alone. (c) Problems to be solved by the invention The problems to be solved by the invention are to perform statistical processing on various information generated by the rolling mill,
The object of the present invention is to provide a method for minimizing information loss, deriving new deterioration information parameters, and thereby comprehensively monitoring and diagnosing the deterioration state of a rolling mill. (d) Means for Solving Problems The rolling mill monitoring method of the present invention includes measuring the rolling load, drive torque, and vibration of the drive system of the rolling mill, and statistically processing the measured values by principal component analysis. The above problem is solved by determining the wear of the universal joint for driving the work roll. It is preferable to measure the vibration of a roll chock in the rolling mill as the vibration of the drive system. It is preferable to use the ratio between the maximum value and the average value of each measurement value as data for principal component analysis. (e) Example The schematic configuration of an apparatus for carrying out the method of the present invention is described in the first example.
As shown in the figure. The information obtained from the rolling mill 1 is usually
Main motor 2 current, main reducer 3 vibration, pinion stand 4 vibration and temperature, mill housing 5 vibration, back-up roll choke 6
vibration, spindle thrust, spindle transmission torque, rolling load, etc. The operational data includes the material of the rolled material, rolling dimensions, rolling temperature, rolling reduction amount, etc. In a conventional rolling mill 1, a work roll 7 is driven by the power of a main motor 2 through a universal joint 8. The analysis device 9 according to the present invention has a function of arbitrarily selecting analysis data for the imported data, and is designed to perform analysis using an ideal combination of variables. In the analysis device 9 of the present invention, the multiplexer 10 operates using the rolling load as a trigger signal for various data obtained from the rolling mill 1, and collects data during rolling (from biting to tailing) of set parameters. Perform import. Figure 2 shows an overview of data import. Each symbol in FIG. 2 represents the following items. A: Data during biting B: Data during steady rolling C: Data during tail-off D: Valid data E: Trigger delay time (can be set arbitrarily) F: Trigger delay time (can be set arbitrarily) The data sampled at the above timing is , and stored in the storage device 12 through the A/D converter 11. The stored data is stored in the arithmetic unit 13
are sequentially calculated and stored in the data storage device 14 as various analysis variables. Calculation contents include bite value, end-of-line value, effective value during steady rolling, average value, maximum value, minimum value, in addition to bite-in value/average value, end-of-line value/average value,
Regarding the vibration value, the koutosis value and skewness value are also calculated. From the analysis variables stored in the data storage device 14, variables to be actually analyzed are extracted and combined.
Using the analyzer 15, for these selected variables,
Perform principal component analysis. The method of principal component analysis is well known and will therefore be omitted. The output of analyzer 15 is input to output device 16 . Analyzer 15 receives further setting inputs from pre-input device 17 . Specific examples are shown below. In the case of this example, the following eight variables were selected and principal component analysis was performed. X1: Rolling load at biting X2: Rolling load biting value/average value X3: Spindle torque at biting X4: Torque biting value/average value X5: Backup, roll, chock, line direction vibration when biting X6: Biting value/average value of back-up/roll/chock/line direction vibration X7: Backup/roll/chock vertical vibration during biting X8: Biting value/average of back-up/roll/chock vertical vibration These 8 Tables 1 to 4 show the results of analyzing 49 data for each variable. Table 1 shows the correlation coefficient of each variable, Table 2 shows the eigenvalue, contribution rate and cumulative contribution rate obtained from the correlation matrix, and Table 3 shows the correlation coefficient of each variable.
Table 4 shows the factor loadings determined from the correlation matrix, and Table 4 shows the results of principal component analysis.

【table】

【table】

【table】

[Table] Figure 3 shows the score calculation results for principal components Z ₁ and Z ₂ , which will be described later. Using this analysis result, the condition of the rolling mill 1 is evaluated. When principal component analysis is applied to the variables X ₁ to X ₈ , Z ₁ to Z ₅ are naturally determined in descending order of contribution rate. However, at this point, it is unclear what meaning Z ₁ to Z ₅ have. In this embodiment, it is from Z ₁ to Z ₅ , but if this is increased, the cumulative contribution rate approaches 1.0.
However, in practice, only three variables can be represented on paper. Furthermore, in order to make it simple and easy to understand two-dimensionally, the limit is two. Therefore, this embodiment also uses two variables, Z ₁ and Z ₂ . Here, the meaning of Z ₁ and Z ₂ is determined by the factor loadings of X ₁ to X ₈ with respect to Z ₁ , as will be described later. Depending on what kind of data is used for X ₁ to _{X 8} ,
Naturally, the meanings of Z ₁ to _{Z 5} will also be different. Now, the meaning that Z ₁ and Z ₂ should have depends on what condition of the rolling mill is mainly monitored, so it is necessary to decide the combination of data to be used for X ₁ to _{X 8} so that the characteristics can be obtained. This varies depending on the type and structure of the rolling mill, and is not a fixed property. The present invention is a monitoring method for determining wear of a work roll drive universal joint. Z ₁ and Z ₂ determined from the data used for X ₁ to _{X 8} must be meaningful for determining the wear of the universal joint. This embodiment is for monitoring the degree of impact at the time of biting and determining whether there is abnormal play in the torque transmission system. Therefore, there is no need to consider variables other than X ₁ to X ₈ . In the case of this example, the cumulative contribution rate of principal components Z ₁ and Z ₂ is 82.2%. This means that the amount of information held by the analytical variables X1 to X8 can be reduced by two new variables, Z ₁ and Z ₂ .
This means that 82.2% can be expressed. When considering the characteristics of Z ₁ and Z ₂ , we pay attention to the factor loading of each analysis variable with respect to Z ₁ and Z ₂ . Z ₁ : Each variable shows a relatively high value, and all take positive values. This means that the value of Z ₁ increases no matter which variable increases, indicating the overall size at the time of biting. Z ₂ : Load and torque take positive values, vibration value takes negative values. In addition, both the load and torque have bite values/
If the average factor loading value is large compared to others and Z ₂ tends to deviate from 0 in the positive direction, it means that the impact load or torque value tends to be large. As the vibration value increases, the negative value tends to increase. FIG. 3 shows the calculated and plotted values of Z ₁ and Z ₂ mentioned above. Specific calculation formulas are shown based on the examples in the text. Z ₁ =0.387X ₁ +0.331X ₂ +0.395X ₃ +0.266X ₄ +0.374X ₅ +0.380X ₆ +0.299X ₇ +0.374X ₈ …(1) Z ₂ =0.187X ₁ +0.467X ₂ +0. 221X ₃ +0.526X ₄ −0.310X ₅ −0.296X ₆ −0.391X ₇ −0.291X ₈ …(2) In Figure 3, the scale sizes of Z ₁ and Z ₂ are
It is arbitrarily determined based on the value obtained from the above-mentioned calculation formula, and is not a fixed numerical value. Note that X ₁ to _{X 8} in the above formulas (1) and (2) are measured values of each data or converted values obtained from the measured values. In the present invention, the measured values are subjected to statistical processing using principal component analysis. In this embodiment, the variables are _X1 to _X8 , but there is no particular limitation. Statistical processing using principal component analysis is limited to the processing performed in the embodiment. The data content used for X ₁ to _{X 8} varies depending on what is mainly monitored and the type of rolling mill, so the Z ₁ The meaning of ~Z ₂ cannot be definitively defined. However, it is necessary to clarify the meaning of Z ₁ to _{Z 5} depending on the monitoring purpose, and to determine X ₁ to X ₈ so that Z ₁ to _{Z 5} have that meaning. Therefore, it is necessary to try several combinations of the frequencies of X ₁ to _{X 8} and find the one Z ₁ to _{Z 5} that has the most meaningful meaning for the purpose of monitoring. Based on the characteristics of these Z ₁ and Z ₂ , the equipment condition is evaluated using the score scatter diagram shown in FIG. In the case of this example, data obtained when the equipment was in a state that was considered to be normal and data immediately before the spindle slipper metal was replaced were analyzed. From the score distribution state, N when normal and A when slipper metal wear progresses.
It can be seen that there is a clear difference in the data trends.
By monitoring the score distribution state of this variable, it is possible to comprehensively monitor the rolling mill including the power transmission system. In the case of this example, the combination of variables was used to monitor the degree of impact at the time of biting, but if it is desired to mainly monitor abnormal vibrations of the mill housing body, other combinations of variables can be made as desired. ing.
Examples of variable combinations are shown below. X1: Torque biting value X2: Torque average value (during steady rolling) X3: Vibration biting value in back-up, roll, chock, and line directions X4: Average value of vibration in back-up, roll, chock, and line directions (during steady rolling) X5: Backup roll chock vertical vibration bite value X6: Backup roll chock vertical vibration average value (during steady rolling) Information can be extracted and equipment status can be accurately determined.

[Brief explanation of the drawing]

FIG. 1 is a schematic structural explanatory diagram of an apparatus for carrying out the method of the present invention. FIG. 2 is an explanatory diagram of importing data obtained from a rolling mill. FIG. 3 is a graph showing the results of principal component score calculations. 1: Rolling mill, 2: Main motor, 3: Main reducer,
4: Pinion stand, 5: Mill housing,
6: Backup roll chock, 7: Work roll, 8: Universal joint, 9: Analysis device.

Claims (1)

[Claims] 1. Measuring the rolling load, drive torque, and vibration of the drive system of a rolling mill, and statistically processing the measured values by principal component analysis to determine the wear of the work roll drive universal joint. A rolling mill monitoring method consisting of: 2. The method according to claim 1, characterized in that vibrations of a roll chock in a rolling mill are measured. 3. The method according to claim 1, wherein the ratio between the maximum value and the average value of each measurement value is used as data for principal component analysis.