JP5158018B2 - Equipment diagnosis apparatus and equipment diagnosis method for production system, equipment diagnosis program, and computer-readable recording medium recording the same - Google Patents

Description

  The present invention relates to an equipment diagnostic apparatus and equipment diagnostic method for diagnosing equipment of a steel production plant and other production systems, an equipment diagnostic program, and a computer-readable recording medium on which the equipment diagnostic program is recorded.

  In the steel production plant and other production systems, normal operation of each facility produces a normal product. Therefore, it is important to diagnose whether each facility is operating normally and to respond promptly when an abnormality occurs. Conventionally, as this kind of equipment diagnostic system, the thing of patent documents 1-4 is known, for example.

  The automatic diagnosis apparatus described in Patent Document 1 performs abnormality diagnosis by comparing the same operation repeated sequentially with reference to a normal signal change (on / off) pattern. In this automatic diagnosis apparatus, a reference pattern is automatically created and corrected instead of manually creating a reference pattern. However, even in such a facility that simply performs a sequential operation, there are many variations in signal changes depending on the state of the facility, making it difficult to cope with such a case.

  The diagnostic device described in Patent Document 2 performs statistical processing on the extracted feature amount, instead of simply determining a threshold value for a difference from the reference pattern, with respect to the automatic diagnostic device described in Patent Document 1 above. It is. Thereby, in this diagnostic device, it is possible to accurately determine even a certain degree of variation by comparing statistics.

  In the facility diagnosis method described in Patent Document 3, a simulation function is executed by software in a PLC (programmable logic controller) during actual operation of the facility, and an operation state input (FB (feedback) signal) from the facility and a pair thereof are performed. The equipment diagnosis is performed by comparing the simulation result corresponding to 1 with the simulation signal.

  The equipment diagnosis apparatus described in Patent Document 4 collects measurement data as raw data according to raw data collection definition data, performs feature value extraction from the collected raw data according to the definition data defined for the raw data, Alternatively, the diagnosis using the extracted feature amount is performed according to the definition data defined for the diagnosis.

JP-A-10-97318 JP 2008-226006 A JP-A-4-284504 JP 2005-141550 A

  The conventional diagnostic apparatuses described in Patent Literatures 1 to 4 can be applied only when a uniform signal is obtained under a certain sequence condition for each past case. In Patent Document 2, although some variation is allowed by statistical processing, it is not significantly different from Patent Documents 1, 3, and 4, and a person defines the feature amount that occurs when the equipment is in a normal state or an abnormal state. I need to do it.

  Also, in general, in diagnosis that focuses on sensing data such as temperature and vibration, or specific process data alone, secular changes are monitored and judged by thresholds, so the difference between normal and abnormal states is significant. If it does not appear, it is impossible to judge. In particular, when the control is dynamic, not the equipment that simply performs sequential operations, such as an iron manufacturing plant, the state of each equipment fluctuates depending on the product specification conditions (operation conditions). In this case, the behavior of the feature quantity is unclear, and it is difficult to determine the feature quantity.

  Therefore, in the present invention, equipment diagnosis apparatus and equipment of a production system capable of diagnosing normality or abnormality even in equipment where the behavior of the characteristic amount is unclear, such as an iron manufacturing plant It is an object of the present invention to provide a diagnostic method, an equipment diagnostic program, and a computer-readable recording medium on which the diagnostic program is recorded.

  The equipment diagnosis apparatus for a production system according to the present invention includes equipment status data indicating the status of each equipment constituting the production system, data collection means for collecting product status data indicating the status of a product produced by the production system, and the data Data classification means for classifying the equipment state data and product state data collected by the collection means into data when the product is normal and data when it is abnormal, and equipment under normal conditions classified by this data classification means Trigger detection means for detecting a feature amount using a predetermined operation of each equipment as a trigger for each of the state data and product state data, and equipment state data and product state data in the case of abnormality, and detected by the trigger detection means Feature values of equipment status data and product status data when normal, and equipment status data when abnormal Database registration means for registering each feature quantity in the feature quantity database only for those that have a difference greater than or equal to a predetermined threshold, and real-time information on each facility during the operation of the production system. A diagnostic means for diagnosing whether a product is normal or abnormal by extracting a characteristic quantity from equipment state data indicating a state and comparing it with a feature quantity database, and an output means for outputting a diagnosis result by the diagnostic means Is included.

  Further, the facility diagnosis method for a production system of the present invention collects facility state data indicating the state of each facility constituting the production system and product state data indicating the state of a product produced by the production system. Equipment status data and product status data are classified into data when the product is normal and data when the product is abnormal, equipment status data and product status data when the classification is normal, and when the product is abnormal For each of the equipment state data and the product state data, a feature amount is detected by using a predetermined operation of each equipment as a trigger, and the detected feature amount of the equipment state data and product state data in the normal case, Compare the feature values of the equipment status data and product status data for each case, and only those that differ by more than a predetermined threshold Whether the product is normal by registering the feature quantity in the feature quantity database, extracting the feature quantity from the equipment status data indicating the real-time status of each equipment during operation of the production system, and comparing it with the feature quantity database It includes diagnosing whether there is an abnormality and outputting the diagnosis result.

  According to these inventions, the equipment state data and the product state data are classified according to whether the product is normal or abnormal, and the feature amount is triggered by a predetermined operation of each equipment in each case of normal or abnormal Are detected and compared, and only those having a difference equal to or greater than a predetermined threshold are registered as significant feature amounts in the feature amount database. By this, even if the behavior of the feature quantity of the equipment state data and the product state data is unclear depending on whether the product is normal or abnormal, the feature quantity database consisting only of the feature quantity in which a clear difference is recognized, It is possible to diagnose whether the product is normal or abnormal.

  Here, it is desirable that the feature amount is detected by classifying the equipment state data and the product state data for each product specification condition. This makes it possible to obtain a feature database consisting of only features that have a clear difference for each product specification condition even in production systems with different equipment status data depending on the product specification conditions even under the same sequence conditions. It becomes possible to diagnose whether the product is normal or abnormal.

  In addition, it is desirable to classify the equipment state data and the product state data by excluding data when the product is normal or abnormal between the normal state and the abnormal state. As a result, it is possible to clearly classify whether the product is normal or abnormal and detect the feature amount, so it is possible to more accurately diagnose whether the product is normal or abnormal. .

  In addition, the production system equipment diagnosis program of the present invention is a data collection for collecting computer status information indicating the status of each equipment constituting the production system and product status data indicating the status of products produced by the production system. And data classification means for classifying the equipment status data and product status data collected by the data collection means into data when the product is normal and data when the product is abnormal, and the data classification means Trigger detection means for detecting a feature amount by using a predetermined operation of each facility as a trigger for each of the equipment state data and product state data in the normal state and the equipment state data and product state data in the case of abnormality, and the trigger Features of normal equipment state data and product state data detected by the detection means A database registration means for comparing feature quantities of equipment state data and product state data in the case of abnormality, and registering each feature quantity in the feature quantity database only for those having a difference of a predetermined threshold or more, and operation of the production system A diagnostic means for diagnosing whether the product is normal or abnormal by extracting the feature quantity from the equipment status data indicating the real-time status of each equipment and comparing it with the feature quantity database, and by this diagnostic means This is for functioning as an output means for outputting a diagnosis result. According to the computer which executed this program, the same operation and effect as the equipment diagnostic device of the production system of the present invention can be obtained. Further, the program can be read and executed by a computer by recording the program on a computer-readable recording medium.

(1) In the present invention, the equipment state data and product state data are classified according to whether the product is normal or abnormal, and the feature amount is triggered by a predetermined operation of each equipment in each case of normal or abnormal. Are detected and compared, and only those having a difference equal to or greater than a predetermined threshold value are registered as significant feature amounts in the feature amount database, so that a feature amount database including only feature amounts in which a clear difference is recognized can be obtained. This makes it possible to diagnose whether the product is normal or abnormal even in a facility where the behavior of the characteristic amount is unclear when it is in a normal state, such as an iron manufacturing plant.

(2) By classifying equipment state data and product state data according to product specification conditions and detecting feature quantities, in production systems where equipment state data differs depending on product specification conditions even under the same sequence conditions In addition, a feature quantity database consisting only of feature quantities in which a clear difference is recognized for each product specification condition is obtained, and it is possible to diagnose whether the product is normal or abnormal.

(3) By categorizing the equipment status data and product status data by excluding data in the middle between normal and abnormal products, it is clear whether the product is normal or abnormal Therefore, it is possible to more accurately diagnose whether the product is normal or abnormal.

It is explanatory drawing which shows the operation support system structure of the steel manufacturing plant as a production system in embodiment of this invention. It is a block block diagram of the equipment diagnostic computer of FIG. It is a flowchart which shows the process of the equipment diagnostic computer of FIG. It is explanatory drawing which shows the example which extracts a feature-value from the electric current and speed of a peripheral installation at the time of coil winding in a rolling process. It is a figure which shows the time change of MD current and PR current. It is a figure which shows the time change of MD speed. It is a figure which shows the time change of PR electric current. It is explanatory drawing which shows the example which extracts a feature-value from the roll electric current value of each roll of a roll table in a rolling process. It is a figure which shows the time change of a roll electric current value.

  FIG. 1 is an explanatory diagram showing the configuration of an operation support system of an iron manufacturing plant as a production system according to an embodiment of the present invention, and FIG. 2 is a block configuration diagram of the equipment diagnosis computer of FIG.

  In FIG. 1, an operation support system for an iron manufacturing plant according to an embodiment of the present invention includes a control valve, a solenoid valve, a motor, and various sensors and pumps for measuring flow rate, pressure, temperature, and the like that constitute each process line of the iron manufacturing plant. PLC (programmable logic controller) 2 as a control device (sequencer) for controlling each facility 1, a general-purpose control network 3 to which the PLC 2 is connected, and a facility state indicating the state of each facility 1 flowing through the general-purpose control network 3 A data collection computer 4 that collects data and a facility diagnosis computer 5 as a facility diagnosis device that diagnoses each facility 1 are configured.

  The facility diagnosis computer 5 is a computer that executes a facility diagnosis program recorded on a computer-readable recording medium (not shown). As shown in FIG. 2, a master information registration unit 10, a data collection unit 11, a data classification unit 12 , Trigger detection means 13, database (DB) registration means 14, diagnosis means 15, output means 16 and the like. The equipment diagnosis computer 5 has a master information database (DB) 17 and a feature quantity database (DB) 18 built in a storage device (not shown).

  The master information registration unit 10 registers master information such as the definition of the equipment 1 to be diagnosed, the definition of the feature quantity extraction method, and the definition of the threshold level in the master information DB 17. The master information registered in the master information DB 17 is referred to by the data collection unit 11, the data classification unit 12, and the trigger detection unit 13, and is used for processing by each unit 11-13.

  The data collection means 11 collects equipment state data collected by the data collection computer 4 and product state data indicating the state of products produced by this steelmaking plant. What data is collected is determined based on the definition of the facility 1 to be diagnosed, which is registered in advance in the master information DB 17 by the master information registration means 10. The equipment state data is various process data that can be grasped in real time such as various achievements and sequences during operation of each equipment 1. On the other hand, the product state data is data obtained by sensing or inspection results that cannot be recognized in real time, such as shape, wrinkles, and post-process information.

  The data classification unit 12 classifies the equipment state data and product state data collected by the data collection unit 11 into data when the product is normal and data when it is abnormal. The determination of normal or abnormal is made based on the definition of each threshold level registered in advance in the master information DB 17 by the master information registration means 10. In the present embodiment, the data classification unit 12 classifies the data by excluding data in the middle between the case where the product is normal and the case where the product is abnormal.

  The trigger detection means 13 performs a predetermined operation of each equipment 1 for each of the normal equipment state data and product state data classified by the data classification means 12 and the abnormal equipment state data and product state data. Is used as a trigger to detect the feature amount. The predetermined operation of each equipment 1 as a trigger is, for example, a change point of sequence data of each equipment 1. Note that what information is used as a feature amount is determined based on the definition of the feature amount extraction method registered in advance in the master information DB 17 by the master information registration unit 10.

  In the present embodiment, the trigger detection means 13 classifies the equipment state data and the product state data according to product specification conditions (for example, operation conditions such as class type, plate width, plate thickness, tension, and speed). Detect feature values. Moreover, in this embodiment, the trigger detection means 13 accumulate | stores the extracted feature-value and calculates | requires statistics, such as an average and a standard deviation.

  The DB registration unit 14 compares the feature amount of the equipment state data and the product state data in the normal case detected by the trigger detection unit 13 with the feature amount of the equipment state data and the product state data in the case of abnormality. Each feature amount is registered in the feature amount DB 18 only for those having a difference equal to or greater than the threshold value. The determination as to whether or not to register in the feature amount DB 18 is made based on the definition of each threshold level registered in advance in the master information DB 17 by the master information registration means 10. Further, registration in the feature amount DB 18 is performed using, for example, a sheet passing condition (a class type, a size, and the like) as a key.

  The diagnosis means 15 extracts whether or not the product is normal or abnormal by extracting the feature amount from the facility state data indicating the real-time state of each facility during the operation of the steel plant and comparing it with the feature amount DB 18. To do. Similar to the trigger detection unit 13, the feature amount extraction is performed based on the definition of the feature amount extraction method registered in advance in the master information DB 17 by the master information registration unit 10. The diagnosis of normality or abnormality is performed based on the definition of each threshold level registered in advance in the master information DB 17 by the master information registration means 10.

  The output means 16 outputs the diagnosis result by the diagnosis means 15 by a screen display, announcement, alarm, a control signal to each facility 1 or the like through a display, a speaker or the like. For example, the output means 16 displays a trend graph or a pass / fail judgment result for the diagnosis result, provides guidance for an assumed coping method registered in advance at the time of abnormality judgment, or automatically outputs a clear measure. Or reflected in the control signal of each facility 1.

  Next, processing by the equipment diagnosis computer 5 having the above configuration will be described with reference to the flowchart of FIG.

  First, in steps S100 and S101, the data collection means 11 uses the product normality data such as the shape, product status data such as bag and post-process information, and various results and sequences during operation of each facility 1 by the data collection means 11. Process data (equipment state data) are collected. Next, in step S102, the data classification unit 12 classifies the collected data into data when the product is normal and abnormal data according to a certain threshold. At this time, the data classifying unit 12 classifies the data by excluding data in the middle between the case where the product is normal and the case where the product is abnormal.

  If the product is normal data, in step S103a, the trigger detection means 13 delimits the mesh with operating conditions such as the class type, the plate width, the plate thickness, the tension, and the speed. Data corresponding to is classified. Subsequently, in step S104a, the trigger detection unit 13 extracts feature quantities from the continuous data using the sequence data of the electric meter control system as a trigger. Next, in step S105a, the extracted feature values are accumulated by the trigger detection means 13, and a statistic such as an average or standard deviation is obtained.

  On the other hand, even in the case of data when the product is abnormal, the same processing as steps S103a, S104a, and S105a is performed in steps S103b, S104b, and S105b.

  Next, in step S106, the DB registration means 14 compares the statistics of the same condition between the group of normal products (good group) and the group of abnormal products (bad group), and there is a difference greater than a certain threshold. It is judged whether or not. Here, a feature quantity for which a difference greater than a certain threshold is not recognized is not registered in the feature quantity DB 18 (step S107).

  In steps S108 and S109, for only those having a difference of a certain threshold value or more, the feature amount when the product is normal and the feature amount when the product is abnormal are registered in the feature amount DB 18, respectively. The above steps S100 to S109 are processed offline.

  The following steps S110 to S113 are processed online. In step S110, the diagnostic means 15 extracts feature quantities in real time from equipment state data indicating the real time state of each equipment 1 during operation of the steel plant. In step S111, the feature quantity extracted in real time is compared with the feature quantity DB 18, and normality or abnormality is diagnosed with a certain threshold value.

  As a result of this diagnosis, if there is an abnormality in step S112, the output means 16 announces it. In addition, possible countermeasures registered in advance for each feature amount are displayed. In addition, a clear countermeasure is automatically reflected in the control signal of each facility 1. On the other hand, in the case of normality, in step S113, the output means 16 gives guidance that it is normal.

  As described above, according to the equipment diagnosis computer 5 in the present embodiment, the equipment state data and the product state data are classified according to whether the product is normal or abnormal. A feature amount is detected using a predetermined operation of the facility as a trigger, and compared, and only those having a difference equal to or greater than a predetermined threshold are registered in the feature amount DB 18 as significant feature amounts. Thus, a feature quantity DB 18 consisting of only is obtained. Thereby, it is possible to diagnose whether the product is normal or abnormal even in the equipment 1 in which the behavior of the characteristic amount is unclear in a normal state like an iron manufacturing plant.

  In addition, the equipment diagnosis computer 5 in the present embodiment classifies the equipment state data and the product state data for each product specification condition and detects the feature amount, thereby detecting the product specification condition even under the same sequence condition. Even in a production system with different equipment state data, a feature value DB 18 consisting only of feature values for which a clear difference is recognized for each product specification condition can be obtained. Therefore, whether the product is normal or abnormal is diagnosed Is possible.

  Further, in the facility diagnosis computer 5 according to the present embodiment, the product is normal by classifying the facility state data and the product state data by excluding data when the product is normal and when it is abnormal. Since the feature amount can be detected by clearly classifying the case and the abnormal case, it is possible to more accurately diagnose whether the product is normal or abnormal.

  Next, as a first embodiment of the equipment diagnosis computer 5, as shown in FIG. 4, in the rolling process of the steel plant, the feature amount is extracted from the current and speed of the peripheral equipment at the time of winding the coil. An example in which slips or irregularities in the whirling shape can be detected will be described.

  As shown in FIG. 4, the sheet passing material 21 that has passed through the final rolling mill 20 in the rolling process is wound up by a take-up device (MD) 23 through a plate pressing pinch roll (PR) 22. At this time, PR current, MD current, and MD speed were collected as the equipment state data, and the cutting shapes of the plate front end and the plate tail end were collected as the product state data, and the feature value DB 18 was formed by the equipment diagnosis computer 5. .

(1) Detection of shape deterioration due to plate loosening at the start of cutting As a feature value registered in the feature value DB 18, a plate presser from the rise of MD current when the plate tip that has passed through the final rolling mill 20 reaches the cutting device 23. When the time difference until the PR current change (see FIG. 5) was extracted, it was possible to estimate the looseness of the plate tip. The longer this time difference, the worse the shape of the wrinkles at the tip of the plate.

(2) Detection of shape deterioration due to initial slip of brazing By extracting MD speed fluctuation (see Fig. 6) at the same timing as (1) above, slip between the coil and MD in the initial stage of brazing can be estimated. Met. When there was a lot of slip, the shape of the ridge at the tip of the plate deteriorated.

(3) Detection of shape deterioration at the time of final rolling mill strip removal By extracting the PR current change pattern (see FIG. 7) when the sheet passing material 21 has passed through the final rolling mill 20 at the final stage of rolling (see FIG. 7). It was estimated that the shape of the plate tail end was deteriorated due to excessive PR pressure. This is because the tension balance between the final rolling mill 20 and the MD is lost after the bottom end, and the PR current swings negatively due to the MD being pulled. Even after the bottom end, if the PR is kept down as it is during rolling, the plate may be displaced, so it is necessary to release the PR at the perfect timing of the bottom end. In a good example (quickly opening), the current that shook negatively immediately after falling off the tail returns. However, as the release timing is delayed, the current return timing is delayed, and the shape of the heel shape at the plate tail end deteriorates.

  As a second embodiment of the equipment diagnosis computer 5, as shown in FIG. 8, in the rolling process of the steel manufacturing plant, by extracting the feature amount from the roll current value of each roll of the roll table, the case where the surface flaw can be detected explain.

  As shown in FIG. 8, in the rolling process, the sheet passing material 30 is conveyed on each roll 32 of the roll table 31. At this time, the roll current value of each roll 32 corresponding to the sheet passing condition was collected as the equipment state data, the surface defect data was collected as the product state data, and the feature amount DB 18 was formed by the equipment diagnosis computer 5. .

  If uniform speed is not maintained by speed control, surface flaws become a problem. That is, even if the speed record calculated from the roll diameter of each roll 32 and the actual rotation speed detected from the pulse generator (PLG) attached to each roll 32 is within the allowable range, the roll diameter error, etc. If the actual roll peripheral speed differs depending on

  Therefore, the comparison value between the allowable range of the roll current value obtained according to the plate passing condition and the actual roll current value is classified into the normal case and the abnormal case based on the surface flaw data. The comparison value was registered in the feature value DB 18 as a feature value. As a result, during operation of the steel plant, it is possible to compare the feature quantity extracted in real time with the feature quantity DB 18 to diagnose whether the product is normal or abnormal.

  The equipment diagnosis apparatus and equipment diagnosis method, the equipment diagnosis program, and the computer-readable recording medium on which the equipment diagnosis program of the present invention is recorded are useful for diagnosing equipment of a steel manufacturing plant and other production systems.

1 Equipment 2 PLC
DESCRIPTION OF SYMBOLS 3 General-purpose control network 4 Data collection computer 5 Equipment diagnostic computer 10 Master information registration means 11 Data collection means 12 Data classification means 13 Trigger detection means 14 DB registration means 15 Diagnosis means 16 Output means 17 Master information DB
18 Feature DB

Claims (6)

Data collection means for collecting facility state data indicating the state of each facility constituting the production system and product state data indicating the state of a product produced by the production system;
Data classification means for classifying the equipment status data and product status data collected by the data collection means into data when the product is normal and data when abnormal,
For each of the normal equipment state data and product state data classified by the data classification means and the abnormal equipment state data and product state data, the feature amount is triggered by a predetermined operation of each equipment. Trigger detecting means for detecting;
The feature amount of the equipment state data and the product state data in the normal case detected by the trigger detection means is compared with the feature amount of the equipment state data and the product state data in the abnormal case, and is equal to or greater than a predetermined threshold value. Database registration means for registering each feature quantity in the feature quantity database only for those that have a difference;
Diagnosing whether the product is normal or abnormal by extracting a feature value from equipment state data indicating the real-time state of each equipment during operation of the production system and comparing it with the feature value database Means,
An equipment diagnosis apparatus for a production system, including output means for outputting a diagnosis result by the diagnosis means.   The equipment diagnosis apparatus for a production system according to claim 1, wherein the trigger detection means detects the feature amount by classifying the equipment state data and product state data for each specification condition of the product.   The equipment diagnosis apparatus for a production system according to claim 1 or 2, wherein the data classification means classifies the data by excluding data when the product is normal between an abnormal case and an abnormal case. Collecting equipment status data indicating the status of each equipment constituting the production system and product status data indicating the status of products produced by the production system;
Classifying the collected equipment state data and product state data into data when the product is normal and data when it is abnormal;
For each of the classified normal equipment state data and product state data and abnormal equipment state data and product state data, detecting a feature amount using a predetermined operation of each equipment as a trigger,
The detected feature amount of the equipment state data and product state data in the normal case is compared with the feature amount of the equipment state data and product state data in the abnormal case, and there is a difference greater than a predetermined threshold value. Registering each feature quantity only in the feature quantity database,
Diagnosing whether the product is normal or abnormal by extracting a feature quantity from the equipment state data indicating the real-time state of each equipment during operation of the production system and comparing it with the feature quantity database ,
A production system facility diagnosis method including outputting the diagnosis result. Data collection means for collecting facility state data indicating the state of each facility constituting the production system and product state data indicating the state of a product produced by the production system;
Data classification means for classifying the equipment status data and product status data collected by the data collection means into data when the product is normal and data when abnormal,
For each of the normal equipment state data and product state data classified by the data classification means and the abnormal equipment state data and product state data, the feature amount is triggered by a predetermined operation of each equipment. Trigger detecting means for detecting;
The feature amount of the equipment state data and the product state data in the normal case detected by the trigger detection means is compared with the feature amount of the equipment state data and the product state data in the abnormal case, and is equal to or greater than a predetermined threshold value. Database registration means for registering each feature quantity in the feature quantity database only for those that have a difference;
Diagnosing whether the product is normal or abnormal by extracting a feature value from equipment state data indicating the real-time state of each equipment during operation of the production system and comparing it with the feature value database Means,
An equipment diagnosis program for a production system for causing a computer to function as output means for outputting a diagnosis result by the diagnosis means.   A computer-readable recording medium on which the facility diagnosis program according to claim 5 is recorded.

Images