Abnormal diagnosis device and abnormal diagnosis method

Abstract

This anomaly diagnosis device comprises: a means for acquiring, from a piece of equipment currently in operation, a signal indicating the state of the equipment as multidimensional data including a signal indicating the current time and a signal indicating a prescribed time period prior to the current time; a means for comparing a signal indicating a state of normal operation of the equipment at a specific time to a signal indicating a prescribed time period prior to the specific time for the acquired multidimensional data, and calculating the distance between the signals; and a means for determining that an anomaly has occurred if the calculated distance is greater than a predetermined threshold.

Description

Technical Field

[0001] The present invention relates to an abnormality diagnosis device and an abnormality diagnosis method.

Background Art

[0002] In the abnormal monitoring of a production line composed of a combination of various facilities such as a steel process, it is important to provide an abnormal monitoring system or the like that can be generally applied without being conscious of the types of monitoring targets and can detect the abnormality with high accuracy.

[0003] For example, in Patent Document 1, a method has been proposed in which a plurality of measured values collected from facilities are regarded as features in a variable space, and abnormality detection is efficiently performed based on the distance from the normal-time data in the variable space.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, since the technique described in Patent Document 1 is a method for detecting an abnormality based on the correlation relationship between a plurality of signals, it is premised that there exists a combination of signals in a characteristic relationship.

[0006] On the other hand, as a concept of abnormality detection, it is also possible to focus on the dynamic characteristics of a single time-series signal itself. Since facilities undergo aging changes due to use, for example, when changes occur in friction or resistance, the facility operation may become slower than normal. In such a case, it is necessary to investigate the time change of the operation in detail. However, since the technique described in Patent Document 1 is based on the correlation between signals, it cannot evaluate the time dynamic characteristics.

[0007] The present invention has been made in view of the above, and by focusing on a single time-series signal, for example, in equipment that is driven batchwise and repeats the same operation during driving, or in equipment that operates continuously while maintaining a certain operation, it is possible to provide an abnormality diagnosis apparatus and an abnormality diagnosis method that can monitor based on the temporal dynamic characteristics of the output signal. Further, an object is to provide an abnormality diagnosis apparatus and an abnormality diagnosis method that can diagnose not only abnormal states experienced in the past regarding the manufacturing state of a manufacturing process but also similar unexperienced abnormal states. Further, for a plurality of pieces of equipment, in equipment that operates in conjunction with a certain time difference, based on a signal indicating the state of the equipment showing the interlocking operation, it is possible to diagnose the abnormal state of the equipment, and an object is to provide an abnormality diagnosis apparatus and an abnormality diagnosis method.

Means for Solving the Problems

[0008] In order to solve the above-described problems and achieve the object, an abnormality diagnosis apparatus according to the present invention includes means for acquiring, as multi-dimensional data including a signal indicating the state of the equipment from the equipment during operation, a signal at the current time, and a signal a predetermined time in the past from the current time; means for comparing the acquired multi-dimensional data with multi-dimensional data indicating the state of the equipment during normal operation and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; and means for determining that it is abnormal when the calculated distance is greater than a predetermined threshold value.

[0009] Further, in the abnormality diagnosis apparatus according to the present invention, in the above invention, the predetermined time is determined based on the time constant of the equipment or, in the case of a periodic signal, based on the period of the signal.

[0010] Further, in the abnormality diagnosis apparatus according to the present invention, in the above invention, the multi-dimensional data indicating the state of the equipment during operation includes data regarding the operating conditions of the equipment, and is composed of a multi-dimensional distribution having the signal at the current time, the signal a predetermined time in the past from the current time, and the data regarding the operating conditions as coordinate axes.

[0011] In addition, in the abnormal diagnosis apparatus according to the present invention, the multidimensional data indicating the state of the facility during normal operation is a signal acquired from the facility operating normally during operation, and is composed of a multidimensional distribution with the signal at a specific time and the signal a predetermined time in the past from the specific time as coordinate axes.

[0012] In addition, in the abnormal diagnosis apparatus according to the present invention, the multidimensional data indicating the state of the facility during normal operation includes data related to the operating conditions of the facility, and is composed of a multidimensional distribution with the signal at a specific time, the signal a predetermined time in the past from the specific time, and the data related to the operating conditions as coordinate axes.

[0013] In order to solve the above-described problems and achieve the object, the abnormal diagnosis apparatus according to the present invention includes means for acquiring, as multidimensional data including a signal at the current time and a signal a predetermined time in the past from the current time, a plurality of signals indicating the state of the facility from a plurality of facilities that operate in conjunction; means for comparing the acquired multidimensional data with multidimensional data indicating the state of the plurality of facilities during normal operation and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; and means for determining that there is an abnormality when the calculated distance is greater than a predetermined threshold value.

[0014] In order to solve the above-described problems and achieve the object, the abnormal diagnosis method according to the present invention includes a step of acquiring, as multidimensional data including a signal at the current time and a signal a predetermined time in the past from the current time, a signal indicating the state of the facility from the facility during operation; a step of comparing the acquired multidimensional data with multidimensional data indicating the state of the facility during normal operation and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; and a step of determining that there is an abnormality when the calculated distance is greater than a predetermined threshold value.

[0015] In order to solve the above-described problems and achieve the object, an abnormality diagnosis method according to the present invention includes: a step of acquiring, as multi-dimensional data, a plurality of signals indicating the states of the facilities from a plurality of facilities that operate in conjunction with each other, the multi-dimensional data including the signal at the current time and the signal at a predetermined time in the past from the current time; a step of comparing the acquired multi-dimensional data with multi-dimensional data indicating the states of the plurality of facilities during normal operation and including the signal at a specific time and the signal at a predetermined time in the past from the specific time, and calculating the distance between the two; and a step of determining that there is an abnormality when the calculated distance is greater than a predetermined threshold value.

Advantages of the Invention

[0016] According to the abnormality diagnosis apparatus and the abnormality diagnosis method of the present invention, by focusing on a single time-series signal, it is possible to monitor based on the temporal dynamic characteristics of the output signal even in facilities that are driven batchwise and repeat the same operation, for example, during driving. Further, according to the abnormality diagnosis apparatus and the abnormality diagnosis method of the present invention, by focusing on a single time-series signal, it is possible to monitor based on the temporal dynamic characteristics of the output signal even in facilities that operate continuously while maintaining a certain operation. Further, according to the abnormality diagnosis apparatus and the abnormality diagnosis method of the present invention, by focusing on the temporal dynamic characteristics of a single signal, it is possible to diagnose not only abnormal states experienced in the past regarding the manufacturing state of the manufacturing process but also similar unexperienced abnormal states. Furthermore, according to the abnormality diagnosis apparatus and the abnormality diagnosis method of the present invention, even for a plurality of facilities, in facilities that operate in conjunction with each other with a certain time difference, it is possible to diagnose the abnormal state of the facility based on the signal indicating the state of the facility showing the interlocking operation.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiments for Carrying Out the Invention

[0018] The abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiments of the present invention will be described with reference to the drawings. Note that the present invention is not limited to the following embodiments, and the constituent elements in the following embodiments include those that can be replaced by those skilled in the art and are easy to replace, or those that are substantially the same.

[0019] (Abnormality Diagnosis System) The abnormality diagnosis system according to the embodiment will be described with reference to FIG. 1. The abnormality diagnosis system 1 includes a setting storage device 10, a data input device 20, an abnormality diagnosis device 30, and a display input device 40. The setting storage device 10, the data input device 20, the abnormality diagnosis device 30, and the display input device 40 are connected by wire or wirelessly via a bus or the like, and are configured to be able to transmit and receive data to and from each other.

[0020] The setting storage device 10 is realized by, for example, a general-purpose database or the like. The setting storage device 10 includes an event detection definition storage unit 11, a processed data definition storage unit 12, a diagnostic parameter storage unit 13, and a reference normal data storage unit 14. Further, the setting storage device 10 includes a diagnostic input history data storage unit 15, a diagnostic result history data storage unit 16, and an abnormality determination parameter storage unit 17.

[0021] The event detection definition storage unit 11 stores an event detection definition (logic) when detecting a diagnostic event in the diagnostic event detection processing unit 31. The processed data definition storage unit 12 stores creation details and the like when creating diagnostic data in the diagnostic data creation processing unit 33.

[0022] The diagnostic parameter storage unit 13 stores various parameters and the like necessary when creating diagnostic data in the diagnostic data creation processing unit 33. The reference normal data storage unit 14 stores normal-time data (hereinafter referred to as "normal data") acquired from equipment operating normally during operation.

[0023] The diagnostic input history data storage unit 15 stores diagnostic conditions and the like input when performing an abnormality diagnosis in the abnormality degree calculation processing unit 34. The diagnostic result history data storage unit 16 stores the results of the abnormality diagnosis in the abnormality degree calculation processing unit 34. The abnormality determination parameter storage unit 17 stores various parameters and the like necessary when performing an abnormality diagnosis in the abnormality degree calculation processing unit 34.

[0024] The data input device 20 is for inputting various information to the abnormality diagnosis device 30. The data input device 20 includes a data input processing unit 21 that inputs sensor signals acquired via a PLC (Programable Logic Controller) or a DCS (Distributed Control System) to the abnormality diagnosis device 30 by continuous processing or batch processing.

[0025] The abnormality diagnosis device 30 collects time-series data from a plant composed of a large number of facilities, and focuses on the temporal dynamic characteristics of the collected time-series data to perform an abnormality diagnosis of the process. Examples of the "process" in the present embodiment include a manufacturing process of steel products, a power generation process of power generation facilities, a conveyance process of conveyance facilities, and the like.

[0026] The abnormality diagnosis device 30 includes a processor (arithmetic processing unit) such as a CPU (Central Processing Unit), and a memory (main storage unit) composed of a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. Then, the processor executes a computer program to control each component and the like, thereby realizing a function that matches a predetermined purpose. The abnormality diagnosis device 30 functions as a diagnosis event detection processing unit 31, a processed data acquisition processing unit 32, a diagnostic data creation processing unit 33, an abnormality degree calculation processing unit 34, and an abnormality determination processing unit 35 through the execution of the above-described computer program.

[0027] Further, the abnormality diagnosis device 30 may separately include a storage unit that stores various processing results and the like. This storage unit is composed of a recording medium such as an EPROM (Erasable Programmable ROM), a hard disk drive (HDD), a solid state drive (SSD), and a removable medium. Examples of the removable medium include disk recording media such as a USB (Universal Serial Bus) memory, a CD (Compact Disc), a DVD (Digital Versatile Disc), and a BD (Blu-ray (registered trademark) Disc).

[0028] The diagnosis event detection processing unit 31 determines whether or not a diagnosis target event has occurred in a time-series signal (hereinafter referred to as "equipment signal") indicating the state of the equipment during operation, based on the event detection definition stored in the event detection definition storage unit 11.

[0029] When the diagnostic event detection processing unit 31 determines that an event to be diagnosed has occurred, the processing data acquisition processing unit 32 acquires processing data (equipment signal) from the data input device 20.

[0030] The diagnostic data creation processing unit 33 creates a diagnostic data set by cutting out the current signal and the signals in the past from the equipment signal. That is, the diagnostic data creation processing unit 33 creates a diagnostic data set composed of the value of the equipment signal in the original coordinate system to be the object of the abnormality diagnosis and the value of the equipment signal in one or more time-delay coordinate systems for the equipment signal acquired by the processing data acquisition processing unit 32.

[0031] Specifically, when the diagnostic data set is configured with two-dimensional data, taking the observed original time-series signal y(t) and representing the time delay as τ, the time-series signal with delay is y(t - τ). Also, the combination of these signals (y(t), y(t - τ)) becomes the target two-dimensional data. The diagnostic data set is generated by acquiring such a set of two-dimensional data or a set of multi-dimensional data in time series. Here, the generated data set may be created by predetermining the data acquisition period from the equipment during operation. The data acquisition time may be, for example, the same as the acquisition time of the normal data described later, but there is no particular problem as long as it is data during operation, even if it is shorter or longer.

[0032] Also, in the present embodiment, the above-mentioned "original coordinate system" and "time-delay coordinate system" have the same value of the observed data represented by the coordinate system, but utilize the difference in the acquisition time of the observed data to represent the value of the observed data according to the acquisition time when the data is distributed and displayed in multiple dimensions. That is, the "original coordinate system" is the coordinate axis representing the value at the original acquisition time of the observed data, and the "time-delay coordinate system" is the coordinate axis representing the value of the observed data at a time delayed by a predetermined time from the original observed data. In this way, the diagnostic data creation processing unit 33 acquires the equipment signal from the equipment during operation as multi-dimensional data including the current signal and the signal a predetermined time in the past from the current time.

[0033] The multidimensional data created by the diagnostic data creation processing unit 33 is, for example, a multidimensional distribution with the equipment signal at the current time and the equipment signal a predetermined time in the past from the current time as coordinate axes. For example, when only one equipment signal a predetermined time in the past from the current time is included in the multidimensional data, it becomes a two-dimensional distribution as in the above example. Also, when a plurality of equipment signals a predetermined time in the past from the current time are included in the multidimensional data, it becomes a distribution of three dimensions or more. Note that "including a plurality of equipment signals a predetermined time in the past from the current time" indicates that, for the original equipment signal, signals delayed by, for example, 5 steps, 10 steps, 15 steps, etc. are included. In this way, by using a diagnostic data set including signals in a plurality of time-delay coordinate systems for the original equipment signal, the characteristics of the waveform of the original equipment signal can be extracted more appropriately.

[0034] The above-mentioned "predetermined time", that is, the magnitude of the time delay amount (also simply referred to as "time delay amount") is determined based on the characteristics (such as waveform, etc.) of the equipment signal to be handled. The predetermined time can be determined, for example, based on the time constant of the equipment to be diagnosed. Also, for example, when the equipment signal is a periodic signal, the predetermined time can be determined, for example, based on the period of the equipment signal.

[0035] By setting the magnitude of the time delay amount to be an integer multiple of the sample period of the equipment signal, for example, the calculation cost can be reduced. Also, for example, when handling an equipment signal that undergoes periodic fluctuations, it is preferable to use a time delay amount shorter than the period, such as 1 / 2 or 1 / 4 of the period of the equipment signal. Also, when capturing the rising (or falling) behavior of a signal change measured along with the acceleration or deceleration of a drive device, it is preferable to use a time delay amount shorter than the rising time (or falling time).

[0036] In addition, in equipment controlled in various operation modes, the behavior (time dynamic characteristics) of the facility signals to be handled may depend on the operation mode and control amount. For example, in the case of a granulation mixer, since the inertial load of the rotating body driven by the motor changes depending on the type and amount of the inserted material, the dynamic characteristics of the torque current of the motor associated with the operation depend on the operating conditions (operation conditions).

[0037] Thus, when diagnosing a facility whose dynamic characteristics of time-series signals change depending on the operating conditions of the facility to be diagnosed, the diagnostic data creation processing unit 33 creates a diagnostic data set with data related to one or more operating conditions (for example, control set values, etc.). In this case, the multidimensional data created by the diagnostic data creation processing unit 33 includes data related to the operating conditions of the facility, and becomes a multidimensional distribution with the signal at the current time, the signal a predetermined time in the past from the current time, and the data related to the operating conditions as coordinate axes.

[0038] Furthermore, even for multiple facilities, in facilities that operate in conjunction with a certain time difference, the abnormal state of the facility can be diagnosed based on the signal indicating the state of the facility showing the interlocking operation. For example, as multiple facilities, the table motors in the conveying table of the hot rolling process can be mentioned.

[0039] The conveying table of the hot rolling process is composed of a plurality of conveying rolls for conveying steel plates. Each of these conveying rolls is driven by a table motor that drives a predetermined number of rolls together, and the table motor is arranged for each predetermined number of conveying rolls to be driven. Therefore, these multiple table motors will operate in conjunction with a similar operation while having a predetermined time difference.

[0040] In such a case, the diagnostic data creation processing unit 33 acquires, as multi-dimensional data, a plurality of signals indicating the states of facilities from a plurality of facilities operating in conjunction, including the signal at the current time and the signal at a predetermined time in the past from the current time. That is, the diagnostic data creation processing unit 33 creates a diagnostic data set by extracting, as facility signals, the signal at one current point in time of the facility to be diagnosed and the signals operating in conjunction in other facilities in the past.

[0041] When the diagnostic data is composed of two-dimensional data, this two-dimensional data is composed of the same type of facility signals of two facilities. When the time difference between the current data and the past data is known in advance for the time lag of the facilities operating in conjunction, a known value can be used. Also, when the time lag of the facilities operating in conjunction is unknown, a large number of time lag axes can be set, and it is possible to adopt the time lag at which the correlation between the standard time axis and the time lag axis is the largest. At this time, when the correlation can be regarded as a linear correlation, the Pearson correlation coefficient may be used. Also, when the correlation is non-linear, the maximum mutual information amount (MIC: Maximal Information Coefficient) representing the strength of the correlation of the non-linear distribution can be used. Note that the diagnostic data may be three-dimensional or higher. Also, the multi-dimensional data may include data related to the operating conditions of the facilities.

[0042] In the above description, the same type of signal such as the peripheral speed of the table motor is being handled. However, for example, signals with a recognized correlation such as motor speed, motor current, or startup command (pulse signal) to the motor can be used as a data set of different types of signals.

[0043] The abnormality degree calculation processing unit 34 compares the multi-dimensional data created by the diagnostic data creation processing unit 33 with the normal data created in advance, and calculates the distance (deviation) between the two as the abnormality degree.

[0044] The above-mentioned "normal data" is equipment signals obtained from equipment operating normally during operation, with the data acquisition time determined in consideration of the equipment operation and dynamic characteristics of the equipment, and is pre-stored in the reference normal data storage unit 14. Further, the normal data indicates the state of the equipment during normal operation, and is multi-dimensional data including the equipment signal at a specific time and the equipment signals at a predetermined time in the past from the specific time. Specifically, this multi-dimensional data is multi-dimensional data with the equipment signal at a specific time and the equipment signals at a predetermined time in the past from the specific time as coordinate axes, and since it is a data group obtained by determining the data acquisition time in consideration of the equipment operation and dynamic characteristics of the equipment as described above, it forms a multi-dimensional distribution. Here, the data acquisition time of the normal data may be determined within a period that spans one cycle if the equipment operation is a periodic operation. Also, if it is a response with a known time constant such as a step response, it may be determined within a range up to the vicinity of the settling. Further, if it is neither of the above, the acquisition time may be determined in accordance with the start and end of the target operation of the equipment.

[0045] In addition, when diagnosing equipment whose dynamic characteristics of the time-series signal change depending on the operating conditions of the equipment to be diagnosed, the abnormality degree calculation processing unit 34 uses normal data including data related to one or more operating conditions. In this case, the normal data used by the abnormality degree calculation processing unit 34 forms a multi-dimensional distribution with the signal at a specific time, the signal at a predetermined time in the past from the specific time, and the data related to the operating conditions as coordinate axes.

[0046] The abnormality degree calculation processing unit 34 calculates the distance (divergence) between the multi-dimensional data created by the diagnostic data creation processing unit 33 and the normal data as the abnormality degree. The abnormality degree calculation processing unit 34 can calculate, for example, the Euclidean distance between the two as the distance between the multi-dimensional data created by the diagnostic data creation processing unit 33 and the normal data.

[0047] In this way, when calculating the distance between the multi-dimensional data created by the diagnostic data creation processing unit 33 and the normal data, the abnormality degree calculation processing unit 34 uses the multi-dimensional distribution in the coordinate system including the original coordinates and the time-lagged coordinates as the geometric features possessed by these data.

[0048] Also, as described above, when the time-varying characteristics of the facility signal to be handled depend on the operating conditions of the facility, the abnormality degree calculation processing unit 34 calculates the distance between multidimensional distributions in which, in addition to the above two coordinate axes, data related to operating conditions such as control set values are added as coordinate axes. In this way, even when the target facility signal has more complex time-varying characteristics, by adding coordinates related to operating conditions, coordinates with different multiple time delay amounts, etc., it becomes possible to handle them as geometric features within the multidimensional distribution.

[0049] When the distance calculated by the abnormality degree calculation processing unit 34 is greater than a predetermined threshold value, the abnormality determination processing unit 35 determines that it is abnormal and detects an abnormal event. Note that the distance calculated here will be described later. Also, when the distance calculated by the abnormality degree calculation processing unit 34 is less than a predetermined threshold value, the abnormality determination processing unit 35 determines that it is normal. Then, the abnormality determination processing unit 35 outputs the abnormality diagnosis result and the detected abnormal event to the display input device 40.

[0050] The display input device 40 is for displaying the abnormality diagnosis result and the like input from the abnormality diagnosis device 30. The display input device 40 is realized by, for example, a general-purpose display such as a liquid crystal display or an organic display.

[0051] (Abnormality Diagnosis Method) The abnormality diagnosis method according to the embodiment will be described with reference to FIG. 2. First, the diagnosis event detection processing unit 31 detects an event to be diagnosed from the facility signal of the facility during operation (step S1). Subsequently, the processing data acquisition processing unit 32 acquires processing data (facility signal) from the data input device 20 (step S2).

[0052] Subsequently, the diagnostic data creation processing unit 33 creates a diagnostic data set composed of the value of the equipment signal in the original coordinate system that is the target of the abnormality diagnosis and the value of the equipment signal in one or more time-delay coordinate systems for the processing data (step S3). Subsequently, the abnormality degree calculation processing unit 34 compares the diagnostic data set (multidimensional data) created by the diagnostic data creation processing unit 33 with the normal data created in advance, and calculates the distance between the two (step S4). In this step, specifically, for example, for each point of the diagnostic data set, the distance from each point of the normal data is calculated, and the nearest neighbor distance from the normal data is calculated.

[0053] Subsequently, the abnormality determination processing unit 35 determines whether the nearest neighbor distance calculated in step S4 is greater than a predetermined threshold value (step S5). Here, the maximum value, minimum value, average value, etc. of the nearest neighbor distance of each point may be calculated respectively, and any one or a plurality of combinations may be used for the determination. In step S5, if it is determined that the calculated distance is greater than the predetermined threshold value (Yes in step S5), the abnormality determination processing unit 35 outputs the detected abnormal event to the display input device 40 (step S6), and completes this processing. On the other hand, in step S5, if it is determined that the calculated distance is less than or equal to the predetermined threshold value (No in step S5), the abnormality determination processing unit 35 returns to the processing of step S1 and repeats the processing after step S1.

[0054] As described above, in the abnormality diagnosis method according to the embodiment, by continuously applying the series of steps of steps S1 to S6 to the time-series signal constantly measured in the equipment, the target signal can be monitored in real time.

[0055] (Example) Examples of the abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiment will be described with reference to FIGS. 3 to 8.

[0056] Figure 3 shows an example of the waveform of a facility signal indicating periodic movement. Examples of facility signals indicating such periodic movement include the current and rotational speed of the drive motor of a facility in which an eccentric body is rotationally driven, such as a granulation mixer.

[0057] In Figure 3, the horizontal axis represents time, and the vertical axis represents the values in the original coordinate system. Also, in the same figure, the normal data indicates the data during the normal operation of the facility. Further, the diagnostic data 1 is the data created by the abnormality diagnostic device and indicates normal data. Also, the diagnostic data 2 is the data created by the abnormality diagnostic device and indicates data with an abnormal amplitude. Also, the diagnostic data 3 is the data created by the abnormality diagnostic device and indicates data with an abnormal period.

[0058] In Figure 3, although the diagnostic data 1 is normal data, when compared with the normal data, the phases of the waveforms do not match. Therefore, for example, even if only the magnitude of the facility signal for a predetermined elapsed time from the specified diagnostic start point (data cut-out time) without considering the phase is compared, it is difficult to determine normal or abnormal.

[0059] On the other hand, Figure 4 shows, for the same waveform as in Figure 3, the data in the time-delay coordinate system created respectively, and shows the correlation between the values in the original coordinate system and the values in the time-delay coordinate system as a two-dimensional distribution. In the same figure, the horizontal axis represents the values in the original coordinate system, and the vertical axis represents the values in the time-delay coordinate system.

[0060] As shown in Figure 4, by removing the information in the time direction from the values in the original coordinate system, it becomes possible to separate and select the normal waveform and the abnormal waveform. That is, since the distribution of the diagnostic data 1 almost overlaps with the distribution of the normal data, it can be determined that the diagnostic data 1 is normal. On the other hand, since the distributions of the diagnostic data 2 and 3 deviate from the distribution of the normal data, it can be determined that the diagnostic data 2 and 3 are abnormal.

[0061] Next, FIG. 5 shows an example of the waveform of the position signal during (before and after driving) the driving of batch-driven equipment. Examples of equipment signals showing such rising movements include, for example, the position signal of the side guide on the hot rolling line in the steel field.

[0062] In FIG. 5, the horizontal axis represents time, and the vertical axis represents the values in the original coordinate system. Also, in the same figure, the normal data shows the data during the normal operation of the equipment. Further, the diagnostic data 1 is the data created by the abnormality diagnostic device and shows normal data. Also, the diagnostic data 2 is the data created by the abnormality diagnostic device and shows abnormal data, for example, when an overload state occurs due to mechanical friction of the drive mechanism.

[0063] In FIG. 5, although the diagnostic data 1 is normal data, when compared with the normal data, since the timing of the start of driving is different, it is difficult to determine that the diagnostic data 1 is normal only by comparing the observed values at each time series point. That is, as shown in the same figure, even when comparing the values in the original coordinate system, it is difficult to separate and select the normal waveform and the abnormal waveform.

[0064] On the other hand, FIG. 6 shows, for the same waveform as in FIG. 5, the data in the time-delay coordinate system is created respectively, and the correlation between the values in the original coordinate system and the values in the time-delay coordinate system is shown as a two-dimensional distribution. In the same figure, the horizontal axis represents the values in the original coordinate system, and the vertical axis represents the values in the time-delay coordinate system.

[0065] As shown in FIG. 6, by removing the information in the time direction from the values in the original coordinate system, it becomes possible to separate and select the normal waveform and the abnormal waveform. That is, since the distribution of the diagnostic data 1 almost overlaps with the distribution of the normal data, it can be determined that the diagnostic data 1 is normal. On the other hand, since the distribution of the diagnostic data 2 deviates from the distribution of the normal data, it can be determined that the diagnostic data 2 is abnormal.

[0066] Here, for example, each conveying roll of the conveying table in the hot rolling process is accelerated to a predetermined speed determined by the conveying conditions of the hot rolled steel sheet before the hot rolled steel sheet produced in batches is conveyed. Also, in the conveying table, a large number of table motors are arranged along the conveying line and operate sequentially from the upstream side. At this time, if there is a conveying table that does not operate or cannot be accelerated to the predetermined speed due to physical interference between each conveying roll, bearing failure, or motor body failure such as insulation degradation, etc., it may cause a failure of the conveying table itself and slip marks on the sheet bar. The abnormality diagnosis apparatus and abnormality diagnosis method according to the embodiment can also be used for diagnosing the abnormal states of a plurality of facilities that interlock with a certain time difference, such as the above table motors.

[0067] For example, FIG. 7 shows an example of the peripheral speeds (rotational speeds) of three different table motors of the conveying table in the hot rolling process. In the figure, the vertical axis represents the peripheral speed of the table motor and shows the ratio (%) to the standard rotational speed. The horizontal axis represents time and shows the number of steps for each sampling period. The sampling frequency can be determined appropriately, but for example, it may be 100 msec. In the figure, it can be seen that the three table motors A, B, and C are accelerating in conjunction with a predetermined time delay. Also, for the table motor B, the data at the time of abnormality is also shown.

[0068] On the other hand, FIG. 8 shows an example of the distribution of two-dimensional data combining the peripheral speed data of the table motor A and the peripheral speed data of the table motor B. (a) of FIG. 8 shows the distribution of two-dimensional data combining the peripheral speed data of the table motor A without time delay with respect to the peripheral speed data of the table motor B. Also, (b) of FIG. 8 shows the distribution of two-dimensional data combining the peripheral speed data of the table motor A with a time delay of 15 steps with respect to the peripheral speed data of the table motor B. Further, (c) of FIG. 8 shows the distribution of two-dimensional data combining the peripheral speed data of the table motor A with a time delay of 35 steps with respect to the peripheral speed data of the table motor B.

[0069] As is clear from FIGS. 7(a) and 7(b), the signal of table motor A and the signal of table motor B have a delay of 35 steps. In FIG. 8(c) showing the locus of two-dimensional data with this delay, the correlation between the two can be clearly understood. Also, since the distribution of the two-dimensional data at the time of abnormality shown in this figure clearly deviates from the distribution of the data at the normal time, it can be diagnosed that the data at the time of abnormality is abnormal.

[0070] According to the abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiment described above, by focusing on a single time-series signal, for example, in equipment that is driven batchwise and repeats the same operation during driving, or in equipment that operates continuously while maintaining a certain operation, it is possible to monitor based on the temporal dynamic characteristics of the output signal. Also, according to the abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiment, by focusing on the temporal dynamic characteristics of a single signal, it is possible to diagnose not only abnormal states experienced in the past regarding the manufacturing state of the manufacturing process but also similar unexperienced abnormal states.

[0071] Also, according to the abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiment, by focusing on a plurality of time-series signals, for a plurality of equipment that repeat the same operation or for equipment that interlocks at regular intervals, it is possible to monitor based on the temporal interlockability of the output signals. Also, according to the abnormality diagnosis apparatus and the abnormality diagnosis method according to the embodiment, by focusing on the temporal interlockability of a plurality of signals, it is possible to diagnose not only abnormal states experienced in the past regarding the manufacturing state of the manufacturing process but also similar unexperienced abnormal states.

[0072] As described above, the embodiments to which the invention made by the present inventor is applied have been described. However, the present invention is not limited by the description and drawings that form a part of the disclosure of the present invention according to this embodiment. That is, other embodiments, examples, operation techniques, etc. made by those skilled in the art based on this embodiment are all included in the scope of the present invention.

Description of Reference Numerals

[0073] 1 Abnormal Diagnosis System 10 Setting Storage Device 11 Event Detection Definition Storage Section 12 Process Data Definition Storage Section 13 Diagnosis Parameter Storage Section 14 Reference Normal Data Storage Section 15 Diagnosis Input History Data Storage Section 16 Diagnosis Result History Data Storage Section 17 Abnormality Judgment Parameter Storage Section 20 Data Input Device 21 Data Input Processing Section 30 Abnormal Diagnosis Device 31 Diagnosis Event Detection Processing Section 32 Process Data Acquisition Processing Section 33 Diagnosis Data Creation Processing Section 34 Abnormality Degree Calculation Processing Section 35 Abnormality Judgment Processing Section 40 Display Input Device

Claims

1. Means for acquiring, as multi-dimensional data including a signal indicating the state of the equipment during operation, a signal at the current time and a signal a predetermined time in the past from the current time; Means for comparing the acquired multi-dimensional data with multi-dimensional data indicating the state during normal operation of the equipment and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; Means for determining an abnormality when the calculated distance is greater than a predetermined threshold value; An abnormality diagnosis device comprising the above.

2. The abnormality diagnosis device according to claim 1, wherein the predetermined time is determined based on the time constant of the equipment or, in the case of a periodic signal, based on the period of the signal.

3. The multi-dimensional data indicating the state of the equipment during operation Includes data related to the operating conditions of the equipment, And consists of a multi-dimensional distribution with the signal at the current time, the signal a predetermined time in the past from the current time, and the data related to the operating conditions as coordinate axes, The abnormality diagnosis device according to claim 1.

4. The multi-dimensional data indicating the state during normal operation of the equipment Is a signal acquired from equipment operating normally during operation, And consists of a multi-dimensional distribution with the signal at a specific time and the signal a predetermined time in the past from the specific time as coordinate axes, The abnormality diagnosis device according to claim 1.

5. The multi-dimensional data indicating the state during normal operation of the equipment Includes data related to the operating conditions of the equipment, And consists of a multi-dimensional distribution with the signal at a specific time, the signal a predetermined time in the past from the specific time, and the data related to the operating conditions as coordinate axes, The abnormality diagnosis device according to claim 4.

6. Means for acquiring, as multi-dimensional data including a plurality of signals indicating the state of the equipment, a plurality of signals from a plurality of interlocking equipment, including a signal at the current time and a signal a predetermined time in the past from the current time; Means for comparing the acquired multi-dimensional data with multi-dimensional data indicating the state during normal operation of the plurality of equipment and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; Means for determining an abnormality when the calculated distance is greater than a predetermined threshold value; Comprising The time difference between the signal at the current time and the signal a predetermined time in the past from the current time Uses a known value when the time delay of the plurality of interlocking equipment is known in advance, And sets a number of delay time axes and uses the time delay at which the correlation between the standard time axis and the delay time axis is the largest when the time delay of the plurality of interlocking equipment is unknown, An abnormality diagnosis device.

7. Obtaining, from the equipment in operation, a signal indicating the state of the equipment as multi-dimensional data including a signal at the current time and a signal a predetermined time in the past from the current time; Comparing the obtained multi-dimensional data with multi-dimensional data indicating the state during normal operation of the equipment and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; Determining that there is an abnormality when the calculated distance is greater than a predetermined threshold value; An abnormality diagnosis method including the above steps.

8. Obtaining, from a plurality of interlocking equipment, a plurality of signals indicating the state of the equipment as multi-dimensional data including a signal at the current time and a signal a predetermined time in the past from the current time; Comparing the obtained multi-dimensional data with multi-dimensional data indicating the state during normal operation of the plurality of equipment and including a signal at a specific time and a signal a predetermined time in the past from the specific time, and calculating the distance between the two; Determining that there is an abnormality when the calculated distance is greater than a predetermined threshold value; Including the above steps, wherein the time difference between the signal at the current time and the signal a predetermined time in the past from the current time is when the time delays of the plurality of interlocking equipment are known in advance, using the known values, when the time delays of the plurality of interlocking equipment are unknown, setting a number of delay time axes and using the time delay at which the correlation between the standard time axis and the delay time axes is the greatest; An abnormality diagnosis method.

Images