JP2016091378A - Diagnostic system, diagnostic method and diagnostic program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a diagnostic system, a diagnostic method and a diagnostic program capable of high accuracy diagnosis to which a diagnostic model common to a plurality of pieces of equipment assumed to be close in operation condition is applied.SOLUTION: An equipment-information calculation unit 12 obtains linked equipment information by linking fixed first linked equipment information and variable second linked equipment information, which are associated with monitoring items including basic equipment information about to-be-monitored equipment, to the basic equipment information. The equipment-information classification unit classifies the linked equipment information. An equipment-state diagnostic unit 13 obtains a feature quantity of an analysis result of analysis of measurement data corresponding to the monitoring items. The equipment-state diagnostic unit diagnoses a state of the to-be-monitored equipment using a diagnostic pattern of the feature quantity of the analysis result corresponding to a classification result of the equipment-information classification unit. The equipment-information classification unit obtains a feature quantity of the linked equipment information. The equipment-information classification unit classifies the linked equipment information to any of a plurality of equipment feature patterns stored in advance on the basis of the feature quantity of the linked equipment information.SELECTED DRAWING: Figure 1

Description

  Embodiments according to the present invention relate to a diagnostic system, a diagnostic method, and a diagnostic program.

  In plant industries such as steel and chemicals, equipment diagnostic techniques such as vibration diagnosis are actively introduced for the purpose of improving the reliability of equipment and reducing running costs.

  In general, when diagnosing the operating state of equipment, calculate the root mean square, effective value, characteristic frequency component value, or the amount of change in those values for the measured value of the equipment operating state, The presence or absence of an abnormality is determined by performing threshold determination. As described above, there is a simple system for diagnosing the operating state of the facility device by threshold determination. However, when precisely diagnosing the operating state of the facility device, it is necessary to consider the influence of the characteristics of each device.

  However, for example, in the case of vibration diagnosis, there are differences in the response characteristics and excitation force of each device due to load specifications, manufacturing errors, installation errors, differences in rigidity due to the combination of installation location and peripheral devices, etc. . For this reason, there has been a problem that it is difficult to determine an abnormality with an absolute value or a change amount.

  In order to solve such problems, it has been proposed to determine an abnormality by simulation combining a change amount (change rate) from a normal state, data and knowledge of diagnosis so far, and an operation parameter of the device. However, in such a proposal, the process up to the determination is complicated, and the difficulty of determination due to the difference in device characteristics is not overcome.

  In addition, the use of dimensionless feature parameters such as skewness and kurtosis that statistically represent the shape of the measured signal eliminates the effects of differences in specific characteristics of each equipment such as the measurement site and operating conditions. There is a diagnostic method.

  In the diagnosis using the dimensionless feature parameter, time waveform distribution such as displacement, vibration speed, vibration acceleration, etc. at one measurement point of the target device is used as a dimensional parameter used for calculation. However, when such a time waveform distribution is used, there are problems such as a change in signal intensity due to noise becomes an error in threshold determination, and the analyzed dimensionless feature parameter can only know whether there is an abnormality in the device. .

  For such a problem, for example, an operating state diagnosis system using a pattern identification method such as a neural network or a support vector machine has been proposed. In this type of diagnostic system, diagnosis is performed by comparing the feature quantity in the abnormal state with the feature quantity in the normal state to detect the presence or absence of abnormality, or by comparing the created case model with the sensor data. ing.

JP 2002-257667 A

  However, in order to diagnose the operating state of the equipment installed at the site with high accuracy, it is necessary to create a diagnostic model for diagnosis from abnormal state data of various actual machines. Moreover, it is difficult to apply the diagnostic sample thus created to other equipment.

  The present invention has been made to solve the above-described problem, and is a diagnostic system, a diagnostic method, and a diagnosis capable of performing a high-accuracy diagnosis by applying a common diagnostic model to a plurality of devices considered to have close operating conditions. The purpose is to provide a program.

  The diagnostic system according to the present embodiment includes a device information classification unit. The device information classification unit relates to the monitoring item including the basic device information of the device to be monitored, and the combined device information obtained by combining the fixed first coupled device information and the variable second coupled device information with the basic device information. To get. The device information classification unit classifies the combined device information. The diagnosis system includes a device state diagnosis unit. The device state diagnosis unit acquires a feature value of an analysis result obtained by analyzing measurement data corresponding to the monitoring item. The device state diagnosis unit diagnoses the state of the device to be monitored using the diagnosis pattern of the feature value of the analysis result corresponding to the classification result of the device information classification unit. The device information classification unit acquires the feature amount of the combined device information. The device information classification unit classifies the combined device information into one of a plurality of device feature patterns stored in advance based on the feature amount of the device information.

  ADVANTAGE OF THE INVENTION According to this invention, the highly accurate diagnosis which applied the diagnostic model common to the some apparatus considered that driving | running conditions are near can be performed.

It is a block diagram of diagnostic system 1 showing a 1st embodiment. It is a figure which shows the list of the information which monitoring data hold | maintain. It is a flowchart of the diagnostic method which shows 1st Embodiment. It is explanatory drawing for demonstrating operation | movement of the apparatus characteristic database 19 and the apparatus information classification | category part 12. FIG. It is explanatory drawing for demonstrating operation | movement of the apparatus state diagnostic part. It is a block diagram of the diagnostic system 1 which shows 2nd Embodiment. It is a flowchart of the diagnostic method which shows 2nd Embodiment. It is a block diagram of the diagnostic system 1 which shows 3rd Embodiment. It is a flowchart of the diagnostic method which shows 3rd Embodiment.

  Embodiments according to the present invention will be described below with reference to the drawings. This embodiment does not limit the present invention.

(First embodiment)
(Constitution)
FIG. 1 is a block diagram of a diagnostic system 1 showing the first embodiment. FIG. 2 is a diagram showing a list of information held in the monitoring data. In FIG. 2, device information held by each monitoring data from first monitoring data D1 to sixth monitoring data D6 described later is indicated by circles. The diagnostic system 1 of this embodiment can be used for diagnosis of the state of equipment provided in industrial facilities such as steel, chemicals, and power plants.

  As shown in FIG. 1, the diagnostic system 1 includes a measurement data analysis unit 11, a device information classification unit 12, a device state diagnosis unit 13, an output unit 14, a learning model generation unit 15, and an input unit 16. A maintenance management database 17, a device management database 18, a device feature database 19, and a diagnostic pattern database 101.

  Each component 11 to 101 of the diagnostic system 1 is hardware (for example, a computer) such as an arithmetic processing device, a storage device, a communication device, and a user interface, but may include software. In addition, each of the constituent units 11 to 101 may be mounted on one device or facility, or a device (for example, a cloud) in which some constituent units can communicate with other constituent units through an external network. On the top server or database).

(Measurement data analysis unit 11)
The measurement data analysis unit 11 includes a data processing unit 111 and an analysis data feature amount extraction unit 112.

(Data processing unit 111)
The data processing unit 111 acquires first monitoring data D1 from a measurement sensor (not shown). As shown in FIG. 2, the first monitoring data D1 includes monitoring items a set in advance for monitoring target devices (facility devices) and measurement data b corresponding to the monitoring items a.

  Here, the monitoring target device is, for example, a device that is a monitoring target based on a facility maintenance management plan. Moreover, the monitoring item a is configured by basic device information and measurement conditions necessary for diagnosing the state of the monitoring target device.

  The basic device information is device information indicating basic attributes of the monitoring target device, and is, for example, the model or installation location of the monitoring target device (including parts of the monitoring target device). The measurement conditions are measurement conditions when obtaining the measurement data b, and are, for example, a measurement direction, a measurement position, a measurement sensor, a measurement parameter, and the like.

  The measurement data b is, for example, the process amount of the monitoring target device necessary for monitoring the status of the monitoring item, and changes when a failure occurs in the monitoring target device. The measurement data b is, for example, vibration, temperature, rotation pulse, flow rate, pressure, etc. of the monitoring target device. The measurement data b is added with the time when the measurement data is acquired in order to determine the long-term deterioration tendency of the device.

  Specific examples of the first monitoring data D1 include the following D1_1 to D1_3. The following ai (i = 1 to 3) indicate different monitoring items (models) depending on the number of the subscript i. bi indicates measurement data corresponding to ai having the same i. For example, the following b1 is operation data of the rotating body a1. In the following, the fact that the monitoring data is composed of a plurality of data is expressed using a “+” sign.

D1_1: Rotating body a1 + operation data b1
D1_2: Electric valve a2 + operation data b2
D1_3: Lubricating oil sample a3 + analysis data b3

  The data processing unit 111 performs analysis processing on the measurement data b in the first monitoring data D1, thereby obtaining the analysis result (analysis data composed of various waveform data) c shown in FIG. The data processing unit 111 associates the analysis result c with the first monitoring data D1, thereby generating second monitoring data D2 used for extracting the feature amount of the analysis result c. The data processing unit 111 outputs the generated second monitoring data D2 to the analysis data feature amount extraction unit 112.

  Specific examples of the second monitoring data D2 include the following D2_1 to D2_3. The following ci shows the analysis result for bi having the same i. For example, the following c1 is a waveform analysis result for the operation data b1 of the rotating body a1.

D2_1: D1_1 + waveform analysis result c1
D2_2: D1_2 + analysis target signal extraction result c2
D2_3: D1_3 + analysis result (chromatographic analysis result, etc.) c3

(Analysis Data Feature Extraction Unit 112)
The analysis data feature quantity extraction unit 112 uses the analysis result c in the second monitoring data D2 input from the data processing unit 111, and the analysis result feature quantity shown in FIG. 2 (hereinafter also referred to as analysis result feature quantity) d. To extract. The analysis result feature amount d is, for example, a parameter indicating the feature of the analysis result. A specific aspect of the analysis result feature quantity d is not particularly limited as long as diagnosis using a diagnosis pattern described later is possible.

  The analysis data feature amount extraction unit 112 associates the extracted analysis result feature amount d with the second monitoring data D2, thereby generating third monitoring data D3 used for combining device information. The analysis data feature amount extraction unit 112 outputs the generated third monitoring data D3 to the device information classification unit 12.

  Specific examples of the third monitoring data D3 include the following D3_1 to D3_3. The following di indicates an analysis result feature amount extracted from ci having the same i. For example, d1 below is a raw waveform, spectrum, sharpness / distortion degree, Lissajous, current, and the like extracted from the waveform analysis result c1 of the operation data b1 of the rotating body a1.

D3_1: D2_1 + raw waveform, spectrum, sharpness / distortion, Lissajous, current, etc. d1
D3_2: D2_2 + motor current, motor voltage, thrust load, torque load, etc. d2
D3_3: D2_3 + viscosity, density, spectrum, etc. d3

  As described above, the measurement data analysis unit 11 analyzes the measurement data b and extracts the analysis result feature quantity d from the analysis result c. By extracting the analysis result feature quantity d, it is possible to diagnose the state of the monitoring target device using the analysis result feature quantity d as a diagnosis target.

(Device information classification unit 12)
The device information classification unit 12 includes a device information combination unit 121, a device information feature amount extraction unit 122, and a device feature determination unit 123.

(Device information combining unit 121)
The device information combining unit 121 acquires combined device information related to the monitoring item. Here, the combined device information is information obtained by combining the specification e of the monitoring target device and the maintenance data f shown in FIG. 2 with the basic device information. The specification e is an example of fixed first coupled device information to be coupled to the basic device information. The maintenance data f is an example of variable second coupled device information to be coupled to the basic device information. The specification e and the maintenance data f increase the amount of device information relative to the basic device information, thereby enabling device information classification by device feature amount pattern identification, which will be described later. The manner of combining the specification e and the maintenance data f with the basic device information is not particularly limited, and for example, a common identifier may be given.

  Here, conventionally, the specification e and the maintenance data f are sometimes used for evaluation of diagnosis results. On the other hand, in the present embodiment, the specification e and the maintenance data f are found to have a new utility of applying the device feature value to the pattern identification, and the specification e and the maintenance data f are used for diagnosis itself (diagnosis flow). Applicable. By applying the specification e and the maintenance data f, it becomes possible to identify the pattern of the device feature quantity based on the detailed data of the actual machine, and thus it is possible to appropriately select a diagnosis pattern to be described later.

  The specification e is, for example, the size or structure of the monitoring target device or parts of the device, data in the drawing, and the like. The specification e is fixed (ie, invariant) information. The specification e is managed by the device management database 18. The device management database 18 is an example of a first management database. For example, the device management database 18 manages the specification e in association with monitoring items (for example, basic device information).

  The maintenance data f is, for example, an operation history for each device such as an inspection plan and inspection results. The maintenance data f is information that is variable according to, for example, an inspection plan or an update (for example, addition or change) of an inspection result. The inspection result includes a diagnosis result of the state diagnosis of the monitoring target device already performed. The maintenance data f may include information such as how many months have passed since the overhaul of the equipment and how many hours the equipment has been operating. The maintenance data f is managed by the maintenance management database 17. The maintenance management database 17 is an example of a second management database. The maintenance management database 17 manages, for example, maintenance data f in association with monitoring items (for example, basic device information).

  The device information combining unit 121 is connected to the maintenance management database 17 and the device management database 18. The device information combination unit 121 acquires the third monitoring data D3 input from the analysis data feature amount extraction unit 112 to the device information classification unit 12. The device information combining unit 121 reads maintenance data f and specification e corresponding to the monitoring item in the acquired third monitoring data D3 from the maintenance management database 17 and the device management database 18, respectively. Then, the device information combining unit 121 generates combined device information by combining the read specification e and maintenance data f with basic device information in the third monitoring data D3. The device information combining unit 121 generates the fourth monitoring data D4 including the combined device information and used for extracting the feature amount of the combined device information simultaneously with the generation of the combined device information. The device information combining unit 121 outputs the generated fourth monitoring data D4 to the device information feature amount extracting unit 122.

  Specific examples of the fourth monitoring data D4 include the following D4_1 to D4_3. The following ei indicates a specification corresponding to ai having the same i. fi indicates maintenance data corresponding to ai having the same i. For example, e1 below is a specification of the rotating body a1. f1 is an operation history of the rotating body a1. The basic device information + e1 + f1 of the rotating body a1 included in D3_1 is combined device information of the rotating body a1.

D4_1: D3_1 + specification e1 + operation history f1
D4_2: D3_2 + specification e2 + driving history f2
D4_3: D3_3 + specification e3 + operation history f3

(Equipment information feature amount extraction unit 122)
The device information feature amount extraction unit 122 uses the combined device information in the fourth monitoring data D4 input from the device information combination unit 121, based on the combined device information feature amount shown in FIG. 2 (hereinafter referred to as device feature amount). Extract g.

  Here, the device feature amount g is, for example, a value obtained by quantifying the combined device information with a dimensionless parameter so that it can be identified and classified with a certain threshold. For example, the device feature amount g may be configured based on a plurality of classification condition items obtained by dividing the specification and the operation history by a range of a certain value (for example, a range of dimensions, a range of total operation time, etc.). In this case, among the classification condition items, an indicator that the combined device information is applicable may be given to the classification condition items to which the specification and the operation history in the combined device information correspond. The classification condition item may be set by statistical processing, a statistical model, or the like. Further, the device feature amount g may include the presence or absence of modification such as support addition to the monitoring target device.

  The device information feature amount extraction unit 122 generates the 4-2 monitoring data D4-2 by associating the extracted device feature amount g with the fourth monitoring data D4. The device information feature amount extraction unit 122 outputs the generated 4-2 monitoring data D4-2 to the device feature determination unit 123.

  Specific examples of the 4-2 monitoring data D4-2 include the following D4-2_1 to D4-2_3. The following gi indicates a device feature amount corresponding to ai having the same i. For example, g1 below is a device feature amount of the rotating body a1.

D4-2_1: D4_1 + device feature amount g1
D4-2_2: D4_2 + device feature amount g2
D4-2_3: D4_3 + device feature amount g3

  In addition, the device information feature amount extraction unit 122 outputs the extracted device feature amount g to the device feature database 19. The device feature database 19 stores the device feature pattern number h shown in FIG. 2 so as to correspond to the device feature amount g. The correspondence relationship between the device feature pattern number h and the device feature amount g in the device feature database 19 is based on knowledge obtained from experimental results for a plurality of devices. The device feature database 19 identifies the device feature amount g input from the device information feature amount extraction unit 122 as a pattern of the device feature amount g, that is, pattern identification. The device feature database 19 extracts a device feature pattern number h corresponding to the device feature amount g, and outputs the extracted device feature pattern number h to the device feature determination unit 123. The operation of the device feature database 19 may actually be performed by a computer that manages the device feature database 19.

(Equipment feature determination unit 123)
The device feature determination unit 123 associates the device feature pattern number h input from the device feature database 19 with the 4-2 monitoring data D4-2 input from the device information feature amount extraction unit 122. That is, the device feature determination unit 123 associates the device feature amount g and the device feature pattern number h with the fourth monitoring data D4. With this association, the device feature determination unit 123 generates the fifth monitoring data D5 used for selecting the diagnosis pattern. The device feature determination unit 123 outputs the generated fifth monitoring data D5 to the device state diagnosis unit 13.

  Specific examples of the fifth monitoring data D5 include the following D5_1 to D5_3. The following hi indicates a device feature pattern number corresponding to gi having the same i. For example, h1 below is a device feature pattern number obtained by pattern-identifying the device feature amount g1 of the rotating body a1.

D5_1: D4_1 + device feature quantity g1 + device feature pattern number h1
D5_2: D4_2 + device feature amount g2 + device feature pattern number h2
D5_3: D4_3 + device feature amount g3 + device feature pattern number h3

  As described above, the device information classifying unit 12 acquires the combined device information related to the preset monitoring item a including the basic device information of the monitoring target device, and classifies the acquired combined device information. The device information classifying unit 12 uses information obtained by combining the specifications of the monitoring target device and maintenance data stored in advance with the basic device information as the combined device information. In addition, the device information classification unit 12 acquires the device feature amount g, and classifies the combined device information into any of a plurality of device feature patterns h stored in advance based on the acquired device feature amount g. According to the device information classifying unit 12 configured as described above, devices with similar operating conditions can be classified into the same device feature pattern. Therefore, devices with similar operating conditions are identified with the same diagnostic pattern by the device state diagnosis unit 13 described later. Can be diagnosed by.

(Equipment status diagnosis unit 13)
The device state diagnosis unit 13 includes a diagnosis pattern determination unit 131, a device diagnosis unit 132, and a diagnosis result verification unit 133.

(Diagnostic pattern determination unit 131)
The diagnosis pattern determination unit 131 acquires the fifth monitoring data D5 input from the device feature determination unit 123. The diagnostic pattern determination unit 131 creates a diagnostic pattern selection value a + h including the device feature pattern number h and the monitoring item a in the acquired fifth monitoring data D5. The diagnostic pattern determination unit 131 outputs the created diagnostic pattern selection value a + h to the diagnostic pattern database 101. Further, the diagnosis pattern determination unit 131 outputs the fifth monitoring data D5 to the device diagnosis unit 132.

Specific examples of the diagnostic pattern selection value a + h include the following.
Selection value corresponding to D5_1: a1 + h1
Selection value corresponding to D5_2: a2 + h2
Selection value corresponding to D5_3: a3 + h3

  In the diagnostic pattern database 101, a plurality of diagnostic patterns (diagnostic pattern files) i of analysis result feature amounts shown in FIG. 2 are stored in advance in a state of being classified for each of a plurality of diagnostic pattern selection values a + h. Since the diagnostic pattern selection value a + h varies depending on the device feature pattern number h, it can be said that the diagnostic pattern i is classified and stored for each device feature pattern. The diagnostic pattern database 101 selects a diagnostic pattern i corresponding to the diagnostic pattern selection value a + h input from the diagnostic pattern determination unit 131. The diagnostic pattern database 101 outputs the selected diagnostic pattern i to the device diagnostic unit 132. The operation of the diagnostic pattern database 101 may actually be performed by a computer that manages the diagnostic pattern database 101.

(Device diagnosis unit 132)
The device diagnosis unit 132 applies the diagnosis pattern i input from the diagnosis pattern database 101 to the analysis result feature amount d in the fifth monitoring data D5 input from the diagnosis pattern determination unit 131, so that the monitoring target device Diagnose the condition. With this diagnosis, the device diagnosis unit 132 acquires the diagnosis result j shown in FIG. The device diagnosis unit 132 associates the acquired diagnosis result j with the fifth monitoring data D5 together with the diagnosis pattern i, thereby generating sixth monitoring data D6 as diagnosis result data. The sixth monitoring data D6 is monitoring data used for verification of the diagnosis result j. The device diagnosis unit 132 outputs the generated sixth monitoring data D6 to the diagnosis result verification unit 133.

  Specific examples of the sixth monitoring data D6 include the following D6_1 to D6_3. The following ii indicates a diagnostic pattern corresponding to the analysis result feature quantity di having the same suffix i. For example, the following i1 is a diagnostic pattern of the frequency value d1 of the spectrum peak. The following ji indicates a diagnosis result corresponding to the analysis result feature quantity di having the same i. For example, j1 below is a diagnosis result of the frequency value d1 of the spectrum peak.

D6_1: D5_1 + diagnosis pattern i1 + diagnosis result j1
D6_2: D5_2 + diagnostic pattern i2 + diagnostic result j2
D6_3: D5_3 + diagnosis pattern i3 + diagnosis result j3

(Diagnosis result verification unit 133)
The diagnosis result verification unit 133 verifies the correctness of the diagnosis result j based on the diagnosis result j in the sixth monitoring data D6 input from the device diagnosis unit 132. The diagnosis result verification unit 133 may also verify the validity of the device feature pattern number h and the diagnosis pattern i used for the diagnosis. An input unit 16 capable of correcting the diagnosis result j is attached to the diagnosis result verification unit 133. If the diagnostic result verification unit 133 determines that the diagnostic result j is incorrect as a result of the verification, the diagnostic result verification unit 133 corrects the diagnostic result j based on the input information of the input unit 16. The diagnostic result verification unit 133 outputs the verified diagnostic result j to the output unit 14, the maintenance management database 17, and the learning model generation unit 15.

  As described above, according to the device state diagnosis unit 13, the device feature pattern h classified by the device information classification unit 12 among the plurality of diagnosis patterns i previously classified and stored for each of the plurality of device feature patterns h. Diagnosis can be performed using the corresponding diagnosis pattern i.

  Here, if a diagnosis is performed in consideration of only characteristics unique to each monitored device, it is necessary to create a diagnosis model for diagnosis from abnormal state data of various actual machines. However, since each diagnostic model created from abnormal state data of various actual machines is a model specialized for each monitored device, it is difficult to apply to the diagnosis of other monitored devices. Also, creating a diagnostic model from a variety of actual machine abnormal state data requires a great deal of labor.

  On the other hand, according to the present embodiment, by applying a diagnosis pattern classified for each device feature pattern, a plurality of devices that are considered to have close operating conditions can be diagnosed with high accuracy using a common diagnostic model. It becomes possible to do. Also, it is possible to reduce the labor for creating a diagnostic pattern.

(Learning model generation unit 15)
When the diagnosis result j after verification input from the diagnosis result verification unit 133 is the corrected diagnosis result j, the learning model generation unit 15 creates a new diagnosis pattern i reflecting the corrected diagnosis result j. Create Creation of a new diagnostic pattern i may involve creation of a new diagnostic pattern selection value corresponding to the new diagnostic pattern i, that is, a new device feature pattern number h. The learning model generation unit 15 updates the diagnostic pattern i by registering the created new diagnostic pattern i in the diagnostic pattern database 101. In addition, the learning model generation unit 15 updates the device feature pattern number h by registering the created new device feature pattern number h in the device feature database 19.

  According to the learning model generation unit 15, the diagnosis pattern i and the device feature pattern number h can be improved so that a diagnosis pattern that more accurately reflects the state of the monitored device is applied. Thereby, diagnosis with higher accuracy is possible. As the measurement data is accumulated, the chance of updating the device feature pattern number h and the diagnostic pattern i increases, so that the diagnostic accuracy can be further improved.

(Function)
FIG. 3 is a flowchart of the diagnostic method showing the first embodiment. The following diagnostic method may be performed by causing the computer to execute a diagnostic program for causing the computer to execute the following procedure (steps). FIG. 4 is an explanatory diagram for explaining the operation of the device feature database 19 and the device information classification unit 12. FIG. 5 is an explanatory diagram for explaining the operation of the device state diagnosis unit 13. Hereinafter, a diagnostic method to which the diagnostic system 1 of FIG. 1 is applied will be described with reference to FIGS.

(Step S1)
First, in step S1 of FIG. 3, the data processing unit 111 analyzes the measurement data b in the input first monitoring data D1 (a + b), and acquires the analysis result c. The data processing unit 111 performs, for example, an order analysis process, a vector analysis process, a crest factor calculation process, a skewness calculation process, a kurtosis calculation process, and the like on the measurement data b, so that the dimensional / non-dimensional parameter c Is calculated. The data processing unit 111 generates the second monitoring data D2 by associating the calculated analysis result c with the first monitoring data D1. Then, the data processing unit 111 outputs the generated second monitoring data D2 to the analysis data feature amount extraction unit 112.

(Step S2)
Next, in step S2, the analysis data feature quantity extraction unit 112 extracts the analysis result feature quantity d from the analysis result c in the second monitoring data D2 input from the data processing unit 111. For example, when the vibration data is subjected to frequency analysis, the analysis data feature amount extraction unit 112 extracts a frequency value d indicating a peak from a vibration waveform having one or a plurality of frequency spectrum peaks. The analysis data feature quantity extraction unit 112 generates the third monitoring data D3 by associating the extracted analysis result feature quantity d with the second monitoring data D2. Then, the analysis data feature amount extraction unit 112 outputs the generated third monitoring data D3 to the device information combination unit 121.

(Step S3)
Next, in step S3, the device information combining unit 121 adds the specification e read from the device management database 18 to the basic device information in the third monitoring data D3 input from the analysis data feature amount extraction unit 112, and the maintenance management database. The maintenance data f read from 17 is combined. Thereby, the device information combining unit 121 generates the fourth monitoring data D4 at the same time as generating the combined device information. Then, the device information combination unit 121 outputs the generated fourth monitoring data D4 to the device information feature amount extraction unit 122.

(Step S4)
Next, in step S4, the device information feature amount extraction unit 122 extracts a device feature amount g from the combined device information in the fourth monitoring data D4 input from the device information combination unit 121. The device information feature amount extraction unit 122 outputs the extracted device feature amount g to the device feature database 19. In addition, the device information feature amount extraction unit 122 outputs the 4-2 monitoring data D4-2 in which the extracted device feature amount g is associated with the fourth monitoring data D4 to the device feature determination unit 123.

(Step S5)
Next, in step S5, the device feature database 19 identifies the device feature amount g input from the device information feature amount extraction unit 122 as, for example, a pattern corresponding to the description of the classification condition item described below.

  In FIG. 4, the device feature amount g input to the device feature database 19 is shown in a table format. The device feature amount g in FIG. 4 includes an index described in an item corresponding to the device feature amount g among the classification condition items (specifications 1 to 3 and operation conditions 1 to 3) set within a certain value range. It has become a thing. The index in FIG. 4 is a circle mark for convenience of explanation, but actually it may be data such as a flag. For example, specifications 1 to 3 in FIG. 4 may indicate ranges of different dimensions and structures (pump output and the like) of the devices. Moreover, the operating conditions 1-3 in FIG. 4 may show the range of the total operation time from which an apparatus mutually differs. The classification condition item may be set by, for example, statistical processing such as a linear discriminant function or Mahalanobis distance, a statistical model, or the like for each piece of connected device information.

  In FIG. 4, a plurality of device feature pattern numbers h stored in the device feature database 19 are shown in a table format. As shown in FIG. 4, each device feature pattern number h is associated with each device feature amount g. The device feature pattern number h may be tabular table data as shown in FIG. The device feature database 19 compares the input device feature amount g with each device feature amount g associated with each device feature pattern number h. The device feature database 19 identifies the device feature amount g by detecting the device feature pattern number h that matches the input device feature amount g.

(Step S6)
Next, in step S6, the device feature database 19 extracts a device feature pattern number h corresponding to the device feature amount g identified as a pattern. For example, as shown in FIG. 4, “000-0005” is extracted as the device feature pattern number h for the device feature quantity g corresponding to the specifications 2, 3 and operating conditions 3. The device feature database 19 outputs the extracted device feature pattern number h to the device feature determination unit 123. The device feature pattern number h serves as an index when selecting an identification pattern to be used when identifying / diagnosing various device operation state data included in the monitoring data.

(Step S7)
Next, in step S7, the device feature determination unit 123 associates the device feature pattern number h input from the device feature database 19 with the 4-2 monitoring data D4-2 input from the device information feature amount extraction unit 122. Thus, the fifth monitoring data D5 is generated. Then, the device feature determination unit 123 outputs the generated fifth monitoring data D5 to the diagnosis pattern determination unit 131.

(Step S8)
Next, in step S8, the diagnostic pattern determination unit 131 creates a diagnostic pattern selection value a + h composed of the device feature pattern number h and the monitoring item a in the fifth monitoring data D5 based on the fifth monitoring data D5. . Then, the diagnostic pattern determination unit 131 outputs the created diagnostic pattern selection value a + h to the diagnostic pattern database 101.

(Step S9)
Next, the diagnostic pattern database 10 selects the diagnostic pattern i based on the diagnostic pattern selection value a + h input from the diagnostic pattern determination unit 131 in step S9. Then, the diagnosis pattern database 101 outputs the selected diagnosis pattern i to the device diagnosis unit 132.

  Here, in FIG. 5, a plurality of diagnostic patterns i stored in the diagnostic pattern database 101 are illustrated in a folder form. Each diagnostic pattern i is given a unique diagnostic pattern selection value a + h as an attribute.

  For example, the device feature pattern number h in the diagnostic pattern selection value a + h is “000-0005” shown in FIG. 5, and the monitoring item a in the diagnostic pattern selection value a + h is “AAA-BBBC” shown in FIG. In some cases, the diagnostic pattern selection value a + h is “000-0005-AAA-BBBC”. When the diagnosis pattern selection value “000-0005-AAA-BBBC” is input, the diagnosis pattern database 101 selects only one diagnosis pattern i having an attribute that matches this value and outputs it to the device diagnosis unit 132. To do.

(Step S10)
Next, the device diagnosis unit 132 applies the diagnosis pattern i input from the diagnosis pattern database 101 to the analysis result feature quantity d in the fifth monitoring data D5 input from the diagnosis pattern determination unit 131, so that the monitoring target Diagnose the operating status of the equipment. And the apparatus diagnostic part 132 produces | generates the 6th monitoring data D6 by associating the diagnostic result j with the 5th monitoring data D5 with the diagnostic pattern i. Then, the device diagnosis unit 132 outputs the generated sixth monitoring data D6 to the diagnosis result verification unit 133.

Here, in FIG. 5, the diagnostic pattern i corresponding to the diagnostic pattern selection value “000-0005-AAA-BBBC” is shown in a table format. The actual diagnosis pattern i may also be tabular table data. The diagnostic pattern i in FIG. 5 shows different events for each pattern of the analysis result feature quantity d. The analysis result feature quantity d of FIG. 5 is divided into items for each parameter 1 to 5 to be extracted as the analysis result feature quantity d if the state of the monitored device is normal. Each item describes an indicator that the corresponding parameter is actually extracted as the analysis result feature amount d. This index is a circle for convenience in FIG. 5, but may actually be data such as a flag.

  For example, when the analysis result feature amount d is a spectrum peak of vibration data of a rotating body such as a shaft system device, the parameters 1 to 5 may be values obtained by multiplying the rotation frequency Fr of the rotating body by different constants. Good. Specifically, parameter 1 is (1/3) × Fr, parameter 2 is (1/2) × Fr, parameter 3 is 1 × Fr, parameter 4 is 2 × Fr, parameter 5 is 3 × Fr,... In this case, if the operating state of the rotating body is normal, any frequency that is a constant multiple of the rotational frequency Fr is extracted as a spectrum peak as the analysis result feature quantity d. On the other hand, when the operating state of the rotating body is not normal, for example, a frequency pattern of a constant multiple different from each other is not extracted as a spectrum peak for each event such as lack of lubricating oil, loosening, rattling or unbalance.

  In the diagnosis pattern i of FIG. 5, for example, when the analysis result feature amount d appears in the parameter 2, the parameter 3, and the parameter 6, the device diagnosis unit 132 sets the state of the monitoring target device to event 4. Diagnose. In this way, the diagnosis result j of the analysis result feature quantity d is obtained.

(Step S11)
Next, in step S11, the diagnosis result verification unit 133 verifies the correctness of the diagnosis result j in the sixth monitoring data D6 output from the device diagnosis unit 132. The diagnosis result verification unit 133 also verifies the validity of the device feature pattern number h and the diagnosis pattern i in the sixth monitoring data D6. The verification may be performed, for example, by a statistical method based on probability or deviation. For example, the diagnosis result verification unit 133 calculates the Mahalanobis distance between the analysis result feature quantity d and the diagnosis pattern i, and when the calculated Mahalanobis distance exceeds a certain threshold, the diagnosis result by the input unit 16 May be accepted. The diagnostic result verification unit 133 outputs the verified diagnostic result to the output unit 14, the maintenance management database 17, and the learning model generation unit 15. At this time, the diagnostic result verification unit 133 also outputs the measurement data b and the analysis result c to the maintenance management database 17.

(Step S12)
Next, in step S12, when the diagnosis result input from the diagnosis result verification unit 133 is a corrected diagnosis result, the learning model generation unit 15 performs a new diagnosis based on the corrected diagnosis result. Create pattern i. Then, the learning model generation unit 15 updates the device feature database 19 and the diagnostic pattern database 101 with the created new diagnostic pattern i. For example, the learning model generation unit 15 may correct the item of the device feature amount g in the device feature pattern number h shown in FIG. 4, and the event (abnormality) in the diagnosis pattern i shown in FIG. Classification items may be modified.

  As described above, according to the present embodiment, as shown in FIG. 4, the device feature value is applied by applying the device feature value to the classification condition item obtained by dividing the specification and the operation condition of the monitored device with a certain threshold value. Patterns can be identified. As a result, it is possible to classify which monitored equipment is operating under operating conditions similar to which other equipment. For example, if two equipment with exactly the same specifications are in operation, and if the operating time after the overhaul is different, it is determined that the equipment is characterized by two equipment with different states. Become.

  By managing devices with similar specifications and operating conditions as a group under a common device feature pattern number, as shown in FIG. 5, using the diagnostic pattern that matches the device feature pattern number, It is possible to efficiently perform highly accurate diagnosis reflecting the above. That is, according to the present embodiment, various knowledge obtained by a plurality of devices can be applied to the diagnosis of the device state, and at the same time, the facility device can be diagnosed based on the abnormal data of the actual device.

(Second Embodiment)
Next, a second embodiment will be described. In the description of the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

(Constitution)
FIG. 6 is a block diagram of the diagnostic system 1 showing the second embodiment. In the diagnostic system 1 of the second embodiment, the maintenance management database 17, the equipment management database 18, the equipment feature database 19, and the diagnostic pattern database 101 of FIG. 1 are unified as one plant database 102. That is, the plant database 102 provides specifications and maintenance data to the device information combining unit 121. Further, the plant database 102 provides a device feature pattern number h to a device feature identification unit 1232 described later. The plant database 102 provides a diagnostic pattern i to a diagnostic pattern selection unit 1312 described later.

  Further, the diagnostic system 1 of the second embodiment includes a device feature identification unit 1232 instead of the device feature determination unit 123 of FIG. The configuration of the device feature identification unit 1232 may be basically the same as that of the device feature determination unit 123 except that the device feature pattern number h stored in the plant database 102 is referred to.

  The diagnostic system 1 of the second embodiment includes a diagnostic pattern selection unit 1312 instead of the diagnostic pattern determination unit 131 of FIG. The configuration of the diagnosis pattern selection unit 1312 may be basically the same as that of the diagnosis pattern determination unit 131 except that the diagnosis pattern i stored in the plant database 102 is referred to.

  In the diagnosis system 1 of the second embodiment, the input unit 16 and the output unit 14 are provided in the plant database 102, respectively. The plant database 102 can correct the diagnosis result based on the input information of the input unit 16 and output the diagnosis result to the output unit 14.

(Function)
FIG. 7 is a flowchart of the diagnostic method showing the second embodiment. The flowchart of FIG. 7 differs from the flowchart of FIG. 3 in that step S21 is added between step S5 and step S6, and step S22 is added between step S8 and step S9.

  In step S21, the device feature identification unit 1232 refers to the device feature pattern number h stored in the plant database 102, and in subsequent step S6, extracts the device feature pattern number h corresponding to the pattern of the device feature amount g.

  In step S22, the diagnostic pattern selection unit 1312 refers to the diagnostic pattern i stored in the plant database 102, and in the subsequent step S9, selects the diagnostic pattern i based on the diagnostic pattern selection value.

  According to the second embodiment, various databases are integrated and managed by the plant database 102, thereby facilitating information management such as information relating to diagnosis of equipment and operation of diagnosis results. Other effects may be the same as those of the first embodiment.

(Third embodiment)
Next, a third embodiment will be described. In the description of the third embodiment, the same components as those in the first and second embodiments are denoted by the same reference numerals, and detailed description thereof is omitted.
(Constitution)
FIG. 8 is a block diagram of the diagnostic system 1 showing the third embodiment. The diagnosis system 1 of the third embodiment is different from the diagnosis system 1 of the second embodiment in that the measurement data analysis unit 11, the device information combination unit 121, the device information feature amount extraction unit 122, the diagnosis result verification unit 133, and The difference lies in the configuration except for the learning model generation unit 15.

  Moreover, in the diagnostic system 1 of 3rd Embodiment, the plant database 102 has memorize | stored operation performance data. Here, the operation result data is data indicating a past diagnosis result of the state of the monitored device. The operation result data includes a diagnosis result j, a monitoring item a used for obtaining the diagnosis result j, measurement data b, an analysis result c, an analysis result feature amount d, coupled device information, and a device feature amount g. The driving record data is associated with time information. The plant database 102 provides the operation result data D30 to the device feature identification unit 1232.

  The plant database 102 stores the latest equipment feature pattern number h and the latest diagnostic pattern i.

(Function)
FIG. 9 is a flowchart of the diagnostic method showing the third embodiment. The flowchart of FIG. 9 differs from the flowchart of FIG. 7 in that steps S1 to S4, 11, and 12 are omitted. The reason why Steps S1 to S4 are omitted is that the information acquired in these steps can be acquired as the operation result data D30. The reason why Steps S11 and S12 are omitted is that the plant database 102 is so enriched that it does not need to be updated based on the diagnosis result of the monitored device.

  According to the third embodiment, it is possible to re-diagnose past diagnosis results acquired before acquisition of the information h and i using the latest device feature pattern number h and diagnosis pattern i. Become. Thus, for example, when certain diagnostic information related to a major failure is obtained, a device that is operating under conditions close to the operating condition in which the major failure occurs by performing a diagnosis again on the past diagnosis result It becomes possible to re-evaluate the state.

  At least a part of the diagnosis system described in the above-described embodiment may be configured by hardware or software. When configured by software, a program for realizing at least a part of the functions of the diagnostic system may be stored in a recording medium such as a flexible disk or a CD-ROM, and read and executed by a computer. The recording medium is not limited to a removable medium such as a magnetic disk or an optical disk, but may be a fixed recording medium such as a hard disk device or a memory. A program that realizes at least a part of the functions of the diagnostic system may be distributed via a communication line (including wireless communication) such as the Internet. Further, the program may be distributed in a state where the program is encrypted, modulated or compressed, and stored in a recording medium via a wired line such as the Internet or a wireless line.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1 Diagnosis system 11 Measurement data analysis part 111 Data processing part 112 Analysis data feature-value extraction part 12 Device information classification | category part 121 Equipment information coupling | bond part 122 Equipment information feature-value extraction part 123 Equipment feature determination part 13 Equipment state diagnosis part
131 Diagnosis pattern determination unit 132 Device diagnosis unit 133 Diagnosis result verification unit 14 Output unit 15 Learning model generation unit 16 Input unit 17 Maintenance management database 18 Device management database 19 Device feature database 101 Diagnosis pattern database

Claims (12)

Relating to the monitoring item including the basic device information of the monitoring target device, obtaining the coupled device information obtained by coupling the fixed first coupled device information and the variable second coupled device information to the basic device information, A device information classification unit for classifying combined device information;
The feature amount of the analysis result for the measurement data corresponding to the monitoring item is acquired, and the state of the monitoring target device is determined using the diagnosis pattern of the feature amount of the analysis result corresponding to the classification result of the device information classification unit. A device condition diagnosis unit for diagnosis,
The device information classification unit acquires a feature amount of the combined device information, and classifies the combined device information into any of a plurality of pre-stored device feature patterns based on the feature amount of the combined device information. , Diagnostic system.   The device state diagnosis unit uses a diagnosis pattern corresponding to the device feature pattern classified by the device information classification unit among the plurality of diagnosis patterns previously classified and stored for each of the plurality of device feature patterns, The diagnosis system according to claim 1, which diagnoses a state of the monitoring target device.   The diagnostic system according to claim 1, wherein the first coupled device information includes a specification of the monitored device, and the second coupled device information includes maintenance data of the monitored device.   The diagnosis system according to claim 2, further comprising a learning model generation unit that acquires a diagnosis result of the device state diagnosis unit and updates at least one of the diagnosis pattern and the device feature pattern based on the diagnosis result. . The device state diagnosis unit verifies the correctness of the diagnosis result, and corrects the diagnosis result based on input information of the input unit when the diagnosis result is incorrect,
The diagnosis system according to claim 4, wherein the learning model generation unit updates at least one of the diagnosis pattern and the device feature pattern based on the diagnosis result corrected by the device state diagnosis unit.   The diagnostic system according to claim 1, further comprising a measurement data analysis unit that analyzes the measurement data and extracts a feature amount from the analysis result. The measurement data analysis unit acquires first monitoring data used for analysis including the monitoring items and the measurement data, analyzes the measurement data in the first monitoring data, and analyzes an analysis result as the first monitoring data. To generate second monitoring data used for extracting the feature value of the analysis result, extract the feature value of the analysis result from the analysis result in the second monitoring data, and determine the feature value of the analysis result Generating third monitoring data used for combining device information by associating with the second monitoring data;
The device information classifying unit includes the combined device information by combining the first coupled device information and the second coupled device information with the basic device information in the third monitoring data. Generating fourth monitoring data used for extracting feature amounts of information, extracting feature amounts of the combined device information from the combined device information in the fourth monitoring data, and extracting the feature amounts of the combined device information; By associating with the fourth monitoring data the device feature pattern number that pattern-identifies the feature amount of the device information, the fifth monitoring data used for selecting the diagnostic pattern is generated,
The device state diagnosis unit creates a diagnosis pattern selection value including the device feature pattern number and the monitoring item in the fifth monitoring data, and acquires the diagnosis pattern selected based on the diagnosis pattern selection value The diagnosis pattern is applied to the feature value of the analysis result in the fifth monitoring data to diagnose the state of the monitoring target device, and the diagnosis result is associated with the fifth monitoring data together with the diagnosis pattern. The diagnostic system of claim 6, wherein the diagnostic system generates data. A first management database for managing the first coupled device information;
A second management database for managing the second coupled device information;
A device feature database that stores the device feature pattern number so as to correspond to the feature amount of the combined device information, and outputs the device feature pattern number corresponding to the feature amount of the combined device information to the device information classification unit;
A diagnostic pattern database for storing the diagnostic pattern corresponding to the diagnostic pattern selection value, and outputting the diagnostic pattern corresponding to the diagnostic pattern selection value created by the device status diagnostic unit to the device information diagnostic unit; The diagnostic system according to claim 7, further comprising:   The diagnostic system according to claim 8, wherein the first management database, the second management database, the device feature database, and the diagnostic pattern database are integrated into one. A plant database,
The plant database includes the monitoring item, the measurement data, the analysis result, the feature amount of the analysis result, the combined device information, the feature amount of the combined device information, and a diagnostic result. While storing as past operation results data indicating the past diagnosis results of the state of the target device, the latest device feature pattern and diagnosis pattern are stored,
The diagnosis system according to claim 2, wherein the device state diagnosis unit re-diagnoses the state of the monitoring target device indicated in the operation result data using the latest device feature pattern and diagnosis pattern. The diagnostic system relates to the monitoring item including the basic device information of the device to be monitored, and combines the coupled device information obtained by combining the fixed first coupled device information and the variable second coupled device information with the basic device information. Obtaining and classifying the combined device information; and
The diagnostic system acquires the feature value of the analysis result for the measurement data corresponding to the monitoring item, and diagnoses the state of the monitored device as the feature value of the analysis result corresponding to the classification result of the combined device information Diagnosing using a pattern, and
In the classifying step, the feature amount of the combined device information is acquired, and the combined device information is classified into any of a plurality of pre-stored device feature patterns based on the feature amount of the combined device information. Diagnosis method. On the computer,
Relating to the monitoring item including the basic device information of the monitoring target device, obtaining the coupled device information obtained by coupling the fixed first coupled device information and the variable second coupled device information to the basic device information, The procedure for classifying the coupling device information,
A feature amount of an analysis result obtained by analyzing measurement data corresponding to the monitoring item is acquired, and a state of the monitoring target device is determined using a diagnosis pattern of the feature amount of the analysis result corresponding to the classification result of the combined device information. Run the procedure to diagnose,
The step of classifying includes acquiring a feature amount of the combined device information, and classifying the combined device information into one of a plurality of pre-stored device feature patterns based on the feature amount of the combined device information. Is a diagnostic program.

Images