JPH0868673A - Method for monitoring state of machine based on distributed measurement data - Google Patents

Abstract

PURPOSE: To monitor and diagnose the state of a machine easily and accurately by obtaining a three-dimensional (3D) profile model through mapping of a two-dimensional distributed measurement data and making a comparison with a past measurement data. CONSTITUTION: A measurement data is taken in, at first, along with measurement conditions and the arrangement of machines is displayed simultaneously by 3D CAD. 3D CAD data of machine is preset. Since an error is present in the measuring conditions, e.g. measuring position, edge is extracted from an image and the image of measurement data is registered with that of 3D CAD. A machine to be monitored is then selected and a corresponding measurement data is mapped onto the surface of 3D CAD thus displaying the parameter distribution of an objective machine. Since the measurement data is obtained only in one direction, operation for obtaining data in different direction is repeated as required. An abnormality decision is then made based on the parameter value thus determining a distributed measurement data required for diagnosis of machine intuitively based on the 3D profile of the machine.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for so-called equipment condition monitoring and abnormality determination in a plant, and more particularly to a method for condition monitoring using distributed measurement data from distributed equipment.

[0002]

2. Description of the Related Art In monitoring the condition of plant equipment, when monitoring the surface temperature and vibration distribution of the equipment by an infrared camera or a laser vibrometer, compared to conventional measurement at a predetermined point. Since it is possible to obtain a two-dimensional data distribution, it becomes possible to grasp the state of the device more finely. That is, if the size of the data at each point of the collected data is displayed in a shape like a contour line in correspondence with changes in color and brightness proportional to it,
The state of the target device can be monitored / diagnosed by capturing such a pattern of changes in color and brightness.

However, in such a two-dimensional measurement, since the measurement data is obtained in a two-dimensional area, not only the data on the target device but also its periphery may be measured. Therefore, it is necessary to perform a process of extracting the data of the target device from the image of the collected data distribution.

Since the target device is a three-dimensional object, it is clear which part of the device the collected data corresponds to, which part of the three-dimensional object should be measured, and which part should be further measured in the future. I don't know. Therefore, if the measurement data can be made to correspond to the shape data of the object, which is originally three-dimensional, the effect of grasping the state in monitoring the device becomes large.

Conventionally, in the case of monitoring a device state using data collected by a remote sensor, a technique for extracting necessary data from two-dimensional data, a technique for two-dimensional pattern recognition, and a target device Since the technology for associating with the dimensional shape has not been established, it was handled by manual interpretation. That is, by synthesizing the information obtained from the sensor into a two-dimensional image and presenting it to the observer, the observer distinguishes the necessary data from the image, and observes the distribution state of the parameters therein, This was handled by monitoring the status of the target device.

However, in the case of constantly monitoring the state of the equipment online, it is considered unrealistic for humans to intervene in the processes from the measurement of data by the sensor to the estimation and judgment of the equipment state based on the data. To be

The present invention has been made in consideration of the above points, and provides a method for monitoring the state of a device, which enables simple and appropriate monitoring and diagnosis of the state of the device based on two-dimensional distribution measurement data extracted from a target device. With the goal.

[0008]

To achieve the above object, in the present invention, two-dimensional distribution measurement data concerning the state of a device including a target device according to claim 1 is fetched, and the two-dimensional distribution measurement data and the device are acquired. The display result of the three-dimensional shape model is aligned and displayed in association with each other, the device portion to be inspected is designated from the displayed contents, and the three-dimensional shape model corresponding to the designated device portion is displayed. And a method of monitoring a device state by distributed measurement data for mapping the measurement data of the device and detecting a change in the measurement data to determine whether or not there is an abnormality in the device. That is, the distribution measurement data relating to the two-dimensional parameter and the display result of the inspection target device based on the three-dimensional shape model such as the three-dimensional CAD are overlapped by matching the characteristic points of the image, and the target device and the measurement thereof are performed. It correlates with the data and automatically extracts only the data necessary for monitoring the state and estimating the state. The extracted two-dimensional distribution measurement data is mapped to a three-dimensional shape model and displayed in a three-dimensional display and a two-dimensional space graph to show information in a more understandable manner, and the absolute value level and past measurement data are compared. By
Monitors the device status and judges abnormalities. Here, the device refers to one including parts.

According to a second aspect of the present invention, the position of the distribution image is coordinated with the three-dimensional model by designating the viewpoint coordinates, the viewpoint direction and the viewing angle.

According to the third aspect of the present invention, a two-dimensional visible image having a known relationship among a viewpoint position, a viewpoint direction and a viewing angle is used as an auxiliary screen for alignment, and the auxiliary screen and a three-dimensional shape model of a device. Align the position of.

According to the method of claim 4, the image and the device
The point to be matched in the three-dimensional shape model is designated on the image, the viewpoint coordinates, the viewpoint direction, and the viewing angle are calculated, and the distribution image and the position of the three-dimensional shape model of the device are matched.

According to the method of claim 5, a two-dimensional visible image having a known relationship with the viewpoint position, viewpoint direction and viewing angle at the time of distribution measurement is used as an auxiliary screen for alignment, and this screen and the device Match the position of the 3D shape model.

According to a sixth aspect of the present invention, the characteristic amount of the distribution measurement data is detected, and based on the detected characteristic amount, a point to be matched between the image and the three-dimensional shape model of the device is determined, and the viewpoint coordinate, the viewpoint direction and the viewing angle are determined. Calculate the distribution image and equipment
Align the position of the dimensional shape model.

According to the method of claim 7, a two-dimensional visible image whose relationship with the viewpoint position, viewpoint direction and viewing angle at the time of distribution measurement is known is used as an auxiliary screen for alignment, and this screen and the device Match the position of the 3D shape model.

According to the method of claim 8, the equipment to be inspected is designated from the piping instrumentation diagram displayed on the screen, and the equipment name corresponding to this is transferred to the three-dimensional shape model for inspection. Specify the device to be used.

In the method according to the ninth aspect, the equipment to be inspected is designated by the three-dimensional CAD diagram displayed on the screen.

In the method according to the tenth aspect, the equipment to be inspected is designated from the image corresponding to the three-dimensional shape model displayed on the screen.

In the method according to the eleventh aspect, the equipment to be inspected is specified from the distribution measurement data corresponding to the three-dimensional shape model displayed on the screen.

In the method according to the twelfth aspect, the equipment to be inspected is designated from the distribution measurement data corresponding to the three-dimensional shape model displayed on the screen.

In the method according to the thirteenth aspect, the equipment to be inspected is designated by the three-dimensional CAD diagram displayed on the screen.

In the method described in claim 14, the equipment to be inspected corresponds to the three-dimensional shape model displayed on the screen.
Specify from the three-dimensional image.

According to a fifteenth aspect of the present invention, a plurality of two-dimensional patterns are used as a three-dimensional shape model of a device for mapping measurement data.
A three-dimensional shape is generated from the three-dimensional image and used.

In the method according to the sixteenth aspect, a three-dimensional shape model of the device is generated from a plurality of two-dimensional images by the triangulation method.

In the method according to the seventeenth aspect, when the distribution measurement data is mapped to the three-dimensional shape model of the device, the data is mapped after correction by the reflectance of the target surface or the relative inclination of the surface.

In the method according to the eighteenth aspect, after the distribution measurement data is mapped to the three-dimensional shape model of the device, it is converted into a front view and a side view and displayed.

In the method according to the nineteenth aspect, after mapping the distribution measurement data to the three-dimensional shape model of the device, the parameters of the measurement data are displayed in a chart along the axial direction of the device.

In the method according to the twentieth aspect, when mapping the distribution measurement data to the three-dimensional shape model of the device, the measurement data of a plurality of screens are mapped and combined with the three-dimensional shape model of the device.

In the method according to the twenty-first aspect, the distribution measurement data is compared with the distribution measurement data measured last time, and an abnormality is determined based on these differences.

In the method according to the twenty-second aspect, the distribution measurement data is compared with a preset abnormality determination level to determine the abnormality.

In the method according to the twenty-fourth aspect, the correlation between the distribution measurement data and the past data is calculated to determine the abnormality.

In the method according to the twenty-fifth aspect, the abnormality determination is performed by combining two or more kinds of measurement data different from each other and determining the combined value.

[0032]

According to the above configuration, the two-dimensional parameter distribution is measured for each target device, and the shape of the inspection target device is simultaneously displayed by this distribution image and the three-dimensional shape model. The positions are aligned so that they overlap each other by the alignment, and the corresponding measurement data is mapped to the three-dimensional shape model. The mapped measurement data or the spatial distribution displayed in the graph is compared and determined with past data to monitor the state of the device and determine an abnormality.

According to the second to seventh aspects, the alignment of the distribution measurement data and the three-dimensional shape model is performed using the viewpoint coordinates, the viewpoint direction, the viewing angle, etc., obtained at the time of measuring the distribution measurement data.

According to the eighth to twelfth aspects, the designation of the device portion to be the target device is performed by the wiring instrumentation diagram, the three-dimensional CAD
This is performed using figures, two-dimensional visible images, and the like.

According to the thirteenth to fifteenth aspects, a three-dimensional CAD is used as an image for mapping the measurement data distribution image,
A two-dimensional image and a three-dimensional image generated from a plurality of two-dimensional images are used.

According to the sixteenth aspect, a three-dimensional model of the device generated from a plurality of two-dimensional images by the triangulation method is used.

According to the seventeenth to twentieth aspects, when mapping the measurement data to the three-dimensional shape model, the reflectance of the target surface and the like are taken into consideration, or the data is converted into a front view or the like, or the axial direction of the device is measured. The parameters of the measurement data are displayed in a chart along with, and the measurement data of multiple screens are combined and mapped.

According to the twenty-first to twenty-fifth aspects, the measured data is compared with the previous data, compared with a preset level, the rate of change is checked, the correlation with the past data is checked, and two types are used. Abnormality determination is performed using a value obtained by combining the above data.

[0039]

EXAMPLE An example of the present invention will be described with reference to FIG. FIG. 1 shows a procedure for surface measurement of vibration distribution data using a laser vibrometer in the present invention, and displaying the measurement result in a three-dimensional CAD by superimposing it. The superposition procedure is roughly divided into the following two parts, as shown in FIG.

1) Alignment of distribution data with three-dimensional CAD 2) Distribution mapping cutout mapping For alignment of distribution data with three-dimensional CAD, the measured two-dimensional parameter distribution is displayed as an image and displayed. The three-dimensional CAD data of the device is displayed in an overlapping manner to show the correspondence between the device and the image, which is performed as follows.

First, in step S1, the measurement data of the vibration distribution and the measurement conditions such as the measurement position at that time are fetched. The vibration distribution data is displayed as image data by converting the parameter value into color or brightness. At the same time, based on the measurement conditions such as the measurement point coordinates, the measurement direction and the viewing angle, the device arrangement is simultaneously displayed by the three-dimensional CAD in step S2. Three-dimensional CAD data regarding the device is created in advance. Since there is an error in the measurement conditions such as the measurement position, the edge of the image is extracted in step S3, and the image of the measurement data and the three-dimensional C in step S4 are extracted.
Align with AD.

Next, in the cut-out mapping of the distribution measurement data, the measurement data corresponding to the device for monitoring the selected state is cut out in step S5 and mapped in the surface of the three-dimensional CAD in step S6 to obtain the parameter distribution of the target device. indicate.

FIG. 2 shows a screen on which a target device to be inspected can be specified, that is, a plant instrumentation diagram (hereinafter referred to as P & ID).
Is displayed, a TV image of the device, a distribution measurement data image, and a three-dimensional CAD screen. Designate and select inspection equipment using these screens. By associating the P & ID with the three-dimensional CAD, the name and characteristics of the inspection device can be known on the spot. In addition, it is possible to directly specify a device to be viewed by a TV image or the like.

After the selection, the measurement data of the corresponding device is mapped to the selected device, and the inspection target device is displayed. Since the measurement data is in only one direction, if it is judged from the measurement data on the device to be inspected and data from another direction is required, these processes are repeated. Then, the abnormality determination based on the parameter value is performed in step S7. As a result, the distribution measurement data necessary for diagnosing the device can be intuitively known based on the three-dimensional shape of the device.

Next, details of the superposition algorithm will be described with reference to FIGS.

The three-dimensional CAD to be superimposed on the measurement data of the vibration distribution is capable of wire frame display and surface display. In order to display the outline of the shape of the device quickly, a simple wire frame display that shows the shape only with lines is suitable, and in order to display the shape of the device with some sense of reality, the outline of the shape is shown. Surface display is suitable. FIG. 3 shows a wire frame display of a three-dimensional CAD of a flange portion provided on a pipe as a simple example, and FIG. 4 shows a surface display thereof.

In both of these figures, since the three-dimensional CAD and the measurement data are displayed in an overlapping manner, the equipment and the measurement data can be associated with each other. At this time, three-dimensional C
Positioning is performed by adjusting the display condition of AD.

There are three methods of alignment, so-called 1) manual adjustment, 2) semi-automatic adjustment, and 3) automatic adjustment. -Positioning by manual adjustment Manual adjustment is a method of manually adjusting the display parameters of the figure by three-dimensional CAD and matching it with the measurement image. These display results can be moved by adjusting the measurement point and viewing angle parameters. By specifying the position of the slider displayed on the screen with the mouse, the x, y, z coordinates of the viewpoint and the measurement can be obtained. You can adjust the direction. -Registration by semi-automatic adjustment Semi-automatic adjustment is a method in which the representative points are manually specified and automatically converted for superposition. In the automatic adjustment, the representative points are extracted as image edges and feature points. It is a method of superimposing it completely automatically.

The semi-automatic adjustment method is such that, when the measurement data and the three-dimensional CAD are matched, a designated point to be matched is determined, and a parameter value for matching is determined by the least square method. If necessary, the edges of the image are extracted in order to make the contour of the measuring object clearer. The edge is extracted by applying a gradient or Sobel operator as shown in FIG. 5 to the display image. And the edge extraction is 3 × 3 of the two-dimensional measurement data.
The differential calculation is performed on the pixels of.

Step A-1 The image and the three-dimensional CAD are displayed simultaneously. If necessary, an edge extraction process of the image is performed so that the image and the three-dimensional CAD can be easily aligned with each other.

Step A-2 Three-dimensional CA to be combined
Pick a point on D and select it. The procedure is as follows.

1) Selective display of characteristic points (vertices appearing on the surface).

2) A feature point pi = (xi, yi, zi) to be matched is selected with a mouse or the like.

3) Save the coordinates of the selected point.

Step A-3 A point on the image to be matched is picked and selected with a mouse or the like. The procedure is as follows.

1) A point qi = (x'i, y'i) corresponding to the selected point Pi is selected with a mouse or the like.

2) A point to be selected is displayed on the image.

3) Save the coordinates of the selected point Qi.

Step A-4 Step A-2, Same A-
Repeat 3 for the required number of points (i = 1, 2, 3, ...,
N).

Step A-5 The viewpoint position vector v and the direction (θ, φ) as shown in FIG. 6 are determined so that the three-dimensional CAD and the image overlap each other. Where qi = PV
Calculate (v, θ, φ) such that (v, θ, φ) Mpi by the method of least squares. M, V, and P are matrices for modeling, viewing, and projection transformation, respectively.

Step A-6 The three-dimensional CAD is moved / converted and displayed again.

The calculation algorithm of the viewpoint and direction parameters in step A-5 by the method of least squares is shown below.

Step 5-1. Initial value X of viewpoint / direction
Enter o.

Xo T = [vo, θo, ψo] (1) where vo: viewpoint initial value (vxo, vyo, vxo) θo, ψo: initial value of viewpoint direction angle.

Here, the parameter to be estimated considers the position and direction of the viewpoint. The viewing angles (αx, αy) are fixed because they are unique data in the measuring instrument, but αx, αy
Even if y is unknown, it can be obtained in the same way as other parameters.

Step 5-2. Enter N pairs of points to match. Each point is defined as follows.

In the image coordinates, yi ^m = (xi ^m , yi ^m ) (2) In the three-dimensional CAD coordinates, xi = (xi, yi, zi), i = 1, 2, 3, ..., N (3) Step 5-3. By the least square method, the viewpoint / viewing angle vector X is changed by ΔX so as to minimize the following evaluation function.

[0068]

[Equation 1] Here, Wi is a weight matrix that represents an error in setting the position.

The calculation procedure is as follows according to the steepest descent method. Other methods may be used in consideration of convergence.

[0070]

[Equation 2] 2) Find the change ΔX of the vector X.

[0071]

(Equation 3) 3) Update vector X.

Xn + 1 = Xn + ΔX (7) 4) | The calculation of 1), 2) and 3) is executed until ΔX | <εo.

Here, the relationship between the point yi on the screen and the point xi on the three-dimensional space is expressed as follows by the viewpoint variable (viewing transformation) and the mapping transformation (perspective transformation) to the two-dimensional screen. It

[0074]

[Equation 4] However, Rx (ψ) is the rotation of the angle ψ about the X axis and Ry (θ) is the rotation of the angle θ about the Y axis.

Therefore, the gradient of equation (5) is

[0076]

(Equation 5) Can be calculated as here,

[0077]

(Equation 6) Is.

Step 5-4. When the converged Xn + 1 is obtained, the viewpoint direction Pn + 1 is obtained and the three-dimensional CAD is displayed again for confirmation.

Pn + 1 = (p′x, p′y, p′z) (12) Here, p′x = vx + cosψsin θ p′y = vy + sinψ (13) p′z = vz + cosψcosθ.

FIG. 7 shows the vibration distribution and the three-dimensional CAD display superimposed on each other in the vicinity of the valve of the pump. A two-dimensional image and a three-dimensional CAD can be superimposed by designating a feature such as a flange.

For the alignment of the two-dimensional image of the measurement data and the three-dimensional model of the device shape, the position and direction of the viewpoint are manually specified, and several points on the screen to be overlaid are designated to determine the position and direction of the viewpoint. A semi-automatic method for calculating and setting is shown. -Alignment by automatic adjustment Fig. 8 shows an algorithm for completely automatic alignment without human intervention.

As a procedure, first, a measurement image is input in step S11, its edge is extracted in step S12, a straight line is traced in step S13, and characteristic points are extracted in step S14.

The feature points include, for example, corners of a rectangular parallelepiped, as shown in FIGS. 9 (a), 9 (b) and 9 (c).

Returning to FIG. 8 again, in step S15, the similarity between the window including the feature points and the image created by the three-dimensional shape model is evaluated by the correlation coefficient, and in step S16, the image with the smallest correlation coefficient is selected. Judge as corresponding points. Then, in step S17, the viewpoint coordinates and the viewpoint position are calculated from the corresponding points of the image by the above-described method.

FIG. 10 shows another method for superimposing the measurement image and the three-dimensional shape model in FIGS. 8 and 9, in which the position with the measuring device, the measuring direction and the measuring viewing angle are known. The position of the TV camera is adjusted.

FIG. 11 shows the mapping display of the image data on the equipment to be inspected. The mapping display of the image data to the inspection target device is the three-dimensional CAD first.
The parts are selected above, and then the measurement data is mapped to the specified primitive.

The algorithm for component selection on the three-dimensional CAD is as shown in 1), 2) and 3) below.

1) Name each component so that it can be picked.

2) Once picked, blink the part.

3) When "confirmed", only that part is displayed.

Here, as described above, the P & ID screen, the TV
Allows specification from the image and measurement data screen.

The mapping of the image data to the inspection target device is performed by mapping the primitives of the three-dimensional CAD constituting the inspection target device as shown in FIG. For example, first, the vertices A, B, C, and D that form the surface of the primitive are converted into the space of the two-dimensional measurement image. The converted measurement data of the area surrounded by the vertices A ′, B ′, C ′ and D ′ is mapped onto the plane A-B-C-D. It is determined whether or not this plane is the front surface of the primitive. If it is the front surface, the image data is mapped to this plane, and if it is the back surface, it is not performed.
If it is hidden by another object, correct it.

The processing algorithm for this modification is summarized below.

1) Select only the surface of the primitives from the surface.

The numbers are given in order from the one on the selected surface (larger z value).

2) Convert the coordinates of the selected surface into image coordinates. The conversion is performed by the following equation.

P'i = PVMPi (14) where P'i: image coordinates = (x'i, y'i) Pi: three-dimensional CAD coordinates = (xi, yi, zi) M: modeling transformation matrix V: Viewing transformation matrix P: Perspective transformation matrix 3) The image is mapped from the upper surface (of the z value) to a polygon which is the minimum unit of three-dimensional CAD. After mapping, the mapped area of the image is "black" (shadow area color)
Apply to.

4) A texture is included for each primitive so that it can be saved and displayed again. As image data,
It is possible to save the mapped data and the image data before mapping.

FIG. 12 shows an example of mapping to a pipe in which cylindrical primitives are combined. By comparing the mapped image data with the past image data, the abnormality of the device can be judged from the difference. Further, it is possible to judge the abnormality of the device by reading the data value of the point of the device to be inspected from this image data and displaying the trend.

FIG. 13 is a diagram in which a plurality of images of a long object such as a pipe are continuously mapped on a three-dimensional CAD, and the distribution of the entire vibration measurement data can be displayed.

FIG. 14 shows the distribution of vibration measurement data.
By plotting the data along a specified direction of piping or equipment as an X-Y graph, the mode is displayed and the vibration characteristics are represented.

Further, although the above description is based on the projection drawing based on the three-dimensional shape, the three-dimensional shape model in which the measurement data is mapped may be represented by a three-sided view such as an upper surface, a front surface, and a side surface.

[Other Embodiments (1)] In the above embodiments,
The state monitoring and abnormality diagnosis method for vibration distribution data has been described, but here, as with the vibration distribution data, temperature measurement data from an infrared camera, visible image data from a TV camera, and sound source distribution data from a stereophonic sensor can also be used. Good. Since the processing contents of these data are basically the same when the value of the parameter is converted into the color or the brightness, the data can be processed in the same manner as the case of the vibration distribution.

[Other Embodiment (2)] Also, in the above embodiments, the three-dimensional CAD for expressing the shape of the equipment by the combination of basic figures as the model of the shape of the equipment to be inspected.
However, in order to express the shape, the actual shape of the device may be measured with a laser range finder and used as a model. The laser range finder, for example, as shown in FIG. 15, amplitude-modulates a continuous wave of a laser, outputs laser light to a target object, detects a phase difference from the reflected light from the target object, and detects the phase difference with the flight time. In other words, it is converted into distance information. By scanning the laser light on the object,
The shape of the object can be obtained.

By mapping the measurement data using this as a model of the device, the state of the device such as the vibration distribution can be easily grasped. In addition, in the method of creating a model using this laser light, data of three-dimensional CAD and the like are not required in advance, so that it becomes possible to flexibly model the device. Further, it is possible to input and display a detailed shape that cannot be expressed by the three-dimensional CAD.

FIG. 16 shows an example of stereoscopic vision in which a three-dimensional shape is measured by a stereoscopic vision method using two TV images. In stereoscopic view, as shown in FIG. 16, the distance to the point P is the corresponding point PL (X
By calculating L, YL) and PR (XR, YR), it is calculated by the following equation.

[0107]

(Equation 7) Here, f is the distance between the camera focus and the image.

For the corresponding points of the two images, the characteristic points of the images should be extracted and superposed on the window portion around these characteristic points, as in the case of superimposing the measurement image and the image by the three-dimensional shape model. Corresponding points are obtained by correlating the windows of the images.

According to this method, since the three-dimensional shape can be known only from the TV image, the shape can be obtained by an inexpensive device.

[Other Embodiments (3)] In the above embodiments,
Although the value measured by the detector does not depend on the inclination of the object to be measured or the reflectance of the surface, if the inclination or the reflectance of the surface is significantly different, it needs to be corrected.

FIG. 17 shows an example in which the object is vibrating in the direction perpendicular to the measurement surface and the inclination of the measurement surface is θ. In the case of the laser vibrometer, if the surface of the object to be measured is perpendicular to the direction of the laser, it is not necessary to correct it, but if the tilt is θ as in this case, the vertical component is cos θ times. The actual vibration speed Vnorm is corrected as follows.

Vnorm = Vmes / cos θ (16) where Vmes is the velocity of the measured vibration.

With this, the vibration of the object can be measured without irradiating the laser beam perpendicularly to the vibrating surface.

Further, in the case of measuring the temperature by the infrared sensor, the temperature can be accurately measured by correcting the reflectance of infrared rays. If the standard reflectance is FO and the reflectance of the object to be measured is Fobs, the reflectance-corrected measurement temperature is given as Tmod as follows.

Tmod = Tmes. (Fobs / FO) (17) where Tmes is the temperature (° C.) directly measured.

[Other Embodiments (4)] In the above embodiments, the abnormality is judged by using only one kind of measured value. However, even if it is judged by using a plurality of measured values here. Good. For example, when the abnormality is independently determined using only the amplitude and temperature of the vibration and when the abnormality is determined by combining the vibration and the temperature, the latter can make a more precise determination.

FIG. 18 illustrates and compares the criteria for judging these abnormalities. In the case of independently judging the amplitude x and the temperature T, x <a is a region showing the normality of the amplitude and T
Assuming that <b is a region showing normal temperature, “x <a
And T <b ”is a normal region judged by these two parameters. This is shown as the area inside the dotted line in FIG. On the other hand, the normal region can be defined based on the relationship between the amplitude indicating the normal value and the temperature.
For example, on the XT plane, a normal region of the current measured value (X, T) is found to have a relationship such that X / a1 + T / b1 <1, and normality / abnormality can be judged from this map.
As a result, the normal / abnormal determination region can be taken more precisely.

That is, when the respective parameters are independently determined, it is necessary to conservatively set the thresholds for abnormality determination, for example, a and b. On the other hand, in the case of determining the abnormality by obtaining the relationship of a plurality of parameters, it is possible to take a larger area than in the case of independently determining the normal region, and the erroneous determination of the abnormality and the wasteful inspection of the device are reduced, which is economical. Be advantageous

The parameters used for the determination can be treated in the same manner even if two or more parameters are used. Further, in the above, the relational expression of X / a1 + T / b1 <1 has been taken as an example, but this relational expression can be similarly used if it is a relational expression representing a normal region.

[0120]

As described above, according to the present invention, the distribution measurement data and the three-dimensional space model are aligned with each other and the distribution measurement data is mapped, so that the operating state of the device can be displayed in an easy-to-understand manner. As a result, it becomes possible to easily make an accurate judgment about the state of the device.
As a result, it is possible to save labor for monitoring the target device and determining an abnormality.

[Brief description of drawings]

FIG. 1 is a flowchart showing the configuration of a distribution data superposition algorithm of the present invention.

FIG. 2 is an explanatory diagram showing an example of a display screen for device selection.

FIG. 3 is an explanatory diagram showing a wire frame display example of a pump by three-dimensional CAD.

FIG. 4 is an explanatory diagram showing a surface display example of a pump by three-dimensional CAD.

FIG. 5 is an explanatory diagram showing a differential operation operator.

FIG. 6 is an explanatory diagram showing viewpoint parameters.

FIG. 7 is an explanatory diagram showing an example of superposed display of vibration distribution and three-dimensional CAD using a pump as an example.

FIG. 8 is a flowchart showing an automatic alignment procedure.

9 (a), 9 (b) and 9 (c) are explanatory views showing examples of characteristic points.

FIG. 10 is an explanatory diagram showing the relationship between the measuring device and the TV camera.

FIG. 11 is an explanatory diagram showing a mapping method.

FIG. 12 is an explanatory diagram of a mapping example of a vibration distribution image on a pipe.

FIG. 13 is an explanatory diagram of a mapping example of a vibration distribution image on a long pipe.

FIG. 14 is an explanatory diagram showing a vibration mode of piping.

FIG. 15 is an explanatory diagram showing a configuration of a laser range finder.

FIG. 16 is an explanatory diagram showing the principle of stereoscopic vision.

FIG. 17 is an explanatory diagram showing the relationship between the inclination of the object and the measurement direction.

FIG. 18 is an explanatory diagram showing the difference in abnormality determination criteria.

[Explanation of symbols]

 v Viewpoint position vector α Viewing angle A, B, C, D Vertices

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Riyoko Haneda 8 Shinsita-cho, Isogo-ku, Yokohama-shi, Kanagawa Toshiba Corporation Yokohama office Machi 1 Co., Ltd. Toshiba R & D Center (72) Inventor Shigeru Motomoto Komukai Toshiba-cho, Kawasaki-shi, Kanagawa 1 Kochi Mukai Toshiba R & D Center

Claims (25)

[Claims] 1. A two-dimensional distribution measurement data regarding a state of a device including a target device is fetched, and the two-dimensional distribution measurement data and a display result of a three-dimensional shape model of the device are aligned and superposed in association with each other. A device part to be inspected is displayed and displayed, the device part to be inspected is specified, the measurement data of the device is mapped on the three-dimensional shape model corresponding to the specified device part, and a change in the measurement data is detected to detect the device. A method for monitoring the device status using distributed measurement data that determines the presence or absence of abnormalities. 2. The method according to claim 1, wherein when the distribution measurement data and the three-dimensional shape model of the device are overlaid, the distribution image and the device of which the viewpoint coordinates, the viewpoint direction and the viewing angle at the time of the distribution measurement are designated are specified. A method for monitoring the device status based on the distribution measurement data by aligning the positions of the three-dimensional shape model. 3. The method according to claim 2, wherein when the distribution measurement data and the three-dimensional shape model of the device are superposed, the relationship between the viewpoint position, the viewpoint direction and the viewing angle at the time of the distribution measurement is known. A method of monitoring the device status by using distribution measurement data in which a three-dimensional visible image is used as an auxiliary screen for alignment and the position of this auxiliary screen and the three-dimensional model of the device are aligned. 4. The method according to claim 1, wherein in superimposing the distribution measurement data and the three-dimensional shape model of the device, a point to be matched between the image and the three-dimensional shape model of the device is designated on the image, and viewpoint coordinates are set. A method of monitoring the device status by calculating the viewpoint direction and the viewing angle and aligning the distribution image with the position of the three-dimensional shape model of the device based on the distribution measurement data. 5. The method according to claim 4, wherein when the distribution measurement data and the three-dimensional shape model of the device are superposed, the relationship between the viewpoint position, the viewpoint direction, and the viewing angle during the distribution measurement is known. A method for monitoring the device status by using distribution measurement data in which a three-dimensional visible image is used as an auxiliary screen for alignment and the position of this screen and the three-dimensional shape model of the device are aligned. 6. The method according to claim 1, wherein when the distribution measurement data and the three-dimensional shape model of the device are overlaid, the feature amount of the distribution measurement data is detected, and based on this, the image and the three-dimensional shape of the device are detected. A device state monitoring method using distribution measurement data in which points to be matched in the model are determined, viewpoint coordinates, viewpoint directions, and viewing angles are calculated, and the positions of the distribution image and the three-dimensional shape model of the device are matched. 7. The method according to claim 6, wherein when the distribution measurement data and the three-dimensional shape model of the device are superposed, the relationship between the viewpoint position, the viewpoint direction and the viewing angle at the time of the distribution measurement is known. A method for monitoring the device status by using distribution measurement data in which a three-dimensional visible image is used as an auxiliary screen for alignment and the position of this screen and the three-dimensional shape model of the device are aligned. 8. The method according to claim 1, wherein the equipment to be inspected is designated from the piping instrumentation diagram displayed on the screen, and the equipment name corresponding to this is transferred to the three-dimensional shape model. A method of monitoring the status of equipment with distributed measurement data that specifies the equipment to be inspected. 9. The method according to claim 1, wherein a device to be inspected is designated by a three-dimensional CAD diagram displayed on a screen, and the device state is monitored by distribution measurement data. 10. The method according to claim 1, wherein the equipment state to be inspected is specified by an image corresponding to the three-dimensional shape model displayed on the screen, and the equipment state is monitored by the distribution measurement data. 11. The method according to claim 1, wherein the equipment to be inspected is specified by the distribution measurement data corresponding to the three-dimensional shape model displayed on the screen. 12. The method according to claim 1, wherein the equipment to be inspected is designated from the distribution measurement data corresponding to the three-dimensional shape model displayed on the screen, and the equipment state is monitored by the distribution measurement data. 13. The method according to claim 1, wherein a device to be inspected is designated by a three-dimensional CAD diagram displayed on the screen. 14. The method according to claim 1, wherein a device to be inspected is designated by a two-dimensional image corresponding to a three-dimensional geometric model displayed on a screen. 15. The method according to claim 1, wherein a three-dimensional shape is generated from a plurality of two-dimensional images as a three-dimensional shape model of the equipment for mapping the measurement data, and the equipment state monitoring method using the distribution measurement data is used. . 16. The method according to claim 1, wherein the three-dimensional shape model of the device is divided into a plurality of two-dimensional shapes by a triangulation method.
A method for monitoring the device status using the distribution measurement data generated from a three-dimensional image. 17. The method according to claim 1, wherein when the distribution measurement data is mapped to a three-dimensional shape model of the device, the data is mapped after correction by the reflectance of the target surface or the relative inclination of the surface. A method for monitoring equipment status using distributed measurement data. 18. The method according to claim 1, wherein the distribution measurement data is mapped to a three-dimensional shape model of the equipment, and then converted into a front view and a side view to be displayed and the device status is monitored. 19. The method according to claim 1, wherein after mapping the distribution measurement data to a three-dimensional shape model of the device, a parameter of the measurement data is displayed in a chart along the axial direction of the device. Monitoring method. 20. The method according to claim 1, wherein when mapping the distribution measurement data to the three-dimensional shape model of the device, the measurement data of a plurality of screens are mapped and synthesized on the three-dimensional shape model of the device. Monitoring method. 21. The method according to claim 1, wherein the distribution measurement data is compared with the previously measured distribution measurement data, and an abnormality is determined based on the difference between them. 22. The method according to claim 1, wherein the distribution measurement data is compared with a preset abnormality determination level, and the equipment state is monitored by the distribution measurement data for determining abnormality. 23. The method according to claim 1, wherein the distribution measurement data is used to determine an abnormality, and the distribution measurement data is used to monitor the device status. 24. The method according to claim 1, wherein the distribution measurement data is used to calculate a correlation between the distribution measurement data and past data, and to determine an abnormality. 25. The method according to claim 1, wherein two or more kinds of measurement data having different determinations of abnormality are combined, and the equipment state is monitored by the distribution measurement data determined by the combined value.