JP2023106032A - Data visualization system, data visualization method, and program - Google Patents

Abstract

To make it possible to grasp a degree of a variant of a value of each of variables in a parallel coordinates plot.SOLUTION: A data visualization system in accordance with an embodiment includes a first calculation unit that calculates a variation coefficient of a value of each of variables of multivariate data that is an object of visualization, a normalization unit that normalizes the value of a variable to a predetermined scale for each variable, a second calculation unit that calculates a correction value, which represents a value obtained by multiplying a normalized value of the variable by the variation coefficient of the variable, for each variable, and a visualization unit that visualizes the multivariate data as a parallel coordinates plot by using the correction value for each of the variables.SELECTED DRAWING: Figure 7

Description

The present invention relates to a data visualization device, a data visualization method, and a program.

A graph called a parallel coordinates plot is known as one of statistical graphs useful for visualizing multivariate data. Also, as techniques related to parallel coordinate plotting, for example, techniques described in Patent Documents 1 to 6 are known.

An interactive operation called brushed highlighting is generally used to investigate relationships between variables in parallel coordinate plots.

Japanese Patent No. 6621124 Japanese Patent No. 6549174 Japanese Patent No. 6661500 Japanese Patent No. 5392635 Japanese Patent No. 4155363 Japanese Patent No. 4819273

However, it is difficult to grasp the degree of variation in the value of each variable at the same time only by an interactive operation called brushed highlighting.

An embodiment of the present invention has been made in view of the above points, and an object of the present invention is to make it possible to grasp the degree of variation in the values of each variable in a parallel coordinate plot.

In order to achieve the above object, a data visualization device according to one embodiment includes a first calculation unit configured to calculate, for each variable of multivariate data to be visualized, a coefficient of variation of the value of the variable; a normalization unit configured to normalize the value of the variable to a predetermined scale for each variable; and a variation of the variable with respect to the normalized value of the variable for each variable. a second calculator configured to calculate a correction value representing a value multiplied by a coefficient; and configured to visualize the multivariate data as a parallel coordinate plot using the correction value for each of the variables. and a visualization unit.

You can grasp the degree of variation in the value of each variable in the parallel coordinates plot.

FIG. 10 is a diagram showing an example of a parallel coordinates plot; It is a figure which shows an example when one certain data is selected in a parallel coordinate plot. It is a figure showing an example of the whole data visualization system composition concerning this embodiment. It is a figure which shows an example of the hardware constitutions of the data visualization apparatus which concerns on this embodiment. It is a figure showing an example of functional composition of a data visualization device concerning this embodiment. 6 is a flowchart showing an example of parallel coordinate plot display processing according to the present embodiment; It is a figure which shows an example of the parallel coordinate plot after data correction|amendment. FIG. 10 is a diagram showing an example of a case where one piece of data is selected in the parallel coordinates plot after data correction;

An embodiment of the present invention will be described below. In this embodiment, as an example, a case will be described in which performance data such as operation data and process data acquired from a plant are targeted and visualized as a parallel coordinate plot.

Here, actual data such as operation data, process data, etc. should be written as x _n (1≦n≦N, where N is the total number of variables) and the actual value of variable x _n at time t as X _n (t). For example, X(t)=(X ₁ (t), . . . , X _N (t)). Below, it is assumed that index i is assigned to performance data X(t) in ascending order of time t, and i-th performance data is X ⁽ⁱ⁾ =(i, t ⁽ⁱ⁾ , X ₁ ⁽ⁱ⁾ , . . , X _N ⁽ⁱ⁾ ). Specific examples of variables include speed, temperature, pressure, torque, flow rate, concentration, command values for operable variables, and the like. Such variables are called, for example, process variables.

However, targeting performance data such as operation data and process data acquired from a plant is an example, and the present embodiment can target arbitrary multivariate data.

<Parallel coordinates plot>
A conventional parallel coordinates plot will now be described. In a parallel coordinate plot, the coordinate axes of each variable are arranged vertically in parallel, and the actual values of all variables are plotted so that the minimum value of that variable is at the bottom and the maximum value is at the top. Actual values on adjacent coordinate axes are connected by line segments with respect to the same actual data.

An example of a parallel coordinates plot is shown in FIG. In the parallel coordinate plot 1000 shown in FIG. 1, the coordinate axes of "d_index" representing the index i of the performance data and "day" representing the day included in the time t ⁽ⁱ⁾ of the performance data are arranged in order from the left, After that, "speed command value 1 (rpm)", "actual speed value 1 (rpm)", "actual torque value 1 (%)", "pressure 1 [MPa]", "pressure 2 [MPa]", "speed Actual value 2 (%)", "Torque actual value 2 (%)", "Pressure 3 [MPa]", "Pressure 4 [MPa]", "Speed command value 3 (rpm)", "Speed actual value 3 ( rpm)”, “actual torque value (%)”, “pressure 5 [MPa]”, “pressure 6 [MPa]”, and “actual speed value 4 (rpm)” are arranged in order from the left. there is

In the parallel coordinate plot, one polygonal line connecting the actual values of adjacent coordinate axes with line segments represents one actual data. For example, in the parallel coordinate plot 1000 shown in FIG. 1, the polygonal line 1100 shown in FIG. 2 represents one performance data (specifically, the performance data whose d_index is "750").

Parallel coordinate plots are useful in that all performance data can be visualized at the same time because the coordinate axes of all variables can be displayed on one screen. It is also useful in that the relationship between variables on two adjacent coordinate axes can be directly known, and the relationship between variables on two non-adjacent coordinate axes can be indirectly known. However, it is difficult to grasp the degree of dispersion of the actual values of each variable (for example, the extent of the index representing the degree of dispersion such as variance). For example, in the parallel coordinates plot 1000 shown in FIG. 1, if one wishes to compare the degree of variability of two variables with respect to their actual values, it is difficult to ascertain which degree of variability is greater. To give a specific example, for example, it is difficult to grasp which degree of variation is greater between the actual value of "actual torque value 1 (%)" and the actual value of "actual torque value 2 (%)". is.

Therefore, below, by correcting the actual value of each variable included in the actual data, when visualized as a parallel coordinate plot, the degree of variation in the actual value of each variable can be grasped at the same time between variables in the parallel coordinate plot. A data visualization system 1 that makes it possible will be described.

<Overall Configuration Example of Data Visualization System 1>
FIG. 3 shows an example of the overall configuration of the data visualization system 1 according to this embodiment. As shown in FIG. 3 , the data visualization system 1 according to this embodiment includes a data visualization device 10 , a data collection device 20 , a control device 30 and a plant 40 . The data visualization device 10 and the data collection device 20 are communicably connected via any communication network. Similarly, the data collection device 20 and the control device 30 are connected via an arbitrary communication network, and the control device 30 and the plant 40 are connected via an arbitrary communication network.

The data visualization device 10 acquires performance data from the data collection device 20 and visualizes all or part of the performance data as a parallel coordinate plot.

The data collection device 20 collects performance data from the control device 30 . The data collection device 20 may not only collect performance data from the control device 30, but may also perform abnormality diagnosis (or abnormality detection) of the plant 40 using these performance data, for example. Such anomaly diagnosis (or anomaly detection) can be realized by a known technique such as multi-variate statistical process control (MSPC).

The control device 30 acquires actual values of various variables from the plant 40 and creates actual data including these actual values. Further, the control device 30 controls the plant 40 using each actual value. Examples of the control device 30 include a PLC (Programmable Logic Controller).

The plant 40 is a device, facility, or the like that performs various processes. Specific examples of the plant 40 include petrochemical plants, steel plants, food plants, and the like.

Note that the overall configuration of the data visualization system 1 shown in FIG. 3 is an example, and other configurations may be used. For example, the data visualization device 10 and the data collection device 20 may be integrally configured.

<Hardware Configuration Example of Data Visualization Device 10>
FIG. 4 shows a hardware configuration example of the data visualization device 10 according to this embodiment. As shown in FIG. 4, the data visualization device 10 according to the present embodiment includes an input device 101, a display device 102, an external I/F 103, a communication I/F 104, a RAM (random access memory) 105, a ROM (Read Only Memory) 106 , an auxiliary storage device 107 and a processor 108 . Each of these pieces of hardware is communicably connected via a bus 109 .

The input device 101 is, for example, a keyboard, mouse, touch panel, physical button, or the like. The display device 102 is, for example, a display, a display panel, or the like.

The external I/F 103 is an interface with an external device such as the recording medium 103a. Examples of the recording medium 103a include a CD (Compact Disc), a DVD (Digital Versatile Disk), an SD memory card (Secure Digital memory card), a USB (Universal Serial Bus) memory card, and the like.

Communication I/F 104 is an interface for connecting data visualization device 10 to a communication network. A RAM 105 is a volatile semiconductor memory (storage device) that temporarily holds programs and data. The ROM 106 is a non-volatile semiconductor memory (storage device) that can retain programs and data even when power is turned off. The auxiliary storage device 107 is, for example, a non-volatile storage device such as a HDD (Hard Disk Drive) or SSD (Solid State Drive), and stores programs and data. The processor 108 is, for example, various arithmetic devices such as a CPU (Central Processing Unit).

Note that the hardware configuration shown in FIG. 4 is an example, and the data visualization device 10 may have other hardware configurations. For example, the data visualization device 10 may have multiple auxiliary storage devices 107 and multiple processors 108, and may have various hardware other than the illustrated hardware.

<Example of functional configuration of data visualization device 10>
FIG. 5 shows a functional configuration example of the data visualization device 10 according to this embodiment. As shown in FIG. 5 , the data visualization device 10 according to this embodiment has an acquisition unit 201 , a correction unit 202 and a visualization unit 203 . Each of these units is implemented by, for example, processing that one or more programs installed in the data visualization device 10 cause the processor 108 or the like to execute. The data visualization device 10 according to this embodiment also has a performance data DB 204 . The DB (database) is implemented by, for example, the auxiliary storage device 107 or the like. Note that the performance data DB 204 may be realized by, for example, a storage device (such as a database server) connected to the data visualization device 10 via a communication network.

The acquisition unit 201 acquires one or more performance data to be visualized from the performance data DB 204 .

The correction unit 202 visualizes the one or more actual data acquired by the acquisition unit 201 as a parallel coordinate plot, and can simultaneously grasp the degree of variation in the actual value of each variable between the variables. Make corrections to More specifically, the correction unit 202 calculates a statistic called a coefficient of variation for each variable of one or more performance data acquired by the acquisition unit 201, normalizes the performance value of the variable, and Multiply the actual value of the variable by the coefficient of variation of the variable. As a result, when each corrected performance data is visualized as a parallel coordinates plot, it is possible to grasp the degree of variation in the performance values of each variable at the same time.

The visualization unit 203 visualizes each performance data corrected by the correction unit 202 as a parallel coordinate plot.

The performance data DB 204 stores performance data acquired from the data collection device 20 . Below, a set of performance data stored in the performance data DB 204 is E={X ⁽ⁱ⁾ =(i,t ⁽ⁱ⁾ , _X1 ⁽ⁱ⁾ ,..., _XN ⁽ⁱ⁾ )|i= 1, . . . , |E|}. I is the number of performance data stored in the performance data DB 204 . Note that the performance data may be acquired from the data collection device 20 via a communication network, for example, or once stored in the recording medium 103a or the like from the data collection device 20, and read from the recording medium 103a. may be obtained.

<Parallel coordinate plot display processing>
In the following, the process of displaying a parallel coordinates plot that allows the degree of variation in the actual values of each variable to be simultaneously grasped when the actual data is visualized will be described with reference to FIG. 6 .

First, the acquisition unit 201 acquires one or more performance data to be visualized from the performance data DB 204 (step S101). Below, {X ⁽ⁱ⁾ = (i, t ⁽ⁱ⁾ , X ₁ (i) , ..., X _N ( ^{i )} )|i = 1, ..., I} Assume that ⊂E is obtained. Note that the performance data to be visualized may be, for example, performance data within a period designated by the user or the like, or may be performance data selected by the user or the like, or may be stored in the performance data DB 204. It may be all performance data that is

Next, the correction unit 202 calculates a statistic called a coefficient of variation for each variable _xn of the performance data X ⁽ⁱ⁾ obtained in step S101 (step S102). Specifically, if an is the coefficient of variation corresponding to the variable _xn , the _correction unit 202 calculates the coefficient of variation _an by _an = _σn / _μn . Here, σ _n is the standard deviation of X _n ⁽¹⁾ , . . . , X _{n (} ^{I) ,} and μ _n is the average of X _n ⁽¹⁾ , . As a result, the coefficient of variation _an is calculated for n=1, . . . , N.

Next, the correction unit 202 normalizes the actual value X _n ^{(i) (} n=1, . ). For example, the correction unit 202 normalizes each actual value X _n ⁽ⁱ⁾ (n=1, . However, this is just an example, and normalization may be performed so as to be 0 or more and 1.0 or less. As a result, each actual value X _n ⁽ⁱ⁾ is normalized to [−1, 1] (or may be [0, 1]) for n=1, . Hereinafter, assuming that the actual value X _n ⁽ⁱ⁾ is normalized to [−1, 1], the normalized actual value will be denoted as Y _n ⁽ⁱ⁾ .

The maximum value of _Xn ⁽ⁱ⁾ for i is Xn _,max , the minimum value of Xn ⁽ⁱ⁾ for i is _{Xn ,min} , the upper end of the normalized actual value is _Ymax , and the lower end is _Ymin . _{Then ,} ^the _actual ^value _{X n ( i _ _} ^{_ )} can be normalized. For example, when the actual value X _n ⁽ⁱ⁾ is normalized to [−1, 1], normalization can be performed using the above formula with Y _max =1 and Y _min =−1.

Next, the correction unit 202 multiplies the normalized actual value Y _n ⁽ⁱ⁾ obtained in step S103 by the coefficient of variation a _n (step S104). Specifically, the correction unit 202 calculates _a n ×Y _n ⁽ⁱ⁾ for n=1, . . . , N and i=1, . Below, _Zn ⁽ⁱ⁾ = _an * _Yn ⁽ⁱ⁾ is calculated. _As ^{a result} , for ⁱ = ₁ , ^. is obtained.

Then, the visualization unit 203 visualizes the post-correction performance data Z ⁽ⁱ⁾ obtained in step S104 as a parallel coordinate plot (step S105). That is, the visualization unit 203 plots each actual value Z _n ⁽ⁱ⁾ on the coordinate axis of the variable x _n , and then plots the actual values Z _n ⁽ⁱ⁾ and Z _n′ ⁽ⁱ⁾ plotted on the adjacent coordinate axes. A polygonal line corresponding to the performance data Z ⁽ⁱ⁾ is visualized by connecting with line segments. At this time, i and t ⁽ⁱ⁾ included in the performance data Z ⁽ⁱ ) may be plotted and connected with a line segment. As a result, a parallel coordinates plot in which a polygonal line corresponding to the corrected actual data Z ⁽ⁱ⁾ is drawn is obtained. This parallel coordinate plot is displayed on a display device 102, such as, for example, a display.

FIG. 7 shows an example of a parallel coordinate plot that visualizes the actual data Z ⁽ⁱ⁾ (i=1, . . . , I) after correction. In the parallel coordinate plot 2000 shown in FIG. 7, the coordinate axes of "d_index" representing the index i of the performance data and "day" representing the day included in the time t ⁽ⁱ⁾ of the performance data are arranged in order from the left, After that, "speed command value 1 (rpm)", "actual speed value 1 (rpm)", "actual torque value 1 (%)", "pressure 1 [MPa]", "pressure 2 [MPa]", "speed Actual value 2 (%)", "Torque actual value 2 (%)", "Pressure 3 [MPa]", "Pressure 4 [MPa]", "Speed command value 3 (rpm)", "Speed actual value 3 ( rpm)”, “actual torque value (%)”, “pressure 5 [MPa]”, “pressure 6 [MPa]”, and “actual speed value 4 (rpm)” are arranged in order from the left. there is

With the parallel coordinates plot 2000 shown in FIG. 7, it is possible to grasp the degree of dispersion of the actual values of each variable simultaneously, easily and instantaneously. For example, between the actual value of "actual torque value 1 (%)" and the actual value of "actual torque value 2 (%)", the degree of variation in the actual value of "actual torque value 2 (%)" is larger. It is possible to grasp easily and instantly.

Also in the parallel coordinates plot 2000 shown in FIG. 7, one polygonal line connecting the actual values of adjacent coordinate axes with line segments represents one actual data. For example, a polygonal line 2100 shown in FIG. 8 represents one performance data (specifically, performance data whose d_index is "800").

<Summary>
As described above, the data visualization device 10 according to the present embodiment focuses on the statistics of the actual values for each variable when visualizing the performance data as a parallel coordinate plot, and uses the statistics to visualize the performance data for each variable. Make a correction that adjusts the scale of the values. As a result, when the performance data after correction is visualized as a parallel coordinates plot, it is possible to grasp the degree of variation in the performance values of each variable simultaneously, easily and instantly. Therefore, for example, in an abnormality factor analysis of the plant 40, it is possible to easily perform an analysis such as judging that a variable having a greater degree of variation than other variables is highly likely to be an abnormality factor.

In the above embodiment, the case of visualizing corrected performance data as a parallel coordinate plot has been described. may be specified as an abnormality factor variable, and a process of controlling the equipment of the plant 40, etc., may be performed based on the specified result.

The present invention is not limited to the specifically disclosed embodiments described above, and various variations, modifications, combinations with known techniques, etc. are possible without departing from the scope of the claims. be.

1 Data Visualization System 10 Data Visualization Device 20 Data Collection Device 30 Control Device 40 Plant 101 Input Device 102 Display Device 103 External I/F
103a recording medium 104 communication I/F
105 RAMs
106 ROMs
107 Auxiliary storage device 108 Processor 109 Bus 201 Acquisition unit 202 Correction unit 203 Visualization unit 204 Performance data DB

Claims (8)

a first calculation unit configured to calculate, for each variable of multivariate data to be visualized, a coefficient of variation of the value of the variable;
a normalization unit configured for each of said variables to normalize the value of said variable to a predetermined scale;
a second calculation unit configured to calculate, for each of the variables, a correction value representing a value obtained by multiplying the normalized value of the variable by a coefficient of variation of the variable;
a visualization unit configured to visualize the multivariate data as a parallel coordinate plot using the correction values for each of the variables;
A data visualization device having The first calculator,
2. The data visualization device according to claim 1, wherein the product of the standard deviation of the values of the variables and the average of the values of the variables is calculated as the coefficient of variation of the variables. The first calculator,
The variables of the multivariate data to be visualized are _xn (n=1, ..., N, N is the total number of variables), i (i = 1, ..., I, I is the multivariate data to be visualized number)-th multivariate data x _n value is X _n ⁽ⁱ⁾ , the standard deviation σ _n of X n ⁽ⁱ ) for i and _{X n} ⁽ⁱ⁾ for i for each variable x _n 3. The data visualization device according to claim 2, configured to calculate a product of the average μ _n of , as the _coefficient of variation an of the variable _xn . The normalization unit
configured to calculate a value _Yn ^{(i) obtained by normalizing the value Xn(i)} _of ^the variable _xn to a predetermined scale for each of the variables _xn ,
The second calculator,
The correction value Z _n (i ⁾ is calculated by multiplying the normalized value Y _n ⁽ⁱ⁾ of the variable x _n by the coefficient of variation a _n for each of the variables x _n . 4. The data visualization device according to claim 3. The visualization unit
By plotting the correction value Z _n ⁽ⁱ⁾ on the coordinate axis of the variable x _n and connecting the points plotted on adjacent coordinate axes with respect to the same i with a line segment, the multivariate data is plotted as a parallel coordinate plot 5. The data visualization device of claim 4, configured to visualize. The normalization unit
6. The data visualization device according to any one of claims 1 to 5, configured to normalize the value of the variable to a scale of -1 to 1 or 0 to 1 or less. a first calculation procedure for calculating a coefficient of variation of the value of the variable for each variable of multivariate data to be visualized;
a normalization procedure for, for each of said variables, normalizing the value of said variable to a predetermined scale;
a second calculation procedure for calculating, for each of the variables, a correction value representing a value obtained by multiplying the normalized value of the variable by a coefficient of variation of the variable;
a visualization procedure for visualizing the multivariate data as a parallel coordinate plot using the correction values for each of the variables;
A computer-implemented data visualization method. a first calculation procedure for calculating a coefficient of variation of the value of the variable for each variable of multivariate data to be visualized;
a normalization procedure for, for each of said variables, normalizing the value of said variable to a predetermined scale;
a second calculation procedure for calculating, for each of the variables, a correction value representing a value obtained by multiplying the normalized value of the variable by a coefficient of variation of the variable;
a visualization procedure for visualizing the multivariate data as a parallel coordinate plot using the correction values for each of the variables;
A program that makes a computer run

Images