JP2006048195A - Visualization method for time-series data, visualization program and visualization system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily find out changing points of time or the situations of the change while preparing a compact image of plotting locus in visualizing data in a long time along a time axis. <P>SOLUTION: When time-series data are assigned to the inside or outside of a plotting locus T for every component, and any change is generated in the time-series data, the time-series data are distributed so that the curvature of the plotting locus T can be changed, and the data are spirally displayed while an outward repulsive force acting between the already plotted plotting locus T and a new plotting locus T to be successively plotted at the top end, and applying a repulsive force to the new plotting locus T and an inward repulsive force acting an internal force to the new plotting locus are operated. It is desired that while the bending of the plotting locus T is absent or small, the plotting locus of the data is reduced to a time axial direction since the change of the data are small. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a time series data visualization method and visualization system. More specifically, the present invention relates to an improvement of a data visualization method suitable for application to sound signal processing, equipment monitoring, equipment diagnosis and the like.

  Analysis of observation data over a long period of time is an issue in many fields such as technological development and elucidation of natural phenomena. For example, it is necessary to analyze the long-term observation data accumulated along the time series in order to diagnose power plant conditions or elucidate degradation phenomena. It is important to find data changes that represent the progress of

  In such data analysis and search, not only statistical analysis and signal processing technology by a computer, but also data visualization technology that utilizes human visual ability plays an important role. For example, when analyzing long-term data, it is important to detect change points such as changes in discharge intensity and changes in insulation characteristics due to leakage current. Visualization technology was used to detect these change points. Discrimination by human eyes is effective.

  Examples of such data visualization techniques include the following. Of these, the linear type, strip type, and spiral type will be described with reference to FIGS.

(A) Linear time series visualization (see FIG. 21)
It is the most widely used time-series data display method at present and can be said to be standard. Here, a long time data can be listed by cutting and arranging the display to the length within the screen range. For example, assuming that n data are displayed, n times as many data as the linear display can be listed. In addition, when there is a correlation between time and data, there is also a feature that a correlation can be easily found by displaying a plurality of visualization results with the time aligned.

(B) Strip-shaped time series visualization (see Fig. 22)
This is a method in which long data is cut and displayed in a strip shape, and is often used next to a linear shape. There is an advantage that information can be displayed densely on the screen, and the tendency of data correlated with time can be easily understood.

(C) Spiral type time series visualization (see FIG. 23)
This is a method of displaying a vortex from the center. There is an advantage that information can be displayed at a density similar to that of a strip type, and unlike the strip type, the information can be displayed without a break in the time axis. However, the display amount within the same screen is smaller than that of the strip type display because of margins at the four corners.

(D) An example of strip-type time series visualization As an example of a strip-type visualization method for statistical data, there is one that changes the size of a rectangle to be displayed according to data and packs it into a band-like area (for example, , See Patent Document 1). It has the property that it is easier to distinguish data than a simple bar graph, line graph, or spectrogram.

(E) An example of extending linear time-series visualization A technique of visualizing respiratory sounds and displaying them on a screen is disclosed (for example, see Patent Document 2). The breathing sound is displayed as a three-dimensional mountain valley having a height, and visualized as if viewed from an oblique direction (for example, see Patent Document 2).

JP 2003-256855 A JP 2004-33254 A

  However, in the conventional visualization method that expresses data in a fixed form, there is a problem that when the data becomes long, the visualization result becomes fine and the change is buried, so that it is difficult to distinguish the change visually. In other words, in the case of the conventional long-term data visualization method, even though there are differences such as straight lines and spirals in the representation of the time axis direction, these extend in the axial direction regardless of the data change. When the data is visualized, the data is continuous in the time axis direction, and the change in the data is less likely to appear as a shape change. For this reason, in order to distinguish the difference in data, it is necessary to distinguish a long and thinly visible pattern, and there is a problem that it becomes difficult to distinguish the difference as the time increases.

Each of the above-described data visualization techniques has the following problems (see advantages and disadvantages shown in Table 1).

  That is, the visualization technique (a) has the disadvantages that although it is currently most widely used, only a part of long data can be seen, and the amount that can be displayed on the screen is small (see FIG. 21). In addition, in order to express long-time data with this display method, there is a drawback that the short-time change becomes invisible as the time is shortened. For this reason, it is often necessary to use a user interface that facilitates browsing of a specific location by scrolling in the time axis direction or zooming in / out.

  In the case of the visualization technique (b), the time-series data is arbitrarily cut and displayed, so that the time axis is cut at the edge of the screen and the data corresponding to the break is not known (see FIG. 22).

  Although the visualization technique (c) has the advantage that there is no break unlike the strip type, it is not a thing that can recognize changes at a glance because it is a regular spiral pattern, and its visibility is low accordingly. There is a drawback. Further, when the displayed data is viewed along the time axis, it is necessary to follow a regular vortex with the eyes, so that there is a drawback that the eyes are easily turned (see FIG. 23).

  In addition, there is a problem that the tendency and characteristics of data become difficult to see as the image is reduced in common with each of the visualization techniques (a), (b), and (c) described above. Furthermore, since every data string has the same shape (strip or vortex), there is a problem that it is difficult to distinguish between a plurality of visualization results.

  In the case of the visualization technique (d), although there is an advantage that data can be easily distinguished from simple bar graphs, line graphs, spectrograms, etc., the strip shape itself does not change, so if it becomes long, it becomes a collection of fine rectangles. In particular, the shortcomings of strips cannot be solved.

  In the case of the visualization technique (e), since the time axis is fixed obliquely and does not bend, there remains a drawback that only a short time can be displayed as in the above-described linear type.

  In view of this, the present invention provides a time series data visualization method and visualization that make it easy to find the time of change or the state of change while making the drawing locus compact when visualizing data over a long time along the time axis. The purpose is to provide a program and visualization system.

  In order to achieve this object, the present inventor has conducted various studies. First, as the display area is reduced, data trends and features become difficult to see, and because all data strings have the same general shape, it is difficult to distinguish between multiple visualization results. We thought that it was common to devise display attributes such as color and shape to deal with the problem. The advantages and disadvantages of each display attribute are summarized (see Table 2). Of these, characterization by coloring is the most commonly used technique, and it is certainly easy to understand when the number of colors is small. However, if the number of colors increases and there is a dense color change, the resulting color mixture will tend to be similar to the human eye. Marks and characters are often used for qualitative data and structured data with clear properties (characteristics). However, for unstructured data, it is difficult to specify marks and characters appropriately. It ’s hard.

  Note that “structured data” in this specification refers to data such as a list, a graph, and a table, in which data that is an individual unit is represented by a set of individually defined items. In many cases, the relationship between these data is defined by links, identification numbers, and the like. For example, employee profiles, schedules, and entry / exit records are representative of structured data. In addition, “unstructured data” as used in this specification refers to structured data that is not divided into units corresponding to defined items, or the relationship between data is not shown even if divided. The data does not satisfy the property. For example, the result of elementary signal processing such as measurement values or FFT, such as recording of input signals from a sensor or video images, is data in which a continuous value string such as a time axis is formed. In some cases, such as in the case of multiple channels, it is a bundle of multiple data strings.

  In addition, there is something that should be striking in the swirl type visualization method in that the data can be visualized as much as possible within the limited display range on the screen, but the above-mentioned causes caused by swirl display are mentioned above. Such drawbacks become a bottleneck. The present inventor, who has repeatedly studied this point, has focused on utilizing information in a form that has not been utilized so far. There are local shapes and global shapes in the information. The local shape means a feature of thickness and size, and a quantitative shape feature such as smooth or jagged, in addition to the above-described marks and the like. The global shape represents the shape of the entire data string expressed as lazy or crowded, and includes deformation of the display structure. As a result of the examination, it was concluded that this type of information, especially the feature representation of data with a global shape, was not utilized in the past. The reason for this is that if the shape is deformed, the overall shape, that is, the display structure is distorted, for example, congestion such as overlap or intersection occurs, the visibility of the visualization result itself decreases, and the amount of information per unit area decreases. Can be mentioned.

  Further examination based on the above, if there is any tendency or feature in the target data, by distorting the visualized image according to the tendency or feature, the point of change or its contents can be changed from the long time data. I have come to know that it will be easy to find or understand.

  The present invention is based on such knowledge, and the invention according to claim 1 is a time-series data visualization method for displaying data over a long time along a time axis as a spiral drawing locus on one screen of an output device. When the data is distributed inside or outside the drawing trajectory for each component and the data changes, the curvature of the drawing trajectory is changed and distributed, and the already drawn drawing trajectory and its tip are distributed. The data is vortexed while acting on the new drawing trajectory that continues to be drawn and applying an outward repulsive force that applies a repulsive force to the new drawing trajectory and an inward repulsive force that applies an inward force to the new drawing trajectory. Is displayed.

  First, the present invention is characterized in that the data is artificially distorted so that changes can be seen at a glance. That is, by distributing data to the inside or outside of the drawing trajectory for each component, when the data changes, the drawing trajectory bends (distorts) the visualized image according to the data change different from the previous one, The change is reflected in the shape, and a visualized image is generated in which the point of time when the tendency or the characteristic of the long-time data changes can be easily discriminated.

  Furthermore, the method using the analogy of repulsive force, that is, the method of controlling the shape of the drawing trajectory by treating it as if the physical repulsive force is acting on the drawing trajectory, The entire visualized image can be stored compactly in the display screen while avoiding overlapping and crossing.

  In the above-described method for visualizing time-series data, as described in claim 2, since the change in the data is small, the drawing locus of the data is not changed in the time axis direction while the drawing locus is not bent or small. It is preferable to reduce the size.

  Furthermore, in the above-described method for visualizing time series data, the repulsive action space in which the outward repulsive force and the inward repulsive force act on the drawing trajectory can be expanded to such an extent that at least overlapping or crossing of the drawing trajectories is avoided. preferable.

  In addition, in the above-described method for visualizing time-series data, it is preferable that the outward repulsive force is applied at regular intervals on the drawn drawing trajectory.

  According to a fifth aspect of the present invention, there is provided a time-series data visualization program for displaying data over a long time along a time axis as a spiral drawing locus on one screen of an output device. If there is a change in the data by distributing inside or outside the drawing trajectory for each component, the process of allocating the curvature of the drawing trajectory to change, and the already drawn drawing trajectory already drawn and its tip A process of applying a repulsive force between the new drawing trajectory to be drawn and applying a repulsive force to the new drawing trajectory; and a process of continuously displaying data while applying an inward repulsive force from the outside to the new drawing trajectory. Is to execute.

  In the above-described time-series data visualization program, it is preferable to execute a process of reducing the drawing trajectory of the data in the time axis direction while the drawing trajectory is not bent or small because the data change is small.

  Further, in the above-described time series data visualization program, a process of expanding the repulsive action space in which the outward repulsive force and the inward repulsive force act on the drawing trajectory to at least an extent that overlap or intersection of the drawing trajectories is avoided. It is preferable to execute.

  In addition, the above-described time-series data visualization program preferably executes a process of causing outward repulsive force to act on the drawn drawing trajectory at regular intervals.

  Furthermore, the invention according to claim 9 is a time-series data visualization system that displays data over a long time along a time axis as a spiral drawing locus on one screen of the output device. Data distribution means for allocating the data so that the curvature of the drawing trajectory changes when the data changes by distributing to the inside or the outside of the drawing, and the newly drawn drawing trajectory already drawn at the tip of the drawn drawing trajectory An outward repulsive force applying means for applying a repulsive force to the new drawing trajectory by applying a repulsive force between the drawing trajectory and an inward repulsive force applying means for applying an inward force to the new drawing trajectory. The data is displayed in a vortex shape while applying an inward repulsive force. Such a time-series data visualization system may include compression means for reducing the drawing locus of the data in the time axis direction while the drawing locus is not bent or is small because the change in the data is small. preferable. It is also preferable to include an adjusting means that expands the repulsive action space in which the outward repulsive force and the inward repulsive force act on the drawing trajectory to such an extent that at least overlapping or crossing of the drawing trajectories is avoided. Furthermore, it is also preferable that the outward giving means described above causes the outward repulsive force to act on the drawn drawing trajectory at regular intervals.

  According to the method for visualizing time-series data according to claim 1, time-series data extending along the time axis can be visualized and displayed on the screen so that deformation can be easily seen according to the data. In such a case, since the changed portion of the data is displayed on the screen in a particularly reflected or emphasized state, it becomes easy to grasp and understand the trend of the data and the change point of the feature. For example, even if the data is so long that it is difficult to distinguish the visualized pattern, the difference can be easily distinguished from the change in shape. In addition, with respect to a plurality of data having different contents, the conventional methods all have the same shape, and the data can be distinguished only by the difference in the pattern, whereas according to the present invention, the shape of the visualization result is made different. It is easy to distinguish between multiple data.

  In addition, in the present invention, both the outward repulsive force and the inward repulsive force are applied in a balanced manner to the newly drawn new locus, so that the data can be displayed as an overall compact spiral locus. In other words, for the drawing trajectory, the inward repulsive force acts to drive the drawing trajectory inward as much as possible, while the outward repulsive force acts to moderately separate the trajectory, so that the drawing trajectory is made as compact as possible. An inward force for display always acts, and an outward repulsive force that does not hinder the action of the inward repulsive force acts moderately so that visibility is not lost. This makes it possible to perform visualization processing that overcomes the conflicting problem of making it easy to find the time of change or the state of the change while collecting the drawing trajectory in a compact manner. In short, in the present invention, it can be said that a new technique has been established in which the trajectory drawn in the visualization technique is intentionally distorted.

  According to the method for visualizing time-series data according to claim 2, the drawing content is made compact by compressing the drawing locus displayed on the screen, and more information is included in the limited display range. Is possible. In the first place, the part where the data change is small or not at all is a part that is not the object of finding (or distinguishing), so compressing does not hinder it, and the contents are more improved while maintaining the changed part. It is reasonable in the sense that it can be made compact.

  Furthermore, according to the method for visualizing time-series data according to claim 3, it is possible to make display contents easy to see by avoiding overlapping or crossing of drawing trajectories. If the repulsive action space, that is, the range in which the outward repulsive force and the inward repulsive force act on the drawing trajectory is narrowed to some extent, the repulsive force acting in a limited narrow range (particularly the inward repulsive force) can cause overlapping of the drawing trajectories. Such a situation can be avoided by setting the repulsive action space to a certain size.

  In addition, according to the method for visualizing time-series data according to claim 4, an outward repulsive force acts on the drawing history (that is, on the drawn drawing locus), and the drawing locus thereafter is already drawn. The drawing is performed so as to avoid the already drawn drawing locus, and there is no overlap or intersection of the locus.

  In addition, according to the inventions described in claims 5 to 9, it is possible to obtain a function and effect unique to the present application. That is, first, according to the visualization program for time-series data according to claim 5, the time-series data extending along the time axis is visualized through a process by a computer so that the deformation can be easily seen according to the data. Can be displayed above. In such a case, since the changed portion of the data is displayed on the screen in a particularly reflected or emphasized state, it becomes easy to grasp and understand the trend of the data and the change point of the feature. For example, even if the data is so long that it is difficult to distinguish the visualized pattern, the difference can be easily distinguished from the change in shape. In addition, with respect to a plurality of data having different contents, the conventional methods all have the same shape, and the data can be distinguished only by the difference in the pattern, whereas according to the present invention, the shape of the visualization result is made different. It is easy to distinguish between multiple data.

  In addition, in the present invention, both the outward repulsive force and the inward repulsive force are applied in a balanced manner to the newly drawn new locus, so that the data can be displayed as an overall compact spiral locus. In other words, for the drawing trajectory, the inward repulsive force acts to drive the drawing trajectory inward as much as possible, while the outward repulsive force acts to moderately separate the trajectory, so that the drawing trajectory is made as compact as possible. An inward force for display always acts, and an outward repulsive force that does not hinder the action of the inward repulsive force acts moderately so that visibility is not lost. This makes it possible to perform visualization processing that overcomes the conflicting problem of making it easy to find the time of change or the state of the change while collecting the drawing trajectory in a compact manner.

  Further, according to the visualization program for time-series data according to claim 6, the drawing content is made compact by compressing the drawing trajectory displayed on the screen, and more information is included in the limited display range. Is possible. In the first place, the part where the data change is small or not at all is a part that is not the object of finding (or distinguishing), so compressing does not hinder it, and the contents are more improved while maintaining the changed part. It is reasonable in the sense that it can be made compact.

  Furthermore, according to the visualization program for time-series data described in claim 7, it is possible to avoid the overlapping or crossing of the drawing trajectories and make the display contents easy to see. If the repulsive action space, that is, the range in which the outward repulsive force and the inward repulsive force act on the drawing trajectory is narrowed to some extent, the repulsive force acting in a limited narrow range (particularly the inward repulsive force) can cause overlapping of the drawing trajectories. Such a situation can be avoided by setting the repulsive action space to a certain size.

  In addition, according to the time-series data visualization system according to claim 8, since the outward repulsive force acts on the drawing history (that is, on the already drawn drawing locus), the subsequent drawing locus is already drawn. The drawing is performed so as to avoid the already drawn drawing locus, and there is no overlap or intersection of the locus.

  Furthermore, according to the time-series data visualization system according to the ninth aspect, the time-series data extending along the time axis can be visualized and displayed on the screen so that the deformation is easily seen according to the data. In such a case, since the changed portion of the data is displayed on the screen in a particularly reflected or emphasized state, it becomes easy to grasp and understand the trend of the data and the change point of the feature. For example, even if the data is so long that it is difficult to distinguish the visualized pattern, the difference can be easily distinguished from the change in shape. In addition, with respect to a plurality of data having different contents, the conventional methods all have the same shape, and the data can be distinguished only by the difference in the pattern, whereas according to the present invention, the shape of the visualization result is made different. It is easy to distinguish between multiple data.

  In addition, in the present invention, both the outward repulsive force and the inward repulsive force are applied in a balanced manner to the newly drawn new locus, so that the data can be displayed as an overall compact spiral locus. In other words, for the drawing trajectory, the inward repulsive force acts to drive the drawing trajectory inward as much as possible, while the outward repulsive force acts to moderately separate the trajectory, so that the drawing trajectory is made as compact as possible. An inward force for display always acts, and an outward repulsive force that does not hinder the action of the inward repulsive force acts moderately so that visibility is not lost. This makes it possible to perform visualization processing that overcomes the conflicting problem of making it easy to find the time of change or the state of the change while collecting the drawing trajectory in a compact manner.

  Hereinafter, the configuration of the present invention will be described in detail based on embodiments shown in the drawings.

  1 to 8 show an embodiment of the present invention. The data visualization method according to the present invention can be applied to a case where data over a long time along the time axis is displayed as a spiral drawing locus on one screen of the output device. If the data is changed by assigning it to the outside (in other words, the right or left side in the direction of travel of the drawing trajectory T), it is distributed so that the curvature of the drawing trajectory T is changed. An outward repulsive force that acts between the drawing trajectory T and a new drawing trajectory T that is subsequently drawn at the tip thereof to impart a repulsive force to the new drawing trajectory T, and an inward force on the new drawing trajectory T The data is displayed in a vortex while applying the inward repulsive force. The screen here is an output screen that can be visually recognized by a user or a user in an external output device such as a display.

  The method for visualizing time-series data according to the present invention makes it possible to easily find a change time point or a state of the change by artificially distorting the drawn locus while collecting the drawing locus T in a compact manner. Is for. For example, in the present embodiment, processing for changing the magnitude of the drawing direction vector v to make the feature stand out (hereinafter also referred to as “elongation processing”) is performed. The elongation process is a process along the time axis direction, such as reducing the data drawing trajectory T in the time axis direction while there is no bending of the drawing trajectory T because the change in data is small or small. is there. In addition, in the present embodiment, a process of bending the drawing path (hereinafter also referred to as “bending process”) is performed by changing the angle θ of the drawing direction vector v according to the data. Further, in the present embodiment, when performing such elongation processing and bending processing, the display area (drawing) by the repulsive element is not impaired without impairing the readability (e.g., easy to see, easy to understand, easy to distinguish) of the entire drawing trajectory T. Confinement to (space). Such a range, that is, “a range in which a repulsive element is applied to confine the repulsive element in the display area” is hereinafter referred to as a “repulsive force operating range”. Here, the “repulsive element” refers to an element that is a source of repulsive force, for example, outward repulsive force or inward repulsive force, and is denoted by reference numeral S in FIGS. 5 and 6 (see FIGS. 5 and 6).

Hereinafter, a method for visualizing time-series data in the present embodiment will be described. In this embodiment, as a method for visualizing long / unstructured data, unlike the conventional method in which the display structure (frame) is fixed, the local characteristics of the data are reflected in the display structure, so that the data trend is also reduced. And a visualization method that makes it easy to understand features (hereinafter referred to as “crawl method” for convenience). Note that this name is simply named for convenience because the visualization result has an atypical curved shape like a trace of insects. Below,
1. Visualization that reflects the magnitude of data changes 2. Reflect local features on the display structure. Each item of introduction of constraints for compact visualization will be explained sequentially.

1. Visualization reflecting the magnitude of data changes The drawing display unit in the crawl method (hereinafter referred to as the “visualization unit”, represented by the symbol U) is a rectangle in which multi-channel information is arranged and identified by color, brightness, etc. Or a sector shape (see FIGS. 1 and 2). The visualization unit U newly drawn at the tip of the drawing locus T becomes a new drawing locus T. Here, “channel” refers to data arranged on the left and right in the drawing direction. For example, in the present embodiment, they are represented as channel 1, channel 2,. By continuously drawing such a visualization unit U along the drawing direction vector (v), time-series data along the time direction is visualized. In this case, if the drawing direction vector v is straight, a linear display is obtained, and if the drawing direction vector v is inclined and moved in a spiral shape, a spiral display is obtained. In other words, if the gradient of the drawing direction vector (v) with respect to the traveling direction is zero, the visualization unit U becomes a rectangular shape that advances straight (see FIG. 1), and if it is inclined to the left or right, it becomes a fan shape that advances while turning (see FIG. 1). (See FIG. 2). In other words, the rectangle (see FIG. 1) can be generalized to be when the angle θ with respect to the traveling direction of the drawing direction vector v is zero.

  In this embodiment, the display width (w) of the visualization unit U is a constant width. On the other hand, the display length (l) or the angle θ is increased or decreased or increased or decreased depending on the drawing direction vector v. “Elongation processing”, which is one of the three features of the crawl method, is processing for changing the size of the drawing direction vector v to make the features stand out as described above (see FIG. 3). ). In this case, the average value of the values of each channel or the rate of change of the value of each channel can be used as the change amount. Further, if the length of the drawing direction vector v is proportional to these change amounts, it is displayed long at a portion where the change amount is large and short at a portion where the change amount is small (see FIG. 3). The overall image drawn in this way is a figure in which the time interval with little change is compressed or omitted, and as a result, the limited display area in the screen can be used effectively, so that the change is less likely to be missed. Is obtained.

2. Reflecting local features to the display structure As described above, the “bending process” in the crawl method is a process in which the angle θ of the drawing direction vector v is changed according to data to bend the drawing trajectory T (or its path). (Refer to FIG. 4). Bending causes changes in the smoothness of the drawing path or undulations in the drawing path in a short time, and the path changes greatly in the long time, so the overall shape of the visualization result changes. As an amount to be reflected in the bending process, a distribution between channels is used. The “distribution between channels” here is a certain value calculated by using all the values of the channels, and in the present embodiment, it is expressed by the following formulas 1, 2, and 3. The term “distribution” is used because the channel value often represents a frequency distribution or a frequency distribution. In the case of sound, the channel represents the amplitude of each frequency band, and represents the frequency distribution at a certain point in time for all channels. Further, as a method of reflecting the distribution, there can be mentioned a method in which the visualization unit U is regarded as a rectangular rigid body and expressed by a moment of force with the value of each channel as an external force. The number of channels is n, the value of each channel is the size, and a vector pointing in the same direction as the drawing direction vector
And the coordinates of the display location of each channel
And The moment of force when the visualization unit U is a rigid body with a constant density is
It becomes. Note that if it is simply considered that a moment of force is applied in a state where there is no frictional force, it will continue to accelerate, which may be against the original purpose of reflecting the value of each channel in the display structure. Therefore, in the present embodiment, a resistance proportional to the speed is provided, and the drawing speed is reduced as the value of each channel is reduced. By doing so, the drawing trajectory T turns to the right when the value of the left channel increases from the center of the visualization unit U toward the traveling direction, and turns to the left when the value of the right channel increases. Even if the change in value is the same, the rotation angle changes if the channel is different. Incidentally, under this analogy, the above-described elongation effect can be considered as a bending effect when the angle θ is constrained to zero.

  As described above, according to the bending effect as described above, the local characteristics of the data are reflected not only in the superficial pattern but also in the drawing path itself, and different data strings have clearly different shapes. As a result, when a long data string is visualized, the same visualization result disappears, and each inflection point represents the tendency and characteristics of the data. Note that when the bending effect is used, not only the right and left paths are shifted, but the entire drawing path is often bent greatly, and as a result, the previously drawn path may be approached and overlapped or crossed. In such a case, it is preferable to expand the repulsive force action space, which is a space where the repulsive element (for example, outward repulsive force and inward repulsive force) S acts. When the repulsive action space, that is, the range in which the outward repulsive force and the inward repulsive force are applied to the drawing trajectory T is narrowed to some extent, the repulsive force (particularly the inward repulsive force) acting in a limited narrow range causes the drawing trajectory T to overlap. In some cases, overlapping and crossing can be avoided by setting the repulsive action space to a certain size by adjusting means or the like.

3. Introduction of constraints for compact visualization By applying the above-mentioned “elongation processing” and “bending processing”, local characteristics of data can be reflected as display contents in the screen. It is desirable to make the visualization as compact as possible within a limited screen (display area). For example, when rendering is performed by applying a bending process without considering the screen size, a sparse visualization result is obtained in some cases. On the other hand, if the conventional strip type or spiral type is used, the effect of the bending process cannot be obtained. For this reason, it is necessary to balance two conflicting items of compactness and free drawing reflecting data. In view of the above points, in the present embodiment, the drawing trajectory T is displayed as compactly as possible in a limited display area by using a repulsive force acting from the outside to the inside in order to balance two conflicting items. (Hereinafter referred to as “confinement” in the display area). This will be described in detail below (see FIGS. 5 and 6).

First of all, as mentioned above, when the bending effect is applied, the visibility may be lowered due to the overlap or intersection of the drawing trajectories T. As described above, the situation is further reduced when the display area is limited. Is likely to occur. On the other hand, in the crawl method, the repulsive force is also used to reduce the overlap of the drawing trajectories T (in this embodiment, the element that is the source of the repulsive force is called “repulsive element”). ). The repulsive element S is also given to the visualization unit U. Here, the repulsive element at a certain point is q ₁ , the repulsive element of the visualization unit U is q ₀ , and the distance between them is r, the external force applied to the visualization unit U is
Where k is a constant. Then, an external force represented by the mathematical formula (mononomial) 4 is applied to the movement of the visualization unit U. In order to limit the drawing trajectory T to a narrow range, in the crawl method, for example, the repulsive element S is arranged around the outside in the display area. When the visualization unit U approaches the arranged repulsive element S, the above-described external force acts as a repulsive force, pushing the drawing trajectory T (visualization unit) inward. In other words, the repulsive element S acts like a wall that rebounds the drawing trajectory T (visualization unit U). By doing so, the drawing trajectory T, which is the visualization result, can be compactly contained (contained) in the display area (see FIGS. 5 and 6).

  As described above, since the influence of the repulsive element S is inversely proportional to the square of the distance, there is almost no influence on the drawing trajectory T while the distance between the repulsive elements S is long, and it is free according to the data trend and characteristics. A trajectory is drawn. On the other hand, when the drawing trajectory T approaches the repulsive force element S that becomes a wall, it receives an action of a large repulsive force, is turned to the inner side, and is pushed inward. By such an action and operation, it is possible to balance the conflicting items of compactness and free drawing.

  In order to reduce the overlap and intersection of the drawing trajectories T, it is also preferable to arrange the repulsive elements S at regular intervals on the history (that is, on the already drawn trajectories). In this way, repulsive force is generated from the once drawn region, and subsequent drawing is performed so as to avoid the already drawn locus, thereby reducing the overlap and intersection of the locus (see FIG. 6).

Note that when performing the confinement as described above, the rough shape of the whole is determined by the arrangement of the repulsive elements S to be used. For example, when the repulsive element S arranged on the circumference is gradually widened, the drawing trajectory T is roughly spiral, and when the repulsive element S arranged on the rectangle is gradually expanded, the drawing trajectory T is rectangular. The shape reciprocates along the long side. Here, “the gradually extending repulsive element S arranged circumferentially” means that the repulsive element S is arranged in a round shape at a constant interval, and the radius of the circle gradually increases as drawing progresses. It is to do. In this case, the shape of the drawing trajectory changes according to the initial radius r ₀ and the spreading speed dr per unit time. When expanding the radius of the circle, it is preferable to increase the number of repulsive elements S so that the distance between the repulsive elements S is constant. By doing so, the repulsive force to the visualization unit U in the peripheral portion of the display area becomes constant. Also, “gradually expanding the repulsive elements S arranged in the rectangle” means that the long side of the rectangle is not moved and the interval between the short sides is increased according to time. Again, coming form the drawing path is changed in accordance with the width increasing amount dl per initial length l ₀ and unit time of the short sides.

  Furthermore, if the description using a figure is added (refer FIG. 7), the visualization unit U will move so that it may escape to the vacant area of the circumference | surroundings. Here, the “vacant place in the vicinity” is determined by the force relationship between the repulsive element (for example, arranged on the outer periphery of the display area and on the drawn drawing trajectory T) S and the visualization unit U. The direction or place where the weakest repulsive force works. If the density of the repulsive element S is thin and the distance to the repulsive element S is far, the repulsive force acting on the place is very weak. Further, even if the density of the repulsive element S is high, the repulsive force that acts if it is far from the visualization unit U is not strong.

Here, a case where the repulsive elements S are arranged in a rectangular shape will be described as an example (see FIG. 7). A symbol T ₁ represents the starting point of the drawing area T. FIG. 7 shows a state in which the distance between the short sides of the rectangle gradually increases as the process proceeds from (a) to (e). At first, since the repulsive force of the upper empty portion in the rectangle is still weak, the visualization unit U is drawn while proceeding in the empty direction along the long side (FIG. 7A). When the visualization unit U approaches the wall (the rectangular side formed by the repulsive element S), the repulsive force from the wall increases and the path is bent in the direction in which the side is widened (FIGS. 7B to 7C). . This is because the repulsive force from the side wall is weakened due to the increase in the distance between the short sides. As the direction change proceeds, the visualization unit U turns downward (FIG. 7 (d)), and eventually it is folded back and moved toward an area that is largely open in the direction in which the visualization unit U has come (FIG. 7). 7 (e)). When such a movement is repeated, the drawing trajectory of the visualization unit U becomes a reciprocating shape.

  Next, a visualization program for executing visualization according to the present embodiment will be described (see FIG. 8). This visualization program is a program for displaying data over a long time along the time axis as a vortex-like drawing locus T on one screen of the output device. The computer displays data for each component inside or outside the drawing locus T. If there is a change in the data, the process of allocating so that the curvature of the drawing trajectory T changes, the already drawn drawing trajectory T that has already been drawn, and the new drawing trajectory T that is continuously drawn at the tip thereof, A process of applying a repulsive force between the new drawing trajectory T and a process of continuously displaying data while applying an inward repulsive force from the outside to the new drawing trajectory T. Is.

  A specific processing example of the visualization program will be described below (see FIG. 8). First, initial settings related to drawing are performed (step 1). Specifically, an initial setting for the visualization unit U, an initial setting for the relationship between data and drawing, an initial setting for a repulsive element group to be a wall, and an initial setting for the repulsive element S on the drawing trajectory T are performed. Further, the initial setting regarding the visualization unit U includes the size of the visualization unit U, the initial position, the initial velocity, and the value of the repulsive force S possessed by the visualization unit U.

  In addition, as an initial setting regarding the relationship between data and drawing, it is determined which channel value of data is to be linked to drawing as a drawing color, brightness, and the like. For example, in the case of a spectrogram of sound, the sound frequency band is divided into several parts, and the frequency band is applied to the hue of the color, and the average value of the amplitude of each band is applied to the lightness.

  Next, a repulsive element S group arranged to control the entire drawing trajectory T is set. The initial arrangement of the repulsive element S group and how to change the arrangement as drawing progresses are set. For example, if the repulsive elements S are arranged in a circle, the initial radius, the extension of the radius per unit time, the value of each repulsive element S arranged on the circumference, and the distance between the repulsive elements S are set. The value of the repulsive element S can be a fixed value, or can be a variable value corresponding to the value of the visualization unit U or the locus of the drawing locus T.

  Further, the repulsive element S on the drawing trajectory T is set. A value of the repulsive element S (a fixed value or a variation value corresponding to the value of the visualization unit U or the locus of the drawing trajectory T) is set, and an interval (time interval or distance interval) at which the repulsive element S is arranged is set. For example, in the case of the crawl method of the present embodiment, either a fixed value or a value proportional to the time average of the average brightness (that is, a larger value if the data value is larger) can be selected as the value of the repulsive element S. Yes. Further, as an example of the arrangement interval of the repulsive elements S, a constant time interval is used in the present embodiment.

  When the above setting (step 1) is completed, it is determined whether all data has been drawn or whether the capture has been completed (step 2). If applicable, the drawing process is terminated (step 10). For example, in the case where data from a file is read and converted into a visualization unit U and displayed, all the data to be drawn has been drawn, or the trace is drawn when the user instructs the drawing to end. It will end. If the input from the microphone, video, sensor, etc. is converted and drawn in real time, the drawing ends when the data capture ends or the user instructs the end of the drawing. I will do it.

  As a result of the end determination in step 2 described above, when the process does not proceed to step 10, that is, when all the data has not been drawn and the capture has not ended, the next data is read or captured (step 3). In the drawing process, data is first read.

  The data read or captured in step 3 may be used for drawing as it is or some signal processing may be performed in advance. Therefore, in step 4, necessary signal processing is performed as an option. For example, when drawing a sound signal, instead of the captured sound signal itself, frequency analysis (FFT) is performed, the average is calculated for each band, and the sound signal changes using the SS method (Spectral Subtraction method). The minute is taken out and the calculation result is used for drawing. This data is distributed to the inside or the outside of the drawing trajectory T (in other words, the right side or the left side in the traveling direction) for each component, such as a high tone and a low tone.

  Subsequently, the repulsive element S group serving as a wall is updated (step 5). That is, the repulsive element group S that is a wall may change the arrangement as drawing progresses. In this case, the repulsive element S is rearranged by calculating a new arrangement every time data is read. For example, when the repulsive element S is arranged in a circle to form a round wall, rearrangement is performed such that the repulsive element S is moved outward so that the circle expands. At this time, the number of repulsive elements S may change.

  Subsequently, the repulsive force S on the drawing trajectory T is updated (step 6). Here, in order not to cross the trajectory drawn so far, as the drawing progresses, for example, a repulsive element S that exerts a repulsive action on the visualization unit U is arranged on the drawn drawing trajectory T. At this time, each time data is read, it is determined whether or not the repulsive element S should be disposed, where it should be disposed, and how much repulsive element S should be disposed, and the placement is executed. To do. For example, in the crawl method according to the present embodiment, the repulsive element S is not arranged on the immediate side of the visualization unit U, but is arranged relatively behind the drawing trajectory T after the drawing progresses for a while. Yes. For example, if the repulsive element S is arranged on the immediate side of the visualization unit U, an extremely large repulsive force is applied to the visualization unit U, and the movement speed of the visualization unit U becomes extremely large, which may hinder smooth drawing. A proper arrangement is preferable in that such a fear is reduced.

  Subsequently, the repulsive force applied to the visualization unit U is calculated from all the repulsive elements S arranged using the above-described mathematical formula (step 7). For example, if 1000 repulsive elements S are arranged, 1000 repulsive force vectors are calculated, and these are combined to calculate the repulsive force vector applied to the visualization unit U.

  If the repulsive force vector is calculated in step 7, the moving direction and moving amount per unit time of the visualizing unit U are calculated from the current velocity of the visualizing unit U and the mass and inertia moment assumed for the visualizing unit U. Further, based on the calculation result, the movement destination of the visualization unit U for each unit time and further the posture there are calculated (step 8).

  Subsequently, the space from the previous position of the visualization unit U to the destination is drawn (step 9). That is, based on the final drawing position drawn at the time of the previous data reading and the posture of the visualization unit U at that time, and the position (movement destination) and posture of the visualization unit U newly calculated this time, the previous final drawing position and the current time Draw to connect the destination of. Specifically, image information such as coordinates indicating the drawing position is sent to the output device and displayed as an image that can be viewed on a display or the like. In addition, since the visualization unit U may bend, generally it is a fan shape or a locus approximate to this. When the drawing is completed, the process returns to step 2 (see FIG. 8).

  As described above, appropriately giving changes such as gradually increasing the wall constituted by the repulsive element S or changing the magnitude of the repulsive force is what changes the target data has. It is possible to display the visualization result as densely as possible in the display area (screen) even if it is a thing, and it is also possible to prevent overlapping or crossing of the drawing trajectories T while making the display dense. To do. Furthermore, these lead to obtaining effective and compact visualization results corresponding to the contents of various data that change.

  By the way, when the wall is enlarged or the magnitude of the repulsive force is changed as described above, the arrangement of the repulsive force S is updated as needed. In general, the “update” described in Steps 5 and 6 is to replace the data such as the arrangement of the repulsive elements S with a new one in this way, but the present invention is not limited to this. "" Also assumes the case of replacing data with no change. In other words, when performing visualization, it is of course possible not to change the size and arrangement of the repulsive element S for the entire time or a certain period of visualization. In such a case, even if the data is updated, There will be no real change.

  The specific processing content or function of the visualization program in the present embodiment has been described above. Incidentally, in recent years, a new field of visual data mining (VDM) has attracted attention in data analysis research. This is regarded as a visualization technology that provides visual support for trend understanding and feature extraction from data, starting from the stage of conventional information visualization, that is, simply converting data into visual representation. However, the present invention is considered to be particularly significant as a new method in such a new field, and for example, is considered suitable as a VDM technique that supports monitoring / diagnosis of electric power facilities.

  The above-described embodiment is an example of a preferred embodiment of the present invention, but is not limited thereto, and various modifications can be made without departing from the scope of the present invention. For example, in the present invention, data over a long time along the time axis, for example, continuous data over a long time such as discharge sound observation data in a power facility is an object to be visualized. It is not limited to things. That is, most of the data over a long period of time is considered to be continuous data, but not only this but also data including a discontinuous section may be present. To illustrate the processing method in such a case, for example, if the discontinuous section is a section without data, the data may be regarded as a continuous section of 0, and the discontinuous section is data from different input sources. In the case where the discontinuity is essentially discontinuous, it can be regarded that the data tendency has simply changed within the range where the number of channels is the same. Furthermore, when the number of channels changes (for example, when a channel that was 10 changes to 20), processing such as simply connecting them at the same scale is also possible. As described above, when a discontinuous section is included, it can be dealt with by appropriately considering when designing data or a program.

  Further, in the above-described embodiment, as an example of the arrangement of the inward repulsive elements S, the case where they are disposed around the outer periphery of the display area is illustrated, but of course this is only an example. S can be arranged inside the display area and around the drawing trajectory T. In these cases, there is no difference in the method itself, but various controls are possible because the functions and operations differ depending on the arrangement. In the same way, regarding how to spread the inward repulsive element S that has been placed once gradually, how to spread it, whether it should be spread with a similar shape, or whether it should be spread while being deformed? There are various methods, and various controls can be performed on the drawing trajectory T.

  With respect to the method for visualizing time series data described in the present embodiment, an evaluation experiment was conducted in order to examine what effects the three functions of stretching processing, bending processing, and repulsive action have on visualization. Specifically, the data visualization program and the visualization method according to the present invention are applied to the discharge sound observation data for 1 hour and 19 minutes, and the frequency components of the sound are used as the respective channels, and different hues from the low sound to the high sound. And an experiment to express the amplitude in each frequency band by brightness. As a result, compared with the conventional typical visualization method, this method keeps the entire image compact, while the trend of data and the point of change of characteristics (the state where the discharge sound continues while changing the intensity or It was confirmed that changes in the phenomenon can be easily identified (see FIG. 9 and the like).

  Here, the contents of FIG. 9 will be outlined. The pattern in FIG. 9 represents the frequency characteristics of the discharge sound. Further, the data is emphasized by turning the drawing trajectory T to the left in the traveling direction when the treble is strong, and turning right when the treble is strong. In addition, the discharge sound data shown in FIG. 9 is a clockwise swirl as a whole because the bass component is strong and bends to the right in the traveling direction. The dark pattern in the figure indicates that the sound is strong. As can be seen from the enlarged view, a streak is drawn from the center toward the bass, which indicates that a discharge sound is generated. In the portion indicated by an arrow A in the figure, the drawing trajectory T is a curved trajectory because the discharge sound is suddenly interrupted. In the figure, the part marked with a ★ mark is a section that is largely bent, indicating that there has been a significant tendency change in the discharge sound.

  In conducting the evaluation experiment, the data of the video image taken in the exposure test conducted for the purpose of preventing the discharge noise of the DC transmission line was targeted for visualization. This video is a part of a video image of a long-term shooting of a DC insulator station installed at the test site over a period of several months, and the sound recorded in the video is 1 m above the ground at a position 1.5 m away from the vertically installed DC insulator station. Was recorded by an omnidirectional microphone installed in This is the SS method (Spectral Subtraction method: A normal sound feature is generated from multiple normal sound data, and only the parts that differ from the normal sound feature are extracted during observation. ), And only the difference from the standard discharge sound remains.

  The records used for visualization are images from May 9th to May 16th, 2001. The shooting period is 17 days, but the time zone when no sound is detected is shortly recorded in silence with a time-lapse video (a video deck that can control the number of recorded video frames), and the actual tape length is It was 1 hour 19 minutes. For comparison, six visualization methods were used by removing No. 3 and No. 7 from the eight methods shown in Table 3. The reason why No. 3 and No. 7 are excluded is that when the bending function is not used, it is visualized along a preset route, and therefore there is no meaning of applying the repulsive element S.

The contents of the visualization of long-term discharge sound observation data are as follows:
1. Visualization software for experiments 2. Visualization of discharge sound Each item to be confirmed will be explained.

1. Visualization software for experiments Visualization software (visualization program) was created for evaluation. FIG. 10 shows an example of a user interface screen of the output device. It operates on Mac OS X, captures video and sound data from a video camera connected to a PC, performs sound signal processing such as SS method, and draws visualization images in real time using each selected visualization method It was. The visualized image can be displayed at an arbitrary magnification between 0.1 times and 2 times (see FIG. 11). In addition, the display area can be moved by clicking anywhere in the visualized image display area with the mouse. Furthermore, the created visualization software also has a function that automatically saves the captured and signal processed data in a file, reads the data later, changes the parameters, changes the visualization method, and redraws. Drawing 5 hours of signal-processed recording data using the Crawl method took about 5 minutes using a PowerBook G4 / 1GHz. Half of the processing time was drawing on the screen, the other half was repulsive calculation processing time, and other processing time was negligibly short.

2. Results of Visualization of Discharge Sounds The results of visualization with each method in this example are shown below. Each visualized image is reduced and displayed so that the long side of the displayed range is 8 cm. This assumes that the visualization result is displayed on a screen of the same size.

  FIG. 12 shows a visualized image in a strip type display (this corresponds to No. 8 in Table 4). It can be seen that the pattern is divided into dark sections and thin sections. The dark pattern is the time when the discharge sound for observation (discharge sound caused by flashover) is observed, and the light pattern has no discharge sound for observation or another sound is mistaken for the discharge sound. This is the recorded time interval. The striped pattern that can be distinguished in the enlarged view is the sound spectrum pattern. In the figure, the upper part represents a low sound and the lower part represents a high sound. It can be seen that the pattern is finely visualized throughout. By the way, the original image is color, and the striped pattern representing the discharge sound is colored brown or green, which is more discriminating than the gray scale.

  FIG. 13 shows a diagram visualized by a spiral display (this corresponds to No. 8 in Table 4). In the spiral type display, the linear velocity was drawn at the same speed as the strip type. Therefore, the drawing angle per unit time is smaller toward the outer periphery. Although it has similar characteristics to the strip type, the visualization unit U is fan-shaped, the outer periphery is widely visualized, and the inner periphery is narrowly visualized, so that the outer peripheral side (high pitch side) is emphasized. ing. It can be confirmed in the enlarged view that the spectrum is drawn finely like the strip type. In both the strip and spiral types, the area of the visualization result shows the elapsed time as it is, so it became possible to intuitively know the duration of the state.

  FIG. 14 is a view visualized by adding an elongation process to the strip type display (this corresponds to No. 4 in Table 4). The stretch process applied here is a process of changing the size of the drawing direction vector v to make the feature stand out as described above. The small value section (the thin pattern section) is shrunk, and the large value section ( It was made to work so as to extend the dark section). In this data, the thin section of the pattern is long in the middle and at the end, so the overall length was reduced to 70%. It can be seen that the pattern is easy to read because the portion with a large change is emphasized (see FIG. 14). In particular, it can be seen that the midway discharge pattern has a higher treble component than the initial discharge pattern. It can also be seen that a sound pattern different from the discharge sound appears near the end of the data (bottom row). This is the data that the observation system incorrectly records a bird cry as a discharge sound. The observation system was designed to record any sound pressure above a certain level.

  FIG. 15 is a view visualized by adding an extension process to the spiral display (this corresponds to No. 4 in Table 4). The linear velocity is the same as in FIG. Even in the swirl display, the processing by the stretching process appeared as an effect of shortening the section, and the entire length was shortened (reduced to 70% of the whole). The point that the pattern was emphasized was the same as that applied to the strip type.

  FIG. 16 shows a diagram visualized using only the bending process without using the repulsive action and the elongation process (this corresponds to No. 6 in Table 4). In this case, it is difficult to see a part of the drawing trajectory T indicating the drawing result. However, since the route represents the change in the data, it is clear that this one-hour data consists of several modes. A relatively strong discharge sound is generated at the beginning, and then it becomes a little quiet, and after a while after the discharge sound disappears, a strong discharge sound that swirls in the lower right of the screen is generated, and then the discharge sound disappears for a long time. Finally, it was confirmed that there is a recording section of a bird call that is bent on the left side of the screen, which is expressed by how the route bends (see FIG. 16). The part indicated by an arrow B in the figure represents a section in which a discharge sound due to flashover is continuously generated, and the part indicated by an arrow C in the figure represents a section in which a bird's cry is recorded incorrectly as a discharge sound. Represents.

  FIG. 17 shows a diagram visualized by using an elongation process and a bending process without using a repulsive action. Because the stretch process was used, the thin section of the pattern with no discharge sound was shortened and the figure was compressed.

  FIG. 18 shows a diagram when visualization is performed by bending and repulsive action without using the elongation treatment (this corresponds to No. 5 in Table 4). The drawing trajectory T here has a spiral shape because the repulsion element S arranged on the circumference is confined in the display area while gradually expanding. The total length of the path is the same as that of the strip type and the spiral type that do not use the elongation treatment. The area of the smallest rectangle including this figure on the inside was 2.58 (ie, 2.58 times that in FIG. 13) when the spiral-shaped visualization result (see FIG. 13) without using the elongation process was 1. That is, since there is a blank (gap) between the drawing trajectories T, the occupation area as a whole is widened. Also, a sharply bent shape (here, “a sharply bent shape” is not a regular bend as shown in FIG. 13 or 15, but is bent to the right or left or irregularly). This means that only the bending effect is used, and the repulsive element S on the outer circumference or path is obstructed when drawing is performed as in the case where the repulsive action is not used (see FIG. 16). This is because it is rarely bent. Therefore, the greater the change in data, the more frequent and steep bends are. On the other hand, when the change in data is small, the arc is drawn relatively straightforwardly. More specifically, the change in data here means both the amount of change in the data value per unit time and the number of changes per unit time. The former is, for example, the amplitude in terms of a signal wave. The latter corresponds to the frequency. In other words, when the frequency fluctuation is large, the channel having a large value is frequently changed per unit time, so that the curvature is frequently changed, which leads to the drawing of “frequent bending”. Further, in the case of FIG. 18, it was confirmed that the tendency of data hardly appears in a global form as compared with the case of FIG. 16.

  FIG. 19 shows the visualization results of the crawl method using all of the elongation treatment, bending treatment, and repulsive action (this corresponds to No. 1 in Table 4). The total length of the drawing path is the same as that of the strip type and the spiral type using the stretching process, and is 70% when the stretching process is not used. The area of the smallest rectangle including this figure on the inside was 2.83 (that is, 2.83 times that in the case of FIG. 15), assuming that the spiral shape using the elongation process (see FIG. 15) is 1. That is, since there is a blank (gap) between the drawing trajectories T, the occupation area as a whole is widened. On the other hand, it can be said that such a margin makes it possible to change the path according to the change in the data, and to make it easier to understand the trend change and the characteristics of each time interval. Specifically, it has been confirmed that the state where the first discharge sound is continuously generated with the center point as the start point is expressed as a winding path (that is, as a continuous derivation section of the discharge sound). After that, a section without a discharge sound visualized in an arc shape continues, and then a time section where a strong discharge sound continues (continuous generation section of a strong discharge sound) is long, and it is expressed by a curved path. I also confirmed that. This discharge sound suddenly stopped, and at that time (when the discharge sound suddenly stopped), it was visualized as an unnaturally bent part (lower left part in the figure). After a while, it was confirmed that the section where bird calls were recorded at the end was visualized as a jagged path (see FIG. 19). In addition, the portion indicated by an arrow D in the figure is a continuous discharge sound generation section, the part indicated by an arrow E in the figure is a continuous discharge sound generation section, and the part indicated by an arrow F in the figure is a discharge sound. When suddenly interrupted, the portions indicated by arrows G in the figure respectively represent sections where bird calls are continuously recorded.

  It was confirmed that the global visualization shape in the crawl method changes depending on the size of the repulsive action space. FIG. 20 shows how the visualization result changes when the size of the repulsive action space is changed. In the middle of the five figures, the result shown in FIG. 19 is reduced in the middle. When the repulsive action space is made wider than this, the path expands and opens as shown on the right side of FIG. It became a shape. On the other hand, when the repulsive action space was narrowed, the path narrowed as shown on the left side of the center, and vortexes were overlapped. Based on the above, it is easier to obtain a visualization result that expresses the characteristics of the data more easily when the repulsive action space is widened. It has been confirmed that the characteristics of are difficult to read. However, the visualization result when the area of the repulsive action space is changed changes so that the kinked string is stretched or shrunk, and the kinking condition itself, that is, the inflection tendency of the path itself changes depending on the size of the repulsive action space. I also confirmed that this is not the case.

3. Items to be confirmed Comparison of the visualization results of the discharge sound recording data showed the characteristics of the three elements in the present invention (elongation treatment, bending treatment, and repulsive treatment).

  First, as an advantage of the stretch processing, it has been confirmed that there is an effect that the changed portion becomes conspicuous and can be expressed in a compact manner, and that the drawing trajectory T or the time series data can be summarized. However, since changes in the time direction are distorted, it may be difficult to grasp the characteristics of the time direction such as the duration, and in such cases, if the emphasis is on elapsed time, visualization is performed without using stretch processing It can also be done.

  Next, with regard to the bending process, when repulsive action was not used together, it was confirmed that the change in the data can be expressed in a straightforward manner and the strength of the change and the degree of change can be reflected in the global shape. However, the drawing paths may overlap or extend in a specific direction. In such a case, the visibility may be reduced.

  When the bending process and the repulsive force action space are used in combination, it is confirmed that the arrangement of repulsive force element S eliminates the overlap and extension of the path, and can be visualized compactly. Even so, it was confirmed that the bends represented the characteristics of the data and that the trend of change was easy to grasp. In particular, when combined with the elongation treatment, it was easily reflected in the global shape, so it was confirmed that it has a shape feature that is likely to remain in the impression even when reduced.

  As described above, it was confirmed that the invention according to the present application is effective as a method for effectively visualizing sensor data having a long time and no structure used in equipment monitoring / diagnosis. In the crawl method described in this embodiment or the present example, the local display feature is reflected in the shape of the visualized image by bending and expanding / contracting the data display path in accordance with the change of the data, and as a result, the reduction is performed. Even so, it was possible to generate a visualized image with easy-to-understand trends and features. In addition, by using a physics analogy, it is possible to avoid overlapping and intersecting images while keeping the entire visualized image compact. In this example, it was found that the present method was applied to the discharge sound observation data for 1 hour, and it was possible to visualize it more clearly with the same screen size compared to the conventional representative visualization method.

It is a figure for demonstrating the visualization unit (rectangle) in embodiment which applied the visualization method of the time series data concerning this invention. It is a figure for demonstrating the visualization unit (fan shape) in embodiment of this invention. It is a figure for demonstrating the effect by "elongation process." It is a figure for demonstrating the effect by "bending process." It is a figure for demonstrating a mode at the time of confining the drawing locus | trajectory T to a display area by applying a repulsive force. It is a figure for demonstrating a mode that a repulsive force is applied and the overlap of the drawing locus | trajectory T is avoided. It is a figure which shows in order from (a) to (e) how the visualization unit U advances when the repulsive element S is arranged in a rectangle and the interval between the short sides increases. It is a flowchart which shows the processing content of the visualization program in this embodiment. It is a figure which shows an example of the drawing locus | trajectory T drawn by applying this invention. It is a figure which shows an example of the user interface screen in the Example of this invention. It is a figure which shows the zoom function of the visualization tool implemented in the present Example. It is a figure which shows the visualization result displayed in strip shape in a present Example. It is a figure which shows the visualization result displayed in the spiral form in a present Example. It is a figure which shows the result visualized by adding elongation processing to strip type display. It is a figure which shows the result visualized by adding the extension process to the spiral display. It is a figure which shows the result visualized using only a bending process, without using a repulsive force action and an elongation process. It is a figure which shows the result visualized using the extension process and the bending process, without using a repulsive effect. It is a figure which shows the result visualized by applying a bending process and a repulsive force effect without using an elongation process. It is a figure which shows the visualization result of the crawl method using all the elongation process, a bending process, and a repulsive effect. It is a figure which shows how a visualization result changes when the area of a repulsive force action space is changed. It is a figure for demonstrating the linear visualization method which is one of the conventional data visualization techniques. It is a figure for demonstrating the strip-shaped visualization method which is one of the conventional data visualization techniques. It is a figure for demonstrating the spiral type visualization method which is one of the conventional data visualization techniques.

Explanation of symbols

T Drawing trajectory U Visualization unit for drawing

Claims (9)

  In a method for visualizing time-series data in which data over a long time along a time axis is displayed as a spiral drawing trajectory on one screen of an output device, the data is distributed to the inside or the outside of the drawing trajectory for each component. If there is a change in the curve, the curvature of the drawing trajectory is distributed so as to change, and this new drawing is performed by acting between the already drawn drawing trajectory and the new drawing trajectory drawn continuously at the tip. A method for visualizing time-series data, wherein the data is displayed in a vortex while applying an outward repulsive force that applies a repulsive force to the locus and an inward repulsive force that applies an inward force to the new drawing locus. .   2. The method for visualizing time-series data according to claim 1, wherein the drawing trajectory of the data is reduced in the time axis direction while the drawing trajectory is not bent or small because the change in the data is small. .   The repulsive action space in which the outward repulsive force and the inward repulsive force act on the drawing trajectory is expanded to at least an extent that overlap or intersection of the drawing trajectories is avoided. Visualization method of described time series data.   4. The time-series data visualization method according to claim 1, wherein the outward repulsive force is made to act on the drawn drawing trajectory at regular intervals. 5.   In a visualization program for time-series data for displaying data over a long time along a time axis as a spiral drawing locus on one screen of an output device, the computer causes the data to be inside or outside the drawing locus for each component. Between the process of allocating so that the curvature of the drawing trajectory changes, and the already drawn drawing trajectory and the new drawing trajectory drawn continuously at the tip of the drawing trajectory. A process of applying a repulsive force to the new drawing trajectory and applying a repulsive force to the new drawing trajectory, and a process of continuously displaying the data while applying an inward repulsive force from the outside to the new drawing trajectory. A time series data visualization program.   6. The time series according to claim 5, wherein a process of reducing the drawing trajectory of the data in the time axis direction is executed while the drawing trajectory is not bent or small because the change in the data is small. Data visualization program.   The processing for expanding a repulsive action space in which the outward repulsive force and the inward repulsive force act on the drawing trajectory is performed to at least an extent that overlap or intersection of the drawing trajectories is avoided. The time-series data visualization program according to 5 or 6.   The time series data visualization program according to any one of claims 5 to 7, wherein a process of causing the outward repulsive force to act on the drawn drawing trajectory at regular intervals is executed. In a time-series data visualization system for displaying data over a long time along a time axis as a spiral drawing trajectory on one screen of an output device, the data is distributed to the inside or outside of the drawing trajectory for each component, and the data When there is a change, the data distribution means for allocating the curvature of the drawing locus to change, and the repulsive force between the already drawn drawing locus already drawn and the new drawing locus drawn at the tip thereof. An outward repulsive force applying means for applying a repulsive force to the new drawing trajectory and an inward repulsive force applying means for applying an inward force to the new drawing trajectory, and applying the outward repulsive force and the inward repulsive force The time-series data visualization system is characterized in that the data is displayed in a spiral shape.