JP2010250366A - Apparatus and method for processing information, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract process capability from time-series data without setting any specification value. <P>SOLUTION: An information processing apparatus includes: a variation coefficient calculation part for calculating a first variation coefficient by dividing the standard deviation of time-series data in a first period just before a certain time by the movement average, and for calculating a second variation coefficient by dividing the standard deviation of the time-series data in a second period shorter than the first period just before the time by the movement average; a variation coefficient calculation part for calculating a variation coefficient ratio by dividing the first variation coefficient by the second variation coefficient; and a process capability evaluation part for evaluating a process capability for the process characterized by the time-series data by referring to the variation coefficient ratio. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an information processing device, an information processing method, and a program, and more particularly, to an information processing device, an information processing method, and a program for trend management.

  In the LSI manufacturing factory, when the standard width is too wide compared to the variation range (actual value) of the measurement value, a sign of a decrease in yield is overlooked and an accident may occur.

  In many factories as well as LSI manufacturing factories, for example, process capability index (Process Capability Index, PCI) or trend management by 6 sigma is performed. The process capability index is expressed as Cp or Cpk. The former is used to evaluate the process capability index when the deviation of the distribution center of the measurement value is small and the latter is large. Further, sigma in 6 sigma represents standard deviation σ.

For the process capability index Cp, when the upper limit (standard upper limit) of the standard value for the measurement value is SU, the lower limit value (standard lower limit) of the standard is SL, the average is AV, and the standard deviation is σ,
Cp = (SU−SL) / 6σ
Given by. In the one-sided standard, the process capability index Cp is
Cp = | SU−AV | / 3σ or | AV−SL | / 3σ
Given by.

Furthermore, when the standard center and the average value of the distribution do not match in the two-sided standard, the process capability index Cpk is
Cpk = (1-K) Cp
Given by. Here, K = | (SU + SL) −2AV | / (SU−SL).

  In the above equation, Cp = 1 corresponds to the management level 3σ, Cp = 1.33 corresponds to the management level 4σ, and Cp = 1.67 corresponds to the management level 5σ. Since the process capability index is in a trade-off relationship with the management cost, it has been considered that setting Cp to 1.67 or more is excessive management, but in recent years, aiming to set Cp to 1.67 or more 6 Sigma management (Cp = 2) has also been proposed.

  Patent Document 1 describes a system and method for detecting an abnormality inside a computerized system or business process.

  Patent Document 2 describes a plant monitoring device that automatically sets a monitoring reference for plant data based on plant data stored in the past.

JP 2004-348740 A JP-A-11-095833

  The following analysis was made by the present inventors. The above management method has the following problems.

  First, although it is necessary to appropriately set the upper limit SU of the standard value and the lower limit SL of the standard value at the beginning of the trend management, it is difficult to set these appropriately.

  When these standards are initially set in a mass production factory, they are often set with reference to the actual values at the time of the latest trial production. However, the population differs between prototypes and mass-produced products. Similarly, in the case of setting with reference to the ability values of similar products being mass-produced at the factory, the populations are also different. Therefore, it becomes difficult to set the management standard appropriately in the early stage of mass production.

When the population follows a normal distribution (Gaussian distribution), the population is represented by a combination of an average value and variation. Here, the variation is represented by standard deviation σ or variance σ ² .

  Second, when the process capability index is improved and the variation (for example, standard deviation σ) becomes small, it is necessary to review the standard value, but it is difficult to properly review the standard value.

  As an example, as is known as the Toyota production system, improvement activities are always carried out in factories, and the manufacturing variation is improved with time as the production shifts from the initial stage of mass production to full-scale mass production. As the manufacturing variation improves, the gap between the standard width set in the initial stage of mass production and the actual value width increases. In such a case, even if a variation at the time of occurrence of a process trouble is added to the actual value, it falls within the standard range, and as a result, the process trouble is overlooked.

  Furthermore, when there are a plurality of measurement values to be managed, it is necessary to review other measurement value standards having a correlation with the measurement value when reviewing the standard of a measurement value. However, it is difficult to predict the presence or absence of such correlation and the magnitude of the correlation in advance. As described above, it is difficult to appropriately review the standard values for all the measurement values to be managed.

  Thirdly, the task of reviewing the management standard is a burden on engineers.

  As an example, in a trend management system that centrally manages many types of measurement values, the number of measurement values is 1,000, the standard review frequency of measurement values is once a month, and the target product types are ten. At this time, the trend management system needs to be reviewed about 100,000 times a year. Therefore, it is a heavy burden for engineers to review the management standards as appropriate in the factory.

  If these problems are not properly solved, the standard range is too wide compared to the variation range (actual value) of the measured values, and a sign of a decrease in yield is overlooked, and an accident may occur. The situation is the same in factories other than the LSI manufacturing factory. Also, in the latter half of mass production, the daily production increases compared to the initial stage of mass production, and the target yield is set high to reduce the selling price. Even if it is the same, a large economic loss occurs.

  Therefore, it becomes an issue to be able to extract process capability from time series data without setting a standard value. The objective of this invention is providing the information processing apparatus, the information processing method, and program which solve this subject.

An information processing apparatus according to the first aspect of the present invention provides:
The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the second coefficient shorter than the first period and immediately before the time is calculated. A coefficient of variation calculation unit that calculates a second coefficient of variation by dividing the standard deviation of the time series data over the period by the moving average;
A variation coefficient ratio calculating unit that calculates a variation coefficient ratio by dividing the first variation coefficient by the second variation coefficient;
A process capability evaluation unit that evaluates the process capability for the process characterized by the time series data with reference to the coefficient of variation ratio.

An information processing method according to the second aspect of the present invention includes:
The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the second coefficient shorter than the first period and immediately before the time is calculated. Calculating a second coefficient of variation by dividing the standard deviation of the time series data over the period by the moving average;
Calculating a coefficient of variation ratio by dividing the first coefficient of variation by the second coefficient of variation;
Referring to the coefficient of variation ratio and evaluating the process capability for the process characterized by the time series data.

The program according to the third aspect of the present invention is:
The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the second coefficient shorter than the first period and immediately before the time is calculated. A process of calculating the second coefficient of variation by dividing the standard deviation of the time series data over the period by the moving average;
A process of calculating a coefficient of variation ratio by dividing the first coefficient of variation by the second coefficient of variation;
With reference to the coefficient of variation ratio, the computer is caused to execute a process for evaluating the process capability for the process characterized by the time series data.

  According to the information processing apparatus, the information processing method, and the program according to the present invention, it is possible to extract process capability from time series data without setting a standard value.

It is a block diagram which shows the structure of the information processing apparatus which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the hardware constitutions of a computer. It is a block diagram which shows the structure of the information processing apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows operation | movement of the information processing apparatus which concerns on the 2nd Embodiment of this invention. It is a figure for demonstrating the calculation method of the variation coefficient ratio in the 2nd Embodiment of this invention. It is a figure which shows an example of the output of the information processing apparatus which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the information processing apparatus which concerns on the 3rd Embodiment of this invention. It is a flowchart which shows operation | movement of the information processing apparatus which concerns on the 3rd Embodiment of this invention. It is a figure for demonstrating operation | movement of the information processing apparatus which concerns on the 3rd Embodiment of this invention.

  The information processing apparatus in the first development form is preferably the information processing apparatus according to the first viewpoint.

  In the information processing apparatus of the second development form, the process capability evaluation unit determines that the process capability is improving when the variation coefficient ratio is greater than 1, and the variation coefficient is smaller than 1. It is preferable to determine that the process capability is decreasing.

  In the information processing apparatus of the third development form, the variation coefficient ratio calculation unit may calculate the variation coefficient ratio for a plurality of times.

  The information processing apparatus according to the fourth development mode calculates the coefficient of variation ratio while changing the length of the first period and the length of the second period, and makes the time variation of the coefficient of variation ratio close to a step shape It is preferable to further have a change part.

  In the information processing apparatus of the fifth development form, it is preferable that the process capability evaluation unit issues a warning when the variation coefficient ratio becomes equal to or less than a predetermined threshold value.

  In the information processing device according to the sixth embodiment, after the coefficient of variation calculation unit adds the minute time-series data with a finite standard deviation and zero moving average to the time-series data in the second period The second variation coefficient may be calculated.

  The information processing apparatus according to the seventh development form preferably further includes a display unit that displays time-series data and a variation coefficient ratio.

  In the information processing apparatus according to the eighth development form, it is preferable that the moving average is at least one of a simple moving average, a load moving average, an exponential moving average, and an exponential weighted moving average.

  The information processing method according to the ninth development form is preferably the information processing method according to the second viewpoint.

  In the information processing method of the tenth development mode, when the variation coefficient ratio is greater than 1, it is determined that the process capability is improving, and when the variation coefficient is smaller than 1, the process capability is It is preferable that the method further includes a step of determining that it is decreasing.

  The program in the eleventh development form is preferably a program according to the third viewpoint.

  The program of the twelfth development mode determines that the process capability is improving when the variation coefficient ratio is greater than 1, and the process capability is decreased when the variation coefficient is smaller than 1. It is preferable to further cause the computer to execute processing for determining that there is a certain item.

(Embodiment 1)
An information processing apparatus according to a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of an information processing apparatus 10 according to the present embodiment.

  Referring to FIG. 1, the information processing apparatus 10 includes a variation coefficient calculation unit 11, a variation coefficient ratio calculation unit 12, and a process capability evaluation unit 13.

  The variation coefficient calculation unit 11 calculates the first variation coefficient by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and immediately before the time and the first The second variation coefficient is calculated by dividing the standard deviation of the time series data over the second period shorter than the period by the moving average.

  The variation coefficient ratio calculation unit 12 calculates the variation coefficient ratio by dividing the first variation coefficient by the second variation coefficient.

  The process capability evaluation unit 13 refers to the coefficient of variation ratio and evaluates the process capability for the process characterized by the time series data. For example, the process capability evaluation unit 13 determines that the process capability is improving when the variation coefficient ratio is larger than 1, and the process capability decreases when the variation coefficient is smaller than 1. Determine that it is on the way.

  FIG. 2 is a block diagram showing a hardware configuration of the computer when the information processing apparatus 10 is realized by a computer. Referring to FIG. 2, the computer 70 includes a CPU 71, a memory 72, a hard disk 73, an input device 74, and an output device 75.

  Each of these units may be connected to a bus line. The input device 74 may include a mouse and a keyboard. The output device 75 may have a display. The hard disk 73 may store a program and time series data. The CPU 71 executes processing in each unit of the information processing apparatus 10.

(Embodiment 2)
An information processing apparatus according to a second embodiment of the present invention will be described with reference to the drawings. FIG. 3 is a block diagram illustrating a configuration of the information processing apparatus according to the present embodiment. Referring to FIG. 3, the information processing device 20 includes a data acquisition unit 21, a data storage unit 22, a noise addition unit 23, an error avoidance unit 31, a variation coefficient calculation unit 32, a variation coefficient ratio calculation unit 24, a display unit 25, and And a process capability evaluation unit 26.

The information processing apparatus 20 implements trend management based on the coefficient of variation ratio R _CV . Here, the variation coefficient ratio R _CV is a ratio of the variation coefficient CV _{L in} the long period to the variation coefficient CV _S in the short period. The variation coefficient CV is a ratio of the standard deviation σ to the moving average MA.

  FIG. 4 is a flowchart showing the operation of the information processing apparatus 20 according to the present embodiment. With reference to FIG. 4, operation | movement of each part of the information processing apparatus 20 is demonstrated.

  Referring to FIG. 4, the data acquisition unit 21 acquires time-series data (moving average> 0) over a predetermined period (step S11).

  The data storage unit 22 stores the acquired time series data (step S12).

The noise adding unit 23 generates minute random noise (standard deviation is finite) that becomes zero (average is zero) when integrated over a certain period, and the time series data described above as necessary. (Step S13). As a result, it is possible to avoid an error (for example, the variation coefficient CV _S or the standard deviation σ _S becomes zero) when calculating the variation coefficient ratio R _CV for time-series data in which the same value continues. Note that it is preferable that the noise amplitude be a small value less than the detection limit so as not to affect the analysis accuracy of the original time-series data.

The error avoiding unit 31 avoids an error that may occur during calculation on time series data. That is, the error avoidance unit 31 avoids moving average MA _S in the calculation of the coefficient of variation CV _S becomes zero, avoids moving average MA _L is zero in the calculation of the coefficient of variation CV _L, coefficient of variation In calculating the ratio R _CV , it is avoided that the coefficient of variation CV _S becomes zero.

The coefficient of variation calculator 32 calculates the coefficient of variation CV _S = σ _S / MA _S based on the moving average MA _S and the standard deviation σ _S for the short-term time series data including the most recent data corrected by the error avoidance unit 31. Is calculated. Variation coefficient calculation unit 32, modified by the error avoidance section 31, varies based on the moving average MA _L and standard deviation sigma _L for long-term time-series data that does not include the most recent data coefficient CV _{L =} σ L _/ MA L Is calculated. The variation coefficient ratio calculation unit 24 calculates a variation coefficient ratio R _CV = CV _L / CV _S which is a ratio between the variation coefficient CV _L and the variation coefficient CV _S (step S14).

  The display unit 25 outputs the time series data and the transition of the calculation result as a graph or a table (step S15).

The process capability evaluation unit 26 determines whether or not the coefficient of variation ratio R _CV falls within the management standard (step S16). As an example, the management standard for the variation count ratio _RCV may be 1. If the transition of the data group deviates from the management standard (Yes in step S16), the process capability evaluation unit 26 issues a warning (step S17).

The process capability evaluation unit 26 automatically determines signs of variation in the time series data, so that the time series data is in a stable state if the variation coefficient ratio R _{CV is} approximately 1, and if it is greater than 1, the time series data The variation is becoming smaller, and if it is smaller than 1, it is determined that the variation of the time series data is becoming larger.

FIG. 5 is a diagram for explaining a method of calculating the coefficient of variation ratio R _CV in the present embodiment. In the present embodiment, instead of the conventional process capability index Cp = (SU-SL) / standard width of 6σ (SU-SL), using the coefficient of variation CV _L for the time-series data over a long period.

A specific calculation method will be described below. Time-series data _{D t-p} contained in the interval p immediately before excluding the measured value _{D t} at time _{t, D t-p + 1} , ..., the moving average of _{D t MA L = (D t} -p + D t-p + 1 + ... + D _t-1 ) / p and the standard deviation σ _L of the time series data D _t-p , D _{t-p + 1} ,..., D _{t-1 included in} the interval p are obtained, and the coefficient of variation CV in the interval p seek the _L = _{σ L} / MA _L.

For example, the section p may be a maintenance interval of the manufacturing apparatus related to the measurement value D. The coefficient of variation CV _L is a constant value when the process capability index is constant over a long period of time, and when the process capability index changes, it shows a smooth change that is a feature of the moving average. Less susceptible to short-term fluctuations (noise) in D.

In the present embodiment, instead of the standard deviation σ of the conventional process capability index Cp = (SU-SL) / 6σ, using the coefficient of variation CV _S for time-series data over a short period.

Specifically, the moving average MA _{S of the} time series data D _{t−q + 1} , D _{t−q + 2} ,..., D _t included in the section q immediately before the time t (where q << p) = ( with obtaining the _{D t-q + 1 + D} t-q + 2 + ... + D t) / q, the time-series data _{D t-q + 1, D} t-q + 2 contained in the section q, ..., a standard deviation sigma _S of _{D t,} section The coefficient of variation CV _S (= σ _S / MA _S ) at q is obtained. The interval q is preferably selected so as to include at least about several tens of data so as not to be affected by short-term fluctuations.

Short-term coefficient of variation CV _S is not influenced by the short-term fluctuation of the measurement values of the most recent measurement value group as the population, and the noise, suitable for trend analysis of measurement variation.

The process capability evaluation unit 26 monitors the transition of the variation coefficient ratio R _CV = CV _L / CV _S which is the ratio of the long-term variation coefficient CV _L to the short-term variation coefficient CV _S. As an example, when the variation coefficient ratio R _CV = 1 is used as a management standard, a change in the process capability index is detected based on whether or not the variation coefficient ratio is 1 or more.

The process capability evaluation unit 26 determines that the process capability is kept constant when the variation coefficient ratio R _CV = 1, and the process capability is determined when the variation coefficient ratio R _CV > 1. If it is determined that the coefficient of variation is R _CV <1, it is determined that the process capability is decreasing. As an example, when the variation coefficient ratio R _CV <0.75, it may be determined that there is an abnormality in the process and a warning may be issued.

  The set values for the long period and the short period depend on the maintenance cycle of the processing apparatus, and do not depend on the type of processing target such as LSI. Further, the maintenance cycle of the processing apparatus often corresponds to any of 24 hours, 7 days, 4 weeks, or 1 year.

  For example, in the case of a device with a short maintenance cycle, the long period is 7 days, the short period is 24 hours, and in the case of a device with a long maintenance cycle, the long period is 4 weeks and the short period is 7 days. Also good. These periods can be set regardless of the type of measurement value. Therefore, for example, even when the number of types of measurement values is about 1000, the frequency of reviewing the standard values of measurement values is once a month, and the number of target varieties is several tens of types, an apparatus with a long maintenance cycle in advance By grouping a group and a short apparatus group, the trend management by the information processing apparatus 20 of the present embodiment becomes possible.

Measurement values may be acquired sequentially, and calculation of the coefficient of variation ratio R _CV may be started when the number of measurement values reaches a predetermined number. Even after the calculation of the coefficient of variation ratio R _CV is started, the measured values are sequentially acquired, so that the distribution shape of the measured values is the same regardless of the long term or the short term while the apparatus is stable. _CV approaches 1.

When the maintenance timing of the apparatus is reached and the apparatus return operation after maintenance is not appropriate, discontinuous changes occur in the measured values before and after the maintenance. Short-term time-series data the most recent past (i.e., immediately after maintenance) the ratio of the measured value of from greater compared to long-term time-series data, the greater short term variation coefficient CV _S. Therefore, the denominator of R _CV is increased and R _CV is smaller than 1.

For example, if the rate of change becomes extremely large (e.g., when a R _{CV <0.75),} by determining an abnormality, it is possible to emit automatically alarm.

  Further, when the process capability is improved by the improvement activity, the variation range of the long-term measurement value is reduced, and thus the system becomes sensitive to the short-term variation. Increasing the sensitivity of the system to measurement value anomalies as the process capability index increases is equivalent to automatically becoming more stringent.

  In addition, when the process capability is reduced due to malfunction of the apparatus, the long-term measurement value variation range becomes large, so the system becomes insensitive to short-term variation. A decrease in the sensitivity of the system to an abnormal measurement value in accordance with the sluggishness of the process capability index corresponds to an automatic loosening of the standard. As a result, it is possible to prevent the alarms from being emitted.

FIG. 6 is a diagram illustrating an example of the output of the information processing apparatus 20 according to the present embodiment. Referring to FIG. 6, the display unit 25 outputs the variation coefficients CV _S and CV _L and the variation coefficient ratio R _CV as a graph in addition to the time series data. Note that the coordinate axes for the variation coefficients CV _S and CV _L are drawn with the increasing direction downward on the right side of FIG. Therefore, when the graph of the coefficient of variation CV _S or CV _L is in a downward trend, it is suggested that the process capability tends to decrease as in the case where the coefficient of variation ratio R _CV is in a downward trend.

(Embodiment 3)
An information processing apparatus according to a third embodiment of the present invention will be described with reference to the drawings. FIG. 7 is a block diagram illustrating a configuration of the information processing apparatus 30 according to the present embodiment. Referring to FIG. 7, the information processing apparatus 30 according to the present embodiment has a configuration in which a period changing unit 28 is added to the information processing apparatus 20 of the second embodiment.

  In the second embodiment, a predetermined short period and long period are used. On the other hand, in this embodiment, a combination of a plurality of periods is used as a short period and a long period, and a period suitable for evaluating the process capability is extracted.

In order to identify the cause of variation of each measurement value group, when the variation coefficient ratio R _CV is used, a combination of a plurality of periods is used as a short period and a long period, and the transition of the obtained variation coefficient ratio R _CV is compared to form a staircase pattern. Select a period that gives a graph showing the transition of When the combination of the selected periods and the cycle of the operation in the manufacturing process are divisors or multiples of each other, it is estimated that the operation is a direct or indirect cause for the variation of the measurement value.

  FIG. 8 is a flowchart showing the operation of the information processing apparatus 30 according to the present embodiment. Steps S11 to S17 in the flowchart of FIG. 8 are the same as steps S11 to S17 in the second embodiment.

In the present embodiment, when the coefficient of variation ratio R _CV does not exceed the standard value (No in step S16), the period changing unit 28 changes the length of the long period and the length of the short period. while calculating the variation coefficient ratio R _CV, temporal change of the variation coefficient ratio is made to be stepped (step S18).

When the restoration work after the maintenance of the apparatus is inappropriate, all of the measurement value groups up to the next maintenance show a distribution different from that at the normal time. When the maintenance cycle and the section for calculating the moving average are the same, only the time series data in a specific maintenance period is included in the section. On the other hand, the time periods before and after this include time-series data in different maintenance periods. When the distribution of the time series data in the adjacent maintenance period is greatly different, the variation of the coefficient of variation ratio R _CV becomes steep before and after the maintenance period is switched. Therefore, the coefficient of variation ratio R _CV shows a stepwise temporal change. . When the cycle of the variation coefficient ratio _RCV changing stepwise coincides with the maintenance cycle of the device, it is possible to detect a lack of maintenance of the device. Further, by investigating the work performed in a cycle when the coefficient of variation ratio R _CV is stepped, it can be used to identify an inappropriate work.

FIG. 9 is a diagram for explaining the operation of the information processing apparatus 30 according to the present embodiment. FIG. 9 shows a case where the period of variation is specified by optimizing the long-term and short-term moving average combination in the variation coefficient ratio R _CV .

FIG. 9A shows the coefficient of variation ratio R _CV when the long period for calculating the coefficient of variation CV _L is 30 days and the short period for calculating the coefficient of variation CV _S is 10 days. On the other hand, FIG. 9B shows the coefficient of variation R _CV when the long period for calculating the coefficient of variation CV _L is 56 days and the short period for calculating the coefficient of variation CV _S is 14 days. .

In calculating the coefficient of variation ratio R _CV for the same time series data, the moving average interval was a multiple of 7, and as a result, the transition of the graph was stepped. From the results shown in FIG. 9, it was found that the cause of fluctuation had a cycle of 4 weeks.

  When trying a combination of a plurality of periods, a plurality of computers may be used, and a computer may be assigned for each combination of periods to perform parallel calculation, thereby reducing the calculation time.

(Embodiment 4)
In the second and third embodiments, the case where the time series data follows a normal distribution (Gaussian distribution) is described. On the other hand, even if it is known that the measured value group follows another distribution (Poisson distribution, negative binomial distribution, etc.), by assuming that the distribution of time series data in the short and long term is the same, The same trend analysis as in the above embodiment can be performed. In other words, in comparison with the distribution of time series data over a long period of time, whether or not the distribution of time series data over a short period of time has greatly fluctuated. If it is decreasing, and it is decreased, it is determined that the process capability is improving.

  As the moving average, instead of using the simple moving average, a load moving average, an exponential moving average, or an exponential weighted moving average may be used. According to the present invention, even when the latter moving average is used, the same effects as when the former simple moving average is used can be exhibited. However, in comparison with the former case where the simple moving average is used, the calculation required for calculating the moving average becomes more complicated when the latter moving average is used, and the calculation takes time.

  Similar to the process capability index Cp, the same replacement as in the second embodiment can be performed for the process capability index Cpk having a biased distribution shape.

For time-series data in which the standard deviation σ _S in the short-term moving average is known not to be zero, the noise adding unit 23 that adds noise to the time-series data can be omitted.

  If the time series data contains both positive and negative values, all by adding a positive constant to the measurement value included in the time series data or squaring the measurement value included in the time series data The method of the present invention may be applied after setting a positive value.

Moreover, in said embodiment, it described about the variation coefficient ratio _RCV regarding a time-axis. On the other hand, the method according to the present invention can also be applied to the coefficient of variation ratio R _CV with the spatial coordinates as an axis (for example, the distance or angle from the starting point).

  As an example, the method of the present invention can be applied when evaluating communication quality within a section according to the distance from the start point of a communication network.

For example, the distance is set to 10 km instead of the short period, and the distance is set to 100 km instead of the long period, and the address is defined every 1 km. At this time, measure the level of the signal for evaluation, the noise level, or the number of simultaneous accesses (collision) for each address, calculate the coefficient of variation from the moving average and standard deviation of the measured values, and calculate the coefficient of variation The coefficient of variation ratio R _CV is obtained based on the above. In this case, it is possible to monitor and specify a signal degradation point, a noise source, and an access concentration section in a specific area of the communication network. For example, it is possible to automatically monitor the communication network by accessing each address every hour and automatically performing the above measurement over the entire length.

  Conventional trend management systems provide standard widths for numerical data having units (dimensions) such as dimensions, angles, number, flow rate, current value, capacity, voltage, weight, temperature, pressure, or humidity. Manage by setting. Therefore, it is necessary to set a standard width for each measurement value. In addition, the standard width set in the initial stage of mass production has to be reviewed as appropriate in accordance with the improvement in process capability.

  According to the information processing apparatus according to the present invention, the standard width automatically changes in accordance with the change in process capability without depending on the type of each measurement value. Therefore, it is possible to prevent a process abnormality that causes a large yield reduction from being overlooked. Furthermore, according to the information processing apparatus according to the present invention, it is possible to specify the cycle of the work that caused the process abnormality.

  Although the above description has been made based on the embodiment, the present invention is not limited to the above embodiment.

10, 20, 30 Information processing apparatus 11, 32 Variation coefficient calculation unit 12, 24 Variation coefficient ratio calculation unit 13, 26 Process capability evaluation unit 21 Data acquisition unit 22 Data storage unit 23 Noise addition unit 25 Display unit 28 Period change unit 31 Error avoidance unit 70 Computer 71 CPU
72 Memory 73 Hard Disk 74 Input Device 75 Output Device

Claims (12)

The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the first variation coefficient is shorter than the first period and immediately before the first time. A variation coefficient calculation unit that calculates the second variation coefficient by dividing the standard deviation of the time-series data over the period of 2 by the moving average;
A variation coefficient ratio calculating unit that calculates a variation coefficient ratio by dividing the first variation coefficient by the second variation coefficient;
An information processing apparatus comprising: a process capability evaluation unit that refers to the variation coefficient ratio and evaluates a process capability for a process characterized by the time series data.   The process capability evaluation unit determines that the process capability is improving when the variation coefficient ratio is greater than 1, and the process capability decreases when the variation coefficient is less than 1. The information processing apparatus according to claim 1, wherein the information processing apparatus is determined to be present.   The information processing apparatus according to claim 1, wherein the variation coefficient ratio calculation unit calculates the variation coefficient ratio for a plurality of times.   The apparatus further includes a period changing unit that calculates the coefficient of variation ratio while changing the length of the first period and the length of the second period, and makes the temporal change of the coefficient of variation ratio approach a step shape. The information processing apparatus according to claim 3.   The information processing apparatus according to claim 1, wherein the process capability evaluation unit issues a warning when the coefficient of variation ratio becomes equal to or less than a predetermined threshold.   The variation coefficient calculation unit adds the minute variation of time series data having a finite standard deviation and a moving average of zero to the time series data in the second period, and then calculates the second variation coefficient. The information processing apparatus according to claim 1, wherein the information processing apparatus calculates the information processing apparatus.   The information processing apparatus according to claim 1, further comprising a display unit that displays the time series data and the variation coefficient ratio.   The information processing apparatus according to any one of claims 1 to 7, wherein the moving average is at least one of a simple moving average, a load moving average, an exponential moving average, and an exponential weighted moving average. The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the first variation coefficient is shorter than the first period and immediately before the first time. Calculating a second coefficient of variation by dividing the standard deviation of the time-series data over a period of 2 by its moving average;
Calculating a coefficient of variation ratio by dividing the first coefficient of variation by the second coefficient of variation;
Referring to the coefficient of variation ratio and evaluating a process capability for a process characterized by the time series data.   When the coefficient of variation ratio is greater than 1, it is determined that the process capability is improving, and when the coefficient of variation is less than 1, the step of determining that the process capability is decreasing. The information processing method according to claim 9, further comprising: The first variation coefficient is calculated by dividing the standard deviation of the time-series data over the first period immediately before a certain time by the moving average, and the first variation coefficient is shorter than the first period and immediately before the first time. A process of calculating the second coefficient of variation by dividing the standard deviation of the time-series data over the period of 2 by the moving average;
A process of calculating a coefficient of variation ratio by dividing the first coefficient of variation by the second coefficient of variation;
A program that causes a computer to execute a process of referring to the coefficient of variation ratio and evaluating a process capability for a process characterized by the time-series data.   When the coefficient of variation ratio is greater than 1, it is determined that the process capability is improving, and when the coefficient of variation is less than 1, a process of determining that the process capability is decreasing is performed. The program according to claim 11, further executed by a computer.

Images