JP2012043262A - Process status analyzer and process status analysis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a process status analyzer, etc. capable of grasping the collapse of the correlation between data.SOLUTION: Contribution degree calculation means calculates a contribution degree of each of data in a data group to the Mahalanobis distance from the reference space by using the Mahalanobis-Taguchi Method for the data group to be analyzed. Distance calculating means calculates the Mahalanobis distance from the reference space for one piece of data selected from the individual data of which the contribution degree is calculated by the contribution calculating means and another piece of data selected from the individual data using the Mahalanobis-Taguchi Method.

Description

  The present invention relates to a process state analysis apparatus that analyzes a process state using a Mahalanobis Taguchi method.

  As a method for analyzing the state of a process in process automation, there is a method of performing statistical analysis based on process data related to process automation, and one of them is a Mahalanobis Taguchi method. Process data related to process automation includes time-series data such as signals of instruments connected to the process, set values of control operations, process information, maintenance information, and disturbance information.

  In the Mahalanobis Taguchi method, a standard is set for these process data values, and the difference between the standard and the analysis target is calculated as the Mahalanobis distance. If normal is used as a reference here, normality and abnormality of the analysis target can be determined based on the calculated Mahalanobis distance. For example, if the Mahalanobis Taguchi method is applied to remarkable process data, the operating state can be quantitatively grasped, so the Mahalanobis Taguchi method is extremely effective for analyzing and improving the plant operating state. It becomes a means.

  If the Mahalanobis Taguchi method is used, the difference between the reference space to be compared and the space to be analyzed can be obtained as an MD (Mahalanobis Taguchi) value, and which parameter causes the MD value to be larger Can be found from the value of contribution required for each parameter.

JP 2009-009300 A

  When the degree of contribution of a certain process data is large, the correlation with other process data is lost even if the deviation of the process data itself is large, as well as when the deviation of the process data itself is small. It is assumed that In the former case, it can be confirmed by a method of displaying a trend graph of a past data group and a trend graph of process data to be analyzed in a superimposed manner. However, in the latter case, there is no effective method for narrowing down which correlation between the process data is broken.

  An object of the present invention is to provide a process state analysis apparatus and the like that can grasp the collapse of the correlation between data.

The process state analysis apparatus of the present invention is a process state analysis apparatus that analyzes the state of a process using the Mahalanobis Taguchi method, and applies the Mahalanobis Taguchi method to a data group to be analyzed, from its reference space. A contribution degree calculating means for calculating a contribution degree for each piece of data in the data group with respect to the Mahalanobis distance, and one data selected from the individual pieces of data for which the contribution degree is calculated by the contribution degree calculating means And a distance calculation means for calculating the Mahalanobis distance from the reference space by applying the Mahalanobis Taguchi method to the other data selected from the individual data. To do.
According to this process state analyzing apparatus, Mahalanobis Taguchi is used for one data selected from the individual data for which the degree of contribution has been calculated and one other data selected from the individual data.・ By applying the method and calculating the Mahalanobis distance from the reference space, it is possible to grasp the collapse of the correlation between the data.

  Data with a large contribution calculated by the contribution calculation means may be selected as the one data.

  A data designation accepting unit that accepts designation of arbitrary data from among the individual data whose contribution is calculated by the contribution degree calculating unit, and the one data is designated via the designation accepting unit; Data may be selected.

  The contribution degree calculating means and the distance calculating means may further include a feature quantity selection receiving means for receiving selection of a feature quantity used when applying the Mahalanobis Taguchi method.

The process state analysis method of the present invention is a process state analysis method that analyzes the state of a process using the Mahalanobis Taguchi method, and applies the Mahalanobis Taguchi method to a data group to be analyzed, from its reference space. A contribution degree calculating step for calculating a contribution degree for each piece of data in the data group with respect to the Mahalanobis distance, and one piece of data selected from the individual pieces of data for which the contribution degree has been calculated in the contribution degree calculating step And a distance calculating step of calculating a Mahalanobis distance from the reference space by applying the Mahalanobis Taguchi method to the other data selected from the individual data. To do.
According to this process state analysis method, Mahalanobis Taguchi is used for one data selected from the individual data for which the degree of contribution is calculated and one other data selected from the individual data.・ By applying the method and calculating the Mahalanobis distance from the reference space, it is possible to grasp the collapse of the correlation between the data.

  According to the process state analysis apparatus of the present invention, Mahalanobis is selected for one data selected from the individual data for which the degree of contribution has been calculated and one other data selected from the individual data.・ By applying Taguchi method and calculating Mahalanobis distance from the reference space, it is possible to grasp the collapse of correlation between data.

  According to the process state analysis method of the present invention, Mahalanobis is selected for one data selected from the individual data for which the degree of contribution has been calculated and one other data selected from the individual data.・ By applying Taguchi method and calculating Mahalanobis distance from the reference space, it is possible to grasp the collapse of correlation between data.

The block diagram which shows the structure of the field control system to which a process state analysis apparatus is applied. The flowchart which shows operation | movement of an analyzer. The figure which shows the example which performs the process of step S1-step S5 with respect to four process data AD. The figure which shows the example which computed Mahalanobis distance about the combination of the process data A and other process data. The figure which shows the feature-value of process data as a table. The figure which shows the example of a setting of a feature extraction frame.

  Hereinafter, an embodiment of a process state analysis apparatus according to the present invention will be described.

  FIG. 1 is a block diagram showing the configuration of a field control system to which the process state analysis apparatus of this embodiment is applied.

  As shown in FIG. 1, the field control system is connected to field controllers 2, 2,... Distributed in a plant and field controllers 2, 2,. And an operation monitoring device 3 for performing operation and monitoring of field device groups connected to 2, 2,.

  Further, the field bus 4 is connected with an analysis apparatus 1 as a process state analysis apparatus of the present embodiment.

  As shown in FIG. 1, the analysis apparatus 1 calculates a Mahalanobis distance from the reference space for a data group to be analyzed, and also calculates a contribution degree calculating means for each piece of data in the data group. 11, and one reference selected from the individual data whose contribution is calculated by the contribution calculation means 11, and another one of the data selected from the individual data. A distance calculating means 12 for calculating the Mahalanobis distance from the space, a designation receiving means 13 for receiving designation of arbitrary data from the individual data whose contribution is calculated by the contribution calculating means 11, and a contribution calculating means 11 and distance calculation means 12 to select the feature quantity of the data group to be analyzed, which is used when applying the Mahalanobis Taguchi method. Only a feature quantity selection receiving means 14 to attach to form a.

  The analysis device 1 includes a display device 15 for displaying calculation results and the like in the contribution calculation unit 11 and the distance calculation unit 12, and a storage unit 16 that sequentially collects process data via the field bus 4. Prepare.

  Next, the operation of the analysis apparatus 1 will be described.

  The storage unit 16 sequentially acquires and accumulates setting values (operation values) for field devices, sensor values obtained from the field devices, and other process data as time series data via the field bus 4. The analysis device 1 performs the later-described analysis using the process data accumulated in the storage unit 16.

  FIG. 2 is a flowchart showing the operation of the analysis apparatus 1.

  Steps S <b> 1 to S <b> 5 in FIG. 2 show processing of the contribution degree calculation means 11.

  In step S1 of FIG. 2, the history of the process data group to be compared with the process data to be analyzed is acquired from the storage unit 16, and the feature amount is extracted. Here, a sample line is applied to each process data to extract a feature amount. Thereby, the time series change of the process data group can be quantified. For example, a normal reference space can be created as will be described later by extracting feature values from a past process data group at normal times.

  Next, in step S2, a reference space based on the feature amount extracted in step S1 is created. For example, it is possible to compare the process data to be analyzed with the normal process data group by creating a reference space for the normal normal process data group.

  Next, in step S3, process data to be analyzed is acquired from the storage unit 16, and feature values are extracted. Here, a sample line is applied to process data to be analyzed, and a feature amount is extracted.

  Next, in step S4, the Mahalanobis Taguchi method is applied to the feature amount extracted in step S3, and the Mahalanobis distance from the reference space is calculated.

  Next, in step S5, the degree of contribution of each process data to be analyzed with respect to the Mahalanobis distance calculated in step S4 is calculated, and the process ends.

  FIG. 3 is a diagram illustrating an example in which the processes in steps S1 to S5 are performed on the four process data A to D.

  In the example of FIG. 3, a reference space is created based on a history data group of process data A to D at normal time, and a Mahalanobis distance between the reference space and the process data A to D as analysis target data is calculated. is doing. Further, the degree of contribution is calculated for each of the process data A to D, which is analysis target data.

  Steps S <b> 11 to S <b> 27 in FIG. 2 indicate processing of the distance calculation unit 12.

  In step S11 of FIG. 2, it is determined whether or not data designation is accepted via the data designation accepting unit 13. If the judgment is affirmed, the process proceeds to step S23, and if the judgment is denied, the process proceeds to step S12.

  In step S12 to step S17, a combination of process data is automatically selected, and the Mahalanobis distance is calculated for each.

  In step S12, the process data with the largest contribution calculated in step S5 is selected. In the example of FIG. 3, the process A is selected because the contribution of the process A is the largest.

  Next, in step S13, one of the other process data not selected in step S12 is selected. In the example of FIG. 3, one of the process data B, C, and D is selected.

  Next, in step S14, a reference space is created for the two process data selected in steps S12 and S13, respectively. For example, when process data A and process data B are selected, a sample line is applied to the process data group (process data A group and process data B group) to be compared, and a feature amount is extracted. A reference space for only process A and process B is created. As a result, it is possible to quantify the relationship that serves as a reference for the time series change for the process data A and the process data B.

  Next, in step S15, a sample line is applied to the process data to be analyzed, and a feature amount is extracted. For example, when process data A and process data B are selected, a sample line is applied to process data A and process data B, and a feature amount is extracted.

  Next, in step S16, the Mahalanobis distance from the reference space created in step S14 is calculated based on the feature amount extracted in step S15. For example, when process data A and process data B are selected, Mahalanobis distances for process data A and process data B are calculated.

  Next, in step S17, it is determined whether or not steps S13 to S16 have been executed for all process data other than the process data selected in step S12. If the determination is affirmative, the process is terminated. If negative, the process returns to step S13. For example, if the process data A is selected in step S12, the process ends when steps S13 to S16 are executed for the process data B, C, and D.

  On the other hand, in step S23, two data designated via the data designation accepting means 13 are selected. The user can designate arbitrary process data via the data designation accepting means 13. For example, process data A and process data B can be designated.

  Next, in step S24, a reference space is created for the two process data selected in step S23. For example, when process data A and process data B are selected, a sample line is applied to the process data group (process data A group and process data B group) to be compared, and a feature amount is extracted. A reference space for only process A and process B is created. As a result, it is possible to quantify the relationship that serves as a reference for the time series change for the process data A and the process data B.

  Next, in step S25, a sample line is applied to the process data to be analyzed, and a feature amount is extracted. For example, when process data A and process data B are selected, a sample line is applied to process data A and process data B, and a feature amount is extracted.

  Next, in step S26, the Mahalanobis distance from the reference space created in step S24 is calculated based on the feature amount extracted in step S25. For example, when process data A and process data B are selected, Mahalanobis distances for process data A and process data B are calculated.

  Next, in step S27, it is determined whether or not the user has instructed the end of the process. If the determination is affirmed, the process ends. If the determination is negative, the process returns to step S11.

  As described above, by calculating the Mahalanobis distance between two process data by the process shown in FIG. 2, it becomes possible to find a combination of process data whose correlation is broken. Specifically, it is possible to determine that the correlation is broken for a combination having a large Mahalanobis distance, and it is possible to quantify the degree of the broken correlation between the process data in the form of the Mahalanobis distance.

  FIG. 4 is a diagram illustrating an example in which the Mahalanobis distance is calculated for the combination of the process data A and other process data in steps S11 to S27. In the example of FIG. 4, Mahalanobis distances are calculated for process data A and process data B, process data A and process data C, and process data A and process data D, respectively. By comparing these Mahalanobis distances, a combination in which the correlation is broken can be grasped.

  Next, an example will be described in which the user's designation regarding the feature amount is received via the feature amount selection receiving unit 14.

  In general, the calculation of the contribution is performed for each element of the feature amount such as the change amount or the existence amount of each process data. For example, among the contributions of the feature amounts belonging to the process data, the largest one can be shown as the contribution of the process data.

  Therefore, for example, when investigating the correlation between the process data A and the process data B, a general method of creating a reference space using all the feature values of the process data A and the process data B may be applied to the present invention. However, it is also possible to create a reference space only with specific feature amounts of the process data A and the process data B and apply the Mahalanobis Taguchi method. As a result, for example, it is possible to analyze and specify the cause in more detail, for example, what part of the process data in which time zone has changed which caused a deviation from the normal state.

  FIG. 5 is a diagram showing the feature amount of the process data as a table. For example, feature quantities A1, A2,... Exist as feature quantities of the process data A, and feature quantities B1, B2,. In this case, the user can select an arbitrary feature amount via the feature amount selection receiving unit 14. For example, as shown in FIG. 4, when examining the correlation between the process data A and the process data B, the feature amount A2 and the feature amount B3 can be designated. The types of feature amounts may be different. For example, the amount of change in process data A and the amount of process data B may be specified. It is also possible to specify different feature quantities for the same process data. For example, by specifying the change amount of the process data A and the existing amount of the process data A, the correlation between the two can be analyzed.

  The user can also specify a feature extraction frame via the feature amount selection accepting unit 14. Thereby, the reference space can be created by combining the feature extraction frames according to the analysis purpose, and the Mahalanobis Taguchi method can be applied.

  For example, when analyzing batch data using the Mahalanobis Taguchi method, the plant improvement staff sets up the required number of feature extraction frames at the notable locations in the process data waveform (trend waveform). Feature quantities can be extracted.

  FIG. 6 is a diagram illustrating an example of setting a feature extraction frame.

  In the example of FIG. 6, a plurality of feature extraction frames (feature extraction frames P, Q, R) are set for certain process data. For example, if the feature extraction frame P is set to capture the rise of process data and the feature extraction frame Q is set to capture the fall of process data, a combination of the feature extraction frame P and the feature extraction frame Q is used. Thus, it is possible to analyze the balance between rising and falling of the process data. For example, it is possible to analyze the difference between the normal state and the balance at the time of abnormality.

  As described above, according to the process state analysis apparatus and the process state analysis method of the present invention, one data selected from the individual data for which the degree of contribution has been calculated and the individual data are selected from the individual data. Since the Mahalanobis Taguchi method is applied to the other data and the Mahalanobis distance from the reference space is calculated, it is possible to grasp the collapse of the correlation between the data.

  The scope of application of the present invention is not limited to the above embodiment. The present invention can be widely applied to a process state analysis apparatus that analyzes a process state using the Mahalanobis Taguchi method.

DESCRIPTION OF SYMBOLS 11 Contribution calculation means 12 Distance calculation means 13 Specification reception means 14 Feature quantity selection reception means

Claims (5)

In a process state analyzer that analyzes the state of a process using the Mahalanobis Taguchi method,
For the data group to be analyzed, contribution calculation means for applying the Mahalanobis Taguchi method and calculating the contribution of each data of the data group to the Mahalanobis distance from the reference space;
Mahalanobis Taguchi method for one data selected from the individual data whose contribution is calculated by the contribution calculating means and another data selected from the individual data And a distance calculation means for calculating the Mahalanobis distance from the reference space,
A process state analyzing apparatus comprising: The process state analysis apparatus according to claim 1, wherein data having a large contribution calculated by the contribution calculation unit is selected as the one data. A data designation accepting means for accepting designation of arbitrary data from the individual data whose contribution is calculated by the contribution calculating means;
The process state analysis apparatus according to claim 1, wherein the data designated through the designation receiving unit is selected as the one data. The feature amount selection accepting unit for accepting selection of a feature amount used when applying the Mahalanobis Taguchi method in the contribution degree calculating unit and the distance calculating unit. The process state analysis apparatus according to item. In the process state analysis method that analyzes the state of the process using the Mahalanobis Taguchi method,
For the data group to be analyzed, the Mahalanobis Taguchi method is applied, and the contribution calculation step of calculating the contribution of each data of the data group to the Mahalanobis distance from the reference space;
Mahalanobis Taguchi method for one data selected from the individual data whose contribution is calculated by the contribution calculation step and another one data selected from the individual data And calculating a Mahalanobis distance from the reference space,
A process state analysis method comprising: