JP2905103B2 - Data analyzer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data analysis apparatus for analyzing data from a plurality of signal sources, and more particularly to a quality control apparatus for managing the quality of a product. When a causal relationship between measurement data (a plurality of channels) from a plurality of signal sources is generated, causal relationship data extracted from the causal relationship between a set of channels is used as a causal relationship between other sets of channels. The present invention relates to a data analysis device that can be used for extraction of data.

[0002]

2. Description of the Related Art In the industry, there is a demand for a quality control device for controlling the quality of manufactured products. For example, in the automobile industry, a sensor is mounted on each part of a manufactured automobile, and analysis data measured by each part sensor in a running state, a starting state, and a stopped state is analyzed, and a predetermined performance such as driving performance and safety is satisfied. Quality control is carried out by inspecting Therefore, in this type of quality control device, an analysis method or the like for easily associating and analyzing a causal relationship between measurement data (measurement data from a plurality of channels) from a large number of sensors (signal sources) is easily created. Capable functions are required.

In the above-described conventional quality control device, a causal relationship between measurement data (a plurality of channels) from a plurality of signal sources measured in the same time series could be generated. The causal relationship data extracted from the causal relationship between the channels cannot be used in extracting the causal relationship between the other sets of channels. That is, when a causal relationship between certain channels is generated, a causal relationship data table between other channels is referred to,
It does not have a function for referencing data or revising data, such as referencing causal relationship data between other channels via a network.

[0004]

Therefore, in quality control of an object such as an automobile having many signal sources (channels), a causal relationship can be generated only from a limited number of signal sources. However, there is a disadvantage that sufficient analysis cannot be performed.

The present invention has been made to solve such a conventional problem. In a quality control apparatus for managing the quality of a product, a plurality of sensors such as sensors mounted on each part of the product are provided. When generating a causal relationship between measurement data from signal sources (between a plurality of channels), causal relationship data extracted from a causal relationship between one set of channels is used to extract a causal relationship between other sets of channels. It is an object of the present invention to provide a data analysis device capable of performing the data analysis.

[0006]

In order to achieve the above-mentioned object, a data analysis apparatus according to the present invention uses the data of a plurality of signal source channels attached to each part of a product to be quality-managed, A data analyzer for creating new data for analyzing / determining the characteristics of each part of an article, wherein the data is measured in the same time series as a measurement data storage unit for storing measurement data output from the plurality of signal sources. An inter-channel causal relationship creating unit that generates a plurality of inter-channel causal relationships based on measurement data from a plurality of signal sources;
An analysis / judgment data creating unit for extracting analysis / judgment data for analyzing / judging the characteristics of the quality control target product from the inter-channel causal relationship, and an inter-channel causality created by the inter-channel causal relationship creating unit An inter-channel causal relationship storage unit for storing the relationship, and an analysis / judgment data storage unit for storing the analysis / judgment data generated by the analysis / judgment data creating unit, and extracting from the inter-channel causal relationship. The extracted analysis / judgment data can be referred to from the inter-channel causal relationship storage unit and / or the analysis / judgment data storage unit when extracting data from another inter-channel causal relationship.

[0007]

In the data analyzer according to the present invention, measurement data from a plurality of signal source channels attached to each part of the quality control target product is stored in the measurement data storage unit.

Next, an arbitrary plurality of channels are selected by an inter-channel causal relationship creating unit, and an inter-channel causal relationship is created based on the measured data. This inter-channel causal relationship is stored in the inter-channel causal relationship storage unit.

Next, the analysis / judgment data creation unit
Analysis / determination data for analyzing / determining the characteristics of the quality control target product is created from the plurality of selected causal relationships between channels. This analysis / judgment data is stored in the analysis / judgment data storage unit.

The data in the measurement data storage unit, the inter-channel causal relationship storage unit, and the analysis / judgment data storage unit can be arbitrarily referred to when extracting the analysis / judgment data from another inter-channel causal relationship. Then, any inter-channel causal data created so far can be read from the inter-channel causal relationship storage unit and referenced.

Therefore, when generating a causal relationship between measurement data (between a plurality of channels) from a plurality of signal sources such as sensors mounted on each part of a product, a causal relationship extracted from a causal relationship between a set of channels is generated. The relational data can be used for extracting a causal relation between other sets of channels.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A data analyzing apparatus according to the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 shows a data analyzer 1 according to the present embodiment.
0 is a block diagram showing the configuration of the data analysis device 10.
Constitutes a quality management device together with other functional parts not shown.

The data analyzer 10 includes a master CPU 12 for performing overall control of the system, a CRT 14 as a display unit (monitor), a touch panel 16 provided as an input unit on the CRT 14, A ROM 18 for storing a program for controlling the analysis device 10 and a RAM 2 for temporarily storing calculation results and the like
0.

The data analysis device 10 includes a measurement unit 22 for processing measurement data output from sensors provided in each unit of the quality control target product based on set measurement conditions and a measurement method, and a plurality of measurement sites. An inter-channel causal relationship creating unit 24 for creating a causal relationship between channels based on the measurement data (channels), and an analysis / judgment data creating unit for creating analysis / judgment data based on the inter-channel causal relationship. From a unit 26, an analysis / judgment data storage unit 28 for storing analysis / judgment data, a measurement data storage unit 30a for storing the measurement data, and an inter-channel causal relationship storage unit 30b for storing the inter-channel causal relationship. A bus line 3 for connecting the data storage / reference unit 30 and the above-mentioned units to exchange control information and data.
And 1.

The measuring section 22 has a ROM 32 for storing a program for setting measurement conditions and a measurement method and a program for executing the measurement, and outputs signals from sensors 1 to n attached to respective parts of the quality control target product. Interface for reading the measured data (hereinafter referred to as I /
F) circuit, and the inter-channel causal relationship creating unit 24 includes a ROM 36 that stores a program for creating a causal relationship between channels and a program for displaying on the CRT 14. The determination data generator 26 stores a program for generating analysis / determination data based on the causal relationship between channels and a program for displaying the data on the CRT 14.
OM38 is provided.

The analysis / judgment data storage unit 28 and the measurement data storage unit 30 constituting the data storage / reference unit 30
a, the inter-channel causal relationship storage unit 30b includes ROMs 40, 42a, and 42b that store programs for displaying stored data on the CRT 14, respectively.

FIG. 2 is a measurement block diagram in the case where a car is taken as an example of a product to be quality-controlled. Sensors are attached to each part of the manufactured car 50, for example, a running state, a start state, and a stop state. The quality control is performed by analyzing the measurement data obtained by the sensors at each section and examining whether or not predetermined performances such as running performance and safety are satisfied.

In FIG. 2, an automobile 50 is mounted so that wheels can be mounted on a rotating drum 52 and the vehicle can be brought into a running state. A sensor for measurement is mounted on each part of the automobile 50. For example, taking the analysis of steering wheel shake as an example, a suspension vibration detection acceleration (hereinafter, acceleration is referred to as G) sensor 54, a steering wheel detection G sensor 56, a tire rotation speed detection sensor 58, and a speed detection sensor 6
0 is attached to each part, and each sensor 54, 56, 58, 60
Are configured to transmit measurement data to the data analyzer 10 via the signal converters 62, 64, 66, and 68, respectively.

The data analyzer 1 configured as described above
With 0, a data analysis method for product quality control is created and data analysis is performed.

Next, the data analyzer 10
Regarding a method of performing data analysis for product quality control, the block diagram of FIG. 1, the measurement block diagram of FIG. 2, the flowcharts of FIGS. 3 and 4, and the CRT 1 shown in FIGS.
This will be described with reference to the display screen diagram displayed in FIG.

In the following description, selection and setting are performed by an operator operating (touching) the touch panel 16.

When the data analyzer 10 is started, the master CPU 12 reads the data of the input condition setting screen from the ROM 18 and displays it on the CRT 14 (FIG. 5). This C
The touch panel 16 provided on the RT 14 is touched by the operator, and for example, a channel (measurement site) to be measured is selected in accordance with the display of “Please select a channel to be used and a place to be set”, and measurement input conditions are set. By making the settings, the signals to be measured (measurement data) are associated, and then, on the measurement condition setting screen (FIG. 6) displayed on the CRT 14, the type of measurement to be performed is set. Here, the average speed of the front wheel right and left is set as a trigger condition in the calculation condition setting, and all items to which the sensors 54, 56, 58, and 60 are attached are measured. The trigger condition is set, for example, such that the measurement starts immediately after the average speed of the right and left front wheels is 150 km / H.

When the above setting is completed, the rotary drum 52 is rotated to bring the automobile 50 into a predetermined running state, and the G sensor 54 for detecting the suspension vibration, the G sensor 56 for detecting the handle shake, the tire rotational speed detecting sensor 58, The measurement of each unit is performed via the speed detection sensor 60. Each sensor 5
The measurement data from 4, 56, 58 and 60 are converted into electric signals by signal converters 62, 64, 66 and 68,
It is sent out by the measurement unit 22 via the / F circuit 34 and stored in the measurement data storage unit 30a.

The inter-channel causal relation creating unit 24 receives a setting (input) from the operator and selects channels included in the inter-channel causal relation, ie, which measurement channels and which measurement Process by linking channels (measurement data extraction)
It is set whether or not to perform (step S1). That is, in the following graph registration procedure, the combination condition of the graph display for extracting the measurement data to be analyzed (vertical axis, horizontal axis, zoom, etc.)
Decide and register.

Taking the analysis of handle shake as an example, as the channels to be associated, the handle shake amount (AD channel 1), suspension vibration amount (AD channel 2), tire rotation speed (AD channel 3), etc. are selected.
From the displayed waveform of the measured data, data that is likely to cause a handle shake is extracted. The combination of channels to be selected involves variously changing the combinations of channels likely to cause handle blur, creating a causal relationship between channels, and extracting and analyzing data.

The inter-channel causal relationship creating unit 24
In one causal relationship data between channels, measurement data of up to three channels can be associated, and up to 32 causal relationship data can be created.

The causal relationship between channels created here is as follows:
It is shown to the operator in a visual form called a graph. When creating a causal relationship between channels, use the CRT
The graph registration screen shown in FIG. 7 is displayed in FIG. 14, and the operator creates a causal relationship between channels by selecting data to be displayed on the graph from the measured data while viewing this screen. The graph registration screen has three functions, and the operator selects the function from the touch panel 16.

First, a channel to be associated is selected by the display data selection function shown in FIG. 7A, and the horizontal axis and the vertical axis of the graph are set. Then, the measurement corresponding to the channel selected from the measurement data storage unit 30a is performed. The data is read (step S2), and the measurement data is shown to the operator in a graph display form in the graph display area (b) of FIG. 7 (step S3), and the operator confirms the measurement data by a graph (step S4). .

When a part necessary for analysis is partially extracted from the displayed measurement data (step S5), a graph extraction range is set using the graph extraction function (c) in FIG. It is set (step S6). At this time, the operator operates the cursor keys at the bottom of the screen to set the cut-out range of the graph displayed in the graph display area B. The measurement data use range data is generated by setting the cut-out range, and the data use range of the graph is enlarged and displayed on the graph display area as needed for the operator (step S7).

Finally, the operator checks the graph cut-out range, that is, the data use range (step S8), and names the causal relationship between channels using the graph name setting function (d) in FIG. (Step S
9) Thereafter, the operator can use the set causal relationship between channels through the graph name.

The inter-channel causal relationship thus created is stored in the inter-channel causal relationship storage unit 30b as inter-channel causal relationship data having a maximum of three measurement channels and a data use range (step S1).
0). If another causal relationship between channels is to be set, the procedure returns to step S1, and the operation is performed by setting the required measurement channel in the above-described procedure (step S11).

Next, a procedure for generating new analysis / judgment data from the causal relationship between channels created as described above and existing analysis / judgment data will be described. The analysis / judgment data includes extraction reference data, extraction target data, extraction trigger conditions, and data names.

When creating analysis / judgment data, the CR
At T14, a graph selection screen in FIG. 8 and an extraction condition setting screen in FIG. 9 are displayed for the operator.
The extraction condition setting screen includes three functions of (a) an extraction reference data setting function, (b) an extraction target data setting function, and (c) an extraction trigger setting function (see FIG. 9), and an operator uses a touch panel. And these condition input settings,
Perform graph selection, etc.

First, the operator checks (a) in FIG.
The graph used to create the analysis / judgment data, that is, the inter-channel causal relationship used to create the analysis / judgment data, is selected using the use graph selection function (step S12). As a result, the analysis / judgment data creation unit 26 reads the inter-channel causal relationship used for creating the analysis / judgment data from the inter-channel causal relationship storage unit 30b. The read causal relationship between channels is displayed as a graph to the operator on the CRT 14 (the graph display area in FIG. 8B).

Next, the operator checks (a) in FIG.
The extraction reference data and the extraction target data are set by using the extraction reference data setting function of (b) and the extraction target data setting function of (b), and the inter-channel causal relation read from the inter-channel causal relation storage unit 30b in step S12. Select from the included use channel numbers (step S1
3. Step S14).

Thereafter, the operator checks (c) in FIG.
The extraction trigger condition is set using the extraction trigger setting function. There are two methods for setting the extraction trigger: a method by directly inputting a numerical value, and a method of referring to the value of other analysis / judgment data. One of them is selected in step S15.

In the setting by direct numerical input (step S
18) Although the extraction trigger condition parameter is a constant value, when referring to the value of other analysis / judgment data, the other analysis / judgment data separately extracted from the analysis / judgment data storage unit 28 is read. (Step S
16, step S17), the extraction trigger condition parameter is a variable that depends on the extraction method of other analysis / determination data and the extraction result. The other analysis / judgment data set here does not need to be generated from the currently set inter-channel causal relationship, and the analysis / judgment data generated from the other inter-channel causal relationship is arbitrarily used. can do.

Thus, the measured data of other channels and the causal relationship between channels which are not included in the causal relationship between channels used to create the data for analysis / judgment are also analyzed /
It can be used for creating the determination data. The analysis / judgment data thus generated is stored in the analysis / judgment data storage unit 28.

That is, a data extraction test is performed according to the data extraction conditions set as described above (step S19). If the data extraction conditions are appropriate (step S20), the extracted analysis / judgment data is analyzed. A name is input (step S21), and the analysis / judgment data storage unit 2 is entered.
8 (step S22). If the setting of the data extraction condition is inappropriate in step S20, the process returns to step S13, and the procedure from the setting of the data extraction condition is repeated. If another analysis / judgment data is to be created in step S23, the process returns to step S12, and the above procedure is repeated from the operation of selecting the inter-channel causal relationship used for creating the analysis / judgment data. become.

More specifically, for example, when the output of the speed detection sensor 60 after signal conversion is AD ₄ and the output of the G sensor 56 for handle shake detection after signal conversion is AD ₁ , AD ₄ is used as a channel to be used in step S 12. ₁ ,
Select AD _4, select the AD ₄ as the extraction reference data in step S13, if you select the AD ₁ as an extraction target data in step S14, that the causal relationship between AD ₁ and AD ₄ is selected becomes, are displayed in the graph display area in AD ₄ and AD ₁ Togazu 9, the amount of the handle blur will be apparent to visually on the same screen for the speed.

Further, the signal-converted suspension vibration
The output of the motion detection G sensor 54 is AD _TwoAnd when
AD for channel₁, AD_TwoThen, in FIG.
AD in graph display area₁And AD_TwoAnd the curve is displayed
The amount of handle shake relative to the amount of suspension vibration
It will be visually recognizable on the same screen.

On the other hand, in FIG. 9, (a) the output of the speed detection sensor 60 is selected as extraction reference data, (b) the output of the handle shake detection G sensor 56 is selected as extraction target data, and (c) extraction is performed. Speed αm / H as trigger
If you select ~βm / H, the output in the region of the velocity αm / H~βm / H based on the output of the speed detecting sensor 60 AD ₁ and AD ₄ is displayed in the graph display area in FIG. Therefore, it can be determined whether or not the steering wheel shake is within the target range in the speed range of αm / H to βm / H.

Further, using the graph selection function shown in FIG. 8A, "A", that is, AD ₁ (AD ₄ ), is used as the causal relationship between channels used to create analysis / judgment data.
Is selected, a display when AD ₁ is a function of AD ₄ is displayed in the graph display area of FIG.
The handle shake amount with respect to the speed can be visually recognized. If “E”, that is, AD ₁ (AD ₂ ) is selected as the causal relationship between channels, a display when AD ₁ is a function of AD ₂ is displayed in the graph display area of FIG. The handle shake amount with respect to the suspension vibration amount can be visually recognized.

By the operation of each unit of the apparatus described above, a plurality of causal relationships between channels are generated, and data for analyzing / determining the characteristics of each signal source (each measurement site) can be extracted from the causal relationships. . Further, when extracting data from the causal relationship between one channel and extracting data from the causal relationship between other channels, the extracted data can be referred to and diverted,
It is possible to comprehensively analyze / determine the characteristics of many signal sources.

[0046]

As described above, according to the data analysis device of the present invention, a plurality of causal relationships between channels are generated,
From the causal relationship, data for analyzing / determining the characteristics of the quality control target product can be extracted. When extracting data from one channel causal relationship and extracting data from another channel causal relationship, It can be referred to and diverted, and the characteristics of the quality control target products can be easily and comprehensively analyzed /
The effect that determination can be obtained is obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a data analysis device embodying the present invention.

FIG. 2 is a measurement block diagram of a quality control target product.

FIG. 3 is a flowchart showing an operation of the data analysis device shown in FIG.

FIG. 4 is a flowchart showing an operation of the data analysis device shown in FIG.

FIG. 5 is a diagram illustrating an input condition setting screen displayed when measurement data from a plurality of signal source channels is input.

FIG. 6 is a diagram illustrating a measurement condition setting screen displayed when conditions for measuring data from a plurality of signal sources are input.

FIG. 7 is a diagram illustrating a graph registration screen displayed when displaying and registering measurement data from a plurality of signal sources.

FIG. 8 is a diagram illustrating a graph selection screen displayed when an arbitrary measurement data is selected from registered measurement data.

FIG. 9 is a diagram illustrating an extraction condition setting screen displayed when an extraction condition for extracting data for analysis / judgment is selected from arbitrary selected measurement data.

[Explanation of symbols]

10 data analyzer 12 master CP
U 14 CRT 16 Touch panel 18 ROM 20 RAM 22 Measurement unit 24 Causal relationship creation unit between channels 26 Analysis / judgment data creation unit 28 Analysis / judgment data storage unit 30a Measurement data storage unit 30b ... Causal relationship creation unit between channels 30 ... Data storage / reference unit 31 ... Bus line 32 ... ROM 34 ... Interface circuit 36,38,40,42a, 42b ... ROM

────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Ichiro Kono 1-10-1 Shinsayama, Sayama-shi, Saitama Honda Engineering Co., Ltd. (58) Field surveyed (Int.Cl. ⁶ , DB name) G05B 23/02 G01D 21/00 G05B 15/02

Claims (2)

(57) [Claims] 1. A data analysis apparatus for creating new data for analyzing / determining characteristics of each part of a quality control target product from data of a plurality of signal source channels attached to each part of the quality control target product. A measurement data storage unit for storing measurement data output from the plurality of signal sources; and a channel for generating a plurality of inter-channel causal relationships based on the measurement data from the plurality of signal sources measured in the same temporal sequence. An inter-causal relationship creating unit; an analyzing / judging data creating unit for extracting analysis / judgment data for analyzing / judging the characteristics of the quality control target product from the inter-channel causal relationship; An inter-channel causal relationship storage unit for storing the inter-channel causal relationship generated by the unit; and an analysis / judgment generated by the analysis / judgment data creating unit. And a analysis / determination data storage unit for storing a over data, extracted from the causal relationship between the channel analysis /
At the time of extracting data for determination from another causal relationship between channels, the causal relationship between channel storage unit and / or
Alternatively, the data analysis device can be referred to from the analysis / judgment data storage unit. 2. The data analysis device according to claim 1, wherein
The data analysis device is configured to include a display unit and an input unit, and an input condition setting screen for inputting measurement data output from the plurality of signal sources, and the measurement data from the plurality of signal sources. A measurement condition setting screen for setting measurement conditions for measurement; a graph registration screen for displaying and registering measurement data from the plurality of signal sources; and selecting any measurement data from the registered measurement data. And an extraction condition setting screen for extracting data for analysis / judgment from the selected arbitrary measurement data on the display means, and a desired selection and setting input from the input means. A data analysis device characterized in that the data analysis device is configured to perform the following.