KR100197715B1 - Method for cleaning status distinction using fuzzy principle - Google Patents

Abstract

The present invention relates to a method for determining cleanliness using a fuzzy principle, wherein the sensory evaluation value measured by a professional panel through an input unit is predicted as a surface reflectivity through a fuzzy logic system, and the surface reflectance predicted through a fuzzy logic system is controlled. Display on a monitor based on a control signal for display from the device. In addition, by inputting the surface reflectance through the input unit, the sensory evaluation value is predicted by the fuzzy logic system by the database of sensory evaluation values measured by the previous specialized panel and provided to the controller, for display from the controller The sensory evaluation value predicted through the fuzzy logic system based on the control signal is monitored and displayed, so that not only the correlation between the surface reflectivity and the sensory evaluation value can be identified, but also the sensory evaluation value according to the human emotion is used for the surface reflectance value. By obtaining a functional relationship that can be predicted by the user, it is possible to maximize emotion control or emotion, and also to cluster data, to prevent duplication of information, to make the data consistent, and to be measured by a specialized panel. When information about sensory evaluation values is stored in a database, human This is to predict the emotional evaluation of.

Description

Determination of Cleanliness Using Fuzzy Principle

The present invention relates to a washing machine, and more particularly, to a washing degree determination method using a purge principle suitable for determining the degree of washing that is an indicator of the improvement of washing performance using the purge principle.

In general, FIG. 1 is a diagram illustrating an apparatus for evaluating washing performance through image analysis of a contaminated cloth according to the prior art. Hereinafter, the washing performance is evaluated through image analysis of an artificial fouling cloth with reference to FIG. The operation process will be described.

First, the surface reflectivity for the front and rear of the artificial contaminated cloth 100 is measured through a spectrophotometer (Spectrophotometer, 102), the washing performance through the computer 104 by the measured surface reflectivity of the artificial contaminated cloth 100, that is, The degree of washing D is calculated.

More specifically, before the contaminated artificial bladder 100 is contaminated, the surface reflectance (Ro) of the raw cloth is measured through the spectrophotometer 102, and the contaminated artificial contaminated cloth 100 is evaluated to evaluate the washing performance before washing. The surface reflectance (Rs) for is measured via spectrophotometer 102. Then, the washing experiment is carried out by the test method specified under a predetermined condition, and the surface reflectance (Rw) of the washed artificial fouling cloth 100 is measured.

The cleaning degree D by each surface reflectance (Ro, Rs, Rw) measured through each measurement process as described above is calculated by Equation 1 below.

The washing performance is evaluated according to the degree of washing (D) as described above.

However, when evaluating the degree of cleaning using the surface reflectivity, it is difficult to set the wavelength of the spectrophotometer 102, and when the artificial fouling cloth 100 is a colored cloth, the method of determining the degree of cleaning using the surface reflectance cannot be applied. There is a problem.

In addition, as the necessity of product development that meets the needs of consumers emerges, the necessity of introducing a sensory evaluation using a human emotion as a sensor for measuring the degree of cleaning is highlighted.

In other words, there is a difference between the cleanliness evaluation that consumers feel with emotion (ie visual) and the mechanical value for artificial stains, that is, the cleanliness evaluation by measuring surface reflectivity, for example due to the visual limitations of the consumer. If the cleanliness evaluation is classified only when the value is 2 or more, if the cleanliness evaluation by the machine value (cleaning degree by surface reflectance) is divided into 1 unit, the consumer does not feel this emotionally, There is no meaning.

Therefore, the present invention has been devised in view of the above-described points, and by introducing a mathematical function using the fuzzy principle, by deriving the correlation between the measured value of the surface reflectance of the artificial bladder and the measured value by the sensory test, It is an object of the present invention to provide a cleaning degree determination method using a purge principle that can measure the degree of cleaning by inspection.

The cleaning degree determination method using the fuzzy principle according to the present invention for achieving the above object, in the cleaning degree evaluation method, selecting a specialized panel through a predetermined identification process and training and managing the panel for a predetermined time and Comparing and analyzing the existing machine value and the sensitivity value measured by the panel, and comparing the standard machine value and the degree of cleaning, and producing a concentration change table according to the comparison result analyzed through the comparison analysis step, It is characterized in that it comprises a step of determining the washing degree recognition ability, a step of predicting the machine value using the sensitivity value measured by the specialized panel and a step of predicting the emotion value using the machine value.

1 is a schematic configuration diagram of a system for evaluating washing performance through image analysis of artificial stain cloth according to the prior art.

2 is a schematic block diagram of hardware for performing a cleaning degree determination method using the purge principle according to the present invention.

Figure 3 is a flowchart for explaining the operation of determining the degree of cleaning using the purge principle in accordance with a preferred embodiment of the present invention.

4 is a flowchart illustrating an operation process of quantifying emotion according to the present invention.

Figure 5 is a flowchart for explaining the operation of the emotional value numerical value in accordance with the present invention.

6 is a view for explaining the coherence coefficient for evaluating the identification ability of artificial blemishes using the number of single triangles.

7 is a view for explaining the improvement of the cleaning degree in the mechanical sense.

8 is a view for explaining the improvement of the washing degree in the emotional sense.

* Explanation of symbols for main parts of the drawings

210: input unit 220: control unit

230: fuzzy logic system 240: monitor

The above object and other advantages of the present invention will become more apparent through the preferred embodiments of the present invention described below with reference to the accompanying drawings by those skilled in the art.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

2 is a schematic block diagram of hardware for performing a cleaning degree determination method using the purge principle according to the present invention.

In FIG. 2, the input unit 210 measures the sensory evaluation value and the surface reflectance measuring device (e.g., evaluated by the sensitivity of a predetermined number of specialized panels (inspectors selected for evaluation of emotional values) selected through a predetermined process (e.g., For example, a surface reflectance measured through a spectrophotometer) is input.

In addition, the control unit 220 builds a fuzzy function by using the sensory evaluation value according to the sensitivity input through the input unit 210 and the surface reflectance measurement value measured by the surface reflectance measuring device to build a fuzzy logic system 230. And display the analysis results analyzed through the fuzzy logic system 230, that is, the surface reflectance value prediction value established by the correlation between the sensory evaluation value and the surface reflectivity or the sensory evaluation value prediction value predicted by the surface reflectance. To generate a control signal.

In addition, the fuzzy logic system 230 predicts the surface reflectance value and the sensory evaluation value using the non-fuzzy function provided from the controller 220 and provides the prediction result to the controller 220, and the monitor 240 controls the controller. Display the predicted result through fuzzy logic system 230 by the display control signal from 220.

With reference to FIGS. 2 to 8, an operation process of the cleaning degree determination method using the purge principle according to the present invention through the hardware composed of the above-described members will be described in more detail.

First, the basic steps for establishing the sensory evaluation of the cleaning degree, firstly, select an inspector or professional panel for the evaluation of the cleaning degree based on the consistency coefficient using Scheffee's Paired Test. (Step 310).

Here, the coherence coefficient is a method of assigning an indicator of discrimination ability by using the number of single triangles, which will be described with reference to FIG.

As shown in FIG. 6A, when there are three objects A, B, and C, when a decision is made that B is superior to A and C is superior to B, a determination is given that C is superior to A. It is natural. However, as shown in FIG. 6B, even though it is determined that B is superior to A and C is superior to B, a case in which A is superior to C is called a pure triangle.

Accordingly, when the result of the paired test of the panel on the panel, the panel can determine that the discriminating ability is consistent when the number of the above-mentioned single triangles is equal to or less than the predetermined number.

Meanwhile, the equation for determining the coherence coefficient ζ is as shown in Equation 2 below.

Here, k denotes the number of individuals to be compared, and d denotes an opening of a single triangle, where the coefficient of consistency ζ is 1 when there is no single triangle.

Next, while training and managing the selected expert panel (ie, the inspector with discrimination ability) through the step 310 for a predetermined time, the experiment is evaluated by evaluating the reproducibility of the judgment, the conformity with the criteria, and the agreement between the panels. It is determined whether to participate (step 320).

Then, by comparing and analyzing the existing surface reflectivity and the sensory evaluation value evaluated by the expert panel (step 330), the measurement value by the surface reflectivity determined to be the best match to the sensory evaluation is set as a reference scale, Are compared and analyzed (step 340).

In detail, a gray scale was prepared to compare and analyze the measured values according to the sensory evaluation (step 350). The state of the gun is set to 100%, and the intervals are divided by 10% to produce 11 levels of change tables, and then the level of cognition of washing for various artificial fouling guns is determined (step 360).

Then, the surface reflectance value is predicted using the sensory evaluation value (step 370), and then the sensory evaluation value is predicted using the surface reflectance value (step 380).

Hereinafter, the operation of predicting the surface reflectance value using the sensory evaluation value of step 370 will be described in more detail with reference to FIG. 4.

First, each sensory evaluation value for each prosthetic bladder was measured by a predetermined number of professionally trained experts (e.g., 10) by a panel of experts. (Step 371), the surface reflectance with respect to the same artificial pollution cloth is measured through a surface reflectance measuring apparatus (step 372).

Then, the sensory evaluation value and the surface reflectance measured through the steps 371 and 372 are input through the input unit 210. At this time, the sensory evaluation value measured by the panel is input data and the surface measured by the surface reflectance measuring device. The reflectances are set as output data and input (step 373).

As an example, Table 1 shows 10 people (X1, X _2, ..., X _10), each of the sensory evaluation value measured by a professional panel and the surface reflectance for the artificial contamination fabric of a predetermined number (618 pieces) The surface reflectance (Y) measured by the measuring apparatus is shown, respectively.

Here, the input data is a sensory evaluation value for each of a predetermined number of artificial staining cloths (for example, 618 artificial staining cloths), and the output data is a surface reflectance measurement value.

Next, the control unit 220 constructs a non-fuzzy function using the input data and the output data input through the input unit 210 (step 374), and the control unit 220 constructs the non-fuzzy function through the step 374. The fuzzy function is provided to the fuzzy logic system 230.

Hereinafter, an operation process in which the non-fuzzy function is built through the control unit 220 will be described in more detail.

First, randomness is divided between the minimum value and the maximum value of each variable, and then a fitness function is assigned. Then, a rule is formed by using a fuzzy label representing the largest fitness value for the input. There are contradictory rules (for example, the front part is the same but the back part is different) among the rules determined through this method, and the goodness-of-fit determined by the former part to select only one of them. ) And the product of the goodness-of-fit values in the rear part.

Using the above rule, a mapping relation from the input data X to the output data Y can be obtained, and the unfuzzy function used at this time is represented by Equation 3 below.

From here, to be.

From here Is the fitness function of the kth variable of the i th rule.

If the goodness-of-fit function is defined as a Gaussian function, the non-fuzzy function of Equation 3 is expressed as Equation 4 below.

From here, Wow Denotes the mean and standard deviation of the Gaussian function used in the test section for the i-th variable of the first rule. Is defined as the median of the fuzzy sets defined in the latter part.

On the other hand, due to inconsistent rules, there are cases where only a small number of rules are derived from the given data pairs, and thus the information contained in the data cannot be sufficiently reflected. In order to prevent such cases, one rule-1 data pair method (One rule for one data pair method).

The 1 rule-1 data pair method described above is a median of the Gaussian function, which is a system variable that determines the input and output nonlinear function relations, and the median value of the back dry function. I / O data given instead , i = 1,2, ..., n, l = 1,2, ..., N can be substituted as shown in Equation 5 below.

Since the above-described method forms a rule as one input-output data pair becomes the median, all the information included in the function derivation data pair can be used, but it can include unnecessary information due to duplication of information. There is a disadvantage, which is solved using adaptive optimal fuzzy logic which will be described below.

In other words, adaptive optimal fuzzy logic is to express one data group by one rule by clustering given input / output data using a circle having a value r as a radius.

First, the median value of one cluster using the first I / O data of In addition, the radius of the cluster is r, and the newly introduced weight variable is defined as follows. In other words,

Next, assuming that k (k = 2,3 ...) data pairs are considered, There will be M clusters with median of To the median of these clusters Nearest cluster computed by, I = 1,2 ‥‥‥ M Determine.

At this time, If is the median of the new cluster Set. In other words, The weight is set as in Equation 6 below.

if, In the case of, it is as shown in Equation 7 below.

Finally, the output of the kth adaptive optimal fuzzy logic can be calculated using Equation 8 below.

Next, the fuzzy logic system 230 predicts the surface reflectance value using the non-fuzzy function (Equation 5) provided from the controller 220 (step 375), and provides the prediction result to the controller 220. do.

In addition, the control unit 220 analyzes the surface reflectance predicted through the fuzzy logic system 230 and the comparative analysis measured by the surface reflectance measuring device (step 376), generates a control signal for displaying the control signal, and the control unit 220. Based on the control signal for display from the control unit 220, the result of the comparative analysis through the control unit 220 is displayed on the monitor 240.

Accordingly, the analysis results of the fuzzy logic system 230 showed that the predicted surface reflectivity measurements showed a constant pattern as the number of panels increased, and the pattern was very close to the actual surface reflectance measurement. This suggests that the sensory evaluation can replace the existing mechanical cleaning method (step 377).

In addition, when the sensory evaluation value is converted into a measured value based on the surface reflectivity using the non-fuzzy function as in the above-described method, it is possible to obtain the minimum increase in the surface reflectivity that the consumer can feel (visually). This means a substantial performance improvement from the consumer's point of view.

In terms of the calculation of the ripple effect of uncertainty, the equation is expressed by the equation (9) below when ni (xi) is the range of cleanliness that the i-th panel can recognize during sensory evaluation. It is reflected in the surface reflectance values f (x1, x2, ..., xi) in the form as shown in equation (10).

As an example, it can be seen that only FIG. 7C has improved the performance that consumers feel under the three expected conditions as shown in FIGs. 7A, B, and C. FIG.

On the other hand, unlike converting the sensory evaluation value into the surface reflectance value using the above equation (5), when converting the surface reflectivity to the sensory evaluation value using the equation (8), by substituting only the surface reflectivity value into the equation (8) In addition, performance improvement can be evaluated in terms of sensuality.

The surface of the operation process as described above is expressed by Equation 10 below.

Here, Rf is a surface reflectance value, δ is a standard deviation of surface reflectance, and is a constant having a relationship independent of surface reflectance. Meaning in the above equation (10) means the degree of cleaning after development ( ) Implies development only when it is greater than the consumer's perception range.

Next, the operation of predicting the sensory evaluation value using the surface reflectance value of step 380 will be described in more detail with reference to FIG. 5.

Meanwhile, in Fig. 5, steps 381 and 382 are performed substantially the same as steps 371 and 372 shown in Fig. 4, so that the description herein is omitted to avoid overlapping substrates. do.

Next, the sensory evaluation value and the surface reflectance measured through the steps 381 and 382 are input through the input unit 210. At this time, the surface reflectance measured by the surface reflectance measuring device is input data and the sensory evaluation measured by the panel. The average of the values is set as output data and input (step 383).

Next, the control unit 220 builds a non-fuzzy function using the input data and the output data input through the input unit 210 (step 384), and the control unit 220 generates the non-fuzzy function through the step 374. The fuzzy function is provided to the fuzzy logic system 230.

Here, the operation process in which the non-fuzzy function is built through the control unit 220 is substantially the same as the operation process as described above, and thus will be omitted here.

Next, the fuzzy logic system 230 predicts the sensory evaluation value using the non-fuzzy function provided from the control unit 220 (step 385), and then provides it to the control unit 220, and provides the Based on the control signal for display, the prediction result provided from the fuzzy logic system 230 through the monitor 240 is displayed.

As an example, of the three cases shown in FIGS. 8A, 8B, and 8C, only FIG. 8C may allow the consumer to feel the improvement in washing performance with the skin. The sensory evaluation value can be converted (step 386).

Therefore, since the sensory evaluation value by an initial specialized panel can be used as a database, the measurement by a specialized panel is unnecessary every time a sensory evaluation value is measured.

As described above, the sensory evaluation value measured by the expert panel through the input unit 210 is predicted as the surface reflectivity through the fuzzy logic system 230, the surface reflectivity predicted by the fuzzy logic system 230 is the control unit ( Displayed to the monitor 240 based on the control signal for display from 220. In addition, when the surface reflectivity is input through the input unit 210, it is predicted as a sensory evaluation value through the fuzzy logic system 230 by the database for sensory evaluation values measured by the previous specialized panel to the controller 220. And sensory evaluation values predicted through the fuzzy logic system 230 based on a control signal for display from the controller 220 are displayed on the monitor 240.

Therefore, by using the present invention, not only the correlation between the surface reflectance and the sensory evaluation value can be identified, but also the emotional control can be obtained by obtaining a functional relationship that predicts the sensory evaluation value according to the human emotion as the surface reflectance value. I have an effect that can maximize the sensitivity. In addition, by clustering data, it is possible to prevent duplication of information and to make the data consistent. When information on sensory evaluation values measured by expert panels is stored in a database, the human emotional evaluation is performed by simple mechanical measurements. Value is predictable.

Claims (3)

CLAIMS 1. A cleaning degree evaluation method comprising: selecting a specialized panel through a predetermined identification process and training and managing the panel for a predetermined time; Comparing and analyzing the existing machine value and the sensitivity value measured by the panel, and comparing and analyzing the reference machine value and the washing degree; Producing a concentration change table according to the comparison result analyzed through the comparative analysis step, and identifying a washing degree recognition ability; Estimating a mechanical value by using the emotional value measured by the specialized panel; Cleanliness evaluation method using the fuzzy principle, characterized in that the step consisting of estimating the emotional value using the mechanical value. The method of claim 1, wherein the machine value predicting step comprises: measuring an emotional value of the artificial pollution degree through the selected specialized panel, and measuring a mechanical value of the artificial contamination cloth through a mechanical value measuring means; Setting the emotional value measured through the step as input data and the mechanical value measured through the machine value measuring means as output data, and then constructing an unfuzzy function using fuzzy principle; Cleaning step using fuzzy principle, characterized in that it comprises a step of predicting the machine value using a non-fuzzy function, and the step of quantifying the emotional value by comparing and analyzing the predicted machine value and the measured machine value Assessment Methods. The method of claim 1 or 2, wherein the emotion value prediction step comprises: measuring the sensitivity value of the artificial pollution degree through the selected specialized panel, and measuring the mechanical value of the artificial contamination cloth through a mechanical value measuring means; And setting the machine values measured by the machine value measuring means as output data and the emotion values measured through the measuring step as input data, respectively, and then constructing a fuzzy function using fuzzy principle. And a step of predicting the mechanical value as an emotional value using the constructed non-fuzzy function.