JP3343666B2 - Temperature compensation method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a measurement technique for detecting and measuring a difference in taste of food and drink using a sensor having a taste, which is one of the five senses of human beings. Food,
For example, it is a technology for detecting a difference in taste of drinking water, alcoholic beverages, and the like to be provided for eating and drinking, and a technology that can be applied to quality control in a factory for producing these foods. In particular, the present invention relates to a technique for temperature compensation to be applied to a measured value of a taste sensor using a lipid membrane, and a technique for implementing the method, in relation to the technology disclosed by the applicants in the patent application. Related to the equipment used.

[0002]

2. Description of the Related Art Saltiness, sweetness, bitterness, sourness, and umami are said to be basic components of taste, and each of them has a different degree. Applicants have disclosed in several patent applications that these taste differences can be measured physically, where human sensation can be assessed. For example, in Japanese Patent Application Laid-Open No. 3-54446, "Taste sensor and method for producing the same", a molecular film of a lipid substance composed of a molecule having a hydrophobic portion and a hydrophilic portion becomes a taste sensor that can replace human taste. That was shown. Since then, based on some heuristic facts, it has been disclosed by patent applications that taste differences can be measured as tastes or taste differences that are physically measurable as electrical quantities such as, for example, potential, potential difference, resistance, current. The list is shown in Table 1. For the purposes of this application, for the disclosure of these applications, only the abstract will be caught,
Try to avoid duplicate descriptions.

[0003]

[Table 1]

[0004]

None of the applicants' disclosures so far refer to temperature compensation for measurement. The first object of the present invention is to perform temperature compensation of a measured value of a taste sensor using a lipid membrane (hereinafter, also simply referred to as a taste sensor), and to enhance reliability and accuracy of a measured value of a taste or a difference in taste. I have.

[0005] Taste sensors are affected by temperature as well as human taste. So, in quality control at a food factory,
When discriminating a subtle taste difference, it is required to perform strict temperature control of the measurement system and measure the taste accurately to capture the essence of the taste. Especially in the production and inspection lines of factories and outdoors (many places where people enjoy the taste), the range in which the temperature can change is wide, and the use of taste sensors is too limited. Therefore, a reliable method for temperature compensation is required. Applying this method,
A technique is provided that enables use in a place with a large temperature difference and comparison of measurement data obtained at different temperatures.

[0006] Depending on the type of taste sensor, the temperature characteristics may vary depending on the sample to be measured. Therefore, a temperature compensation method that can be universally applied to samples having unknown temperature characteristics is required. In the case of a taste sensor that captures electric potential as a physical measurement quantity, for example, drift can be seen significantly when the elapsed measurement time is set to hours and minutes, so this type of problem can also be addressed. A method is desired.

[0007]

[Means for Solving the Problem] In general, considering a technique for compensating for temperature, a temperature change characteristic of an object to be measured near a measurement temperature is known, and in many cases, the characteristic is represented by a temperature coefficient, and a temperature difference is represented. A compensation method is used in which the amount of change in physical property temperature corresponding to the product of the temperature and the temperature coefficient is added to or subtracted from the measured value. Here, looking at the temperature coefficient that is the basis of compensation, it is the temperature differential value of the physical quantity, which is a temperature function, that is, the value obtained by dividing the difference between the physical quantities at two different temperatures by the temperature difference. That is, temperature compensation using a quantity having a dimension obtained by dividing [measured physical quantity] by [temperature]. On the other hand, the present invention is characterized in that the temperature compensation is performed by using the difference in temperature coefficient as a source in order to cope with the unique characteristics of the taste sensor using the lipid membrane.

Further, in the present invention, the temperature coefficient difference is not used as it is in the calculation for calculating the compensation amount as it is, but the value obtained by normalizing the temperature coefficient with the physical quantity to be measured (temperature coefficient difference normalized value) It is characterized in that temperature compensation is performed using an amount having a dimension of minus one power of [temperature].

[0009]

The method of temperature compensation according to the present invention will be described in detail using mathematical expressions. It has been found that the following three assumptions are established from various experiments on taste sensors using lipid membranes.

(1) The taste sensor uses a lipid membrane,
Since the lipid can respond to various kinds of tastes, the temperature compensation of the taste sensor should be strictly analyzed using matrix algebra, and the difference in the temperature characteristics of the taste sensor depending on the taste type should be considered. I need to examine it. For example, since saline (saltiness) and sucrose solution (sweetness) have completely different types of taste, the temperature characteristics of the taste sensor for both samples are different. This can be experienced by the human sense, and it should be said that there is a difference between salty and sweet tastes depending on the temperature.

However, since many uses of taste sensors are for quality control in a certain food factory, the measurement target is a single variety, such as sake, and even if the taste quality differs slightly depending on the brand, After all, it is the range of sake, and the difference in the temperature characteristics of the taste sensor due to the difference in taste quality will be put on hold for a while, and a simple model will be considered. (Actually, as will be described later, in the experimental example of sake, it has been found that this shelving is appropriate and that the difference in the temperature characteristics of the taste sensor due to the difference in taste quality can be ignored.) In the discussion, it is assumed that the change in the temperature characteristic of the taste sensor due to the taste type or taste quality is ignored. In other words, the output f of the taste sensor depends on two factors, the concentration C of the taste substance and the temperature T [° C]. To perform the temperature compensation means to perform the temperature compensation on the actually measured values f (Cs, Ts) and f (Cr, Tr) of the sample solution and the reference solution, respectively, and calculate the calculated value (estimated value) f ( C
s, To) and f (Cr, To).

(2) The output f (C, T) of the taste sensor is represented by the following expression using g (T) and h (T) whose variables are functions of the temperature T. It was discovered that it was good to see. f (C, T) = g (T) × logC + h (T) (1) The meaning of Equation 1 is that if the measurement target is narrowed down to one variety (sake), in such a narrow range, the temperature is not affected. It shows that the output of the sensor changes linearly with the value (logC).

(3) For a simple idea, it is assumed that the temperature characteristic of the taste sensor is linear with respect to the temperature T.
With the understanding of the technical idea of the present invention, it will be understood that this assumption is not always necessary. Since the temperature sensitivity (temperature coefficient) S [mV / ° C] of the taste sensor is not a function of the temperature T, in the equation 2 obtained by differentiating the equation 1 with the temperature, S = (∂f / ∂T) = g ′ (T) × log C + h ′ (T) (2) where g ′ (T) and h ′ (T) are constants.

FIG. 1 is a diagram for explaining the concept of the technical concept of the present invention. The horizontal axis represents the temperature (liquid temperature) [° C.] of a liquid (eg, sake) having a taste to be measured, and the vertical axis represents the temperature. It represents the output potential [mV] of the sensor. The straight line A in the figure indicates the temperature characteristic of the output of a certain taste sensor for the reference liquid. Each of the three points on the temperature scale is a reference temperature (the taste at this temperature is being measured with a taste sensor).
To, the temperature Tr exhibited by the reference liquid at the time of measurement, and the temperature Ts exhibited by the sample liquid to be measured. In the present invention,
First, the temperature coefficient S (Cr) of the output value of the taste sensor when measuring the reference liquid is determined. The gradient of the straight line A in FIG. 1 is S (Cr). Thereby, the sample liquid temperature Tr
The amount S of the white arrow indicating the size between the points PA and PO at
(Cr) (Ts-Tr) is required. Next, the sensor output versus liquid temperature characteristic of the same taste sensor is measured for another type of liquid (for example, another type of sake), and a straight line B is obtained. In the figure, the straight line B has a larger gradient than the straight line A. In other words, since the temperature coefficient is large, the value of PB (sensor output value) has a larger value than PA when viewed at the liquid temperature Ts.
How to see this difference is an important point of the present invention.

Assuming that the difference between the straight line B and the straight line A at the reference temperature To is a unit amount 1, the difference between the straight line B and the straight line A (the difference between PB and PA) at the liquid temperature Ts is expressed as follows.
(PB -PA) = (PB -P1) + (P1 -PA). Here, from equation 2, S (Cs) = g ′ (T) log Cs + h ′ (T) = g ′ (T) log (Cs / Cr) + S (Cr) = {g ′ (T) / g (To)} × {F (Cs, To) -f (Cr, To)} + S (Cr) (3) Since g '(T) is a constant, g' (T) / g (To)
Is also a constant. If this is set to a, (PB -P1) = a
(Ts−To), and (PB−PA) is 1 + a (Ts
−To). Here, a is a value obtained by measuring the difference between the temperature coefficients of the liquid from which the straight line A was obtained and the liquid from which the straight line B was obtained on a scale of 1 unit. Can be.

[0016] Regarding other samples of sake U and V,
The same type of temperature characteristic straight lines C and D can be obtained. In the temperature compensation of the present invention, the sensor output value U at the sample liquid temperature
s, Vs, the sensor output values Uo, V at the reference temperature
o will be estimated. Assumptions (1), (2),
Under (3), it can be seen that the relationship between the straight lines A and B holds both for the straight line C (for extrapolation) and for the straight line D (for interpolating). The following relationship is established between the temperature sensitivity S (Cr) of the sensor and the temperature sensitivity S (Cs) of the taste sensor in the sample. Δf (To) = f (Cs, To) −f (Cr, To) = [{f (Cr, Ts) −f (Cr, Tr)} − S (Cr) (Ts−Tr)] ÷ ｛1 + a (Ts-To)｝ (4)

As can be inferred from the above description, when obtaining the normalized value a of the temperature coefficient difference, it is not always necessary to grasp the relationship between the straight line A and the straight line B. , B, C, and D, it is sufficient to use at least two types of relationships.
The average value of the mutual relationship between the straight lines A, B, C, and D may be obtained. However, it is not necessary to do so much trouble, and any of B, C, and D is used as a substitute for the reference liquid characteristic line A. Is also good. In the above description, the example in which the temperature change is a straight line has been described for simplicity. However, it is apparent that the concept of the present invention can be applied even if the temperature characteristic of the sensor output is a curve if the differential concept is introduced. Will.

[0018]

【Example】

[First Embodiment] First, an experiment was conducted on the basic taste in order to confirm whether the assumptions (2) and (3) described in the section of the operation are correct. Here, an example of saltiness will be described. In the experiment, the temperature coefficient (temperature sensitivity) of the sensor output was changed by changing the concentration of the KCl aqueous solution. The taste sensor is a multi-channel type, and the concentration of the sample solution, which is an aqueous KCl solution, is 10, 100, 1000.
mM and the temperature of the sample is approximately 10 ° C, 20 ° C,
30 ° C. The potential of the taste sensor is plotted on the ordinate and the temperature difference from the reference temperature is plotted on the abscissa, and the results for the first to fourth channels are shown in FIGS. 2 (a), (b) and (c). The results for the channels are shown in FIGS.
(C). FIG. 4 is a plot of the temperature coefficient (temperature sensitivity) for each channel versus the sensor output of the test solution as a value at approximately 20 ° C.
(A), (b), (c) and FIGS. 5 (a), (b),
This is shown in (c). FIG. 4 (a) corresponds to the first channel, and FIG. 5 (c) corresponds to the sixth channel.

The results of FIG. 2 and FIG.
Can be expressed as 4 and 5, the value obtained by normalizing the rate of change [1 / ° C.] of the temperature sensitivity (temperature coefficient) of the taste sensor due to the change in the concentration of the test liquid with the output of the taste sensor due to the concentration difference. Table 3 shows the results.

[0020]

[Table 2]

[0021]

[Table 3]

There is a so-called Webber ratio as an index for evaluating a person's discrimination ability in sensuality. It is an index that measures the percentage of the initial stimulus by changing the amount of the stimulus and how different from the initial stimulus can be distinguished from the initial stimulus. In the world of taste, the intensity of stimulation is expressed by the concentration of a taste substance. That is, two solutions (concentration C and concentration C + c) having the same taste quality but different intensities (concentrations) are compared by drinking and discriminating the difference between them (however, it is not necessary to discriminate which is darker or lighter). The Weber ratio in this case is c / C × 100
%. The Weber ratio for taste is
It is said that about 20% is 5% for a taster (paneler) who performs quality control by identifying differences between factories and lots. Therefore, considering the industrial use of the taste sensor, it is necessary to identify the taste sensor at a Weber ratio of 5%.

The taste sensor output is proportional to the logarithm of the density (log C) as in the case of humans. Further, when the density changes by a factor of 10, the output changes by about 40 mV. The change is about 40 x log 1.05
= 0.85 mV. In order to identify the Weber ratio of 5%, the error due to temperature fluctuation must be sufficiently smaller than 0.85 mV. However, looking at Table 2, there is almost the same fluctuation per 1 ° C. of temperature change of the solution. From this fact, it is necessary to control the temperature with an accuracy of about 0.1 ° C., but it is actually very difficult, and temperature compensation is definitely required. Further, from Table 2, it can be seen that the temperature sensitivity is greatly different when the sample liquid is different. Therefore, it is understood that the error is not sufficient if the temperature compensation is performed only from the conventional temperature sensitivity without using the general sample liquid.

[Second Embodiment] The development of the embodiment described for the basic taste was also carried out for sake. A certain brand of Japanese sake (a commercial product of a sake brewing company with a factory in Fushimi-ku, Kyoto, where production is well controlled) is used as a reference solution, diluted with pure water, and diluted with sake such as mirin. By adding a small amount of a sweet substance similar to the above, the characteristic curve of the taste sensor output versus temperature can be represented by the straight line groups A, B,
A group of characteristic curves as shown by C and D is obtained.

FIG. 6 shows characteristic curves of sensor output potential versus temperature for four taste sensors for the above-mentioned specific brand. FIG. 7 shows the output potential versus temperature characteristics of the same sensor for four types of sake. In the figure, (a) is what is called Junmai sake, (b) is what is called Honjoshu,
(C) shows data on what is called Alps Ginjo, and (d) shows data on what is called synthetic sake. FIGS. 6 and 7 clearly show that the characteristic curve is represented by a straight line for any type of sake. From these measurement results, the temperature coefficient (temperature sensitivity) of the taste sensor output is plotted against the taste sensor output, and FIG. This proves that the assumption of the linear approximation described in the operation is reasonable.

[Third Embodiment] Based on the experimental facts shown in the first and second embodiments, the normalized value a (1 / ° C.) of the difference in temperature sensitivity (temperature coefficient) is used. The flow of the temperature compensation method is as follows. See the flow diagram (FIG. 9) and claim 1. The temperature compensation method of the measured value of the taste sensor using the lipid membrane according to the present invention goes through the following steps. The temperature coefficient of the output value of the taste sensor when measuring the 1 ° reference liquid is determined. Calculate or measure the output of the taste sensor at the desired temperature To of the 2 ° reference liquid. Temperature coefficient S (Cs) for samples other than 3 ° reference solution
Ask for. The output of the taste sensor at the desired temperature To of the 4 ° sample is calculated or measured in advance. The temperature coefficient obtained at 5 ° 1 ° may be used as one of them, but after all, at least 2
A normalized value of the difference between the temperature coefficients is obtained from the temperature coefficient of the output value of each taste sensor when measuring the type of liquid.
At this time, if there are variations in the measured values, an average value method such as the least squares method may be used in the calculation. 6 ° The output value of the taste sensor when measuring the reference liquid and the temperature of the reference liquid are obtained. 7 ° The output value of the taste sensor when measuring the liquid to be measured and the temperature of the liquid to be measured are obtained. The difference between the output value of the taste sensor when measuring the reference liquid at a desired temperature and the output value of the taste sensor when measuring the liquid to be measured is calculated using the value obtained in the above step at 8 °. Thus, the output value of the taste sensor at the temperature To, that is, the taste is measured.

Fourth Embodiment FIG. 10 shows a functional block diagram of a main part of a temperature compensator according to a second embodiment of the present invention. The subtraction unit 11 in FIG. 10 inputs the temperature coefficient of the taste sensor in each sample liquid for two sample liquids A and B (one of which may be a reference liquid) having different taste sensor outputs when the temperature is the same. And calculate the difference between the temperature coefficients. The normalization unit 12 receives the output of the subtraction unit 11 and the difference between the sensor outputs for the solution A and the solution B at the temperature To, normalizes the former by dividing the former by the latter, and calculates a constant a. The calculation unit 20 outputs the output a of the normalization unit 12,
The temperature coefficient S (Cr) of the taste sensor in the reference liquid, the temperatures Ts and Tr of the test liquid and the reference liquid, and the difference Δf (Cs, Ts) between the taste sensor outputs when the test liquid and the reference liquid are measured. f
(Cr, Tr)}, the equation 4 is calculated, and the difference {f (Cs, To) −f (Cr, To) of the taste sensor output when the test solution and the reference solution are measured at the temperature To. ) Calculate｝.

[Fifth Embodiment] In the above description, it has been assumed that the characteristic curve is linear for easy understanding of the invention, as described in the operation section. However, if the temperature coefficient is regarded as a temperature sensitivity represented by a differential value instead of a constant, it is obvious that the present invention can be applied to a case where the taste sensor output versus temperature characteristic is non-linear.

In the case where the temperature characteristic of the taste sensor output is non-linear with respect to the temperature T, for the sake of simplicity, g in Equation 1 is used.
Description will be made on the assumption that (T) and h (T) can be approximated by a quadratic expression of the temperature T. However, the following discussion can be similarly performed by approximation using a general n-order expression. First, assuming that the temperature characteristic of the taste sensor output in the reference liquid can be approximated by a quadratic expression of the temperature T,
A temperature characteristic curve can be obtained by measuring the output of at least three points having different temperatures (however, one of the three liquid temperatures may be To). At least (n + 1) different outputs may be measured). Specifically, the temperature characteristic curve of the taste sensor output in the reference liquid is represented by e (T). Sensor outputs at different temperatures T0, T1, and T2 in the reference liquid are measured. For each temperature Ti (i = 0 to 2), e (T) can be obtained from e (Ti) = {f (C0, Ti) -f (C0, T0)} (5).

Since the compensation value of the taste sensor output at an arbitrary temperature T can be obtained from the temperature characteristic curve, the compensation value f (Cr, Ts) of the taste sensor output obtained by adjusting the temperature T to the liquid temperature Ts of the sample is obtained. Experimental value f (Cr, Tr) and temperature T
Can be calculated as follows from the quadratic expression e (T). f (Cr, Ts) = f (Cr, Tr) − {e (Tr) −e (Ts)} (6) When the temperature of the sample liquid is equal to the temperature of the reference liquid, using the equation 1, f (Cs, Ts) −f (Cr, Ts) = g (Ts) × log (Cs / Cr) = {g (Ts) / g (T0)} f (Cs, T0) −f (Cr, T0)} (7 ), The compensation value Δf (Cs, T0) −f (Cr, T
0)｝ only needs to know g (T). From equation 1, δf (C, T) / δlogC = g (T) (8)

That is, g (T) means the slope of the density of the taste sensor with respect to the logarithm (log C). Since there is no variable of the concentration C, the slope is smaller than the sensor output difference (the two samples have the same liquid temperature) in at least two types of samples having different concentrations (one of the two may be a reference solution). You can ask. Assuming that g (T) can be approximated by a quadratic expression of the temperature T, g is obtained by calculating the slope at at least three points having different liquid temperatures (one of the three liquid temperatures may be T0). (T) can be specified (in the case of approximation of n-th order, at least (n +
1) The inclination at the point may be obtained).

For example, a sample solution having the same taste quality as the reference solution and having a concentration k times (k ≠ 1) is prepared, and the reference solution and the sample solution have different T0 and T0, respectively.
1. Measure the sensor output at the temperature of T2. From equation (1), for each temperature Ti, (i = 0 to 2) g (T) = {f (kC0, Ti) −f (C0, Ti)} / log (k) (9) Ask. However, in practice, the taste sensor reacts to various tastes, and the sensitivity for each taste is also different. Therefore, g (T) depends on which taste is used to obtain g (T).
Results are different. If the concentration Ci for each taste and the sensitivity of the taste sensor for that taste are αi, and the concentration C is represented by the sum of the concentrations of each taste, that is, if C = Σαi * Ci, Equation 8 is δf (C, T) / δlogCi = αi × g (T) (10), which is different for each taste for which the concentration is changed. However, there is no problem because αi disappears from the ratio of g (Ts) and g (T0).

As a result, the compensation value Δf (Cs, T0) −f (C
From Equations 6 and 7, f (Cs, T0) -f (Cr, T0) = {f (Cs, Ts) -f (Cr, Tr) + e (Tr) -e (Ts) ｝ × {g (T0) / g (Ts)} (11) (However, e (T) and g (T) can be obtained by Expressions 5 and 9.) e (T) is the taste of the reference liquid. This is the output characteristic of the sensor, and e (T) = S (Cr) (T−T0) (12) in the case of the first order of the temperature T. The part of the molecule of formula 4 corresponds to it. g (T)
Is the concentration gradient of the taste sensor, and in the case of the first order of the temperature T, a in Equation 4 corresponds to g ′ (T) / g (T0), and 1 + a (Ts−T0) of the denominator of Equation 4 is g (Ts) / g (T0)
Is equivalent to

[0034]

According to the present invention, there has been described a method and an apparatus for estimating a taste value at a reference temperature different from a measured temperature by performing temperature compensation on an output value of a taste sensor using a lipid membrane. The method of temperature compensation is not to simply add (decrease) the product of the temperature coefficient and the temperature difference to the measured value as in the conventional technology, but to calculate the value obtained by normalizing the difference between the two temperature coefficients with the output value of the sensor. It was decided to use it. That is, for the value obtained by dividing the difference between the output values at two different temperatures, called the temperature coefficient, by the temperature,
By taking the difference between the two types and using the normalized value, the following features were obtained.

1. The value of taste at a reference temperature can be obtained from the value measured by a taste sensor using a lipid membrane. 2. It is now possible to evaluate the industrial measurement of taste by adding temperature as a parameter. 3. The correspondence between the measured values and the human taste was made clearer, and the Weber ratio was clearly shown. 4. By knowing the temperature characteristics of two types of foods of the same system (for example, sake), it became possible to perform temperature compensation by interpolation or extrapolation for other foods. In other words, since the difference between the differences is an important factor in the compensation, the temperature compensation can be made universal and a temperature compensation method with a small error can be provided.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the concept of the technical concept of the present invention.

FIG. 2 shows first to fourth measurements of a KCl aqueous solution.
It is a graph which shows the temperature coefficient of the taste sensor output of a channel, (a) is KCl10mM, (b) is KCl1.
(C) is a graph when KCl is 1000 mM when the concentration is 00 mM.

FIGS. 3A and 3B are graphs showing temperature coefficients of the taste sensor outputs of the fifth and sixth channels when a KCl aqueous solution is measured. FIG. 3A is a graph showing KCl 10 mM, and FIG.
(C) is a graph when KCl is 1000 mM when the concentration is 00 mM.

FIG. 4 is a graph showing the temperature coefficient of the first to third channels at about 20 ° C. versus the sensor output of the test liquid;
(A) is the first channel, (b) is the second channel,
(C) is a graph of the third channel.

FIG. 5 is a graph showing the temperature coefficient of the fourth to sixth channels at about 20 ° C. versus the sensor output of the test liquid;
(A) is the fourth channel, (b) is the fifth channel,
(C) is a graph of the sixth channel.

FIG. 6 is a graph showing characteristic curves of output potential versus temperature of four taste sensors for a specific brand of sake.

FIG. 7 is a graph showing characteristic curves of output potential versus temperature of four taste sensors for four types of sake.

FIG. 8 shows characteristic curves of taste sensor output versus temperature coefficient from the measurement results of five types of sake. It is rough.

FIG. 9 is a diagram showing the flow of the temperature compensation method of the present invention.

FIG. 10 is a functional block diagram of a main part of the temperature compensation device of the present invention.

[Explanation of symbols]

 Reference Signs List 10 temperature coefficient difference normalized value calculation means 11 subtraction unit 12 normalization unit 20 calculation unit

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Kiyoshi Toko 2-8-3-2 Miwadai, Higashi-ku, Fukuoka City, Fukuoka Prefecture (56) References JP-A-3-54446 (JP, A) JP-A-3 163351 (JP, A) Japanese Patent Laid-Open No. 5-232083 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01N 27/416 G01N 33/02

Claims (2)

(57) [Claims] 1. A temperature compensation method for a measured value of a taste sensor using a lipid membrane, wherein a step (1) of obtaining a temperature coefficient of an output value of the taste sensor when a reference liquid is measured; Obtaining a normalized value of the difference in temperature coefficient from the temperature coefficient of the output value of each taste sensor when at least two types of liquids having different output values are measured (2,
(3, 4, 5), a step (6) of obtaining an output value of the taste sensor when the reference liquid is measured and a temperature of the reference liquid, and a step (6) of the taste sensor when the liquid to be measured is measured. (7) obtaining an output value and the temperature of the liquid to be measured; and calculating the output value of the taste sensor and the liquid to be measured when the reference liquid at a desired temperature is measured using the values obtained in the various steps. Calculating the difference between the measured value and the output value of the taste sensor (8). 2. A temperature coefficient difference normalizer for calculating a normalized value of a temperature coefficient difference from a temperature coefficient of an output value of each taste sensor for at least two types of liquids having different output values of a taste sensor using a lipid membrane. A temperature compensating device for a taste sensor, comprising a chemical value calculating means (10).