JP2004093447A - Smell measuring device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve a calculating precision of an index value representing the degree of a strength of smell such as an odor index, etc., of unknown smell. <P>SOLUTION: Reference vectors formed by measuring many standard smells prepared corresponding to various types of the smells are stored in a storage unit 53. At an unknown sample measuring time, a similarity judging unit 57 judges a similarity of the unknown smell to the standard smell from positional relations between a measured point of the unknown sample and all reference vectors in 10-dimensional space formed by detecting outputs of 10 smell sensors 30-39. A selector 58 in the standard selects a small number of the standards having a large contribution to the unknown smell from the result, obtains a considerable concentration in each standard by using the reference level, and calculates the odor strength from the concentration. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
TECHNICAL FIELD The present invention relates to an odor measuring device for measuring what kind of odor or how strong an odor (including all odors and odors) is. The odor measuring device according to the present invention can be applied to a wide range of fields such as an odor measuring device, a device for performing quality inspection and quality evaluation of foods and medicines, and the like.
[0002]
[Prior art]
Heretofore, smell identification and evaluation has generally been performed using the actual sense of smell of humans. However, since it is necessary to assume that the sense of smell fluctuates depending on the individual difference of the person (panel) who actually smells and the physical condition of the day, in order to obtain objective results with high accuracy, secure a certain number of panels or more Also, sufficient consideration must be given to the environment at the test site. Therefore, labor and time are enormous. Even with such considerations, it is difficult to always make a definitive decision based on a certain standard because human olfaction has a characteristic of adapting to smell.
[0003]
On the other hand, in recent years, an odor measurement device using an odor sensor having a response characteristic to odor substances has been disclosed and commercialized in Patent Documents 1 and 2 (for example, FF-Shimadzu Corporation). 1 etc.). In such an odor measuring device, based on detection signals acquired by a plurality of odor sensors, cluster analysis, various multivariate analysis processes such as principal component analysis, or nonlinear analysis using a neural network, etc., are performed. The separation distance (whether or not the odor is in a close category) of a plurality of samples can be determined.
[0004]
[Patent Document 1]
JP-A-11-352088
[Patent Document 2]
JP-A-2002-22692
[0005]
[Problems to be solved by the invention]
However, the conventional odor measuring device as described above is suitable for judging the similarity of a plurality of odors belonging to a relatively narrow category, for example, for the purpose of quality inspection of foods. It was difficult to objectively determine the similarity of a wide variety of odors. Further, since the response sensitivity of the odor sensor is different from the sensitivity of the human nose, in some cases, a result that does not coincide with the sensory test using olfaction may be obtained. In view of these points, the present applicant has proposed a novel configuration of an odor measuring device in Japanese Patent Application Nos. 2002-41465 and 2002-254975.
[0006]
This odor measuring device considers an m-dimensional odor space in which detection outputs of m odor sensors are respectively taken as axes, and positions a vector created from a measurement result of a known standard odor in the m-dimensional odor space. Since the unknown sample to be measured can also be expressed as a vector, the angle between the unknown odor vector and the standard odor vector is determined, and the smaller the angle, the higher the similarity between the two odors. The similarity between the unknown odor and the standard odor of the unknown sample is calculated below. When using a standard odor curve that does not consider the directionality in the m-dimensional odor space, the similarity to the standard odor is calculated by calculating the shortest distance from the measurement point of the unknown sample to each standard odor curve. Is calculated.
[0007]
Now, when trying to apply such an odor measuring device to, for example, an odor measuring device in the atmosphere, the types of odors are very wide, even if it is a odor. Therefore, it is unavoidable that the number of types of standard smells to be measured in advance is increased to some extent. However, since the odor sensor mounted on such an odor measuring device usually has a response sensitivity to a plurality of odor species, the number of types of standard odors is increased, and the unknown odor due to the similarity to the standard odor is increased. It has been found that trying to express the nature of the odor causes a problem that it is rather difficult to capture the characteristics of the smell. Further, it was also found that calculating the concentration, odor intensity, odor index, and the like of the standard odor that contributes to the unknown odor based on such similarity may increase the error.
[0008]
The present invention has been made to solve such problems, and its main purpose is to be able to more accurately express the quality of an unknown odor of an unknown sample, and to improve the concentration and concentration of the unknown odor. It is an object of the present invention to provide an odor measuring device capable of improving the calculation accuracy of an index value indicating the degree of odor intensity.
[0009]
[Means for Solving the Problems]
In order to solve the above problems, the odor measurement device according to the present invention does not necessarily use all of the reference data obtained by measuring a plurality of types of standard odors, such as the type and classification of an unknown sample to be measured. According to the standard, an appropriate standard odor to express the degree of the intensity of the unknown odor is appropriately selected, and the degree of the odor intensity is determined using only the reference data corresponding to the selected standard odor. The index value to represent is calculated. As an operation of selecting an appropriate standard smell, there are a case where a measurer instructs the selection and a case where the apparatus judges the quality of the unknown smell and selects the standard smell.
[0010]
Then, the first odor measuring device according to the present invention is:
a) m (m is an integer of 2 or more) odor sensors having different response characteristics;
b) In an m-dimensional space formed by the measurement results of the m odor sensors, n reference odor vectors or references represented by n (n is an integer of 2 or more) known standard odor measurement results. Reference data acquisition means for creating an odor curve and storing data representing the vector or the curve,
c) Selection for the measurer to select and instruct h (1 ≦ h <n integer) standard odors among the n standard odors in accordance with the type or classification of the unknown sample to be measured. Indicating means;
d) positioning the measurement result of the unknown sample in the m-dimensional space, and determining the unknown value based on the positional relationship with the h reference odor vectors or the standard odor curves corresponding to the selected and designated h odors. An index value indicating the similarity between the unknown odor and the standard odor of the sample, and / or an index value calculating means for calculating an index value indicating the degree of the intensity of the unknown odor;
It is characterized by having.
[0011]
Further, the second odor measuring device according to the present invention comprises:
a) m (m is an integer of 2 or more) odor sensors having different response characteristics;
b) In an m-dimensional space formed by the measurement results of the m odor sensors, n reference odor vectors or references represented by n (n is an integer of 2 or more) known standard odor measurement results. Reference data acquisition means for creating an odor curve and storing data representing the vector or the curve,
c) positioning the measurement result of the unknown sample in the m-dimensional space, and determining the unknown value of the unknown sample based on a positional relationship with n reference odor vectors or reference odor curves corresponding to the n kinds of standard odors. Similarity determining means for determining the similarity between odor and standard odor,
d) standard odor selecting means for selecting h (1 ≦ h <n integer) standard odors out of the n standard odors in accordance with the similarity determination result;
e) the positional relationship between the measurement result of the unknown sample positioned in the m-dimensional space and h reference odor vectors or reference odor curves corresponding to the h kinds of standard odors selected by the standard odor selecting means; An index value calculating means for calculating an index value representing the degree of the strength of the unknown odor,
It is characterized by having.
[0012]
Embodiments and effects of the present invention
When one standard odor is measured by m odor sensors, the intensity signal is output from each odor sensor, so that m data are generated. These are mathematically represented by one point in the m-dimensional scent space. When a plurality of standard samples having different concentrations of the standard odor are measured, a point in the odor space moves in a certain direction with the change in the concentration, so that one vector connecting the points can be considered. . This odor vector has a specific orientation for that odor type. In addition, when the directionality of the point movement with respect to the density change is not taken into consideration, it can be regarded as a curve instead of a vector. In the odor measuring device according to the present invention, the reference data acquiring means creates such a reference odor vector or reference odor curve for each of the n types of standard odors, and stores data representing the vector or the curve in a memory or the like. Here, as the standard odor, for example, an odor representative of various series (aromatic, sulfur, hydrocarbon, etc.) is appropriately selected so as to correspond to various odor species.
[0013]
A measurer who intends to perform measurement by the first odor measuring device according to the present invention has previously obtained some information about the type and classification of the unknown odor of the unknown sample to be measured. For example, when trying to measure a bad smell, if you know where the unknown sample was taken, it is easy to deduce what kind of factory emitted the bad smell. Can be. Therefore, in the first odor measuring device, the measurer inputs the standard odor considered appropriate for performing the odor measurement of the unknown sample by using the selection instructing means based on the information obtained in advance. Preferably, reference information such as a list describing what standard odor is appropriate to be selected according to the type and division of the odor is presented to the measurer.
[0014]
The measurement result of the unknown sample is positioned as a certain point (measurement point) in the m-dimensional space in which the reference odor vector or the reference odor curve is formed as described above. Accordingly, the indicated value calculating means is obtained by changing the positional relationship between the measurement point and the h reference odor vectors corresponding to the standard odor designated by the selection indicating means, or by changing the concentration of the unknown sample. The relationship between the direction of the unknown odor vector and the direction of the h reference odor vectors, and in the case of a standard odor curve, the unknown odor vector according to the shortest distance from the measurement point to each of the h reference odor curves. Determine the similarity to the standard odor. Since the position of each point on the reference odor vector or the reference odor curve corresponds to the concentration of the standard odor component, the position of the reference odor vector or the point on the reference odor curve that gives the shortest distance from the measurement point, etc. From, the converted concentration corresponding to the standard odor in the unknown odor is obtained.
[0015]
On the other hand, in the second odor measuring device according to the present invention, the similarity determining means determines the positional relationship between the measurement points of the unknown sample and n reference odor vectors corresponding to all n types of standard odors, and the unknown sample. The relationship between the direction of the unknown odor vector obtained by measuring while changing the concentration of n and the direction of the n reference odor vectors, and in the case of the standard odor curve, from the measurement point to each of the h reference odor curves The similarity between the unknown odor and the standard odor is determined according to the shortest distance or the like. Then, the standard odor selecting means selects h types of standard odors, which are considered to have a high degree of contribution to the odor intensity of the unknown odor, using the similarity determination result. Alternatively, it is determined which category of the odor belongs to the table created separately for the unknown odor, and a standard odor associated with the category is selected. That is, the similarity determining means and the standard odor selecting means fulfill the functions of prior information on the unknown sample, determination by the measurer, and selection instruction means in the first odor measuring device.
[0016]
Thereafter, the indicated value calculating means obtains a converted concentration corresponding to the standard odor in the unknown odor from the position of a point on the reference odor vector which gives the shortest distance from the measurement point. Then, based on an olfactory threshold reflecting the characteristics of human olfaction or an arithmetic expression including a value equivalent thereto, an index value representing the degree of odor intensity such as odor intensity or odor index is calculated from the converted concentration. I do.
[0017]
In addition, as one embodiment of the first and second odor measuring devices according to the present invention, a suitable standard for calculating the index value is provided for each of a plurality of various odor species to be measured. A table describing a combination of odors is prepared, and a standard odor can be selected by searching for a corresponding section in the table.
[0018]
As described above, according to the odor measuring device according to the present invention, the quality of the unknown odor is expressed using a relatively small number of standard odors, except for the standard odor that contributes little to the intensity of the unknown odor to be measured. Or an index value indicating the degree of the unknown odor intensity is calculated. Therefore, the odor quality can be expressed more accurately and clearly than when expressing the quality of an unknown odor using a large number of standard odors including those with low or little contribution. However, it is easier for the measurer to understand the odor quality.
[0019]
In addition, when estimating the degree of the intensity of an unknown odor, if a standard odor with a low or little contribution is also taken into consideration, an overestimate (overestimation) is made especially when the odor is relatively weak. However, according to the odor measuring device of the present invention, such an overestimate is eliminated or reduced, and the odor intensity can be calculated with higher accuracy than before.
[0020]
【Example】
Hereinafter, an odor measuring device of a first embodiment, which is one embodiment of the present invention, will be described with reference to the drawings. FIG. 1 is a block diagram of the odor measuring device according to the present embodiment.
[0021]
The odor measuring device of the present embodiment has a suction port 1 for aspirating a sample, a pre-processing unit 2 for pre-processing the aspirated sample, and a response characteristic for measuring an unknown sample or a standard sample containing a standard odor component. , A plurality of (in this example, ten, but not limited to) sensor cells 3 having odor sensors 30 to 39, a pump 4 for drawing an unknown sample or a standard sample into the sensor cells 3, and odor sensors 30 to 39 A signal processing unit 5 for analyzing the detection signal, a display unit 6 for displaying the output of the analysis process on a display screen, a control unit 7 for controlling the operation of the entire apparatus, a standard odor selection instruction unit 8 including a keyboard, etc. Is done.
[0022]
In the pre-processing unit 2, removal of water contained in the sample, concentration / dilution of the sample, removal of interfering gas, and the like are performed. The odor sensors 30 to 39 are, for example, oxide semiconductor sensors whose resistance changes according to odor components, such as a conductive polymer sensor, a sensor having a gas adsorption film formed on the surface of a quartz oscillator or a SAW device, or the like. Alternatively, a sensor using another detection method may be used. The signal processing unit 5 (and the control unit 7) is mainly composed of a personal computer. By executing a predetermined program on the personal computer, a peak extraction unit 51, a vector calculation unit 52, a reference vector storage unit 53, and a standard smell are executed. It functions as a converted concentration calculation unit 54, an odor index calculation unit 55, an olfactory threshold storage unit 56, and the like.
[0023]
Note that an odor substance, such as hydrogen sulfide or ammonia, which is difficult to be detected by the oxide semiconductor sensor, may be measured in a portion different from the sensor cell 3 to increase the odor discriminating power.
[0024]
Next, a measurement procedure by the odor measurement device of the present embodiment will be described. In the present odor measuring device, a standard sample containing each of a plurality of types of odor components is measured in advance and stored as reference data. The standard sample is prepared, for example, by mixing and diluting a raw material gas generated from a cylinder (gas), liquid, or solid. As for the raw material gas, a gas at room temperature may be sealed in a gas cylinder, and a predetermined amount may be taken out and used. In the case of a liquid, it may be placed in a glass container or the like, and the odor may be generated by maintaining the temperature at a predetermined value or bubbling nitrogen gas. In addition, a solid substance may generate an odor by maintaining it at a predetermined temperature.
[0025]
Here, assuming that the odor in the atmosphere is mainly measured, three types of aromatic (such as toluene), hydrocarbon (such as heptane), and ester (such as ethyl acetate) are used as the standard of good quality. As the standard smell of malignancy, six types are used: amine (such as trimethylamine), aldehyde (such as butyraldehyde), sulfur (such as methyl mercaptan), organic acid (such as butyric acid), ammonia, and hydrogen sulfide. .
[0026]
With respect to the above nine types of standard odors, standard samples prepared by changing the concentrations are supplied to the inlet 1, and the standard sample is drawn into the sensor cell 3 by operating the pump 4. When the standard sample introduced into the sensor cell 3 comes into contact with the odor sensors 30 to 39, different detection signals are output in parallel from the odor sensors 30 to 39, respectively. In the signal processing unit 5, the peak extracting unit 51 captures a local maximum point (peak top) in the temporal variation of the output signal of one system, and obtains one detection signal for each of the odor sensors 30 to 39. Of course, besides using the peak top as the detection signal, appropriate modifications can be made, such as using the peak area. Therefore, the peak extraction unit 51 obtains a total of ten measurement data DS1, DS2, DS3, DS4, DS5, DS6, DS7, DS8, DS9, and DS10 for one standard sample. Considering a ten-dimensional odor space formed by using detection signals from the ten odor sensors 30 to 39 as axes in different directions, the ten measurement data are stored at one point (DS1, DS2, DS3) in the odor space. , DS4, DS5, DS6, DS7, DS8, DS9, DS10).
[0027]
By measuring a standard sample containing the same standard odor having a different concentration, the measurement point positioned in the 10-dimensional odor space moves in a certain direction according to the change in the concentration. One reference odor vector is created based on a plurality of measurement points obtained for the odor. Then, nine different standard odor vectors S1 to S9 are created by measuring the nine types of standard odors, and data expressing the reference odor vectors are stored in the reference vector storage unit 53. The data stored in the reference vector storage unit 53 is used as reference data for estimating the properties and intensity of the unknown odor contained in the unknown sample.
[0028]
Since it is difficult to illustrate the 10-dimensional space, here, in order to facilitate understanding, two signals formed by detection signals DS1 and DS2 of the first and second two odor sensors as shown in FIG. Let's think in a dimensional space. In this two-dimensional odor space, two measurement data of a certain standard sample by the first and second odor sensors are represented by one point (DS1, DS2), and one type of standard odor is measured. One reference odor vector S created based on the measurement result is positioned, for example, as shown in FIG.
[0029]
Prior to measuring the unknown sample to be measured, the measurer 9 specifies the type of standard odor used for the odor measurement of the unknown sample by the standard odor selection instructing section 8. At this time, the measurer 9 uses, for example, a standard smell selection example list table 10 provided by the device maker.
[0030]
FIG. 3 is a diagram showing an example of the standard smell selection example list table 10. As shown in FIG. Here, the types and divisions of the odors are defined as odor sources, and as shown in the figure, the odor sources are classified into, for example, resin manufacturing plants, fish processing plants, and so on. For each offensive odor sources, a maximum of four types of standard odors suitable for measuring the odor generated from the odor source are designated, and are indicated by a triangle. For example, when the odor source is a landfill, four standard odors suitable for measuring the odor are hydrogen sulfide, amine type, organic acid type, and aldehyde type. Such a standard smell selection example list 10 can be created using, for example, the smell component analysis data and the threshold value of each malodorous substance, and FIG. 3 is merely an example. The standard odors specified in the standard odor selection example list 10 are those that have a high degree of contribution to the sensual odor intensity, regardless of the odor generated from each odor source, regardless of the concentration. You can see that.
[0031]
Even if it is an unknown sample, if the odor source is specified or the odor source can be analogized, up to four types of standard odors can be used using the standard odor selection example list 10 as described above. Can be selected. Then, the selection information given from the standard odor selection instructing section 8 is input to the signal processing section 5 via the control section 7 and held in the standard odor conversion density calculating section 54.
[0032]
When measuring an unknown sample, the unknown sample is supplied to the suction port 1 and the pump 4 is operated to draw the unknown sample into the sensor cell 3. At this time, the output signals from the ten odor sensors 30 to 39 are given to the peak extraction unit 51 as described above, and one detection signal is obtained for each of the odor sensors 30 to 39, and the standard odor conversion concentration calculation unit 54 Is entered.
[0033]
The unknown odor component of the unknown sample is unknown, but as an assumption for estimating the quality of the odor, the synthetic odor of up to four of the nine standard odors selected as described above is used. Suppose there is. Under this assumption, a maximum conversion amount (conversion density) corresponding to each standard odor constituting the unknown odor is calculated using a maximum of four reference odor vectors stored in the reference vector storage unit 53. Specifically, the standard odor conversion density calculation unit 54 calculates the conversion density corresponding to the standard odor by, for example, the following method.
[0034]
For the sake of simplicity, the description will be made in the two-dimensional odor space shown in FIG. First, a point Q on the vector S is searched such that the distance from the measurement point P located in the two-dimensional odor space by the measurement of the unknown sample to the reference odor vector S is the shortest. Since each point on the reference odor vector S is associated with the density of the standard odor, a provisional density corresponding to the standard odor for the unknown odor is calculated based on the position of the point Q (or the length L on the vector S). You can ask. In addition, the shorter the distance d between the measurement point P and the point Q, the higher the similarity between the unknown and the standard can be regarded as high. Therefore, the standard determined as described above according to the distance d can be considered. The temporary concentration corresponding to the odor is corrected and determined as the converted concentration corresponding to the standard odor.
[0035]
The method for obtaining the converted concentration corresponding to the standard from the measurement results of the unknown sample is not limited to this, and various methods can be used. As another example, an unknown odor vector for an unknown sample is created in the odor space in the same manner as the reference odor vector by changing the concentration of the unknown sample, and the angle formed between the unknown odor vector and the reference odor vector. Alternatively, the conversion density corresponding to the standard odor may be calculated based on the length of a vector obtained by orthogonally projecting the unknown odor vector onto the reference odor vector.
[0036]
In each of the odor sensors 30 to 39, when the relationship between the concentration of the odor component and the detection output is linear, the odor vector has a linear shape as shown in FIG. Is nonlinear, the odor vector has a curved shape as shown in FIG. 6 in a two-dimensional (actually ten-dimensional) odor space. However, even in such a case, based on the above-described purpose, the point Q on the vector S where the distance d from the measurement point P to the reference odor vector S is shortest is the same.
[0037]
In addition, the odor vector considers the direction of movement of the measurement point with respect to the density change (density increase), but if the directionality is not considered, the odor vector can be regarded as a simple odor curve. Therefore, it can be said that the reference odor vector is a reference odor curve. That is, as shown in FIG. 7, a reference odor curve is created from measurement points based on sensor outputs when the concentration of the standard odor component is changed, and the shortest distance d from the measurement point P of the unknown odor to the reference odor curve is found. . Then, based on the shortest distance to each reference odor curve, the similarity to the unknown odor can be determined, and the converted density corresponding to the standard odor can be calculated from the density corresponding to the point Q. .
[0038]
Although FIG. 2 shows only the relationship between the measurement point P of the unknown sample and one reference odor vector S, there are actually up to four reference odor vectors in a 10-dimensional odor space. As an example, assuming that the odor source is a refuse treatment plant as described above, as schematically shown in FIG. 5, hydrogen sulfide, amine-based, organic acid-based and aldehyde-based odors are present in a 10-dimensional odor space. There are four reference odor vectors S1, S4, S5, S6. Therefore, the standard odor conversion concentration calculator 54 calculates the distances from the measurement point P of the unknown sample to the reference odor vectors S1, S4, S5, and S6 on the respective vectors S1, S4, S5, and S6 so as to be the shortest. Search for points Q1, Q4, Q5, Q6 respectively. Then, the distances d1, d4, d5, and d6 between the measurement point P of the unknown sample and the points Q1, Q4, Q5, and Q6 are obtained, and the provisional density and distance d1 corresponding to the positions of the points Q1, Q4, Q5, and Q6 are obtained. , D4, d5, and d6, the conversion densities C1, C4, C5, and C6 corresponding to each standard are calculated.
[0039]
The standard odor conversion density calculation unit 54 executes the above-described arithmetic processing, and sends the obtained conversion density to the odor index calculation unit 55. The olfactory threshold storage unit 56 stores olfactory thresholds TH1 to TH9 (not necessarily the olfactory thresholds for the standard odor itself) corresponding to the above nine types of standard odors. This smell threshold may be input by the measurer, or a standard value may be stored by the manufacturer at the time of shipment of the present apparatus.
[0040]
The odor index calculating unit 55 reads the odor threshold corresponding to the selected standard odor from the odor threshold storage unit 56. Then, the converted concentration of each standard odor is divided by an olfactory threshold to convert the odor intensity into an index value of odor intensity reflecting human olfactory characteristics. Further, the odor intensity of the unknown odor is obtained using the odor intensity of each standard odor, and this is converted into an odor index using a predetermined calculation formula. When the maximum value model is used as the intensity of the total odor of the unknown odor, the maximum one of the standard odor components may be selected.On the other hand, when the additive model is used, each standard odor component may be selected. After obtaining the odor index, the odor index is once converted into an odor concentration, and all of them are added, and then the odor index is calculated again. The signal processing unit 5 displays the calculated odor index and other various index values on the screen of the display unit 6 and presents them to the measurer 9.
[0041]
As described above, the odor measuring device of the first embodiment accurately calculates the odor index of the odor of the unknown sample from among a large number of standard odors prepared in advance, based on the input instruction by the measurer. A suitable standard odor is selected as appropriate, and the odor index of the unknown odor is calculated using the measurement result of the standard odor. Although a measurement example that has demonstrated the effect will be described later, the accuracy can be significantly improved as compared with a case where an odor index is calculated using all of a large number of prepared odors. Further, even when the characteristics of an odor are represented by similarities to the respective standard odors, the characteristics of the unknown odor become clearer by limiting the types of the standard odors.
[0042]
Next, a second embodiment which is one embodiment of the odor measuring device according to the second invention will be described. FIG. 8 is a block diagram of the odor measuring device according to the second embodiment. The same components as those of the first embodiment are denoted by the same reference numerals, and description thereof is omitted.
[0043]
The essential difference from the first embodiment is that, in the first embodiment, the work of selecting the standard smell was performed by the measurer, but in the second embodiment, the signal processing unit 5 automatically required the work. Is to choose a good standard smell. In order to achieve such an operation, the signal processing unit 5 functionally includes a similarity determination unit 57 and a standard odor selection unit 58.
[0044]
The characteristic operation of the odor measuring device of the second embodiment will be described. Until a plurality of types (here, nine types) of standard odors are measured in advance, nine reference odor vectors S1 to S9 are created, and data representing the vectors S1 to S9 is stored in the reference vector storage unit 53. Is the same as in the first embodiment.
[0045]
In the measurement of an unknown sample, if there is no information on the unknown odor of the unknown sample (for example, the odor source described above), the measurer cannot give an instruction to select an appropriate standard odor. Therefore, in the present apparatus, after acquiring the detection signals of the odor sensors 30 to 39 for the unknown sample, the similarity determination unit 57 determines the reference odor vectors S1 to S9 based on all nine types of standard odors without selecting the standard odor. , The similarity to each standard odor is calculated from the positional relationship between the measurement point P of the unknown sample in the 10-dimensional odor space and each of the reference odor vectors S1 to S9. Specifically, as shown in FIG. 4, the shortest distances between the measurement point P and the reference odor vectors S1 to S9 are obtained, and the similarity is defined according to the distances. Alternatively, an unknown odor vector may be created by changing the concentration of the unknown sample to measure, and the similarity may be defined according to the angle formed between the unknown odor vector and the reference odor vector.
[0046]
In this way, the similarities to the nine types of standard odors are obtained. Since these similarities can be regarded as representing the quality of unknown odors, the standard odor selecting unit 58 determines, for example, based on the magnitude relation between the similarities, that the unknown odors are classified into odor species as shown in FIG. Here, it is determined which of the two types is classified. Then, a standard odor predetermined according to the classification of the odor species is selected. For example, if the unknown odor is classified as a bad smell of a rubber factory based on the similarity, two types of hydrogen sulfide and sulfur type are selected as the standard odor.
[0047]
The selection instruction information is sent to the standard odor conversion density calculating unit 54, and the odor index of the unknown odor is calculated using only the reference odor vector based on the selected standard odor. The method is the same as in the first embodiment. Therefore, in the odor measuring device of the second embodiment, even when there is no prior information about the unknown sample, the odor index of the unknown odor and other index values relating to the odor intensity are highly accurate as in the first embodiment. Can be calculated.
[0048]
In both the first and second embodiments, the odor index of the unknown odor is obtained by limiting the types of the standard odors. However, the case where the odor index is actually calculated by using all nine types of the standard odors, The difference in calculation accuracy between the case where the four types of standard odors with high contributions are selected and the odor index is calculated will be described with reference to actual measurement examples.
[0049]
FIG. 9A shows a case where nine types of standard odors are used (that is, a case where standard odors are not selected), and FIG. 9B shows a case where only four types of standard odors having a high contribution are used. It is a graph showing the error of each odor index. In these graphs, the abscissa indicates the odor index obtained by a sensory test performed by six panels, and the ordinate indicates the odor index calculated by the present apparatus (here, referred to as the predicted odor index). That is, the straight lines indicated by the dotted lines in FIGS. 9A and 9B are in a state where the both odor indexes match, and the error is zero.
[0050]
As shown in FIG. 9 (A), when the analysis process is performed using all nine types of standard odors prepared to cope with various types of odors, a region having a particularly low odor index, that is, a odor is compared. When the target is in an extremely weak state, the tendency that the predicted odor index becomes excessive and the error increases is remarkable. This is considered to be because the odor intensity (or odor index) of the unknown odor was overestimated by adding the odor intensity contributed by each standard odor when the number of the standard odors is large.
[0051]
By limiting the types of standard odors, such overestimates are eliminated, and as shown in FIG. 9 (B), even in the low odor index region, the predicted odor index does not become excessive, and the error is reduced to reduce the predicted odor index. It can be seen that the calculation accuracy has been improved. Thus, by narrowing down the types of standard odors according to the odor to be measured, the calculation accuracy of the odor index and other index values relating to the intensity of the odor is improved, and the possibility of being able to substitute sensory tests in particular is remarkable. Increase.
[0052]
It is to be noted that each of the above embodiments is an example of the present invention, and it is apparent that changes and modifications can be made as appropriate within the scope of the present invention.
[0053]
For example, in the second embodiment, the standard odor selecting unit 58 selects a standard odor suitable for calculating the odor index from the similarity of the unknown odor, and then notifies the measurer of the odor by display or the like, and the measurer determines the odor index based on the notification. Alternatively, a standard odor may be selected and instructed. Further, a process of directly selecting a standard odor can be performed without going through a procedure such as classification into odor species.
[0054]
Further, in the above embodiment, it was assumed that the measurement of the standard sample was completed in advance and the reference odor vector was created.For example, in an apparatus set so that a plurality of standard odors can be automatically measured, After the selection instruction is made by the measurer, the measurement of the selected standard odor is performed to create a required reference odor vector, and the odor index of the unknown odor is calculated using the reference odor vector. Good.
[Brief description of the drawings]
FIG. 1 is a block diagram of an odor measuring device according to a first embodiment of the present invention.
FIG. 2 is an explanatory diagram of a measurement principle of the odor measurement device of the first embodiment.
FIG. 3 is a view showing an example of a standard odor selection list used for measurement by the odor measuring device of the first embodiment.
FIG. 4 is a conceptual diagram showing an example of the arrangement of all nine reference odor vectors in a 10-dimensional odor space in the odor measurement device of the first embodiment.
FIG. 5 is a conceptual diagram showing an example of the arrangement of four selected reference odor vectors in a 10-dimensional odor space in the odor measurement device of the first embodiment.
FIG. 6 is a supplementary explanatory view of the measurement principle of the odor measurement device of the first embodiment.
FIG. 7 is an auxiliary explanatory diagram of the measurement principle of the odor measurement device of the first embodiment.
FIG. 8 is a block diagram of an odor measuring device according to a second embodiment of the present invention.
FIG. 9 is a diagram for explaining a difference in accuracy between a case where an odor index is calculated using all nine standard odors and a case where an odor index is calculated by selecting four types of standard odors having a high contribution. Graph.
[Explanation of symbols]
1. Inlet
2 ... Pre-processing unit
3. Sensor cell
30-39… Odor sensor
4 ... Pump
5 ... Signal processing unit
51 ... Peak extraction unit
52 ... vector operation unit
53: Reference vector storage unit
54: Standard odor conversion concentration calculator
55: Odor index calculator
56: Olfactory threshold storage unit
57 ... Similarity determination unit
58 ... Standard odor selection unit
6 Display unit
7 ... Control unit
8 ... Standard smell selection indicator
9 ... Measurer
10. Standard smell selection list

Claims (3)

a) m (m is an integer of 2 or more) odor sensors having different response characteristics;
b) In an m-dimensional space formed by the measurement results of the m odor sensors, n reference odor vectors or references represented by n (n is an integer of 2 or more) known standard odor measurement results. Reference data acquisition means for creating an odor curve and storing data representing the vector or the curve,
c) Selection for the measurer to select and instruct h (1 ≦ h <n integer) standard odors among the n standard odors in accordance with the type or classification of the unknown sample to be measured. Indicating means;
d) positioning the measurement result of the unknown sample in the m-dimensional space, and determining the unknown value based on a positional relationship with h reference odor vectors or reference odor curves corresponding to the selected and designated h types of standards. An index value indicating the similarity between the unknown odor and the standard odor of the sample, and / or an index value calculating means for calculating an index value indicating the degree of the intensity of the unknown odor;
An odor measuring device comprising: a) m (m is an integer of 2 or more) odor sensors having different response characteristics;
b) In an m-dimensional space formed by the measurement results of the m odor sensors, n reference odor vectors or references represented by n (n is an integer of 2 or more) known standard odor measurement results. Reference data acquisition means for creating an odor curve and storing data representing the vector or the curve,
c) positioning the measurement result of the unknown sample in the m-dimensional space, and determining the unknown value of the unknown sample based on a positional relationship with n reference odor vectors or reference odor curves corresponding to the n kinds of standard odors. Similarity determining means for determining the similarity between odor and standard odor,
d) standard odor selecting means for selecting h (1 ≦ h <n integer) standard odors out of the n standard odors in accordance with the similarity determination result;
e) the positional relationship between the measurement result of the unknown sample positioned in the m-dimensional space and h reference odor vectors or reference odor curves corresponding to the h kinds of standard odors selected by the standard odor selecting means; An index value calculating means for calculating an index value representing the degree of the strength of the unknown odor,
An odor measuring device comprising: Create a table that describes a combination of standard odors suitable for calculating the index value for each of the various categories of various odor species to be measured, and search for the corresponding category in the table. The odor measuring device according to claim 1, wherein a standard odor is selected by the following.

Images