JP4016611B2 - Odor identification device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus for outputting (displaying, printing out, etc.) the results measured by an odor measuring device, a bad odor measuring device, a gas alarm device, etc. so as to be easily understood by a measurer or a user.
[0002]
[Prior art]
The odor sensor measures only the intensity of the odor by measuring the concentration of a specific substance with a specific gas sensor, and also identifies the type of odor as well as the odor intensity by using multiple gas sensors. There is.
[0003]
In the latter type of odor sensor, conventionally, only the relative relationship between the measured sensors is represented based on the detection signal acquired by each gas sensor. Specifically, sample A and sample B are close, but sample C is distant. The identification method includes, for example, a method for determining the similarity of a combination pattern of detection outputs of a plurality of gas sensors, a method using various multivariate analysis methods such as cluster analysis and principal component analysis, and a method using a neural network. It was used.
[0004]
[Problems to be solved by the invention]
In the latter type of odor sensor, the relative comparison between the measured samples was easy, but an absolute measurement could not be performed.
[0005]
The present invention has been made to solve these problems, and its object is to provide an absolute representation of various actual odors with relatively few types of standard odors. An object of the present invention is to provide an apparatus that can perform the above.
[0006]
[Means for Solving the Problems]
An odor identification device according to the present invention, which has been made to solve the above problems ,
a) m (m is an integer of 2 or more) odor sensors;
b) storage means for storing the measurement results of the m odor sensors for each of n (n is an integer of 2 or more) standard odors;
c) In an m-dimensional odor space formed by the measurement results of the m odor sensors, the odor vector represented by the measurement result of the sample and the n odor vectors represented by the measurement results of the stored standard odor. Vector computing means for computing the relationship with the standard odor vector;
d) an output means for visually outputting the calculation result;
The vector calculation means calculates the angle between the odor vector of the sample and each standard odor vector, and outputs the figure representing the odor vector of the sample only with the standard odor vector whose angle is a predetermined value or less. It is characterized by being sent to the means.
[0007]
DETAILED DESCRIPTION OF THE INVENTION
When one standard odor is measured by m odor sensors, the intensity is output from each odor sensor, so m data are generated. These can be considered mathematically as representing one vector (odor vector) in an m-dimensional space (odor space). Accordingly, when n standard odors are measured, n odor vectors in an m-dimensional odor space are generated. The storage means stores these data.
[0008]
Next, when the sample is measured with the same m odor sensors, the data is also represented by one vector in the m-dimensional odor space. By calculating the relationship between the odor vector of the sample and the n standard odor vectors described above by various methods and visually outputting the result, the user can detect the odor of the sample and n standard odors. Can be grasped visually. As a result, for example, it is possible to easily visually grasp which standard odor the sample is close to, or which standard odor is close to mixing with which standard.
[0009]
【The invention's effect】
In the odor identification apparatus according to the present invention, conventionally, it can be expressed only by simple numerical values (odor index, odor intensity) and sensory words ("odor close to burnt odor", "odor like rot smell of fish", etc.). The odor of the missing sample can be visually represented. This makes it possible to grasp the relationship with existing standards more quantitatively and intuitively. In addition, since the relationship between standards can be visually expressed in this way, it is not always possible to express not only an odor that is close to an existing standard but also a single standard, but multiple standards. The odor of the sample that can be expressed in relation to the odor can also be expressed appropriately. In other words, the odor of the sample can be identified with only a relatively small number of standards without preparing a large number of standards.
[0010]
【Example】
More detailed contents and effects of the present invention will be described with reference to specific examples. FIG. 1 is a block diagram showing an example of an odor discriminating apparatus 10 embodying the present invention. The apparatus includes a suction port 11 for sucking a sample, a pretreatment unit 12 for preprocessing the sucked sample, a standard odor generating unit 13 for generating a standard odor, a three-way valve 14 for switching between the sample and the standard odor, a sample or The sensor unit 15 having a plurality of odor sensors for measuring a standard odor, a suction pump 16 for drawing a sample or standard odor toward the sensor unit 15, and the operation of the entire apparatus are controlled. It comprises a control / processing unit 17 and the like for inputting and analyzing the measured results.
[0011]
The pretreatment unit 12 performs removal of moisture, concentration / dilution of the sample, removal of interfering gas, and the like. In the standard odor generating unit 13, various standard odors are produced by mixing source gases generated from cylinders (gas), liquid, and solid. As for the raw material gas, if it is a gas at normal temperature, it is sealed in a gas cylinder and a certain amount is taken out and used. In the case of liquid, it is put in a glass container or the like, and an odor is generated by maintaining a predetermined temperature or bubbling nitrogen gas. Solid materials generate odors by maintaining a predetermined temperature. The control / processing unit 17 generates a necessary type of standard odor as it is or by controlling the mixing ratio.
[0012]
The control / processing unit 17 operates the pump 16 and switches the three-way valve 14, thereby drawing the first standard odor S ₁ thus produced into the sensor unit 15. In the sensor unit 15, each of the plurality of odor sensors measures the strength of the first standard odor S ₁ and outputs the measurement result to the control / processing unit 17. Here, assuming that the number of odor sensors is ten, for example, ten measurement data DS _1,1 , DS _2,1 ,..., DS _10,1 are sent from the sensor unit 15 to the control / processing unit 17. Become. The control / processing unit 17 stores the transmitted data in a storage device (memory, HDD, etc.). These 10 data constitute one vector F _S1 (DS _1,1 , DS _2,1 ,..., DS _10,1 ) in a 10-dimensional odor space formed by the outputs of 10 sensors. Can think.
[0013]
Next, the standard odor generating unit 13 is controlled to generate the second standard odor S ₂ and pulled into the sensor unit 15. Accordingly, another 10 pieces of measurement data DS _1,2 , DS _2,2 ,..., DS _10,2 are sent from the sensor unit 15 to the control / processing unit 17 and stored therein. These data constitute another vector F _S2 (DS _1,2 , DS _2,2 ,..., DS _10,2 ) in the 10-dimensional odor space.
[0014]
Thus, for example, when eight kinds of standard odors are generated and measured by ten odor sensors, 80 data of DS _1,1 , DS _2,1 ,..., DS _10,8 , DS _10,8 are controlled and It is stored in the storage device of the processing unit 17. These are 8 vectors F _S1 (DS _1,1 , DS _2,1 , ..., DS _10,1 ), ..., F _S8 (DS _1,8 , DS _2,8 , ..., DS in a 10-dimensional odor space _10,8 ). The direction of each standard odor vector F _Sk (k = 1 to 8) represents the fragrance of the standard odor S, and the magnitude of the vector F _Sk represents the intensity of the standard odor.
[0015]
In general, since different types of odors are produced as standard odors, these standard odor vectors F _Sk (k = 1 to 8) are mathematically independent vectors in a 10-dimensional odor space. A space formed by these eight standard odor vectors F _{S1 to} F _S8 (a subspace of the 10-dimensional odor space) is called a standard odor space.
[0016]
Next, the three-way valve 14 is switched, and the odor of the sample to be measured (sample 1) is drawn into the sensor unit 15 and measured. The measurement results of the sample 1 by the ten odor sensors are also sent to the control / processing unit 17 and stored in the storage device. These data also constitute one vector F _U1 (DU _1,2 , DU _2,2 ,... DU _10,2 ) in the odor space.
[0017]
In the control / processing unit 17, data DU _1,1 , DU _2,2 ,... DU _10,2 of the measurement result of the sample and the standard odor data measured and stored in advance are as follows. Perform the operation.
T _k = {(F _U1・ F _Sk ) / | F _Sk | ² } ・ F _Sk
Here, F _U1 represents the odor vector of sample 1, F _Sk (k = 1 to 8) represents each standard odor vector, and (F _U1 · F _Sk ) represents the inner product of both vectors. T _k obtained by the above formula is mathematically an orthogonal projection vector of the odor vector F _U1 of the sample to the standard odor vector F _Sk (FIG. 2).
[0018]
The magnitude | T _k | of the orthogonal projection vector T _k (k = 1 to 8) to each standard odor vector F _Sk (k = 1 to 8) of the odor vector of the sample 1 It can be considered as the contribution rate of the standard odor (the unit is the magnitude of the standard odor vector F _Sk ). An example of the contribution ratio of each standard odor for the two types of samples is shown in FIG.
[0019]
Further, the control / processing unit 17 outputs the value of the contribution rate thus calculated in a form that is easy for the user to understand. An example of the output is shown in FIG. Note that the value of the contribution ratio of each standard odor to the odor of the sample may change when the odor intensity of the sample (the size of the odor vector of the sample | F _U |) changes. That is, the odor quality of a sample may change depending on the intensity of the odor. Therefore, when outputting the measurement result of the odor of the sample, it is desirable to output the odor intensity of the sample at the same time. Alternatively, a predetermined vector length may be determined, and the concentration of the sample may be adjusted so that the vector representing the measurement result becomes the length.
[0020]
Although FIG. 4 shows a bar graph format, it can be output in various formats such as a pie chart (pie graph) and a web (spider web) graph. As an output destination, it is desirable to enable printout in addition to the display. With these visual outputs, the user can easily know what kind of standard odor the sample is and how it is composed.
[0021]
Another output example is described next. In the above example (FIG. 4), all the contribution ratios of each standard odor are output. For example, when the odor vector of the sample is close to a vector of a certain standard odor (the angle between the two is small), such a contribution ratio. Is also meaningful, but if it is far from the standard odor vector (the angle between the two is large), the odor of the sample is qualitatively different from the standard odor, even if the contribution value is large. Can be considered different. When the absolute value of the standard odor vector | F _Sk | is small, it may occur that the angle is large but the contribution ratio is large. For example, in the case of FIG. 2, the contribution rate of the standard odor S ₂ is large, but the sample odor vector and the standard odor vector S ₂ are considerably separated from each other, and the standard odor S ₂ has some relationship with the sample odor. Is not reasonable.
[0022]
Therefore, the control / processing unit 17 calculates the angle (θ ₁ , θ _{2 in} FIG. ₂ ) between the odor vector of the sample and each standard odor vector, and the angle is a certain value (for example, 5 degrees or 10 degrees). If it is above, the contribution rate of the standard odor may not be output. An example is shown in FIG. As a result, the odor of the sample can be expressed more concisely, and the user can more easily understand and judge the quality of the sample odor.
[Brief description of the drawings]
FIG. 1 is a block diagram of a configuration example of an odor discriminating apparatus according to the present invention.
FIG. 2 is an explanatory diagram showing a relationship between a sample odor vector and a standard odor vector.
FIG. 3 is a calculation example of contribution ratios of eight kinds of standards to Sample 1 and Sample 2.
FIG. 4 shows an example of the output of the odor discriminating apparatus according to the present invention.
FIG. 5 shows another output example of the odor discriminating apparatus according to the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... Odor identification apparatus 11 ... Suction port 12 ... Pre-processing part 13 ... Standard odor generating part 14 ... Three-way valve 15 ... Sensor part 16 ... Suction pump 17 ... Control and processing part

Claims (2)

a) m (m is an integer of 2 or more) odor sensors;
b) storage means for storing the measurement results of the m odor sensors for each of n (n is an integer of 2 or more) standard odors;
c) In an m-dimensional odor space formed by the measurement results of the m odor sensors, the odor vector represented by the measurement result of the sample and the n odor vectors represented by the measurement results of the stored standard odor. Vector computing means for computing the relationship with the standard odor vector;
d) an output means for visually outputting the calculation result;
The vector calculation means calculates the angle between the odor vector of the sample and each standard odor vector, and outputs the figure representing the odor vector of the sample only with the standard odor vector whose angle is a predetermined value or less. An odor discriminating apparatus characterized by being sent to a means. The vector calculation means calculates the contribution ratio of each standard odor to the sample odor based on the inner product of the sample odor vector and each standard odor vector, and the standard odor vector whose angle with the sample odor vector is a predetermined value or less. The odor discriminating apparatus according to claim 1 , wherein a diagram representing the odor vector of the sample using the contribution ratio is sent to the output means.

Images