CN102654494A - Method for establishing quality identification and detection standard for agricultural products - Google Patents
Method for establishing quality identification and detection standard for agricultural products Download PDFInfo
- Publication number
- CN102654494A CN102654494A CN2012101090294A CN201210109029A CN102654494A CN 102654494 A CN102654494 A CN 102654494A CN 2012101090294 A CN2012101090294 A CN 2012101090294A CN 201210109029 A CN201210109029 A CN 201210109029A CN 102654494 A CN102654494 A CN 102654494A
- Authority
- CN
- China
- Prior art keywords
- information
- sensor
- agricultural product
- quality
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 230000004927 fusion Effects 0.000 claims abstract description 26
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 22
- 210000002569 neuron Anatomy 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 13
- 230000005540 biological transmission Effects 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 238000010561 standard procedure Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 230000001537 neural effect Effects 0.000 claims description 5
- 230000008030 elimination Effects 0.000 claims description 4
- 238000003379 elimination reaction Methods 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 4
- 238000000513 principal component analysis Methods 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 7
- 238000010606 normalization Methods 0.000 abstract description 6
- 238000005070 sampling Methods 0.000 abstract 2
- 230000001594 aberrant effect Effects 0.000 abstract 1
- 230000000007 visual effect Effects 0.000 abstract 1
- 241000251468 Actinopterygii Species 0.000 description 6
- 241000252230 Ctenopharyngodon idella Species 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 239000013505 freshwater Substances 0.000 description 3
- 235000013372 meat Nutrition 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 244000025254 Cannabis sativa Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 235000021393 food security Nutrition 0.000 description 1
- 235000020997 lean meat Nutrition 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000284 resting effect Effects 0.000 description 1
- 238000003307 slaughter Methods 0.000 description 1
- 210000001835 viscera Anatomy 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention relates to a method for establishing a quality identification and detection standard for an agricultural products, belonging to the technical field of quality identification. The method comprises the following steps of: acquiring characteristic information by using a characteristic sensor, and displaying the acquired information on an LabVIEW visual interface; firstly removing aberrant points from the acquired information, and then extracting characteristics from the information through SPSS software, and carrying out normalization processing on the characteristic data; carrying out characteristic layer fusion on the normalized characteristic data to obtain a fusion result; and sampling the characteristic information of the same agricultural product multiple times and solving the result of each characteristic layer fusion, establishing a quality identification standard according to the fusion results of sampling multiple times, and establishing an expert system according to the established standard. According to the method, a system can acquire more thorough information based on the multi-sensor information fusion technology, thereby realizing comprehensive and accurate understanding on an object to be measured so that the established identification standard is more accurate and the detection result is more precise..
Description
Technical field
The present invention relates to a kind of quality of agricultural product recognition detection establishment of standard method, belong to technical field of quality identification.
Background technology
With the increase in agricultural products in China market, the quality of agricultural product is also increasingly paid close attention to by people.The quality of agricultural product directly determines its mouthfeel, therefore, has highly important meaning to the Quality Detection research of agricultural product.The detection to quality of agricultural product was all to use single detection method in the past, and the Testing index of quality of agricultural product is many, such as color, smell, pH value and other characteristics, single detection can not include information in all directions, there is limitation to the understanding of recognition result, and there may be error.Multi-sensor information fusion technology, utilize difference of the various sensors in performance and the complementary defect for making up single detection method, overcome the limitation of single detection method, system is obtained more fully information, so as to realize the comprehensive and accurate understanding to measurand.
Meanwhile, quality of agricultural product Standard System Construction is related to the problems such as rural economic development, food security, raising of people's quality of life.Agricultural products in China quality standard weak foundation, existing standard is not complete, especially it is the absence of influential standard in the world, it is related to ten major classes and dominates quality of agricultural product standard and its planning standard, specification etc., comprehensive standard system is not yet formed, and the agricultural product with local characteristic lack complete careful provincial standard；Agricultural product quality detection architecture is unsound, and agricultural product safety management grade lack of standardization is all current more distinct issues.Therefore, the method for setting up quality of agricultural product discrimination standard and its detection seems necessary.
The content of the invention
For the deficiency of existing issue, the present invention provides a kind of quality of agricultural product recognition detection establishment of standard method, this method is based on multi-sensor information fusion technology, this multi-sensor information fusion technology can make system obtain more fully information, so as to realize the comprehensive and accurate understanding to measurand, so that the discrimination standard set up is more accurate, testing result is more accurate.
The technical solution adopted by the present invention is：The foundation of quality of agricultural product discrimination standard is carried out according to below step：
The first step, data acquisition；
According to the Feature Selection feature sensor of agricultural product to be measured itself, using feature sensor acquisition characteristics information, the information collected is shown in LabVIEW visualization interfaces；
Second step, data prediction；
The information collected is carried out to wild point rejecting processing first and then by SPSS softwares, utilize principal component analysis feature extraction formula
3rd step, Fusion Features；
Normalized characteristic is subjected to Feature-level fusion using BP artificial neural networks, by Matlab softwares, fusion results are obtained；
4th step, set up standard, expert system；
Repeatedly same agricultural product characteristic information is sampled, and try to achieve the result of each Feature-level fusion, set up quality discrimination standard according to the fusion results of multiple repairing weld, expert system is set up according to the standard of foundation.
The feature sensor of selection is smell sensors array or other feature sensors, and the feature sensor of selection can be one or more of.
The wild point elimination method of use is specially that the information that each sensor is collected is averaged first, then being compared each value collected and average, the larger information of those deviation averages is rejected, the information to each sensor carries out wild point rejecting processing.
BP artificial neural networks are according to Kolmogorov theorems using a N
(2N+1)
M 3 layers of BP networks, wherein N represents the sample number of input layer, is determined according to the number of features of selection；M represents that output layer passes through the sample number of neural metwork training, is determined according to the grade of agricultural product, and the information detected point training set and test set are substituted into network simulation is trained, and by training, just quality grade can be made a distinction by output result.
The transmission function selection S type tans of intermediate layer neuron, the transmission function selection S logarithmic functions of output layer neuron, Select Error target is 0.001, is set for the training of 5000 steps, and Studying factors are set to 0.01, and factor of momentum is set as 0.9.
The feature sensor of selection can be smell sensors array, feature（Non- smell）Sensor etc., the feature sensor of selection can be one or more of.
The wild point elimination method of use is specially that the information that each sensor is collected is averaged first, then being compared each value collected and average, the larger information of those deviation averages is rejected, the information to each sensor carries out wild point rejecting processing.
BP artificial neural networks are according to Kolmogorov theorems using a N(2N+1)
M 3 layers of BP networks, wherein N represents the sample number of input layer, is determined according to the number of features of selection；M represents that output layer passes through the sample number of neural metwork training, is determined according to the grade of agricultural product, and the information detected point training set and test set are substituted into network simulation is trained, and by training, just quality grade can be made a distinction by output result.
The transmission function selection S type tans of intermediate layer neuron, the transmission function selection S logarithmic functions of output layer neuron, Select Error target is 0.001, is set for the training of 5000 steps, and Studying factors are set to 0.01, and factor of momentum is set as 0.9.
Beneficial effects of the present invention：Based on multi-sensor information fusion technology, this multi-sensor information fusion technology can make system obtain more fully information, so as to realize the comprehensive and accurate understanding to measurand so that the discrimination standard of foundation is more accurate, and testing result is more accurate.
Brief description of the drawings
Fig. 1 is flow chart of data processing schematic diagram of the invention；
Fig. 2 is neural network topology structure schematic diagram of the invention；
Fig. 3 is the flow chart of data processing schematic diagram of the embodiment of the present invention；
Fig. 4 is the structural schematic block diagram of the embodiment of the present invention.
In figure：The smell sensors array of many gas sensor compositions of 1-, 2-PCI1716 data collecting cards, 3- feature sensors, 4-STC89C52 single-chip microcomputers, 5-RS232 serial communications line, 6- host computers computer, 7- Virtual instrument LabVIEWs.
Embodiment
The present invention is described further with reference to the accompanying drawings and examples, to facilitate the technical staff to understand.
As shown in Figure 1：The foundation of quality of agricultural product discrimination standard is carried out according to below step：
The first step, data acquisition；
According to the Feature Selection feature sensor of agricultural product to be measured itself, using feature sensor acquisition characteristics information, the information collected is shown in LabVIEW visualization interfaces；
Second step, data prediction；
The information collected is carried out to wild point rejecting processing first and then by SPSS softwares, utilize principal component analysis feature extraction formula
3rd step, Fusion Features；
Normalized characteristic is subjected to Feature-level fusion using BP artificial neural networks, by Matlab softwares, fusion results are obtained；
4th step, set up standard, expert system；
Repeatedly same agricultural product characteristic information is sampled, and try to achieve the result of each Feature-level fusion, set up quality discrimination standard according to the fusion results of multiple repairing weld, expert system is set up according to the standard of foundation.
The feature sensor chosen in the above-mentioned methods is smell sensors array or other feature sensors, and the feature sensor of selection can be one or more of.
The wild point elimination method of use is specially that the information that each sensor is collected is averaged first, then being compared each value collected and average, the larger information of those deviation averages is rejected, the information to each sensor carries out wild point rejecting processing.
As shown in Figure 2：BP artificial neural networks are according to Kolmogorov theorems using a N(2N+1)M 3 layers of BP networks, wherein N represents the sample number of input layer, is determined according to the number of features of selection；M represents that output layer passes through the sample number of neural metwork training, is determined according to the grade of agricultural product, and the information detected point training set and test set are substituted into network simulation is trained, and by training, just quality grade can be made a distinction by output result.
The transmission function selection S type tans of intermediate layer neuron, the transmission function selection S logarithmic functions of output layer neuron, Select Error target is 0.001, is set for the training of 5000 steps, and Studying factors are set to 0.01, and factor of momentum is set as 0.9.
Embodiment
Using Kunming grass carp as agricultural product to be measured, after grass carp is bought from the market of farm produce, after slaughtering, except scaling, internal organ, head, tail, and clear water wash clean is used, take fish back lean meat as sample sample.
Experiment is, in order to which the freshwater fish meat to different freshness is identified, to have carried out testing for 6 days altogether, and 30 parts of testing samples are determined daily.Due to sample height corruption at the 5th day, then measure nonsensical through housing the sample of 6 days to oneself, therefore only preceding 5 days samples are analyzed.
The first step, data acquisition；
According to the characteristics of grass carp, odor characteristics, chemical feature-pH value and the physical features of grass carp-electrical conductivity tripartite's region feature are the important symbols of the quality judgement of grass carp, so gathering information from smell sensors array, pH value sensor, conductivity sensor, the odor characteristics of wherein grass carp are most important, are first principal component value.
As shown in Figure 3：Utilize smell sensors array, pH value sensor, conductivity sensor collection information.
Second step, data prediction；
As shown in Figure 4：The data message of information, conductivity sensor to smell sensor and the data message of pH value sensor carry out wild point and rejected and normalized；Detailed process is the same, and carrying out wild point by the data message of pH value sensor below rejects with being illustrated exemplified by normalized.
The pH value of first day 30 parts of sample is：
6.4 6.3 6.5 6.6 6.5 6.7 6.4 6.6 6.5 6.3
6.5 6.6 6.5 6.7 6.7 6.3 6.6 6.4 6.7 6.6
6.5 6.6 6.7 6.5 6.8 6.6 6.8 6.5 6.7 6.4
To above-mentioned 30 number according to averaging, it is 6.55 to be computed its average, then carries out wild point and rejects, is weeded out 6.55 big data are deviateed.Through analysis, 6.8 and 6.3 are rejected, three 6.3 and two 6.8 altogether, is rejected by open country point, also remains 25 data, it is as follows：
6.4 6.5 6.6 6.5 6.7 6.4 6.6 6.5
6.5 6.6 6.5 6.7 6.7 6.6 6.4 6.7
6.6 6.5 6.6 6.7 6.5 6.6 6.5 6.7 6.4
To pH value sensor the 2nd, 3,4, the data of 5 days also carry out above-mentioned same data and reject process.
Data after being rejected through wild point for 2nd day:
6.8 6.8 6.9 6.9 6.8 6.8 6.9 6.8
6.9 6.9 6.8 7.0 6.9 6.8 6.9 7.0
6.8 6.9 6.9 6.8 7.0 6.9 7.0 6.8 6.9
Data after being rejected through wild point for 3rd day:
7.1 7.1 7.2 7.1 7.2 7.2 7.1 7.2
7.2 7.3 7.2 7.1 7.2 7.1 7.1 7.3
7.2 7.1 7.3 7.2 7.2 7.3 7.1 7.2 7.1
Data after being rejected through wild point for 4th day:
7.4 7.5 7.4 7.6 7.4 7.5 7.4 7.4
7.6 7.4 7.5 7.5 7.6 7.5 7.4 7.4
7.4 7.6 7.5 7.5 7.6 7.4 7.5 7.4 7.6
Data after being rejected through wild point for 5th day:
7.8 7.7 8.0 7.8 7.9 7.7 7.8 7.9
8.1 7.7 8.0 7.9 7.7 7.7 7.8 7.9
8.0 8.0 8.1 7.9 7.8 7.7 7.7 8.0 7.7
Then totally 125 data that the process data of 5 days all to pH value sensor are rejected again are normalized, and its formula is：, wherein
=8.1,
=6.4, the data after wild point is rejected are normalized and obtained：
Data after the normalization in the 1st day of pH value sensor：
0.0000 0.0588 0.1176 0.0588 0.1765 0.0000 0.1176 0.0588
0.0588 0.1176 0.0588 0.1765 0.1765 0.1176 0.0000 0.1765
0.1176 0.0588 0.1176 0.1865 0.0588 0.1176 0.0588 0.1765 0.0000
Data after normalization in 2nd day：
0.2353 0.2353 0.2941 0.2941 0.2353 0.2353 0.2941 0.2353
0.2941 0.2941 0.2353 0.3529 0.2941 0.2353 0.2941 0.3529
0.2353 0.2941 0.2941 0.2353 0.3529 0.2941 0.3529 0.2353 0.2941
Data after normalization in 3rd day：
0.4118 0.4118 0.4706 0.4118 0.4706 0.4706 0.4118 0.4706
0.4706 0.5294 0.4706 0.4118 0.4706 0.4118 0.4118 0.5294
0.4706 0.4118 0.5294 0.4706 0.4706 0.5294 0.4118 0.4706 0.4118
Data after normalization in 4th day
0.5882 0.6471 0.5882 0.7059 0.5882 0.6471 0.5882 0.5882
0.7059 0.5882 0.6471 0.6471 0.7059 0.6471 0.5882 0.5882
0.5882 0.7059 0.6471 0.6471 0.7059 0.5882 0.6471 0.5882 0.7059
Data after normalization in 5th day
0.8235 0.7647 0.9412 0.8235 0.7724 0.7647 0.8235 0.8824
1.0000 0.7647 0.9412 0.8824 0.7647 0.7647 0.8235 0.8824
0.9412 0.9412 1.0000 0.8824 0.8235 0.7647 0.7647 0.9412 0.7647
It is first principal component value because the odor characteristics of grass carp are most important, so paper carries out the process of feature extraction using principal component analytical method to the information of gas sensor here.
The data of 125 samples of all 5 days are carried out with SPSS softwares main into analysis totally, the data of logical 7 gas sensors of each sample are represented.Through analysis obtain first principal component accounts for total contribution rate 88.7%, it has been indicated that all 7 gas sensors certainly most information, therefore can only with first principal component value as odiferous information, odiferous information can be expressed as s=pc
, obtained by the computing of software,
=6.209,
=0.996,
=0.968,=0.979,=0.799,
=0.953,
=0.941. =0.942 will
、、、
、
、、
And
Substitute into formula（1）
=0.3997,
=0.3884,
=0.3929,
=0.3206,=0.3824,
=0.3776,The response of wherein 7 gas sensors of=0.3780. is used respectively、
、
、
、
、
、
Represent.Will=0.3997,
=0.3884,=0.3929,
=0.3206,
=0.3824,
=0.3776,
=0.3780,
、
、
、、
、
、Substitute into formula（2）It can obtain
Value, s=
, such odiferous information means that out.Then the process being normalized according to pH value, odiferous information is normalized.
3rd step, Fusion Features；
Using the characteristic information of the flesh of fish aspect of sample three obtained in said process as artificial neural network input, in BP network training process, with the odor characteristics of fish body（First principal component value）, chemical feature（PH value）And physical features（Electrical conductivity）As the input of neutral net, the fresh water fish sample of different freshness is as output sample, and with 100 00 desired outputs for representing first day, with 010 00 desired outputs for representing second day, the rest may be inferred for the desired output of the 3-5 days.This paper test specimens are the resting period different freshwater fish meats with different freshness, daily 25 samples, altogether 125 samples, wherein 100 samples are used as training set, 25 samples are used as test set.According to Kolmogorov theorems using a N
（2N+1）M 3 layers of BP networks, wherein N represents the sample number of input layer, is that 3, M represents that output layer passes through the sample number of neural metwork training to this experiment, is 5. to this experiment so the BP network structures of this experimental data are:Input layer has 3 neurons, and there are 7 neurons in intermediate layer, and output layer has 5 neurons.The transmission function selection S type tans of intermediate layer neuron, the transmission function selection S logarithmic functions of output layer neuron, Select Error target is 0.001, is set for the training of 5000 steps, and Studying factors are set to 0.01, and factor of momentum is set as 0.9.
Target output is set as：
T=[1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0; 0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0; 0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0; 0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 0 1 0; 0 0 1 0 0;0 0 1 0 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0; 0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1; 0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1];
The reality output of training set：
T=[1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0; 0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0; 0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0; 0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 1 0 0;0 0 0 1 0; 0 0 1 0 0;0 0 1 0 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0; 0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 1 0;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1; 0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1];
The reality output of test set：
[1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0;1 0 0 0 0; 0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0;0 1 0 0 0; 0 0 1 0 0;0 0 1 0 0;0 0 1 0 0; 0 0 1 0 0; 0 0 1 0 0; 0 0 0 1 0;0 0 0 1 0;0 0 0 1 0; 0 0 0 1 0;0 0 0 1 0; 0 0 0 0 1;0 0 0 0 1;0 0 0 0 1;0 0 0 0 1; 0 0 0 0 1;]
4th step, set up standard, expert system；
By multiple training result, a discrimination standard is obtained, specific standard is as follows：Five different stages are defined first：
The sample qualities of first day are fresh；
The sample qualities of second day are secondary fresh；
The sample qualities of the 3rd day are corrupt early stage；
The sample qualities of the 4th day are corrupt mid-term；
The sample qualities of the 5th day are the corrupt later stage.
Obtained specific standards are as follows：
| Quality grade | PH value | Electrical conductivity | Odiferous information |
| It is fresh | 6.4-6.7 | 517-539 | 3.0824-3.1451 |
| It is secondary fresh | 6.8-7.0 | 813-845 | 3.2968-3.3560 |
| Corrupt early stage | 7.1-7.3 | 1230-1253 | 6.4811-6.5385 |
| Corrupt mid-term | 7.4-7.6 | 1835-1895 | 6.5662-6.8714 |
| The corrupt later stage | 7.7-8.0 | 2570-2710 | 7.5537-7.8554 |
According to the discrimination standard set up, quality differentiation can be carried out：According to the discrimination standard of foundation, after agricultural product sample being put into test casing, the information collected according to the present invention just can easier determine the quality grade of agricultural product with reference to the expert system of foundation.
For example randomly choose the 5 groups of data collected to enter to differentiate, such as according to as follows：
3.0977 6.6 535
3.1009 6.6 526
3.2978 6.8 835
6.5057 7.3 1252
7.7965 7.7 2673
Contrasted by the standard with foundation, can determine whether that this five groups of data belong to fresh level, fresh level, secondary fresh level, corrupt mid-term, corrupt later stage successively.
The present invention is illustrated by specific implementation process, without departing from the present invention, various conversion and equivalent replacement can also be carried out to patent of the present invention, therefore, patent of the present invention is not limited to disclosed specific implementation process, and should include whole embodiments for falling into scope of the patent claims of the present invention.
Claims (5)
1. a kind of quality of agricultural product recognition detection establishment of standard method, it is characterised in that：The foundation of quality of agricultural product discrimination standard is carried out according to below step：
The first step, data acquisition；
According to the Feature Selection feature sensor of agricultural product to be measured itself, using feature sensor acquisition characteristics information, the information collected is shown in LabVIEW visualization interfaces；
Second step, data prediction；
The information collected is carried out to wild point rejecting processing first and then by SPSS softwares, utilize principal component analysis feature extraction formula
3rd step, Fusion Features；
Normalized characteristic is subjected to Feature-level fusion using BP artificial neural networks, by Matlab softwares, fusion results are obtained；
4th step, set up standard, expert system；
Repeatedly same agricultural product characteristic information is sampled, and try to achieve the result of each Feature-level fusion, set up quality discrimination standard according to the fusion results of multiple repairing weld, expert system is set up according to the standard of foundation.
2. a kind of quality of agricultural product recognition detection establishment of standard method according to claim 1, it is characterised in that：The feature sensor of selection is smell sensors array or other feature sensors, and the feature sensor of selection can be one or more of.
3. a kind of quality of agricultural product recognition detection establishment of standard method according to claim 1, it is characterised in that：The wild point elimination method of use is specially that the information that each sensor is collected is averaged first, then being compared each value collected and average, the larger information of those deviation averages is rejected, the information to each sensor carries out wild point rejecting processing.
4. a kind of quality of agricultural product recognition detection establishment of standard method according to claim 1, it is characterised in that：BP artificial neural networks are according to Kolmogorov theorems using a N(2N+1)
M 3 layers of BP networks, wherein N represents the sample number of input layer, is determined according to the number of features of selection；M represents that output layer passes through the sample number of neural metwork training, is determined according to the grade of agricultural product, and the information detected point training set and test set are substituted into network simulation is trained, and by training, just quality grade can be made a distinction by output result.
5. a kind of quality of agricultural product recognition detection establishment of standard method according to claim 4, it is characterised in that：The transmission function selection S type tans of intermediate layer neuron, the transmission function selection S logarithmic functions of output layer neuron, Select Error target is 0.001, is set for the training of 5000 steps, and Studying factors are set to 0.01, and factor of momentum is set as 0.9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101090294A CN102654494A (en) | 2012-04-16 | 2012-04-16 | Method for establishing quality identification and detection standard for agricultural products |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012101090294A CN102654494A (en) | 2012-04-16 | 2012-04-16 | Method for establishing quality identification and detection standard for agricultural products |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102654494A true CN102654494A (en) | 2012-09-05 |
Family
ID=46730150
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012101090294A Pending CN102654494A (en) | 2012-04-16 | 2012-04-16 | Method for establishing quality identification and detection standard for agricultural products |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102654494A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194587A (en) * | 2017-05-24 | 2017-09-22 | 暨南大学 | It is a kind of to be circulated based on block chain and the art work of expert system and identification register method |
| CN107767221A (en) * | 2017-10-26 | 2018-03-06 | 口碑(上海)信息技术有限公司 | Item Information exchange method and device based on smell |
| CN107884526A (en) * | 2017-11-07 | 2018-04-06 | 雅安蒋氏蜜蜂园有限公司 | A kind of honey parameter detection method and system based on deep learning |
| CN108500498A (en) * | 2018-03-26 | 2018-09-07 | 华中科技大学 | A kind of appearance of weld quality monitoring method |
| CN110517160A (en) * | 2019-08-02 | 2019-11-29 | 重庆邮电大学 | A kind of quality grading method and quality grading system of agricultural product |
| CN110850028A (en) * | 2019-10-09 | 2020-02-28 | 南京所云人工智能科技有限公司 | Fruit quality detection method based on machine learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101008621A (en) * | 2007-01-12 | 2007-08-01 | 浙江大学 | Method and device for detecting fruit defects based on multi-sensor information fusion |
| CN201421447Y (en) * | 2008-09-02 | 2010-03-10 | 昆明理工大学 | Identification device of tobacco scent |
| CN102084794A (en) * | 2010-10-22 | 2011-06-08 | 华南农业大学 | Method and device for early detecting crop pests based on multisensor information fusion |
-
2012
- 2012-04-16 CN CN2012101090294A patent/CN102654494A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101008621A (en) * | 2007-01-12 | 2007-08-01 | 浙江大学 | Method and device for detecting fruit defects based on multi-sensor information fusion |
| CN201421447Y (en) * | 2008-09-02 | 2010-03-10 | 昆明理工大学 | Identification device of tobacco scent |
| CN102084794A (en) * | 2010-10-22 | 2011-06-08 | 华南农业大学 | Method and device for early detecting crop pests based on multisensor information fusion |
Non-Patent Citations (3)
| Title |
|---|
| V. STEINMETZ,ET AL: "A Methodology for Sensor Fusion Design: Application to Fruit Quality Assessment", 《JOURNAL OF AGRICULTURE ENGNEERING RESEARCH》 * |
| 吴新垣: "信息融合及其智能化途径探讨", 《舰船指挥控制系统》 * |
| 王新宇: "基于计算机视觉和电子舌技术的绿茶分类分级研究", 《中国优秀硕士学位论文全文数据库（电子期刊）工程科技Ⅰ辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194587A (en) * | 2017-05-24 | 2017-09-22 | 暨南大学 | It is a kind of to be circulated based on block chain and the art work of expert system and identification register method |
| CN107767221A (en) * | 2017-10-26 | 2018-03-06 | 口碑(上海)信息技术有限公司 | Item Information exchange method and device based on smell |
| CN107884526A (en) * | 2017-11-07 | 2018-04-06 | 雅安蒋氏蜜蜂园有限公司 | A kind of honey parameter detection method and system based on deep learning |
| CN108500498A (en) * | 2018-03-26 | 2018-09-07 | 华中科技大学 | A kind of appearance of weld quality monitoring method |
| CN110517160A (en) * | 2019-08-02 | 2019-11-29 | 重庆邮电大学 | A kind of quality grading method and quality grading system of agricultural product |
| CN110850028A (en) * | 2019-10-09 | 2020-02-28 | 南京所云人工智能科技有限公司 | Fruit quality detection method based on machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102654494A (en) | Method for establishing quality identification and detection standard for agricultural products | |
| CN106769052B (en) | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on clustering | |
| Tsai et al. | A data-mining framework for exploring the multi-relation between fish species and water quality through self-organizing map | |
| CN108875913B (en) | Tricholoma matsutake rapid nondestructive testing system and method based on convolutional neural network | |
| CN103278609B (en) | A kind of Examination of Meat Fresh Level merged based on multi-source perception information | |
| CN107977671A (en) | A kind of tongue picture sorting technique based on multitask convolutional neural networks | |
| CN106568907B (en) | A kind of steamed crab freshness lossless detection method based on semi-supervised identification projection | |
| CN110796284A (en) | Method and device for predicting pollution level of fine particulate matters and computer equipment | |
| CN109962982A (en) | A kind of river and lake water ecological environment monitoring system based on Internet of Things | |
| CN102998350B (en) | Method for distinguishing edible oil from swill-cooked dirty oil by electrochemical fingerprints | |
| CN113658640B (en) | Fresh water ecological system health evaluation method | |
| CN107808358A (en) | Image watermark automatic testing method | |
| CN102628794A (en) | Method for quickly measuring total quantity of livestock meat bacteria based on hyperspectral imaging technology | |
| CN110189383A (en) | Chinese medicine tongue color coating colour quantitative analysis method based on machine learning | |
| CN107045013A (en) | Beef taste quality automatic classification detector and its detection method | |
| WO2024021359A1 (en) | Built environment dominant color measurement method and system based on image eeg sensitivity data | |
| CN116611000B (en) | Intelligent hairy crab culture environment monitoring system based on machine learning | |
| Chen et al. | Effects of fixed-count size on macroinvertebrate richness, site separation, and bioassessment of Chinese monsoonal streams | |
| Barman et al. | Prediction of soil pH using smartphone based digital image processing and prediction algorithm | |
| CN117456463A (en) | Disease and pest identification system based on big data and deep learning | |
| CN103868821B (en) | Adopt the evaluation method based on hyperacoustic Portable fishing meat freshness detection device | |
| CN110726813B (en) | Electronic nose prediction method based on double-layer integrated neural network | |
| CN116466408B (en) | Artificial neural network superbedrock identification method based on aeromagnetic data | |
| CN106530175A (en) | Urban ecosystem health diagnosis system | |
| CN110197120A (en) | Personnel target identification method for unattended sensor system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120905 |