CN111141836B - Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information - Google Patents
Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information Download PDFInfo
- Publication number
- CN111141836B CN111141836B CN202010033329.3A CN202010033329A CN111141836B CN 111141836 B CN111141836 B CN 111141836B CN 202010033329 A CN202010033329 A CN 202010033329A CN 111141836 B CN111141836 B CN 111141836B
- Authority
- CN
- China
- Prior art keywords
- spectrum
- pear
- sound vibration
- domain
- near infrared
- 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.)
- Active
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 88
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 88
- 238000001228 spectrum Methods 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000001514 detection method Methods 0.000 title claims abstract description 42
- 230000004927 fusion Effects 0.000 title claims abstract description 29
- 238000002329 infrared spectrum Methods 0.000 title claims abstract description 24
- 235000014443 Pyrus communis Nutrition 0.000 claims abstract description 81
- 238000009659 non-destructive testing Methods 0.000 claims abstract description 13
- 235000013399 edible fruits Nutrition 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000005516 engineering process Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 16
- 239000000523 sample Substances 0.000 claims description 16
- 238000004458 analytical method Methods 0.000 claims description 14
- 238000004433 infrared transmission spectrum Methods 0.000 claims description 14
- 238000000411 transmission spectrum Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 10
- 230000004044 response Effects 0.000 claims description 10
- 238000010801 machine learning Methods 0.000 claims description 9
- 238000003909 pattern recognition Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 7
- 239000000725 suspension Substances 0.000 claims description 6
- 241000228143 Penicillium Species 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 5
- 238000007637 random forest analysis Methods 0.000 claims description 5
- 238000012706 support-vector machine Methods 0.000 claims description 5
- 238000005520 cutting process Methods 0.000 claims description 4
- 238000007499 fusion processing Methods 0.000 claims description 4
- 238000001845 vibrational spectrum Methods 0.000 claims description 4
- 241001164374 Calyx Species 0.000 claims description 3
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 210000004556 brain Anatomy 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000011081 inoculation Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 239000013307 optical fiber Substances 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 claims description 3
- 238000002347 injection Methods 0.000 claims 1
- 239000007924 injection Substances 0.000 claims 1
- 238000012847 principal component analysis method Methods 0.000 claims 1
- 241000220324 Pyrus Species 0.000 abstract description 84
- 235000021017 pears Nutrition 0.000 abstract description 16
- 238000011161 development Methods 0.000 abstract description 5
- 230000001737 promoting effect Effects 0.000 abstract description 2
- 241000220225 Malus Species 0.000 description 9
- 238000010521 absorption reaction Methods 0.000 description 6
- 235000021016 apples Nutrition 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 244000179970 Monarda didyma Species 0.000 description 5
- 235000010672 Monarda didyma Nutrition 0.000 description 5
- 238000010276 construction Methods 0.000 description 5
- 238000012937 correction Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 208000019622 heart disease Diseases 0.000 description 5
- 230000003044 adaptive effect Effects 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 230000005284 excitation Effects 0.000 description 4
- 235000015203 fruit juice Nutrition 0.000 description 4
- 239000003053 toxin Substances 0.000 description 4
- 231100000765 toxin Toxicity 0.000 description 4
- 108700012359 toxins Proteins 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000003595 spectral effect Effects 0.000 description 3
- 239000011550 stock solution Substances 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000004497 NIR spectroscopy Methods 0.000 description 2
- 238000005481 NMR spectroscopy Methods 0.000 description 2
- 238000000149 argon plasma sintering Methods 0.000 description 2
- 230000002860 competitive effect Effects 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 239000001963 growth medium Substances 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000008223 sterile water Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 1
- 241001123663 Penicillium expansum Species 0.000 description 1
- 235000010724 Wisteria floribunda Nutrition 0.000 description 1
- 244000052616 bacterial pathogen Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004820 blood count Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000012258 culturing Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000007865 diluting Methods 0.000 description 1
- 238000010790 dilution Methods 0.000 description 1
- 239000012895 dilution Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 235000019990 fruit wine Nutrition 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 229910052739 hydrogen Inorganic materials 0.000 description 1
- 239000001257 hydrogen Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 235000010482 polyoxyethylene sorbitan monooleate Nutrition 0.000 description 1
- 229920000053 polysorbate 80 Polymers 0.000 description 1
- 235000021039 pomes Nutrition 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000007790 scraping Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/46—Processing the detected response signal, e.g. electronic circuits specially adapted therefor by spectral analysis, e.g. Fourier analysis or wavelet analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4481—Neural networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
Abstract
The invention discloses a nondestructive testing method and device for early internal diseases of pears based on fusion of sound vibration multi-domain spectrums and near infrared spectrum information. By the method and the device, the pear with the early internal diseases can be detected rapidly and accurately, the early detection of the internal diseases is realized as much as possible, and the method and the device have great significance in improving commodity rate and market competitiveness of the pear and promoting rapid and healthy development of pear industry.
Description
Technical Field
The invention belongs to the technical field of rapid nondestructive testing of agricultural product quality, and particularly relates to a nondestructive testing method and device for early internal diseases of pears based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information.
Background
China is the largest pear producing country in the world, xinjiang is one of the main pear producing areas, the yields of Xinjiang bergamot pears and apples in 2017 are 123.09 ten thousand tons and 144.17 ten thousand tons respectively, and the Xinjiang pear producing area is an important income-increasing and money-increasing fruit for Xinjiang fruit farmers. The apples and the bergamot pears have different degrees of internal diseases, the mildew heart rate of the red Fuji apples is 27.78%, and the maximum kurla bergamot pears with the black heart in the storage period can reach 30%. When the lesion volume of the pear is lower than 5%, the pulp is intact, the eating is not affected, and the pear still has commodity value, but due to the concealment of diseases in the pear, the pear cannot be found at all early, the pear is easy to be developed to rot and deteriorate, even the whole box and stack of healthy fruits are infected, and serious economic loss is caused for fruit manufacturers and fruit fresh-keeping enterprises. In addition, the disease fruit pulp is enriched with a large amount of toxins, and the toxins are mixed into deep processing links such as fruit juice, fruit wine, fruit can heads and the like, so that the toxins of the fruit juice exceed the standard (the European Union standard prescribes that the toxins Mmax of the fruit juice are less than or equal to 50 mug/Kg), which not only seriously endangers the health of human bodies, but also makes the export rate of the fruit juice of China extremely low. Therefore, the nondestructive detection method for the internal diseases of the pears is explored, the early detection of the internal diseases is realized as much as possible, and the method has great significance in improving commodity rate and market competitiveness of the pears and promoting the rapid and healthy development of the pear industry.
In recent years, researchers at home and abroad try to develop a great deal of basic research on the problem of nondestructive detection of internal diseases of pomes based on different detection technologies. The nondestructive detection method for the internal diseases of the pear mainly comprises an X-ray imaging method, a nuclear magnetic resonance technology, a near infrared spectrum analysis technology, an acoustic vibration technology and the like. From the technical means adopted, each method has a certain limitation, and no nondestructive testing method with high accuracy and high detection speed for early internal diseases of pears has been developed so far. The X-ray imaging method has high safety protection requirement, and is unfavorable for popularization and application. The nuclear magnetic resonance technology has the defects of excessively high equipment, low imaging speed and complex detection and analysis process. The near infrared spectroscopy is limited in improvement of early slight disease recognition accuracy due to interference of light scattering effect, spectrum peak overlapping, baseline drift and other factors. The acoustic vibration method is a traditional nondestructive testing method with relatively low technical equipment, the problem group Xu Hubo and the like (2017) are used for distinguishing and researching the black heart disease of the bergamot pears based on the resonance frequency of the acoustic vibration method, when the disease degree is relatively high, the resonance frequency difference between healthy fruits and disease fruits is large, the detection accuracy of the black heart disease can reach 96.7%, and the detection accuracy is higher than the identification accuracy of near infrared transmission spectrums of the (2016) such as the Shendereyetal (2010) and the Guo Zhiming on the internal diseases of the apple approximation degree. However, when the disease degree is low, the resonance frequency difference between healthy fruits and diseased fruits is small, and it is difficult to judge whether the resonance frequency change is caused by internal diseases, and the accuracy of judging the black heart disease is only 3%. Obviously, only extracting frequency domain characteristic parameters from the sound vibration response signals, failing to fully mine more abundant and comprehensive other potential information related to internal diseases of pears, is probably one of reasons for lower detection precision of the sound vibration method on the early internal disease pears. Since the multi-domain spectral characteristic parameters of the acoustic vibration response signals are sensitive to the physical information change of pear tissues, the near infrared spectrum is better than the acquisition of the chemical characteristic information change of hydrogen-containing groups of organic molecules of the pear tissues. If the sound vibration method is combined with the near infrared spectroscopy, the advantages of the two detection methods are fully utilized, the advantages and the advantages are compensated, the physical and chemical information fusion of the internal diseases of the pear is realized, the development degree of the internal diseases of the pear can be more comprehensively known, the detection precision of the early internal diseases of the pear can be greatly improved, the rapid nondestructive detection of the early internal diseases of the pear is a new research trend, and the method is a brand new thought and has no related patent literature.
Disclosure of Invention
The invention mainly solves the technical problems of providing a nondestructive testing method and device for early internal diseases of pears based on the fusion of sound vibration multi-domain spectrums and near infrared spectrum information, particularly adopting different information fusion technologies to respectively construct disease discrimination models fused with sound vibration multi-domain spectrums and near infrared transmission spectrum information, realizing accurate detection of early internal diseases of the pears, providing necessary scientific basis for the research and development of on-line detection technology of the internal diseases of the pears in future, and solving the problem of lower detection precision of the early internal diseases of the pears in the current grading and storage processes of the pears.
In order to solve the technical problems, the invention adopts the following technical scheme: the method comprises the following steps:
step one: the preparation method of the disease pear sample comprises the following steps: first, the concentration was 1.0X10 by using a microinjector on an ultra clean bench by wound inoculation 10 cfu/mL of the penicillium spore suspension is injected into the pear pit area along the end of the pear calyx, and the injected penicillium spore suspension is injectedThe pear samples are transferred into a constant temperature and humidity box, and internal disease development is carried out in an environment with the temperature of 25 ℃ and the relative humidity of 90%.
Step two: under the condition that the experimental device is stable and reliable, the time domain/frequency domain information and the spectrum information are acquired by utilizing the sound vibration detection device and the near infrared transmission spectrometer respectively.
Step three: after the multi-domain spectrum and the transmission spectrum of the sound vibration signal are collected, cutting the pear from the central line part, photographing the cut section by using a Canon-EOS 750D digital camera in the range of 30-100 nm, processing the image by using a MatlabR2018a digital image processing toolbox, and obtaining the disease degree index of the pear by adopting an iterative threshold segmentation method;
step four: and respectively utilizing a signal processing technology and a spectrum analysis technology to process the data of the obtained sound vibration multi-domain spectrum and spectrum information, and respectively extracting sound vibration response signal multi-domain spectrum characteristics and transmission spectrum characteristics which can effectively represent the pear with internal diseases so as to establish a sound vibration multi-domain spectrum/transmission spectrum information characteristic set.
Step five: based on the established sound vibration multi-domain spectrum/transmission spectrum information characteristic set, a prediction model for distinguishing the internal diseases of the pear is established on the characteristic layer by utilizing a machine learning algorithm.
Step six: based on the single-source sound vibration signal multi-domain spectrum and near infrared transmission spectrum information feature subsets, a pear internal disease series discriminant analysis model is built by using a machine learning algorithm, probability outputs of the built independent series prediction models are fused by using a D-S evidence theory, and a disease pear discriminant model fused on a decision level is built.
The concentration in the first step is 1.0x10 10 The preparation method of the cfu/mL penicillium spore suspension comprises the following steps: randomly cutting pulp pieces 5mm from the rotting position of pear with a sterile knife on an SW-CF-1F ultra-clean workbench, mixing with appropriate amount of sterile water to obtain stock solution, and sequentially diluting the stock solution to 10 -1 ~10 -6 100. Mu.L of stock solution with each dilution is coated on PDA culture medium, and cultured in an intelligent constant temperature incubator at 25 ℃ for 7d. Selection of different colonies Using plate streaking in New PDARepeatedly inoculating on the culture medium until pure strain is obtained. The purified strain was then inoculated onto the inclined surface of PDA test tube, incubated at 25℃for 7 days, and then stored at low temperature. Then inoculating Penicillium expansum onto fresh PDA slant, culturing at 25deg.C for 14d, adding sterile water containing 0.05% Tween80, scraping spores on colony surface with inoculating loop, filtering with 4 layers of gauze to remove impurities such as mycelium, observing with a blood cell counting plate under DM2000 type optical microscope, and adjusting spore suspension concentration to 1.0X10 10 cfu/mL for use.
The method for extracting the multi-domain spectrum characteristic parameters of the sound vibration signal in the fourth step comprises the following steps: extracting common time-frequency domain statistical characteristic parameters such as mean value, root mean square value, variance, peak value, waveform index, pulse index, margin index, average frequency, center frequency, root mean square frequency and the like from time-frequency domain curves by adopting a time-domain/frequency domain analysis method; for a time-frequency domain distribution image obtained by time-frequency analysis processing of short-time Fourier transform (STFT), continuous Wavelet Transform (CWT), S Transform (ST) and Adaptive Optimal Kernel (AOK), characteristics such as time-frequency domain energy distribution, information entropy, image texture and the like are extracted by wavelet packet decomposition, integrated empirical mode analysis, gray level symbiotic matrix method and local binary mode method. In order to further select the characteristic parameters which are more sensitive to early internal diseases of the pear and eliminate irrelevant and redundant information, a popularity learning Method (ML), a distance evaluation method (DET) or a maximum correlation minimum redundancy method (mRMR) is adopted to perform characteristic evaluation on the basis of the characteristic extraction so as to obtain a multi-domain spectral characteristic subset of the acoustic vibration response signals which has moderate dimension and is fully necessary for subsequent internal disease information fusion detection.
In the fourth step, in order to eliminate the influence of high-frequency random noise, baseline drift, light scattering, difference of component content information among samples and the like, and improve the convergence performance of a model, the original spectrum data is required to be preprocessed, the scheme tries to process the spectrum data by adopting vector normalization (Norm), standard normal variable transformation (SNV), savitzky Golay convolution smoothing (SG-2-Der), multiple Scattering Correction (MSC) and other common preprocessing methods on all collected original spectrum variables of the samples, and then the method adopts a variable extraction method such as a joint interval partial least square method (is-PLS), a Genetic Algorithm (GA), a continuous projection method (SPA), a Wavelet Transformation (WT), a competitive adaptive weighting algorithm (CARS), and the like, and finally discloses the difference of transmission spectrum information of pear with different morbidity degrees based on the absorption frequency, the absorption intensity, the peak area, the absorption intensity ratio of different wavelengths and the like, and finally the optimized variable sensitive to the internal disease pear is selected to form a transmission near infrared characteristic subset.
In the fifth step, the construction thought of the pear internal disease prediction model based on the feature layer multisource information fusion is as follows: (1) aiming at the established characteristic set capable of effectively representing the internal diseases of the pear, a predictive model for judging the internal diseases of the pear is respectively constructed on a characteristic layer by adopting a BP neural network (BP-ANN), a Support Vector Machine (SVM), an Extreme Learning Machine (ELM), a Random Forest (RF) and other traditional machine learning algorithms; (2) the method utilizes the advantages of deep learning on self-adaptive extraction and self-learning of internal disease features of the pear, adopts deep learning algorithms such as a Deep Belief Network (DBN), a stacked automatic encoder (SDAE), a Deep Convolutional Neural Network (DCNN) and the like, takes an original multidimensional high-domain pear internal disease information data set as input, and respectively builds a detection model for distinguishing the pear internal disease at a feature layer by adjusting parameters such as weight, hidden layer node number, learning rate, noise factor and the like.
In the sixth step, the construction thought of the pear internal disease prediction model based on decision layer multisource information fusion is as follows: two construction ideas are adopted: (1) and (3) attempting to fuse probability outputs of mutually independent series prediction models constructed by respectively extracting characteristic information by using a D-S evidence theory and a near infrared transmission spectrum method, establishing a primary discrimination result obtained by fusing different information characteristic layers by using a disease pear discrimination model (2) fused on a decision level, and fusing information of different information multi-decision models by using the D-S evidence theory to obtain a disease pear discrimination result fused with multi-source information and the multi-source decision model.
And D, analyzing and comparing the predictive performance of the disease pear by the discrimination models constructed in different layers by the sample division verification set and the correction set on each model constructed in the step five and the step six, further adopting an ROC curve to carry out comparative analysis on the accuracy of discriminating the disease pear of each model, and finally determining an optimal model for qualitatively and quantitatively analyzing the existence of the disease in the pear and classifying the disease degree by a multi-source information fusion technology, thereby realizing the accurate detection of the early-stage internal disease of the pear.
The computer performs fusion and mode processing on the extracted characteristic signals, wherein the data fusion adopts a traditional machine learning algorithm and a modern deep learning algorithm to construct a high-precision real-time mode classification system to process sound vibration and transmission spectrum data, and the high-precision real-time mode classification system is connected with a database to perform learning and training to obtain a knowledge base, so that the developed system can judge whether diseases of a tested sample exist or not.
The pear early internal disease nondestructive testing device based on the fusion of the sound vibration multi-domain spectrum and the near infrared spectrum information comprises a sound vibration testing system, a near infrared spectrum system and a pattern recognition and data fusion processing system; the sound vibration detection system comprises a signal generator, a voltage amplifier, a piezoelectric beam acceleration sensor, a VibPilot vibration control and dynamic signal acquisition analyzer and SOAnalyzer4.1 software matched with the VibPilot vibration control and dynamic signal acquisition analyzer; the near infrared transmission spectrum system comprises a spectrum acquisition instrument, an optical fiber probe, a detection table, a light source and a shading camera bellows; the spectrum acquisition instrument and the matched OceanView spectrum analysis software are respectively connected with a computer; the computer internal signal fusion and pattern recognition system simulates human brain to fuse and pattern recognition processing the near infrared spectrum data and the sound vibration multi-domain spectrum data, and finally the computer determines whether the diseases of the pear to be detected exist or not.
The beneficial effects of the invention are as follows:
according to the sound vibration characteristics and optical characteristics of the pear with internal diseases, the sound vibration analysis technology and the spectrum analysis technology are organically combined, the multi-source information fusion processing method is adopted, sound vibration multi-domain spectrum information and near infrared transmission spectrum information are fused, a high-precision real-time mode classification system is utilized to process spectrum and sound vibration data, and the spectrum and sound vibration data are compared and judged with information in a database established through learning to carry out nondestructive testing on the existence of early internal diseases of the pear.
The method is used for grading and selecting pear quality and controlling the quality of the processing process, can timely and accurately sort out early-stage disease pear before apple warehouse entry, prevents early-stage disease pear germs from spreading and infecting healthy fruits in a large area, effectively reduces the incidence rate of fruit production, processing and storage periods, simultaneously assists and replaces professional detection personnel, liberates labor force, eliminates human subjective factors, improves the production efficiency, and can be popularized and applied to rapid nondestructive detection of diseases in other agricultural products.
The invention applies the deep learning algorithm to the construction of the early internal disease detection model of the pear by combining the multi-source information fusion technology, fully plays the advantages of the deep learning of self-extracting features from the original data, better characterizes the internal information with rich data, can analyze the sound vibration and spectrum data fusion in a short time and then output the result, meets the requirements of quick and real-time analysis of big data, realizes the intelligent grading treatment of the early internal disease of the pear, and provides necessary technical support for the construction of the modern intelligent forest and fruit industry system.
Drawings
FIG. 1 is a flow chart of a method of fusion of vibro-acoustic information and transmission spectrum information in accordance with the present invention.
Fig. 2 is a flowchart of a method for integrating vibro-acoustic information and transmission spectrum information according to the present invention (DBN is taken as an example).
FIG. 3 is a schematic diagram of a detection system apparatus of the present invention.
The components in the drawings are marked as follows: 1. a sound spectrum and spectrum test analysis and reporting software system; 2. vibration control and dynamic signal acquisition analyzer; 3. a voltage amplifier; 4. a near infrared spectrum acquisition instrument; 5. and (5) a test platform.
FIG. 4 is a schematic view of the structure of the test platform of the present invention.
The components in the drawings are marked as follows: 1. a horizontal slide rail; 2. a horizontal slider; 3. a sensor and a bracket; 4. a ball screw; 5. a vertical sliding table; 6. a vertical slide rail; 7. a sensor; 8. a fixing nut; 9. a shading dark box; 10. a light source; 11. apple sample; 12. a height-adjustable stage; 13. a collimator lens; 14. supporting table
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making clear and defining the scope of the present invention.
1-2, and example implementation apparatus referring to FIGS. 3-4, embodiments of the present invention include:
taking bergamot pear black heart as an example, nondestructive detection of other internal diseases of fruits can refer to the method of the embodiment, and a new knowledge base is established specifically aiming at the evaluation standard of the detected samples, so that nondestructive detection can be carried out on the products.
The pear early internal disease multisource information fusion detection system consists of an acoustic vibration detection system, a near infrared spectrum system and a pattern recognition and data fusion processing system; the sound vibration detection system comprises a sound spectrum and spectrum test analysis and report software system, a VibPilot vibration control and dynamic signal acquisition analyzer, a voltage amplifier and a piezoelectric beam acceleration sensor, and the near infrared transmission spectrum system comprises a spectrum acquisition instrument, an optical fiber probe, a detection table, a light source and a shading dark box; the spectrum acquisition instrument and the matched OceanView spectrum analysis software are respectively connected with a computer; the computer internal signal fusion and pattern recognition system simulates human brain to fuse and pattern recognition processing the near infrared spectrum data and the sound vibration multi-domain spectrum data, and finally the computer determines whether the diseases of the pear to be detected exist or not. The method comprises the following specific steps:
first, the concentration was 1.0X10 by using a microinjector on an ultra clean bench by wound inoculation 10 cfu/mL of penicillium spore suspension is injected into a pear pit area along the end of a pear calyx, the injected pear sample is transferred into a constant temperature and humidity box, internal disease development is carried out in an environment with the temperature of 25 ℃ and the relative humidity of 90%, and the backup measurement is carried out for 72 hours.
And then the apples to be tested are placed on a height-adjustable object stage, and are respectively and effectively contacted with 2Q 220-A4-303YB piezoelectric beam type sensors on two sides of the opposite equatorial part (1/3 of the middle of the cantilever beam). Half sine with peak voltage of 2.5VThe pulse signal is linearly amplified into an excitation signal VA having a peak voltage of 80V by a voltage amplifier of the HA-405 type. The signal is input to an excitation sensor in contact with the equatorial portion of the pear sample to vibrate the sample, and the cantilever Liang Suixiang pear is vibrated to deform due to the contact of the cantilever beam of the sensor with the pear, and the response signal V is generated by the vibration of the sample RE (equatorial response signal) and V RC The piezoelectric beam type sensor at the sensing end receives the excitation signal and the response signal, the excitation signal and the response signal are collected by a VibPilot vibration control and dynamic signal collecting analyzer in a mode that the sampling frequency is 51200Hz and the duration is 0.16s, and finally the analysis processing is carried out by a software system SOAnalyzer 4.1.
Meanwhile, in order to ensure the stability of the light source, the light source is turned on for preheating 15min before each spectrum data acquisition. As known from preliminary experiments, the light source is 130mm away from the surface of the object stage, the spectrasuite software sets the integration time to 100ms, the average scanning times to 10, the smoothness to 5, dark noise is removed, and the spectrum data acquired during nonlinear correction and stray light correction are optimal. When data are collected, the camera bellows door is kept in a closed state, a pear sample is placed on the height-adjustable objective table, and the axial direction of the fruit handle is perpendicular to the irradiation direction of the light source. And 3 points which are uniformly distributed and have no defects are selected at the equatorial plane of each sample to acquire spectrum information, 120 degrees are formed between the points, and finally, the average value of 3 times of data is taken as the spectrum data of the sample. And collecting the spectra of all the disease apples under the same condition, and immediately collecting near infrared transmission spectra after the collection of the sound vibration multi-domain spectrum information of the internal disease apples is completed.
After the acoustic vibration multi-domain spectrum and the near infrared spectrum data are acquired by the acoustic vibration detection system and the near infrared spectrum system, a time domain/frequency domain analysis method is adopted to extract common time domain and frequency domain statistical characteristic parameters such as average value, root mean square value, variance, peak-to-peak value, waveform index, pulse index, margin index, average frequency, center frequency, root mean square frequency and the like from the time domain and frequency domain curves; for a time-frequency domain distribution image obtained by time-frequency analysis processing of short-time Fourier transform (STFT), continuous Wavelet Transform (CWT), S Transform (ST) and Adaptive Optimal Kernel (AOK), characteristics such as time-frequency domain energy distribution, information entropy, image texture and the like are extracted by wavelet packet decomposition, integrated empirical mode analysis, gray level symbiotic matrix method and local binary mode method.
And performing feature evaluation by adopting a popularity learning Method (ML), a distance evaluation method (DET) or a maximum correlation minimum redundancy method (mRMR) to obtain a multi-domain spectral feature subset of the sound vibration response signal which has moderate dimension and is fully necessary for subsequent internal disease information fusion detection. The method comprises the steps of processing spectrum data by adopting a vector normalization (Norm), standard normal variable transformation (SNV), savitzkyGolay convolution smoothing (SG-2-Der), multiple Scattering Correction (MSC) and other common preprocessing methods, then adopting a variable extraction method such as a joint interval partial least square method (is-PLS), a Genetic Algorithm (GA), a continuous projection method (SPA), wavelet Transformation (WT), a competitive adaptive weighting algorithm (CARS) and the like, revealing the difference of transmission spectrum information of disease fruits with different disease degrees based on characteristic information such as absorption frequency, absorption intensity, peak area, different wave band integral areas, absorption intensity ratio of different wavelengths and the like, forming a near infrared transmission spectrum feature subset by optimizing and screening effective wavelength variables sensitive to the internal disease fruits, and finally constructing a pear early-stage black heart nondestructive detection database.
By utilizing the multi-domain acoustic vibration spectrum information and near infrared spectrum information multi-level fusion technology, a BP neural network (BP-ANN), a Support Vector Machine (SVM), an Extreme Learning Machine (ELM), a Random Forest (RF) and other traditional machine learning algorithms, a Deep Belief Network (DBN), a stacked automatic encoder (SDAE), a Deep Convolutional Neural Network (DCNN) and other deep learning algorithms and other methods are adopted to carry out acoustic vibration and spectrum information fusion, a high-precision real-time mode classification system is constructed to process acoustic vibration and transmission spectrum data and is connected with a database for learning and training, a knowledge base is obtained, a black heart disease pear discrimination model fused on a feature layer/decision layer is established, and finally the discrimination of the existence of early black heart disease pear is realized.
The foregoing description is only illustrative of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes or direct or indirect application in other related technical fields are included in the scope of the present invention.
Claims (5)
1. The pear early internal disease nondestructive testing method based on the fusion of the sound vibration multi-domain spectrum and the near infrared spectrum information is characterized by comprising the following steps of:
step one: the preparation method of the disease pear sample comprises the following steps: first, the concentration was 1.0X10 by using a microinjector on an ultra clean bench by wound inoculation 10 Injecting cfu/mL penicillium spore suspension into a pear pit area along the end of a pear calyx, transferring the pear sample after injection into a constant temperature and humidity box, and developing internal diseases in an environment with the temperature of 25 ℃ and the relative humidity of 90%;
step two: under the condition that an experimental device is stable and reliable, synchronously acquiring time domain/frequency domain information and spectrum information by utilizing an acoustic vibration detection device and a near infrared transmission spectrometer respectively;
step three: after the multi-domain spectrum and the transmission spectrum of the sound vibration signal are collected, cutting pear from a central position, transversely cutting pear along an equatorial position, shooting pulp by using an intelligent scanner ET18, and calculating the percentage of the rotten pulp area in the image to the whole sectional area by using image processing software Photoshop;
step four: respectively utilizing a signal processing technology and a spectrum analysis technology to process the data of the obtained sound vibration multi-domain spectrum and spectrum information, and respectively extracting sound vibration response signal multi-domain spectrum characteristics and transmission spectrum characteristics which can effectively represent the pear with internal diseases so as to establish a sound vibration multi-domain spectrum/transmission spectrum information characteristic set;
step five: based on the established sound vibration multi-domain spectrum/transmission spectrum information characteristic set, a prediction model for distinguishing the internal diseases of the pear is established on the characteristic layer by utilizing a machine learning algorithm;
step six: based on the single-source sound vibration signal multi-domain spectrum and near infrared transmission spectrum information feature subsets, a pear internal disease series discriminant analysis model is built by using a machine learning algorithm, probability outputs of the built independent series prediction models are fused by using a D-S evidence theory, and a disease pear discriminant model fused on a decision level is built.
2. The nondestructive testing method for early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information of the invention is characterized by comprising the following steps: and (3) repeating the second step to acquire sound vibration multi-domain spectrum and near infrared transmission spectrum information of the healthy fruits according to the requirement of the quantity ratio of the diseased fruits to the healthy fruits of 1:1.
3. The nondestructive testing method for early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information of the invention is characterized by comprising the following steps: and step four, unifying the information dimensionalities of the multi-domain sound vibration spectrum and the near infrared transmission spectrum on the basis of completing the extraction of the multi-domain sound vibration signal spectrum characteristics and the near infrared transmission spectrum characteristic variables, carrying out normalization processing, and if information is repeatedly expressed in each characteristic parameter of the sound vibration multi-domain spectrum/near infrared spectrum, carrying out dimensionality reduction by adopting a principal component analysis method to ensure that the selected principal components are uncorrelated with each other, and the variance accumulated contribution rate reaches more than 80 percent, and finally constructing a multi-domain spectrum/spectrum information characteristic set capable of effectively representing internal diseases according to the number and the scoring condition of the selected principal components.
4. The nondestructive testing method for early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information of the invention is characterized by comprising the following steps: the machine learning algorithm in the fifth step comprises a BP neural network BP-ANN, a support vector machine SVM, an extreme learning machine ELM, a random forest RF traditional machine learning algorithm, a deep confidence network DBN, a stacked automatic encoder SDAE and a deep convolutional neural network DCNN deep learning algorithm.
5. A device for realizing the pear early internal disease nondestructive testing method based on the fusion of sound vibration multi-domain spectrum and near infrared spectrum information according to claim 1, which is characterized by comprising a sound vibration detection system, a near infrared spectrum system and a pattern recognition and data fusion processing system; the sound vibration detection system comprises a signal generator, a voltage amplifier, a piezoelectric beam acceleration sensor, a VibPilot vibration control and dynamic signal acquisition analyzer and SOAnalyzer4.1 software matched with the VibPilot vibration control and dynamic signal acquisition analyzer; the near infrared transmission spectrum system comprises a spectrum acquisition instrument, an optical fiber probe, a detection table, a light source and a shading camera bellows, wherein the spectrum acquisition instrument and the OceanView acquisition analyzer are respectively connected with a computer; the computer internal signal fusion and pattern recognition system simulates human brain to fuse and pattern recognition processing the near infrared spectrum data and the sound vibration multi-domain spectrum data, and finally the computer determines whether the diseases of the pear to be detected exist or not.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010033329.3A CN111141836B (en) | 2020-01-13 | 2020-01-13 | Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010033329.3A CN111141836B (en) | 2020-01-13 | 2020-01-13 | Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111141836A CN111141836A (en) | 2020-05-12 |
CN111141836B true CN111141836B (en) | 2023-07-18 |
Family
ID=70524757
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010033329.3A Active CN111141836B (en) | 2020-01-13 | 2020-01-13 | Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111141836B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111693488B (en) * | 2020-06-08 | 2022-12-06 | 济南大学 | Fruit grade classification method and system based on DS evidence theory fusion |
CN111855800B (en) * | 2020-07-17 | 2022-04-01 | 西南科技大学 | Method for rapidly and nondestructively measuring shelf life or optimal edible period of fruit by acoustic vibration |
CN113295674B (en) * | 2021-04-29 | 2022-10-14 | 中国科学院沈阳自动化研究所 | Laser-induced breakdown spectroscopy characteristic nonlinear processing method based on S transformation |
CN113484738A (en) * | 2021-05-25 | 2021-10-08 | 北京航空航天大学 | Circuit fault diagnosis method based on multi-feature information fusion |
CN114092710A (en) * | 2021-11-23 | 2022-02-25 | 沈阳农业大学 | Construction method of DCNN leaf blast classification model based on fusion characteristics |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101228513B1 (en) * | 2012-11-29 | 2013-01-31 | 케이엔디이주식회사 | A system of non-destructive inspection |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5402782A (en) * | 1993-03-05 | 1995-04-04 | University Of Kentucky Research Foundation | Apparatus and method for noninvasive chemical analysis |
AU2002314969A1 (en) * | 2001-06-06 | 2002-12-16 | University Of Kentucky Research Foundation | Multimode structure for current measurement, ultrasonic and near-ir spectrum collection and processing |
CN103063585B (en) * | 2013-01-05 | 2015-09-02 | 石河子大学 | Melon and fruit degree of ripeness Rapid non-destructive testing device and detection system method for building up |
JP5855693B2 (en) * | 2014-02-28 | 2016-02-09 | 富士重工業株式会社 | Vibration detection apparatus and vibration detection method |
US20170356842A1 (en) * | 2016-06-09 | 2017-12-14 | Adelphi University | Real-time trace gas sensor using a multi-mode diode laser and multiple line integrated cavity enhanced absorption spectroscopy |
CN109100323B (en) * | 2018-08-20 | 2020-11-20 | 江苏大学 | Nondestructive quantitative evaluation method for transmission spectrum of apple water core disease |
CN109932333B (en) * | 2019-04-04 | 2022-09-13 | 浙江大学 | Fruit firmness measuring system and method with acoustic vibration and near infrared spectrum fused |
CN110376159A (en) * | 2019-08-22 | 2019-10-25 | 华东交通大学 | Yali pear black heart method for quick identification based on near-infrared diffusing transmission spectrum |
-
2020
- 2020-01-13 CN CN202010033329.3A patent/CN111141836B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101228513B1 (en) * | 2012-11-29 | 2013-01-31 | 케이엔디이주식회사 | A system of non-destructive inspection |
Also Published As
Publication number | Publication date |
---|---|
CN111141836A (en) | 2020-05-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111141836B (en) | Method and device for nondestructive detection of early internal diseases of pome based on fusion of sound vibration multi-domain spectrum and near infrared spectrum information | |
CN108663339B (en) | On-line detection method for mildewed corn based on spectrum and image information fusion | |
JP6339244B2 (en) | Method for predicting sugar content and acidity of fruit using multivariate statistical analysis of FT-IR spectrum data | |
JPH06505815A (en) | Neural network system using spectral analysis to identify complex organic structures | |
Qiao et al. | Analysis and detection of decayed blueberry by low field nuclear magnetic resonance and imaging | |
CN105158186A (en) | Method for detecting black core of white radish based on hyperspectral image | |
CN101158651A (en) | Device and method for detecting apples inside and outside quality based on laser image analyse | |
CN108801971A (en) | Mould based on high light spectrum image-forming technology infects the detection method of Chinese chestnut | |
CN111157511A (en) | Egg freshness nondestructive testing method based on Raman spectrum technology | |
Zhaoyong et al. | Detection of moldy core in apples and its symptom types using transmittance spectroscopy | |
Yin et al. | Non-destructive detection of foreign contaminants in toast bread with near infrared spectroscopy and computer vision techniques | |
CN104297136A (en) | Hyperspectral image-based method for forecasting growth of pseudomonas aeruginosa | |
CN104749249B (en) | A kind of method of the detection physical purity of seed based on ultrasonic technology | |
CN105911000A (en) | Characteristic wave band based blood spot egg on-line detection method | |
Fu et al. | Discrimination of pear varieties using three classification methods based on near-infrared spectroscopy | |
CN105651727B (en) | The method that near-infrared spectrum analysis based on JADE and ELM differentiates apple shelf life | |
Chu et al. | A Vis/NIR spectra-based approach for identifying bananas infected with Colletotrichum musae | |
CN112086137B (en) | Method for quantitatively analyzing sorbose content in fermentation liquor | |
Su et al. | Identification of damage in pear using hyperspectral imaging technology | |
CN112067577A (en) | Method, device and equipment for identifying overproof cream pigment based on support vector machine | |
CN112801172A (en) | Chinese cabbage pesticide residue qualitative analysis method based on fuzzy pattern recognition | |
CN113418889B (en) | Real-time detection method for water content and total colony count of dried vegetables based on deep learning | |
Hu et al. | Detection of moldy cores in apples with near-infrared transmission spectroscopy based on wavelet and BP network | |
CN100494982C (en) | Non-destructive detection method for quickly detecting inner quality of pear | |
CN113791049B (en) | Method for rapidly detecting freshness of chilled duck meat by combining NIRS and CV |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |