WO2019041055A1 - Method for estimating the oil of individual olives using non-destructive technologies - Google Patents
Method for estimating the oil of individual olives using non-destructive technologies Download PDFInfo
- Publication number
- WO2019041055A1 WO2019041055A1 PCT/CL2018/050071 CL2018050071W WO2019041055A1 WO 2019041055 A1 WO2019041055 A1 WO 2019041055A1 CL 2018050071 W CL2018050071 W CL 2018050071W WO 2019041055 A1 WO2019041055 A1 WO 2019041055A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- olives
- oil
- nir
- color
- sample
- 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.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
- G01J3/02—Details
-
- 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
-
- 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
-
- 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
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/02—Food
- G01N33/03—Edible oils or edible fats
Definitions
- the present invention relates to a method of estimating oil of individual olives based on non-destructive technologies, which can be used in the agricultural industry.
- the reference method for the estimation of oil is the so-called SOHXLET method.
- This method delivers the oil content from a dry sample that comes from the processing of a group of olives (minimum 10 olives).
- the problem that exists with the current method is that the method destroys the samples of olives and that it is not possible to determine with precision the contribution of oil from each one of the olives.
- the SOXHLET method is a chemical method that is very accurate, but generates polluting waste, requires laboratory infrastructure, and requires at least 6 hours according to the standard to obtain the oil content.
- NIR near infrared reflectance
- WO2016141451 A1 discloses a new spectroscopic method of near-infrared Fourier fast transform (FT-NIR) to detect the authenticity of extra virgin olive oils (EVOO) and to determine the kind and quantity of an adulterant in EVOO.
- FT-NIR near-infrared Fourier fast transform
- the document "The Detection and Quantification of Adulteration in Olive Oil by Near-Infrared Spectroscopy and Chemometrics” (Christy et al) describes a new procedure for the classification and quantification of the adulteration of pure olive oil by soybean oil, flower oil of sun, corn oil, walnut oil and hazelnut oil. The procedure is based on a chemometric analysis of the near infrared (NIR) spectra of olive oil mixtures, which contained different adulterant.
- NIR near infrared
- the present invention corresponds to a method for estimating the oil content in individual olives based on non-destructive technologies.
- groups of olives with similar characteristics are formed considering the color and NIR wavelengths of light absorption of the acid grades. Since the olives of the groups are similar, the average of the characteristics is representative of the set of olives (low standard deviation), and therefore it is appropriate to associate said average characteristics with the oil content.
- the resulting model allows to estimate with an acceptable error by the oil production company the content of olive oil based on NIR and color characteristics which are measured without the need to destroy the olives.
- the method allows to estimate the oil of individual olives without destroying them, which avoids the laboratory work that involves destroying the sample, and can be associated by a model the characteristics of the olives measures (Color and NIR) with the obtained oil content as a reference through SOXHLET.
- Figure 1 represents a near infrared spectrum NIR with a spectrometer in reflectance mode, according to one embodiment of the invention.
- Figure 2 represents obtaining similar bags of olives by Kmeans ++, according to one embodiment of the invention.
- Figure 3 represents equipment for extracting oil in parallel with six tanks for the dry samples, according to one embodiment of the invention.
- Figure 4 represents equipment for extracting oil in parallel with six tanks for the dry samples, according to one embodiment of the invention.
- the present invention corresponds to a method for estimating the oil content of individual olives based on NIR and Color characteristics.
- Said method estimating the oil content comprises the following stages: ) Obtain characteristics NIR and Color of Olives: To a set of olives that come from the field of the company producing of oil obtains characteristics him of color and NIR.
- the olives of the bags from the field are placed in a tray, preferably a transparent plastic tray.
- a tray preferably a transparent plastic tray.
- the segmentation of each of the olives is carried out by means of image processing techniques, obtaining as a result the individualization of each of the olives.
- the individualized olives are obtained the average color in a RGB color model and its transformation to the model d c2c3.
- the c1 c2c3 model is used as a color characteristic since it corresponds to a color model that is invariant to the illumination, then it makes it possible to ensure the robustness of the color characteristic against possible variations in lighting.
- the phenomenon studied is the absorption of light by fatty acids, and art indicates that the major fatty acids have a light absorption centered at 1728 nm. Therefore, the chosen range varies between 1710 nm and 1735 nm, corresponding to 1 1 points of the total points of the spectra obtained. The previous one allows to justify in a pertinent way the use of a group of wavelengths and not the whole spectrum.
- the olive is represented with 3 points of the color model d c2c3 plus the 1 1 points of the spectrum where the absorption of the major fatty acids takes place.
- the descriptor VIS-NIR of the Olive has 14 points.
- the olive boxes are frozen at a temperature of -20 ° C to stop the evolution of olive degradation.
- the state of the art indicates that the freezing of olives does not alter the oil content (but not its color characteristics and quality parameters). In this sense it is possible to keep the frozen olive trays awaiting the identification stage of similar olives.
- Group olives based on color characteristics and NIR This grouping allows to properly associate characteristics of an olive to the oil content delivered by the SOHXLET to obtain oil from a group of very similar olives. This allows that the olives considered to obtain the dry sample have a low standard deviation of their characteristics.
- the VIS-NIR descriptor of the olive corresponds to a vector of 14 elements constituted by the 3 channels of the model c1 c2c3 and the 1 1 points of the NIR spectrum selected for main fatty acids. It is noted that each descriptor was normalized in the scale to give equal importance to both types of data, considering a normalization with respect to the mean of each descriptor element. They are grouped of similar olives by grouping the standard VIS-NIR descriptor. The grouping procedure is based on the kmeans ++ algorithm, which groups descriptors according to similarity and needs as a parameter the number of groups.
- a grouping algorithm was developed to find the maximum number of possible groups, each group having a certain minimum size.
- the SOXHLET oil extraction standard requires a minimum mass of approximately 15 olives per extraction of an oil value, in addition to carrying out measurements in duplicate as a safety measure (sample and counter-sample), this means that each bag has to Have a minimum number of 30 olives for sample and counter-sample verification. From the boxes where the olives are stored, the corresponding olives indicated by the grouping are removed and stored in bags as shown in Figure 2. The bags are refrozen while awaiting the SOXHLET oil extraction stage.
- Extract oil through SOXHLET Each of the bags frozen in the previous stage, has at least 30 olives, it is possible to make at least 2 measurements of oil per bag, one for sample and another for counter-sample.
- the countersample has a sense of verification of the value of obtained oil, since to be of similar olives, sample and counter-sample must have very close oil values.
- the SOXHLET oil extraction method is standardized by the International Olive Council, and consists of extracting the fats from a dry sample using an apolar solvent (hexane), by continuously circulating the hexane through the sample for a minimum of 6 hours.
- the number of hours of use of extractor is variable according to the number of dry samples, but to make a Robust model will require hundreds of hours in extracting the ground from the truth of oil content. It is precisely this characteristic of the process that is improved with the developed method. While it is true it is necessary to have the oil references to build the model, subsequent estimates do not require oil extraction, and the times involved are those that take the measurements of the color and the NIR spectrum. To reduce the extraction time, equipment that has the capacity to extract oil in parallel is used, as is the case of Figure 3, which has 6 tanks for the dry samples.
- the preparation of the sample consists of taking a group of olives, and grinding them with a pulverizing mill, and weighing the sample in fresh weight. Then the sample is taken to an oven to remove the water and leave only the dry matter which is also weighed. Dry matter is extracted between 2 and 5 grams (according to the norm of the SOXHLET method) for the extraction of fat.
- the dry sample is entered into a cellulose extraction thimble which is immersed in hexane, and the solvent is recirculated through the sample for 6 hours to extract oil. At the end of the process it is separated from the sample, in a glass with hexane. This glass is taken to a stove to evaporate the hexane and leaving only the oil.
- the positive characteristics of the SVR are that it is not necessary to assume a polynomial model, that its computation algorithm is very well defined mathematically and ensures to find an optimal model, and that it has few hyper-parameters to be defined (2 in the case of the SVR).
- the experimental validation method corresponds to that mentioned in Stage 4.
- N groups of similar olives which were grouped by their 14 characteristics (3 Color + 1 1 NIR) oil was extracted.
- N estimation models were made, leaving one outside, and the average of the N models is considered as a result of the statistical test.
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- General Physics & Mathematics (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Engineering & Computer Science (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
La presente invención se relaciona con un método de estimación de aceite de olivas individuales en base a tecnologías no destructivas, el cual puede ser utilizado en la industria. El método se lleva a cabo en base a características NIR y Color, comprendiendo las siguientes etapas: a. Seleccionar un conjunto de olivas que provienen del campo; b. Obtener características NIR y Color de Olivas; c. Agrupar olivas en base a características de color y NIR; d. Extraer aceite mediante un método de extracción; y e. Desarrollar un modelo de estimación.The present invention relates to a method of estimating oil of individual olives based on non-destructive technologies, which can be used in the industry. The method is carried out based on NIR and Color characteristics, comprising the following stages: a. Select a set of olives that come from the field; b. Get NIR and Olive Color features; c. Group olives based on color and NIR characteristics; d. Extract oil by an extraction method; and e. Develop an estimation model.
Description
MÉTODO DE ESTIMACIÓN DE ACEITE DE OLIVAS INDIVIDUALES EN BASE A METHOD OF ESTIMATING OIL OF INDIVIDUAL OLIVES BASED ON
TECNOLOGÍAS NO DESTRUCTIVAS. NON-DESTRUCTIVE TECHNOLOGIES.
CAMPO DE APLICACIÓN La presente invención se relaciona con un método de estimación de aceite de olivas individuales en base a tecnologías no destructivas, el cual puede ser utilizado en la industria agrícola. FIELD OF APPLICATION The present invention relates to a method of estimating oil of individual olives based on non-destructive technologies, which can be used in the agricultural industry.
ANTECEDENTES BACKGROUND
En la actualidad, el método de referencia para la estimación de aceite, recomendado por el Consejo Olivícola Internacional, es el denominado método SOHXLET. Dicho método entrega el contenido de aceite desde una muestra seca que proviene del procesamiento de un grupo de olivas (mínimo 10 olivas). El problema que existe con el método actual es que el método destruye las muestras de olivas y que no se puede determinar con precisión el aporte de aceite de cada una de las olivas. El método SOXHLET es un método químico que es muy exacto, pero que genera desechos contaminantes, se requiere infraestructura de laboratorio, y requiere al menos 6 horas según norma para obtener el contenido de aceite. Currently, the reference method for the estimation of oil, recommended by the International Olive Council, is the so-called SOHXLET method. This method delivers the oil content from a dry sample that comes from the processing of a group of olives (minimum 10 olives). The problem that exists with the current method is that the method destroys the samples of olives and that it is not possible to determine with precision the contribution of oil from each one of the olives. The SOXHLET method is a chemical method that is very accurate, but generates polluting waste, requires laboratory infrastructure, and requires at least 6 hours according to the standard to obtain the oil content.
Actualmente, algunas empresas olivícolas están utilizando tecnología NIR (near infrared reflectance, en ingles) destructiva para la estimación de aceite. Consiste en un sensor NIR (de diferentes fabricantes) que toma el espectro de una masa de olivas molidas. Esta propuesta aunque utiliza una tecnología no destructiva, el proceso es destructivo, pues muele las olivas. Además mantiene la falta de representatividad fundamental del método SOXHLET de la estimación de aceite, pues el aceite extraído corresponde a un conjunto de olivas. Currently, some olive companies are using destructive NIR (near infrared reflectance) technology for oil estimation. It consists of a NIR sensor (from different manufacturers) that takes the spectrum from a mass of ground olives. Although this proposal uses a non-destructive technology, the process is destructive, as it grinds the olives. It also maintains the lack of representativeness fundamental of the SOXHLET method of oil estimation, since the extracted oil corresponds to a set of olives.
El documento WO2016141451 A1 describe un nuevo método espectroscópico de transformada rápida de Fourier cercano al infrarrojo (FT-NIR) para detectar la autenticidad de los aceites de oliva virgen extra (EVOO) y determinar la clase y cantidad de un adulterante en EVOO. WO2016141451 A1 discloses a new spectroscopic method of near-infrared Fourier fast transform (FT-NIR) to detect the authenticity of extra virgin olive oils (EVOO) and to determine the kind and quantity of an adulterant in EVOO.
El documento "The Detection and Quantification of Adulteration in Olive Oil by Near- Infrared Spectroscopy and Chemometrics" (Christy et al) describe un nuevo procedimiento para la clasificación y cuantificación de la adulteración de aceite de oliva puro por aceite de soja, aceite de flor de sol, aceite de maíz, aceite de nuez y aceite de avellana. El procedimiento se basa en un análisis quimiométrico de los espectros del infrarrojo cercano (NIR) de mezclas de aceite de oliva, que contenían diferentes adulterante. The document "The Detection and Quantification of Adulteration in Olive Oil by Near-Infrared Spectroscopy and Chemometrics" (Christy et al) describes a new procedure for the classification and quantification of the adulteration of pure olive oil by soybean oil, flower oil of sun, corn oil, walnut oil and hazelnut oil. The procedure is based on a chemometric analysis of the near infrared (NIR) spectra of olive oil mixtures, which contained different adulterant.
RESUMEN DE LA INVENCIÓN La presente invención corresponde a un método para la estimación del contenido de aceite en olivas individuales en base a tecnologías no destructivas. En pocas palabras, se forman grupos de olivas con características similares considerando el color y las longitudes de onda NIR de absorción de luz de los ácidos grados. Dado que las olivas de los grupos son similares, el promedio de las características es representativo del conjunto de olivas (baja desviación estándar), y por ello es adecuado asociar dichas características promedio con el contenido de aceite. El modelo resultante permite estimar con un error aceptable por la empresa de producción de aceite el contenido de aceite de una oliva en base a características NIR y de color las cuales se miden sin necesidad de destruir las olivas. SUMMARY OF THE INVENTION The present invention corresponds to a method for estimating the oil content in individual olives based on non-destructive technologies. In short, groups of olives with similar characteristics are formed considering the color and NIR wavelengths of light absorption of the acid grades. Since the olives of the groups are similar, the average of the characteristics is representative of the set of olives (low standard deviation), and therefore it is appropriate to associate said average characteristics with the oil content. The resulting model allows to estimate with an acceptable error by the oil production company the content of olive oil based on NIR and color characteristics which are measured without the need to destroy the olives.
El método permite estimar el aceite de olivas individuales sin destruirlas, lo que evita el trabajo de laboratorio que implica el destruir la muestra, y se pueden asociar mediante un modelo las características de las olivas medidas (Color y NIR) con el contenido de aceite obtenido como referencia mediante SOXHLET. The method allows to estimate the oil of individual olives without destroying them, which avoids the laboratory work that involves destroying the sample, and can be associated by a model the characteristics of the olives measures (Color and NIR) with the obtained oil content as a reference through SOXHLET.
BREVE DESCRIPCIÓN DE LAS FIGURAS BRIEF DESCRIPTION OF THE FIGURES
La figura 1 represente unespectro de infrarrojo cercano NIR con un espectrómetro en modalidad de reflectancia, de acuerdo a una modalidad de la invención. Figure 1 represents a near infrared spectrum NIR with a spectrometer in reflectance mode, according to one embodiment of the invention.
La figura 2 representa la obtención de bolsas de olivas similares mediante Kmeans++, de acuerdo a una modalidad de la invención. Figure 2 represents obtaining similar bags of olives by Kmeans ++, according to one embodiment of the invention.
La figura 3 represente un equipos qpara extraer aceite en paralelo con seis depósitos para las muestras secas, de acuerdo a una modalidad de la invención. La figura 4 represente un equipos qpara extraer aceite en paralelo con seis depósitos para las muestras secas, de acuerdo a una modalidad de la invención. Figure 3 represents equipment for extracting oil in parallel with six tanks for the dry samples, according to one embodiment of the invention. Figure 4 represents equipment for extracting oil in parallel with six tanks for the dry samples, according to one embodiment of the invention.
DESCRIPCIÓN DETALLADA DE LA INVENCIÓN DETAILED DESCRIPTION OF THE INVENTION
La presente invención corresponde a un método para estimar el contenido de aceite de olivas individuales en base a características NIR y Color. Dicho método estimador del contenido de aceite comprende las siguientes etapas: ) Obtener características NIR y Color de Olivas: A un conjunto de olivas que provienen del campo de la empresa productora de aceite se le obtienen características de color y NIR. The present invention corresponds to a method for estimating the oil content of individual olives based on NIR and Color characteristics. Said method estimating the oil content comprises the following stages: ) Obtain characteristics NIR and Color of Olives: To a set of olives that come from the field of the company producing of oil obtains characteristics him of color and NIR.
Las olivas de las bolsas provenientes del campo se ponen en una bandeja, preferentemente una bandeja plástica transparente. Mediante un ambiente iluminación difusa controlada, se obtiene una imagen sin golpes de luz ni sombras (defectos típicos de las imágenes adquiridas con luz direccional). The olives of the bags from the field are placed in a tray, preferably a transparent plastic tray. By means of a controlled diffuse lighting environment, an image is obtained without light hits or shadows (typical defects of the images acquired with directional light).
Se realiza la segmentación de cada una de las olivas mediante técnicas de tratamiento de imágenes obteniéndose como resultado la individualización de cada una de las olivas. The segmentation of each of the olives is carried out by means of image processing techniques, obtaining as a result the individualization of each of the olives.
A las olivas individualizadas se les obtiene el color promedio en un modelo de color RGB y su transformación al modelo d c2c3. Se utiliza como característica de color el modelo c1 c2c3 pues corresponde a un modelo de color invariante a la iluminación, luego permite asegurar robustez de la característica de color frente a posibles variaciones de iluminación. The individualized olives are obtained the average color in a RGB color model and its transformation to the model d c2c3. The c1 c2c3 model is used as a color characteristic since it corresponds to a color model that is invariant to the illumination, then it makes it possible to ensure the robustness of the color characteristic against possible variations in lighting.
A cada una de las olivas de las cajas, se le obtiene su espectro de infrarrojo cercano NIR con un espectrómetro en modalidad de reflectancia tal como se muestra en la Figura 1 Los espectros obtenidos tienen la cantidad de puntos y resolución por punto en [nm] dependiendo del equipo utilizado. To each one of the olives of the boxes, their NIR near infrared spectrum is obtained with a spectrometer in reflectance mode as shown in Figure 1 The spectra obtained have the number of points and resolution per point in [nm] depending on the equipment used.
El fenómeno estudiado es la absorción de luz por los ácidos grasos, yestado del arte indica que los ácidos grasos mayoritarios tienen una absorción de luz centrada en 1728 nm. Por lo anterior, el rango escogido varía entre los 1710 nm y 1735 nm que corresponden a 1 1 puntos del total de puntos de los espectros obtenidos. Lo anterior permite justificar de forma pertinente el uso de un grupo de longitudes de onda y no todo el espectro. The phenomenon studied is the absorption of light by fatty acids, and art indicates that the major fatty acids have a light absorption centered at 1728 nm. Therefore, the chosen range varies between 1710 nm and 1735 nm, corresponding to 1 1 points of the total points of the spectra obtained. The The previous one allows to justify in a pertinent way the use of a group of wavelengths and not the whole spectrum.
Según la descripción anterior, la oliva se representa con 3 puntos del modelo de color d c2c3 más los 1 1 puntos del espectro donde se produce la absorción de los ácidos grasos mayoritarios. En total el descriptor VIS-NIR de la Oliva tiene 14 puntos. According to the previous description, the olive is represented with 3 points of the color model d c2c3 plus the 1 1 points of the spectrum where the absorption of the major fatty acids takes place. In total the descriptor VIS-NIR of the Olive has 14 points.
Luego de obtener las características VIS-NIR las cajas de olivas se congelan a una temperatura de -20 °C para detener la evolución de la degradación de la oliva. El estado del arte indica que el congelamiento de las olivas no altera el contenido de aceite (no así sus características de color y parámetros de calidad). En este sentido es posible guardar las bandejas de olivas congeladas en espera de la etapa de identificación de olivas similares. After obtaining the VIS-NIR characteristics, the olive boxes are frozen at a temperature of -20 ° C to stop the evolution of olive degradation. The state of the art indicates that the freezing of olives does not alter the oil content (but not its color characteristics and quality parameters). In this sense it is possible to keep the frozen olive trays awaiting the identification stage of similar olives.
) Agrupar olivas en base a características de color y NIR: Esta agrupación permite asociar adecuadamente características de una oliva al contenido de aceite entregado por el SOHXLET para obtener aceite de un grupo de olivas muy similares. Esto permite que las olivas consideradas para obtener la muestra seca tengan una baja desviación estándar de sus características. ) Group olives based on color characteristics and NIR: This grouping allows to properly associate characteristics of an olive to the oil content delivered by the SOHXLET to obtain oil from a group of very similar olives. This allows that the olives considered to obtain the dry sample have a low standard deviation of their characteristics.
La obtención de grupos de olivas similares se obtiene mediante una técnica de agrupación o clustering. El descriptor VIS-NIR de la oliva corresponde a un vector de 14 elementos constituido por los 3 canales del modelo c1 c2c3 y los 1 1 puntos del espectro NIR seleccionados para ácidos grasos principales. Se hace notar cada descriptor fue normalizado en la escala para darle igual importancia a ambos tipos de datos, considerando una normalización respecto de la media de cada elemento del descriptor. Se agrupan de olivas similares mediante el agrupamiento el descriptor VIS-NIR normalizado. El procedimiento de agrupamiento se basa en el algoritmo kmeans++, el cual agrupa descriptores según similaridad y necesita como parámetro el número de grupos. Se desarrollo un algoritmo de agrupamiento que permite encontrar el número máximo de grupos posibles teniendo cada un grupo un tamaño mínimo determinado. El estándar de extracción de aceite SOXHLET requiere una masa mínima de aproximadamente 15 olivas por extracción de un valor de aceite, además de la realización de mediciones en duplicado como medida de seguridad (muestra y contra-muestra), esto significa que cada bolsa tiene que tener un número mínimo de 30 olivas para muestra y contra-muestra de verificación. De las cajas donde están guardadas las olivas se sacan las olivas correspondientes indicadas por el agrupamiento y se almacenan en bolsas tal como se muestra en la Figura 2. Las bolsas se vuelven a congelar a la espera de la etapa de extracción de aceite por SOXHLET. ) Extraer aceite mediante SOXHLET: Cada una de las bolsas congeladas en la etapa anterior, tiene mínimamente 30 olivas, es posible realizar por bolsa al menos 2 mediciones de aceite, una para muestra y otra para contra-muestra. La contramuestra tiene un sentido de verificación del valor de obtenido de aceite, pues al ser de olivas similares, muestra y contra-muestra tienen que tener valores de aceite muy cercanos. Obtaining groups of similar olives is obtained by means of a grouping or clustering technique. The VIS-NIR descriptor of the olive corresponds to a vector of 14 elements constituted by the 3 channels of the model c1 c2c3 and the 1 1 points of the NIR spectrum selected for main fatty acids. It is noted that each descriptor was normalized in the scale to give equal importance to both types of data, considering a normalization with respect to the mean of each descriptor element. They are grouped of similar olives by grouping the standard VIS-NIR descriptor. The grouping procedure is based on the kmeans ++ algorithm, which groups descriptors according to similarity and needs as a parameter the number of groups. A grouping algorithm was developed to find the maximum number of possible groups, each group having a certain minimum size. The SOXHLET oil extraction standard requires a minimum mass of approximately 15 olives per extraction of an oil value, in addition to carrying out measurements in duplicate as a safety measure (sample and counter-sample), this means that each bag has to Have a minimum number of 30 olives for sample and counter-sample verification. From the boxes where the olives are stored, the corresponding olives indicated by the grouping are removed and stored in bags as shown in Figure 2. The bags are refrozen while awaiting the SOXHLET oil extraction stage. ) Extract oil through SOXHLET: Each of the bags frozen in the previous stage, has at least 30 olives, it is possible to make at least 2 measurements of oil per bag, one for sample and another for counter-sample. The countersample has a sense of verification of the value of obtained oil, since to be of similar olives, sample and counter-sample must have very close oil values.
El método de extracción de aceite SOXHLET esta normalizado por el Consejo Olivícola Internacional, y consiste en extraer las grasas de una muestra seca utilizando un solvente apolar (hexano), haciendo circular en forma continua el hexano por la muestra durante un mínimo de 6 horas. El número de horas de uso de extractor es variable según el número de muestras secas, pero para realizar un modelo robusto se van a requerir centenas de horas en la extracción del terreno de la verdad de contenido de aceite. Es justamente esta una característica del proceso que se mejora con el método desarrollado. Si bien es cierto es necesario tener las referencias de aceite para construir el modelo, las estimaciones posteriores no requieren de extracción de aceite, y los tiempos involucrados son los que llevan las mediciones del color y del espectro NIR. Para disminuir el tiempo de extracción se utilizan equipos que tienen la capacidad de extraer aceite en paralelo, tal como es el caso de la Figura 3, que tiene 6 depósitos para las muestras secas. The SOXHLET oil extraction method is standardized by the International Olive Council, and consists of extracting the fats from a dry sample using an apolar solvent (hexane), by continuously circulating the hexane through the sample for a minimum of 6 hours. The number of hours of use of extractor is variable according to the number of dry samples, but to make a Robust model will require hundreds of hours in extracting the ground from the truth of oil content. It is precisely this characteristic of the process that is improved with the developed method. While it is true it is necessary to have the oil references to build the model, subsequent estimates do not require oil extraction, and the times involved are those that take the measurements of the color and the NIR spectrum. To reduce the extraction time, equipment that has the capacity to extract oil in parallel is used, as is the case of Figure 3, which has 6 tanks for the dry samples.
La preparación de la muestra consiste en tomar un grupo de olivas, y molerlas con un molino pulverizador, y pesar de la muestra en peso fresco. Luego se lleva la muestra a un horno para sacarle el agua y dejar solo la materia seca la cual también se pesa. De la materia seca se extraen entre 2 a 5 gramos (según la norma del método SOXHLET) para extracción de grasa. La muestra seca se ingresa a un dedal de extracción de celulosa el cual se sumerge en hexano, y se hace recircular el solvente a través de la muestra por 6 horas para extraer aceite. Al final del proceso queda separado de la muestra, en un vaso con hexano. Este vaso se lleva a una estufa para evaporar el hexano y quedando así solo el aceite. Para calcular la cantidad de aceite se hace la diferencia entre el peso del vaso con aceite y el peso del vaso sin aceite. Los resultados de contenido de aceite están expresados en aceite de las olivas base peso seco. ) Desarrollar un modelo de estimación: Para el caso particular del procedimiento de la presente invención, se logran 62 muestras para extracción de aceite. Este valor es importante para poder indicar la técnica de validación del modelo a implementar. El modelo que asocia las 14 características VIS-NIR de una oliva individual con el contenido de aceite de las muestras corresponde a un modelo de regresión. En este caso, se desarrolla un modelo basado en Support Vector Machine for Regresión (SVR). La literatura muestra SVR corresponden a máquinas de aprendizaje que tienen ventajas sobre las Redes Neuronales y los modelos regresionales que definen polinomios. Dentro de las características positivas de los SVR se encuentran que no es necesario asumir un modelo polinomial, que su algoritmo de computo está muy bien definido matemáticamente y asegura encontrar un modelo óptimo, y que tiene pocos hiper-parámetros a definir (2 en el caso del SVR). The preparation of the sample consists of taking a group of olives, and grinding them with a pulverizing mill, and weighing the sample in fresh weight. Then the sample is taken to an oven to remove the water and leave only the dry matter which is also weighed. Dry matter is extracted between 2 and 5 grams (according to the norm of the SOXHLET method) for the extraction of fat. The dry sample is entered into a cellulose extraction thimble which is immersed in hexane, and the solvent is recirculated through the sample for 6 hours to extract oil. At the end of the process it is separated from the sample, in a glass with hexane. This glass is taken to a stove to evaporate the hexane and leaving only the oil. To calculate the amount of oil, the difference between the weight of the glass with oil and the weight of the glass without oil is made. The results of oil content are expressed in olive oil base dry weight. ) Develop an estimation model: For the particular case of the process of the present invention, 62 samples are obtained for oil extraction. This value is important to be able to indicate the validation technique of the model to be implemented. The model that associates the 14 VIS-NIR characteristics of an individual olive with the oil content of the samples corresponds to a regression model. In this case, a model based on Support Vector Machine for Regression (SVR) is developed. The literature shows SVR correspond to learning machines that have advantages over Neural Networks and the regression models that define polynomials. Among the positive characteristics of the SVR are that it is not necessary to assume a polynomial model, that its computation algorithm is very well defined mathematically and ensures to find an optimal model, and that it has few hyper-parameters to be defined (2 in the case of the SVR).
Debido al conjunto de datos, para validar el modelo se realiza Validación Cruzada Dejando Uno a fuera (Leave One-Out Cross Validation) LOOCV. Esta estrategia de validación permite validar modelos cuando existen pocos datos. En este caso LOOCV se ajusta al problema de estimación del contenido de aceite de olivas individuales, pues debido al costo de tiempo y logística necesaria para obtener las referencias de aceite, no es posible obtener cientos de datos para construir el modelo. El funcionamiento del LOOCV se presenta en la Figura 4, si se tienen N datos, se deja uno afuera y se estima un modelo con los restantes N-1 datos. En LOOCV se estiman N modelos, y evaluación del esquema corresponde al promedio de los N modelos. Due to the data set, to validate the model Cross Validation is performed Leaving One out (Leave One-Out Cross Validation) LOOCV. This validation strategy allows to validate models when there is little data. In this case LOOCV fits the problem of estimating the oil content of individual olives, because due to the cost of time and logistics necessary to obtain the oil references, it is not possible to obtain hundreds of data to build the model. The functioning of the LOOCV is presented in Figure 4, if we have N data, we leave one outside and we estimate a model with the remaining N-1 data. In LOOCV, N models are estimated, and evaluation of the scheme corresponds to the average of the N models.
EJEMPLO DE APLICACIÓN APPLICATION EXAMPLE
El método de validación experimental corresponde al mencionado en la Etapa 4. Para N grupos de olivas similares las cuales fueron agrupadas por sus 14 características (3 Color + 1 1 NIR) se les extrajo aceite. Mediante LOOCV se realizaron N modelos de estimación dejando uno afuera, y se considera como resultado de la prueba estadística el promedio de los N modelos. The experimental validation method corresponds to that mentioned in Stage 4. For N groups of similar olives which were grouped by their 14 characteristics (3 Color + 1 1 NIR) oil was extracted. Through LOOCV, N estimation models were made, leaving one outside, and the average of the N models is considered as a result of the statistical test.
Se realizaron experimentos con 2 grupos de datos: Experiments were performed with 2 groups of data:
Grupo 1: Se obtuvieron N=62 mediciones de aceite de olivas similares cuya muestra seca si contiene elementos del hueso. Group 1: N = 62 olive oil measurements were obtained whose sample dried if it contains elements of the bone.
Grupo 2: Se obtuvieron N=62 mediciones de aceite de olivas similares cuya muestra seca no contiene elementos del hueso. Group 2: N = 62 olive oil measurements were obtained whose dry sample does not contain bone elements.
Los resultados del error de la estimación de aceite considerando el promedio de los 62 modelos SVR se muestran en la siguiente tabla.
De los resultados se observa que no es necesario sacar el hueso a las olivas, pues el aporte de acierte del hueso es mínimo (1 %). Finalmente, sin sacar el hueso se obtiene un error porcentual de estimación de aceite del 7%. From the results it is observed that it is not necessary to remove the bone from the olives, since the contribution of the bone is minimal (1%). Finally, without removing the bone, a percentage error of oil estimate of 7% is obtained.
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CL2192-2017 | 2017-08-29 | ||
| CL2017002192A CL2017002192A1 (en) | 2017-08-29 | 2017-08-29 | Method for estimating individual olive oil based on non-destructive technologies. |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2019041055A1 true WO2019041055A1 (en) | 2019-03-07 |
Family
ID=63046495
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CL2018/050071 Ceased WO2019041055A1 (en) | 2017-08-29 | 2018-08-24 | Method for estimating the oil of individual olives using non-destructive technologies |
Country Status (2)
| Country | Link |
|---|---|
| CL (1) | CL2017002192A1 (en) |
| WO (1) | WO2019041055A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111795944A (en) * | 2020-06-30 | 2020-10-20 | 江南大学 | A NIR Spectral Wavelength Screening Method with Improved Team Progress Algorithm |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ES2431955T3 (en) * | 2000-10-30 | 2013-11-28 | Monsanto Technology Llc | Procedure and device for analyzing agricultural products |
-
2017
- 2017-08-29 CL CL2017002192A patent/CL2017002192A1/en unknown
-
2018
- 2018-08-24 WO PCT/CL2018/050071 patent/WO2019041055A1/en not_active Ceased
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ES2431955T3 (en) * | 2000-10-30 | 2013-11-28 | Monsanto Technology Llc | Procedure and device for analyzing agricultural products |
Non-Patent Citations (9)
| Title |
|---|
| CAYUELA, J. ET AL.: "Sorting olive oil based on alpha-tocopherol and total tocopherol content using near-infra-red spectroscopy (NIRS) analysis", JOURNAL OF FOOD ENGINEERING, vol. 202, 2017, pages 79 - 88, XP055581884, Retrieved from the Internet <URL:http://dx.doi.org/10.1016/j.jfoodeng. 2017.01.01 5> * |
| GARCIA, A. ET AL.: "Estudio comparativo de distintas tecnicas analiticas (espectroscopia ' de NIR y RMN y extracci6n mediante Soxhlet) para la determination del contenido graso y de humedad en aceitunas y orujo de Jaen", GRASAS Y ACEITES, vol. 56, no. 3, 2005, pages 220 - 227, XP055581903, ISSN: 0017-3495 * |
| GIOVENZANA, V. ET AL.: "Postharvest Characterization of Olive Oil Fruits Texture by NIR and Vis/NIR Spectroscopy", CHEMICAL ENGINEERING TRANSACTIONS, vol. 44, 2015, XP055581889, ISSN: 2283-9216, ISBN: 978-88-95608-35-8 * |
| GUZMAN, E. ET AL.: "Determination of the olive maturity index of intact fruits using image analysis", J FOOD SCI TECHNOL, vol. 52, no. 3, March 2015 (2015-03-01), pages 1462 - 1470, XP035461865, DOI: doi:10.1007/s13197-013-1123-7 * |
| GUZMAN, E. ET AL.: "Using a Visible Vision System for On-Line Determination of Quality Parameters of Olive Fruits", FOOD AND NUTRITION SCIENCES, vol. 4, no. 7, 2013, pages 90 - 98, XP055581882, Retrieved from the Internet <URL:http://dx.doi.org/10.4236/fns.2013.47A011> * |
| LEON. L. ET AL.: "Analisis de aceituna intacta mediante espectroscopia en el infrarrojo cercano (NIRS): una herramienta de utilidad en programas de mejora de olivo", GRASAS Y ACEITES, vol. 54, no. 1, 30 March 2003 (2003-03-30), pages 41 - 47, XP055581893, Retrieved from the Internet <URL:https://doi.org/10.3989/gya.2003.v54.i1.275> * |
| MARLEY, M.: "Near-infrared spectroscopy and hyperspectral imaging: non-destructive analysis of biological materials", CHEM. SOC. REV., vol. 43, 2014, pages 8200 - 8214, XP055378134, DOI: doi:10.1039/C4CS00062E * |
| SALGUERO-CHAPARRO, L. ET AL.: "On-line versus off-line NIRS analysis of intact olives", LWT - FOOD SCIENCE AND TECHNOLOGY, vol. 56, 2014, pages 363 - 369, XP055581724, Retrieved from the Internet <URL:http://dx.doi.org/10.1016/j.lwt. 2013.11.03 2> * |
| VARMUZA KURT: "Chemometrics in Practical Applications", 23 March 2012, INTECH, ISBN: 978-953-51-0438-4, article GUIDETTI, R. ET AL.: "Chemometrics in Food Technology", pages: 217 - 252, XP055581880 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111795944A (en) * | 2020-06-30 | 2020-10-20 | 江南大学 | A NIR Spectral Wavelength Screening Method with Improved Team Progress Algorithm |
| CN111795944B (en) * | 2020-06-30 | 2021-06-25 | 江南大学 | A NIR Spectral Wavelength Screening Method with Improved Team Progress Algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CL2017002192A1 (en) | 2018-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mishra et al. | Near-infrared hyperspectral imaging for non-destructive classification of commercial tea products | |
| Huang et al. | Improved generalization of spectral models associated with Vis-NIR spectroscopy for determining the moisture content of different tea leaves | |
| Darvishzadeh et al. | Inversion of a radiative transfer model for estimation of rice canopy chlorophyll content using a lookup-table approach | |
| CN107271375B (en) | A kind of high spectral image detecting method of quality of mutton index | |
| CN106404692A (en) | Method for detecting freshness grade of instant sea cucumber by using hyperspectral imaging technology | |
| Wang et al. | Remote sensing of fuel moisture content from ratios of narrow-band vegetation water and dry-matter indices | |
| CN103018179B (en) | Device and method for near-infrared polarization detection of crop water stress | |
| BR112019012178A2 (en) | texture classification and soil content by near infrared spectroscopy | |
| CN104255118B (en) | Rapid lossless testing method based on near infrared spectroscopy technology for paddy rice seed germination percentage | |
| Yue et al. | A dynamic soil endmember spectrum selection approach for soil and crop residue linear spectral unmixing analysis | |
| CN103389255A (en) | Method for predicting water content of tea on basis of near-infrared hyperspectral textural feature modeling | |
| CN103196838A (en) | Hyperspectral remote sensing monitoring method for coastal estuary eutrophication | |
| CN107664621B (en) | Abnormal sample identification method and system, server and electronic equipment | |
| Mobasheri et al. | Leaf Equivalent Water Thickness assessment using reflectance at optimum wavelengths | |
| CN105823752B (en) | The method that near infrared spectroscopy quickly differentiates edible oil type | |
| Taous et al. | Geographical classification of authentic Moroccan Argan oils and the rapid detection of soya and sunflower oil adulteration with ATR-FTIR spectroscopy and chemometrics | |
| CN116519693B (en) | Quantitative measurement method for Hetian jade color | |
| Chen et al. | New approach of simultaneous, multi‐perspective imaging for quantitative assessment of the compactness of grape bunches | |
| CN107219184A (en) | Meat identification method and device applied to origin tracing | |
| CN114972160A (en) | Natural pasture grass yield rapid measurement method and system based on unmanned aerial vehicle | |
| Onoyama et al. | Potential of hyperspectral imaging for constructing a year-invariant model to estimate the nitrogen content of rice plants at the panicle initiation stage | |
| WO2019041055A1 (en) | Method for estimating the oil of individual olives using non-destructive technologies | |
| CN105548455A (en) | Method for evaluation of winter wheat blade nitrogen accumulation amount from green turning to heading | |
| CN104502307A (en) | Method for quickly detecting content of glycogen and protein of crassostrea gigas | |
| CN111122635A (en) | Method and system for online evaluation of drying state of edible fungi |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18849532 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 18849532 Country of ref document: EP Kind code of ref document: A1 |