Search

[EN] In this study, an industrial washing process for harvested products is modeled, simulated, and optimized. For this purpose, a flume washing tank is considered, where we assume that the longitudinal spatial dimension retains the fundamental dynamics of free chlorine (FC), organic matter (COD), and bacterial load (Xw). The model is used to optimize the addition of the disinfectant, which enters the tank from three injection ports, ensuring that the microbial load at the end of the flume tank is below a considered safe threshold while minimizing the maximum amounts of disinfectant that could produce hazardous byproducts, [ES] En este estudio se modela, simula y optimiza un proceso de lavado industrial de productos de cosecha. Para ello se considera un tanque de lavado en forma de canal asumiendo que la dimensión espacial longitudinal retiene las dinámicas fundamentales de las dinámicas de cloro libre (FC), materia orgánica (COD) y carga bacteriana (Xw). El modelo se utiliza para optimizar la adición del desinfectante, que entra a lo largo del tanque en tres puntos de inyección, de forma que se garantice que la carga microbiana a la salida del tanque está por debajo de un umbral considerado seguro, mientras se minimiza las cantidades máximas de desinfectante que pudieran producir productos secundarios peligrosos

[EN] In this work, we propose a new mathematical model for the description of the sterilization process of packaged food in steam retorts. The main advantage of such a model with respect to those existing in the literature is that it has been developed to be used in the framework of nolinear model predictive control (NMPC), which requires solving on-line dynamic optimization problems (periods shorter than 5 seconds). In addition, the proposed model describes a number of physical phenomena that had not been taken into account in previous models. In particular, it considers the possibility that: (i) air enters through the purge; (ii) gases exit through the drain; (iii) liquid water and steam in the autoclave are not in equilibrium. The main contribution is the use of the Fischer-Burmeister equation to model the discrete state changes derived from (iii) in an efficient way, [ES] En este trabajo proponemos un nuevo modelo matemático para la descripción del proceso de esterilización de alimentos envasados en autoclaves de vapor. La principal ventaja de dicho modelo con respecto a los existentes en la literatura es que ha sido desarrollado para ser usado en el marco del control predictivo no lineal basado en modelos (NMPC, por sus siglas en inglés), lo que requiere resolver problemas de optimización dinámica en línea (periodos inferiores a 5 segundos). Además, el modelo propuesto describe una serie de fenómenos físicos que no se habían tenido en cuenta en modelos anteriores. En particular, se considera la posibilidad de que: (i) el agua y el vapor en el autoclave no estén en equilibrio; (ii) haya entrada de aire por la purga; (iii) haya salida de gases por el drenaje. La principal contribución es la utilización de la ecuación de Fischer-Burmeister para modelar los cambios de estado discretos derivados de (i) de forma eficiente

[EN] Developing predictive antimicrobial resistance (AMR) models for supporting optimal treatment design poses one of the biggest challenges for mathematical biology. Birth-death (BD) processes constitute an intuitive and robust approach for modelling stochastic dynamics of biological systems, which are regaining attention due to rapid advances in computational techniques. In this work, we propose a BD model of bacterial heteroresistance, defined as the presence of many subpopulations with reduced antimicrobial sensitivities. The model accounts for competition between subpopulations so that the equations for the moment dynamics of the BD process are coupled, forming an infinite system. We propose a moment closure assuming a log-normal distribution to approximate the behaviour of the heteroresistant population under antimicrobial stress and compare the results with stochastic simulations of the actual BD process for a typical time-kill assay, [ES] El desarrollo de modelos predictivos de Resistencia Antimicrobiana (RAM) que apoyen el diseño óptimo de tratamientos para combatir las “superbacterias” es uno de los mayores retos que debe afrontar actualmente la biología matemática. Los procesos de Nacimiento-Muerte (NM) constituyen un enfoque intuitivo y potente para modelar las dinámicas estocásticas propias de los sistemas biológicos, el cual está recuperando atención gracias a los rápidos avances en las técnicas computacionales. En este trabajo, se presenta un modelo NM de heterorresistencia bacteriana, un fenotipo según el cual se distinguen diferentes subpoblaciones con sensibilidad reducida al antimicrobiano en comparación con la población principal. El modelo considera competencia entre las distintas subpoblaciones, provocando que las dinámicas para los momentos del proceso NM se encuentren totalmente acopladas, formando un sistema infinito de ecuaciones. Así, proponemos una clausura de momentos basada en la distribución log-normal para aproximar el comportamiento de la población heterorresistente bajo estrés antimicrobiano, y comparamos los resultados con simulaciones estocásticas del proceso NM real para un típico ensayo de tiempo-muerte

Scheduling tasks in production facilities are usually hybrid optimization problems of a large combinatorial nature. They involve solving, in near-real time, the integration of the operation of several batch units of continuous dynamics with the discrete manufacture of items in processing lines. Moreover, one has to deal with uncertainty (process delays, unexpected stops) and the management of shared resources (energy, water, etc.) including decisions made by plant operators: still, some tasks in the scheduling layers are done manually. Manufacturing Execution Systems (MESs) are intended to support plant personnel at this level. However, there is still much work to do in terms of performing automatic scheduling, computed in real time, that guides managers to achieve an optimal operation of such complex cyber-physical systems. This work proposes a closed-loop approach to handle the uncertainty arising when facing the online scheduling of supply lines and parallel batch units. These units often share some resources, so effects due to concurrent resource consumption on the system dynamics are explicitly considered in the presented formulation. The proposed decision support system is tested onsite in a tuna cannery, to handle short-term online scheduling of sterilization processes that deal with limited steam, carts, and operators as shared resources

This repository contains the bibliometric search results discussed in the manuscript: García, MR, Férez-Rubio, JA, Vilas, C. (2022). Assessment and Prediction of Fish Freshness using Mathematical Modelling: A Review. MDPI Foods, 11(15):2312. DOI:10.3390/foods11152312.

[EN] The rapid development of bacterial antimicrobial resistance still promotes the research on mathematical modelling of the mechanisms leading to the emergence, transmission and selection of resistance. The most common approach to explain the dynamics of an antimicrobial-treated population is to classify bacteria into sensitive and resistant. In reality, however, we can find infinite resistance degrees in the population, known as heteroresistance. In this paper, we propose a mathematical model to explain the dynamics of a heteroresistant population, introducing the resistance degree as a random variable indicating the effectiveness of the antimicrobial. We analyse the model’s capabilities and establish its structural observability and identifiability, [ES] El rápido avance de la resistencia bacteriana a los antimicrobianos continúa fomentando la investigación en torno al modelado matemático de los mecanismos que conducen a la emergencia, transmisión y selección de resistencia. El enfoque más extendido para explicar las dinámicas de una población tratada con antimicrobiano consiste en clasificar las bacterias en sensibles y resistentes. Sin embargo, en la realidad existen infinidad de grados de resistencia intermedios en la población, lo que se conoce como heterorresistencia. En este trabajo, se propone un modelo matemático para explicar las dinámicas de una población heterorresistente, introduciendo el grado de resistencia como una variable aleatoria indicadora de la efectividad del antimicrobiano sobre la población. Se analizan las capacidades del modelo y se establecen, además, su observabilidad e identificabilidad estructurales

[EN] Food waste is a significant sustainability problem. Fish, in particular, is a very perishable product. In fact, spoilage increases rapidly when the storage temperature is not adequately controlled. Mathematical models are useful tools to predict fish quality evolution during transport and storage. This knowledge will help to develop strategies to reduce food waste and to increase the product value. In this work, we propose a new methodology to predict the evolution of quality, during transport and storage, of hake fillets packed in modified atmospheres. It consists of two parts: a model based on the process mechanisms to describe the evolution of two highly used quality indicators (bacterial content and volatile bases), and a machine learning model to correlate such indicators with one of the most employed sensory indexes: the Quality Index Method (QIM). Moreover, we propose a Power Law model to describe volatile bases evolution. We estimate the unknown parameters using experimental data. In the machine learning part, we compare random forest models and neural networks using several statistical techniques. Overall, this methodology is an improvement in fish quality prediction because it makes QIM prediction using just temperature data with promising results, [ES] El desperdicio alimentario supone un gran problema de sostenibilidad. El pescado, en concreto, es un producto especialmente perecedero, degradándose con mucha rapidez cuando la temperatura de almacenamiento no se controla adecuadamente. Los modelos matemáticos son herramientas útiles para predecir la evolución de la calidad del pescado durante el transporte y almacenamiento. Conocer la dinámica de la degradación con antelación también nos permitir´a definir estrategias para reducir el desperdicio alimentario y aumentar el valor añadido del pescado. En este trabajo se presenta una nueva metodología para la descripción de la evolución de calidad de filetes de merluza (Merluccius merluccius) envasada en atmósferas modificadas durante el transporte y almacenamiento. Dicha metodología consta de dos partes: un modelo basado en los mecanismos del proceso que describe la evolución de dos de los indicadores de calidad más utilizados (contenido bacteriano y bases volátiles), y un modelo de aprendizaje automático que permite correlacionar dichos indicadores con un indicador sensorial, el QIM (del inglés, Quality Index Method). Para la evolución de las bases volátiles se propone un nuevo modelo basado en la ley de potencias. Además, en este trabajo, los parámetros desconocidos del modelo se estiman utilizando datos experimentales. En el caso de los modelos de aprendizaje automático, se comparan técnicas de redes neuronales y árboles de decisión utilizando diferentes indicadores del error. El modelo combinado supone un avance en la predicción de calidad, ya que permite predecir el QIM a partir, exclusivamente, de datos de temperatura y obtener resultados satisfactorios

The implementation and fully compliance of the Common Fisheries Policy (CFP) of the EU depends largely on the ability to quantify total catches on board commercial fishing vessels. To this aim, the use of electronic devices is gaining relevance and vision-based electronic monitoring technologies have emerged as a more cost-effective and efficient way to monitor fishing activity. In this work, we present the iObserver 2.0, a device that uses Deep Learning image recognition to automatically identify and quantify in real time the entire catch on board fishing vessels. It builds upon two previous prototypes, improving image quality by using line scan technology. Two neural networks are used for fish species segmentation, identification, and length regression tasks. As main results of this disruptive technology, the iObserver 2.0 distinguishes more than twice the number of species than previous version, works with area scan and line scan camera images, and it is evaluated with a test set incorporating more complex images. An experimental fishing survey has been conducted to assess the system’s performance in real-life conditions, showing promising results in terms of total catch registration of target and discard fish species

The implementation and fully compliance of the Common Fisheries Policy (CFP) of the EU depends largely on the ability to quantify total catches on board commercial fishing vessels. To this aim, the use of electronic devices is gaining relevance and vision-based electronic monitoring technologies have emerged as a more cost-effective and efficient way to monitor fishing activity. In this work, we present the iObserver 2.0, a device that uses Deep Learning image recognition to automatically identify and quantify in real time the entire catch on board fishing vessels. It builds upon two previous prototypes, improving image quality by using line scan technology. Two neural networks are used for fish species segmentation, identification, and length regression tasks. As main results of this disruptive technology, the iObserver 2.0 distinguishes more than twice the number of species than previous version, works with area scan and line scan camera images, and it is evaluated with a test set incorporating more complex images. An experimental fishing survey has been conducted to assess the system’s performance in real-life conditions, showing promising results in terms of total catch registration of target and discard fish species., Peer Reviewed

El uso extensivo de compuestos antimicrobianos (antibióticos, desinfectantes y conservantes) durante las últimas décadas, especialmente en las industrias farmacéutica y alimentaria, ha acelerado la aparición de bacterias resistentes o “superbacterias”. El problema de la resistencia a los antimicrobianos (AMR, por las siglas en inglés de Antimicrobial Resistance) se agrava a un ritmo alarmante, sobrepasando ampliamente nuestra capacidad para desarrollar nuevos tratamientos. Tan sólo en el año 2019, estudios recientes estiman en 1.27 millones los fallecimientos directamente relacionados con infecciones provocadas por bacterias resistentes, mientras que, para 2050, se prevé que esta será la causa de unos 10 millones de decesos anuales alrededor del mundo (Murray et al., 2022)

El pescado fresco es un producto que se degrada con rapidez (Tavares et al., 2021), lo que conlleva que cerca de un 15% del pescado capturado se desperdicie. Si las condiciones de transporte y almacenamiento (principalmente temperatura) no son adecuadas, el desperdicio puede llegar al 35% (Corradini, 2018). Los modelos matemáticos que describen la degradación de calidad de pescado fresco a lo largo de toda la cadena de valor son herramientas útiles que nos permiten determinar los efectos de posibles abusos de temperatura. Además, también permiten proporcionar al consumidor información relevante sobre el estado del producto y su vida útil, aumentando el valor añadido del mismo, sin necesidad de realizar una evaluación experimental. Cabe destacar que dicha evaluación es un proceso invasivo, lento, y que tiene que ser realizado por personal especializado

Several factors influence consumers’ choices of food products. While price remains the main criterion, quality, pleasure, convenience, and health are also important driving factors in food market evolution. Food enterprises are making significant efforts to manufacture products that meet consumers’ demands without compromising on safety standards. Additionally, the food industry also aims to improve the efficiency of transformation and conservation processes by minimizing energy consumption, process duration, and waste generation. However, foods are highly complex systems in which: (i) non-linear dynamics and interactions among different temporal and spatial scales must be considered; (ii) a wide range of physical phenomena (such as evaporation, mechanical changes, thawing, energy/mass transport, and color changes) occur; (iii) different food matrices (such as meat, vegetables, cereal, milk, and juices) with different microstructures and properties are involved; and (iv) the number of quality and safety indicators (such as bacteria, total volatile basic nitrogen, color, texture, odor, and sensory characteristics) is substantial. Mathematical modeling and simulation are key elements that allow us to gain a deeper understanding of food processes and enable the use of tools such as optimization and real-time control to improve their efficiency. This special issue aims to gather research on the development of dynamic mathematical models that describe the relevant factors in food processes from the perspectives of food safety (chemical or microbiological), food quality (organoleptic or nutritional), or resource consumption. Additionally, the development of model-based tools to improve food processes is also considered. This includes decision-making and optimization tools, the characterization of uncertainty/variability in model predictions, model simulation techniques, software sensors, and software development. The contributions published in this Special Issue can be gr

This is a reprint of articles from the Special Issue published online in the open access journal Foods (ISSN 2304-8158)

The successful implementation of the Common Fisheries Policy (CFP) depends, at a large extent, on the capacity to quantify total catches on board commercial vessels Because of the large number of fishing vessels and the high number of trips to be monitored, classic monitoring methods mainly based on inspections, are not effective --> The use of electronic devices to quantify fishing catches is gaining relevance The data provided by such devices, in combination with mathematical models, may be used to assess the state of the different fishing stocks and to optimize the fishing activity Increasingly though, technology has quickly developed during the last years to provide vision and artificial intelligence based remote Electronic Monitoring (REM or EM systems) at lower costs, and with more potential to cover large areas than traditional monitoring strategies

[EN] Fish consumers demand high quality products. Quality determine the commercial value of fishing products. Therefore, theuse of rapid, reliable, non-destructive tools able to assess quality indicators in fish is becoming critical. In this regard, techniques based on hyperspectral images (HSI) are among the most promising in this field and their interest is increasing among the scientific community. Nevertheless, most of the contributions focus on processed fish, namely filets. Besides, HSI data are difficult to interpret, therefore, mathematical models able to provide estimations of fish quality from HSI data must be obtained. In this work, we derive a mathematical model, by combining HSI and partial least squares regression, to assess quality in unprocessed fish. In particular, we consider turbot from an aquaculture farm. Results show that the model is able to satisfactorily asses the QIM, one of the most used indicators, [ES] Los consumidores de pescado demandan productos de calidad, la cual determina su valor comercial. Por lo tanto, existe un gran interés en el desarrollo de herramientas capaces de evaluar la calidad del pescado de forma rápida, fiable, y queno suponga la destrucción del alimento. En este sentido, la tecnología de imágenes hiperespectrales (HSI, por sus siglas en inglés) es de las más prometedoras y está experimentando un gran auge entre la comunidad científica. Sin embargo, en lo que se refiere a pescado, la mayoría de artículos se centran en pescado procesado, principalmente filetes. Por otra parte, dado que los datos HSI no son sencillos de interpretar, se recurre a modelos matemáticos que proporcionen estimaciones de la calidad del pescado a partir de dichos datos. En este trabajo se propone un modelo matemático, obtenido mediante la combinación de HSI y regresión por mínimos cuadrados parciales, para la evaluación de calidad en pescado entero, concretamente rodaballo de acuicultura. Los resulados muestran que dicho modelo reproduce de forma satisfactoria el QIM, uno de los indicadores de calidad más utilizados

La capacidad que poseen las bacterias para adaptarse a su entorno y desarrollar resistencia a los compuestos antimicrobianos actualmente conocidos (antibióticos, desinfectantes y conservantes) plantea un serio problema, económico y de salud pública, el cual se agrava a pasos agigantados. Tan solo en el año 2019, informes oficiales de las Naciones Unidas cifran en 700 000 los fallecimientos que se pueden atribuir directamente a la infección provocada por algún tipo de bacteria resistente a los antibióticos. Por desgracia, algunos estudios recientes sugieren que estos datos pueden resultar incluso más desalentadores, rondando los 1.27 millones de fallecidos en este mismo año (Murray et al., 2022). El rápido desarrollo de resistencia a antimicrobianos que presentan algunas bacterias afecta ya, de manera particularmente negativa, a las industrias médica y alimentaria, entre muchas otras. Este problema, lejos de tener solución, parece avanzar mucho más rápidamente que nuestra capacidad para entender las causas e implicaciones, del mismo. Por lo tanto, se hace cada vez más necesario ampliar nuestro conocimiento sobre los mecanismos que conducen al origen, selección y transmisión de resistencia a antimicrobianos en poblaciones bacterianas, con el fin de diseñar terapias de desinfección eficaces para atajar este indeseable fenómeno

The fast increase in bacterial resistance to current antimicrobials Antimicrobial Resistance or AMR) poses a paramount economic and health concern that seems far from being solved. Recent studies [1] estimate 1,27 million the number of deaths worldwide that are directly attributable to AMR only in 2019. Understanding the main mechanisms involved in the emergence, selection, and spread of AMR is critical to determine the optimal therapies to deal with this undesirable phenomenon

Fish freshness can be considered as the combination of different nutritional and organoleptic attributes that rapidly deteriorate after fish capture, i.e., during processing (cutting, gutting, packaging), storage, transport, distribution, and retail. The rate at which this degradation occurs is affected by several stress variables such as temperature, water activity, or pH, among others. The food industry is aware that fish freshness is a key feature influencing consumers’ willingness to pay for the product. Therefore, tools that allow rapid and reliable assessment and prediction of the attributes related to freshness are gaining relevance. The main objective of this work is to provide a comprehensive review of the mathematical models used to describe and predict the changes in the key quality indicators in fresh fish and shellfish during storage. The work also briefly describes such indicators, discusses the most relevant stress factors affecting the quality of fresh fish, and presents a bibliometric analysis of the results obtained from a systematic literature search on the subject

In this manuscript, we summarize some of the works initiated, developed and supervised by Prof. Antonio A. Alonso in the context of process and bioprocess engineering. In particular, we focus on those works related to modeling, optimization and control in which the Thermodynamics played a key role, namely: (i) the thermal sterilization of food products using steam and super-heated water retorts; (ii) modeling, model reduction and control based on first and second laws of thermodynamics; and (iii) plant-wide control of large dimensional systems using the link among thermodynamics, passivity and Lyapunov theory

[EN] The quality and freshness of fishery products has traditionally been determined by sensory evaluation methods, as well as by various analytical and instrumental techniques. However, all of them have several limitations. The first ones are subjective, as they depend on the personal judgment of a panel, while the second ones are generally invasive and destructive, requiring the selection of samples from the production batch and being unable to be applied to a complete production in an online way. Thus, a continuous effort is being made to find efficient methods and tools that allow the determination of quality and freshness in an objective, fast, reliable and non-destructive approach, in order to satisfy the expectations of consumers. In this regard, hyperspectral imaging (HSI) technology is one of the most promising in this field since, combined with multivariate analysis, it can result in a fast, non-invasive and non-destructive technique for the evaluation of fish quality indicators. However, most of the existing studies focus on processed fish, namely filets. In this work, correlation models using partial least squares regression (PLSR) have been developed to predict various quality indicators (quality index (QI) and colorimetric variables L, a and b of the CIELAB color space) in aquaculture whole turbot samples. Satisfactory results have been obtained for the prediction of QI using hyperspectral information collected from the dark side of turbots. This model was developed for 32 samples and 10 components and efficiently estimates the QI value of 9 samples of the external evaluation set, resulting in a normalized error of 1.0212 and a MSE of 8.7929., [ES] La calidad y frescura de los productos de la pesca se ha determinado tradicionalmente mediante métodos de evaluación sensorial, así como con diversas técnicas analíticas e instrumentales. Sin embargo, todos ellos adolecen de diversas limitaciones. Los primeros son subjetivos, pues dependen de la opinión personal de un panel, en tanto que los segundos son, por lo general, invasivos y destructivos, por lo que requieren la selección de muestras de un mismo lote y son incapaces de aplicarse a una producción completa de forma online. Así pues, existe una continua búsqueda de métodos y herramientas eficientes que permitan la determinación de la calidad y frescura de manera objetiva, rápida, fiable y que no lleven asociada la destrucción del alimento, con el fin de alcanzar las expectativas de los consumidores. En este contexto, la tecnología de imágenes hiperespectrales (HSI) es de las más prometedoras ya que, combinada con el análisis multivariable, puede resultar en una técnica rápida, no invasiva y no destructiva para la evaluación de indicadores de calidad del pescado. Sin embargo, la mayoría de las investigaciones existentes se centran únicamente en pescado procesado, principalmente filetes. En este trabajo se han elaborado modelos de correlación mediante regresión por mínimos cuadrados parciales (PLSR) para predecir varios indicadores de calidad (índice de calidad (QI) y variables colorimétricas L, a y b del espacio de color CIELAB) en muestras de rodaballo entero de acuicultura. Se han obtenido resultados satisfactorios para la predicción del QI a partir de la información hiperespectral recogida de la cara oscura. Este modelo, elaborado para 32 muestras y 10 componentes, logra estimar de forma eficaz el valor del QI de 9 muestras pertenecientes al conjunto externo de evaluación, obteniéndose un error normalizado de 1,0212 y un MSEE de 8,7929, [GA] A calidade e frescura dos produtos pesqueiros determináronse tradicionalmente mediante métodos de avaliación sensorial, así como diversas técnicas analíticas e instrumentais. Non obstante, todos sofren varias limitacións. Os primeiros son subxectivos, xa que dependen da opinión persoal dun panel, mentres que os segundos son xeralmente invasivos e destrutivos, polo que requiren a selección de mostras dun mesmo lote e son incapaces de ser aplicados á produción completa en liña. Así pois, existe unha busca continua de métodos e ferramentas eficientes que permitan determinar a calidade e a frescura de forma obxectiva, rápida, fiable e que non leve asociada a destrución do alimento, co fin de atender ás expectativas dos consumidores. Neste contexto, a tecnoloxía de imaxes hiperespectrais (HSI) é unha das máis prometedoras xa que, combinada coa análise multivariante, pode dar como resultado unha técnica rápida, non invasiva e non destrutiva para a avaliación dos indicadores de calidade do peixe. Non obstante, a maior parte das investigacións existentes céntranse unicamente no peixe procesado, principalmente filetes. Neste traballo desenvolvéronse modelos de correlación mediante regresión de mínimos cadrados parciais (PLSR) para predicir varios indicadores de calidade (índice de calidade (QI) e variables colorimétricas L, a e b do espazo cromático CIELAB) en mostras de rodaballo enteiro de acuicultura. Obtivéronse resultados satisfactorios para a predición do QI a partir da información hiperespectral recollida do lado escuro. Este modelo, desenvolvido para 32 mostras e 10 compoñentes, consegue estimar de forma eficiente o valor QI de 9 mostras pertencentes ao conxunto externo de avaliación, obténdose un erro normalizado de 1,0212 e un MSE de 8,7929

The development and effective compliance of efficient fishing policies that guarantee both the sustainability of marine resources and fishing activity is one of the main challenges that policymakers nowadays face. At EU level, successful implementation of the Common Fisheries Policy (CFP) depends, at a large extent, on the capacity to quantify catches on board commercial vessels. Because of the large number of fishing vessels and the high number of trips to be monitored classic, monitoring methods, mainly based on inspections, are not effective. Therefore, the use of electronic devices to quantify fishing catches is gaining relevance. The data provided by such devices, in combination with mathematical models, may be used to assess the state of the different fishing stocks and to optimize the fishing activity. In this work, we consider different algorithms based on Deep Learning (DL) for species identification and length estimation. On the one hand, for the instance segmentation task, we have adapted the Mask R-CNN algorithm to the problem of fish species identification. On the other hand, the MobileNet-V1 convolutional neural network is used for the estimation of the length of each individual. The results show that, when overlapping among individuals is moderate to low, both the identification and length estimation models are able to satisfactorily quantify the catch. In situations where overlapping among individuals is large, results need further improvements

Thermal processing of canned products is subject to variability among the different items in a batch, which results in differences in terms of food quality and safety. Such variability is mainly caused by heterogeneity in the thermal processing unit as well as differences in the quantity of food, its compaction, and the quantity of packing liquid from item to item. Besides, for a proper design of the process operation strategies, different conflicting objectives, such as maximization of product quality and uniformity or minimization of processing time and energy consumption, must be considered. In this paper, we present a suitable formulation to efficiently address the multi-objective optimization problem of thermal processing of canned food under the presence of product variability. Such variability has been characterized from experimental data provided by an industrial cannery. Our formulation exploits the monotonic and convex behavior of the quality/safety indicators and the product kinetics with respect to the uncertain thermal coefficients, avoiding thus the computationally-expensive stochastic approach via Monte-Carlo simulations. In particular, customizable retort temperature profiles are considered as a beneficial alternative to the standard constant retort temperature practice

Industrial canneries are subject to perturbations that may compromise food safety requirements. In such cases, plant operators typically increase the processing time, leading to undesirable large processing cycles and excessive quality degradation. In addition, differences among the items in a batch lead to variability in terms of quality and safety which, if not explicitly considered in the processing strategy, forces the use of conservative operation policies. In this work, we present an event-based dynamic optimization approach that combines available plant measurements and mathematical model predictions to anticipate the effect of plant perturbations on food safety. A safety software sensor is build upon an on-line predictive simulation and a previous food-variability characterization such that, if any perturbation during the sterilization compromises food safety, a new processing strategy that optimizes a trade off among quality, uniformity and processing time is recomputed and implemented. Such multi-objective dynamic optimization problem under food product variability is efficiently addressed by taking advantage of the monotonicity and convexity properties of the food quality/safety dynamics

Monitoring plays a key role in all aspects of fsheries management, including those related to sustainable management of resources, the economic performance of the fshery, and the distribution of benefts from the exploitation of the fshery and environment. In this work, software improvements made on the remote electronic monitoring (REM) device iObserver are described towards the improvement of fsheries monitoring by precisely identifying and quantifying fshing catches on board commercial vessel´s. To this aim, we exploit deep learning and convolutional neural networks (CNNs) capabilities and potential., Peer Reviewed

Sustainability is a basic premise for the economic and social future of fisheries and the main objective of fishing policies, such as the Common Fisheries Policy of the European Union. An immense challenge faced by sustainable fisheries management and policies is that of finding cost-effective monitoring methods. Several remote electronic monitoring devices (REM) systems are available to that purpose; however, many of them exhibit a number of drawbacks that prevent its generalization among fleets (e.g. off-line evaluation of catches by the so-called dry observers in land, crew interferences and distrust, etc.). To overcome these drawbacks, robust and reliable innovative technologies for registering captures are required. In this regard, we have developed the iObserver, an electronic device for automatic identification and quantification of the whole catch on board fishing vessels. The iObserver is installed in the fishing park, over the conveyor belt, just before the fishing separation zone. The system takes images of everything that crosses this conveyor belt during the separation process. The recognition software automatically analyzes every image, identifies all the individuals, estimates their length and generates a report containing the results. The Deep Learning algorithms for species recognition algorithms included in the iObserver use Convolutional Neural Networks. Two models were created with transfer learning and data augmentation techniques: an instance segmentation model for specimen detection and classification and a regression model for fish length estimation. To train and test both models, a set of about 6,000 images were labeled with mask, species and length of each individual for a selection of 14 of the most relevant species for the local fisheries. Preliminary results were promising: precision and recall of 98% and 95% respectively for species recognition and 3.2% mean absolute percentage error for length estimation in the test data set. Species i

The main control measures used in EU fisheries are the use of logbooks, monitoring of vessel geographic positions and the inspections of the vessels at sea (patrol vessel surveillance) and at the ports (inspection of the landings). Many studies indicate that this traditional control measures are not effective within the current Landing Obligation rules. The Remote Electronic Monitoring (REM) is currently considered as one of the best future alternatives for the control of fishing activity. In fact, many fleets have already incorporated this technology into their activity to respond to the requirements of regional fishing organizations and control authorities in many countries of the world. The aim of this work was to test the implementation of innovative technologies based on artificial vision devices for catch composition determination and data management technologies installed on board. A study case including onboard observer trials and an automatic device was carried out to characterize discards and record unwanted species occurrence in commercial vessels. The “iObserver system” was implemented to improve the quality and availability of data and consequently to deepen knowledge on the status of the fisheries resources. This technology is aimed to be able to identify and quantify the catch (targeted and discarded) without interfering with the activity of fishermen. Once the data (species and biomass estimation) is acquired by iObserver, information is pre-processed and transmitted to land (to management servers). This system will allow real-time decision making for the fishing activity in order to eventually perform a more selective fishing. Automatic estimates of discards by species allow to take real-time decisions, avoid areas/times with high discards rates and potentially to comply with landing obligation

In this work, a co-simulation approach has been used to simulate a sterilization plant corresponding to a canned fish factory. This study is part of the project Integrated plant-wide Control a nd Optimization for Industry4.0 (InCO4In) and has the objective of integrating modelling, optimization and optimal control, to apply the results to the process industry. In this context, it is desired to optimize the operation of the cited sterilization plant

Monitoring plays a key role in all aspects of fsheries management, including those related to sustainable management of resources, the economic performance of the fshery, and the distribution of benefts from the exploitation of the fshery and environment. In this work, software improvements made on the remote electronic monitoring (REM) device iObserver are described towards the improvement of fsheries monitoring by precisely identifying and quantifying fshing catches on board commercial vessel´s. To this aim, we exploit deep learning and convolutional neural networks (CNNs) capabilities and potential

Quality and safety in the food industry, as well as the processes involved in the production, packaging, and transport of food, may be optimised using predictive microbiology. This discipline models the behaviour of microorganisms and uses the socalled primary models to describe bacterial growth. One of the main challenges consists of explaining and simulating the growth when the number of initial bacteria (inoculum) is reduced because of the stochastic behaviour. There exist different alternatives to modelling such stochastic processes depending on: (1) the required level of detail for the solution; (2) the computational time; and (3) the possibility of estimating parameters from fast and inexpensive measurements. In this work, we compare the solutions, computational times and structural identifiability obtained using different alternatives: (1) microscopic scale considering individual-based models; and (2) mesoscopic scale considering the modified Fokker-Planck equation

This paper addresses the problem of closed-loop operation of scheduling, together with the interaction of control and scheduling, for a class of processes that appear very often in industry: those that combine continuous production lines with parallel batch units that share some resources. The paper presents a novel approach to this problem, including batching and a new type of precedence in the assignment problem. It also considers the effect of shared resources on the duration of the cycle time of the batch units. The approach is illustrated with a real-life example of a canned tuna factory

Monitoring plays a key role in all aspects of fisheries management, including those related to sustainable management of resources, the economic performance of the fishery, and the distribution of benefits from the exploitation of the fishery and environment. In this manuscript, an electronic device (the iObserver) is described, which aims to improve fisheries monitoring by identifying and quantifying fishing catches on board commercial vessels. This device is located over the conveyor belt in the fishing sorting area to automatically take pictures of the entire catch during fish separation. Each picture is analyzed using open source image recognition software to identify the number of individuals, the species and length of each individual based on skin descriptors (color, texture), and shape. The iObserver is equipped with a graphical and user-friendly interface. The information provided by the iObserver is sent to the RedBox software, where it is aggregated and augmented with vessel instrumentation data, such as location, velocity, and course. Then, the data are sent to a shore-based center to be used for different purposes, including the following: feeding mathematical models describing stock evolution; identifying those regions with a large presence of individuals below a Minimum Conservation Reference Size (MCRS); and supporting administrative decisions about a given fishing region

Smart active packaging is an innovative packaging system that combines the benefits of measuring, estimating or predicting different aspects of food quality or safety with the release of an active substance that extends product shelf life. Nevertheless, in its typical configuration, the active packaging and the smart packaging are not connected, and the information provided is not exploited to design the release of the active substance. In this work, we demonstrate how smart active packaging systems using predictive mathematical models allow the automatic optimisation of food packaging design and the prediction of the expected shelf life along the food chain. On the one hand, the system calculates the best design of the active packaging and the concentration of the active substance in the different layers that maximise food quality and safety. On the other hand, the model allows to calculate and update shelf life values along the food chain under unexpected changes in the storage conditions. Shelf life estimations and prediction will help distributors and sellers to adjust the product market prices. For example, prices can be lowered to avoid food losses when the product is close to its use-by date. Hake (Merluccius merluccius) represents an example of a highly relevant and perishable food that can be conserved using natural antimicrobials. Therefore, the case study selected to illustrate the proposed methodology consists of the smart active packaging of hake using carvacrol as the active substance (antimicrobial). Besides, different polymers are considered as possible active packaging materials. The Matlab™ codes required to perform the simulations of the models described in this work as well as the optimisations for packaging design are available at https://doi.org/10.5281/zenodo.3244153

Background: Freeze-drying or lyophilization is a dehydration process employed in high added-value food and biochemical goods. It helps to maintain product organoleptic and nutritional properties. The proper handling of the product temperature during the operation is critical to preserve quality and to reduce the process duration. Methods: Mathematical models are useful tools that can be used to design optimal policies that minimize production costs while keeping product quality. In this work, we derive an operational mathematical model to describe product quality and stability during the freeze-drying process. Model identification techniques are used to provide the model with predictive capabilities. Then, the model is used to design optimal control policies that minimize process time. Results and conclusion: Experimental measurements suggest splitting the process into two subsystems, product and chamber, to facilitate the calibration task. Both models are successfully validated using experimental data. Optimally designed control profiles are able to reduce the process duration by around 30% as compared with standard policies. The optimization task is introduced into a real time scheme to take into account unexpected process disturbances and model/plant mismatch. The implementation of the real time optimization scheme shows that this approach is able to compensate for such disturbances.

Current market trends and business globalization have steered industries towards large production volumes, complex alternative recipes and a larger product portfolio, thus leading to processes of increased complexity. Optimal production scheduling has a direct impact on the overall efficiencyof processing facilities and can lead to increased productivity and minimal costs. Although a plethora of works study the optimal production scheduling problem [1], academic developments have been mostly tested on generic but small problem instances. This is mainly due to the computational complexity of real industrial problems, which easily become intractable. This gap between theory and industrial practice has been identified by the scientific community [2] and some recent works have successfully investigated medium to large problem instances [3]. In this work different solution strategies are proposed for a large real-life food industry

In this work an operational mathematical model for the freeze-drying process is derived. The model describes the state variables related to product quality and stability. Computational issues associated to the presence of a moving front are approached by using the Landau transform. Unknown model parameters are estimated using experimental data. The model is used to design optimal operation policies that reduce the process duration ensuring product quality

A key requirement for the successful implementation of the Landing Obligation is the need to monitor and regulate unwanted catches at sea. This issue is particularly challenging because of the large number of vessels and trips that need to be monitored and the remoteness of vessels at sea. Several options exist in theory, ranging from patrol vessels to onboard observers and self-sampling. Increasingly though, technology is developing to provide remote Electronic Monitoring (EM) with cameras at lower costs. This chapter first provides an overall synthesis of the pro’s and con’s of several monitoring tools and technologies. Four EM technologies already trialled in EU fisheries are then summarised. We conclude that it is now possible to conduct reliable and cost-effective monitoring of unwanted catches at sea, especially if various options are used in combination. However, effective monitoring is a necessary condition for the successful implementation of the Landing Obligation but insufficient unless it is implemented with a high level of coverage and with the support of the fishing industry

Mathematical models, in particular, physics-based models, are essential tools to food product and process design, optimization and control. The success of mathematical models relies on their predictive capabilities. However, describing physical, chemical and biological changes in food processing requires the values of some, typically unknown, parameters. Therefore, parameter estimation from experimental data is critical to achieving desired model predictive properties. This work takes a new look into the parameter estimation (or identification) problem in food process modeling. First, we examine common pitfalls such as lack of identifiability and multimodality. Second, we present the theoretical background of a parameter identification protocol intended to deal with those challenges. And, to finish, we illustrate the performance of the proposed protocol with an example related to the thermal processing of packaged foods.

Sterilization process is aimed to inactivate potentially harmful microorganisms. To that purpose the packaged food is subject to a time/temperature profile. In the canning industry such profiles are chosen based on the experience of the operator. In the presence of perturbations, such as steam supply problems, operators must react and design new profiles which, in most of the cases, are too conservative and/or may lead to the batch rejection, either because of quality or safety reasons. In order to overcome this problem, we propose in this work a model based real time optimization (RTO) strategy. The model, which describes the different relevant aspects of the plant (retort/can temperature, color dynamics, microorganism lethality, energy consumption, etc.) is used to predict the behavior of the system. Plant measurements are taken periodically and, in the event of a perturbation, an ptimization procedure is run to compute a new time/temperature profile based on the past measurements

In this work we develop a model describing the evolution of the adenosine triphosphate (ATP) degradation products, typically used as early indicators of fish quality loss, during storage and transport conditions. The model is constructed following a modular approach that includes essentially three mechanisms: (1) enzymatic transformation of inosine 5′-monophosphate (IMP), inosine (Ino) and hypoxanthine (Hx) with some reactions catalyzed by bacteria; (2) bacterial growth and (3) nucleotide diffusion through the food matrix. This approach allows us to combine the different underlying mechanisms to account for other fish species and conditions. We compare alternative mechanisms explaining the catalytic effect of Pseudomonas and Shewanella populations on the reaction linking IMP, Ino and Hx. The selection is carried out in terms of the Akaike Information Criteria. The predictive capabilities of the selected model are demonstrated with experiments

Listeria monocytogenes is a food-borne pathogen that can persist in food processing plants by forming biofilms on abiotic surfaces. The benefits that bacteria can gain from living in a biofilm, i.e., protection from environmental factors and tolerance to biocides, have been linked to the biofilm structure. Different L. monocytogenes strains build biofilms with diverse structures, and the underlying mechanisms for that diversity are not yet fully known. This work combines quantitative image analysis, cell counts, nutrient uptake data and mathematical modeling to provide a mechanistic insight into the dynamics of the structure of biofilms formed by L. monocytogenes L1A1 (serotype 1/2a) strain. Confocal laser scanning microscopy (CLSM) and quantitative image analysis were used to characterize the structure of L1A1 biofilms throughout time. L1A1 forms flat, thick structures; damaged or dead cells start appearing early in deep layers of the biofilm and rapidly and massively loss biomass after 4 days. We proposed several reaction-diffusion models to explain the system dynamics. Model candidates describe biomass and nutrients evolution including mechanisms of growth and cell spreading, nutrients diffusion and uptake and biofilm decay. Data fitting was used to estimate unknown model parameters and to choose the most appropriate candidate model. Remarkably, standard reaction-diffusion models could not describe the biofilm dynamics. The selected model reveals that biofilm aging and glucose impaired uptake play a critical role in L1A1 L. monocytogenes biofilm life cycle.