Tanya Titova | University of Food Technologies - Plovdiv (original) (raw)

Papers by Tanya Titova

2022 International Conference on Communications, Information, Electronic and Energy Systems (CIEES)

TEM Journal

Modelling of sensor signals is a key element in implementing the procedures for estimation and tr... more Modelling of sensor signals is a key element in implementing the procedures for estimation and tracking, sensor data fusion, fault detection and diagnosis, etc. In many cases, using traditional approaches to solve this problem is impossible due to lack of prior information about the observed process or particular sensor characteristics. Starting only from the finite rate of change of the signal at the output of real sensors, the Weierstrass theorem assumes the existence of a polynomial that approximates the signal, but does not specify how to find it. The present work attempts to solve this problem by assuming that the approximating polynomial has to retain that information in the signal which would allow its Bayesian estimation. The described approach has been successfully applied to model a humidity sensor signal, but it can be used in other cases without any problem.

2022 International Conference on Communications, Information, Electronic and Energy Systems (CIEES)

Purpose of this exercise is to introduce the device , principle of operation and testing equipmen... more Purpose of this exercise is to introduce the device , principle of operation and testing equipment to measure hydrostatic pressure.

Compressed air is a reliable source of energy that is widely used in industry and service sectors... more Compressed air is a reliable source of energy that is widely used in industry and service sectors thanks to the ease and safety of production and processing. In most industrial facilities, compressed air is needed for many aspects of their operations. The nature of the respective production (industry), its scale and the structure of the technological equipment, the life cycle stage, make the task of optimizing pneumatic systems multivariate. In view of this, it is difficult to find a single solution for improving the energy efficiency of using compressed air. The present work aims at systematizing the main factors, the degree of their influence and the possibilities of saving of pneumatic energy. The analysis and evaluation were performed on the basis of quantitative indicators from leading industrial companies in Bulgaria. Keywords – pneumatic systems optimization, estimate energy saving, reducing compressed air use.

IOP Conference Series: Materials Science and Engineering, 2020

By eliminating leaks in compressed air systems in manufacturing plants, up to 50% of energy can b... more By eliminating leaks in compressed air systems in manufacturing plants, up to 50% of energy can be saved. With a well-designed inspection and maintenance plan, eliminating the majority of leaks can be a routine and effective practice. This paper describes the results of a pneumatic leakage audits in manufacturing facilities from the bottling industry in Bulgaria. The results are analyzed and are defined major groups elements of the pneumatic system, which can be potential sources of leakage leading to significant losses of compressed air and determining energy efficiency in the production.

International Journal of Reasoning-based Intelligent Systems, 2015

Hybrid neuro-genetic networks are a subclass of neural networks combining random-search methods w... more Hybrid neuro-genetic networks are a subclass of neural networks combining random-search methods with adaptive optimisation with the direct analogy of natural selection and genetics in biological systems. This paper tries to improve the efficiency of automated classifiers in the systems for automated quality determination and sorting via hybrid structures.

The paper presents a self-learning algorithm for a qualityrecognition (diagnostics) Bayesian clas... more The paper presents a self-learning algorithm for a qualityrecognition (diagnostics) Bayesian classifier used in the AQS 602 automatic sorting system. The algorithm was created on the basis of a method for extraction of information from data in a regime of online classification and sorting of products. The new algorithm produces promising results providing the possibility for increasing the efficiency of the system which are likely to be of help for work with sorting equipment.

2022 57th International Scientific Conference on Information, Communication and Energy Systems and Technologies (ICEST)

The purpose of developing a graphical interface for support students in mastering course "Op... more The purpose of developing a graphical interface for support students in mastering course "Optimization Methods".

In this paper is proposed self-organizing artificial neural network for multifactorial classifica... more In this paper is proposed self-organizing artificial neural network for multifactorial classification of tea type "English breakfast" in five different commercial brands. Experimental study of tea samples is based on spectral analysis in the visible and near infrared region of the electromagnetic spectrum of light (VIS/NIR spectral analysis). In parallel with the spectral characteristics were measured and the characteristics of color and pH of the tea samples. Experimental data were analyzed by the method stepwise Linear Discriminant Analysis in order to reduce the dimensionality of the output feature space and to extract the most valuable of informative perspective characteristics of tea samples. Practical applications The paper has implemented self-organizing artificial neural network trained to recognize "English breakfast" tea from five different brands. Tea samples were classified successfully in five cluster groups, achieved precision work of the neural net...

Proceedings of the International Conference on Computer Systems and Technologies and Workshop for PhD Students in Computing - CompSysTech '09, 2009

The paper presents the results of a research on the possibilities of improving the automatic clas... more The paper presents the results of a research on the possibilities of improving the automatic classifiers of ASM 501, Bulgarian sorting machines for fruit and vegetables. The research was carried out along the lines of eliminating heuristic approaches in forming the classifier feature space through strict mathematical proofs of the synthesis procedures using the Discrete Wavelet Transform (DWT). The paper

This paper concerns the use of evolutionary algorithms for improving the support vector machine (... more This paper concerns the use of evolutionary algorithms for improving the support vector machine (SVM) classification. Method genetic algorithms and genetic programming in the design of an SVM in learning stages: feature selection and kernel function selection are summarized. Experimental results of the SVM method application and adaptation to the recognition of unidimensional spectral realization of potatoes have been presented. In this practical non-destructive quality estimation task, the purpose of this work was to achieve maximum objectivity and accuracy by reducing the heuristic approach to feature selection and kernel SVM classifier synthesis.

Abstract. The paper analyzes the possibilities to non-destructively determine food quality (potat... more Abstract. The paper analyzes the possibilities to non-destructively determine food quality (potatoes, eggs) by means of the spectra of transmission in the visible and near-infrared regions of the electromagnetic spectrum. The research includes the creation and testing of a training sample of representative samples and the evaluation of the possibilities for classification using Neural Classifier and Support Vector Machines method (SVM). Key words: non-destructive quality evaluation, pattern recognition, food quality, classifiers

"Electronic tongue" (e-tongue) is instrumental system are designed to crudely mimic human taste s... more "Electronic tongue" (e-tongue) is instrumental system are designed to crudely mimic human taste sensory organs and are composed of an array of sensors. Complex data sets from "e-tongue" signals combined with multivariate statistics represent rapid and efficient tools for classification, recognition and identification of samples, also for the prediction of concentrations of different compounds. A wide variety of sensors can be employed into the design of these instrumental systems, especially that of "e-tongues", offering numerous practical applications. In this study are review, characteristics of sensors and possibilities "e-tongue" applications in the food industry. Practical applications: The "e-tongue" can be used in various applications, including on quality control in the food industry and pharmacy.

In this paper are considered basic methods for research the properties of multicomponent mixtures... more In this paper are considered basic methods for research the properties of multicomponent mixtures based multifactor statistical regression models.

Wavelet neural networks are a subclass of neural networks which combine wavelet transform methods... more Wavelet neural networks are a subclass of neural networks which combine wavelet transform methods. The present research tries to enhance the efficiency of classifiers in automated quality determination and sorting systems based on images that have been obtained via spectroscopy in the visual and near-infrared regions (VIS-NIR) of the electromagnetic spectrum. The problem at hand has to do with non-destructive egg freshness determination through a transmitted light flow, which is one of the major indicators of egg quality. The research has the following aspects: synthesis and training of wavelet neural networks for different wavelet functions and nondestructive egg freshness determination problems. Three types of wavelet neural networks have been studied, with Morlet, Gaussian and Mexican Hat wavelets as transfer functions in the wavelet neural layer.

This paper looks at some of the most important aspects related to sensory characteristics and exa... more This paper looks at some of the most important aspects related to sensory characteristics and examples of applications of color characteristics to define the quality of food products. The purpose of the study is exploring the possibilities of combining data from different sensors in order to increase the accuracy of classification of food products. For the assessment of quality there is used probabilistic neural networks. The procedure has been successfully tested to increase the accuracy in data experiments for quality classification citrus juices. The results show the potential of the proposed type of classifiers to be used as a rapid, objective and non-destructive tool for quality assessment on real recognition systems in the near future.

Faults are unwanted events in any industrial production system. Early detection and diagnosis of ... more Faults are unwanted events in any industrial production system. Early detection and diagnosis of faults in automated systems is important in order to prevent equipment damage, loss of performance and profits. For this purpose, more and more sophisticated and complex systems for observation and monitoring of basic characteristics in automated processes are being built. Preconditions for increasing their efficiency are processing and analysis of process information is obtained through a significant number of sensors. For pneumatic systems in addition to the identification of certain faults that may affect the normal production process, it is important to consider the possibilities to improve their energy efficiency. In this regards, the work focuses on the detection of leaks. The fault detection is based on the measurement of the compressed air consumption at the inlet of the pneumatic module and synchronization with signal from the PLC to the valve, and controlled the pneumatic cylin...

2022 International Conference on Communications, Information, Electronic and Energy Systems (CIEES)

TEM Journal

Modelling of sensor signals is a key element in implementing the procedures for estimation and tr... more Modelling of sensor signals is a key element in implementing the procedures for estimation and tracking, sensor data fusion, fault detection and diagnosis, etc. In many cases, using traditional approaches to solve this problem is impossible due to lack of prior information about the observed process or particular sensor characteristics. Starting only from the finite rate of change of the signal at the output of real sensors, the Weierstrass theorem assumes the existence of a polynomial that approximates the signal, but does not specify how to find it. The present work attempts to solve this problem by assuming that the approximating polynomial has to retain that information in the signal which would allow its Bayesian estimation. The described approach has been successfully applied to model a humidity sensor signal, but it can be used in other cases without any problem.

2022 International Conference on Communications, Information, Electronic and Energy Systems (CIEES)

Purpose of this exercise is to introduce the device , principle of operation and testing equipmen... more Purpose of this exercise is to introduce the device , principle of operation and testing equipment to measure hydrostatic pressure.

Compressed air is a reliable source of energy that is widely used in industry and service sectors... more Compressed air is a reliable source of energy that is widely used in industry and service sectors thanks to the ease and safety of production and processing. In most industrial facilities, compressed air is needed for many aspects of their operations. The nature of the respective production (industry), its scale and the structure of the technological equipment, the life cycle stage, make the task of optimizing pneumatic systems multivariate. In view of this, it is difficult to find a single solution for improving the energy efficiency of using compressed air. The present work aims at systematizing the main factors, the degree of their influence and the possibilities of saving of pneumatic energy. The analysis and evaluation were performed on the basis of quantitative indicators from leading industrial companies in Bulgaria. Keywords – pneumatic systems optimization, estimate energy saving, reducing compressed air use.

IOP Conference Series: Materials Science and Engineering, 2020

By eliminating leaks in compressed air systems in manufacturing plants, up to 50% of energy can b... more By eliminating leaks in compressed air systems in manufacturing plants, up to 50% of energy can be saved. With a well-designed inspection and maintenance plan, eliminating the majority of leaks can be a routine and effective practice. This paper describes the results of a pneumatic leakage audits in manufacturing facilities from the bottling industry in Bulgaria. The results are analyzed and are defined major groups elements of the pneumatic system, which can be potential sources of leakage leading to significant losses of compressed air and determining energy efficiency in the production.

International Journal of Reasoning-based Intelligent Systems, 2015

Hybrid neuro-genetic networks are a subclass of neural networks combining random-search methods w... more Hybrid neuro-genetic networks are a subclass of neural networks combining random-search methods with adaptive optimisation with the direct analogy of natural selection and genetics in biological systems. This paper tries to improve the efficiency of automated classifiers in the systems for automated quality determination and sorting via hybrid structures.

The paper presents a self-learning algorithm for a qualityrecognition (diagnostics) Bayesian clas... more The paper presents a self-learning algorithm for a qualityrecognition (diagnostics) Bayesian classifier used in the AQS 602 automatic sorting system. The algorithm was created on the basis of a method for extraction of information from data in a regime of online classification and sorting of products. The new algorithm produces promising results providing the possibility for increasing the efficiency of the system which are likely to be of help for work with sorting equipment.

2022 57th International Scientific Conference on Information, Communication and Energy Systems and Technologies (ICEST)

The purpose of developing a graphical interface for support students in mastering course "Op... more The purpose of developing a graphical interface for support students in mastering course "Optimization Methods".

In this paper is proposed self-organizing artificial neural network for multifactorial classifica... more In this paper is proposed self-organizing artificial neural network for multifactorial classification of tea type "English breakfast" in five different commercial brands. Experimental study of tea samples is based on spectral analysis in the visible and near infrared region of the electromagnetic spectrum of light (VIS/NIR spectral analysis). In parallel with the spectral characteristics were measured and the characteristics of color and pH of the tea samples. Experimental data were analyzed by the method stepwise Linear Discriminant Analysis in order to reduce the dimensionality of the output feature space and to extract the most valuable of informative perspective characteristics of tea samples. Practical applications The paper has implemented self-organizing artificial neural network trained to recognize "English breakfast" tea from five different brands. Tea samples were classified successfully in five cluster groups, achieved precision work of the neural net...

Proceedings of the International Conference on Computer Systems and Technologies and Workshop for PhD Students in Computing - CompSysTech '09, 2009

The paper presents the results of a research on the possibilities of improving the automatic clas... more The paper presents the results of a research on the possibilities of improving the automatic classifiers of ASM 501, Bulgarian sorting machines for fruit and vegetables. The research was carried out along the lines of eliminating heuristic approaches in forming the classifier feature space through strict mathematical proofs of the synthesis procedures using the Discrete Wavelet Transform (DWT). The paper

This paper concerns the use of evolutionary algorithms for improving the support vector machine (... more This paper concerns the use of evolutionary algorithms for improving the support vector machine (SVM) classification. Method genetic algorithms and genetic programming in the design of an SVM in learning stages: feature selection and kernel function selection are summarized. Experimental results of the SVM method application and adaptation to the recognition of unidimensional spectral realization of potatoes have been presented. In this practical non-destructive quality estimation task, the purpose of this work was to achieve maximum objectivity and accuracy by reducing the heuristic approach to feature selection and kernel SVM classifier synthesis.

Abstract. The paper analyzes the possibilities to non-destructively determine food quality (potat... more Abstract. The paper analyzes the possibilities to non-destructively determine food quality (potatoes, eggs) by means of the spectra of transmission in the visible and near-infrared regions of the electromagnetic spectrum. The research includes the creation and testing of a training sample of representative samples and the evaluation of the possibilities for classification using Neural Classifier and Support Vector Machines method (SVM). Key words: non-destructive quality evaluation, pattern recognition, food quality, classifiers

"Electronic tongue" (e-tongue) is instrumental system are designed to crudely mimic human taste s... more "Electronic tongue" (e-tongue) is instrumental system are designed to crudely mimic human taste sensory organs and are composed of an array of sensors. Complex data sets from "e-tongue" signals combined with multivariate statistics represent rapid and efficient tools for classification, recognition and identification of samples, also for the prediction of concentrations of different compounds. A wide variety of sensors can be employed into the design of these instrumental systems, especially that of "e-tongues", offering numerous practical applications. In this study are review, characteristics of sensors and possibilities "e-tongue" applications in the food industry. Practical applications: The "e-tongue" can be used in various applications, including on quality control in the food industry and pharmacy.

In this paper are considered basic methods for research the properties of multicomponent mixtures... more In this paper are considered basic methods for research the properties of multicomponent mixtures based multifactor statistical regression models.

Wavelet neural networks are a subclass of neural networks which combine wavelet transform methods... more Wavelet neural networks are a subclass of neural networks which combine wavelet transform methods. The present research tries to enhance the efficiency of classifiers in automated quality determination and sorting systems based on images that have been obtained via spectroscopy in the visual and near-infrared regions (VIS-NIR) of the electromagnetic spectrum. The problem at hand has to do with non-destructive egg freshness determination through a transmitted light flow, which is one of the major indicators of egg quality. The research has the following aspects: synthesis and training of wavelet neural networks for different wavelet functions and nondestructive egg freshness determination problems. Three types of wavelet neural networks have been studied, with Morlet, Gaussian and Mexican Hat wavelets as transfer functions in the wavelet neural layer.

This paper looks at some of the most important aspects related to sensory characteristics and exa... more This paper looks at some of the most important aspects related to sensory characteristics and examples of applications of color characteristics to define the quality of food products. The purpose of the study is exploring the possibilities of combining data from different sensors in order to increase the accuracy of classification of food products. For the assessment of quality there is used probabilistic neural networks. The procedure has been successfully tested to increase the accuracy in data experiments for quality classification citrus juices. The results show the potential of the proposed type of classifiers to be used as a rapid, objective and non-destructive tool for quality assessment on real recognition systems in the near future.

Faults are unwanted events in any industrial production system. Early detection and diagnosis of ... more Faults are unwanted events in any industrial production system. Early detection and diagnosis of faults in automated systems is important in order to prevent equipment damage, loss of performance and profits. For this purpose, more and more sophisticated and complex systems for observation and monitoring of basic characteristics in automated processes are being built. Preconditions for increasing their efficiency are processing and analysis of process information is obtained through a significant number of sensors. For pneumatic systems in addition to the identification of certain faults that may affect the normal production process, it is important to consider the possibilities to improve their energy efficiency. In this regards, the work focuses on the detection of leaks. The fault detection is based on the measurement of the compressed air consumption at the inlet of the pneumatic module and synchronization with signal from the PLC to the valve, and controlled the pneumatic cylin...