Категория: АВТОМАТИКА И ИНФОРМАТИКА 1, 2018 г.

Key Words: Anaerobic degradation; biogas; pilot bioreactor; software sensors; LabVIEW; Beckhoff controller; system for monitoring and control.  

Abstract. In this paper the pilot biogas plant of the Stephan Angeloff Institute of microbiology, BAS is presented. This pilot biogas plant includes computer system for monitoring and control and was designed for development and scale-up of different biotechnologies for anaerobic degradation of organic wastes. Two options of the computer system are developed – research option based on PC and LabVIEW and industrial option based on PLC controller of Beckhoff. Data from sensors is stored and visualized on PC and is used for calculation of unmeasured technological parameters (via software sensors) and for optimal values of control algorithms parameters determination. The developed monitoring system includes software sensors for calculation of specific growth rates of two microbial populations and their biomass concentrations. Bang-bang algorithms are tested for bioreactor temperature regulation (with the controller of Beckhoff) and regulation of the biogas flow-rate (both with LabVIEW and the controller of Beckhoff). The obtained results are acceptable for the temperature regulation and poor for the regulation of the biogas flow-rate.

Повече

Key Words: Food quality; sensory analysis; e-nose; SVM classification; chicken meat.

Abstract. This paper discusses some more important aspects related to sensory characteristics, in particular the implementation of an „electronic nose“ system. The aim of the study is to investigate the possibilities for assessing the quality and authenticity of food products, in particular fresh chicken meat. By using multisensory  data fusion, an attempt has been made to overcome some of the difficulties and shortcomings inherent in organoleptic assessments. For the purposes of classification into classes used one-class SVM classifier. The procedure has been successfully tested to increase the accuracy of the classification by selecting the most informative sensors. The results show the potential of the proposed classifier that could be used as a quick, objective and non-destructive tool for assessing the quality of real-world recognition systems.

Повече

Key Words: Risk including; copper converter; predictive maintenance; case-based reasoning; rule–based reasoning; decision support.

Abstract. The purpose of the investigation is increasing the Remaining Useful Life (RUL) of Peirce-Smith (PS) copper converters on the base of risk assessment. During the operation, on the time of the last third of the cycles in the convertor campaign, there is a danger of a breakthrough in the tuyeres. An approach for the current individual risk assessment of each one tuyere is adopted and on this basis the real risk assessment for the tuyeres line is determined. A two-step predictive maintenance procedure is proposed based on the value of the current breakthrough risk in the tuyeres. When risk assessment is under 80%, the main task is to align the profile of the tuyeres line, and supportive actions are limited to determining which tuyeres should be blocked or unblocked. When the risk rises above 80%, then move to another strategy, which takes into account four risk components: breakthrough in the tuyeres, breakthrough in the body of the converter, non-conventional composition of the black copper, and failure of the intended yield. Decision making for predictive maintenance in this case is supported by Case Based Reasoning and Rule Based Reasoning systems and technological actions and maintenance actions are proposed. 

Повече

Кey Words: Logs; modeling; freezing; free water; icing degree; latent thermal energy.

Abstract. An approach for the computation of the specific (for 1 m³ wood) latent thermal energy of the free water, Q_Lheat-fw, in logs subjected to freezing has been suggested. The approach takes into account to a maximum degree the physics of the freezing processes of the free water in wood. It reflects the influence on the mentioned energy of the wood density, the icing degree formed by the freezing of free water in the logs, as well as the influence of the fiber saturation point of each wood species. Mathematical description of the specific thermal energy Q_Lheat-fw, which is released in logs during the free water freezing in the range from 0 ^oC to –1 ^oC, has been executed. This description is introduced in own 2D non-linear mathematical model of the freezing process of logs. For the solution and verification of the model and for the computation of the energy Q_Lheat-fw, a software program based on the suggested approach and mathematical description was prepared in FORTRAN, which was input into the calculation environment of Visual Fortran Professional. With the help of the program computations have been carried out for determination of the energy Q_Lheat-fw of two beech and two poplar logs with a diameter of 0.24 m, length of 0.48 m, and moisture content above the hygroscopic range during their many hours freezing in a freezer at approximately –30 ^oC. The information about Q_Lheat-fw is needed for scientifically based computing the energy consumption of the freezing and also of the subsequent defrosting processes of logs aimed at their plasticizing in the production of veneer.

Повече

Key Words: Convolutional neural networks; artificial neural networks; machine learning; object classification and recognition; computer vision.

Abstract. The field of machine learning has undergone rapid development with the rise of artificial neural networks (ANNs), over the past years. Some of the recently gained popularity models of the ANN are the so-called convolutional neural networks (CNNs). Impressive results in image recognition and object detection are achieved by the latest generation of CNN’s architectures, which unravel the significant interest in them from various professional communities. This paper presents the structure and basic principles of functioning and training of CNNs. The latest results in the field of development and application of such models have been discussed. The presentation has an informal, intuitive character and implies that the reader is familiar with the basics of machine learning and artificial neural networks.

Повече

Key Words: Internet of Things (IoT); big ocean data; machine learning; linear regression; SVM; Apache Spark; result analysis.

Abstract. The scientific paper has presented the various methods for collecting ocean data in Internet of Things Ecosystem. Most of the big ocean data is associated with sea surface temperature, water flows, air mass movement and their ocean-atmosphere interaction, sea level, sea-ice concentration, ocean topography and their impact on meteorological conditions. All these features of ocean data are of great importance and impact on climate change and its impact on human life. This paper is proposed a method for big data analytics and knowledge discovery of ocean water masses based on machine learning. The experimental framework is based on the Apache Spark environment and uses a PYTHON programming language optimized for big data processing. The experimental investigations have been performed using machine learning algorithms: linear regression and supporting vector machines. The paper has been presented the obtained results and their analysis.

Повече

Key Words: Energy management systems; Internet of Things; optimization; ISO 50001; metallurgy plant.

Abstract. Energy management systems (EMS) are complete solutions for optimization of energy consumption and energy processes in enterprises. They encompass specialized hardware and software components and services directed towards monitoring, measurement and management of energy consumption. Тhe advanced Industrial Internet of Things (IIoT) paradigm may be successfully used to improve the functionality and quality of EMS ensuring reliable data collection and sharing, ubiquitous computing, and computing clouds using powerful resources to solve a variety of decision-making and scheduling tasks that abound in the system. The basic requirements for advanced energy management systems based on the ISO 50001 standard are analyzed. The architecture and functionality of currently used energy management system for non-ferrous metallurgy plants are presented and the weaknesses of this system are analyzed. An improved framework of the energy management system based on the concept and technologies of the Industrial Internet of Things is proposed and discussed.

Повече