— Anomaly detection refers to the problem of finding patterns in data that do not conform to expected behavior. It is very important to timely detect parameter anomalies in real-world running thermal power plant system, which is one of the most complex dynamical systems. Artificial neural networks are one of anomaly detection techniques. This paper presents the Elman recurrent neural network as method to solve the problem of parameter anomaly detection in selected sections of thermal power plant (steam superheaters and steam drum). Inputs for neural networks are some of the most important process variables of these sections. In addition to the implementation of this network for anomaly detection, the effect of key parameter change on anomaly detection results is also shown. Results confirm that recurrent neural network is good approach for anomaly detection problem, especially in real-time industrial applications.

Anomalies are integral part of every system's behavior and sometimes cannot be avoided. Therefore it is very important to timely detect such anomalies in real-world running power plant system. Artificial neural networks are one of anomaly detection techniques. This paper gives a type of neural network (probabilistic) to solve the problem of anomaly detection in selected sections of thermal power plant. Selected sections are steam superheaters and steam drum. Inputs for neural networks are some of the most important process variables of these sections. It is noteworthy that all of the inputs are observable in the real system installed in thermal power plant, some of which represent normal behavior and some anomalies. In addition to the implementation of this network for anomaly detection, the effect of key parameter change on anomaly detection results is also shown. Results confirm that probabilistic neural network is excellent solution for anomaly detection problem, especially in real-time industrial applications.

Abstract Intelligent Transport Systems (ITS) fall in the framework of cyberphysical systems due to the interaction between physical systems (vehicles) and distributed information acquisition and dissemination infrastructure. With the accelerated development of wireless Vehicle-to-Vehicle (V2V) and Vehicle-to Infrastructure (V2I) communications, the integrated acquiring and processing of information is becoming feasible at an increasingly large scale. Accurate prediction of the traffic information in real time, such as the speed, flow, density has important applications in many areas of Intelligent Transport systems. It is a challenging problem due to the dynamic changes of the traffic states caused by many uncertain factors along a travelling route. In this paper we present a V2V based Speed Profile Prediction approach (V2VSPP) that was developed using neural network learning to predict the speed of selected agents based on the received signal strength values of communications between pairs of vehicles. The V2VSPP was trained and evaluated by using traffic data provided by the Australian Centre for Field Robotics. It contains vehicle state information, vehicle-to-vehicle communications and road maps with high temporal resolution for large numbers of interacting vehicles over a long time period. The experimental results show that the proposed approach (V2VSPP) has the capability of providing accurate predictions of speed profiles in multi-vehicle trajectories setup.

As an inspiration for robot behavior, it is possible to make analogies between behavior of biological organisms and robot models. Inspired by behavior of the bat, bee and dor, we designed reactive, sensor-based behavior for a mobile robot. In this paper, we will present the motor control results obtained through experiments, which confirm the effectiveness of the control based on behavior algorithm of living organisms. We demonstrate too, how the finite state machine approach may be applied in practical applications of biological based multi-robot teams.

The main task of coal producers is to provide for sufficient quantities of coal of required quality with minimum costs of excavation. Therefore, the prediction of energy values is the most important task aiming to secure the optimal usage of coal energy value. The goal of paper is to identify, examines and evaluate most influential artificial intelligence algorithms that have been widely used in the data mining community, on a real problem of predicting coal quality. Our researches are based on the data achieved under laboratory conditions during the period of five years (2005-2010), and include 33256 coal samples from ’’Kreka’’ Coal Mine Company. The goal of the research is to build, based on the described data, a prediction model that will be used for predicting the coal quality class of unknown samples. Four algorithms have been identified: C4.5, kNN, Naive Bayes and Multilayer Perceptron (MLP). The idea is to find the best model through the following stages: each of the algorithms is calibrated in order to find appropriate model division techniques which maximize the performance of the algorithms, to assess the importance of input attributes and ultimate comparison of the algorithm orders them with respect to their performance. The final evaluation of the outcomes allowed singling out the MLP to be the best predicting methods for the given domain with optimal structure for input, hidden and output layer.

This paper investigates the genetic based re-planning search strategy, using neural learned vibration behavior for achieving tolerance compensation of uncertainties in robotic assembly. The vibration behavior was created from complex robot assembly of cogged tube over multistage planetary speed. Complex extensive experimental investigations were conducted for the purpose of finding the optimum vibration solution for each planetary stage reducer in order to complete the assembly process in defined real-time. However, tuning those parameters through experimental discovering for improved performance is a time consuming process. Neural network based learning was used to generate wider scope of parameters in order to improve the robot behavior during each state of the assembly process. As a novel modelling formalism of reactive hybrid automata, we propose the Wormhole Model with both learning and re-planning capacities (WOMOLERE). For our application, the states of hybrid automaton include amplitudes and frequencies of robot vibration module. The transition action is a function of minimal distance and uncertainty effects due to jamming during the assembly process. The results suggest that the methodology is adequate and could be recognized as an idea for designing of robot surgery assistance methods, especially in soft-robotics.

As an inspiration for robot behavior, it is possible to make analogies between behavior of biological organisms and robot models. Inspired by the behavior of the beetles, we designed reactive, sensor-based behavior for mobile robot. The focus is on intelligent control algorithm, which approximates the behavior of the dor beetle with a Finite State Machine-based Design method. This finite state machine–based approach could be usefull as methodologie for the improved planning of the real-time complex tasks in robot-based manufactoring systems, using information from factory sensor networks, and taking into account the constraints from factory environments. In this paper, we will present the motor control results obtained through experiments, which confirm the effectiveness of the control based on behavior algorithm of living organisms. INTRODUCTION A biomimetic system is basically a system that mimics/copiess biology, in a way that it adopts mechanisms found in natural organisms and applies them in engineering [Na et al., 2009]. In recent years, the field of biomimetics made a great progress in related fields, including mechanical intelligence, materials science, biological science, understanding the wonders of nature and biology. These new findings have enabled scientists and engineers to apply the reverse engineering on the many functions of the animal world and to implement these technological capabilities [Argyros et. al., 2004]. Currently in robotics, a lot of research has been done in biomimetics. Biomimetic robots do not copy the complete behavior of the animals. Only the most useful abilities are extracted and modified so they can be used with the body structure of the robot and in this way the robot reacts more practically. The key features of biomimetic robots are their capacity to adapt to the environment and the ability to learn and react faster. A biomimetic robot also involves mechanisms and manipulator structures capable of meeting the requirements of enhanced performance [Vepa, 2009]. The biomimetic robots are designed to be substantially more compliant and stable than conventionally controlled robots and they use the advantage of new developments in materials, microsystems technology and also the developments that have led to a deeper understanding of biological behavior. For small, insect-like robots, applications include autonomous or semi-autonomous tasks such as reconnaissance and de-mining, and for human interaction tasks at a larger scale.

This paper presents a visual/motor behavior learning approach, based on neural networks. We propose Behavior Chain Model (BCM) in order to create a way of behavior learning. Our behavior-based system evolution task is a mobile robot detecting a target and driving/acting towards it. First, the mapping relations between the image feature domain of the object and the robot action domain are derived. Second, a multilayer neural network for offline learning of the mapping relations is used. This learning structure through neural network training process represents a connection between the visual perceptions and motor sequence of actions in order to grip a target. Last, using behavior learning through a noticed action chain, we can predict mobile robot behavior for a variety of similar tasks in similar environment. Prediction results suggest that the methodology is adequate and could be recognized as an idea for designing different mobile robot behaviour assistance.

This paper presents implementation of optimal search strategy (OSS) in verification of assembly process based on neural vibration learning. The application problem is the complex robot assembly of miniature parts in the example of mating the gears of one multistage planetary speed reducer. Assembly of tube over the planetary gears was noticed as the most difficult problem of overall assembly. The favourable influence of vibration and rotation movement on compensation of tolerance was also observed. With the proposed neural-network-based learning algorithm, it is possible to find extended scope of vibration state parameter. Using optimal search strategy based on minimal distance path between vibration parameter stage sets (amplitude and frequencies of robots gripe vibration) and recovery parameter algorithm, we can improve the robot assembly behaviour, that is, allow the fastest possible way of mating. We have verified by using simulation programs that search strategy is suitable for the situation of unexpected events due to uncertainties.

This paper presents a visual/motor behavior learning approach, based on neural networks. The behavior description is introduced in order to create behavior learning. Our behavior-based system evolution task is a mobile robot detecting a target and driving/acting towards it. First, the mapping relations between the image feature domain of the object and the robot action domain are derived. Second, a multilayer neural network for off-line learning of the mapping relations is used. This learning structure through neural network training process represents a connection between the visual perceptions and motor sequence of actions in order to gripe a target. Last, using behavior learning through a noticed action chain, we can predict mobile robot behavior for a variety of similar tasks in similar environment. Prediction results suggest the methodology is adequate and could be recognized as an idea for designing mobile robot assistance for blind people guiding or for performing manipulation tasks, as a helping hand to disabled people.

The applications of robot are very extended and have already become classic in different branches of mass industrial production such as welding, painting by spraying, antirust protection, etc. Though the operations performed by robots in these fields are very complex, the operations of assembly are even more complex. In fact, robot assembly operations involve the process of direct solving the conflicting situations being not within the classic repetitive work. Investigations treating typical assembly duties started forty years ago (Bohman, 1994). In the meantime, it was offered a series of control mechanism of mating date. Performing assemblies depends on sensation of and appropriate reaction to the forces of contact between mating components date (Wei, 2001). It is shown that with the intelligent techniques, example components can be assembled faster, gentle and more reliably. In order to create robot behaviours that are similarly intelligent, we seek inspiration from human strategies date (Chan, 1995). The working theory is that the human accomplishes an assembly in phases, with a defined behaviour and a subgoal in each phase. The human changes behaviours according to events that occur during the assembly and the behaviour is consistent between the events. The human’s strategy is similar to a discrete event system in that the human progresses through a series of behavioural states separated by recognizable physical events. In achieving acceptably fast robot behavior with assuring contact stability, many promising intelligent-control methods have been investigated in order to learn unstructured uncertainties in robot manipulators date (Chan, 1995), (Miyazaki et al., 1993), (Brignone et al., 2001). For example, (Newman et al., 2001) work describes intelligent mechanical assembly system. First phase for assembly is blind search. In this phase multiple parameters are assigned to rotational search attractor. If sensors register force values higher then thresholds, new parameters are assigned. Intelligent layer is represented on 22-dimensional space of trajectories, and based on blind search parameters (correct and incorrect) neural network is made. Correct assembly path is chosen by using form of Genetic algorithm search, so the new vectors are evolved from most successful “parents”. Using this process, the robot was allowed to generate and test its own program modifications. The primary source of difficulty in automated assembly is the uncertainty in the relative position of the parts being assembled (Vaaler, 1991). The crucial thing in robot assembly is how to enable a robot to accomplish a task successfully in spite of the inevitable uncertainties

Focus of this paper is on the development of the action-oriented education approach. Results from secondary data analysis and different empirical studies indicate that for the different levels of education (for example, tertiary practical education as well as operational professional training), enterprises and educational experts expect more competence extension in the future. The concept of action-oriented education which includes informing-process, planning, decision-making, evaluation, examination and achievement should lead the trainee to the development of their own key qualifications, predominantly their own methods and social competence to meet requirements. The concept of action-oriented education is explained by using learning purpose formulation related to the example of the assembly and programming of the microcontroller steered mobile robot. In this context, the role of the trainer is explained, who should be a moderator during his/her instructor's activity in order to increasingly advance the trainees learning and to qualify them in single handed project work and also in groups.

Benford's law gives expected patterns of the digits in numerical data. It can be used as a tool to detect outliers for example in as a test for the authenticity and reliability of transaction level accounting data. Based on Benford's law tests for first two digits, first three digits, last digits, last digit, last two digits have been derived as an additional analytical tool. Benford's law is known as a 'first digit law', 'digit analysis' or 'Benford-Newcomb phenomenon'. The second order test is an analysis of the digit frequencies of the differences between the ordered (ranked) values in a data set. The digit frequencies of these differences approximates the frequencies of Benford's law for most distributions of the original data. From some auditor's point of view it is very important if it is possible to use Benford's law to trace siginificant changes in data in some period of time or detect some trends within them.

This paper investigates the use of autonomous learning in the problems of complex robot assembly of miniature parts in the example of mating the gears of one multistage planetary speed reducer. Assembly of tube over the planetary gears was noticed as the most difficult problem of overall assembly and favourable influence of vibration and rotation movement on compensation of tolerance was also observed. With the proposed neural network based learning algorithm, it is possible to find extended scope of vibration state parameter. Using deterministic search strategy based on minimal distance path action between vibration parameter stage sets and recovery parameter algorithm, we can improve the robot assembly behaviour, i.e. allow the fastest possible way of mating.