Volume 23 (2018): Issue 2 (December 2018)

Production networks have been established in many industrial domains with globalized supply structures, sourcing strategies or cooperation environments. Knowledge management in such networks requires a context-oriented approach in order to accommodate for individual and organizational needs when providing relevant knowledge for complex tasks, such as value creation. The focus of this paper is on the procedure by which to determine what actually has to be taken into account as part of the ‘context’ for establishing context-oriented knowledge management, how to capture this context, and how to use it. The variability of organisational and individual tasks both at design time and at runtime is essential for understanding context in production networks. The main contribution of the present study is a context modelling method, including variability identification. The application of this method is demonstrated by using an example of a production network from automotive industries.

Data has evolved into a large-scale data as big data in the recent era. The analysis of big data involves determined attempts on previous data. As new era of data has spatiotemporal facts that involve the time and space factors, which make them distinct from traditional data. The big data with spatiotemporal aspects helps achieve more efficient results and, therefore, many different types of frameworks have been introduced in cooperate world. In the present research, a qualitative approach is used to present the framework classification in two categories: architecture and features. Frameworks have been compared on the basis of architectural characteristics and feature attributes as well. These two categories project a significant effect on the execution of spatiotemporal data in big data. Frameworks are able to solve the real-time problems in less time of cycle. This study presents spatiotemporal aspects in big data with reference to several dissimilar environments and frameworks.

Collaborative learning is a process that involves a group of peers collaborating with the aim to acquire new knowledge or skills. Collaborative learning environment enables such interactions by means of ICT. The paper focuses on affective collaborative learning environments, i.e., collaborative learning environments that are additionally aware of user’s emotions and moods. Based on the analysis of existing research, a general architecture of an affective collaborative learning environment has been proposed in the paper and the main challenges for developing such an environment have been identified, namely, nonintrusive and safe detection of user’s emotions, the adaptation of tutoring strategies, as well as modelling of artificial peers. This study can be considered the first step for the development of the collaborative learning environment that takes into account various affective aspects during the collaborative learning process.

Along with new artifact development, software engineering also includes other tasks. One of these tasks is the reverse engineering of binary artifacts. This task can be performed by using special “decompiler” software. In the present paper, the author performs a comparison of four different Java programming language decompilers that have been chosen based on both personal experience and results of a software developer survey.

Ad-hoc networks are a set of mobile nodes that are connected via a wireless channel. Some of the nodes in this network behave selfishly and do not send data to other nodes so that in order to increase network performance these nodes must be identified. A credit-based algorithm is proposed to detect the selfish nodes. Three watchdog nodes are selected to monitor suspicious nodes in each cluster. The cluster head nodes detect the existence of selfish nodes by controlling general features of network, such as delay, the total number of sent packets, the total number of received packets, throughput, and network traffic. The watchdog nodes send their comment on selfishness or cooperation of the node to the cluster head. Cluster head makes decisions with a majority vote on a suspicious node. The simulation results show that the rate of detection accuracy and the life time of network are considerably high and the false alarm rate and energy consumption are low comparing to that of similar methods.

A number of PCB defects, though having passed successfully the defect identification procedure, can potentially grow into critical defects under the influence of various external and (or) internal influences. The complex nature of the development of defects leading to PCB failures demands developing and updating the data measuring systems not only for detection but also for the prediction of future development of PCB defects considering the external influences. To solve this problem, it is necessary to analyse the models of defect development, which will allow predicting the defect growth and working out the mathematical models for their studies.

The study uses the methods of system analysis, theory of mathematical and imitation modelling, analysis of technological systems. The article presents four models for determining the theoretical stress concentration factor for several types of common defects, considering the strength loss of PCB elements. For each model the evaluation of parameters determining its quality is also given. The formulas are given that link the geometry of defects and the stress concentration factor, corresponding to four types of defects. These formulas are necessary for determining the number of cycles and time to failure, fatigue strength coefficient.

The chosen models for determining the values of the stress concentration factor can be used as a database for identifying PCB defects. The proposed models are used for software implementation of the optical image inspection systems.

The article shows the relevance of the application of discrete models of rod structures of radio-electronic means (REM) for the study of their behaviour under transient loading. A discrete model of the propagation of harmonic waves in the rod and the study of standing waves are proposed. Computational experiments using the proposed model are conducted. The results show that the model accurately reflects qualitative dynamics of the physical processes in the elastic rod while the waves of elastic deformations are passing through. The proposed models are used for software implementations of systems of mechanical simulation of the behaviour of rod structures.

A complex classification task as scene recognition is considered in the present research. Scene recognition tasks are successfully solved by the paradigm of transfer learning from pretrained convolutional neural networks, but a problem is that the eventual size of the network is huge despite a common scene recognition task has up to a few tens of scene categories. Thus, the goal is to ascertain possibility of a size reduction. The modelling recognition task is a small dataset of 4485 grayscale images broken into 15 image categories. The pretrained network is AlexNet dealing with much simpler image categories whose number is 1000, though. This network has two fully connected layers, which can be potentially reduced or deleted. A regular transfer learning network occupies about 202.6 MB performing at up to 92 % accuracy rate for the scene recognition. It is revealed that deleting the layers is not reasonable. The network size is reduced by setting a fewer number of filters in the 17^th and 20^th layers of the AlexNet-based networks using a dichotomy principle or similar. The best truncated network with 384 and 192 filters in those layers performs at 93.3 % accuracy rate, and its size is 21.63 MB.

The relevance and nature of a new technology for measurement of vibrational displacement of a material point through normal toward the object plane are stated in the article. This technology provides registration and processing of images of a round mark or a matrix of round marks, which are applied to the surface of a control object. A measuring signal here is the module of radius increment of the round mark image at vibrational blurring of this image. The method for calculation of the given error of measurements, as a function of a number of pixels of the round mark image, has been developed and proven in the present research. The results of pilot studies are given. Linearity of transformation of the measured size into a measuring signal has been proven. The conditions of a technical compromise between the field of view area of a recording device during distribution measurement of vibrational displacements along the surface of a control object, and the accuracy of this measurement are determined. The results are illustrated with numerical examples of calculations of the given error of measurements in the set field of view and the one at the given maximum set error of measurements.