The thesis presents usefulness of duality between graph and his adjacency matrix. The teoretical part provides the basis of graph theory and matrix algebra mainly focusing on sparse matrices and options of their presentation witch takes into account the number of nonzero elements in the matrix. The thesis includes presentation of possible operations on sparse matrices and algorithms that basically work on graphs, but with help of duality between graph and his adjacency matrix can be presented with sequence of operations on matrices.
Practical part presents implementation of some algorithms that can work both with graphs or their adjacency matrices in programming language Java and testing algorithms that work with matrices.
It focuses on comparison in efficiency of algorithm working with matrix written in standard mode and with matrix written in format for sparse matrices. It also studies witch presentation of matrices works beter for witch algorithm.

Graph algorithms in the language of linear algebra

The revolution in information technologies, and the spread of the Internet of Things (IoT) and smart city industrial systems, have fostered widespread use of smart systems. As a complex, 24/7 service, healthcare requires efficient and reliable follow-up on daily operations, service and resources. Cloud and edge computing are essential for smart and efficient healthcare systems in smart cities. Emergency departments (ED) are real-time systems with complex dynamic behavior, and they require tailored techniques to model, simulate and optimize system resources and service flow. ED issues are mainly due to resource shortage and resource assignment efficiency. In this paper, we propose a resource preservation net (RPN) framework using Petri net, integrated with custom cloud and edge computing suitable for ED systems. The proposed framework is designed to model non-consumable resources and is theoretically described and validated. RPN is applicable to a real-life scenario where key performance indicators such as patient length of stay (LoS), resource utilization rate and average patient waiting time are modeled and optimized. As the system must be reliable, efficient and secure, the use of cloud and edge computing is critical. The proposed framework is simulated, which highlights significant improvements in LoS, resource utilization and patient waiting time.

An Edge Computing Based Smart Healthcare Framework for Resource Management.

Rapid Prototyping technologies, especially 3D printing are becoming increasingly popular due to their usability and the constant decrease in price of printing equipment and materials. The article focuses on the study of selected mechanical strength properties of 3D-printed elements, which are not very important if the element is only a model for further manufacturing techniques, but which are important when 3D-printed elements will be a part of a functioning device, e.g. a part of unique scientific equipment. The research was carried out on a set of standardised samples, printed with low-cost standard materials (ABS), using a cheap 3D printer. The influence of parameters (such as the type of infill pattern, infill density, shell thickness, printing temperature, the type of material) on selected mechanical properties of the samples, were tested. The obtained results allows making conscious decisions on the printing of elements to be durable enough, either on a non-professional printer, or when to ordered by a professional manufacturer.

https://iopscience.iop.org/article/10.1088/1757-899X/227/1/012033/pdf

The influence of printing parameters on selected mechanical properties of FDM/FFF 3D-printed parts

At present times, the real-time requirement on the multiaccess healthcare monitoring system, information mining, and efficient disease diagnosis of health conditions is a difficult process. The recent advances in information technology and the internet of medical things (IoMT) have fostered extensive utilization of the smart system. A complex, 24/7 healthcare service is needed for effective and trustworthy monitoring of patients on a daily basis. To accomplish this need, edge computing and cloud platforms are highly required to satisfy the requirements of smart healthcare systems. This paper presents a new effective training scheme for the deep neural network (DNN), called ETS-DNN model in edge computing enabled IoMT system. The proposed ETS-DNN intends to facilitate timely data collection and processing to make timely decisions using the patterns that exist in the data. Initially, the IoMT devices sense the patient’s data and transfer the captured data to edge computing, which executes the ETS-DNN model to diagnose it. The proposed ETS-DNN model incorporates a Hybrid Modified Water Wave Optimization (HMWWO) technique to tune the parameters of the DNN structure, which comprises of several autoencoder layers cascaded to a softmax (SM) layer. The SM classification layer is placed at the end of the DNN to perform the classification task. The HMWWO algorithm integrates the MWWO technique with limited memory Broyden–Fletcher-Goldfarb-Shannon (L-BFGS). Once the ETS-DNN model generates the report in edge computing, then it will be sent to the cloud server, which is then forwarded to the healthcare professionals, hospital database, and concerned patients. The proposed ETS-DNN model intends to facilitate timely data collection and processing to identify the patterns exist in the data. An extensive set of experimental analysis takes place and the results are investigated under several aspects. The simulation outcome pointed out the superior characteristics of the ETS-DNN model over the compared methods.

/pdf/an-effective-training-scheme-for-deep-neural-network-in-edge-38ow5qsefm.pdf

An Effective Training Scheme for Deep Neural Network in Edge Computing Enabled Internet of Medical Things (IoMT) Systems

We consider the problem of finding cyclic schedules for a job shop in which all jobs are identical. Specifically, we assume that a single product is produced on a finite number of machines. Each part is manufactured by performing a given set of operations in a pre-determined sequence. Each operation can be performed on exactly one machine. 
 
Key structural properties of cyclic schedules are studied. A search algorithm based on these properties is proposed, and preliminary computational tests of the algorithm are reported.

/pdf/optimal-cyclic-schedules-for-job-shops-with-identical-jobs-2nf7syfz99.pdf

Optimal Cyclic Schedules for Job Shops with Identical Jobs

In the process of multiassortment repetitive production planning the designation of operating system parameters for the steady state (in which after the last operation the return to the first operation in the sequence of productive resources occurs) is required. The length of steady state of the system is determined by the work of the critical resource. Determination of resources sequence for the steady state gives possibility to determine the processing times for processes in the system without taking into account the transitional phases. During the execution of this type of production, in addition to the phase in which production takes place in the steady state, transition phases connected with starting (start-up phase) and finishing work in the system (cease phase), and the phases associated with the changes in ongoing sequences in steady states can be distinguished. The article presents the problem of estimating the duration of the start-up phase and the cease phase in the concurrent multiassortment production systems, in which access to the resources is regulated by the local dispatching rules. A method and a procedure of creating a schedule for the transition phases using the integration between the systems of production planning and scheduling is presented. The demonstrated approach is illustrated with the examples using KbRS and SWZ systems.

Integration production planning and scheduling systems for determination of transitional phases in repetitive production

In this paper a solution of soft real time scheduling in manufacturing systems is presented. The basic requirements of scheduling as a real time system are discussed. The proposed rescheduling method uses predictive-reactive strategy and multi thread searching approach with rule-based heuristics, meta-heuristics and random modules.

Predictive - Reactive Strategy for Real Time Scheduling of Manufacturing Systems

In this paper the hybrid method of manufacturing knowledge representation in the CAPP knowledge based system is presented. In the presented solution the manufacturing knowledge as well as the product design is represented by means of the object method. This paper shows only results obtained at the object design stage (OOD). Currently the scope of decision problems solved by proposed system is limited to axi---symmetric products, but basis on the achievements to date it is thought to develop this system in order to widen the subjected products group. Moreover the detailed description of the inference engine of the CAPP system is given. In the inference engine the two groups of rules were distinguished. In the paper content the examples for each group of rules are shown.

The hybrid method of knowledge representation in a CAPP knowledge based system

Fast growth of the SME sector, increasing demands of customers and dynamic market force the producers to lower production costs and to use new IT tools in production. The paper presents method of integration of preparation of production and production planning systems, such as SWZ and PROEDIMS. The system constituted in this process will enable SME entrepreneurs to take correct decisions connected with planning and controlling production. The integration will be achieved by methods of data transformation and data mapping using XML language. Due to this integration PROEDIMS system will be enriched with the module supporting verification of production orders, using constraints satisfaction and depth-first search (DFS) algorithm with backtracking.

Production Preparation and Order Verification Systems Integration Using Method Based on Data Transformation and Data Mapping

The method for determining dispatching rules for transient phases, based on designation of a local dispatching rule for steady state, which determines the rest of its components has been presented in the paper. The algorithm for determining the transient rules has been presented and dependencies resulting from the presented sufficient conditions have been formulated. For the proposed transitional-rules generation algorithm the simulation model has been prepared. Simulation experiments to verify the correctness and effectiveness of the presented method have been conducted. The experiments have been prepared for input data describing the production system and the production order. In the verification process Enterprise Dynamics computer simulation system has been used. The results were compared with the results obtained for transitional phases realized through the S0 state of the production system.

Damian Krenczyk

Papers

Integration production planning and scheduling systems for determination of transitional phases in repetitive production

Predictive - Reactive Strategy for Real Time Scheduling of Manufacturing Systems

The hybrid method of knowledge representation in a CAPP knowledge based system

Production Preparation and Order Verification Systems Integration Using Method Based on Data Transformation and Data Mapping

Transient States of Cyclic Production Planning and Control