1,721,003 research outputs found
Research and development of classification method of embryo cleavage stages. .
No full textThis paper represents an embryo cleavage-stage classification algorithm. There are used statistical feature extraction methods and two classification methods: Classification with training and classification without training. The main problem of this work is detection of early embryo cleavage stages. The aim is to adapt the proper classification method. The first part of this paper represents the analysis of the literature, and the methods used by other researchers examining similar issues. The second part of this research represents the proposed algorithm. There are introduced proposed methods. For the feature extraction proposed statistical methods: entropy, invariant moments and principal components analyses. For the classification are used neural networks and K-nearest neighbor method. The proposed method is checked by experiment. It is expected that this method will work well in video sequences. The Master's thesis consists of an introduction, three chapters, references and author of publications on the topic of the Master. General master's thesis consists of 70 pages, numbered 26 formulas, 103 pictures and 8 tables. References list includes sources
Development and research of respiratory masks classification method.
No full textThe classification method of respiratory mask is presented in master thesis. Presented method detects a mask on moving conveyer and recognize the type of respiratory mask. The state of the art methods are present in the first part of work. The second section discusses the application of these methods to the test system. The results of experimental investigation are given in the third part. SIFT and SURF features and three types of classifiers, i.e., Euclidian distance, K-NN nearest neighbor and decision tree, are experimentally investigated in the work. Final conclusions, recommendations and references are presented on the end of this work
Modernization of the water supply control system in electricity production process.
No full textIn this work I have analysed water supply cascade control system, which was based on three PID‘s, for electricity production process. This kind of cascade control systems have complex structure and cause difficulties which appear while tuning systems. Due to these conditions it was decided to modernize system by removing one control loop. I have used Ovation 3.2 DCS on “VMware Workstation” test-and-development environment for making this system model. At first, water supply control system for electricity production process was created including water supply, drum level control, pumps and fluid couplings, valves control logic and graphic interface. Then processes on three PID cascade control system was monitored and compared with modernised and tuned cascade control system, which consist of 2 PID controllers
Development of control system of 3D printer.
No full textThis paper describes real work, which main objective is to create three dimensional printer. This printer was created for the purpose of demonstrating modern motion technologies and to help understand positioning and movement programming. This three dimensional printer has three main functions: printing with plastic, milling and drawing. For convenient human machine interface (HMI) there is SCADA visualisation, in which there is accessible all the functions of this printer. For finding the optimal parameters it was mandatory to perform several experiments with motion and heating. When found these parameters the printer performed in raised criteria margins
Finansinių tendencijų vertinimui skirto prognozavimo modelio tyrimas.
No full textIn this thesis, chart pattern influence on the GA-LSTM-CNN model is investigated. Chart patterns are a branch of technical analysis that uses stock price movement shapes or candlesticks to predict future price movement. The GA-LSTM-CNN model was selected due to its combination of several neural network algorithms that achieve self-structuring, temporal and spatial pattern recognition. As it is not clear which features should be chosen for stock trading due to numerous trading strategies, a genetic algorithm provides an automatic search for features that best fit the use case. LSTM and CNN combinations provide the model with temporal and spatial pattern recognition capabilities, respectively. That is needed in stock trading, as patterns exist between features and in time for the same feature. Before the preliminary tests were conducted, chart patterns were selected from financial data using analytical rules outlined in the analysed research paper. Searching of chart patterns was done backwards to not give the model any knowledge of financial events in the future. Preliminary tests showed that the model selected chart patterns as one of the features. The results were similar to those of the model without chart patterns. Final tests revealed that chart patterns do not have a significant impact on individual model performance, but could be useful in an ensemble scenario. This thesis is separated into seven sections. The first section is a brief introduction to the problem, object of experiments and experimental steps taken. The second section is state-of-the-art research, where 29 sources are analysed to get a comprehensive overview of automated stock market trading. The third section contains explanations of used methods, models and metrics. The fourth section has descriptions of the training and testing environments, data used for experiments. The fifth section has comparison of model testing results. The sixth section is a discussion on problems encountered in the experiments, successes and future work. Finally, conclusion is given on model performances in the last section. In total, this work contains 41 figures and 53 pages
Research of diseased wheat recognition methods.
No full textAgro – industry sector plays an important role in today‘s economy. Agricultural production is easily affected by various plant diseases that cause ecological and economical damage. So, as in many areas, no exceptions in agriculture, the aim is to apply advanced technical solutions to achieve greater efficiency. The paper analyzes the literature, which describes the review of research in the world of science on a related topic. This work also includes studies to evaluate the effectivnes of different structures in recognizing wheat leaves and stems affected by rust disease
Feasibility study of artificial inteliligence methods application for automatic trading robot.
No full textThe paper provides an overview of the stock market, it’s patterns and possibilities of algorithmic systems usability. Created system works independently, without human intervention with the main goal to make successful financial decisions in the global stock market and simulate trading actions. In order to implement the task selected the ranking methodology, created by the famous American financial advisor Ted Allrich. Modified version of this ranking methodology is adapted for this work. Two ranking methods (rigid and don’t rigid) and 6 different observation time of investment portfolios were analyzed. MS Excel software package and integrated VBA programming environment are selected for algorithm implementation. The program code is written and install user interface using Excel forms in this environment. Created system is capable under the user described criteria to login to the website \"finviz.com\" and using duplicate connections tactics to collect selected stock data. In order to test system 5-month period is chosen. The results, advantages and disadvantages of this system are presented in Master's thesis
Research and development of algorithm of solid image resolution.
No full textIn this final master work was developed algorithm for focal plane stacking, to increase depth of field in photo. From 28 mathematical methods was chosen 8, which showed best performance in information detection in photos. These methods was used to detect focal plane in photos. For focal plane stacking was used Laplace pyramid. Was measured quality of results, using mathematical methods. Was set the most appropriate method for detecting focal plane
Research of forest fire detection system.
No full textAs global warming gains momentum and average global temperatures are constantly rising, the number of forest fires is rising along with it, due to more frequent droughts. Forest fires can destroy entire ecosystems, permanently altering landscapes and soil structure. Rapid and accurate smoke detection is essential in order to reduce the damage caused by wildfires. There are currently several forest fire prevention methods on the market, but not all of them are accurate, fast and autonomous. This paper describes how effective YOLOv5 method can be for fire localization and evaluation tasks. This method was successfully used for solving traffic, human detection and tracking problems. In this work, the speed and accuracy of the algorithm are investigated. The dependency of the accuracy is carried out using different base models on different databases and testing equipment. An additional study is performed in order to analyze the dependence of the accuracy with different union and intersection parameters
Medinių kotelių vizualinės patikros sistemos kūrimas ir tyrimas.
No full textThe purpose of this study was mainly due to a problem statement provided by a company manufacturing small wooden dowels. A visual inspection system to identify the presence of defect in the stick will be required. The focus will be in the algorithm development and proposal for the system design will be given. The previous literature assumed a particular color space to be the best and worked on selecting the effective features that could be extracted from them. The main focus of this thesis is to identify the best color space that could be used for all defect detection and grading system related to wood. Not well studied Neural Network structure for wood defect application will be used as the classifier and the reaction of the number of inputs and number of hidden units to produce effective results was studied At first a simple and effective algorithm for finding the defect was developed. This algorithm was tested with different colour space for different number of hidden neurons. The effectiveness of each color space was found and the best color space was chosen. The best NN structure and the outputs for different neuron numbers was also found. There is no exact rule for selecting the number of neurons so far, so based on the previous assumptions and studies the neurons were selected and checked if the assumptions holds good in our case. The system was designed based on the requirements and the setup was made. The algorithm part was developed with MATLAB. From the experiments the effective color space RGB and YCbCr produced results with 96% accuracy and the combination of them produced about 99% accuracy for the chosen neuron numbers. The not very well studied NN structure was studied. Algorithm was made effective by minimizing the time for processing as much as possible
- …
