A flexible workcell controller has been developed using a three level control hierarchy (workcell, workstation, equipment). The cell controller is automatically generated from a model input by the user. The model consists of three sets of graphs. One set of graphs describes the process plans of the parts produced by the manufacturing system, one set describes movements into, out of and within workstations, and the third set describes movements of parts/transporters between workstations. The controller uses an event driven Petri net to maintain state information and to communicate with lower level controllers. The control logic is contained in an artificial neural network. The Petri net state information is used as the input to the neural net and messages that are Petri net events are output from the neural net. A genetic algorithm was used to search over alternative operation choices to find a "good" solution. The system was fully implemented and several test cases are described.

Šiame darbe tiriama sritis yra kompiuteriniuose žaidimuose naudojamas dirbtinis intelektas. Konkrečiai gilinamasi į metodus, kurie valdo daugybę veikėjų žaidime, siekiančių tam tikro tikslo. Dėl konkretiems žaidimams unikalių mechanikų, šie metodai paprastai būna labai glaudžiai susiję su žaidimo aplinka ir taisyklėmis. Tyrimo tikslas yra sukurti ir ištirti metodą, skirtą daugybės veikėjų pajėgų valdymui ir jų veiksmų modeliavimui virtualioje aplinkoje. Analizuojami metodai skirti pavienių veikėjų veiksmų įvertinimui ir modeliavimui, metodai skirti optimalių sprendimų žinių bazei sudaryti ir metodai toms žinioms pritaikyti paskirstant veikėjus aplinkoje. Pagal analizės rezultatus sukuriamas projektas daugelio veikėjų valdymui realiu laiku virtualioje aplinkoje. Lokalių veikėjų veiksmų modeliavimui naudojamas procedūrinis taktinių veiksmų parinkimo metodas. Veikėjų judėjimui aplinkoje modeliuoti naudojamas neuroninis tinklas. Jis apmokomas pagal genetiniu algoritmu sudarytus optimalius sprendimus. Suprojektuota sistema realizuojama ir testuojama. Atliekamas eksperimentas su sistemos veikimo metu gautais rezultatais. Eksperimente nustatoma, kad šis sprendimo būdas gali tikslingai reaguoti į situacijas, susidarančias realaus laiko virtualioje aplinkoje, ir modeliuoti veikėjų veiksmus joje. / The research area of this paper is artificial intelligence used in computer games. Specifically it is focused on methods for controlling a group of agents with a specific goal. Because of the uniqueness of individual game mechanics, those kinds of methods are usually closely related to that games environment and rules. The goal of this study is to design and test a method that could control a group of multiple agents in a virtual environment. Methods for evaluating and selecting individual agent actions in a local environment, for gathering a database of optimal solutions and for applying that knowledge in distributing agents across the environment are analyzed. A design for controlling the actions of multiple agents in a real time virtual environment is designed, based on the results. Dynamic procedural combat tactics is used to model individual agent actions in a local environment. A neural network is used to model the movement of multiple agents in an environment. It is trained using optimal solutions, generated by a genetic algorithm. Designed system is implemented and tested. Using data that the system generates, an experiment is conducted. It shows that this solution is capable of correctly reacting to situations, occurring in a real time virtual environment, and of modeling multiple agent actions in it.

An accelerated high cycle fatigue testing approach is presented to determine the fatigue endurance limit of materials at high frequencies. Base excitation of a tapered plaque driven into a high frequency resonance mode allows the test to be completed in a significantly shorter time. This high cycle fatigue testing is performed using the tracked sine resonance search and dwell strategy. The controller monitors the structural health during the test. Any change in the dynamic response indicates crack initiation in the material. In addition, a shape optimization finite element model is conducted for the design of the tapered plaques. An integrated neural (Neural-Network) genetic (NSGA_II) optimization technique is implemented to carry out the shape optimization for this component. This process results in a significant reduction in the computational cost. A Pareto set is then produced that meets the designer’s requirements and provides the decision maker several alternatives to choose from.

Portfelio parinkimo uždavinys yra viliojantis technikos moksluose, tiesioginiuose („online“) algoritmuose ir žinoma financiniuose skaičiavimuose. Šiame darbe buvo naudotas algoritmas, kuris nebando atspėti laimėtojus. ANTICOR algoritmo optimizavimui buvo naudotas genetinis algoritmas ir algoritmo stabdymas. / The portfolio selection problem is a challenging problem for machine learning, online algorithms and of course, computational finance. In this work was used a portfolio selection algorithm, which does not try to predict winners. There were used a genetic algorithm and algorithm stopping, trying to optimize the ANTICOR algorithm.

A fuzzy logic controller (FLC) for the dissolved oxygen (DO) concentration of a wastewater bioreactor is presented. The FLC is developed and tested based on simulations using first order plus dead time models obtained from experiments with an actual wastewater bioreactor. The FLC uses feedback of the error in DO concentration and rate of change of the DO concentration and manipulates the stem position of the flow control valves (FCVs) supplying air to the bioreactor. The proposed FLC is tested for robustness across several process models, two of which include proposed worst-case process conditions. The performance of the proposed hand tuned FLC is compared to that of a similarly tuned proportional-integral-derivative controller. The FLC is implemented as a lookup table for speed and ease of deployment. The disturbances present in the experimental step testing data are characterized and used as the basis for disturbing the control loop during controller performance testing. A low-pass filter is then included to subsequently smooth the feedback signal. The nonlinear relationship between the FCV stem position and output flow is modelled and included in the controller performance testing. A genetic algorithm (GA) is developed that manipulates the membership functions of the FLC to yield an optimal controller for the ensemble of process models. The ability of the GA to converge on an optimal FLC is verified through repeated trials. The performance of the GA optimized FLC is observed under realistic process conditions and is benchmarked against a manually optimized PID controller.

Simulation of animal disease spread is essential for understanding and controlling the outbreak of disease among herds of livestock (in particular cattle and poultry). Using a computerized system or simulator, animal health professionals or epidemiologists often spend many hours determining the set of input parameters that most accurately represent a disease spread or an outbreak scenario. A parameter can be a simple boolean value, or a scientific or often hypothetically derived range of real numbers. Many times, an epidemiologist chooses a value provisionally in a random fashion and repeats the simulation until a viable solution is achieved. This tedious process is inefficient and lengthy. To assist and improve this laborious practice in a concise and timely manner, a Genetic Algorithm is employed to determine a population based solution consisting of input parameters using the North American Animal Disease Spread Model (NAADSM).

Šiame darbe išanalizuoti genetinių algoritmų (GA) veikimo principai. Analizuojamos egzistuojančios modeliavimo aplinkos ir genetiniais algoritmais pagrįsti modeliavimo įrankiai. Kuriant modeliavimo įrankį, nustatyti funkciniai ir nefunkciniai reikalavimai. Realizuotas sukurtas įrankis ir atliktas pasirinktos esybės evoliucijos modeliavimas. "16th International Conference on Information and Software Technologies" konferencijoje buvo pristatytas pranešimas “Genetic Algorithm Modeling Approach for Mobile Malware Evolution Forecasting”. Panaudojus jame pateiktus parametrų duomenis, atlikti bandymai su sukurtu modeliavimo įrankiu. Taip pat XIV jaunųjų mokslininkų konferencijoje „Mokslas - Lietuvos ateitis“ 2011, pristatytas straipsnis „Informacijos saugos dalykinės srities esybių evoliucijos modeliavimo įrankio, paremto genetiniais algoritmais, kūrimas “, šis straipsnis buvo parašytas remiantis šiuo darbu. Darbą sudaro: 7 skyriai, 26 paveikslai, 7 lentelės, 2 priedai. Literatūros sąraše 52 šaltiniai. / This study analyzes operating principles of the genetic algorithms (GA), also submit proposals for the calculation of GA. Discuss the existing simulation environment and tools to implement GA. Towards a modeling tool to determine the functional and non-functional requirements. Marketed developed tool and to carry out tests for selected test of evolutionary analysis. ‘16th International Conference on Information and Software Technologies’ was presented the article ‘Genetic Algorithm Modeling Approach for Mobile Malware Evolution forecasting’ using the parameters details of this article perform the tests with simulation tool. Also XIV Conference of Young Scientists ‘Science - The future of Lithuania‘ 2011, was introduced the article ‘Development of the subject area of the information security beings evolutionary modeling tool based on genetic algorithms’. This article was written on the basis of this work. Thesis consist of: 7 chapters, 26 pictures, 7 tables, 2 appendixes, 52 bibliographical entries.

Unconventional reservoirs are typically characterized by very low permeabilities, and thus, the pressure depletion from a producing well may not propagate far from the well during the life of a development. Currently, two approaches are widely utilized to perform unconventional reservoir analysis: analytical techniques, including the decline curve analysis and the pressure/rate transient analysis, and numerical simulation. The numerical simulation can rigorously account for complex well geometry and reservoir heterogeneity but also is time consuming. In this thesis, we propose and apply an efficient technique, fast marching method (FMM), to analyze the shale gas reservoirs. Our proposed approach stands midway between analytic techniques and numerical simulation. In contrast to analytical techniques, it takes into account complex well geometry and reservoir heterogeneity, and it is less time consuming compared to numerical simulation. The fast marching method can efficiently provide us with the solution of the pressure front propagation equation, which can be expressed as an Eikonal equation. Our approach is based on the generalization of the concept of depth of investigation. Its application to unconventional reservoirs can provide the understanding necessary to describe and optimize the interaction between complex multi-stage fractured wells, reservoir heterogeneity, drainage volumes, pressure depletion, and well rates. The proposed method allows rapid approximation of reservoir simulation results without resorting to detailed flow simulation, and also provides the time-evolution of the well drainage volume for visualization. Calibration of reservoir models to match historical dynamic data is necessary to increase confidence in simulation models and also minimize risks in decision making. In this thesis, we propose an integrated workflow: applying the genetic algorithm (GA) to calibrate the model parameters, and utilizing the fast marching based approach for forward simulation. This workflow takes advantages of both the derivative free characteristics of GA and the speed of FMM. In addition, we also provide a novel approach to incorporate the micro-seismic events (if available) into our history matching workflow so as to further constrain and better calibrate our models.

The purpose of this work is to optimize the chemical composition as well as the heat treatment for improving the mechanical performance of the TRIP steel by employing the theoretical models. TRIP steel consists of the microstructure with ferrite, bainite, retained austenite and minor martensite. Austenite contributes directly to the TRIP effect as its transformation to martensite under the external stress. In order to stabilize austenite against the martensitic transformation through the heat treatment, the two-step heat treatment is broadly applied to enrich the carbon and stabilize the austenite. During the first step of the heat treatment, intercritical annealing (IA), a dual phase structure (ferrite+austenite) is achieved. The austenite can be initially stabilized because of the low carbon solubility of ferrite. The bainite isothermal treatment (BIT) leads to the further carbon enrichment of IA-austenite by the formation of carbon-free ferrite. Comparing to the experiments, the thermodynamic and kinetic models are the lower and upper bounds of the carbon content of retained austenite. The mechanical properties are predicted using the swift model based on the predicted microstructure. In this work, a theoretical approach is coupled to a Genetic Algorithm-based optimization procedure to design (1) the heat treated temperatures to maximize the volume fraction of retained austenite in a Fe-0.32C-1.42Mn-1.56Si alloy and the chemical composition of (2) Fe-C-Mn-Si and (3) Fe-C-Mn-Si-Al-Cr-Ni alloy. The results recommend the optimum conditions of chemical composition and the heat treatment for maximizing the TRIP effect. Comparing to the experimental results, this designing strategy can be utilized to explore the potential materials of the novel alloys.

No description available.