Metal droplets are basic elements of many droplet based innovative manufacturing technologies which nowadays are highly demanded in different industrial applications. In the presented laser droplet generation process an annular laser beam pulse is used to melt the wire-end of the vertically fed metal wire. Due to the interplay of surface tension, gravity force, and light-metal interaction, undulating pendant droplet is formed from the molten wire-end. The critical phase of the laser droplet generation process, that essentially influences the process dynamics, is pendant droplet detachment from the solid part of the wire-end. In the paper, different possible detachment scenarios, related nonlinear phenomena, dynamics and process instabilities observed in the forced drop-on-demand and continuous droplet generation process are presented. A nonlinear model of continuous mass spring like resonant detachment regime is described, results of which are in good agreement with the experiments.
B.04 Guest lecture
COBISS.SI-ID: 14026523In laser droplet generation process a laser pulse is used to form and detach a droplet from a metal wire. Using a sequence of laser pulses, dripping of metal droplets can be generated, exhibiting complex dynamics. In this paper the dynamics of various dripping regimes in dependence on detachment pulse power is considered. The dynamics is analyzed and characterized based on experimental time series. By means of state space reconstruction from the time series and Lyapunov spectra, it is shown that the process should be considered as a deterministic low-dimensional chaotic dynamical system. Subsequently, recurrence plots and corresponding quantification analysis are employed to strengthen this result and also to further extend it by inferring that the transition from spontaneous to forced dripping regime has characteristics of a chaos-to-chaos transition.
B.04 Guest lecture
COBISS.SI-ID: 14027035This work deals with the topological approach to mobile robot navigation using recurrent neural networks (RNNs). We applied a special architecture of RNNs to enhance the basic topological approach. A novel RNN architecture selectively latches presumably relevant input information and ignores presumably irrelevant input information. Simple types of reactive behavior are supplemented with random decisions to switch between them at decision points. The RNN is trained on a sequence of sensory contents and actions. This approach is applicable to multi-step prediction of sensory information and the travelled distances between decision points, given a sequence of decisions at decision points. Thus, the optimal path to a specified goal can be sought. A problem of this approach is that due to inherent inability to design a perfect reactive behaviour, unwanted situations may appear, such as redundant decision points and unreliable switching among behaviours. We demonstrate that the selective RNN lowers the impact of faulty decision points and thus improves the prediction.
F.02 Acquisition of new scientific knowledge
COBISS.SI-ID: 1536707523Various forecasting models are considered and compared for shortterm heat load forecasting in a district heating system. Heat load data and weather related influential variables for five subsequent winter seasons of district heating in Ljubljana (Slovenia) are applied in this study, and additional informative features are extracted to improve the forecasting accuracy. Forecasting models include linear autoregressive and stepwise regression models, neural networks and extreme learning machines. The models are developed with the objective to forecast the future daily heat load with the forecasting horizon one day ahead. Evaluation of the forecasting models is based on generalization error, obtained on an independent testing data set. Comparison of forecasting models reveals good forecasting performance of a linear stepwise regression model (SR) that utilizes only the most relevant input variables. The operation of SR model was improved by using neural network (NN) models, and also NN models with direct linear link (NNLL). The best forecasting result was obtained by using extreme learning machine (ELM) model. The results demonstrate the applicability of NN, NNLL and ELM models to accurately forecast the heat load data, but also reveal practical considerations in designing NN-based and ELM models. Namely, random initializations of NN-based and ELM models require multiple iterations of a learning procedure in order to achieve good forecasting results. Furthermore, ELM models are sensitive to the range of input variables because hidden layer weights are not tuneable but randomly chosen. Only if properly designed and trained, NN-based and ELM models offer a good forecasting tool for the district heating market.
F.15 Development of a new information system/databases
COBISS.SI-ID: 14083867Forecasting energy consumption in district heating networks is crucial for wise energy resource management and sustainable energy supply. We investigated which factors most affected the heat consumption and how the forecasting accuracy can be improved. We found that air temperature, past heat consumption and solar radiation had the greatest impact on heat consumption. Nine empirical models were developed for heat consumption forecasting in the district heating system. Random-walk model and temperature correlation model were applied as reference models, and then more complex regression model and several types of autoregressive models were developed. The quality of empirical models was estimated by cross validation procedure based on forecasting error which represented the mean absolute error normalized by the maximum transfer capacity of the district heating network. Forecasting results have been improved by introducing the population variables for the days of the week and additional variables describing the system growth and seasonal cycle. The best forecasting result was obtained by an empirical model from a group of autoregressive models (model SR), which was developed by a stepwise regression method.
F.15 Development of a new information system/databases
COBISS.SI-ID: 13917723