A5036367968- Advanced machining processes and optimization
- Adaptive Control of Nonlinear Systems
- Neural Networks and Applications
- Advanced Machining and Optimization Techniques
- Iterative Learning Control Systems
- Fault Detection and Control Systems
- Tunneling and Rock Mechanics
- Fuzzy Logic and Control Systems
- Dental Implant Techniques and Outcomes
- Advanced Control Systems Optimization
- Micro and Nano Robotics
- Engineering Technology and Methodologies
- Nanotechnology research and applications
- Robotic Path Planning Algorithms
- Advanced Surface Polishing Techniques
- Molecular Communication and Nanonetworks
- Advanced Algorithms and Applications
- Drilling and Well Engineering
- Lubricants and Their Additives
- Sensor Technology and Measurement Systems
- Industrial Technology and Control Systems
- Orthopaedic implants and arthroplasty
- Engineering Diagnostics and Reliability
- Manufacturing Process and Optimization
- Modular Robots and Swarm Intelligence
University of Zagreb
2012-2024
Faculty of Mechanical Engineering and Naval Architecture in Zagreb
2020
Forecasting performances of feed-forward and recurrent neural networks (NN) trained with different learning algorithms are analyzed compared using the Mackey-Glass nonlinear chaotic time series. This system is a known benchmark test whose elements hard to predict. Multi-layer Perceptron NN was chosen as network because it still most commonly used in financial forecasting models. It modified version so-called Dynamic characterized dynamic neuron model, i.e., Auto Regressive Moving Average...
The existing controllers for robot manipulators with uncertain gravitational force can globally stabilize only revolute joints. main obstacles to the global stabilization of mixed and prismatic joints are unboundedness inertia matrix Jacobian gravity vector. In this note, a class stable is proposed. asymptotic achieved by adding nonlinear proportional derivative term linear proportional-integral-derivative (PID) controller. By using Lyapunov's direct method, explicit conditions on controller...
In this paper, a new class of finite-dimensional repetitive controllers for robot manipulators is proposed. The global asymptotic stability proved the unperturbed system. passivity-based design proposed controller avoids problem tight conditions and slow convergence conventional, internal model-based, controllers. passive interconnection nonlinear mechanical systems provides same as with exact feed-forward compensation dynamics. simulation results on three degrees freedom spatial manipulator...
This work considers the application of radial basis function neural network (RBFNN) for tool wear determination in milling process. Tool wear, i.e., flank zone widths, have been estimated two phases using types RBFNN algorithms. In first phase, pattern recognition algorithm is used order to classify features three level classes (initial, normal and rapid wear). On behalf these results, second regression utilized estimate average amount widths. were extracted time frequency domain from...
This paper proposes a new algorithm for the automatic generation of toolpaths machining complex geometric positions, such as molds used in orthosis production. The production individualized orthoses often requires use multi-axis systems, five-axis machines or industrial robots. Typically, and expensive CAD/CAM systems are to generate these machines, requiring definition strategy each surface. While this approach can achieve reliable high-quality process, it is very time-consuming makes...
This paper investigates the energy efficiency of a Direct Driven Hydraulic (DDH) system and proportional electrohydraulic system. A detailed analysis their is carried out based on experimental results obtained in laboratory conditions using fully loaded cylinder sine wave reference trajectory. To ensure fairness testing process, same with initial used for both systems. The velocity estimated online novel algebraic differentiator approach measured position. power each component calculated...
In this paper, an improved Levenberg-Marquardt-based feedforward neural network, with variable weight decay, is suggested. Furthermore, parallel implementation of the network on graphics processing unit presented. Parallelization achieved two different levels. First level parallelism data set level, where parallelization possible due to inherently structure networks. Second Jacobian computation level. Third parallelism, i.e. optimization search steps, not implemented which makes third...
Control of autonomous robot motion in radial mass density field is presented. In that sense the described as function parameters. The between maximal and minimal density. Between these two limited values one can use n points (n = 1, 2, . nmax) calculate related for each point. at gravitational radius radius. This conclusion valid Planck scale, but also scales are less or higher one. Using ratio it generated energy conservation constant with value κ 0.99993392118. Further, this theory...
Drill bits with internal cooling capabilities are still not employed in stone machining practices within shop floor environments. Therefore, a conventional industrial drill bit used was subject to redesign wherein an axial channel machined throughout its body. A comparison drawn between the standard without and redesigned bit, which compressed air as medium. The experiment performed by drilling three types of samples varying hardness nine combinations cutting speed feed rate. During process,...