The paper is concerned about the system of automatic detecting of wear and damages of the end mill tool by the use of computer vision. By using the algorithms developed for image segmentation and by an innovative approach to extraction of features describing individual end mill tool tooth the information representing significant features of the individual tool tooth is effectively gained from the captured image. The proposed approach to feature extraction is robust and independent of the scale and rotation of the end mill tool in the image. The features vectors have been classified by two approaches, i.e., k-nearest neighbor algorithm and artificial neural network. Both approaches have been tested by the test base of tool images and the results mutually compared. The classification being validated by 10-fold cross-validation method. The best precision of classification (92.63 %) has been reached by the use of artificial neural network. Simulation results have confirmed that the proposed approach can improve monitoring of tool wear and damages and, consequently, the effectiveness and reliability of CNC milling machine tool.
COBISS.SI-ID: 19214358
In this paper, the effect of process parameters was analysed on the laser cut quality of an uncommon alloy, the tungsten alloy (W ≈ 92.5% and the remainder Fe and Ni) sheet with thickness of 1mm. This paper introduces a developed back-propagation artificial neural network (BP- ANN) model for the analysis and prediction of cut quality during the CO2 laser cutting process. In the presented study, three input process parameters were considered such as laser power, cutting speed and assist gas type, and two output parameters such as kerf width and average surface roughness. Based on the results of the study, it was shown that the proposed artificial neural network model could be a useful tool for analysing and predicting surface roughness and kerf width during CO2 laser cutting processes.
COBISS.SI-ID: 19154966
This paper describes a genetic based approach for the modelling of elongation in cold drawn copper alloy. Genetic programming is one of the most general genetic based methods and was used in our research. It is an automated evolutionary computation method for creating a working computer programme from a problem%s high-level statement. Genetic programming does this by breeding a population of computer programmes genetically using the principles of Darwinian%s natural selection and biologically inspired operations. In our research, material was formed by drawing using different process parameters and then determining elongation of the specimens. On the basis of a training data set, various different genetic models for the elongation distribution were developed during simulated evolution. The accuracies of the best models were proved by a testing data set and comparison between the genetic and regression models was carried out.
COBISS.SI-ID: 18521110
A visual cutting chip control system is designed to automatically adjust feed rate in order to maintain constant surface roughness in ball-end milling. The proposed visual control system has a modular structure, consisting of an optical vision system (OVS), an adaptive cutting chip size-control loop for a feed servo and a surface roughness in-process prediction model. The OVS is employed to acquire the cutting chip sizes form the camera. A division controller is used to control the chip size by modifying the feed rate and consequently maintaining surface roughness constant. Surface roughness is predicted based on the detected chip size. The efficiency of the chip control strategy is tested by series of simulation with various step changes in the cutter/workpiece contact area. For simulation purposes an experimentally validated milling plant simulator with an adopted feed servo drive model and a cutting chip size model is employed. An adaptive neural inference system (ANFIS) is established to effectively simulate the cutting chip size in ball end-milling. In simulation, the reference chip size and consequently the reference surface roughness are well maintained when the cutting-depth profile of a workpiece is varying step-wise or continuously.
COBISS.SI-ID: 18907670
In the field of hydraulic drive technology various power supply systems are used within different power unit set-ups. The both of two mostly used drive concepts in modern electrohydraulic systems, a variable displacement pump driven by constant speed electric motor and fixed displacement pump driven by a variable speed electric motor, have some disadvantages. Especially regarding the increasing demand for maximum efficiency of the entire power unit without lowering the high dynamics. The combination of a variable pump and speed-controlled electric motor, offers the option of setting two parameters of the drive, the rotational speed of the motor and the pump flow-rate. Such a combination allows all power unit components to operate within the areas of their maximum efficiencies, the so-called maximum efficiency drive. A prerequisite for designing suitable controllers that would ensure the operations of individual components within the areas of maximum efficiency, regardless of the current operating point, is certainly knowledge about the efficiency area of the entire power unit. The paper presents a procedure for determining areas of efficiency, first on the basis of simulation and detailed models of each component, and later verification of the model using an experiment.
COBISS.SI-ID: 19213846