Article deals with design of appropriate control strategy for prediction of surface roughness as one of the important indicators of machined surface quality applying artificial intelligence. Test sample was nickel based super alloy UDIMET 720, which is used as material of jet engines components such as discs etc. Experimental data collected from tests were used as input parameters into neural network to identify the sensitivity among cutting conditions, tool wear and monitoring parameters and surface roughness. Selected parameters were used to design a suitable algorithm for control and monitoring of the drilling process. Moreover, the developed software for implementation to machine tool control system for surface roughness on-line identification through monitoring indices is described

Shaped surfaces are increasingly used in the field of mold-making for casting or injection molding, where future products include shapes with different curvatures. These are surfaces that form convex curves, concave curves, or a combination thereof. Given these machined surfaces, it is important to know the impact of the finishing strategies on these surfaces. This paper deals with the comparison of finishing milling strategies in the production of shaped surfaces and the analysis of different methods for the evaluation of surface topography. In addition, the experimental results evaluate the roughness of the machined surface and surface shape variations. The material used for the experiments was AlCu4Mg aluminum alloy, and Constant Z, spiral and spiral circle strategies were chosen for the finishing strategies. The evaluation of surface topography and surface roughness was carried out at three different specimen heights with respect to the tool’s contact with the machined surface.

Freeform surfaces are present on an increasing number of engineering products. Three- and multi-axis computer numerical control milling machines are commonly used for improved production. Computer-aided manufacturing (CAM) systems are used almost exclusively for the creation of programs for a variety of machining centers. This study compared the quality of freeform surfaces made by 3- and 5-axis milling using three commonly used strategies (linear, offset, and spiral). The CAM system-predicted surfaces were also compared with the actual surfaces. A test sample with a freeform surface was used for the experiments. Considering the size and distribution, the discrepancy between the predicted surface deviations and the deviations in the produced samples was proven. Maximum negative surface deviations, when 5-axis milling, employed linear and spiral strategy of 29% and 71% less than those produced by the 3-axis milling. On the contrary, positive deviations were 48% smaller. A comparison of the scans showed that the two strategies (linear and spiral) yielded better results for 5-axis milling, and the offset strategy was better for 3-axis milling. Evaluation of the achieved surface roughness showed that the milling method did not significantly affect the surface quality in the linear strategy. However, other two strategies (offset and spiral) achieved better results with 5-axis milling compared to 3-axis milling. The proposed method of evaluating the accuracy of machined free form surfaces can be used in experimental as well as production activities.

There is a variety of reasons for the installation of a monitoring system in a manufacturing process. Hole-making mainly drilling is one of the most common operation used and usually is carried out as one of the last steps in the production process. Holes in rotating turbine and compressor disks are among the most highly-stressed geometric features of jet-engines. For manufacturers of jet-engine components it is important to assess the quality of these at an early stage in the manufacturing of the product. The use of commercially available monitoring systems in hole-making has been successful in individual cases so far. Major reasons for this lack of effectiveness are the large material variations within one production batch, the overall difficult machinability of the materials applied, the small lot size which makes “teach-in” operations ineffective. The paper describes a design of adaptive control system for drilling process of aerospace critical components. The proposed system is directed towards the real time control of selected surface roughness parameter. Proposed model for monitoring and control consists of two subsystems: surface roughness prediction subsystem and decision making subsystem. The artificial neural network was employed to calculate surface roughness parameters throughout process monitoring indices such as torque Mz, force Fz, power P and cutting conditions feed f, cutting speed vc. Due to ability to predict nonlinear behaviour and quickly calculate future values, artificial neural networks are ideal for both predictive and adaptive controllers. Test samples were nickel based super alloy Udimet 720 used in discs for gas turbine engines. The experimental results show that predicted values of surface roughness are very close to the values measured experimentally. Advantages of the proposed subsystem for surface roughness prediction are simplicity, computational power and speed, capacity and ability to learn from system changes as they become