Sensorlose Überwachung der Einzelteilfertigung/Spindle-current-based process monitoring using artificial intelligence

Durch die Messung von Spindelströmen lassen sich Informationen aus spanenden Fertigungsprozessen ohne zusätzliche Sensorik erfassen. Diese Daten sind aus der Maschinensteuerung mit geringem Aufwand abrufbar. Jedoch stellt die Klassifikation von Fehlerereignissen bei Einzelteilen ohne Vergleichsdaten eine erhebliche Herausforderung dar. Untersucht wurde die Berechnung von Toleranzgrenzen durch ein neuronales Netzwerk basierend auf den Daten einer Materialabtragssimulation.   Spindle current measurement allows acquiring process information without the need for additional sensors. Digital machine controls allow accessing the data with low effort. However, precise classification of process errors is a non-trivial task, especially for complex, single item workpieces without reference data. This work presents an approach to predict the spindle current and calculate tolerance limits by using a neuronal network based on a material removal simulation.

Deep Anomaly Detection for CNC Machine Cutting Tool Using Spindle Current Signals

In recent years, industrial production has become more and more automated. Machine cutting tool as an important part of industrial production have a large impact on the production efficiency and costs of products. In a real manufacturing process, tool breakage often occurs in an instant without warning, which results a extremely unbalanced ratio of the tool breakage samples to the normal ones. In this case, the traditional supervised learning model can not fit the sample of tool breakage well, which results to inaccurate prediction of tool breakage. In this paper, we use the high precision Hall sensor to collect spindle current data of computer numerical control (CNC). Combining the anomaly detection and deep learning methods, we propose a simple and novel method called CNN-AD to solve the class-imbalance problem in tool breakage prediction. Compared with other prediction algorithms, the proposed method can converge faster and has better accuracy.

On-Line Fuzzy Monitoring of Micro-Hole Drilling Based on Genetic Algorithm

In automated micro-hole drilling system, in order to improve the drilling performance and reduce of production costs by maximizing the use of drill life and preventing drill failures, the drill bit wear state monitoring is more important. However, drill bit wear is difficult to measure in drilling process. By observation, wear failure of the drill bit could cause related changes of the spindle current signal, so construct fuzzy control mathematical models with the relationship between drill bit wear and spindle current, genetic algorithm and fuzzy control theory are applied to micro-drilling system in this paper .The membership functions of fuzzy control model are optimized by genetic algorithms. Through calculation, we can get drill bit wear value which used as monitoring threshold value in micro-hole drilling on-line monitoring system to avoid the drill breakage and improve the monitoring reliability.

Investigation of Tool Failure Modes and Machining Disturbances Using Monitoring Signals

The ACCENT Project (FP7-AAT-2007-RTD-1) will allow the European Aero Engine manufacturers to improve their competitiveness by applying adaptive control techniques to the manufacturing of their components. For the critical rotating parts of aircraft engines, the surface integrity generated after machining is a key factor on the life cycle. In this context, one particular attention has to be carried out on tool condition. The aim of this paper is to define a monitoring approach able to detect the tool condition and machining disturbances. The main failure modes characterizing this particular Nickel base drilling and the apparition of embedded chips over the machined surface were identified. By experimental techniques, cartography of failure modes was performed. The results show that flank wear and notch are the main failure modes limiting the tool life. For some cutting conditions, the tool failure occurs after the first hole due to the important cutting forces. Some interesting combinations are made between the spindle current/accelerometers/ thrust force and flank wear, tool breakage and notch. Before these correlations, a detailed signal analysis was performed, considering different disturbing phenomena, such as chips evacuation problem. Finally, a “synopsis” for process monitoring is proposed, considering the analyzed phenomena.

Research on Modeling Mapping Relationship between Cutting Load and Spindle Motor Current

The relationship between cutting load and motor current of CNC machine tools is an important part for research in intelligent control system. In this paper, the process of processing the error spindle current signal averaging, filtering, signal preprocessing, At the base of feature analysis and extraction, using system identification theory get the mapping model from current of the spindle to cutting load. Finally, the actual cutting data used to verify the model is reasonable.