Data Reducing in Machine Diagnostics With Neural Networks

Abstract

The paper presents a method to collect and save only relevant data from a machine. The data can be used for training the system to identify running situations that have already occurred in the monitored machine or similar types of machines. This method can be used to give direct notice or alarm in a control room with clear indication of the problem. Modern software development tools, like Java for instance, offer new technologies for operation monitoring of machines. The system can be accessed via Internet using common browsers. The paper presents practical experiences in the application of the method. It demonstrates, how the system will adapt into various environments and running situations. The method presented in this paper is applied in operation monitoring of machines, but it can be also used for other monitoring purposes.