An intelligent diagnostic system from Siemens can monitor the condition of mechanical equipment just by analyzing the noises it makes. The system's sensors listen to machine noises in the same way that a doctor listens to a patient's heart and lungs. The system learns to tell the difference between normal and faulty operation by analyzing noises and vibrations. The sensors it requires are simple, and the system can be used on many different kinds of machines and equipment. Three prototypes of STEVE (Siemens Tremor EValuation Equipment) - as the Siemens' researchers have dubbed their system - are currently undergoing testing in power plants in Morocco and the U.S., as well as at Siemens' gas turbine test center in Berlin.
Many technical facilities, such as power plants, are so complex that it isn't possible to continuously check all of the running machines without leaving a few gaps in the process. If there is a breakdown, the replacement of critical components such as turbines, generators, transformers, and important support systems is very complicated and can cost hundreds of thousands of euros. Power companies contract Siemens Energy to operate around 25 power plants worldwide, with a combined capacity of more than 15 gigawatts. It is important for Siemens to be able to guarantee the highest possible operational availability of these facilities.
As a result, experts from Siemens Energy cooperated with the global research department Corporate Technology (CT) in Princeton to develop the mobile monitoring system. STEVE registers malfunctions before they can cause an interruption in operations, thus reducing down time. It is equipped with coin-sized sensors that can be affixed at various places on just about every machine. When STEVE identifies an abnormality, specialists from Siemens can assist operators at the plant with their analysis by cell phone.
The system detects noises, or rather structure-borne sounds, at a rate of almost one million measurements per second - that's 25 times faster than the human ear. Additionally, STEVE is programmed to "learn" which machine noises and vibrations are characteristic of different operational states. After a data collection phase of about one week, STEVE can tell the difference between noises and vibrations that are normal and those that indicate a malfunction. The system is mobile, weatherproof, and easily transferable from one component to another, so it can periodically take measurements on many technical devices. Because of STEVE's ability to learn, it can be installed on practically any machine that emits vibrations.
Explore further: ESA investigates an alternative, environmental-friendly method of corrosion resistance