Hilbert transform technique for signal demodulation and application to valve diagnostics

Proper maintenance of motor-operated valves in the power and process industries is critical to the functioning and operability of the plant. Non-intrusive diagnostics of motor-operated valves has received considerable interest, since the current methods and procedures to maintain these valves need improvements in terms of automating the diagnostics procedure. Analysis of the motor current drawn by a motor-operated valve provides insight into the condition of the valve. The motor current needs to be demodulated before it can be used as a signature to perform non-intrusive diagnostics of the valve. A methodology to perform digital demodulation of double-sideband suppressed-carrier type of amplitude modulated signals, using the Hilbert transform, has been developed and is applied in this research project. Advantages and limitations of the proposed method of signal demodulation are established. The SMB000 type of motor-operated valve was installed in the Nuclear Engineering Department of the University of Tennessee. Motor current drawn by the SMB000 was digitized for several test runs, under normal and simulated abnormal conditions. The proposed method of signal demodulation was applied to demodulate motor current signatures obtained from the SMB000. The expert system for motor-operated valve diagnostics, MOVEX, was studied and suitably modified to provide diagnostics information for different families of motor-operated valves. Motor current signatures obtained from the SMB000 and demodulated by the proposed method were analyzed by the MOVEX system. The signal preprocessing technique presented here is very effective and its suitability is demonstrated using both simulated signals and valve operational data.