Development Of The Operation Diagnosis System For Centrifugal Pump Based On Sensorless Technology

A centrifugal pump long-term operates under off-design operation conditions would lead to flow instability,low efficiency,additional energy loss,increasing wear or even potential mechanical damage.Early detection and remedial measures are of great importance to its operation efficiency and safety.Traditional diagnosis methods for centrifugal pumps usually require dozens of sensors to be installed for data acquisition,which brings the limitations of increasing diagnosis cost,intrusive installation and hard maintenance.Sensorless technology is a cost-effective and non-intrusive diagnosis technology for motor-driven devices.It takes the driving induction motors as torque sensors,and extracts the operation information of machines from the input electrical signals.Considering its great advantage and application potential,this dissertation aimed at developing a reliable diagnosis methods for centrifugal pumps based on such technology.Theoretical analysis and experiments were conducted to study the sensorless diagnosis mechanism.Operation condition indicators for centrifugal pumps derived from the motor stator current signature were established.Furthermore,the intelligent diagnosis system based on the proposed indicators was designed and manufactured.The main research achievements and technical innovations of this dissertation are as follows:1.Considering the coupling effects among current,flux and torque for motor are rather complicated to analyze,transfer regularity of torque fluctuation characteristics was investigated based on the simplified mathematical model of induction motor,and firstly validated by the variable torque experiment.Theoretical analysis and experimental verification indicated that the torque fluctuation frequency would be reflected in the form of sidebands in motor stator current through power frequency modulation.The stronger the torque fluctuation intensity is,the higher the reflected sidebands' amplitudes are.2.Existing researches on the operation diagnosis for centrifugal pumps tend to overlook the excessive calculation complexity brought by the current developed methods.Meanwhile,most study focuses on fault diagnosis and stays short of identifying the specific internal flow status of centrifugal pumps.In response to the above limitations,this dissertation set up several operation condition indicators with less computation cost for centrifugal pumps combining with the flow features analysis under different operation conditions.More specifically,physical implication of root mean square(RMS),impulse factor,kurtosis factor,total harmonic distortion(THD),signal-to-noise ratio(SNR),signal to noise and distortion ratio(SINAD)were refered and normalized to bulid the diagnosis indicators,which were employed for motor stator current analysis to reflect the flow status and detect the off-design operation for centrifugal pumps.Experiments with a centrifugal pump operating under off-design flowrates and different cavitation stages were conducted.Experimental results show that the RMS indicator of motor stator current shares a similar tendency versus flowrate with that of the shaft power.Impulse factor and kurtosis factor indicators of motor stator current are helpful for identifying the optimal operation range of the centrifugal pump.THD,SNR and SINAD indicators of motor stator current all show good detection ability on both off-design flowrates and different cavitation stages.3.Based on the proposed indicators,an intelligent operation condition diagnosis system for centrifugal pumps was developed through embedded design.Function structure of each hardware module was constructed in detail.Working logic of the diagnosis system was analyzed and the corresponding function realization programs were written.The field test results shows that this operation diagnosis system could finish the work of indicators calculation with error less than 1.8%.