SYSTEM AND METHOD FOR TURBOMACHINERY PREVENTIVE MAINTENANCE AND ROOT CAUSE FAILURE DETERMINATION

• US 20190191287A1
• Filed: 02/26/2019
• Published: 06/20/2019
• Est. Priority Date: 01/17/2015
• Status: Active Grant

• Current Assignee: Machinesense LLC
• Original Assignee: Machinesense LLC
• Inventors: Pal, Biplab, Zinski, James
• Granted Patent: US 10,638,295 B2
• Time in Patent Office: Days
• CPC Class Codes: G05B 19/4185 characterised by the network communication
• CPC Class Codes: G05B 2219/31457 Factory remote control, monitoring through internet
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
• CPC Class Codes: G05B 23/0272 Presentation of monitored results, e.g. selection of status reports to be displayed; Filtering information to the user
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06
• CPC Class Codes: H04W 4/80 Services using short range communication, e.g. near-field communication [NFC], radio-frequency identification [RFID] or low energy communication
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
• CPC Class Codes: Y02P 90/80 Management or planning

Abstract

Predicting maintenance needs and analyzing preventative maintenance requirements in electrically powered turbomachinery with multi-parameter sensors and power quality sensors, both of the Fog-type, providing time domain output data and transforming data samples into the frequency domain to detect a root cause of failure of the machinery.

Claims

1. A method for determining root causes of failure in electrically powered turbomachinery, comprising:

a) sensing physical parameters, in the course of operation of the machine, including at least one of motor speed, vibration, magnetic field presence and strength, temperature, relative humidity, infrared radiation, input voltage, input current and phase characteristics of input power;

b) providing time domain output data indicative of the sensed physical parameters to first and second parallel processing blocks;

c) processing the time domain output data in the second processing block by;

i) selecting a parameter data sample;

ii) transforming the data sample into the frequency domain by performing Fourier transformation thereon;

iii) extracting metadata for selected characteristics of the sensed physical parameters from the transformed data;

d) continuously monitoring the time domain data and performing trend analysis thereon for selected parameter characteristics; and

e) upon detecting deviation of a selected characteristic in the time domain data from a preselected value, sampling the frequency domain metadata for a preselected time interval and then analyzing the frequency domain selected parameter characteristic according to a preselected algorithm relative to a base value to detect as a root cause of failure in the turbomachine any variation of the parameter from a base value.

2. A system for determining root causes of failure in electrically powered turbomachinery, comprising:

a) sensor device connected to an electric motor powering the turbomachinery, for sensing physical parameters including at least one of motor speed, vibration, magnetic field presence and strength, temperature, relative humidity, infrared radiation, input voltage, input current and phase characteristics of input power, and providing output data indicative of the sensed physical parameters;

b) a communication network connected to the device for receiving the output data therefrom;

c) a first computing device connected to the communication network for receiving the output data therefrom, having first and second parallel processing blocks each receiving time domain sensed data from the network, the second processing block processing the time domain data by;

i) selecting a sensed physical parameter data sample size and duration to be within the processing capacity of the first computing device;

ii) transforming the data into the frequency domain by performing Fourier transformation thereon;

iii) extracting metadata for selected characteristic(s) of the sensed parameter(s) from the transformed data;

d) an edge cloud computing device connected to the first computing device by the communication network, receiving from the first processing block time domain data for analysis thereof and receiving from the second processing block frequency domain metadata extracted from the transformed data, for;

i) continuously monitoring and performing trend analysis on the time domain data for selected parameter characteristics;

ii) upon detecting deviation of a selected characteristic in the time domain data from a preselected value, sampling the frequency domain metadata for a preselected time interval and then analyzing the selected parameter characteristic according to preselected algorithm relative to a base value to detect as a root cause of failure in the turbomachinery any variation of the parameter from the base value.

3. A system identifying root causes of vibration induced failures in electrically powered turbo machinery, comprising:

a) a vibration sensor connected to an electric motor powering the turbomachinery and providing time domain output data indicative of sensed vibration parameters;

b) a communication network connected to the sensor for receiving the output data therefrom;

c) a first computing device connected to the communication network for receiving the output data therefrom, having first and second parallel processing blocks each receiving the time domain data from the network, the second processing block processing the time domain data by;

i) selecting a data sample of preselected size and duration;

ii) transforming the data into the frequency domain by performing Fourier transformation thereon;

iii) extracting metadata for selected vibration parameters from transformed data;

d) an edge cloud computing device connected to the first computing device by the communication network, receiving from the first processing block time domain data for analysis thereof and receiving from the second processing block frequency domain metadata extracted from the transformed data, for;

i) continuously monitoring and performing trend analysis on the time domain data for selected characteristics indicative of status of the vibration parameters;

ii) upon detecting variance of the selected characteristic in the time domain data from a preselected value, sampling the frequency domain metadata for a preselected time interval and analyzing the sampled frequency domain metadata for the selected characteristics according to a preselected algorithms relative to base values of the characteristics to detect as a root cause of a vibration induced turbomachinery failure any deviations of the characteristics from the base values.

4. In a system for determining root causes of failure in electrically powered turbomachinery, having a sensor device connected to an electric motor powering the turbomachinery, for sensing at least one physical parameter, and providing output data indicative of the sensed physical parameter, and a communication network connected to the device for receiving the output data therefrom, the improvement consisting of:

a) a first computing device connected to the communication network for receiving the output data therefrom, having parallel processing blocks each receiving time domain sensed data from the network, one of the processing blocks processing the time domain data by;

i) selecting a sample size and duration to be within the processing capacity of the first computing device;

ii) transforming the data into the frequency domain by performing Fourier transformation thereon;

iii) extracting metadata for selected characteristics of the sensed parameter from the transformed data; and

b) an edge cloud computing device connected to the first computing device by the communication network, receiving from the first processing block time domain data for analysis thereof and receiving from the second processing block frequency domain metadata extracted from the transformed data, for;

i) continuously monitoring and performing trend analysis on the time domain data for at least one characteristic of the physical parameter;

ii) upon detecting deviation of the selected characteristic in the time domain data from a preselected value, sampling the frequency domain metadata for a preselected time interval and then analyzing a the selected parameter characteristic according to a preselected algorithm relative to a base value to detect as a root cause of failure in the turbomachinery any variation of the parameter from a base value.

5. A method for predicting maintenance needs, analyzing preventative maintenance e requirements and determining toot causes of failure in electrically powered turbomachinery, comprising:

a) mounting a multi-parameter sensor, of the Fog-type having micro-computer component, on the turbomachine;

b) attaching a power quality sensor, of the Fog-type having a micro-computer component, to the electrical line providing electrical power to the turbomachine;

c) transmitting data from the multi-parameter sensor and date from the power quality sensor in the time domain to a computing device which is connected to the cloud via the Internet;

d) providing time domain output data indicative of the sensed physical parameters to first and second parallel processing blocks;

e) processing the time domain output data in the second processing block by;

i) selecting sensed physical parameter time domain data samples;

ii) transforming the selected data samples into the frequency domain by performing Fourier transformation thereon;

iii) extracting metadata for selected characteristics of the sensed physical parameters from the transformed data samples;

f) continuously monitoring the time domain output data and performing trend analysis thereon for selected characteristics of the sensed physical parameters; and

g) upon detecting deviation of a selected characteristic in the time domain data from a preselected value, sampling the frequency domain metadata for the selected characteristic for a preselected time interval and then analyzing the frequency domain selected parameter characteristic data according to a preselected algorithm relative to a base value to detect as a root cause of failure in the turbomachine any variation of the parameter from the base value.

6. The method of claim 5 wherein the time domain output data further comprises collected vibration data and the method further comprises:

a) analyzing in the time domain the collected vibration data to determine long term trends of vibration parameters including but not limited to amplitude, frequency, voltage, phase, and harmonics;

b) correlating the resulting vibration parameter long term trends with different kinds of turbomachine rotor failures;

c) using the results of the correlation to generate an early indication of turbomachine degradation to operators; and

d) providing the results of the correlation to the multi-parameter sensor and to the power quality sensor for subsequent computations thereby.

7. The method of claim 6 further comprising performing frequency domain Fourier transform-based analysis on successive batches of the collected vibration data to ascertain root causes of turbomachine rotor failure until computed error is reduced to within an acceptable statistical limit.

8. The method of claim 7 further comprising:

a) transforming the results of the correlation into the frequency domain;

b) analyzing the infrared radiation temperature and magnetic field parameters time domain data in the frequency domain in the edge or public cloud to provide further details as to root cause of failure in the turbomachine.

9. The method of claim 8 further comprising in the edge or public cloud, correlating the parameter data for harmonics, noise an phase imbalance in the time domain, with the temperature and magnetic field data results and applying frequency domain analysis thereto to isolate an exact cause of degradation of turbomachine performance.

10. The method of claim 9 further comprising:

a) in the edge or public cloud further correlating the result obtained by correlating the parameter data for harmonics, noise an phase imbalance in the time domain, with the temperature and magnetic field data results;

b) applying frequency domain analysis thereto;

c) accepting use data furnished by a mobile/web feedback device and a rule-based and/or machine learning based recommendation engine to define appropriate degradation mitigation steps; and

d) furnishing those steps to third party customer and the like via the Internet or other wireless communication.