The invention provides an adaptive manifold probability distribution-based bearing fault diagnosis method, including constructing transferable domains and transfer tasks; converting a data sample in each transfer task into frequency domain data via Fourier transform, inputting the frequency domain data into a GFK algorithm model, and calculating a manifold feature representation matrix related to a bearing fault in each transfer task by using the GFK algorithm model; calculating a cosine distance between centers of a target domain and a source domain in each transfer task according to a manifold feature representation, and defining a target function of in-domain classifier learning; then solving the target function, to obtain a probability distribution matrix of the target domain; and selecting a label corresponding to the largest probability value corresponding to each data sample in the target domain from the probability distribution matrix as a predicted label of the data sample in the target domain.

A vibration detection and correction device includes a housing and a sensor. The housing includes one or more inputs, a processor, and a memory connected to the processor. The sensor is connected to at least one input of the one or more inputs. The sensor is configured to measure rotation data of a shaft of a rotary machine. The processor is configured to determine dynamic vibration data of the shaft based on the rotation data, determine a coarse runout amount of the shaft based on the rotation data measured each time the rotary machine is coming to a stop, and correct the dynamic vibration data by removing the coarse runout amount from the dynamic vibration data so as to obtain coarse-adjusted vibration data. The memory is configured to store the coarse-adjusted vibration data.

A method for sensing damage of bearing of engine using a vibration signal may include separating a vibration signal of an engine sensed by a vibration sensor installed at one side of the engine of a vehicle into a vibration signal by combustion knocking and a vibration signal of a bearing installed between a crank pin and a connecting rod; extracting, by a signal processing filter, a signal of a predetermined natural frequency band from the vibration signal of the bearing; determining whether the vibration signal of the bearing is higher than a predetermined bearing damage threshold, in a predetermined engine state condition in order to sense breakage of the bearing during operation of the engine; and confirming that the bearing has been damaged.

A method for sensing damage of bearing of engine using a vibration signal may include separating a vibration signal of an engine sensed by a vibration sensor installed at one side of the engine of a vehicle into a vibration signal by combustion knocking and a vibration signal of a bearing installed between a crank pin and a connecting rod; extracting, by a signal processing filter, a signal of a predetermined natural frequency band from the vibration signal of the bearing; determining whether the vibration signal of the bearing is higher than a predetermined bearing damage threshold, in a predetermined engine state condition in order to sense breakage of the bearing during operation of the engine; and confirming that the bearing has been damaged.

A vibration measurement and analysis system identifies faulty bearings in a machine based on spectral vibration data. The system includes vibration sensors attached to the machine that generate vibration signals. A vibration data collector generates vibration spectral data based on the vibration signals. The vibration spectral data comprises vibration amplitude versus frequency data that includes peak amplitudes at corresponding peak frequencies. At least some of the peak amplitudes are associated with vibration generated by the faulty bearings. A vibration analysis computer processes the vibration spectral data to (1) locate the largest peak amplitudes, (2) search a bearing fault frequency library to generate a list of identified bearings having bearing fault frequencies matching the peak frequencies of the largest peak amplitudes, (3) determine a normalized accuracy error for each of the identified bearings, and (4) select from the list one of the identified bearings having a smallest normalized accuracy error.

A vibration measurement and analysis system identifies faulty bearings in a machine based on spectral vibration data. The system includes vibration sensors attached to the machine that generate vibration signals. A vibration data collector generates vibration spectral data based on the vibration signals. The vibration spectral data comprises vibration amplitude versus frequency data that includes peak amplitudes at corresponding peak frequencies. At least some of the peak amplitudes are associated with vibration generated by the faulty bearings. A vibration analysis computer processes the vibration spectral data to (1) locate the largest peak amplitudes, (2) search a bearing fault frequency library to generate a list of identified bearings having bearing fault frequencies matching the peak frequencies of the largest peak amplitudes, (3) determine a normalized accuracy error for each of the identified bearings, and (4) select from the list one of the identified bearings having a smallest normalized accuracy error.

A remaining life prediction system includes: a training information obtaining unit that obtains training deterioration amount information, a training feature vector, and a training remaining life; a first regression model training unit that trains a first regression model that estimates a deterioration amount based on the training feature vector and the training deterioration amount information; a second regression model training unit that trains a second regression model that estimates a remaining life based on the training deterioration amount information and the training remaining life; an evaluation information obtaining unit that obtains an evaluation feature vector; and a remaining life derivation unit that estimates evaluation deterioration amount information by the first regression model, and derives a remaining life by the second regression model.

A remaining life prediction system includes: a training information obtaining unit that obtains training deterioration amount information, a training feature vector, and a training remaining life; a first regression model training unit that trains a first regression model that estimates a deterioration amount based on the training feature vector and the training deterioration amount information; a second regression model training unit that trains a second regression model that estimates a remaining life based on the training deterioration amount information and the training remaining life; an evaluation information obtaining unit that obtains an evaluation feature vector; and a remaining life derivation unit that estimates evaluation deterioration amount information by the first regression model, and derives a remaining life by the second regression model.

Apparatus and associated methods relate to a vibrational sensing system (VSS) including an accelerometer and a data processor, which determines an “operational state” of a mechanical drive unit, the processor further employing the “operational state” to gate learning of long-term vibrational data to exclude collection of non-operational data, the long-term data collected to calculate alarm thresholds. For example, vibrations from a target motor are sensed by a coupled accelerometer. Vibrational data from the accelerometer is fed into a data processor which determines the operational state of the motor. The operational state (e.g., on/off indication) may gate data collection such that data is only acquired during on-time, which may advantageously create accurate baselines from which alarm thresholds may be generated, and nuisance alarms may be avoided.

Apparatus and associated methods relate to a vibrational sensing system (VSS) including an accelerometer and a data processor, which determines an “operational state” of a mechanical drive unit, the processor further employing the “operational state” to gate learning of long-term vibrational data to exclude collection of non-operational data, the long-term data collected to calculate alarm thresholds. For example, vibrations from a target motor are sensed by a coupled accelerometer. Vibrational data from the accelerometer is fed into a data processor which determines the operational state of the motor. The operational state (e.g., on/off indication) may gate data collection such that data is only acquired during on-time, which may advantageously create accurate baselines from which alarm thresholds may be generated, and nuisance alarms may be avoided.