公开(公告)号：US20240085274A1

公开(公告)日：2024-03-14

申请号：US18462471

申请日：2023-09-07

Applicant: UNIVERSITY OF SOUTH CAROLINA

Inventor： GUANGXING NIU ,  BIN ZHANG

IPC: G01M13/045 ,  G06N3/045 ,  G06N3/08

CPC classification number: G01M13/045 ,  G06N3/045 ,  G06N3/08

Abstract: A system and method concerns accurate bearing fault diagnosis and prognosis (FDP), critical for optimal maintenance schedules, safety and reliability. Existing methods face challenges: the bearing condition is healthy in most of the service time, so it is critical to detect the occurrence of faults and the start point for prognosis in real applications. Due to differences in manufacturing quality, assembly quality, and different operating conditions, it is difficult to describe the fault dynamic using one single fault model. A hybrid Bayesian estimation-based bearing FDP framework with fault detection and automatic fault model selection is disclosed. A convolutional neural network is used to detect fault and select the appropriate fault dynamic model. To improve performance with different bearings under different operating conditions, continuous wavelet coefficient matrices power spectrum of vibration are fused with operating conditions to build information maps for fault detection and model selection. After a fault is detected, a Bayesian estimation based FDP method is triggered to estimate the fault state and predict the remaining useful life. In the prognostic process, Dempster-Shafer theory is employed to fuse prediction results from different models if necessary.

公开(公告)号：US20230351177A1

公开(公告)日：2023-11-02

申请号：US18191991

申请日：2023-03-29

Applicant: UNIVERSITY OF SOUTH CAROLINA

Inventor： GUANGXING NIU ,  BIN ZHANG

IPC: G06N3/08 ,  G06N3/0464 ,  G05B23/02

CPC classification number: G06N3/08 ,  G06N3/0464 ,  G05B23/0283

Abstract: Deep learning-based diagnosis methods currently face some challenges and open problems. First, domain knowledge of fault modes and operating conditions are not integrated in most existing approaches, which results in low diagnosis accuracy and training efficiency. Second, existing methods treat all features with indiscriminate attention, which causes unnecessary computation and even false diagnosis results in some cases. Third, multi-task diagnosis becomes more important for health maintenance. To address these challenges, a deep residual convolutional neural network is provided with an enhanced discriminate feature learning capability and information fusion for multi-task bearing fault diagnosis. Domain knowledge is integrated with monitoring data to build the information map. Two attention modules are introduced to enhance the discriminate feature learning ability, and two classifiers are employed for multi-task diagnosis, providing significant improvements in diagnostic accuracy and training efficiency.

公开(公告)号：US20230003802A1

公开(公告)日：2023-01-05

申请号：US17854980

申请日：2022-06-30

Applicant: UNIVERSITY OF SOUTH CAROLINA

Inventor： ENHUI LIU ,  BIN ZHANG

IPC: G01R31/367 ,  G01R31/3835 ,  G01R31/392

Abstract: Method provides accurate state-of-health (SOH) diagnostics and prognostics during the whole-life-service of a lithium-ion battery by considering the effects of state- of-charge (SOC) and SOH on certain parameters (such as consideration of nonlinearity of the terminal voltage) during the process of SOC diagnostics and prognostics. The method integrates Lebesgue sampling and equivalent circuit model (ECM) analysis, which greatly decreases computation cost and uncertainty accumulation to provide efficient acquisition of open circuit voltage (OCV) determinations for the ECM process. The OCV curve of the battery was obtained during Hybrid Pulse Power Characterization testing by fitting a series of selected OCV points after enough rest of the subject battery. Identified parameters of ECM are updated according to terminal voltage measurement to enable accurate SOC estimation and prediction during the period from full charge to full discharge of the battery. Parameter identification is re-conducted and an initial condition for SOC estimation is updated according to SOH to enable accurate SOC estimation during the whole-life-service of battery.

公开(公告)号：US20230375636A1

公开(公告)日：2023-11-23

申请号：US18317472

申请日：2023-05-15

Applicant: UNIVERSITY OF SOUTH CAROLINA

Inventor： GUANGXING NIU ,  BIN ZHANG

IPC: G01R31/392 ,  G01R31/367 ,  G01R31/36

CPC classification number: G01R31/392 ,  G01R31/367 ,  G01R31/3648

Abstract: Fault diagnosis and prognosis (FDP) is critical for ensuring system reliability and reducing operation and maintenance (O&M) costs. Lebesgue sampling based FDP (LS-FDP) is an event-based approach with the advantages of cost-efficiency, uncertainty management, and less computation. In previous works, LS-FDP approaches are mainly model-based. However, fault dynamic modeling is difficult and time consuming for some complex systems and this severely hinders the applications of LS-FDP. To address this problem, this present disclosure presents a data-driven based LS-FDP framework in which deep belief networks (DBN) and particle filter (PF) are integrated to achieve fault state estimation and remaining useful life (RUL) prediction. In the proposed approach, DBN learns the state evolution model and the Lebesgue time transition model, which are used as diagnostic and prognostic models in PF for FDP. The proposed approach has higher efficiency in terms of computation and better performance in terms of FDP accuracy and precision.