Single-phase 25 kV traction networks of electrified alternating current(AC)railways create electromagnetic fields(EMFs)with significant levels of intensity.The most intense magnetic fields occur when short circuits ex...Single-phase 25 kV traction networks of electrified alternating current(AC)railways create electromagnetic fields(EMFs)with significant levels of intensity.The most intense magnetic fields occur when short circuits exist between the contact wire and rails or ground.Despite the short duration of exposure,they can adversely affect electronic devices and induce significant voltages in adjacent power lines,which is dangerous for operating personnel.Although numerous investigations have focused on modeling the EMF of traction networks and power lines,the challenge of determining the three-dimensional electromagnetic fields near metal supports during the flow of a short-circuit current through them is yet to be resolved.In this case,the field has a complex spatial structure that significantly complicates the calculations of intensities.This study proposes a methodology,algorithms,software,and digital models for determining the EMF in the described emergency scenarios.During the modeling process,the objects being studied were represented by segments of thin wires to analyze the distribution of the electric charge and calculate the intensities of the electric and magnetic fields.This approach was implemented in the Fazonord software,and the modeling results show a substantial increase in EMF levels close to the support,with a noticeable decrease in the levels as the distance from it increases.The procedure implemented in the commercial software Fazonord is universal and can be used to determine electromagnetic fields at any electrical power facility that includes live parts of limited length.Based on the proposed procedure,the EMF near the supports of overhead power lines and traction networks of various designs could be determined,the EMF levels at substations can be calculated,and the influence of metal structures located near traction networks,such as pedestrian crossings at railway stations,can be considered.展开更多
A new method is proposed to assess the condition of structures under unknown support excitation by simultaneously detecting local damage and identifying the support excitation from several structural dynamic responses...A new method is proposed to assess the condition of structures under unknown support excitation by simultaneously detecting local damage and identifying the support excitation from several structural dynamic responses. The support excitation acting on a structure is modeled by orthogonal polynomial approximations, and the sensitivities of structural dynamic response with respect to its physical parameters and orthogonal coefficients are derived. The identification equation is based on Taylor's first order approximation, and is solved with the damped least-squares method in an iterative procedure. A fifteen-story shear building model and a five-story three-dimensional steel frame structure are studied to validate the proposed method. Numerical simulations with noisy measured accelerations show that the proposed method can accurately detect local damage and identify unknown support excitation from only several responses of the structure. This method provides a new approach for detecting structural damage and updating models with unknown input and incomplete measured output information.展开更多
Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine (LRE).Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.T...Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine (LRE).Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.Thus it is important to differentiate abnormal or unknown patterns from normal pattern with novelty detection methods.One-class support vector machine (OCSVM) that has been commonly used for novelty detection cannot deal well with large scale samples.In order to model the normal pattern of the turbopump with OCSVM and so as to monitor the condition of the turbopump,a monitoring method that integrates OCSVM with incremental clustering is presented.In this method,the incremental clustering is used for sample reduction by extracting representative vectors from a large training set.The representative vectors are supposed to distribute uniformly in the object region and fulfill the region.And training OCSVM on these representative vectors yields a novelty detector.By applying this method to the analysis of the turbopump's historical test data,it shows that the incremental clustering algorithm can extract 91 representative points from more than 36 000 training vectors,and the OCSVM detector trained on these 91 representative points can recognize spikes in vibration signals caused by different abnormal events such as vane shedding,rub-impact and sensor faults.This monitoring method does not need fault samples during training as classical recognition methods.The method resolves the learning problem of large samples and is an alternative method for condition monitoring of the LRE turbopump.展开更多
The methods combined by test, field monitoring and theoretical analysis were adopted to do the systemic research on the rock mass from micro-structure to macro-deformation, and rheological model of Jinchuan rock mass ...The methods combined by test, field monitoring and theoretical analysis were adopted to do the systemic research on the rock mass from micro-structure to macro-deformation, and rheological model of Jinchuan rock mass was established to discuss the reasonable supporting time. Results show that supporting after suitable stress and displacement release can benefit for the long-term stability of surrounding rock.展开更多
Real-time video transport over wireless Internet faces many challenges due to the heterogeneous environment including wireline and wireless networks. A robust network condition classification algorithm using multiple ...Real-time video transport over wireless Internet faces many challenges due to the heterogeneous environment including wireline and wireless networks. A robust network condition classification algorithm using multiple end-to-end metrics and Support Vector Machine (SVM) is proposed to classify different network events and model the transition pattern of network conditions. End-to-end Quality-of-Service (QoS) mechanisms like congestion control, error control, and power control can benefit from the network condition information and react to different network situations appropriately. The proposed network condition classifica- tion algorithm uses SVM as a classifier to cluster different end-to-end metrics such as end-to-end delay, delay jitter, throughput and packet loss-rate for the UDP traffic with TCP-friendly Rate Control (TFRC), which is used for video transport. The algorithm is also flexible for classifying different numbers of states representing different levels of network events such as wireline congestion and wireless channel loss. Simulation results using network simulator 2 (ns2) showed the effectiveness of the proposed scheme.展开更多
The State Council Information Office published on September 27, 2009, a white paper entitled China' s Policies toward Minorities and Common Prosperity and Development of All Nationalities. The white paper says that t...The State Council Information Office published on September 27, 2009, a white paper entitled China' s Policies toward Minorities and Common Prosperity and Development of All Nationalities. The white paper says that the policies well fit the national conditions and are correct. They are instrumental to promoting harmonious coexistence of all nationalities that work with one heart and one mind and in mutual cooperation; they have helped open up a good national situation characterized by economic development, political stability,展开更多
Downhole working conditions of sucker rod pumping wells are automatically identified on a computer from the analysis of dynamometer cards. In this process, extraction of feature parameters and pattern classification a...Downhole working conditions of sucker rod pumping wells are automatically identified on a computer from the analysis of dynamometer cards. In this process, extraction of feature parameters and pattern classification are two key steps. The dynamometer card is firstly divided into four parts which include different production information according to the "four point method" used in actual oilfield production, and then the moment invariants for pattern recognition are extracted. An improved support vector machine (SVM) method is used for pattern classification whose error penalty parameter C and kernel function parameter g are optimally chosen by the particle swarm optimization (PSO) algorithm. The simulation results show the method proposed in this paper has good classification results.展开更多
The submersible pumping unit is a new type of pumping system for lifting formation fluids from onshore oil wells, and the identification of its working condition has an important influence on oil production. In this p...The submersible pumping unit is a new type of pumping system for lifting formation fluids from onshore oil wells, and the identification of its working condition has an important influence on oil production. In this paper we proposed a diagnostic method for identifying the working condition of the submersible pumping system. Based on analyzing the working principle of the pumping unit and the pump structure, different characteristics in loading and unloading processes of the submersible linear motor were obtained at different working conditions. The characteristic quantities were extracted from operation data of the submersible linear motor. A diagnostic model based on the support vector machine (SVM) method was proposed for identifying the working condition of the submersible pumping unit, where the inputs of the SVM classifier were the characteristic quantities. The performance and the misjudgment rate of this method were analyzed and validated by the data acquired from an experimental simulation platform. The model proposed had an excellent performance in failure diagnosis of the submersible pumping system. The SVM classifier had higher diagnostic accuracy than the learning vector quantization (LVQ) classifier.展开更多
Gear fault diagnosis technologies have received rapid development and been effectively implemented in many engineering applications.However,the various working conditions would degrade the diagnostic performance and m...Gear fault diagnosis technologies have received rapid development and been effectively implemented in many engineering applications.However,the various working conditions would degrade the diagnostic performance and make gear fault diagnosis(GFD)more and more challenging.In this paper,a novel model parameter transfer(NMPT)is proposed to boost the performance of GFD under varying working conditions.Based on the previous transfer strategy that controls empirical risk of source domain,this method further integrates the superiorities of multi-task learning with the idea of transfer learning(TL)to acquire transferable knowledge by minimizing the discrepancies of separating hyperplanes between one specific working condition(target domain)and another(source domain),and then transferring both commonality and specialty parameters over tasks to make use of source domain samples to assist target GFD task when sufficient labeled samples from target domain are unavailable.For NMPT implementation,insufficient target domain features and abundant source domain features with supervised information are fed into NMPT model to train a robust classifier for target GFD task.Related experiments prove that NMPT is expected to be a valuable technology to boost practical GFD performance under various working conditions.The proposed methods provides a transfer learning-based framework to handle the problem of insufficient training samples in target task caused by variable operation conditions.展开更多
Aiming at solving shield attitude rectification failure problem,a method of shield working condition classification based on support vector data description( SVDD) was introduced. Shield attitude mechanics model conta...Aiming at solving shield attitude rectification failure problem,a method of shield working condition classification based on support vector data description( SVDD) was introduced. Shield attitude mechanics model containing priori knowledge was helpful to feature selection. SVDD handled the one class classification problem and a decision function for attitude rectification was proposed. Experimental results indicate that the method is able to accomplish the shield attitude working condition classification.展开更多
Pd/LaxPbyMnOz, Pd/C, Pd/molecular sieve and Pd-heteropoly acid catalysts for direct synthesis of diphenyl carbonate (DPC) by heterogeneous catalytic reaction were compared and the results of DPC synthesis indicated th...Pd/LaxPbyMnOz, Pd/C, Pd/molecular sieve and Pd-heteropoly acid catalysts for direct synthesis of diphenyl carbonate (DPC) by heterogeneous catalytic reaction were compared and the results of DPC synthesis indicated that the catalyst Pd/LaxPbyMnOz had higher activity. The Pd/LaxPbyMnOz catalyst and the support was characterized by XRD, SEM and TEM, the main phase was La0.62Pb0.38MnO3 and the average diameter could be about 25.4 nm. The optimum conditions for synthesis of DPC with Pd/LaxPbyMnOz were determined by orthogonal experiments and the experimental results showed that reaction temperature was the first factor of effect on the selectivity and yield of DPC, and the concentration of O2 in gas phase also had significant effect on selectivity of DPC. The optimum reaction conditions were catMyst/phenol mass ratio 1 to 50, pressure 4.5 MPa,volume concentration of O2 25%, reaction temperature 60° and reaction time 4 h. The maximum yield and average selectivity could reach 13% and 97% respectively in the batch operation.展开更多
In order to realize the real-time and precise test for a weapon system of a certain type of fighter,a signal classification method according to attributes is proposed,common input channels for multiple signals are con...In order to realize the real-time and precise test for a weapon system of a certain type of fighter,a signal classification method according to attributes is proposed,common input channels for multiple signals are configured optimally,and a test adapter and an adaptive signal conditioning module is designed. The hardware of conditioning module can be configured flexibly and the programmable test range can be adjusted owing to programmable multiplexer. An FPGA adaptive filter is designed by the calculated filter coefficient vectors with LMS method to solve the problem of parallel test of fighter weapon system in electromagnetic interference environment. The adaptive signal conditioning technology is characterized by high efficiency,precision and integration. Its application makes the test system successful to conduct real-time and parallel test for a weapon system,which is developed based on VXI bus and virtual-instrument technology.展开更多
Aging adults with chronic conditions rely heavily on an informal network of caregivers to remain within their communities of choice. This reliance can take a significant toll on caregivers through the lens of physical...Aging adults with chronic conditions rely heavily on an informal network of caregivers to remain within their communities of choice. This reliance can take a significant toll on caregivers through the lens of physical and psychological problems, financial issues, and social isolation. These variables may then lead to less desirable outcomes for care recipients. This review highlights existing support services in their many forms, including: psychosocial interventions, environmental interventions, respite care, and health information technology as a method of delivery. Given the current trend with informal caregivers assuming increased responsibility in healthcare, programs and services supporting these caregivers must be understood and trialed to ensure that their needs are not overlooked.展开更多
The fault detection and diagnosis of diesel engine valve clearance can effectively improve the availability and safety of diesel engine and have extremely important value and significance.Diesel engines generally oper...The fault detection and diagnosis of diesel engine valve clearance can effectively improve the availability and safety of diesel engine and have extremely important value and significance.Diesel engines generally operate in various stable operating conditions,which have important influence on the fault diagnosis.However,many fault diagnosis methods have been put forward under specific stable operating condition based on vibration signal.As the result of great impact caused by operating conditions,corresponding diagnosis models cannot deal with the fault diagnosis under different operating conditions with required accuracy.In this paper,a fault diagnosis of diesel engine valve clearance under variable operating condition based on soft interval support vector machine(SVM)is proposed.Firstly,the fault features with weak condition sensitivity have been extracted according to the influence analysis of fault on vibration signal.Moreover,soft interval constraint has been applied to SVM algorithm to reduce the random influence of vibration signal on fault features.In addition,different machine learning algorithms based on different feature sets are adopted to conduct the fault diagnosis under different operating conditions for comparison.Experimental results show that the proposed method is applicable for fault diagnosis under variable operating condition with good accuracy.展开更多
文摘Single-phase 25 kV traction networks of electrified alternating current(AC)railways create electromagnetic fields(EMFs)with significant levels of intensity.The most intense magnetic fields occur when short circuits exist between the contact wire and rails or ground.Despite the short duration of exposure,they can adversely affect electronic devices and induce significant voltages in adjacent power lines,which is dangerous for operating personnel.Although numerous investigations have focused on modeling the EMF of traction networks and power lines,the challenge of determining the three-dimensional electromagnetic fields near metal supports during the flow of a short-circuit current through them is yet to be resolved.In this case,the field has a complex spatial structure that significantly complicates the calculations of intensities.This study proposes a methodology,algorithms,software,and digital models for determining the EMF in the described emergency scenarios.During the modeling process,the objects being studied were represented by segments of thin wires to analyze the distribution of the electric charge and calculate the intensities of the electric and magnetic fields.This approach was implemented in the Fazonord software,and the modeling results show a substantial increase in EMF levels close to the support,with a noticeable decrease in the levels as the distance from it increases.The procedure implemented in the commercial software Fazonord is universal and can be used to determine electromagnetic fields at any electrical power facility that includes live parts of limited length.Based on the proposed procedure,the EMF near the supports of overhead power lines and traction networks of various designs could be determined,the EMF levels at substations can be calculated,and the influence of metal structures located near traction networks,such as pedestrian crossings at railway stations,can be considered.
基金National Natural Science Foundation of China Under Grant No.50579008Joint Research Fund for Overseas Chinese, Hong Kong and Macao Young Scholars Under Grant No.50429802+1 种基金Program for New Century Excellent Talents in University by State Education Commission Under Grant No.NCET-04-0323a research grant from the Hong Kong Polytechnic University
文摘A new method is proposed to assess the condition of structures under unknown support excitation by simultaneously detecting local damage and identifying the support excitation from several structural dynamic responses. The support excitation acting on a structure is modeled by orthogonal polynomial approximations, and the sensitivities of structural dynamic response with respect to its physical parameters and orthogonal coefficients are derived. The identification equation is based on Taylor's first order approximation, and is solved with the damped least-squares method in an iterative procedure. A fifteen-story shear building model and a five-story three-dimensional steel frame structure are studied to validate the proposed method. Numerical simulations with noisy measured accelerations show that the proposed method can accurately detect local damage and identify unknown support excitation from only several responses of the structure. This method provides a new approach for detecting structural damage and updating models with unknown input and incomplete measured output information.
基金supported by National Natural Science Foundation of China (Grant No. 50675219)Hu’nan Provincial Science Committee Excellent Youth Foundation of China (Grant No. 08JJ1008)
文摘Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine (LRE).Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.Thus it is important to differentiate abnormal or unknown patterns from normal pattern with novelty detection methods.One-class support vector machine (OCSVM) that has been commonly used for novelty detection cannot deal well with large scale samples.In order to model the normal pattern of the turbopump with OCSVM and so as to monitor the condition of the turbopump,a monitoring method that integrates OCSVM with incremental clustering is presented.In this method,the incremental clustering is used for sample reduction by extracting representative vectors from a large training set.The representative vectors are supposed to distribute uniformly in the object region and fulfill the region.And training OCSVM on these representative vectors yields a novelty detector.By applying this method to the analysis of the turbopump's historical test data,it shows that the incremental clustering algorithm can extract 91 representative points from more than 36 000 training vectors,and the OCSVM detector trained on these 91 representative points can recognize spikes in vibration signals caused by different abnormal events such as vane shedding,rub-impact and sensor faults.This monitoring method does not need fault samples during training as classical recognition methods.The method resolves the learning problem of large samples and is an alternative method for condition monitoring of the LRE turbopump.
基金Supported bythe National Natural Science Foundation of China (50904024) the State Key Laboratory Research Fund of Coal Resources and Mine Safety of China University of Mining & Technology (10KF02) the Doctoral Fund of Henan Polytechnic University (B2009-66)
文摘The methods combined by test, field monitoring and theoretical analysis were adopted to do the systemic research on the rock mass from micro-structure to macro-deformation, and rheological model of Jinchuan rock mass was established to discuss the reasonable supporting time. Results show that supporting after suitable stress and displacement release can benefit for the long-term stability of surrounding rock.
基金Project supported by the Croucher Foundation Fellowship fromHong Kong, China
文摘Real-time video transport over wireless Internet faces many challenges due to the heterogeneous environment including wireline and wireless networks. A robust network condition classification algorithm using multiple end-to-end metrics and Support Vector Machine (SVM) is proposed to classify different network events and model the transition pattern of network conditions. End-to-end Quality-of-Service (QoS) mechanisms like congestion control, error control, and power control can benefit from the network condition information and react to different network situations appropriately. The proposed network condition classifica- tion algorithm uses SVM as a classifier to cluster different end-to-end metrics such as end-to-end delay, delay jitter, throughput and packet loss-rate for the UDP traffic with TCP-friendly Rate Control (TFRC), which is used for video transport. The algorithm is also flexible for classifying different numbers of states representing different levels of network events such as wireline congestion and wireless channel loss. Simulation results using network simulator 2 (ns2) showed the effectiveness of the proposed scheme.
文摘The State Council Information Office published on September 27, 2009, a white paper entitled China' s Policies toward Minorities and Common Prosperity and Development of All Nationalities. The white paper says that the policies well fit the national conditions and are correct. They are instrumental to promoting harmonious coexistence of all nationalities that work with one heart and one mind and in mutual cooperation; they have helped open up a good national situation characterized by economic development, political stability,
基金support from the Key Project of the National Natural Science Foundation of China (61034005)Postgraduate Scientific Research and Innovation Projects of Basic Scientific Research Operating Expenses of Ministry of Education (N100604001)
文摘Downhole working conditions of sucker rod pumping wells are automatically identified on a computer from the analysis of dynamometer cards. In this process, extraction of feature parameters and pattern classification are two key steps. The dynamometer card is firstly divided into four parts which include different production information according to the "four point method" used in actual oilfield production, and then the moment invariants for pattern recognition are extracted. An improved support vector machine (SVM) method is used for pattern classification whose error penalty parameter C and kernel function parameter g are optimally chosen by the particle swarm optimization (PSO) algorithm. The simulation results show the method proposed in this paper has good classification results.
文摘The submersible pumping unit is a new type of pumping system for lifting formation fluids from onshore oil wells, and the identification of its working condition has an important influence on oil production. In this paper we proposed a diagnostic method for identifying the working condition of the submersible pumping system. Based on analyzing the working principle of the pumping unit and the pump structure, different characteristics in loading and unloading processes of the submersible linear motor were obtained at different working conditions. The characteristic quantities were extracted from operation data of the submersible linear motor. A diagnostic model based on the support vector machine (SVM) method was proposed for identifying the working condition of the submersible pumping unit, where the inputs of the SVM classifier were the characteristic quantities. The performance and the misjudgment rate of this method were analyzed and validated by the data acquired from an experimental simulation platform. The model proposed had an excellent performance in failure diagnosis of the submersible pumping system. The SVM classifier had higher diagnostic accuracy than the learning vector quantization (LVQ) classifier.
基金Supported by National Natural Science Foundation of China(Grant No.51835009).
文摘Gear fault diagnosis technologies have received rapid development and been effectively implemented in many engineering applications.However,the various working conditions would degrade the diagnostic performance and make gear fault diagnosis(GFD)more and more challenging.In this paper,a novel model parameter transfer(NMPT)is proposed to boost the performance of GFD under varying working conditions.Based on the previous transfer strategy that controls empirical risk of source domain,this method further integrates the superiorities of multi-task learning with the idea of transfer learning(TL)to acquire transferable knowledge by minimizing the discrepancies of separating hyperplanes between one specific working condition(target domain)and another(source domain),and then transferring both commonality and specialty parameters over tasks to make use of source domain samples to assist target GFD task when sufficient labeled samples from target domain are unavailable.For NMPT implementation,insufficient target domain features and abundant source domain features with supervised information are fed into NMPT model to train a robust classifier for target GFD task.Related experiments prove that NMPT is expected to be a valuable technology to boost practical GFD performance under various working conditions.The proposed methods provides a transfer learning-based framework to handle the problem of insufficient training samples in target task caused by variable operation conditions.
基金National Basic Research Program of China ( 973 Program) ( No. 2007CB714006)
文摘Aiming at solving shield attitude rectification failure problem,a method of shield working condition classification based on support vector data description( SVDD) was introduced. Shield attitude mechanics model containing priori knowledge was helpful to feature selection. SVDD handled the one class classification problem and a decision function for attitude rectification was proposed. Experimental results indicate that the method is able to accomplish the shield attitude working condition classification.
基金National Natural Science Foundation of China(No.20076036Tianjin University C1 National Laboratory Project
文摘Pd/LaxPbyMnOz, Pd/C, Pd/molecular sieve and Pd-heteropoly acid catalysts for direct synthesis of diphenyl carbonate (DPC) by heterogeneous catalytic reaction were compared and the results of DPC synthesis indicated that the catalyst Pd/LaxPbyMnOz had higher activity. The Pd/LaxPbyMnOz catalyst and the support was characterized by XRD, SEM and TEM, the main phase was La0.62Pb0.38MnO3 and the average diameter could be about 25.4 nm. The optimum conditions for synthesis of DPC with Pd/LaxPbyMnOz were determined by orthogonal experiments and the experimental results showed that reaction temperature was the first factor of effect on the selectivity and yield of DPC, and the concentration of O2 in gas phase also had significant effect on selectivity of DPC. The optimum reaction conditions were catMyst/phenol mass ratio 1 to 50, pressure 4.5 MPa,volume concentration of O2 25%, reaction temperature 60° and reaction time 4 h. The maximum yield and average selectivity could reach 13% and 97% respectively in the batch operation.
基金Sponsored by the Key Equipment Research Project of Air Force of China (KJZ06119)
文摘In order to realize the real-time and precise test for a weapon system of a certain type of fighter,a signal classification method according to attributes is proposed,common input channels for multiple signals are configured optimally,and a test adapter and an adaptive signal conditioning module is designed. The hardware of conditioning module can be configured flexibly and the programmable test range can be adjusted owing to programmable multiplexer. An FPGA adaptive filter is designed by the calculated filter coefficient vectors with LMS method to solve the problem of parallel test of fighter weapon system in electromagnetic interference environment. The adaptive signal conditioning technology is characterized by high efficiency,precision and integration. Its application makes the test system successful to conduct real-time and parallel test for a weapon system,which is developed based on VXI bus and virtual-instrument technology.
文摘Aging adults with chronic conditions rely heavily on an informal network of caregivers to remain within their communities of choice. This reliance can take a significant toll on caregivers through the lens of physical and psychological problems, financial issues, and social isolation. These variables may then lead to less desirable outcomes for care recipients. This review highlights existing support services in their many forms, including: psychosocial interventions, environmental interventions, respite care, and health information technology as a method of delivery. Given the current trend with informal caregivers assuming increased responsibility in healthcare, programs and services supporting these caregivers must be understood and trialed to ensure that their needs are not overlooked.
基金Supported by the National Key Research and Development Plan(No.2016YFF0203305)the Fundamental Research Funds for the Central Universities(No.JD1912,ZY1940)Double First-rate Construction Special Funds(No.ZD1601).
文摘The fault detection and diagnosis of diesel engine valve clearance can effectively improve the availability and safety of diesel engine and have extremely important value and significance.Diesel engines generally operate in various stable operating conditions,which have important influence on the fault diagnosis.However,many fault diagnosis methods have been put forward under specific stable operating condition based on vibration signal.As the result of great impact caused by operating conditions,corresponding diagnosis models cannot deal with the fault diagnosis under different operating conditions with required accuracy.In this paper,a fault diagnosis of diesel engine valve clearance under variable operating condition based on soft interval support vector machine(SVM)is proposed.Firstly,the fault features with weak condition sensitivity have been extracted according to the influence analysis of fault on vibration signal.Moreover,soft interval constraint has been applied to SVM algorithm to reduce the random influence of vibration signal on fault features.In addition,different machine learning algorithms based on different feature sets are adopted to conduct the fault diagnosis under different operating conditions for comparison.Experimental results show that the proposed method is applicable for fault diagnosis under variable operating condition with good accuracy.