Fault diagnosis of low-speed heavy load super large rolling bearing based on deep learning
By Simin Li, Hongchao Wang
The conventional eigenvalue alarm mode has a high rate of false alarm and missed alarm for the low-speed heavy load super large rolling bearing. Besides, the traditional signal processing method such as envelope spectral analysis is difficult to extract its fault characteristic frequencies, resulting in a high rate of false diagnosis and missed diagnosis. In order to solve the above problems, an intelligent diagnosis method for the low-speed heavy load super large rolling bearing based on deep learning is proposed. The proposed method mainly utilizes the strong robustness of deep learning algorithm to the quality of original vibration data in the field of fault diagnosis. Firstly, an effective signal acquisition scheme is designed to solve the problem that the signal characteristics of low-speed heavy load super large rolling element bearing are difficult to be acquired. Then, the collected data are randomly divided into training sets, verification sets and test sets by using data enhancement technology. Subsequently, input the divided training set samples into 1-dimensional convolution neural network (1DCNN) deep learning model for learning and training to construct the 1DCNN learning model and set network structure parameters. Meanwhile, the optimal training model is obtained by validating the updating effect of model parameters through validation set. Finally, the test data is input into the trained model to realize intelligent diagnosis. Effectiveness of the proposed method is verified by the vibration data of a wind power main bearing.
A convolutional neural network method based on Adam optimizer with power-exponential learning rate for bearing fault diagnosis
By Youming Wang, Zhao Xiao, Gongqing Cao
Fault diagnosis and health management of bearings in rotating equipment based on vibration analysis – a review
By Adnan Althubaiti, Faris Elasha, Joao Amaral Teixeira
A review on wind turbines gearbox fault diagnosis methods
By H. Gu, W. Y. Liu, Q. W. Gao, Y. Zhang
Bearing fault diagnosis based on improved VMD and DCNN
By Ran Wang, Lei Xu, Fengkai Liu
Mathematical Models in Engineering
Optimization of palm methyl ester and its effect on fatty acid compositions and cetane number
This paper proposes the Taguchi based optimization technique for the production of biodiesel from palm oil. For this purpose, L9 orthogonal array was successfully used for better yield estimation by using Minitab-18 software. Different process variables like molar ratio, catalyst concentration, reaction time and reaction temperature were studied. A predicted yield of 92.06 % was achieved by regression analysis by maintaining the process variables of molar ratio 5:1 (methanol to oil), 4 grams of catalyst concentration (NaOH), 180 minutes of reaction time and 44°C of reaction temperature. Experimentations were conducted on the same process variables and achieved a yield of 91.65 %. By this it is clear that both experimentation and regression analysis by Taguchi are in good agreement with an error of 0.41 % which may be acclaimed as experimental error. The fatty acid compositions (FAC) were also analyzed and it is found that 37.12 % saturated and 62.88 % unsaturated fatty acids present in the palm methyl ester (PME). By using the FAC of PME the Cetane number was predicted as 55.38. The predicted Cetane number from FAC tally with experimental Cetane number. The PME is characterized for different fuel properties by following the international standards. And it is concluded that catalyst concentration and reaction temperature are the important parameters which influence PME yield, Taguchi based optimization technique will helps in predicting the maximum yield with minimum number of experiments.
Performance of PID-Fuzzy control for cab isolation mounts of soil compactors
To improve the soil compactor ride comfort, a combined control method of Fuzzy and PID control is proposed to control the cab isolation system of soil compactor based on the non-linear vehicle dynamic model. The vibration excitation sources are concerned by the vibrator drum and elastoplastic soil (EPS) interactions in the compression process. The power-spectral-density (PSD) and weighted root-mean-square (RMS) of acceleration responses of both the vertical driver’s seat and pitching cab angle are chosen as the objective functions. The research results show that both the PSD and weighted RMS values of the vertical driver’s seat and pitching cab angle are significantly reduced by using the PID-Fuzzy control under various EPSs in the low-frequency region, especially on the EPS with high-density.
Optimization and modelling of mahua oil biodiesel using RSM and genetic algorithm techniques
In this present investigation, four important process parameters of catalyst concentration, molar ratio, reaction time, and reaction temperature were studied and optimized using Box Behnken assisted response surface method (RSM) and Genetic Algorithm (GA) to achieve the maximum mahua oil biodiesel yield. For this purpose, 27 experiments were conducted randomly based on the design matrix using statistical software MiniTab®2019. A maximum yield of 91.32 % is achieved in RSM, catalyst concentration and reaction time are identified as influence parameters in biodiesel yield. GA modelling show an improvement of 4.96 % in biodiesel yield compared to RSM approach. Both techniques are successfully tested in prediction and modelling the biodiesel yield from mahua oil. The obtained biodiesel from the transesterification process is blended with standard diesel fuel at various proportions (B10 to B90) and tested for different fuel properties. All the biodiesel blends are observed within the limits of international standards of ASTMD-6751 and EN-14214. The results indicate that the chosen models are highly accurate in achieving maximum biodiesel yield and mahua biodiesel is recommended as the best alternative fuel to diesel engines without any major modifications in the engine design.
Fault identification and remaining useful life prediction of bearings using Poincare maps, fast Fourier transform and convolutional neural networks
Bearings are integral components of rotating machinery and their failure tends to be a catastrophic failure of the machine. Poincare Maps are used to detect bearing failures using the concept of non-linear dynamics. Each time-domain vibration signature array has its own Poincare Map over a period of time. Fast Fourier Transform (FFT) is a method of analysing the frequency plots of a bearing signature. Convolutional Neural Networks (CNN) process the bearing Continuous Wavelet Transform images and provide the Remaining Useful Life (RUL) of the bearing. The Poincare Maps and FFT plots are used to diagnose the type and location of the fault in the bearing, whereas the CNN helps to provide the fraction of Remaining Useful Life. The study concludes that a combination of Poincare Maps, FFT analysis and Convolutional Neural Networks constitutes a robust and precise method of monitoring bearing conditions.
Mathematical Models in Engineering
Mathematical results and models specifically applicable to engineering science, technology, and their practical applications across various disciplines
Best of Theme
December 31, 2018
An intelligent fault diagnosis method of rotating machinery using L1-regularized sparse filtering
By Weiwei Qian, Shunming Li, Jinrui Wang, Zenghui An, Xingxing Jiang
September 30, 2019
Fault diagnosis method for energy storage mechanism of high voltage circuit breaker based on CNN characteristic matrix constructed by sound-vibration signal
By Shutao Zhao, Erxu Wang, Jiawei Hao
April 2, 2020
Artificial neural network based fault diagnostics for three phase induction motors under similar operating conditions
By Abhisar Chouhan, Purushottam Gangsar, Rajkumar Porwal, Christopher K. Mechefske
December 2, 2021
Mathematical modeling of forced oscillations of semidefinite vibro-impact system sliding along rough horizontal surface
By Vitaliy Korendiy, Volodymyr Gursky, Oleksandr Kachur, Volodymyr Gurey, Oleksandr Havrylchenko, Oleh Kotsiumbas
You might also like
Dynamic analysis of slider-crank mechanism with clearance fault
By Shungen Xiao, Mengmeng Song, Zexiong Zhang
In this paper, the dynamic behavior of the slider-crank mechanism with clearance fault is investigated. The revolute joint with clearance is equivalent to a virtual massless rod, and then the dynamic equation of the crank slider mechanism with clearance is established by the Lagrangian method. In addition, a three-dimensional dynamic model of the crank slider mechanism with clearance is also established by ADAMS. The numerical results show that the clearance affects the displacement and velocity response of the crank-slider mechanism in a weak way, but influences the acceleration response of the mechanism in a significant manner. Due to the existence of the clearance, the revolute joint of the mechanism produces a rub-impact phenomenon, and the larger the clearance, the greater the impact strength. During the rub-impact process, there are three kinds of motion states of separation, collision and contact occur.
Variational mode decomposition denoising combined with the Euclidean distance for diesel engine vibration signal
By Gang Ren, Jide Jia, Jianmin Mei, Xiangyu Jia, Jiajia Han
Variational mode decomposition (VMD) is a recently introduced adaptive signal decomposition algorithm with a solid theoretical foundation and good noise robustness compared with empirical mode decomposition (EMD). There is a lot of background noise in the vibration signal of diesel engine. To solve the problem, a denoising algorithm based on VMD and Euclidean Distance is proposed. Firstly, a multi-component, non-Gauss, and noisy simulation signal is established, and decomposed into a given number K of band-limited intrinsic mode functions by VMD. Then the Euclidean distance between the probability density function of each mode and that of the simulation signal are calculated. The signal is reconstructed using the relevant modes, which are selected on the basis of noticeable similarities between the probability density function of the simulation signal and that of each mode. Finally, the vibration signals of diesel engine connecting rod bearing faults are analyzed by the proposed method. The results show that compared with other denoising algorithms, the proposed method has better denoising effect, and the fault characteristics of vibration signals of diesel engine connecting rod bearings can be effectively enhanced.
Research on unbalance response characteristics of gas turbine blade-disk rotor system
Taken a gas turbine rotor system as the research object, this paper has established dynamical model based on transfer matrix method. Natural vibration characteristics are solved with different transfer matrix models, and are compared with that gotten by finite element method. Then unbalanced response is researched. The effects of eccentric factors are discussed on unbalanced vibration. Furthermore, the application is explored in rotor dynamic balancing. The results show there is much minor deviation between natural frequencies of transfer matrix method and finite element method, and vibration modes of two methods coincide quite well. It illustrates the model established in the research is reliable. For input side of blade-disk rotor, unbalanced vibration occurs more easily in lower frequency. While unbalanced vibration of output side is more sensitive at higher frequency. With the increase of eccentric mass, unbalanced vibration of blade-disk rotor is more heavily, and the displacement rotates around a fixed node. The application of opposite unbalanced mass can decrease the amplitude of vibration in lower frequency. Hence, it can be used to achieve the balance of rotor dynamics. This research provides the theoretical basis and technological fundament for gas turbine rotor dynamics design.
Mathematical modeling of first order process with dead time using various tuning methods for industrial applications
The proposed research involves Mathematical modeling for first order process with dead time using various tuning methods for industrial applications. Different tuning methods are proposed. Proposed method selection depends on plant operating conditions and also depending upon the process dynamics. The PID controller is most widely used for industrial process control. Modeling were developed for modified internal control model [4-6] in this proposed research. The proposed work is the modeling and simulation of three different first order processes with dead time. The standard controller tuning method is used to obtain the steady state response of first order with dead time.