A deep learning-based multi-sensor data fusion method for degradation monitoring of ball screws

As ball screw has complex structure and long range of distribution, single signal collected by one sensor is difficult to express its condition fully and accurately. Multi-sensor data fusion usually has a better effect compared with single signal. Multi-sensor data fusion based on neural network(BP) is a commonly used multi-sensor data fusion method, but its application is limited by local optimum problem. Aiming at this problem, a multi-sensor data fusion method based on deep learning for ball screw is proposed in this paper. Deep learning, which consists of unsupervised learning and supervised learning, is the development and evolution of traditional neural network. It can effectively alleviate the optimization difficulty. Parallel superposition on frequency spectra of signals is directly done in the proposed deep learning-based multi-sensor data fusion method, and deep belief networks(DBN) are established by using fused data to adaptively mine available fault characteristics and automatically identify the degradation condition of ball screw. Test is designed to collect vibration signals of ball screw in 7 different degradation conditions by using 5 acceleration sensors installed on different places. The proposed fusion method is applied in identifying the degradation degree of ball screw in the test to demonstrate its efficacy. Finally, the multi-sensor data fusion based on neural network is also applied in degradation degree monitoring. The monitoring accuracy of deep learning-based multi-sensor data fusion is higher compared with that of neural network-based multi-sensor data fusion, which means the proposed method has more superiority.

[1]  Geoffrey E. Hinton.  A Practical Guide to Training Restricted Boltzmann Machines , 2012, Neural Networks: Tricks of the Trade.

[2]  Yee Whye Teh,et al.  A Fast Learning Algorithm for Deep Belief Nets , 2006, Neural Computation.

[3]  Paul Smolensky,et al.  Information processing in dynamical systems: foundations of harmony theory , 1986 .

[4]  Geoffrey E. Hinton,et al.  Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[5]  Yoshua Bengio,et al.  Greedy Layer-Wise Training of Deep Networks , 2006, NIPS.

[6]  Dong Yu,et al.  Deep Learning and Its Applications to Signal and Information Processing [Exploratory DSP] , 2011, IEEE Signal Processing Magazine.

[7]  M. S. Safizadeh,et al.  Using multi-sensor data fusion for vibration fault diagnosis of rolling element bearings by accelerometer and load cell , 2014, Inf. Fusion.

[8]  Pingfeng Wang,et al.  Failure diagnosis using deep belief learning based health state classification , 2013, Reliab. Eng. Syst. Saf..

[9]  Geoffrey E. Hinton,et al.  A Learning Algorithm for Boltzmann Machines , 1985, Cogn. Sci..

[10]  Xin Zhou,et al.  Deep neural networks: A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data , 2016 .

[11]  Andrew D. Ball,et al.  An approach to fault diagnosis of reciprocating compressor valves using Teager-Kaiser energy operator and deep belief networks , 2014, Expert Syst. Appl..

[12]  R J. Kuo,et al.  Multi-sensor integration for on-line tool wear estimation through radial basis function networks and fuzzy neural network , 1999, Neural Networks.