Haifeng Huang

Multifeatures Fusion and Nonlinear Dimension Reduction for Intelligent Bearing Condition Monitoring

Condition-based maintenance is critical to reduce the costs of maintenance and improve the production efficiency. Data-driven method based on neural network (NN) is one of the most used models for mechanical components condition recognition. In this paper, we introduce a new bearing condition recognition method based on multifeatures extraction and deep neural network (DNN). First, the method calculates time domain, frequency domain, and time-frequency domain features to represent characteristic of vibration signals. Then the nonlinear dimension reduction algorithm based on deep learning is proposed to reduce the redundancy information. Finally, the top-layer classifier of deep neural network outputs the bearing condition. The proposed method is validated using experiment test-bed bearing vibration data. Meanwhile some comparative studies are performed; the results show the advantage of the proposed method in adaptive features selection and superior accuracy in bearing condition recognition.

Machinery vibration signal denoising based on learned dictionary and sparse representation

Mechanical vibration signal denoising has been an import problem for machine damage assessment and health monitoring. Wavelet transfer and sparse reconstruction are the powerful and practical methods. However, those methods are based on the fixed basis functions or atoms. In this paper, a novel method is presented. The atoms used to represent signals are learned from the raw signal. And in order to satisfy the requirements of real-time signal processing, an online dictionary learning algorithm is adopted. Orthogonal matching pursuit is applied to extract the most pursuit column in the dictionary. At last, denoised signal is calculated with the sparse vector and learned dictionary. A simulation signal and real bearing fault signal are utilized to evaluate the improved performance of the proposed method through the comparison with kinds of denoising algorithms. Then Its computing efficiency is demonstrated by an illustrative runtime example. The results show that the proposed method outperforms current algorithms with efficiency calculation.

Screw Performance Degradation Assessment Based on Quantum Genetic Algorithm and Dynamic Fuzzy Neural Network

To evaluate the performance of ball screw, screw performance degradation assessment technology based on quantum genetic algorithm (QGA) and dynamic fuzzy neural network (DFNN) is studied. The ball screw of the CINCINNATIV5-3000 machining center is treated as the study object. Two Kistler 8704B100M1 accelerometers and a Kistler 8765A250M5 three-way accelerometer are installed to monitor the degradation trend of screw performance. First, screw vibration signal features are extracted both in time domain and frequency domain. Then the feature vectors can be obtained by principal component analysis (PCA). Second, the initialization parameters of the DFNN are optimized by means of QGA. Finally, the feature vectors are inputted to DFNN for training and then get the screw performance degradation model. The experiment results show that the screw performance degradation model could effectively evaluate the performance of NC machine screw.

Artificial neural network for screw life prediction

The life change of the screw in High-end CNC machine tool in the process has some features such as non-linear, dynamic and uncertainty. A screw online life prediction system was designed to monitor the performance of lead screw by vibration sensors and temperature sensors which installed at different locations of Lead Screw Pair and reflected the trend of changes of different processing conditions, lifting wavelet transform was used to extract the most sensitive characteristics of screw performance. RBF neural network was used to build the non-linear relationship between screw vibration signal changes and screw life. Eventually constructed screw life prediction model based on RBF neural network bring into effect of effective assessment of residual life of lead screw. The results show that the performance degradation model can predict the remaining life of screw effectively.

Tool wear monitoring based on novel evolutionary artificial neural networks

In order to improve the accuracy and speed of on-line tool wear monitoring system, an evolutionary neural network using variable string genetic algorithm (VGA) was developed to construct the relations between tool wear and signal features extracted from cutting forces, vibrations, and acoustic emission by different signal processing methods. The system could automatically evolve the appropriate architecture of neural network and find a near-optimal set of connection weights globally. Then the conformable connection weights for model could be found with back-propagation (BP) algorithm, the multi-model finally completed calculation of tool wear. The experimental results show that the system proposed in the paper has higher classification precision and calculating speed.

Screw performance degradation model based on novel neural networks

A screw performance degradation model based on neural network which was optimized by improved genetic algorithm was proposed to predict screw life accurately and provide active maintenance proof. Key factors which related to screw life were analyzed by screw motion mechanism. Three vibration sensors were installed on different position of screw and vibration signal were processed by EMD, time domain analysis, frequency domain analysis and wavelet packet analysis. The most sensitive features to screw life were selected by correlation coefficient and evaluation index. The relation between screw life and features was built by neural network that constructed by BP training algorithm, and screw life was calculated. The long practical results show that the screw life prediction model can meet the need of active maintenance and reduce maintenance cost.

Tool Wear Monitoring Based on Localized Fuzzy Neural Networks for Turning Operation

On-line tool wear monitoring is essential to automatic machining process. In order to predict tool wear accurately and reliably under different cutting conditions, a novel tool wear monitoring system (TWMS) is proposed by using localized fuzzy neural networks(LFNN) in this study which may improve classification accuracy of tool states and the computing speed compared with BPNN and normal fuzzy neural networks in the process of turning. By analyzing cutting forces signals and acoustic emission signals in time domain, frequency domain, and time-frequency domain, a series of features that sensitive to tool states were selected as inputs of neural networks according to synthesis coefficient. The nonlinear relations between tool wear and features were modeled by using integrated neural network (INN) that constructed and optimized through LFNN trained by an adaptive learning algorithm. The experimental results show that the monitoring system based on LFNN is provided with high precision, rapid computing speed and good multiplication.