In recent years, learning algorithms have been proved to learn abstract hierarchical pattern in the data and provides a better generalization. In this paper, intricate fuzzy patterns in the data are learned with the help of fuzzy network using Mamdani fuzzy inference system. The underlying architecture constituted with layers of Mamdani fuzzy inference system as nodes is called Mamdani Fuzzy Network (MFN). This architecture has an input layer, hidden layers and an output layer. Each layer has Mamdani fuzzy inference system as its basic building unit. A training is carried out using Broyden-Fletcher-Goldfarb-Shanno (BFGS) optimization algorithm to achieve the identification of all parameters in the architecture according to training data. The proposed architecture is implemented in two class air compressor vibration data containing 1200 samples with 5 features. These features are passed to fuzzy network to learn abstract fuzzy representations in the form of multiple distinct fuzzy rule bases which intelligible to human begins. The identified fuzzy rule bases consist of linguistic information of IF-THEN rules with both premise and consequent parts as linguistic labels. The effectiveness of proposed methodology is validated on vibration data sets collected from air compressor. These vibration data sets have been recorded from most sensitive positions of air compressor under various health conditions.

Condition based monitoring, Data acquisition, Deep learning, Fuzzy inference system, Mamdani fuzzy network
2018 Prognostics and System Health Management Conference, PHM-Chongqing 2018
Department of Mechanical and Aerospace Engineering

Sharma, A.K. (Arun K.), Singh, V. (Vikas), Verma, N.K. (Nishchal K.), & Liu, J. (2019). Condition Based Monitoring of Machine Using Mamdani Fuzzy Network. In Proceedings - 2018 Prognostics and System Health Management Conference, PHM-Chongqing 2018 (pp. 1159–1163). doi:10.1109/PHM-Chongqing.2018.00204