A Novel Generative Method for Machine Fault Diagnosis

Deep learning is widely used in fault diagnosis of mechanical equipment and has achieved good results. However, these deep learning models require a large number of labeled samples for training, which is difficult to obtain enough labeled samples in the actual production process. However, it is easier to obtain unlabeled samples in the industrial environment. To overcome this problem, this paper proposes a novel method to generative enough label samples for training deep learning models. Unlike the generative adversarial networks, which required complex computing time, the calculation of the proposed novel generative method is simple and effective. First, we calculate the Euclidean distance between the training sample and the test sample; then, the weight coefficient between the training sample and the test sample is settled to generate pseudosamples; finally, combine with the pseudosamples, the deep learning method is training for machine fault diagnosis. In order to verify the effectiveness of the proposed method, two experiment datasets with planetary gearboxes and wind gearboxes are carried out with different activation functions. Experimental results show that the proposed method is effective for most activation function models.

A Generative Method for Indoor Localization Using Wi-Fi Fingerprinting

Indoor localization is an enabling technology for pervasive and mobile computing applications. Although different technologies have been proposed for indoor localization, Wi-Fi fingerprinting is one of the most used techniques due to the pervasiveness of Wi-Fi technology. Most Wi-Fi fingerprinting localization methods presented in the literature are discriminative methods. We present a generative method for indoor localization based on Wi-Fi fingerprinting. The Received Signal Strength Indicator received from a Wireless Access Point is modeled by a hidden Markov model. Unlike other algorithms, the use of a hidden Markov model allows ours to take advantage of the temporal autocorrelation present in the Wi-Fi signal. The algorithm estimates the user’s location based on the hidden Markov model, which models the signal and the forward algorithm to determine the likelihood of a given time series of Received Signal Strength Indicators. The proposed method was compared with four other well-known Machine Learning algorithms through extensive experimentation with data collected in real scenarios. The proposed method obtained competitive results in most scenarios tested and was the best method in 17 of 60 experiments performed.