Sonal Dixit

Intelligent Condition Based Monitoring Using Acoustic Signals for Air Compressors

Intelligent fault diagnosis of machines for early recognition of faults saves industry from heavy losses occurring due to machine breakdowns. This paper proposes a process with a generic data mining model that can be used for developing acoustic signal-based fault diagnosis systems for reciprocating air compressors. The process includes details of data acquisition, sensitive position analysis for deciding suitable sensor locations, signal pre-processing, feature extraction, feature selection, and a classification approach. This process was validated by developing a real time fault diagnosis system on a reciprocating type air compressor having 8 designated states, including one healthy state, and 7 faulty states. The system was able to accurately detect all the faults by analyzing acoustic recordings taken from just a single position. Additionally, thorough analysis has been presented where performance of the system is compared while varying feature selection techniques, the number of selected features, and multiclass decomposition algorithms meant for binary classifiers.

Generating feature sets for fault diagnosis using denoising stacked auto-encoder

Recent advancements in sensor technologies and data driven model based techniques have made intelligent diagnostic systems prominent in machine maintenance frameworks of industries. The performance of such systems immensely relies upon the quality of features extracted and the classifier model learned. Traditionally features were handcrafted, where engineers would manually design them with statistical parameters and signal transforms based energy distribution analysis. Recently, deep learning techniques have shown new ways of obtaining useful feature representation that provide state of the art results in image and speech processing applications. This paper first presents a brief survey of traditional handcrafted features and later presents a short analysis of handcrafted features v/s features learned by deep neural networks (DNN), for doing fault diagnosis. The DNN based features in this paper were generated in 3 phases: 1) extracted handcrafted features using traditional techniques 2) initialized the weights of DNN by learning de-noising sparse auto-encoders with the handcrafted features in unsupervised fashion and 3) applied two generic fine tuning heuristics that tailor DNNs weights to give good classification performance. The experimentation and analysis were performed on 5 datasets: one each on Air compressor monitoring, Drill bit monitoring and Steel plate monitoring, and two on bearing fault monitoring data. The results clearly show the prospects of DNN obtaining good feature representations and good classification performance. Further, it also finds that Fast Fourier Transform based features with DNN are more suited for Support Vector Machine as classifier than Random Forest.

Android app for intelligent CBM

Smartphone applications have changed the traditional way of using cellphones. They are not only used for calling and messaging, but also for specialized and multiple engineering applications like face recognition, navigation, driving style recognition, etc. In this paper we present a scalable android application which enables smartphones to diagnose faults in rotating machines. With this ability of fault detection, it can be used for Condition Based Monitoring (CBM), which is a popular maintenance strategy used in industry. The smartphone performs fault detection by analyzing acoustic signatures generated by a rotating machine in running condition. The acoustic signature is recorded and analyzed by the inbuilt microphone and processor respectively, thus enabling the smartphone to be a complete fault detection and recognition system. The advantage of this is that we get an industrial fault detection system which is portable, economically viable and easily deployable. The performance of the system has been assessed by training and testing on an industrial air compressor acoustic data for three different machine conditions. Observed fault recognition accuracies were approximately 93%.

Detection and classification for faults in drilling process using vibration analysis

In this era of flexible manufacturing systems, increase in demand of automatic and unattended machining process is very high. Thus arise the need for proper online tool condition monitoring methods, in order to minimize error and waste of work-material. In this study, Support Vector Machine (SVM), Artificial Neural Network (ANN) and Bayes classifier are used to develop such a system for automatic drilling operations with the help of vibration signals. The performances of models generated by these classifiers are compared with each other in order to establish the best method. As the vibration signals were acquired under different drilling parameters, this study also tries to understand the events in drilling process that help in ease of fault classification. Three different kinds of wears were studied and later compared to understand the degree or magnitude of effect of wears on the drilling process and signals.

Cost-benefit and reliability analysis of prognostic health management systems using fuzzy rules

Prognostic Health Management (PHM) systems have been found to be profitable in most industrial ventures as compared to Time-based Maintenance (TBM) or Corrective Maintenance (CM). However, while undertaking a Cost-Benefit Analysis (CBA), the profit obtained for a PHM system cannot be accurately known in all cases due to the vagueness present in certain parameters involved in the analysis. Our paper focuses on modeling the vagueness or uncertainty in these quantities using Fuzzy Rule-Based Systems (FRBS). Reliability being a major desirable objective, the influence of the quality of PHM on reliability analysis has been outlined. Fuzzy rules have also been employed considering both tangible and intangible benefits to present an accurate comprehensive model of the overall benefit. To evaluate the practicality of the cost-benefit analysis, an optimization condition has been derived, maximizing net benefit for given constraints. Over and above this analysis, how far the PHM system is advisable is determined by computing the risk factor associated with a PHM venture.

Windows mobile and tablet app for acoustic signature based machine health monitoring

With the availability of new APIs and IDE, App development has become easier than ever before. Fast processors, better RAM, better OS functionality and better storage options have given Smartphones and Tablets the capability to perform most functions for which previously a larger computing device was used. In this paper, we discuss about an application for performing Condition Based Monitoring (CBM) meant for Industrial Machines developed on two platforms namely Windows Mobile OS and Windows Tablet OS (Windows 8 Pro). The challenges faced in each platform and how they were encountered for making this into a successful app on Windows based smartphone & tablet for industrial use, have been thoroughly discussed in this paper. The application uses acoustic recordings and data mining techniques to distinguish between healthy and faulty states of a machine. The developed smartphone application was trained successfully on a reciprocating type air compressor for distinguishing Leakage Inlet Valve fault from Healthy state and was able to accurately detect the fault.

Thermal imaging based condition based monitoring on Android platform

The use of thermal images for condition based monitoring is becoming very popular in industries. This paper presents a simple approach for condition based monitoring of machines using thermal Images on Android Platform. This approach is non-contact, fast and precise for CBM of rotating and non-rotating machines. The proposed approach has been successfully implemented on Android smartphone for CBM of prototype machine. The application uses standard OpenCV library functions to implement processing logic. For this the temperature profile of overall machine from different views was captured and analyzed using thermal images. Regression model was developed to find relation between local temperature and pixel intensity which helped in recognizing the condition of prototype machine.