Data Mining with Shapelets for Predicting Valve Failures in Gas Compressors

Abstract

Gas compressor failures are frequently caused by breakdown of valves. Since production is dependent on rotating equipment, it is useful to minimize downtime caused by such valve failures, and try to predict them in advance. This is a challenging problem, which we address using Big Data analysis of the data gathered by a large number of sensors deployed on various parts of the compressor. These sensors take periodic readings (at every few minutes) of various physical properties of the compressors including motor winding temperatures, compressor vibrations, and pressure and temperature for both suction and discharge at various compression stages. We frame this problem as a multivariate time series classification task, and propose a novel machine learning approach to solve it.Our proposed approach is based on the concept of shapelets, which are discriminative subsequences extracted from time series. This approach does not make assumptions about the nature of the dataset (crucial for real industrial datasets) and has very fast classification times. These shapelets act as a 'signature' capturing the characteristics and differences between sensor data related to normal valve function versus failed valve function. Shapelets are increasingly being used for univariate (single dimension data read by one sensor) time series data mining. But there have been few efforts to solve the problem of multivariate time series classification using shapelets due to the additional challenges emanating from multiple sensors in terms of the size and variety of data. Specifically, the existing approaches make the assumption that the reading of sensors are independent, which is not the case for sensor data in gas compressors as variation or anomaly in a valve affect the reading of adjacent sensors. Since all the sensors record data synchronized in time, the temporal dependencies across them need to be captured.In this work, we propose a method, which attempts to incorporate these dependencies into the final shapelet-based classification framework. We achieve this using a heuristic of inter-leaving time series data across the sensors. This helps us reduce the multivariate time series data to a univariate format such that existing univariate shapelet extraction methods can be applied directly on the data. We evaluate our approach on real sensor data taken from gas compressors in an oil field in North America. Our results illustrate that time series approaches based on shapelet mining are valuable for fast prediction of failures from sensor data in oil and gas fields. These approaches provide key insights into the functioning of the individual sensors as well as deliver a visual aid to domain experts for further root cause analysis.