Consequently, it is crucial to forecast equipment life. Early identification of failure and damage is often critical in determining the Remaining Useful Life (RUL) of the equipment. If the equipment’s life can be measured with accuracy ahead of time, necessary maintenance and management measures can be taken to ensure proper reliability and effectiveness of the equipment. With continued operation and progress in time, the probability of system failure increases, eventually resulting in system degradation leading to a catastrophic failure. This stable health status continues to persist until a critical time and eventually reduces when an early incipient fault condition manifests. During the initial stage of usage, the system’s health status will be generally stable, and every operational trajectory will have its fixed health level. At the initial level of a complex system’s lifetime, maintenance of the system is not necessary as the components of the system will be in proper working condition. A new model of analyzing sequential data to enable pattern recognition is required.

Finding the relation between independent variables and outcome variable has to be done using multiple models. They result from time-series measurements at successive time intervals. Most of the data points collected from significant sources of IoT applications are not independent and identically distributed. As the data is generated from sensors of various research fields and the plethora of digital machines, high-performance sequence methods are needed to apply on them. Time series of sensors are to be analyzed to find the patterns in the data, which are useful to generate alarms or trigger indications to avoid catastrophic failure. Hence, there is a need to develop methods to evaluate the sequence of events and their consequences on the device pertaining to its status. Such a study can lead to avoiding the sudden breakdown of machines during operation. The aim of finding out the health status of a machine is to maximize its operational availability by adopting functions of condition monitoring, state assessment, prognostics, failure progression, and diagnostics. To extract the health status of a machine under consideration, a systematic relation is to be established between the failure mechanisms and the life-cycle management of a system. The objective of IoT systems is primarily to monitor the failure mechanisms to illustrate the way that complex machines run for ages and fail.

Machine condition monitoring based on sensor data is crucial in generating insights from sensor data in multiple domains. Keywords: Multivariate sensor data TimeImageNet Remaining life estimation machine learning 2D image Convolutional Neural NetworkĬommercial Modular Aero-Propulsion System Simulationĭata Generated from Internet of Things (IoT) systems is increasing enormously due to the availability of various sensors and low-cost Internet. Better accuracy in the classification of components is achieved using the TimeImagenet-based approach. The current methodology is useful in avoiding the health index generation for predicting the remaining useful life of the industrial components. Further, the proposed net is also used for predicting the number of engines that would fail in the given time window. Using these created images, a model for estimating the remaining life of the aircraft is developed. The created images were fed to Convolutional Neural Network, which includes both time variation and space variation of each sensed parameter.

Converted Images would represent the health of a system in higher-dimensional space. The time series of each sensor is converted to a 2D image with a specific time window. The methodology is based on different sampling sizes to obtain optimum results with great accuracy. In the present work, the non-linear multivariate sensor data is used to understand the health status and anomalous behavior. The run to failure data includes states like new installation, stable operation, first reported issue, erroneous operation, and final failure. Aircraft engine data from run to failure is used in the current study. These models are further used to predict the possible downtime for proactive action on the system condition. Sensor data of all possible states of a system are used for building machine learning models. Abstract: Internet of Things systems generate a large amount of sensor data that needs to be analyzed for extracting useful insights on the health status of the machine under consideration.

Timenet time series classification