A Scalable Sensor Network for Pipeline Vibration Analysis

点击翻译按钮获取中文摘要

The pipelines used in the mining, oil and gas industry deteriorate over time. Pipeline maintenance is expensive and often based on best guesses of the condition of individual pipeline segments. We propose to use a network of wireless sensors that monitor and report the vibrations sensed at each pipeline segment. The relative amplitudes of the vibrations at various frequencies, and across the length of a pipeline, form a vibration profile for the pipeline. We will develop the sensor platform and protocols for the acquisition, communication, and study of the vibration profiles and their changes over time as they reflect the changing condition of the pipeline. Using the platform, pipeline operators will be able to determine trouble spots as early as possible. Machine learning techniques will be used to classify the profiles as normal or problematic, to infer the most likely location of recent change that could mean need for maintenance, and to determine the points in time where the profiles change to a degree that warrants inspection and maintenance. We will use advanced data compression and aggregation techniques adapted to the limited computation and communication resources of the wireless sensors nodes. We will integrate a widely available Internet-of-Things (IoT) protocol, LoRaWAN, with 5G IoT protocols to aggregate and relay the sensed data in compressed form to monitoring stations. With the help of the partner organizations, we will test the platform and its networking aspects to ensure that it meets the performance envelope of the heterogeneous, LoRaWAN and 5G, communication environment. The overall objective is to automate the process of identifying, early and effectively, pipeline segments in need of maintenance, thus reducing the corresponding operating costs.