Anomaly detection in oil refinery equipment usingmulti-layer perceptron neural network learning

Equipment reliability is a critical aspect in the oil and gas industry because the failure of suchequipment can lead to catastrophic results at various levels and create huge costs for theindustry. In oil refineries as process industries, if a failure occurs at any stage of production, itcan have a wide impact on the entire process and even stop production. In these industries,we need safe, accurate and cost-effective operation of equipment, which requires reliableestimation and identification of failure in order to improve performance and ensure efficient andeffective maintenance strategy. Therefore, failure detection in the oil and gas industry is criticalto increasing component life and reducing unexpected equipment failures, thereby preventingcostly plant shutdowns and equipment damage. In this article, we present an intelligentmachine learning model using multilayer perceptron neural network classification and PSOparticle optimization algorithm to detect unexpected failures. In this article, process equipmentworks in two classes, normal and abnormal. The goal is to identify errors that may not beidentified due to large volume, variety and human error.