... inspection costs. An outcome of those regularly performed assessments is the basis for engineers to choose the most appropriate maintenance strategy on a case by case basis. Abstract The key idea is to deliver an automated, machine-learning (ML) based algorithm to combine the results from different non...

... variations that may occur throughout the year. Machine learning algorithms can be developed to recognize patterns in historical readings in relation to other operation or relatable data – such as soil type, seasonality and precipitation – ultimately providing prescriptive insights into what future thresholds...

...EXPERIMENTAL APPROACH Machine Learning Modeling Approach Based on the data collected from MSPs during simplified environmental cycles with medium salt loading, machine learning was used to build regression models to predict free and galvanic corrosion current using environmental severity data...

... at a perceived risk of external corrosion. Despite its simplicity, the prototype achieves a high performance, and with further development it is anticipated that the network could be employed for even more sophisticated integrity management decisions. data analytics Bayesian networks machine learning in-line...

... a similar technological approach with machine learning to make the estimated corrosion rates actionable to operators by focusing on wall loss percentages. This paper proposes an algorithmic approach that combines both historical ILI assessments with environmental conditions to yield specific risk...

... presented herein investigates the impact of dataset size on Deep Learning for automatic detection of corrosion on steel assets. Dataset creation is typically one of the first steps when applying Machine Learning methods to a new task; and the real-world performance of models hinges on the quality...

.... Therefore, they are significant when determining if mass transfer is controlling the erosion-corrosion process. To begin building a machine learning/multi-label SVM model, an original dataset was formed by calculating all of these parameters for erosion-corrosion tests performed previously by the authors...

.... Since CGRs are the most fundamental piece of information required for a pipeline corrosion growth assessment, the study aims to use Bayesian networks to enhance CGR predictions. Bayesian Networks Algorithms that can automatically learn patterns from historical data are known as ‘machine learning...

... and using the information available. Risk based inspection Expert system Machine learning Predictive analytics Prediction Efficiency Cost-reduction Data-driven Predictive modelling Statistical model Decision support INTRODUCTION Across much of industry, there is a drive towards using or at least...

... methodology improvement and future development of the machine learning within the tool. Furthermore, a separately developed feature uses digital twins of the assets, linked with assets’ integrity data and other relevant data, providing more efficient communication and work processes. The goal of the tool...

...DATASETS USED FOR MODELING To train the Machine Learning models, a large array of different dependent variables including atmospheric conditions, human activity, and alternating current (AC) interference are included. Atmospheric Conditions Table 1 Quantities of Data with and without ILI...

... inhibitors which are responsible for inhibition of steel in acidic solution. The framework consists of modules like data preprocessing, descriptor selection and model building. A robust predictive model for multiple class of corrosion inhibitors was developed using advanced machine learning algorithm...

... Network Dirichlet Distribution Predictive Analytics Risk Based Inspection Risk Management Corrosion Prediction Inspection Optimisation Piping Vessels Artificial Intelligence Machine Learning Industrial Revolution INTRODUCTION Industry 4.0 is a term that in the widest sense describes...

... rates using the k_p-p model. We systematically analyzed the correlation between elemental alloy compositions and the manually fitted k_p and p values to select high-ranking features to be included in a machine learning analysis. The machine...