A Hierarchical Approach to Improve the Interpretability of Causality Maps for Plant-Wide Fault Identification

Modern mineral processing plants utilise fault detection and diagnosis to minimise time spent under faulty conditions. However, a fault originating in one plant section (PS) can propagate throughout the entire plant, obscuring its root cause. Causality analysis identifies the cause–effect relationships between process variables and presents them in a causality map to inform root cause identification. This paper presents a novel hierarchical approach for plant-wide causality analysis that decreases the number of nodes in a causality map, improving interpretability and enabling causality analysis as a tool for plant-wide fault diagnosis. Two causality maps are constructed in subsequent stages: first, a dimensionally reduced, plant-wide causality map used to localise the fault to a PS; second, a causality map of the identified PS used to identify the root cause. The hierarchical approach accurately identified the true root cause in a well-understood case study; its plant-wide map consisted of only three nodes compared to 15 nodes in the standard causality map and its transitive reduction. The plant-wide map required less fault-state data, time series in the order of hours or days instead of weeks or months, further motivating its application in rapid root cause analysis.

fastBMA: Scalable Network Inference and Transitive Reduction

BACKGROUND:Inferring genetic networks from genome-wide expression data is extremely demanding computationally. We have developed fastBMA, a distributed, parallel and scalable implementation of Bayesian model averaging (BMA) for this purpose. fastBMA also includes a novel and computationally efficient method for eliminating redundant indirect edges in the network.FINDINGS:We evaluated the performance of fastBMA on synthetic data and experimental genome-wide yeast and human datasets. When using a single CPU core, fastBMA is up to 100 times faster than the next fastest method, LASSO, with increased accuracy. It is a memory efficient, parallel and distributed application that scales to human genome wide expression data. A 10,000-gene regulation network can be obtained in a matter of hours using a 32-core cloud cluster.CONCLUSIONS:fastBMA is a significant improvement over its predecessor ScanBMA. It is orders of magnitude faster and more accurate than other fast network inference methods such as LASSO. The improved scalability allows it to calculate networks from genome scale data in a reasonable timeframe. The transitive reduction method can improve accuracy in denser networks. fastBMA is available as code (M.I.T. license) from GitHub (https://github.com/lhhunghimself/fastBMA), as part of the updated networkBMA Bioconductor package (https://www.bioconductor.org/packages/release/bioc/html/networkBMA.html) and as ready-to-deploy Docker images (https://hub.docker.com/r/biodepot/fastbma/).