Showing papers by "Mahdi Aliyari Shoorehdeli published in 2020"
••
TL;DR: Two types of variable thresholding are introduced and a novel approach for performance assessment of VTASs using Priority-AND gate and semi-Markov process is proposed, allowing the proposed approach to consider industrial measurements with non-Gaussian distributions.
Abstract: In large industrial systems, alarm management is one of the most important issues to improve the safety and efficiency of systems in practice. Operators of such systems often have to deal with a numerous number of simultaneous alarms. Different kinds of thresholding or filtration are applied to decrease alarm nuisance and improve performance indices, such as Averaged Alarm Delay (ADD), Missed Alarm and False Alarm Rates (MAR and FAR). Among threshold-based approaches, variable thresholding methods are well-known for reducing the alarm nuisance and improving the performance of the alarm system. However, the literature suffers from the lack of an appropriate method to assess performance parameters of Variable Threshold Alarm Systems (VTASs). This study introduces two types of variable thresholding and proposes a novel approach for performance assessment of VTASs using Priority-AND gate and semi-Markov process. Application of semi-Markov process allows the proposed approach to consider industrial measurements with non-Gaussian distributions. In addition, the paper provides a genetic algorithm based optimized design process for optimal parameter setting to improve performance indices. The effectiveness of the proposed approach is illustrated via three numerical examples and through a comparison with previous studies.
28 citations
••
TL;DR: A novel method for clustering referred to as Reward-Based Online Clustering (RBOC) which is formed based on the reinforcement learning algorithm and can automatically detect the clusters while there is no knowledge about the number of clusters.
10 citations
•
TL;DR: A new transfer learning (TL) based regression method, called Domain Adversarial Neural Network Regression (DANN-R), is proposed and employed for designing transferable soft sensors that can successfully adapt to new plants and new working conditions.
Abstract: In this paper, a new approach is proposed for designing transferable soft sensors. Soft sensing is one of the significant applications of data-driven methods in the condition monitoring of plants. While hard sensors can be easily used in various plants, soft sensors are confined to the specific plant they are designed for and cannot be used in a new plant or even used in some new working conditions in the same plant. In this paper, a solution is proposed for this underlying obstacle in data-driven condition monitoring systems. Data-driven methods suffer from the fact that the distribution of the data by which the models are constructed may not be the same as the distribution of the data to which the model will be applied. This ultimately leads to the decline of models accuracy. We proposed a new transfer learning (TL) based regression method, called Domain Adversarial Neural Network Regression (DANN-R), and employed it for designing transferable soft sensors. We used data collected from the SCADA system of an industrial power plant to comprehensively investigate the functionality of the proposed method. The result reveals that the proposed transferable soft sensor can successfully adapt to new plants.
7 citations
••
6 citations
••
01 Dec 2020TL;DR: In this article, a new Fault Detection and Isolation (FDI) approach based on Transfer Learning (TL) was introduced for improving health monitoring systems of gas turbines under varying working conditions.
Abstract: In this paper, we introduce a new Fault Detection and Isolation (FDI) approach based on Transfer Learning (TL) for improving health monitoring systems of gas turbines under varying working conditions. Nowadays, researchers have found intelligent algorithms a reliable tool for condition monitoring of mechanical systems and processes. In this regard, modern automation systems in many industries, including power plants, are heavily utilizing machine learning algorithms. However, the performance of data-driven methods depends on the consistency of data distribution. Unfortunately, this assumption might not be satisfied with real-world problems. In this research, we contribute to finding a solution to this problem, which is a crucial barrier to many intelligent condition monitoring systems. We used domain adversarial training of neural networks to find models that can adapt to new working conditions of gas-turbines. Accordingly, a well-known gas-turbine simulator is employed to simulate the process behavior under various working conditions, and it is illustrated that even small variations in working conditions cause a dramatic decline in the performance of models. We demonstrate that the proposed TL-based FDI approach can be successfully employed to cope with the inconsistency of data distribution in process systems.
5 citations
••
TL;DR: The TLBO algorithm is modified based on stability analysis and the modified TLBO is compared with the standard TLBO, particle swarm optimization (PSO), real genetic algorithm (RGA), and gravitational search algorithm (GSA).
Abstract: Teaching–learning-based optimization (TLBO) algorithms is one of the swarm-based optimization search algorithms. It develops based on the teaching–learning procedures at a classroom to solve multi-dimensional and nonlinear problems. In this paper, convergence and stability analysis of TLBO are studied. The stability of individual dynamics is analyzed by Lyapunov stability theorem and the concept of system dynamics. Stability conditions are achieved and utilized for adapting parameters of the TLBO. The TLBO algorithm is modified based on stability analysis. The modified TLBO is compared with the standard TLBO, particle swarm optimization (PSO), real genetic algorithm (RGA), and gravitational search algorithm (GSA). Simulation results confirm the validity and feasibility of the proposed modified TLBO. The appropriate performance is achieved for multi-dimensional and nonlinear standard bench functions.
5 citations
•
TL;DR: This paper shows that the proposed method for Transfer Learning based on Gibbs Sampling can successfully enhance target classification by a considerable ratio and has the advantage over common DA methods that it needs no target data during the process of training of models.
Abstract: In this paper, we present a new idea for Transfer Learning (TL) based on Gibbs Sampling Gibbs sampling is an algorithm in which instances are likely to transfer to a new state with a higher possibility with respect to a probability distribution We find that such an algorithm can be employed to transfer instances between domains Restricted Boltzmann Machine (RBM) is an energy based model that is very feasible for being trained to represent a data distribution and also for performing Gibbs sampling We used RBM to capture data distribution of the source domain and use it in order to cast target instances into new data with a distribution similar to the distribution of source data Using datasets that are commonly used for evaluation of TL methods, we show that our method can successfully enhance target classification by a considerable ratio Additionally, the proposed method has the advantage over common DA methods that it needs no target data during the process of training of models
4 citations
••
TL;DR: A robust anomaly detection filter is proposed for continuous linear Roesser systems using dynamic observer framework and its sensitivity to anomaly as well as its robustness to disturbances are addressed via linear matrix inequalities (LMIs).
•
TL;DR: With the help of a few sensors and data mining approaches, an anomaly detection system is built for Osmotron water purifier and extensive experiments are conducted to show the accuracy of the data-driven and model-based anomaly detection methods.
Abstract: Industry 4.0 will make manufacturing processes smarter but this smartness requires more environmental awareness, which in case of Industrial Internet of Things, is realized by the help of sensors. This article is about industrial pharmaceutical systems and more specifically, water purification systems. Purified water which has certain conductivity is an important ingredient in many pharmaceutical products. Almost every pharmaceutical company has a water purifying unit as a part of its interdependent systems. Early detection of faults right at the edge can significantly decrease maintenance costs and improve safety and output quality, and as a result, lead to the production of better medicines. In this paper, with the help of a few sensors and data mining approaches, an anomaly detection system is built for CHRIST Osmotron water purifier. This is a practical research with real-world data collected from SinaDarou Labs Co. Data collection was done by using six sensors over two-week intervals before and after system overhaul. This gave us normal and faulty operation samples. Given the data, we propose two anomaly detection approaches to build up our edge fault detection system. The first approach is based on supervised learning and data mining e.g. by support vector machines. However, since we cannot collect all possible faults data, an anomaly detection approach is proposed based on normal system identification which models the system components by artificial neural networks. Extensive experiments are conducted with the dataset generated in this study to show the accuracy of the data-driven and model-based anomaly detection methods.