An extensive reference dataset for fault detection and identification in batch processes

Abstract

Close process monitoring (i.e., detection and identification of disturbances) is important to achieve high process efficiency and safety. The Tennessee Eastman process is an extensive benchmark dataset for fault detection and identification, but it is only representative for continuous processes because it does not contain the inherent nonstationarity that complicates monitoring of batch processes. Nevertheless, batch processes also play an important role in many types of industry. This paper therefore presents an extensive reference dataset for benchmarking data-driven methodologies for fault detection and identification in batch processes. The original Pensim model [10] is expanded with sensor noise. By changing the properties of the initial conditions and/or model parameters, four subsets of different complexity are generated, each containing 400 batches with normal operation. To correctly assess the fault detection and identification in batch processes, 15 faults are simulated with various amplitudes and onset times for a total of 22,200 faulty batches for each subset, or 90,400 batches in total. Analysis of the data indicates that the presented types of process faults and their various amplitudes in each of the four subsets present a suitable benchmark for fault detection and identification in batch processes. The dataset is freely available at http://cit.kuleuven.be/biotec/batchbenchmark.

Monitoring batch processes with dynamic time warping and k-nearest neighbours

“…The MPCA-Q TDR plot for Fault 6 reproduces closely the results presented inFig. 14(d)of[16] with minor differences likely due to choice of MPCA infilling method, control limits, and our use of 5 batch repetitions rather than 50.…”

“…In the following sections DTW-NN is outlined in detail. It is applied to the extensive datasets from [16] of a simulated penicillin production batch process and performance is contrasted with that of a benchmark MPCA scheme.…”

“…To evaluate the proposed method, the four simulated datasets provided by [16] were used. These datasets, which extend the Pensim model of [25], represent data from batch production of penicillin.…”

Big data approach to batch process monitoring: Simultaneous fault detection and diagnosis using nonlinear support vector machine-based feature selection

“…On the other hand, despite the wide use of batch reactor processes in chemical, food, and pharmaceutical industries, most novel techniques for fault detection and diagnosis have focused on continuous processes. The main reason for this is the challenging characteristics of batch process data 35,36 such as (i) involvement of a considerable number of interconnected variables, (ii) inherent non-stationarity, (iii) finite duration, (iv) nonlinear response, and (v) batch-to-batch variability. Additionally, the dimensionality of batch process data further obstructs and complicates the monitoring.…”

“…Section 3 presents the proposed theoretical and algorithmic developments in feature selection using a nonlinear SVM formulation which can be implemented in various engineering problems. In Section 4, we adopt a recent extensive benchmark, simulation dataset on penicillin production process model, PenSim model, 35,38 . Section 5 presents the step by step implementation of the developed nonlinear SVM-based feature selection algorithm in the fault detection and diagnosis of batch process setting.…”