A frequency‐localized recursive partial least squares ensemble for soft sensing

Abstract

We report the use of a frequency‐localized adaptive soft sensor ensemble using the wavelet coefficients of the responses from the physical sensors. The proposed method is based on building recursive, partial least squares soft sensor models on each of the wavelet coefficient matrices representing different frequency content of the signals from the physical sensors, combining the predictions from these models via static weights determined from an inverse‐variance weighting approach, and recursively adapting each of the soft sensor models in the ensemble when new data are received. Wavelet‐induced boundary effects are handled by using the undecimated wavelet transform with the Haar wavelet, an approach that is not subject to wavelet boundary effects that would otherwise arise on the most recent sensor data. An additional advantage of the undecimated wavelet transform is that the wavelet function is defined for a signal of arbitrary length, thus avoiding the need to either trim or pad the training signals to dyadic length, which is required with the basic discrete wavelet transform. The new method is tested against a standard recursive partial least squares soft sensor on 3 soft‐sensing applications from 2 real industrial processes. For the datasets we examined, we show that results from the new method appear to be statistically superior to those from a soft sensor based only on a recursive partial least squares model with additional advantages arising from the ability to examine performance of each localized soft sensor in the ensemble.

Prediction analysis of slope stability due to soft and weak interlayers based on partial least squares method

“…In recursive partial least squares with a forgetting factor, the forgetting factor impacts the prediction results, and different values  are chosen to achieve different forgetting effects on historical data [19][20]. In the range of  the larger the value, the stronger the prediction effect on stability and the smaller the error in the prediction results, but the slower the convergence rate leads to a reduction in the algorithm's predictive power.…”

Control de caudal con un sensor virtual basado en técnicas de deep learning

“…Las metodologías desarrolladas son muy diversas y su comportamiento dependerá de cómo se implementen. Por un lado, existen implementaciones mediante un análisis estadístico multivariable [7], o con modelos de primer principio donde el sistema está modelado mediante los principios matemáticos y físicos que lo definen [8]. Otra implementación diferente sería con observadores basados en datos [9] o técnicas de filtrado como son los filtros de Kalman [10].…”

Soft Sensing Model Visualization: Fine-tuning Neural Network from What Model Learned

“…In order to obtain effective quality indicators, the industry has developed soft-sensing technology to estimate and predict the quality variables [2]. The soft-sensing technology has been widely applied to various industries for real-time quality monitoring and early reports before time-consuming laboratory analysis, including chemical production [3], [4], Figure 1: A high-level simplified overview of wafer manufacturing process drug industry [5], petroleum refining [6], mechanical industry [7], fault detection and diagnosis [8], [9] and semiconductor manufacturing industry [10].…”