Consensus-Based Sequential Estimation of Process Parameters via Industrial Wireless Sensor Networks.

Feilong Lin,Liyong Yuan,Wenbai Li

doi:10.3390/s18103338

Abstract

Process parameter estimation, to a large extent, determines the industrial production quality. However, limited sensors can be deployed in a traditional wired manner, which results in poor process parameter estimation in hostile environments. Industrial wireless sensor networks (IWSNs) are techniques that enrich sampling points by flexible sensor deployment and then purify the target by collaborative signal denoising. In this paper, the process industry scenario is concerned, where the workpiece is transferred on the belt and the parameter estimate is required before entering into the next process stage. To this end, a consensus-based sequential estimation (CSE) framework is proposed which utilizes the co-design of IWSN and parameter state estimation. First, a group-based network deployment strategy, together with a TDMA (Time division multiple access)-based scheduling scheme is provided to track and sample the moving workpiece. Then, by matching to the tailored IWSN, the sequential estimation algorithm, which is based on the consensus-based Kalman estimation, is developed, and the optimal estimator that minimizes the mean-square error (MSE) is derived under the uncertain wireless communications. Finally, a case study on temperature estimation during the hot milling process is provided. The results show that the estimation error can be reduced to less than within a limited time period, although the measurement error can be more than in existing systems with a single-point temperature sensor.

Highlights

Industrial wireless sensor networks (IWSNs) have been considered to be fundamental technologies that will promote the industry revolution, such as Industry 4.0 and Industrial Internet.By leveraging the advantages of IWSNs, in terms of flexible deployment and wireless routing, monitoring in hostile environments becomes tractable and the ubiquitous conception on the plant floor becomes feasible, both of which are bottleneck problems in traditional wired monitoring systems.more and more efforts have be paid to developing new solutions with IWSNs to improve industrial production
The results show that state consensus-based distributed Kalman-consensus filter (DKCF) gains a lower mean-square error (MSE) than the others
The results demonstrate that consensus-based sequential estimation (CSE) can greatly improve the estimation performance within a limited sampling time period, and gains much better estimation performance than the unscented Kalman filter (UKF)-based estimation

Summary

Introduction

Industrial wireless sensor networks (IWSNs) have been considered to be fundamental technologies that will promote the industry revolution, such as Industry 4.0 and Industrial Internet.By leveraging the advantages of IWSNs, in terms of flexible deployment and wireless routing, monitoring in hostile environments becomes tractable and the ubiquitous conception on the plant floor becomes feasible, both of which are bottleneck problems in traditional wired monitoring systems.more and more efforts have be paid to developing new solutions with IWSNs to improve industrial production. Industrial wireless sensor networks (IWSNs) have been considered to be fundamental technologies that will promote the industry revolution, such as Industry 4.0 and Industrial Internet. By leveraging the advantages of IWSNs, in terms of flexible deployment and wireless routing, monitoring in hostile environments becomes tractable and the ubiquitous conception on the plant floor becomes feasible, both of which are bottleneck problems in traditional wired monitoring systems. More and more efforts have be paid to developing new solutions with IWSNs to improve industrial production. We expect to obtain an accurate estimate of the specific parameter’s value with the aid of IWSN before the workpiece enters into the process stage. To a large extent, determines the quality of industrial production [1,2]

Objectives

Results

Conclusion