Uwe Mönks

Sensor fusion by two-layer conflict solving

Many of the existing fusion approaches based on Dempster-Shafer Theory (DST) tend to be unreliable in various scenarios. Therefore, this topic is still in discussion. In this work a Two-Layer Conflict Solving (TLCS) data fusion scheme is proposed which is based on Dempster-Shafer Theory and on Fuzzy-Pattern-Classification (FPC) concepts. The aim is to provide an approach to data fusion which provides a stable conflict scenario handling. Furthermore, the scheme can easily be extended to fuzzy classification and is applicable to sensor fusion applications. Therefore, the suggested approach will contribute as a novel fuzzy fusion method.

Sensorless drive diagnosis using automated feature extraction, significance ranking and reduction

Systems for process automation become increasingly complex and also tend to be composed of autonomous subsystems, which is strongly driven by the progress made in information technology. An active field of research is the implementation of monitoring and control at sub-system level using cognitive approaches. In this paper we present a method for autonomous and sensorless condition monitoring of an electric drive train. Based on experiment design we measured phase currents of a physical demonstrator device including mechanical defects and extracted signal features using proper orthogonal decomposition. In favor of classification of different defect states we performed a linear discriminant analysis, which yields appropriate data for a Fuzzy-Pattern-Classification algorithm. As a result we were able to identify different reference defect states as well as previously unknown states.

Machine conditioning by importance controlled information fusion

Sensor and information fusion is recently a major topic which becomes important in machine diagnosis and conditioning for complex production machines and process engineering. It is a known fact that distributed automation systems have a major impact on signal processing and pattern recognition for machine diagnosis. Therefore, it is necessary to research and develop smart diagnosis methods which are applicable for distributed systems like resource-limited cyber-physical systems. In this paper we propose an new approach for sensor and information fusion based on Evidence Theory and socio-psychological decision-making. We show that context based condition monitoring is instantiated even in conflict situations, oc-curing in real life scenarios permanently. A simple but effective importance measure is proposed which controls the significance of conditioning propositions in a system.

An extended perspective on evidential aggregation rules in machine condition monitoring

In this paper we propose a novel, extended perspective on evidential aggregation rules in machine condition monitoring. First, aspects regarding the interconnections between Dempster-Shafer, Fuzzy Set, and Possibility Theory are shown. Subsequently, a novel approach for direct determination of basic probability assignments using Fuzzy membership functions is proposed. Finally, it is applied to a pipe extrusion lines condition monitoring system, considering and reducing pairwise conflicts.

Information Fusion of Conflicting Input Data

Sensors, and also actuators or external sources such as databases, serve as data sources in order to realise condition monitoring of industrial applications or the acquisition of characteristic parameters like production speed or reject rate. Modern facilities create such a large amount of complex data that a machine operator is unable to comprehend and process the information contained in the data. Thus, information fusion mechanisms gain increasing importance. Besides the management of large amounts of data, further challenges towards the fusion algorithms arise from epistemic uncertainties (incomplete knowledge) in the input signals as well as conflicts between them. These aspects must be considered during information processing to obtain reliable results, which are in accordance with the real world. The analysis of the scientific state of the art shows that current solutions fulfil said requirements at most only partly. This article proposes the multilayered information fusion system MACRO (multilayer attribute-based conflict-reducing observation) employing the μBalTLCS (fuzzified balanced two-layer conflict solving) fusion algorithm to reduce the impact of conflicts on the fusion result. The performance of the contribution is shown by its evaluation in the scope of a machine condition monitoring application under laboratory conditions. Here, the MACRO system yields the best results compared to state-of-the-art fusion mechanisms. The utilised data is published and freely accessible.

Fuzzy-Pattern-Classifier Training with Small Data Sets

It is likely in real-world applications that only little data is available for training a knowledge-based system. We present a method for automatically training the knowledge-representing membership functions of a Fuzzy-Pattern-Classification system that works also when only little data is available and the universal set is described insufficiently. Actually, this paper presents how the Modified-Fuzzy-Pattern-Classifier’s membership functions are trained using probability distribution functions.

Fast evidence-based information fusion

Information fusion systems are crucial for the success of the upcoming fourth industrial revolution. In this emerging field, cyber-physicals systems play a major role. These are physical processing systems equipped with sensory devices which interconnect over communication networks for distributed cognitive information processing applications. Cyber-physical systems are generally limited in computational resources. Due to this fact, signal processing algorithms cannot be implemented one-to-one. Instead, efforts must be spent in algorithm optimisation towards resource efficiency and reduced computational complexity. In this contribution, we present our optimisation approach by matrix decomposition of an evidence-based conflict-reducing fusion approach which after optimisation is applicable in resource-limited devices for cognitive signal processing. We evaluate the results by comparison with the algorithms original definition and show the improvements achieved.

Fuzzy-Pattern-Classifier Based Sensor Fusion for Machine Conditioning

Sensor and Information fusion is recently a major topic not only in traffic management, military, avionics, robotics, image processing, and e.g. medical applications, but becomes more and more important in machine diagnosis and conditioning for complex production machines and process engineering. Several approaches for multi-sensor systems exist in the literature (e.g. Hall, 2001; Bosse, 2007). In this chapter an approach for a Fuzzy-Pattern-Classifier Sensor Fusion Model based on a general framework (e.g. Bocklisch, 1986; Eichhorn, 2000; Schlegel, 2004; Lohweg, 2004; Lohweg, 2006; Hempel, 2008; Herbst 2008; Monks, 2009; Hempel 2010) is described. An application of the fusion method is shown for printing machines. An application on quality inspection and machine conditioning in the area of banknote production is highlighted. The inspection of banknotes is a high labour intensive process, where traditionally every note on every sheet is inspected manually. Machines for the automatic inspection and authentication of banknotes have been on the market for the past 10 to 12 years, but recent developments in technology have enabled a new generation of detectors and machines to be developed. However, as more and more print techniques and new security features are established, total quality, security in banknote printing as well as proper machine conditions must be assured (Brown, 2004). Therefore, this factor necessitates amplification of a sensorial concept in general. Such systems can be used to enhance the stability of inspection and condition results for user convenience while improving machine reliability. During printed product manufacturing, measures are typically taken to ensure a certain level of printing quality. This is particularly true in the field of security printing, where the quality standards, which must be reached by the end-products, i.e. banknotes, security documents and the like, are very high. Quality inspection of printed products is conventionally limited to the optical inspection of the printed product. Such optical inspection can be performed as an off-line process, i.e. after the printed product has been processed in the printing press, or, more frequently, as an in-line process, i.e. on the printing press, where the printing operation is carried out. Usually only the existence or appearance of colours and their textures are checked by an optical inspection system. 14

An Approach to Automated Fusion System Design and Adaptation

Industrial applications are in transition towards modular and flexible architectures that are capable of self-configuration and -optimisation. This is due to the demand of mass customisation and the increasing complexity of industrial systems. The conversion to modular systems is related to challenges in all disciplines. Consequently, diverse tasks such as information processing, extensive networking, or system monitoring using sensor and information fusion systems need to be reconsidered. The focus of this contribution is on distributed sensor and information fusion systems for system monitoring, which must reflect the increasing flexibility of fusion systems. This contribution thus proposes an approach, which relies on a network of self-descriptive intelligent sensor nodes, for the automatic design and update of sensor and information fusion systems. This article encompasses the fusion system configuration and adaptation as well as communication aspects. Manual interaction with the flexibly changing system is reduced to a minimum.

Fast classification in industrial Big Data environments

Many modern industrial applications, e.g. those incorporating hundreds or thousands of electrical sensors and actuators, must be categorised into Big Data environments, in which it is essential to design suitable information processing models. Central data processing in such environments is impossible and must be carried out in a distributed way on resource-limited cyber-physical systems. One of the challenging tasks for machine learning is thus the design of a classifier which is simple, accurate and has an acceptable realisation time. We present ComRef-2D-ConvHull method for linear classification optimisation in lower-dimensional feature space, which is based on ComRef from [1]. Compared to original ComRef, we consider only classification optimisation in 2-dimensional feature spaces in ComRef-2D-ConvHull. Due to the decreased time complexity for calculations in 2-dimensional feature space, we expect many industrial Big Data enviroments to profit from our method. Tests regarding the generalisation ability of ComRef-2D-ConvHull on several reference data sets and on a real-world industrial dataset show promising results.