Development of Performance Measurement System using Internet of Things

Abstract

The ability to measure operational performance is an important factor for competing enterprises in the global market. Performance measurement helps in the evaluation of the long term effects of outputs for improving competitiveness and decision-making power. A companys competitiveness and profits are reduced by a consistent continuation of subpar performance, as this eventually leads to a failure to meet customer need. In this overall perspective, using performance measurement to understand the companys circumstances is necessary for the manufacturing system to have rapid reactive ability. Although manufacturing companies have used information systems to manage performance, there has been the difficulty of capturing real-time data to depict real situations. The recent rapid proliferation of Internet of Things (IoT) has enabled the resolution of this problem. With the maturity of IoT devices and databases technology, manufacturers are able to assess productivities and obtain real-time feedback from all production lines through IoT data. As IoT-based environment is well established, Industry 4.0 has evolved. It is the fourth stage of industrialization, and is also referred to as smart factory. Indubitably, in a smart factory environment, the complexity of information system network has increased, because manufacturing systems consist of multiple servers and client applications. Interoperability among manufacturing information systems is a rising issue for a manufacturer who developed the inter-connected systems and systematic obedience. OPC-UA (Open Platform Communication Unified Architecture) is a set of industrial standards providing a common interface for communications and represents a method to transmit any kinds of data. This thesis follows OPC-UA standard and explains how IoT data are exchanged among heterogeneous systems. Moreover, complexity of network causes IoT fault. If an IoT fault occurs, the performance measurement results cannot describe the production situation appropriately, because data-driven measurement is strongly connected with acquired IoT data. In other words, a reasonable value for Key Performance Indicators cannot be derived, if the IoT data have an error value. An IoT data anomaly detection and mitigation process is therefore required in response to the problem. To resolve enumerate backgrounds and problems, the dissertation comprised five steps: (1) Development of an smart factory performance measurement model consistent with the ISA-95 and ISO-22400 standards, which define manufacturing processes and performance indicator formulas

(2) Identification of IoT applicable parts in ISO-22400 standard and selection of the Key Performance Indicators of the Net-Overall Equipment Effectiveness (OEE)

(3) Configuration of the smart factory architecture and performance measurement process using Business Process Modelling, and adaptation of data exchange protocol by referencing OPC-UA

(4) Implementation of an IoT fault case classification and data anomaly detection and mitigation algorithm, using k-means and statistical inference methods

and (5) Validation of the proposed system through experimental simulation. The experimental simulation results showed that the proposed system represented the timestamp data acquired by IoT and captured the entire production process. In addition, these results indicated that the proposed data anomaly detection and mitigation algorithm have a positive impact on IoT data anomaly identification, thus enabling the determination of real-time performance indicators.