Trafikverket, the agency in charge of state road maintenance in Sweden, have a number of so-called Road Weather Information Systems (RWIS). The main purpose of the stations is to provide winter road maintenance workers with information to decide when roads need to be plowed and/or salted. Each RWIS have a number of sensors which make road weather-related measurements every 30 minutes. One of the sensors is dug into the road which can cause traffic disturbances and be costly for Trafikverket. Other RWIS sensors fail occasionally. This project aims at modelling a set of RWIS sensors using supervised machine learning algorithms. The sensors that are of interest to model are: Optic Eye, Track Ice Road Sensor (TIRS) and DST111. Optic Eye measures precipitation type and precipitation amount. Both TIRS and DST111 measure road surface temperature. The difference between TIRS and DST111 is that the former is dug into the road, and DST111 measures road surface temperature from a distance via infrared laser. Any supervised learning algorithm trained to model a given measurement made by a sensor, may only train on measurements made by the other sensors as input features. Measurements made by TIRS may not be used as input in modelling other sensors, since it is desired to see if TIRS can be removed. The following input features may also be used for training: road friction, road surface condition and timestamp. Scikit-learn was used as machine learning software in this project. An experimental approach was chosen to achieve the project results: A pre-determined set of supervised algorithms were compared using different amount of top relevant input features and different hyperparameter settings. Prior to achieving the results, a data preparation process was conducted. Observations with suspected or definitive errors were removed in this process. During the data preparation process, the timestamp feature was transformed into two new features: month and hour. The results in this project show that precipitation type was best modelled using Classification And Regression Tree (CART) on Scikit-learn default settings, achieving a performance score of Macro-F1test = 0.46 and accuracy = 0.84 using road surface condition, road friction, DST111 road surface temperature, hour and month as input features. Precipitation amount was best modelled using k-Nearest Neighbor (kNN); with k = 64 and road friction used as the only input feature, a performance score of MSEtest = 0.31 was attained. TIRS road surface temperature was best modelled with Multi-Layer Perceptron (MLP) using 64 hidden nodes and DST111 road surface temperature, road surface condition, road friction, month, hour and precipitation type as input features, with which a performance score of MSEtest = 0.88 was achieved. DST111 road surface temperature was best modelled using Random forest on Scikit-learn default settings with road surface condition, road friction, month, precipitation type and hour as input features, achieving a performance score of MSEtest = 10.16.

Inom industrin utvärderas maskiners tillstånd av enskilda arbetare för att avgöra behovet av underhåll. Dessa beslut baseras på antaganden och den enskilda arbetarens erfarenhet, vilket kan leda till felaktiga beslut. Beslut som leder till onödigt underhåll påverkar företagens ekonomi negativt. Genom att använda sensorer installerade på maskiner tillsammans med ett system för att resonera om värden från dessa kan maskinernas tillstånd avgöras. Genom att använda Augmented Reality för att visa detta tillstånd för arbetarna kan mer informerade beslut om underhåll tas. Den här rapporten undersöker de olika teknologier som behövs för att göra detta möjligt, Augmented Reality, Reasoning Systems, och Internet of Things. En prototypapplikation som utnyttjar dessa har skapats för att visa på vad som är möjligt med de enheter vi alla bär med oss varje dag. / Inspection workers in industries, evaluates the state of machines based on assumptions to decide if a need for service exists. These assumptions varies depending on the person performing the evaluation, which can cause the wrong decision to be made. These decisions on machine service affect the economy of the industry. By using sensors mounted to the machines and a reasoning system to evaluate the data from these sensors, the condition of the machines can be determined. Augmented Reality can then be used to display this condition to the inspection worker, leading to more informed decisions about the need for service being made. This thesis examines the different technologies needed to make this possible, Augmented Reality, Reasoning Systems, and Internet of Things. A prototype application is created using these to show what is possible using the mobile devices we all carry.

FundaÃÃo Cearense de Apoio ao Desenvolvimento Cientifico e TecnolÃgico / Este trabalho apresenta o desenvolvimento e avaliaÃÃo de um sistema de monitoramento com tecnologia Java e Bluetooth para o auxilio no gerenciamento da condiÃÃo de transformadores de distribuiÃÃo a partir de modelos tÃrmicos. Sabe-se que monitorar a condiÃÃo dos transformadores tem sido objeto de preocupaÃÃo no que diz respeito a evitar as perdas econÃmicas causadas por falhas no equipamento e a consequente descontinuidade do serviÃo. à possÃvel, atravÃs da gestÃo do sistema: maximizar a vida Ãtil dos equipamentos monitorados; melhorar a confiabilidade da rede; reduzir os custos de manutenÃÃo. AtravÃs do levantamento bibliogrÃfico de sistemas de monitoramento atuais, de suas funcionalidades, em funÃÃo de recentes tecnologias de software e hardware, e do estudo de modelos tÃrmicos aplicados a transformadores, foi desenvolvido um sistema de coleta de dados e uma interface para anÃlise de dados. Este trabalho traz uma contribuiÃÃo para o desenvolvimento desses sistemas, propondo o uso de dispositivos mÃveis como uma estratÃgia economicamente viÃvel para coletar dados de transformadores de distribuiÃÃo, e a utilizaÃÃo dos dados coletados por estes dispositivos para gerar modelos tÃrmicos (das temperaturas do topo do Ãleo e do ponto quente) que permitam: (i) estabelecer limites mÃximos de carga para o equipamento, e (ii) estimar sua perda de vida Ãtil. Estas informaÃÃes fornecidas pelo sistema de monitoramento proposto devem servir como suporte à implementaÃÃo de planos de aÃÃo da concessionÃria de energia elÃtrica para: (i) a eventual substituiÃÃo de equipamentos, (ii) a adequaÃÃo na distribuiÃÃo das cargas, (iii) a priorizaÃÃo no caso de necessidade de atuaÃÃo do sistema de proteÃÃo. / This paper presents the development and evaluation of a monitoring system with Bluetooth and Java technology to aid managing the condition of distribution transformers through thermal models. It is well known that monitor the condition of transformers has been the subject of concern in relation to avoid economic losses caused by equipment failure. It is possible, through the system management: (i) maximize the life time of the monitored equipment, (ii) improve the reliability of the electrical grid and (iii) reduce its maintenance costs. Through a bibliographical survey of transformer condition monitoring systems and study of thermal models applied to transformers, was developed a system for data acquisition and an interface for data analysis. This work brings a contribution to the development of these systems, proposing the use of mobile devices as a viable strategy to collect data of distribution transformers and using data collected by those devices for generating thermal models (to estimate top-oil and hotspot temperature) that will let us: (i) establish maximum load for the equipment, and (ii) estimate the loss of life. The information provided by the condition monitoring system proposed should serve as support to electric utility company to implement action plans to help possible exchange of equipment, do a more balanced load in their distribution transformers and help prioritizing, in case of any necessary actuation of the system protection.

Reliability of wind turbine components and maintenance optimisation are among the critical aspects of wind power development closely related to profitability and future development. The main reason for research in these areas is lowering the cost of energy production for wind power, specifically important in offshore environment. Continuous monitoring of specific wind turbine components can be valuable for wind farm operators and, subsequently, wind farm owners.  Also, health assessment of critical components can be useful in estimating the possibilities for life extension of wind turbines. Expensive Condition Monitoring Systems (CMSs) are not always available, particularly in older wind farms, and additionally installing CMSs on wind turbines is not always economically feasible. However, most of modern wind turbines are equipped with the Supervisory Control And Data Acquisition (SCADA) system which is recording 10-minute average values of parameters that depict operation of the turbine. That being said, SCADA data contains a vast amount of information that can be used for analysis of wind turbine components health. Therefore, this project will present an application of previously published methodology for SCADA data condition monitoring on real wind farm data. The goal of this project is to investigate on the possibilities of the SCADA monitoring methodology and what can be the added value of the application for wind farm operators, owners and other stakeholders. The methodology for condition monitoring through SCADA data was applied on real data gathered from two wind farms in Germany and one in the Netherlands. During the project the methodology had to be modified in order to ensure the best possible industrial application. Results of the project showed that the SCADA data condition monitoring approach is not capable of predicting failures. However, the technique has been proven successful for detecting the changes of trends in dependencies of working parameters, specifically monitoring parameters related to the turbine generators. Continuously monitoring the dependencies of working parameters can be used as an additional source of information for maintenance scheduling and assessment of components health. The approach presented in this paper can be valuable to asset managers and wind farm owners.

This thesis presents a multi-physics-based approach to the design and analysis of smart cutting tools for emerging industrial requirements, within an innovative design process. The design process is in stages according to design specifications and requires analysis, conceptual design, detailed design, prototype production and service testing. The research presented in the thesis follows the design process but focuses on the detailed design of the smart turning tool, including mechanical design, electrical wiring and sensor circuitry, embedded algorithms development, and multi-physics-based simulation for the tool system integration, design analysis and optimisation. The thesis includes the introduction of the research background, a critical literature review of the research topic, a multi-physics-based design and analysis of the smart cutting tool, a mechanical structural detail design of the prototype smart turning tool, the electrical system design focusing on cutting force measurement and embedded wireless communication features, and the final experimental testing and calibration of the smart cutting tool. The contributions to knowledge are highlighted in the conclusions chapter towards the end of the thesis. The research proposes multi-physics-based design and analysis concepts for a smart turning tool, which can measure the cutting forces on a 0.1 N scale and can also be used to monitor the tool condition, particularly for ultraprecision and micro-machining purposes. The smart turning tool is a sensored tool, constructed with wireless and plug-and-produce features. The tool design modelling and simulation was undertaken within a multi-physics modelling and analysis environment-based on COMSOL. This integrates the piezoelectric physics with mechanical structural design and radio frequency electronic communications of cutting force signals. The multi-physics simulation method takes account of all design-mechanics-physics-electronics analysis and transformations simultaneously within one computational environment, including FEA analysis, modal analysis, structural deformation, lead piezoelectric effect and wireless data/signal simulation. With the multi-physics simulation developed, the integrated design of the smart turning tool and its performance can be physically analysed and optimised in a virtual environment. The tool design process follows the total design methodology, which can be strictly executed in several design stages. Both mechanical and electrical design of the smart cutting tool are embodied into the tool detail design. The tool mechanical structure is systematically built from the selection of the tool material, through the structure analysis and further progressed with static force – strain/stress transformation, equivalent force measurement and calibration. The electrical circuitry was systematically developed from developing the customised charge amplifier, detail design of the main circuitry and coding development procedure, preliminary PCB fabrication and multi-sensor port PCB development, as well as the real-time cutting force monitoring programming and interface coding. The experiment calibrations and cutting trials with the tool system are also designed in light of the total design methodology. The experiment procedure for using the smart turning tool is further presented in two different sections. The thesis concludes with a further discussion on the main research findings, which are further supported by the highlighted contributions to knowledge and recommendations for future work.

This thesis project investigates approaches for malfunction prediction using unsupervised, self-organized models, with an orientation on bus fleets. Certain bus malfunctions are not predictable with conventional methods and preventive replacements are too costly and time consuming. Malfunctions that could result in interruption of service or on degradation of safety  are of high priority to predict.The settings of the desired application define the following constraints: definition of a model by an expert is not desirable as it is not a scalable solution, ambient conditions or usage schedule must not affect the prediction, data communication between the systems is limited so data must be compressed with relevant to the problem features. In this work, definition of normal or faulty operation is not handled by an expert, but using the Wisdom of the crowd idea and Consensus Self-organized models for fault detection (COSMO), or by the system's past state by monitoring an autoencoder's reconstruction error. In COSMO each system constructs a model describing its condition and then all distances between models are estimated to find the Most Central Pattern (MCP), which is considered the normal state of the system. The measure of deviation is the tendency of a system's model to be farther from the MCP after a sequence of observations, expressed as a probability that the deviation is incidental.  Factors that apply to the total of systems, such as the weather conditions are thus minimized.The algorithms approach the problem from the scopes of: linear and non linear relations between signals, distribution of values of a single signal, spectrum information of a single signal. This is achieved by constructing relevant models of each observed system (bus). The performance of the implemented algorithms is investigated using ROC curves and real bus fleet data, targeting at predicting a set of malfunctions of the air pressure system.More tests are performed using artificial data with injected malfunctions, to evaluate the performance of the methods. By applying the method on artificial data, the ability of different methods to detect different malfunctions is exhibited.

Advances in machine condition monitoring technologies are driven by the rise in complexity of modern machines and the increased demand for product reliability. Condition monitoring research tends to focus on the development of signal processing algorithms that are sensitive to machine faults, robust under time-varying operating conditions, and informative regarding the nature and extent of machine faults. A significant challenge remains for monitoring the condition of machines that are subject to time-varying operating conditions. The here presented work is concerned with the development of cost effective condition monitoring algorithms. It is investigated how empirical models (including probability density distributions and regression functions) may be used to extract diagnostic information from machine response signals that have been generated under fluctuating operating conditions. The proposed methodology is investigated on a number of case studies, including gearboxes, alternator end windings, and haul roads. It is shown how empirical models for machine condition monitoring may generally be implemented according to one of two basic approaches. The two approaches are referred to as discrepancy analysis and waveform reconstruction. Discrepancy analysis is concerned with the comparison of a novel signal to a reference model. The reference model is sufficiently expressive to represent vibration response as measured on a healthy machine over a range of operating conditions. The novel signal is compared to the reference model in such a manner that a discrepancy signal transform is obtained. A discrepancy signal is sensitive to faults, robust to time-varying operating conditions, and inherently simple. As such it may further beWaveform reconstruction implements a regression function to model machine response as a function of different state space variables. The regression function may subsequently be exploited to extract diagnostic information. The machine response may for instance be reconstructed at a specified steady state operating condition. This renders the signal wide-sense stationary so that Fourier analysis may be applied. analysed in order to extract periodicities and magnitudes as diagnostic markers. / Dissertation (MEng)--University of Pretoria, 2013. / gm2014 / Electrical, Electronic and Computer Engineering / unrestricted

It is widely acknowledged that bearing failures are the primary reason for breakdowns in rotating machinery. These failures are extremely costly, particularly in terms of lost production. Roller bearings are widely used in industrial machinery and need to be maintained in good condition to ensure the continuing efficiency, effectiveness, and profitability of the production process. The research presented here is an investigation of the use of acoustic emission (AE) to monitor bearing conditions at low speeds. Many machines, particularly large, expensive machines operate at speeds below 100 rpm, and such machines are important to the industry. However, the overwhelming proportion of studies have investigated the use of AE techniques for condition monitoring of higher-speed machines (typically several hundred rpm, or even higher). Few researchers have investigated the application of these techniques to low-speed machines (<100 rpm), This PhD addressed this omission and has established which, of the available, AE techniques are suitable for the detection of incipient faults and measurement of fault growth in low-speed bearings. The first objective of this research program was to assess the applicability of AE techniques to monitor low-speed bearings. It was found that the measured statistical parameters successfully monitored bearing conditions at low speeds (10-100 rpm). The second objective was to identify which commonly used statistical parameters derived from the AE signal (RMS, kurtosis, amplitude and counts) could identify the onset of a fault in either race. It was found that the change in AE amplitude and AE RMS could identify the presence of a small fault seeded into either the inner or the outer races. However, the severe attenuation of the signal from the inner race meant that, while AE amplitude and RMS could readily identify the incipient fault, kurtosis and the AE counts could not. Thus, more attention needs to be given to analysing the signal from the inner race. The third objective was to identify a measure that would assess the degree of severity of the fault. However, once the defect was established, it was found that of the parameters used only AE RMS was sensitive to defect size. The fourth objective was to assess whether the AE signal is able to detect defects located at either the centre or edge of the outer race of a bearing rotating at low speeds. It is found that all the measured AE parameters had higher values when the defect was seeded in the middle of the outer race, possibly due to the shorter path traversed by the signal between source and sensor which gave a lower attenuation than when the defect was on the edge of the outer race. Moreover, AE can detect the defect at both locations, which confirmed the applicability of the AE to monitor the defects at any location on the outer race.

The aim of this study is to apply computed order tracking with subsequent rotation domain averaging and statistical analysis to typical mining environments. Computed order tracking is a fault detection method that is unaffected by varying speed conditions often found in industry and has been proven effective in laboratory conditions. However in the controlled environment of a laboratory it is difficult to test the robustness of the order-tracking procedure. The need thus exists to adjust the order tracking procedure so that it will be effective in the mining environment. The procedure needs to be adjusted to function with a two pulse per revolution speed input. The drag gear aboard a dragline rotates in two directions. This gives the unique opportunity to observe the performance of the order tracking method in a bi-directional rotating environment allowing relationships between the results of each operating direction to be investigated. A monitoring station was set up aboard a dragline and data was captured twice daily for a period spanning one year. The data captured consisted of accelerometer and proximity sensor data. The key on the shaft triggers the proximity sensors allowing speed and direction to be measured. The rudimentary measured speed is interpolated using various documented speed interpolation techniques and by a newly developed speed interpolation technique. The interpolated speed is then used to complete the order tracking procedure that re-samples the vibration data with reference to the speed. The results indicate that computed order tracking can be successfully implemented in typical mining environments. Furthermore there is a distinct relationship between vibration data taken in both rotational directions: one direction provides a better indication of incipient failure. It is thus important not to choose a direction randomly when monitoring rotating machinery of this kind. / Dissertation (MEng)--University of Pretoria, 2008. / Mechanical and Aeronautical Engineering / unrestricted

The growth of global competition has led to remarkable changes in the way manufacturing companies operate. These changes have affected maintenance and made its role even more crucial for business success. To remain competitive, manufacturing companies must continuously increase the effectiveness and efficiency of their production processes. Furthermore, the introduction of lean manufacturing has increased concerns regarding equipment availability and, therefore, the demand for effective maintenance. That maintenance is becoming more important for the manufacturing industry is evident in current discussions on national industrialization agendas. Digitalization, the industrial internet of things (IoT) and their connections to sustainable production are identified as key enablers for increasing the number of jobs in industry. Agendas such as “Industry 4.0” in Germany and “Smart Industry” in Sweden are promoting the connection of physical items such as sensors, devices and enterprise assets, both to each other and to the internet. Machines, systems, manufactured parts and humans will be closely interlinked to collaborative actions. Every physical object will formulate a cyber-physical system (CPS), and it will constantly be linked to its digital fingerprint and to intensive connection with the surrounding CPSs of its on-going processes. That said, despite the increasing demand for reliable production equipment, few manufacturing companies pursue the development of strategic maintenance. Moreover, traditional maintenance strategies, such as corrective maintenance, are no longer sufficient to satisfy industrial needs, such as reducing failures and degradations of manufacturing systems to the greatest possible extent. The concept of maintenance has evolved over the last few decades from a corrective approach (maintenance actions after a failure) to a preventive approach (maintenance actions to prevent the failure). Strategies and concepts such as condition based maintenance (CBM) have thus evolved to support this ideal outcome. CBM is a set of maintenance actions based on the real-time or near real-time assessment of equipment conditions, which is obtained from embedded sensors and/or external tests and measurements, taken by portable equipment and/or subjective condition monitoring. CBM is increasingly recognized as the most efficient strategy for performing maintenance in a wide variety of industries. However, the practical implementation of advanced maintenance technologies, such as CBM, is relatively limited in the manufacturing industry. Based on the discussion above, the objective of this research is to provide frameworks and guidelines to support the development and implementation of condition based maintenance in manufacturing companies.  This thesis will begin with an overall analysis of maintenance management to identify factors needed to strategically manage production maintenance. It will continue with a focus on CBM to illustrate how CBM could be valued in manufacturing companies and what the influencing factors to implement CBM are. The data were collected through case studies, mainly at one major automotive manufacturing site in Sweden. The bulk of the data was collected during a pilot CBM implementation project. Following the findings from these efforts, a formulated maintenance strategy is developed and presented, and factors to evaluate CBM cost effectiveness are assessed. These factors indicate the benefits of CBM, mostly with regard to reducing the probability of experiencing maximal damage to production equipment and reducing production losses, particularly at high production volumes. Furthermore, a process of CBM implementation is presented. Some of the main elements in the process are the selection of the components to be monitored, the techniques and technologies for condition monitoring and their installation and, finally, the analysis of the results of condition monitoring. Furthermore, CBM of machine tools is presented and discussed in this thesis, focusing on the use of vibration monitoring technique to monitor the condition of machine tool spindle units. / INNOFACTURE - innovative manufacturing development