Detailed performance monitoring that can provide timely feedback on technique and physiological capability is a major issue within sport domains, particularly at the elite level where personalised, reconfigurable training programmes are required. The current practice of monitoring parameters as measures of athlete performance requires significant human interaction with the aid of several independent and isolated devices. The research included collaborations with British Swimming (the national governing body for swimming in the U.K.) and British Cycling (the national governing body for cycling in the U.K.), to drive the requirements and direction of research. This thesis details the research undertaken towards improving performance-related feedback in swimming and cycling domains. A multi-sensor enabled integrated monitoring system using a range of sensor modalities to monitor the different phases of elite swimmer performance (i.e. starts, free swimming and turns) has been designed to deliver performance indicative features to sports scientists/coaches in real-time at the pool-side to enhance the feedback mechanism. Within the cycling application, a novel automated cycling ergometer that: (1) allows faster, automated and more accurate configuration for different cyclists' preferred bicycle geometry setup via motor controlled actuators and (2) provides a motor driven pedalling load enabling variable resistance profiles has been designed. The cycling system also enables simultaneous access of biomechanical and kinematical feedback via the integration and synchronous feedback from multiple sensor modes.Structured methodology using CIMOSA reference architecture has been proposed to (1) design the user interfaces for the system and (2) decompose the system to identify potential software components. Several software components have been implemented adhering to object-oriented architecture principles to (1) acquire data from the sensors, (2) process and extract performance features via the use of signal processing techniques and (3) store the raw signals as well as the performance features. The flexibility of the methodologies was validated by applying it to both applications. The development of the systems is presented in detail and the validation of the functional and non-functional requirements is also presented. Symbolic regression, which is a process of obtaining mathematical models that fit a given data set, has been applied to predict the turn performance within the swimming domain. The integrated, multi-sensor based system has been developed and tested to prove its ability to produce useful, quantitative feedback information. Each of the sensor technologies were not used in isolation but were supported by other synchronous data capture. The integrated system was found to be capable of providing greater insight into performance that has not been previously possible using the current state of the art techniques. Future work should look recruiting elite/sub-elite cyclists for extensive trials using the integrated cycling performance monitoring tool. Additionally, future work should also involve the refinement of a couple of software components to enable their rapid reuse within other domains involved in product/process monitoring.
© Copyright 2017 Veröffentlicht von Loughborough University. Alle Rechte vorbehalten. / Übersetzt mit https://www.DeepL.com/Translator