Coaches and exercise physiologist have found that structured approaches to training athletes consistently produce superior results. Such approaches require the collection of large amounts of data on the athlete’s training regimen. This is particularly true for cyclists, where the availability and sophistication of monitoring devices has increased substantially. The new generation of cycling computers can record data such as heart-rate, power, altitude, cadence, location, time and speed at rates approaching 10 hz. This mass of training data may contain valuable information about the athlete’s current state (e.g., level of conditioning, over-training, the onset of illness, and so forth). Existing approaches to analyzing the data largely depend on a coach’s or athlete’s ability to interpret simple graphical representations of the data. There has been no effort in applying the sophisticated analysis techniques that are in widespread use in the process industries to this application.
The process industries have taken advantage of developments in the areas of monitoring, fault detection and diagnosis to enhance their performance over the past three decades. The techniques that havebeen developed for these industries have proven their ability to extract useful information about the current state of a process from large volumes of data. In this project, we well extend these existing methods to athletic performance data and develop new methods for data analysis. One of the key differences that this research program will have to confront is that the process industries tend to operate for very long periods of time at a specified set of operating conditions; whereas, an athlete’s training regiment typically involves frequent changes in “operating level”, which is commonly called interval training.
We are currently working with the world’s leading manufacturer of computers used by competitive cyclists. These are the devices that all of the elite cycling athletes use.
