Last year I made a thread* where I tried to compare Bannister's training impulse model with a modern power based model (performance manager). While I still use the training impulse model to some extent it is difficult to determine the various constants in the model for a given sport/athlete.
http://www.bikeforums.net/showthread.php?t=289060
If you have a 5 zone HRM it is easy to calculate each workouts TRIMP. However, due to the aforementioned difficulties in fitting Bannister's model, I suggest you just use the obtained TRIMP scores in place of TSS and use the exponentially weighted moving average system to model the training dose-response, as done by the power based model (CTL, ATL, TSB).
*Ensure you read the whole thread as corrections were made in later posts.