Mystery chess position III: my first endgame study

The position on the left with black (me) to move is from one of my recent games at ChessWorld. I ended up loosing the game because of an outrageous blunder at move 47 that was followed by immediate checkmate (Congratulations to my opponent beartrostle). Anyway, I noticed that I tend to loose less material lately and as a consequence a larger fraction of my games is decided in the endgame. Unfortunately, I am utterly clueless on using my pawns properly. Here, for instance, I played 45. ... f5 with the idea to attack the lonely white b-pawn. This turned out to be a big mistake! As usual, click here for the fully annotated game.

Loosing the game motivated me to take a closer look at this endgame. During the game, I thought that black has better chances because the white rook is tied up defending the pawn on the h-file and black has 2 vs. 1 pawn on the other side of the board. After playing through several variations with the help of Shredder (see my recent review), however, it looks more like a draw to me. Black can check the white K in order to remove the defender from the white b-pawn but after move sequences like 45. ... Ra3+ 46.Kg2 Kg4 47.Rg8+ Kf5 48.Ra8 Ra2+ 49.Kh3 Kg5 50.Rg8+ things are basically back to the starting point (if black makes sure to prevent white from promoting the h-pawn, which I did not). Does anyone see a winning strategy for either side? As far as my understanding goes, the key defensive strategy for black is to prevent white from advancing the h-pawn. Once it is on the 7th rank, white can use the rook to check the black king and promote the pawn on the next move. After that, black is forced to capture the newly born queen with ... Rxa8 and white ends up winning a rook by recapturing with Rxa8.

The biggest problem for me is to learn how to analyze this kind of situation. Endgames consist of much more moves than standard tactical problems. I tried to learn from observing what the computer plays from the above starting position, but most moves simply did not make much sense to me. Therefore, I might be better off spending time playing through more simple situations from beginner-level endgame books than to look at my own games (at least for now). Speaking of that, see PE's chess cast blog for a nice movie explaining the principles behind "bad" bishops in endgames.