If you completed Cheung CX - Ronnestad –
…and came away with any sort of “easier” appraisal then you have a latent BT waiting to be expressed. 
You can prove that next time by riding the Ronnestad sets continuously in Slope mode, extending duration or upping watts or adding another interval until a failure point is reached by the end of each set.
Cheung should force this issue if your signature is valid –
You want the same to occur with A Pain for at least one set –
1 watt is normal noise. So is 5 watts down if decay does its thing without sufficient rise in TL or punching through MPA now and then. Those punches can be anywhere along your curve.
Three months without a BT is too long. The stale warning appears after three weeks. That means do something which can be any sort of max effort under fatigue.
My new favorite method is to leverage XMB for doing that (Garmin or Android) – which you can do at any time. A Ronnestad workout in Slope mode would be my second choice.
Every progression requires an upward trend in one table value or another. If not, the prediction reflects a lateral move that is unlikely to produce much if any change.
If limited by max hours, you’ll have to add more punch (high/peak) into the equation as Scott suggests.
IMO the best way to peak is to start in the valley. 
How much downtime do you schedule during a season?
How toothy are your year-over-year progressions? (TL deltas)
Here’s my five peak 5-year profile (XPMC set to TL) –
These are all BBP progressions including periods of rising TP due solely to the increase in TL.
I did insert BT efforts occasionally on HIT days if the 3-week warning appeared.
I also perform a fitness test of some sort at the end of a program to verify where I’m at.