I have gotten this to work!!
The problem I was having was with inrush current; incandescent light bulbs have an initial inrush current around 10x the normal operating current during the heating of the elements. Even though my transistors were rated well over the operating current of around 6A, they were reaching avalanche breakdown during the current inrush.
My first and most primitive solution to this problem was to install a simple toggle switch, allowing the lights to start from line power, and then be switched to power through transistor after the bulbs were heated. The result: Success! The problem: in this circuit the lights cannot be dimmed to 0% without completely resetting the switches and starting the process over.
The next solution is to insert a 1ohm NTC thermistor in the line before my transistor, the thermistor caps the current inrush at 9A, well below the 18A capacity of my transistors. This would allow the circuit to run without any further tampering or startup procedures. I'll be testing soon to confirm.
A note on using PWM with incandescent lights:
Incandescent lights work exceptionally well with PWM by way of MOSFET transistor. With a 200V 18A transistor I was able to control a 150w halogen light without any other components. The bulbs dim very smoothly, and with a great deal of control available. The only limitation on current seems to depend on the current limitation of the NTC thermistor; 9A has been the highest current I can find one rated for. I have researched many different approaches for dimming high voltage high current lights, and this looks far easier and cheaper than anything else I have come across. The circuit should cost no more than $5 with heat sync included. I decided to make my own heat syncs out of 18ga copper sheet, and was able to keep the temperature of the transistors below 80degF with cooling fan, and approx. 110degF without.
Also, I was able to control these lights up to 6A without a resistor or buffer transistor on the gate supply, however I think I will be including both in future tests.
Hope this is of use to someone, hopefully I haven't missed any crucial details