Well... everyone seems to be indicating that the market needs a large heated-bed solution whose on/off, power-hungry needs don't create banding in the printed material itself. Since most of us are inventors rather than artists, you'd think that we'd just invent a solution. Kudos for the various solutions as suggested above but as I hear all this I'm thinking that I could create a bed for the Robo 3D C2 which is a snap-in-place replacement. I'd start with one of these...
...I'd then cut each of the coils in half, buy more diodes of the same rating that appear to be on that chassis and then drive them digitally through relays with a car battery. I think I would arrange the coils in such a way that individual coils could be running while others are off. It would be nice to favor the area under where the print head is at any moment but heat propagation occurs much more slowly—you could program this to distribute heat to where the print head will be later but that's overkill. Heating the footprint area of the object to be printed is really all that's necessary.
From what I'm reading between-the-lines, heating the bed is mostly necessary for part adhesion (but may also take a part in preventing warpage). If it's just the former, then I'm guessing that the bed only needs to be heated until the first few millimeters of the base are printed and then may be ramped down from there, right? Is there really a need to heat the bed for four hours? I doubt it.
I'm guessing that milling some cooling fins into the bottom surface of an aluminum bed top would help to distribute the heat across the surface area so that nothing turns into a hotspot. In fact, you could also install one or two fans inside to circulate the hot air in a circle and even mill some air guides into the aluminum to support that, too. (Think "convention oven".)
I'm thinking at least one K probe under the surface is required, as coupled to the control circuit which doesn't technically have to be connected at all with the printer. A single dedicated Raspberry Pi Zero W inside the bed could run this show with a web interface. Dial in the temperature and the zone(s)/coil(s) to turn on, set a timer for how long it's supposed to run, set a ramp-down curve and finally, set a notification event for when it's ready to go. A cheap version could probably be built for under $100 without the milling, methinks.