The selection of MOSFET depends upon various parameters in which one important parameter is Vds.
Max drain to source voltage is the rating given to the parts ability to block voltage applied to it when it is off. A good rule of thumb is to choose a part whose voltage rating is twice that of the expected voltage applied to the drain. The reasoning behind this is that short voltage spikes well above the input voltage are common in electrical systems where switching MOSFETs are present.

So if I have a DC voltage of 350V from the DC link, according to what I have understood from the above, I have to choose a mosfet whose Vds_max is twice the voltage applied to drain which is 350V*2=700V Vds max. Can anybody tell me if it's the correct way to do?

 

There is no reason to select a FET with a voltage rating twice what is expected. 20% margin, in most cases, is quite sufficient.

In certain applications, like automotive, what is expected under unusual but distinctly possible conditions may be much higher than the nominal system voltage, but it is still expected. For example, "load dump" with a battery in poor condition might allow a transient much higher than the usual max of about 14 volts in a nominally 12 V system (alternator regulators are pretty slow). Boosting with 24 volts, which might be nearer 30 V, is sometimes done, but very foolish in a modern car.

I have routinely used FETs rated at 500 V in high-power switch mode power supplies where the nominal operating voltage was about 400 V (half bridge circuit). A flyback converter operating from 400 V would typically require a FET rating of 600 to 700 V, which was an expected repetitive level (actually a design target for the transformer). I've deliberately allowed the FET in a small "housekeeping supply" flyback converter to "avalanche" due to voltage higher than it's nominal rating while discharging energy in the leakage inductance of the transformer primary - but I knew how much energy was involved and the FET was rated to withstand this.

You must carefully analyze what the maximum expected voltage is going to be in the circuit, meaning not just for normal operation but for predictable unusual circumstances. The 2x rule of thumb isn't a rule. It will result in a more expensive, lower performing part.

 

Thanks for you kind and clear explanation. Specifically speaking about renewable energy application, in which the mosfets are used in converters, my application targets Rds ON to be less than 1m ohm. With this selection there are no mosfets with Vds more than 60V.
I am attaching the image of the circuit which I designed using LTspice. Could you take it as a example and help me in coming to a conclusion on choosing the mosfet for this applicaiton?