You should know the end application of welded component in order to decide the type of fatigue test (stress controlled or strain controlled). Unless you specify the purpose for which the plate is being fabricated, it is difficult to comment (perhaps you may not require a fatigue test at all whereas a fracture test is more appropriate).

Anyways, typically you can go for push-pull fatigue loading, using a dog bone specimen (as suggested in  the earlier post). I would suggest to go for a whole weld specimen. That is the entire specimen extracted from  the weld metal alone. This will tell you if your weld metal is having sufficient fatigue resistance compared to the base metal or not. You have not mentioned the plate or weld dimensions. I therefore do not know if the extraction of whole weld specimen is possible or not.

This, however, is not enough to comment on the integrity of the welded region as a whole. You may have to look into the fatigue behaviour of heat affected zone as well. Prepare a similar dog bone specimen in such a manner that the HAZ is located in the centre.

You refer to ASTM E466 and E606 for the stress and strain controlled fatigue testing. Test procedure and specimen drawings are available in these standards. All that you require is a dynamic test frame with a facility to grip a flat specimen.

You also may require to look into the fatigue crack growth behaviour of weld and HAZ regions to comment about the performance of the weld region. Refer ASTM E 647 for this purpose.

As I begin with it all depends up on  your requirement/application.

check this: https://books.google.co.in/books?id=QrDrtHo47nMC&pg=PA67&lpg=PA67&dq=fatigue+testing+of+flat+plates&source=bl&ots=WYmCjvE7tp&sig=PWFmyxRgZ1qpB9NArU-6sGqfGzs&hl=en&sa=X&ei=vcK0VPq9MNSQuATM0oLgBA&ved=0CDMQ6AEwBQ#v=onepage&q=fatigue%20testing%20of%20flat%20plates&f=false

I think you can go for a standard fatigue test specimen of dog bone shape. You have to cut the same using a wire cut EDM. A transverse weld specimen (weld line is perpendicular to the loading direction) can be used to find the number of cycles for breakage corresponding to a particular R ratio. By doing for different R ratio's you can generate an SN curve for that particular weld and which will help you in modelling and predicting the fatigue life of the weld.   

You should know the end application of welded component in order to decide the type of fatigue test (stress controlled or strain controlled). Unless you specify the purpose for which the plate is being fabricated, it is difficult to comment (perhaps you may not require a fatigue test at all whereas a fracture test is more appropriate).

Anyways, typically you can go for push-pull fatigue loading, using a dog bone specimen (as suggested in  the earlier post). I would suggest to go for a whole weld specimen. That is the entire specimen extracted from  the weld metal alone. This will tell you if your weld metal is having sufficient fatigue resistance compared to the base metal or not. You have not mentioned the plate or weld dimensions. I therefore do not know if the extraction of whole weld specimen is possible or not.

This, however, is not enough to comment on the integrity of the welded region as a whole. You may have to look into the fatigue behaviour of heat affected zone as well. Prepare a similar dog bone specimen in such a manner that the HAZ is located in the centre.

You refer to ASTM E466 and E606 for the stress and strain controlled fatigue testing. Test procedure and specimen drawings are available in these standards. All that you require is a dynamic test frame with a facility to grip a flat specimen.

You also may require to look into the fatigue crack growth behaviour of weld and HAZ regions to comment about the performance of the weld region. Refer ASTM E 647 for this purpose.

As I begin with it all depends up on  your requirement/application.

Unless you specify what type of loading that is exerted on the weld, it is hard to give you an accurate answer. The push pull test with a load transverse to the weld, as suggested earlier, and the resulting S-N curve would be a model that should be used for fatigue life prediction restricted only to that type of loading.

ASTM E466 and E468 are applicable for push-pull test.