In　this　paper,　the　optimal　beamforming　problem　of　multi-input　single-output　(MISO)　cognitive　radio　(CR)　downlink　networks　with　simultaneous　wireless　information　and　power　transfer　is　studied.　Due　to　the　nonconvexity　of　the　objective　function,　the　considered　nonconvex　optimization　problem　is　firstly　transformed　to　an　equivalent　subtraction　problem　and　then　an　approximated　convex　optimization　problem　is　obtained　by　using　the　successive　convex　approximation　(SCA).　When　the　instantaneous　channel　state　information　(CSI)　of　the　eavesdropping　link　is　unknown　to　the　legitimate　transmitter,　another　interruption-constrained　energy　efficiency　optimization　problem　is　proposed　and　the　Bernstein-type　inequality　(BTI)　is　used　to　conservatively　approximate　the　probability　constraint.　The　paper　proposes　a　two-level　iterative　algorithm　based　on　Dinkelbach　to　find　the　optimal　solution　of　the　EE　maximization　problem.　Numerical　results　validate　the　effectiveness　and　convergence　of　the　proposed　algorithm.