Assuming when you say transfer function you are not referring to an S-domain "transfer function" for the behaviour (as this is actually really hard and relies on alot of heaviside functions) but more of a relationship of input to output.
ASUMMUMING the supply is an ideal supply with each voltage sources being 120degree's separated and their amplitudes & freq are all the same and that there is no supply feeder impedance
With the firing angle = 0 and thus the SCR's act like diodes, we know that:
Vd = \$\frac{3\sqrt{2}}{\pi}V_L\$
Where Vd = DClink voltage and \$V_L\$ = the Line-Line (rms) voltage from the supply
We want to know what Vd is with regards to some arbitary firing angle \$\alpha\$
\$V_\alpha = Vd - \frac{A_\alpha}{\pi/3} \$
\$A_\alpha\$ is the volts-second area that occurs every 60degrees which reduces the average DClink.
We know that:
\$V_a = \sqrt{2} V_L Sin(\omega t)\$
Thus
\$A_\alpha = \int\limits_0^\alpha \sqrt{2}V_LSin(\omega t) d(\omega t) \$
\$ = \sqrt{2}V_L(1-cos\alpha)\$
Thus
\$ V_\alpha = \frac{3\sqrt{2}}{\pi} V_Lcos\alpha \$