You can consider translating the weights to frequency domain to obtain the appropriate frequency response. Then you can compute the pattern at the given frequency.
M = 16;
wfreq = fft(reshape(yweights,16,[]).',M);
freqvec = [0:7 -8:-1]*fs/M;
bandidx = 5;
freqbin = 2e3;
hstv = phased.SteeringVector('SensorArray',h,'PropagationSpeed',c);
sv = step(hstv,freqbin,incidentAngle);
figure
pattern(h,freqbin,-180:180,0,'PropagationSpeed',c,'Weights',sv.*wfreq(bandidx,:).',...
'CoordinateSystem','polar','Type','powerdb');
HTH
Best Answer