The main issue you have is that you are only randomly sampling from a small set of frequencies and you are doing that 18 times, then you don't vary the phase at all. Therefore, you're going to generate the same frequency more than once. You're adding sine waves at the same frequency and phase with different amplitudes. Therefore the resulting amplitude is going to be the sum of the individual amplitudes.
I'll verify this for a frequency of 4 Hz.
Fs = 200;
t= 0:1/Fs:180-1/Fs;
y_in=zeros (1, length(t));
for i = 1:18
F = randi ([4 7], 1);% frequency
freqv(i) = F;
A = randi ([75 200],1);% amplitude A=75~200 uV.
amp(i) = A;
y_tmp =A*sin (2*pi*F*t);
y_in=y_in+ y_tmp;
end;
L=length(y_in);
Y_new = fft(y_in);
Y_new = Y_new(1:length(y_in)/2+1)/L;
Y_new(2:end-1) = 2*Y_new(2:end-1);
f_new = Fs/2*linspace(0,1,length(y_in)/2+1);
plot(f_new,abs(Y_new));
title('Single-Sided Amplitude Spectrum of y_in(t)')
xlabel('Frequency (Hz)')
ylabel('|y_in(f)|')
Note that if we find the number of times, 4 Hz is added and sum the amplitudes of those components, you get your answer.
indices = find(freqv == 4);
amp4 = sum(amp(indices));
You see that amp4 is the amplitude of your component at 4 Hz in the plot.
