Hi Phillippe
If you have Statistics Toolbox or Curve Fitting Toolbox its relatively easy to force this type of constraint.
Here's a simple example using Curve Fitting Toolbox where I constrain the intercept to take the value "4" by setting the upper and lower bound for the regression coefficient to be 4.
X = 1:100
X = X'
Y = 3* X + 5 + randn(100,1)
foo = fit(X, Y, 'poly1', 'Lower', [-Inf 4], 'Upper', [Inf 4])
