Nonlinear
Equality Constraints
The
basic FSQP algorithm was designed for nonlinear inequality
constraints only. To handle nonlinear equality constraints,
FSQP incorporates a modification of a scheme due to Mayne
& Polak. Equality constraints
are
turned into inequality constraints
The
algorithm generates iterates that are feasible for the new
inequalities Negative values are penalized, i.e. the objective
function is replaced with
where
are
positive penalty parameters (iteratively increased, but bounded).
The
result is a differentiable exact penalty
function.
Back
to the top |
