Why do I receive different optimized values by varying the order of nonlinear constraints in FMINCON

I am applying fmincon function on motion optimization of robotic arm. The only constraints that I consider, are the set of nonlinear constraint. The objective and set of constrains are written below.

objective = @(h) h(1)+h(2)+h(3)+h(4)+h(5)+h(6)+h(7)+h(8)+h(9);

subjected to following constraints

1. Ve(i)=abs[(w(i+1)/2)*h(i)+(q(i+1)-q(i))/h(i)+((w(i)-w(i+1))*h(i))/6] =<400
2. Jerk(i)= abs[(w(i+1)-w(i))/h(i)]=<240
3. A(i)= abs[(w(i)/(h(i)))*(t(i+1)-tx) + (w(i+1)/(h(i))) * (tx-t(i))] =<250

I have obtained different optimized values based on the different sequences of constraints (1,2,3). In 1st case I use the sequence (1,2,3) and in 2nd case , constraint’s sequence is (3,1,2). In both cases, different optimization values obtained even though the input values are same in both cases.

Case one results are;

1.7363 2.7841 2.0678 1.4065 0.6189 1.7987 2.4236 1.7065 1.3710

Final Objective: 15.9134

Case second results are;

1.0e+03 *

0.1922 0.3243 0.1647 0.0914 0.0828 0.9730 2.5193 1.8097 -0.0403

Final Objective: 6117.0441

May I know the reason behind this huge variations in the optimized values.

[Matt J]

Does it happen from all initial guesses?

[Muhammad Saeed]

I provide only one initial guess.And it remain same for all different orders.

[Matt J]

I think you should test different initial guesses.

[dpb]

Indeed. Also need to investigate It's possible can get "stuck" in local minima.