Look up table optimizer

Dear All, I've a simulink model of non-Linear equation with a lookup table in it. Say for example consider the following equation, X=a.x1+b.x2+c.(sqrt(x1));

Where the a,b,c are the independent variable ; X is dependent variable(model value) x1,x2 are the output of two 1D- lookup tables which depends on other independent variable(d,e).

Here the goal is, i've the measured data of X,a,b,c,d,e. I need to optimize the lookup tables for reducing the error between the model (output of the equation) and the measured values of X. In all the solvers available on matlab i could use only nlinfit solver to solve the above problem as many solver doesn't run because of the usage of interp function inside the solver function. Following is the way i use in nlinfit

Lookuptable_BreakPoints=1:8;
Lookuptable1=[10,20,30,40,50,60,70,80]; % Initial dummy value
Lookuptable2=[10,20,30,40,50,60,70,80]; % Initial dummy value
B= [Lookuptable1,Lookuptable2]; % Initial Guess
data= [a,b,c,d,e]; Output=X;
nlinfit( data,Output,@solver_function,B) 

function Output= solver_function(B,data)
a=data(:,1);b=data(:,2);c=data(:,3);d=data(:,4);e=data(:,5);
Lookuptable1=B(1:8);
Lookuptable2=B(8:16);
x1=interp1(1:8,Lookuptable1,d);
x2=interp1(1:8,Lookuptable2,e);
Output=a.*x1+b.*x2+c.*(sqrt(x1));

This process is perfectly optimizing my data, but the problem here is i need both the Lookuptables values to be in increasing order only even if some error is available it is ok . Here for reducing the SSE the solver gives a very random value even in first iteration itself

Is there any solver available which increases/decreases the initial guess in a step wise manner so that i can stop the iterations once i find the trend of B changes more ? Or is there any other way to solve using lookup table optimizer?