trying to use a neural network in a genetic optimiser

hi, i have created a neural net using 6 inputs and one output, its for a project to optermize intake length and valve timings for an engine

i am happy that the network is working as i intended but i dont know how to generate a function from the network and then use this funtion in the genetic algorithm optimizer, any help would be appieciated

here is my code for the network

filename3 = 'data for network.xlsx';

%x1 = rpm

x1 = xlsread(filename3,'C4:AGW4');

%x2 = length

x2 = xlsread(filename3,'C5:AGW5');

%x3 = IVO timing

x3 = xlsread(filename3,'C6:AGW6');

%x4 = IVC timing

x4 = xlsread(filename3,'C7:AGW7');

%x5 = EVO timing

x5 = xlsread(filename3,'C8:AGW8');

%x6 = EVC timing

x6 = xlsread(filename3,'C9:AGW9');

%x = {x1;x2;x3;x4;x5;x6};

x25 = [x1;x2;x3;x4;x5;x6];

%y1 = hp

y1 = xlsread(filename3,'C10:AGW10');

net_hp = feedforwardnet(10);

net_hp = configure(net_hp,x25,y1);

net_hp = init(net_hp);

[net_hp,tr] = train(net_hp,x25,y1);

Answer 1

[Abdolkarim Mohammadi]

If I have understood you well, you want ga() to train your feedforward ANN. If so, you can read about it here:

https://www.mathworks.com/matlabcentral/answers/100323-how-can-i-use-the-genetic-algorithm-ga-to-train-a-neural-network-in-neural-network-toolbox

[luke haworth]

not quite the network is already trained

the network has targets also (horsepower)

i need to optimize the network and then get it display the input values for its best optimization

[luke haworth]

basically i want the network/optimizer to find the best possible value for horsepower (highest value) and then show the network inputs for the best best (highest) output

[Abdolkarim Mohammadi]

I think you have a surrogate model, that is, you have trained a network that gets inputs and returns the horsepower. If so, your job is very easy. Just put the network inside a function

function ObjectiveFunctionValue = ObjectiveFunction (x, net)
ObjectiveFunctionValue = net (x);
end

and pass it as the objective function to ga() or anything else. For example:

fun = @ (x) ObjectiveFunction (x, net);
nvars = % number of decision variables (inputs of the network) 
[x, fval] = ga (fun, nvars);

Read more in ga() documentation.