├── .gitattributes
├── ANN_DataPreprocessing.py
├── Abstract.pdf
└── Multivariate_windspeed_prediction_ANN.m


/.gitattributes:
--------------------------------------------------------------------------------
1 | # Auto detect text files and perform LF normalization
2 | * text=auto
3 | 


--------------------------------------------------------------------------------
/ANN_DataPreprocessing.py:
--------------------------------------------------------------------------------
 1 | f = open("temp.csv","r")
 2 | str=f.read()
 3 | f1=open("means.csv","w")
 4 | li=str.split('\n')
 5 | i=1
 6 | while len(li) > 9:
 7 | ti=li[:10]
 8 | try:
 9 | ti=[float(x) for x in ti]
10 | except ValueError,e:
11 | print ("error",e,"on line",i)
12 | li=li[10:]
13 | m=sum(ti)/10
14 | print(m,file=f1)
15 | i=i+1
16 | f.close()
17 | f1.close()
18 | 


--------------------------------------------------------------------------------
/Abstract.pdf:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/Sohan-Rai/Multivariate-windspeed-prediction-using-ANN/53eb205e721424e65d04b8bf22417655c9f05dd7/Abstract.pdf


--------------------------------------------------------------------------------
/Multivariate_windspeed_prediction_ANN.m:
--------------------------------------------------------------------------------
 1 | Y=xlsread('apr5speed.xlsx');
 2 | t=Y';
 3 | Y1=xlsread('apr5direction.xlsx');
 4 | t1=Y1';
 5 | X=con2seq(t1);
 6 | T=con2seq(t);
 7 | N = 144; 
 8 | inputseries  = X(1:end-N);
 9 | targetseries = T(1:end-N);
10 | inputseriesvalidation  = X(end-N+1:end);
11 | targetseriesvalidation = T(end-N+1:end); 
12 | delay =200;
13 | hiddenlayer =20;
14 | net = narxnet(1:delay,1:delay,hiddenlayer);
15 | [Xs,Xi,Ai,Ts] = preparets(net,inputseries,{},targetseries); 
16 | net = train(net,Xs,Ts,Xi,Ai);
17 | Y = net(Xs,Xi,Ai); 
18 | perf = perform(net,Ts,Y);
19 | inputseriesprediction = [inputSeries(end-delay+1:end),inputseriesvalidation];
20 | targetseriesprediction = [targetseries(end-delay+1:end), con2seq(nan(1,N))];
21 | netc = closeloop(net);
22 | [Xs,Xi,Ai,Ts] = preparets(netc,inputseriesprediction,{},targetseriesprediction);
23 | prediction = netc(Xs,Xi,Ai);
24 | perf = perform(net,prediction,targetseriesvalidation);
25 | figure;
26 | plot([cell2mat(prediction);    cell2mat(targetseriesvalidation)]')
27 | legend('Network Predictions','Expected Outputs')
28 | 


--------------------------------------------------------------------------------