MATLAB: Norm form and prediction

Deep Learning Toolboxnnstartnntoolnormalize

i want to use ann for forecasting water table

i have two set of data. water table from past and temperature.

it means two matrix with 12 column and 15 row

12 column means 12 month

and each row are separated data from a year(15 year).

it mean i have two data for month and year

i devided my data into 2 matrix

first 11 year for input and 4 year for output

i want to use nntool to forecast and predict water table for future years with under consideration of temperature. now my questions:

some of my data don't have value and matlab give them 0 in my table. is this affect the network? and if it is what should i do?
do i need to normalize my data before using nntool and how?
when i start nnstart, do i need to use time series?
how i consider the effect of temperature?

sorry for my long question. I am new in matlab and have a lot of question. thanks a lot.

Best Answer

I would unfold your data into a single 2 dimensional input time series matrix of size

[ 2 180 ] = size(series)

where the inputs are waterlevel and temperature.

The target is just the future waterlevel.

To estimate how far into the future you can predict, obtain the autocorrelation function of waterlevel and crosscorrelation function of waterlevel and temperature.

From the correlation functions you can determine the significant lags of the local correlation peaks.

Then use a subset of the significant lags in a narxnet.

Hope this helps.

Thank you for formally accepting my answer

Greg

Related Solutions

MATLAB: Prediction of remaining useful life

Sir , this is my proposed result where the no of cycles are predicted from a particular point on x axix . the graph is between cycle and capacity. Also for your reference i am attaching my code. I have used 4 battery dataset . As told earlier , input are voltage current tempertaure and capacity whereas output is capacity. i trained my nn by utilizing 2 dataset as input and testing the datset from which i wish to acquire results. But i am unable to figure out the prediction as shown in fig attached from 50, 60 and 70 cycles which i wish to perform.


 
clc
clear all
load B0005
load B0006
load B0007
load B0018
%% load the value of capacity from B0005 B0006 B0007 and B0018
D_cycle1={B0005.cycle.type};
D_cycle2={B0006.cycle.type};
D_cycle3={B0007.cycle.type};
D_cycle4={B0018.cycle.type};
D_cycle1=D_cycle1';
D_cycle2=D_cycle2';
D_cycle3=D_cycle3';
D_cycle4=D_cycle4';
index1 = find(ismember(D_cycle1, 'discharge'))
index2 = find(ismember(D_cycle2, 'discharge'))
index3 = find(ismember(D_cycle3, 'discharge'))
index4 = find(ismember(D_cycle4, 'discharge'))
for i=1:length(index1)
    capacity1(i)=B0005.cycle(index1(i,1)).data.Capacity
end
index=[]
for i=1:length(index2)
    capacity2(i)=B0006.cycle(index2(i,1)).data.Capacity
end
index2=[]
for i=1:length(index3)
    capacity3(i)=B0007.cycle(index3(i,1)).data.Capacity
end
index3=[]
for i=1:length(index4)
    capacity4(i)=B0018.cycle(index4(i,1)).data.Capacity
end
index4=[]
capacity1 = capacity1'
capacity2 = capacity2'
capacity3 = capacity3'
capacity4 = capacity4'
capacity = [capacity1;capacity2;capacity3;capacity4]
%% load the value of voltage current and temperature from B0005 B0006 B0007 and B0018 each cycle(10 samples per cycle)
C_cycle1={B0005.cycle.type};
C_cycle2={B0006.cycle.type};
C_cycle3={B0007.cycle.type};
C_cycle4={B0018.cycle.type};
C_cycle1= C_cycle1'; %%B0005



C_cycle2= C_cycle2'; %%B0006




C_cycle3= C_cycle3'; %%B0007




C_cycle4= C_cycle4'; %%B0018



index5= find(ismember(C_cycle1,'charge')) %%5
index6= find(ismember(C_cycle2,'charge')) %%6
index7= find(ismember(C_cycle3,'charge'))  %%7
index18= find(ismember(C_cycle4,'charge')) %%18
%%B0005
cnt=0
for i=1:length(index5);
    for ii=1:round(length(B0005.cycle(index5(i)).data.Voltage_measured)/10):length(B0005.cycle(index5(i)).data.Voltage_measured);
     cnt=cnt+1   ;
            voltage1(i,cnt)=B0005.cycle(index5(i)).data.Voltage_measured(ii);
            current1(i,cnt)=B0005.cycle(index5(i)).data.Current_measured(ii);
            temperature1(i,cnt)=B0005.cycle(index5(i)).data.Temperature_measured(ii);
            
    end
    cnt=0;
end
index5=[]
%%B0006
for i=1:length(index6);
    for ii=1:round(length(B0006.cycle(index6(i)).data.Voltage_measured)/10):length(B0006.cycle(index6(i)).data.Voltage_measured);
     cnt=cnt+1   ;
            voltage2(i,cnt)=B0006.cycle(index6(i)).data.Voltage_measured(ii);
            current2(i,cnt)=B0006.cycle(index6(i)).data.Current_measured(ii);
            temperature2(i,cnt)=B0006.cycle(index6(i)).data.Temperature_measured(ii);
            
    end
    cnt=0;
end
index=[]
%%B0007
for i=1:length(index7);
    for ii=1:round(length(B0007.cycle(index7(i)).data.Voltage_measured)/10):length(B0005.cycle(index7(i)).data.Voltage_measured);
     cnt=cnt+1   ;
            voltage3(i,cnt)=B0007.cycle(index7(i)).data.Voltage_measured(ii);
            current3(i,cnt)=B0007.cycle(index7(i)).data.Current_measured(ii);
            temperature3(i,cnt)=B0007.cycle(index7(i)).data.Temperature_measured(ii);
            
    end
    cnt=0;
end
index=[]
%%B0018
for i=1:length(index18);
    for ii=1:round(length(B0018.cycle(index18(i)).data.Voltage_measured)/10):length(B0018.cycle(index18(i)).data.Voltage_measured);
     cnt=cnt+1   ;
            voltage4(i,cnt)=B0018.cycle(index18(i)).data.Voltage_measured(ii);
            current4(i,cnt)=B0018.cycle(index18(i)).data.Current_measured(ii);
            temperature4(i,cnt)=B0018.cycle(index18(i)).data.Temperature_measured(ii);
            
    end
    cnt=0;
end
index=[]
%%delete 11th column from B0005 as only 10 samples are required per cycle
voltage1(:,11)=[] 
current1(:,11)=[]
temperature1(:,11)=[]
%%delete 11th column from B0006 as only 10 samples are required per cycle
voltage2(:,11)=[] 
current2(:,11)=[]
temperature2(:,11)=[]
%%delete 11th column from B0007 as only 10 samples are required per cycle
voltage3(:,11)=[] 
current3(:,11)=[]
temperature3(:,11)=[]
%%delete 11th column from B0018 as only 10 samples are required per cycle
voltage4(:,11)=[] 
current4(:,11)=[]
temperature4(:,11)=[]
%% combine the values of voltage current temperature and capacity  
%%B0005   
for vk1=1:size(voltage1,1)
for i = 1 : find(voltage1(vk1,:),1,'last')
   D_voltage1(vk1).voltage1(i)=voltage1(vk1,i)
   D_voltage1(vk1).current1(i)=current1(vk1,i)
   D_voltage1(vk1).temperature1(i)=temperature1(vk1,i)
end
if vk1<=length(capacity1)
    D_voltage1(vk1).capacity1=capacity1(vk1)
else
    continue   
end
end
%%B0006
for vk2=1:size(voltage2,1)
for i = 1 : find(voltage2(vk2,:),1,'last')
   D_voltage2(vk2).voltage2(i)=voltage2(vk2,i)
   D_voltage2(vk2).current2(i)=current1(vk2,i)
   D_voltage2(vk2).temperature2(i)=temperature2(vk2,i)
end
if vk2<=length(capacity2)
    D_voltage2(vk2).capacity2=capacity2(vk2)
else
    continue   
end
end
%%B0007
for vk3=1:size(voltage3,1)
for i = 1 : find(voltage3(vk3,:),1,'last')
   D_voltage3(vk3).voltage3(i)=voltage3(vk3,i)
   D_voltage3(vk3).current3(i)=current3(vk3,i)
   D_voltage3(vk3).temperature3(i)=temperature3(vk3,i)
end
if vk3<=length(capacity3)
    D_voltage3(vk3).capacity3=capacity3(vk3)
else
    continue   
end
end
%% B0018
for vk4=1:size(voltage4,1)
for i = 1 : find(voltage4(vk4,:),1,'last')
   D_voltage4(vk4).voltage4(i)=voltage4(vk4,i)
   D_voltage4(vk4).current4(i)=current4(vk4,i)
   D_voltage4(vk4).temperature4(i)=temperature4(vk4,i)
end
if vk4<=length(capacity4)
    D_voltage4(vk4).capacity4=capacity4(vk4)
else
    continue   
end
end
%%delete 169th and 170th row as data until 168 cycle is required
D_voltage1(169:170) = [];
D_voltage2(169:170) = [];
D_voltage3(169:170) = [];
D_voltage4(133:134) = [];
%%combine the values of voltage current temperature and capacity
%%B0005
for i=1:length(D_voltage1)
input1(i,:)=[D_voltage1(i).voltage1 D_voltage1(i).current1 D_voltage1(i).temperature1 D_voltage1(i).capacity1]
end
%%B0006
for i=1:length(D_voltage2)
input2(i,:)=[D_voltage2(i).voltage2 D_voltage2(i).current2 D_voltage2(i).temperature2 D_voltage2(i).capacity2]
end
%%B0007
for i=1:length(D_voltage3)
input3(i,:)=[D_voltage3(i).voltage3 D_voltage3(i).current3 D_voltage3(i).temperature3 D_voltage3(i).capacity3]
end
%%B0018
for i=1:length(D_voltage4)
input4(i,:)=[D_voltage4(i).voltage4 D_voltage4(i).current4 D_voltage4(i).temperature4 D_voltage4(i).capacity4]
end
input = [ input1; input2; input3; input4 ] %% complete input merged to form one input
output1=capacity1; %%B0005
output2=capacity2; %%B0006
output3=capacity3; %%B0007
output4=capacity4; %%B0018
%% normalize the input
for u=1:size((input),2)
normalize(:,u) = ((input(:,u)-min(input(:,u)))./(max(input(:,u))-min(input(:,u))).*2)-1;
end
input = normalize; 
target = capacity;
%% divide the data
data_2 = input(169:504,:)  %% 2 denotes two batteries for training(B0006 B0007)
data_1 = input(1:168,:)    %% 1 denotes one battery for testing(B0005)
% Separate to training and test data
data_train = input(169:504,:)';
data_test  = input(1:168,:)';
target_train=target(169:504,:)'
target_test=target(1:168,:)
%% BPNN algorithm 
 for pp=1:2
   tic  
net=newff(minmax(data_train),[20 1],{'logsig','purelin'},'trainlm');
net.trainParam.epochs=1000;
net.trainParam.goal=1e-6;
net.trainParam.lr= 0.5;
net.performFcn = 'mse';
   
[net, tr] = train(net, data_train, target_train);
capacity_estimated=(sim(net, data_test))';
% 
capacity_error = gsubtract(target_test,capacity_estimated);
RMSE = rms(capacity_error)*100
% MSE = mse(net,target_test,SOC_estimated)
MSE=(mean((capacity_estimated-target_test).^2))*100
MAPE=(sum(abs((capacity_estimated-target_test)./(capacity_estimated)))./length(target_test))*100
MAPE2=(sum(abs((target_test-capacity_estimated)./(capacity_estimated)))./length(target_test))*100
MAE=(sum(abs(capacity_estimated-target_test))./length(target_test))*100
SD=(std(capacity_estimated-target_test))*100
        check1(pp)=RMSE;
        check2(pp)=MSE;
        check3(pp)=MAPE;
        check4(pp)=MAE;
        check5(pp)= SD
        check6(pp,:)=capacity_estimated
        check7(pp,:)=capacity_error
end
[RMSE_BPNN,clm]=min(check1)
MSE_BPNN=min(check2(:,clm))
MAPE_BPNN=min(check3(:,clm))
MAE_BPNN=min(check4(:,clm))
SD_BPNN=min(check5(:,clm))
capacity_estimated_BPNN = check6(clm,:)
capacity_error_BPNN= check7(clm,:)       
toc

MATLAB: How to use weights and bias from Neural Network Toolbox for classification

I found the answer (output and yy are equal):

% Simulated a Neural Network
clear
close
clc
rng default
[x,y] = simpleclass_dataset;
neurons = 10;
net = feedforwardnet(neurons);
net = train(net,x,y);
outputs = sim(net,x);
% outputs = round(outputs);
figure, plotconfusion(y,outputs)
w1 = net.IW{1,1};
w2 = net.LW{2,1};
b1 = net.b{1};
b2 = net.b{2};
xx = x;
for ii = 1:length(net.inputs{1}.processFcns)
    xx = feval(net.inputs{1}.processFcns{ii},...
        'apply',xx,net.inputs{1}.processSettings{ii});
end
a1 = tanh(w1*xx + b1);
yy = purelin(w2*a1 + b2);
for mm = 1:length(net.outputs{1,2}.processFcns)
    yy = feval(net.outputs{1,2}.processFcns{mm},...
        'reverse',yy,net.outputs{1,2}.processSettings{mm});
end
figure, plotconfusion(y,yy)

Best Answer

Related Solutions

MATLAB: Prediction of remaining useful life

MATLAB: How to use weights and bias from Neural Network Toolbox for classification

Related Question