-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathOCP_NumericApproach_1_QF.m
189 lines (151 loc) · 4.75 KB
/
OCP_NumericApproach_1_QF.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
%%
% NMPC -- Theory and Applications
% Course at Karlsruhe Institute of Technology
%
% Optimal loop Optimal control
%
%
% Autor: Hesham Hendy
% Email: [email protected]
% Date: 08-10-2019
% Status: final
clear;
clc;
addpath('C:\Program Files\MATLAB/casadi-windows-matlabR2016a-v3.5.0');
import casadi.*;
opti = casadi.Opti();
const = constants();
%% System limits
% Notice (for me) compare to direct_single_shooting.m codes, also in CasADi’s examples collection
t_0 = 0;
N = 70; % % number of control intervals % Time horizon %horzions (optimizied)
%% optimized Variables
% The control trajectory is parameterized using some piecewise smooth approximation, typically piecewise constant.
x = opti.variable(4, N + 1); % Optimization of 4 states (amount of parameters) on N control intervals + 1 predicted state
u = opti.variable(2, N); % Optimization of 2 inputs controls on N control intervals
tendopt = opti.variable(); % Optimization of the value of the time step ( the optimized final time)
timeSteps = tendopt / N;
%% objective function
S = 1000*0.1; % weight matrix to act on time
R = [0.1 0;
0 0.1]; % weight matrix to act on input torque
R_ = [0.1 0;
0 0.1];
Q = [0.001 0 0 0;
0 0.001 0 0;
0 0 0.001 0;
0 0 0 0.001]; % weight matrix to act on path
% Notice: Infeasible Problem Detected sometimes !
J = S * tendopt; % responsible for time optimality
t = t_0;
x_0 = const.x_0;
x_end = const.x_end;
for i = 1:N
% continuity constraint
x_next = explicitEuler(x(:, i), u(:, i), timeSteps, tendopt);
opti.subject_to(x(:,i+1) == x_next);
%t = [t, t(end) + timeSteps];
J = J +(x(:, i + 1))' * Q * (x(:, i + 1)) + u(:, i)' * R * u(:, i) + (u(:, i) - u(:, i-1))' * R_ * (u(:, i) - u(:, i-1)) ; %through that ,a compromise bet minimal time and minimal control energy and minimum path
end
% (x(:, i + 1) - x_end)' * Q * (x(:, i + 1) - x_end)
opti.minimize(J);
%% constraints formulation
opti.subject_to(x(:, 1) == x_0);
%opti.subject_to(x(:, end) == x_end);
opti.subject_to(const.omega_lb <= x(3, :) <= const.omega_ub);
opti.subject_to(const.omega_lb <= x(4, :) <= const.omega_ub);
opti.subject_to(const.u_lb / 1000 <= u(1, :) <= const.u_ub / 1000);
opti.subject_to(const.u_lb / 1000 <= u(2, :) <= const.u_ub / 1000);
opti.subject_to(0 < tendopt <= 3);
%% solve
opti.solver('ipopt'); % Ipopt, a library for large-scale nonlinear optimization.
sol = opti.solve();
% Remeber I became an error by running and after placing diffrent break
% points I discovered that the input is tiny so I decided to magnifiy (scale) it by
% solving and unscale it again !
% Run on command window
% opti.debug.value(x_next)
% opti.debug.value(x,opti.initial())
% opti.debug.show_infeasibilities()
x_sol = full(sol.value(x));
u_sol = full(sol.value(u)) .* [1000; 1000];
t_opt_sol = full(sol.value(tendopt));
timeSteps = t_opt_sol/ N; % length of a control interval
t = 0:timeSteps:t_opt_sol;
%% plot in q1-q2-plane
figure;
subplot(2, 2, 1);
hold on;
grid on;
plot(t, x_sol(1, :),'g', 'LineWidth', 2);
xlabel('sec');
ylabel('rad');
title('Angles on q1 plane');
subplot(2, 2, 2);
hold on;
grid on;
plot(t, x_sol(2, :),'g', 'LineWidth', 2);
xlabel('sec');
ylabel('rad');
title('Angles on q2 plane');
subplot(2, 2, 3);
hold on;
grid on;
plot(t, x_sol(3, :),'g', 'LineWidth', 2);
xlabel('sec');
ylabel('rad/s');
title('Angualr Velocity on q1 plane');
subplot(2, 2, 4);
hold on;
grid on;
plot(t, x_sol(4, :), 'g','LineWidth', 2);
xlabel('sec');
ylabel('rad/s');
title('Angualr Velocity on q2 plane');
%% plot in x-y-plane
figure;
x_ = const.l1 .* cos(x_sol(1, :)) + const.l2 .* cos(x_sol(1, :) + x_sol(2, :));
y_ = const.l1 .* sin(x_sol(1, :)) + const.l2 .* sin(x_sol(1, :) + x_sol(2, :));
subplot(2, 1, 1);
hold on;
grid on;
plot(t, x_, 'LineWidth', 2);
xlabel('sec');
ylabel('m/sec');
title('Motion on x-plane');
subplot(2, 1, 2);
hold on;
grid on;
plot(t, y_, 'LineWidth', 2);
xlabel('sec');
ylabel('m/sec');
title('Motion on y-plane');
%% plot inputs
figure;
fig3 = gcf;
fig3.PaperUnits='inches';
fig3.PaperPosition = [0 0 8 8];
subplot(2, 1, 1);
axis([0, 3, -1500, 1500]);
hold on;
grid on;
stairs(t(2:length(t)), u_sol(1, :), 'LineWidth', 2);
xlabel('sec');
ylabel('N.m');
title('Applied Torque by U1');
subplot(2, 1, 2);
axis([0, 3, -1500, 1500]);
hold on;
grid on;
stairs(t(2:length(t)), u_sol(2, :), 'LineWidth', 2);
xlabel('sec');
ylabel('N.m');
title('Applied Torque by U2');
%% ========================================================================
% subfunctions
% =========================================================================
function x_next = explicitEuler(x, u, h ,t_opt)
%Solving an ODR using one-Step Method
% Computation is performed without iteration
x_next = x + h .* stateSpace_ODE(t_opt, x, u); % revise Buchter Diagram
end