def test_RotatingMasses(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
mhe = template_mhe(model)
"""
Set initial state
"""
np.random.seed(99)
# Use different initial state for the true system (simulator) and for MHE / MPC
x0_true = np.random.rand(model.n_x)-0.5
x0 = np.zeros(model.n_x)
mpc.x0 = x0
simulator.x0 = x0_true
mhe.x0 = x0
mhe.p_est0 = 1e-4
# Set initial guess for MHE/MPC based on initial state.
mpc.set_initial_guess()
mhe.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = mhe.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, mhe], 'results_rotatingMasses', overwrite=True)
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_rotatingMasses.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i], ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i], ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(mhe.data.__dict__[test_i], ref['estimator'].__dict__[test_i])
self.assertTrue(check)
def test_oscillating_masses_discrete(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
np.random.seed(99)
x0 = np.random.rand(model.n_x) - 0.5
mpc.x0 = x0
simulator.x0 = x0
estimator.x0 = x0
# Use initial state to set the initial guess.
mpc.set_initial_guess()
# This is only meaningful for DAE systems.
simulator.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, estimator], 'results_oscillatingMasses_dae', overwrite=True)
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results(
'./results/results_oscillatingMasses_dae.pkl')
test = ['_x', '_u', '_aux', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
def test_CSTR(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
# Set the initial state of mpc and simulator:
C_a_0 = 0.8 # This is the initial concentration inside the tank [mol/l]
C_b_0 = 0.5 # This is the controlled variable [mol/l]
T_R_0 = 134.14 #[C]
T_K_0 = 130.0 #[C]
x0 = np.array([C_a_0, C_b_0, T_R_0, T_K_0]).reshape(-1, 1)
mpc.x0 = x0
simulator.x0 = x0
mpc.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, estimator], 'results_CSTR')
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_CSTR.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
def test_batch_reactor(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
X_s_0 = 1.0 # This is the initial concentration inside the tank [mol/l]
S_s_0 = 0.5 # This is the controlled variable [mol/l]
P_s_0 = 0.0 #[C]
V_s_0 = 120.0 #[C]
x0 = np.array([X_s_0, S_s_0, P_s_0, V_s_0])
mpc.x0 = x0
simulator.x0 = x0
estimator.x0 = x0
mpc.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_batch_rector.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
"""
import matplotlib.gridspec as gridspec import time from template_model import template_model from template_mpc import template_mpc from template_simulator import template_simulator """ User settings: """ show_animation = False store_results = False """ Get configured do-mpc modules: """ model = template_model() mpc = template_mpc(model) simulator = template_simulator(model) estimator = do_mpc.estimator.StateFeedback(model) """ Set initial state """ X_s_0 = 0.0 # This is the initial concentration inside the tank [mol/l] Y_s_0 = 0.0 # This is the controlled variable [mol/l] #V_s_0 = 0.0 #[C] T_s_0 = 0.0 #[C] x0 = np.array([X_s_0, Y_s_0, T_s_0]) mpc.x0 = x0 simulator.x0 = x0 estimator.x0 = x0
def test_industrialpoly(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
# Set the initial state of mpc and simulator:
x0 = model._x(0)
delH_R_real = 950.0
c_pR = 5.0
x0['m_W'] = 10000.0
x0['m_A'] = 853.0
x0['m_P'] = 26.5
x0['T_R'] = 90.0 + 273.15
x0['T_S'] = 90.0 + 273.15
x0['Tout_M'] = 90.0 + 273.15
x0['T_EK'] = 35.0 + 273.15
x0['Tout_AWT'] = 35.0 + 273.15
x0['accum_monom'] = 300.0
x0['T_adiab'] = x0['m_A'] * delH_R_real / (
(x0['m_W'] + x0['m_A'] + x0['m_P']) * c_pR) + x0['T_R']
mpc.set_initial_state(x0, reset_history=True)
simulator.set_initial_state(x0, reset_history=True)
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, estimator], 'results_industrial_poly')
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_industrial_poly.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
def test_industrialpoly(self):
"""
Get configured do-mpc modules:
"""
model = template_model()
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
delH_R_real = 950.0
c_pR = 5.0
# x0 is a property of the simulator - we obtain it and set values.
x0 = simulator.x0
x0['m_W'] = 10000.0
x0['m_A'] = 853.0
x0['m_P'] = 26.5
x0['T_R'] = 90.0 + 273.15
x0['T_S'] = 90.0 + 273.15
x0['Tout_M'] = 90.0 + 273.15
x0['T_EK'] = 35.0 + 273.15
x0['Tout_AWT'] = 35.0 + 273.15
x0['accum_monom'] = 300.0
x0['T_adiab'] = x0['m_A'] * delH_R_real / (
(x0['m_W'] + x0['m_A'] + x0['m_P']) * c_pR) + x0['T_R']
# Finally, the controller gets the same initial state.
mpc.x0 = x0
# Which is used to set the initial guess:
mpc.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, estimator], 'results_industrial_poly', overwrite=True)
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_industrial_poly.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
def test_dip(self):
"""
Get configured do-mpc modules:
"""
# Define obstacles to avoid (cicles)
obstacles = [
{
'x': 0.,
'y': 0.6,
'r': 0.3
},
]
model = template_model(obstacles)
mpc = template_mpc(model)
simulator = template_simulator(model)
estimator = do_mpc.estimator.StateFeedback(model)
"""
Set initial state
"""
simulator.x0['theta'] = np.pi
simulator.x0['pos'] = 0
x0 = simulator.x0.cat.full()
mpc.x0 = x0
estimator.x0 = x0
mpc.set_initial_guess()
"""
Run some steps:
"""
for k in range(5):
u0 = mpc.make_step(x0)
y_next = simulator.make_step(u0)
x0 = estimator.make_step(y_next)
"""
Store results (from reference run):
"""
#do_mpc.data.save_results([mpc, simulator, estimator], 'results_dip', overwrite=True)
"""
Compare results to reference run:
"""
ref = do_mpc.data.load_results('./results/results_dip.pkl')
test = ['_x', '_u', '_time', '_z']
for test_i in test:
# Check MPC
check = np.allclose(mpc.data.__dict__[test_i],
ref['mpc'].__dict__[test_i])
self.assertTrue(check)
# Check Simulator
check = np.allclose(simulator.data.__dict__[test_i],
ref['simulator'].__dict__[test_i])
self.assertTrue(check)
# Estimator
check = np.allclose(estimator.data.__dict__[test_i],
ref['estimator'].__dict__[test_i])
self.assertTrue(check)
