def parameters(self, system='Pendulum-v0'):
self.system = system
self.env = gym.make(self.system)
self.env.reset() # to reset the environment state
self.x0 = self.env.state
self.nx = self.x0.shape[0]
self.action_sample = self.env.action_space.sample()
self.nu = np.asarray([self.action_sample]).shape[0]
# default simulation setup
self.U = np.zeros([(self.nsim - 1), self.nu])
self.U[int(self.nsim / 8)] = 0.1
# randomIndList =[]
# for i in range(0, 100):
# # any random numbers from 0 to 1000
# ind = random.randint(0, self.nsim-1)
# self.U[ind] = random.uniform(-1, 1)
self.U = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 40)),
xmax=0.5,
xmin=-0.5)
# self.U = SplineSignal(nsim=self.nsim, values=None, xmin=-2.0, xmax=2.0)
# print(self.U.shape)
if type(self.action_sample) == int:
self.U = self.U.astype(int)
def parameters(self):
self.N = 10000 # population
# initial number of infected and recovered individuals
self.e_0 = 1 / self.N
self.i_0 = 0.00
self.r_0 = 0.00
self.s_0 = 1 - self.e_0 - self.i_0 - self.r_0
self.x0 = np.asarray([self.s_0, self.e_0, self.i_0, self.r_0])
self.nx = 4
self.nu = 1
self.t_incubation = 5.1
self.t_infective = 3.3
self.R0 = 2.4
self.alpha = 1 / self.t_incubation
self.gamma = 1 / self.t_infective
self.beta = self.R0 * self.gamma
# default simulation setup
self.U = Steps(nx=self.nu,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 100)),
xmax=1,
xmin=0)
def parameters(self, system='Pendulum-v0'):
self.system = system
self.env = gym.make(self.system)
self.env.reset() # to reset the environment state
self.x0 = self.env.state
self.nx = self.x0.shape[0]
self.action_sample = self.env.action_space.sample()
self.nu = np.asarray([self.action_sample]).shape[0]
# default simulation setup
self.U = np.zeros([(self.nsim - 1), self.nu])
self.U[int(self.nsim / 8)] = 0.1
self.U = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 40)),
xmax=0.5,
xmin=-0.5)
if type(self.action_sample) == int:
self.U = self.U.astype(int)
def parameters(self):
self.rho = 1000.0 # water density (kg/m^3)
self.A = 1.0 # tank area (m^2)
# Initial Conditions for the States
self.x0 = 0
# default imputs
c = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 400)),
xmax=55,
xmin=45)
valve = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 400)),
xmax=100,
xmin=0)
self.U = np.vstack([c.T, valve.T]).T
self.nu = 2
self.nx = 1
def parameters(self):
super().parameters()
self.ts = 1
self.c1 = 0.08 # inlet valve coefficient
self.c2 = 0.04 # tank outlet coefficient
# Initial Conditions for the States
self.x0 = np.asarray([0, 0])
# default simulation setup
pump = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 400)),
xmax=0.5,
xmin=0)
valve = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 400)),
xmax=0.4,
xmin=0)
self.U = np.vstack([pump.T, valve.T]).T
self.nu = 2
self.nx = 2
def parameters(self):
# Volumetric Flowrate (m^3/sec)
self.q = 100
# Volume of CSTR (m^3)
self.V = 100
# Density of A-B Mixture (kg/m^3)
self.rho = 1000
# Heat capacity of A-B Mixture (J/kg-K)
self.Cp = 0.239
# Heat of reaction for A->B (J/mol)
self.mdelH = 5e4
# E - Activation energy in the Arrhenius Equation (J/mol)
# R - Universal Gas Constant = 8.31451 J/mol-K
self.EoverR = 8750
# Pre-exponential factor (1/sec)
self.k0 = 7.2e10
# U - Overall Heat Transfer Coefficient (W/m^2-K)
# A - Area - this value is specific for the U calculation (m^2)
self.UA = 5e4
# Steady State Initial Conditions for the States
self.Ca_ss = 0.87725294608097
self.T_ss = 324.475443431599
self.x0 = np.empty(2)
self.x0[0] = self.Ca_ss
self.x0[1] = self.T_ss
# Steady State Initial Condition for the Uncontrolled Inputs
self.u_ss = 300.0 # cooling jacket Temperature (K)
self.Tf = 350 # Feed Temperature (K)
self.Caf = 1 # Feed Concentration (mol/m^3)
# dimensions
self.nx = 2
self.nu = 1
self.nd = 2
# default simulation setup
self.U = self.u_ss + Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 100)),
xmax=6,
xmin=-6)
class GymWrapper(EmulatorBase):
"""
wrapper for OpenAI gym environments
https://gym.openai.com/read-only.html
https://github.com/openai/gym
# TODO: include visualization option for the trajectories or render of OpenAI gym
"""
def __init__(self, nsim=1000, ninit=0, system='Pendulum-v0', seed=59):
super().__init__(nsim=nsim, ninit=ninit)
self.system = system
def parameters(self, system='Pendulum-v0'):
self.system = system
self.env = gym.make(self.system)
self.env.reset() # to reset the environment state
self.x0 = self.env.state
self.nx = self.x0.shape[0]
self.action_sample = self.env.action_space.sample()
self.nu = np.asarray([self.action_sample]).shape[0]
# default simulation setup
self.U = np.zeros([(self.nsim - 1), self.nu])
self.U[int(self.nsim / 8)] = 0.1
self.U = Steps(nx=1,
nsim=self.nsim,
values=None,
randsteps=int(np.ceil(self.nsim / 40)),
xmax=0.5,
xmin=-0.5)
if type(self.action_sample) == int:
self.U = self.U.astype(int)
def equations(self, x, u):
if type(self.action_sample) == int:
u = u.item()
self.env.state = x
x, reward, done, info = self.env.step(u)
return x, reward
def simulate(self, nsim=None, U=None, x0=None, **kwargs):
"""
:param nsim: (int) Number of steps for open loop response
:param U: (ndarray, shape=(self.nu)) control actions
:param x0: (ndarray, shape=(self.nx)) Initial state. If not give will use internal state.
:return: The response trajectories, X
"""
if nsim is None:
nsim = self.nsim
if U is None:
U = self.U
if x0 is None:
x = self.x0
else:
assert x0.shape[0] == self.nx, "Mismatch in x0 size"
x = x0
X, Reward = [], []
N = 0
for u in U:
x, reward = self.equations(x, u)
X.append(x) # updated states trajectories
Reward.append(reward) # updated states trajectories
N += 1
if N == nsim:
break
Xout = np.asarray(X)
Yout = np.asarray(Reward).reshape(-1, 1)
Uout = np.asarray(U)
return {'X': Xout, 'Y': Yout, 'U': Uout}