import numpy as np
from copy import deepcopy
from foxes_opt.core import FarmVarsProblem, FarmOptProblem
from foxes.models.turbine_models import Calculator
from foxes.config import config
import foxes.variables as FV
import foxes.constants as FC
def _calc_func(valid, P, ct, algo, mdata, fdata, st_sel):
"""helper function for Calculator turbine model"""
return (valid, P * valid, ct * valid)
[docs]
class RegularLayoutOptProblem(FarmVarsProblem):
"""
Places turbines on a regular grid and optimizes
its parameters.
Attributes
----------
min_spacing: float
The minimal turbine spacing
initial_values: dict
Initial values for opt variables, key:
spacing_x, spacing_y, offset_x, offset_y, angle
:group: opt.problems.layout
"""
SPACING_X = "spacing_x"
SPACING_Y = "spacing_y"
OFFSET_X = "offset_x"
OFFSET_Y = "offset_y"
ANGLE = "angle"
[docs]
def __init__(
self,
name,
algo,
min_spacing,
initial_values=None,
**kwargs,
):
"""
Constructor.
Parameters
----------
name: str
The problem's name
algo: foxes.core.Algorithm
The algorithm
min_spacing: float
The minimal turbine spacing
initial_values: dict, optional
Initial values for opt variables, key:
spacing_x, spacing_y, offset_x, offset_y, angle
kwargs: dict, optional
Additional parameters for `FarmVarsProblem`
"""
super().__init__(name, algo, **kwargs)
self.min_spacing = min_spacing
self.initial_values = initial_values
[docs]
def initialize(self, verbosity=1, **kwargs):
"""
Initialize the object.
Parameters
----------
verbosity: int
The verbosity level, 0 = silent
kwargs: dict, optional
Additional parameters for super class init
"""
self._mname = self.name + "_calc"
for t in self.algo.farm.turbines:
if self._mname not in t.models:
t.add_model(self._mname)
self._turbine = deepcopy(self.farm.turbines[-1])
self.algo.mbook.turbine_models[self._mname] = Calculator(
in_vars=[FC.VALID, FV.P, FV.CT],
out_vars=[FC.VALID, FV.P, FV.CT],
func=_calc_func,
)
b = self.farm.boundary
assert b is not None, f"Problem '{self.name}': Missing wind farm boundary."
pmax = b.p_max()
pmin = b.p_min()
self._pmin = pmin
self._xy0 = pmin
self._halfspan = (pmax - pmin) / 2
self._halflen = np.linalg.norm(self._halfspan)
self.max_spacing = 2 * (self._halflen + self.min_spacing)
self._halfn = int(self._halflen / self.min_spacing)
if self._halfn * self.min_spacing < self._halflen:
self._halfn += 1
self._nrow = 2 * self._halfn + 1
self._nturb = self._nrow**2
if verbosity > 0:
print(f"Problem '{self.name}':")
print(f" xy0 = {self._xy0}")
print(f" span = {np.linalg.norm(self._halfspan * 2):.2f}")
print(f" min spacing = {self.min_spacing:.2f}")
print(f" max spacing = {self.max_spacing:.2f}")
print(f" n row turbns = {self._nrow}")
print(f" n turbines = {self._nturb}")
print(f" turbine mdls = {self._turbine.models}")
iniv = self.initial_values
self.initial_values = {
self.SPACING_X: self.min_spacing,
self.SPACING_Y: self.min_spacing,
self.OFFSET_X: 0.0,
self.OFFSET_Y: 0.0,
self.ANGLE: 0.0,
}
if iniv is not None:
mins = {
v: m for v, m in zip(self.var_names_float(), self.min_values_float())
}
maxs = {
v: m for v, m in zip(self.var_names_float(), self.max_values_float())
}
for k, v in iniv.items():
assert k in self.initial_values, (
f"Invalid initial value key: '{k}', expected one of {list(self.initial_values.keys())}."
)
assert v >= mins[k], (
f"Initial value for '{k}' too small: {v} < {mins[k]}."
)
assert v <= maxs[k], (
f"Initial value for '{k}' too large: {v} > {maxs[k]}."
)
self.initial_values[k] = v
if verbosity > 0:
print(" initial values:")
for k, v in self.initial_values.items():
print(f" {k:12s} = {v}")
super().initialize(
model_vars=[FV.X, FV.Y, FC.VALID],
verbosity=verbosity,
**kwargs,
)
if self.farm.n_turbines < self._nturb:
new_turbines = self.farm.turbines
for i in range(self._nturb - self.farm.n_turbines):
ti = len(new_turbines)
new_turbines.append(deepcopy(self._turbine))
new_turbines[-1].index = ti
new_turbines[-1].name = f"T{ti}"
self.farm.reset_turbines(self.algo, new_turbines)
elif self.farm.n_turbines > self._nturb:
new_turbines = self.farm.turbines[: self._nturb]
self.farm.reset_turbines(self.algo, new_turbines)
[docs]
def var_names_float(self):
"""
The names of float variables.
Returns
-------
names: list of str
The names of the float variables
"""
return [
self.SPACING_X,
self.SPACING_Y,
self.OFFSET_X,
self.OFFSET_Y,
self.ANGLE,
]
[docs]
def initial_values_float(self):
"""
The initial values of the float variables.
Returns
-------
values: numpy.ndarray
Initial float values, shape: (n_vars_float,)
"""
return list(self.initial_values.values())
[docs]
def min_values_float(self):
"""
The minimal values of the float variables.
Use -numpy.inf for unbounded.
Returns
-------
values: numpy.ndarray
Minimal float values, shape: (n_vars_float,)
"""
return np.array(
[
self.min_spacing,
self.min_spacing,
0.0,
0.0,
0.0,
],
dtype=config.dtype_double,
)
[docs]
def max_values_float(self):
"""
The maximal values of the float variables.
Use numpy.inf for unbounded.
Returns
-------
values: numpy.ndarray
Maximal float values, shape: (n_vars_float,)
"""
return np.array(
[
self.max_spacing,
self.max_spacing,
1.0,
1.0,
90.0,
],
dtype=config.dtype_double,
)
[docs]
def opt2farm_vars_individual(self, vars_int, vars_float):
"""
Translates optimization variables to farm variables
Parameters
----------
vars_int: numpy.ndarray
The integer optimization variable values,
shape: (n_vars_int,)
vars_float: numpy.ndarray
The float optimization variable values,
shape: (n_vars_float,)
Returns
-------
farm_vars: dict
The foxes farm variables. Key: var name,
value: numpy.ndarray with values, shape:
(n_states, n_sel_turbines)
"""
dx, dy, ox, oy, a = vars_float
n_states = self.algo.n_states
nx = self._nrow
ny = self._nrow
a = np.deg2rad(a)
nax = np.array([np.cos(a), np.sin(a), 0.0], dtype=config.dtype_double)
nay = np.cross(np.array([0.0, 0.0, 1.0], dtype=config.dtype_double), nax)
pts = np.zeros((nx, ny, 2), dtype=config.dtype_double)
pts[:] = (
self._xy0[None, None, :]
+ (ox + np.arange(nx)[:, None, None]) * dx * nax[None, None, :2]
+ (oy + np.arange(ny)[None, :, None]) * dy * nay[None, None, :2]
)
pts = pts.reshape(nx * ny, 2)
valid = self.farm.boundary.points_inside(pts)
farm_vars = {}
for v, d in zip([FV.X, FV.Y, FC.VALID], [pts[:, 0], pts[:, 1], valid]):
a = np.zeros((n_states, nx * ny), dtype=config.dtype_double)
a[:] = d[None, :]
farm_vars[v] = a
return farm_vars
[docs]
def opt2farm_vars_population(self, vars_int, vars_float, n_states):
"""
Translates optimization variables to farm variables
Parameters
----------
vars_int: numpy.ndarray
The integer optimization variable values,
shape: (n_pop, n_vars_int)
vars_float: numpy.ndarray
The float optimization variable values,
shape: (n_pop, n_vars_float)
n_states: int
The number of original (non-pop) states
Returns
-------
farm_vars: dict
The foxes farm variables. Key: var name,
value: numpy.ndarray with values, shape:
(n_pop, n_states, n_sel_turbines)
"""
n_pop = len(vars_float)
n_turbines = self.farm.n_turbines
dx = vars_float[:, 0]
dy = vars_float[:, 1]
ox = vars_float[:, 2]
oy = vars_float[:, 3]
nx = self._nrow
ny = self._nrow
a = vars_float[:, 4]
N = self._nturb
a = np.deg2rad(a)
nax = np.stack([np.cos(a), np.sin(a), np.zeros_like(a)], axis=-1)
naz = np.zeros_like(nax)
naz[..., 2] = 1
nay = np.cross(naz, nax)
pts = np.zeros((n_pop, nx, ny, 2), dtype=config.dtype_double)
pts[:] = (
self._xy0[None, None, None, :]
+ (ox[:, None, None, None] + np.arange(nx)[None, :, None, None])
* dx[:, None, None, None]
* nax[:, None, None, :2]
+ (oy[:, None, None, None] + np.arange(ny)[None, None, :, None])
* dy[:, None, None, None]
* nay[:, None, None, :2]
)
qts = np.zeros((n_pop, n_turbines, 2))
qts[:, :N] = pts.reshape(n_pop, N, 2)
del pts
valid = self.farm.boundary.points_inside(
qts.reshape(n_pop * n_turbines, 2)
).reshape(n_pop, n_turbines)
farm_vars = {}
for v, d in zip([FV.X, FV.Y, FC.VALID], [qts[:, :, 0], qts[:, :, 1], valid]):
a = np.zeros((n_pop, n_states, n_turbines), dtype=config.dtype_double)
a[:] = d[:, None, :]
farm_vars[v] = a
return farm_vars
[docs]
def finalize_individual(self, vars_int, vars_float, verbosity=1):
"""
Finalization, given the champion data.
Parameters
----------
vars_int: np.array
The optimal integer variable values, shape: (n_vars_int,)
vars_float: np.array
The optimal float variable values, shape: (n_vars_float,)
verbosity: int
The verbosity level, 0 = silent
Returns
-------
problem_results: Any
The results of the variable application
to the problem
objs: np.array
The objective function values, shape: (n_objectives,)
cons: np.array
The constraints values, shape: (n_constraints,)
"""
farm_vars = self.opt2farm_vars_individual(vars_int, vars_float)
sel = np.where(farm_vars[FC.VALID][0])[0]
x = farm_vars[FV.X][0, sel]
y = farm_vars[FV.Y][0, sel]
turbines = [t for i, t in enumerate(self.farm.turbines) if i in sel]
for i, t in enumerate(turbines):
t.xy = np.array([x[i], y[i]], dtype=config.dtype_double)
t.models = [m for m in t.models if m not in [self.name, self._mname]]
t.index = i
t.name = f"T{i}"
self.farm.reset_turbines(self.algo, turbines)
return FarmOptProblem.finalize_individual(
self, vars_int, vars_float, verbosity=1
)