# RCAIDE/Library/Methods/Aerodynamics/Vortex_Lattice_Method/train_AVL_surrogates.py
#
# ----------------------------------------------------------------------------------------------------------------------
# IMPORT
# ----------------------------------------------------------------------------------------------------------------------
# RCAIDE imports
import RCAIDE
from RCAIDE.Framework.Mission.Common import Results
from RCAIDE.Library.Methods.Aerodynamics.Athena_Vortex_Lattice.run_AVL_analysis import run_AVL_analysis
# Package imports
import os
import numpy as np
from shutil import rmtree
# ----------------------------------------------------------------------------------------------------------------------
# train_AVL_surrogates
# ----------------------------------------------------------------------------------------------------------------------
[docs]
def train_AVL_surrogates(aerodynamics):
"""Call methods to run VLM for sample point evaluation.
Assumptions:
None
Source:
None
Args:
aerodynamics : VLM analysis [unitless]
Returns:
None
"""
run_folder = os.path.abspath(aerodynamics.settings.filenames.run_folder)
vehicle = aerodynamics.vehicle
training = aerodynamics.training
AoA = training.angle_of_attack
Mach = training.Mach
side_slip_angle = aerodynamics.settings.side_slip_angle
roll_rate_coefficient = aerodynamics.settings.roll_rate_coefficient
pitch_rate_coefficient = aerodynamics.settings.pitch_rate_coefficient
lift_coefficient = aerodynamics.settings.lift_coefficient
atmosphere = RCAIDE.Framework.Analyses.Atmospheric.US_Standard_1976()
atmo_data = atmosphere.compute_values(altitude = 0.0)
len_AoA = len(AoA)
len_Mach = len(Mach)
CM = np.zeros((len_AoA,len_Mach))
CL = np.zeros_like(CM)
CD = np.zeros_like(CM)
e = np.zeros_like(CM)
Cm_alpha = np.zeros_like(CM)
Cn_beta = np.zeros_like(CM)
NP = np.zeros_like(CM)
# remove old files in run directory
if os.path.exists(aerodynamics.settings.filenames.run_folder):
if aerodynamics.settings.new_regression_results:
rmtree(run_folder)
for i,_ in enumerate(Mach):
# Set training conditions
run_conditions = Results()
run_conditions.expand_rows(len_AoA)
run_conditions.aerodynamics.angles.alpha = np.array([AoA]).T
run_conditions.freestream.density = np.ones_like(run_conditions.aerodynamics.angles.alpha)*atmo_data.density
run_conditions.freestream.gravity = np.ones_like(run_conditions.aerodynamics.angles.alpha)*9.81
run_conditions.freestream.speed_of_sound = np.ones_like(run_conditions.aerodynamics.angles.alpha)*atmo_data.speed_of_sound[0,0]
run_conditions.freestream.velocity = np.ones_like(run_conditions.aerodynamics.angles.alpha)*Mach[i] * run_conditions.freestream.speed_of_sound
run_conditions.freestream.mach_number = np.ones_like(run_conditions.aerodynamics.angles.alpha)*Mach[i]
run_conditions.aerodynamics.angles.beta = np.ones_like(run_conditions.aerodynamics.angles.alpha)*side_slip_angle
run_conditions.static_stability.coefficients.roll = np.ones_like(run_conditions.aerodynamics.angles.alpha)*roll_rate_coefficient
if lift_coefficient == None:
run_conditions.aerodynamics.coefficients.lift.total= lift_coefficient
else:
run_conditions.aerodynamics.coefficients.lift.total= np.array([lift_coefficient]).T
run_conditions.static_stability.coefficients.pitch = np.ones_like(run_conditions.aerodynamics.angles.alpha)*pitch_rate_coefficient
# Run Analysis at AoA[i] and Mach[i]
run_AVL_analysis(aerodynamics,run_conditions)
CL[:,i] = run_conditions.aerodynamics.coefficients.lift.total[:,0]
CD[:,i] = run_conditions.aerodynamics.coefficients.drag.induced.total[:,0]
e [:,i] = run_conditions.aerodynamics.coefficients.drag.induced.efficiency_factor[:,0]
CM[:,i] = run_conditions.static_stability.coefficients.pitch[:,0]
Cm_alpha[:,i] = run_conditions.static_stability.derivatives.CM_alpha[:,0]
Cn_beta[:,i] = run_conditions.static_stability.derivatives.CN_beta[:,0]
NP[:,i] = run_conditions.static_stability.neutral_point[:,0]
if aerodynamics.training_file:
# load data
data_array = np.loadtxt(aerodynamics.training_file)
# convert from 1D to 2D
CL_1D = np.atleast_2d(data_array[:,0])
CD_1D = np.atleast_2d(data_array[:,1])
e_1D = np.atleast_2d(data_array[:,2])
CM_1D = np.atleast_2d(data_array[:,3])
Cm_alpha_1D = np.atleast_2d(data_array[:,4])
Cn_beta_1D = np.atleast_2d(data_array[:,5])
NP_1D = np.atleast_2d(data_array[:,6])
# convert from 1D to 2D
CL = np.reshape(CL_1D, (len_AoA,-1))
CD = np.reshape(CD_1D, (len_AoA,-1))
e = np.reshape(e_1D , (len_AoA,-1))
CM = np.reshape(CM_1D, (len_AoA,-1))
Cm_alpha = np.reshape(Cm_alpha_1D, (len_AoA,-1))
Cn_beta = np.reshape(Cn_beta_1D , (len_AoA,-1))
NP = np.reshape(NP_1D , (len_AoA,-1))
# Save the data for regression
if aerodynamics.settings.new_regression_results:
# convert from 2D to 1D
CL_1D = CL.reshape([len_AoA*len_Mach,1])
CD_1D = CD.reshape([len_AoA*len_Mach,1])
e_1D = e.reshape([len_AoA*len_Mach,1])
CM_1D = CM.reshape([len_AoA*len_Mach,1])
Cm_alpha_1D = Cm_alpha.reshape([len_AoA*len_Mach,1])
Cn_beta_1D = Cn_beta.reshape([len_AoA*len_Mach,1])
NP_1D = Cn_beta.reshape([len_AoA*len_Mach,1])
np.savetxt(vehicle.tag+'_stability_data.txt',np.hstack([CL_1D,CD_1D,e_1D,CM_1D,Cm_alpha_1D, Cn_beta_1D,NP_1D ]),fmt='%10.8f',header=' CM Cm_alpha Cn_beta NP ')
# Store training data
# Save the data for regression
training_data = np.zeros((7,len_AoA,len_Mach))
training_data[0,:,:] = CL
training_data[1,:,:] = CD
training_data[2,:,:] = e
training_data[3,:,:] = CM
training_data[4,:,:] = Cm_alpha
training_data[5,:,:] = Cn_beta
training_data[6,:,:] = NP
# Store training data
training.coefficients = training_data
