Fatigue_Life/LHS_Sampling.py

import numpy as np
from scipy.stats import qmc

# 1. Define the number of new samples you want to generate
num_samples = 26

# 2. Define the parameter space [min, max] for each variable
# [Large Diameter D, Small Diameter d, Fillet Radius r]
param_bounds = [
    [14.0, 16.0],  # Range for D
    [5.0, 7.5],   # Range for d
    [1.0, 3.5]    # Range for r
]

# 3. Create the Latin Hypercube Sampler
# The sampler generates points in a normalized space (between 0 and 1)
# We use a seed for reproducibility, so you always get the same "random" set
sampler = qmc.LatinHypercube(d=len(param_bounds), seed=42)
unit_hypercube_samples = sampler.random(n=num_samples)

# 4. Scale the normalized samples to your actual parameter ranges
scaled_samples = qmc.scale(unit_hypercube_samples, 
                           [b[0] for b in param_bounds], 
                           [b[1] for b in param_bounds])

# 5. Print the results in your desired format
print("--- Generated 26 New Shaft Combinations using LHS ---")
for i, sample in enumerate(scaled_samples):
    # Ensure that d is always less than D
    # If by chance LHS generates d > D, we can swap them or adjust.
    # A simple approach is to ensure d is at least a certain amount smaller than D.
    D_val = sample[0]
    d_val = sample[1]
    
    # Simple constraint: ensure d is at most 95% of D
    if d_val >= D_val:
        d_val = D_val * 0.95
        
    r_val = sample[2]
    
    # Format the output to one decimal place for clarity
    print(f"{i+1:2d}: life_D{D_val:.1f}_d{d_val:.1f}_r{r_val:.1f}")