""" optimize_yagi.py — Example: Optimize a 2m Yagi with GA and PSO. Optimizes 3 director half-lengths and spacings for: - SWR < 1.5 at 144.2 MHz - Gain > 10 dBi - F/B ratio > 20 dB Runs both GA (50 gen, 30 pop) and PSO (100 iter, 20 particles). Compares results and writes winner to yagi_optimized.nec. Run: python optimize_yagi.py """ import logging import sys from pathlib import Path logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s: %(message)s") sys.path.insert(0, str(Path(__file__).parent.parent / "scripts")) from nec_generator import Yagi, NECModel from nec_runner import NECRunner from genetic_optimizer import GeneParameter, FitnessFunction, GeneticOptimizer from pso_optimizer import PSOOptimizer, PSOConfig # ─── Parameters ─────────────────────────────────────────────────────────────── # Optimize 3 directors: half-lengths (m) and their boom spacings (m) # Also optimize driven element half-length and reflector half-length TARGET_FREQ_MHZ = 144.2 C_LIGHT = 299.792458e6 WL = C_LIGHT / (TARGET_FREQ_MHZ * 1e6) # ~2.08m params = [ GeneParameter("refl_hl", 0.46 * WL/2, 0.52 * WL/2), # Reflector half-length GeneParameter("driv_hl", 0.44 * WL/2, 0.50 * WL/2), # Driven half-length GeneParameter("d1_hl", 0.42 * WL/2, 0.47 * WL/2), # Director 1 half-length GeneParameter("d2_hl", 0.40 * WL/2, 0.46 * WL/2), GeneParameter("d3_hl", 0.38 * WL/2, 0.44 * WL/2), GeneParameter("refl_sp", 0.15 * WL, 0.25 * WL), # Reflector spacing (behind driven) GeneParameter("d1_sp", 0.15 * WL, 0.30 * WL), # Director 1 spacing GeneParameter("d2_sp", 0.20 * WL, 0.35 * WL), GeneParameter("d3_sp", 0.20 * WL, 0.40 * WL), ] objectives = { "swr_target": 1.5, "swr_weight": 3.0, "gain_min_dBi": 10.0, "gain_weight": 2.0, "fb_min_dB": 20.0, "fb_weight": 1.5, } # ─── Model Builder ──────────────────────────────────────────────────────────── def build_custom_yagi(genes: dict, freq_mhz: float = TARGET_FREQ_MHZ) -> NECModel: """Build a Yagi from optimized gene dict.""" import math from nec_generator import Wire, Excitation, FrequencyCard, RadiationPattern, GroundPlane m = NECModel(title=f"Optimized Yagi {freq_mhz} MHz", freq_mhz=freq_mhz) h = 6.0 # Height above ground (m) r = 0.004 # Wire radius (m) # Element X positions (boom axis) x_refl = -genes["refl_sp"] x_driv = 0.0 x_d1 = genes["d1_sp"] x_d2 = genes["d1_sp"] + genes["d2_sp"] x_d3 = genes["d1_sp"] + genes["d2_sp"] + genes["d3_sp"] elements = [ (x_refl, genes["refl_hl"], False, "Reflector"), (x_driv, genes["driv_hl"], True, "Driven"), (x_d1, genes["d1_hl"], False, "Director_1"), (x_d2, genes["d2_hl"], False, "Director_2"), (x_d3, genes["d3_hl"], False, "Director_3"), ] driven_tag = None for tag_base, (x, hl, is_driven, name) in enumerate(elements, start=1): segs = max(9, int(2 * hl / WL * 10)) if segs % 2 == 0: segs += 1 sh = segs // 2 + 1 t1 = tag_base * 2 - 1 t2 = tag_base * 2 m.add_wire(x, -hl, h, x, 0, h, radius=r, segs=sh, tag=t1) m.add_wire(x, 0, h, x, hl, h, radius=r, segs=sh, tag=t2) if is_driven: driven_tag = t1 if driven_tag: wire = next(w for w in m.wires if w.tag == driven_tag) m.excitations.append(Excitation(tag=driven_tag, segment=wire.segs)) m.add_frequency(freq_mhz) m.set_ground(type=0) # Free space for optimization speed m.add_pattern(theta_inc=10, phi_inc=10) # Coarser pattern for speed during opt return m # ─── Run Optimization ───────────────────────────────────────────────────────── def run_ga(runner): print("\n--- Genetic Algorithm (50 gen, 30 pop) ---") fitness_fn = FitnessFunction(objectives) opt = GeneticOptimizer( params=params, fitness_fn=fitness_fn, model_builder=build_custom_yagi, runner=runner, pop_size=30, n_generations=50, mutation_rate=0.12, crossover_rate=0.8, elitism=2, early_stop_gens=15, ) result = opt.run() print(f"GA best fitness: {result.best_fitness:.3f}") print(f"GA best genes: {result.best_genes}") result.to_csv("ga_convergence.csv") return result def run_pso(runner): print("\n--- Particle Swarm (100 iter, 20 particles) ---") fitness_fn = FitnessFunction(objectives) cfg = PSOConfig(n_particles=20, n_iterations=100, early_stop_iters=20) opt = PSOOptimizer( params=params, fitness_fn=fitness_fn, model_builder=build_custom_yagi, runner=runner, config=cfg, ) result = opt.run() print(f"PSO best fitness: {result.best_fitness:.3f}") print(f"PSO best genes: {result.best_genes}") result.to_csv("pso_convergence.csv") return result # ─── Main ───────────────────────────────────────────────────────────────────── if __name__ == "__main__": print(f"Yagi Optimizer — {TARGET_FREQ_MHZ} MHz, λ={WL*1000:.0f}mm") print(f"Objectives: SWR<{objectives['swr_target']}, " f"Gain>{objectives['gain_min_dBi']} dBi, " f"F/B>{objectives['fb_min_dB']} dB\n") runner = NECRunner() ga_result = run_ga(runner) pso_result = run_pso(runner) # Pick winner if ga_result.best_fitness >= pso_result.best_fitness: winner, winner_name = ga_result, "GA" else: winner, winner_name = pso_result, "PSO" print(f"\n=== WINNER: {winner_name} (fitness={winner.best_fitness:.3f}) ===") print("Best genes:") for k, v in winner.best_genes.items(): print(f" {k:12s}: {v*1000:.1f} mm") # Write final optimized NEC file final_model = build_custom_yagi(winner.best_genes, TARGET_FREQ_MHZ) # Switch to fine-grained pattern for final output final_model.patterns.clear() final_model.add_pattern(theta_inc=5, phi_inc=5) final_model.write("yagi_optimized.nec") print("\nOptimized design written: yagi_optimized.nec") print("Verify with: nec2c yagi_optimized.nec yagi_optimized.out")