How Harris Hawks Optimization Mimics Eagle Hunting to Solve Complex Problems

Harris Hawks and Optimization Algorithm

The Harris hawk (Parabuteo unicinctus) is a raptor that hunts cooperatively in families, using coordinated “surprise” attacks to capture prey such as rabbits.

Inspired by this behavior, Seyedali Mirjalili and Amir H. Gandomi proposed the Harris Hawks Optimization (HHO) algorithm in 2019 to solve complex optimization problems.

Basic Idea

The algorithm simulates the exploration and exploitation phases of a hawk swarm: during exploration the swarm searches the solution space, and during exploitation it adapts its hunting strategy based on the prey’s response to converge on optimal solutions.

Mathematical Model

Hawks fly over a large area, randomly changing direction and altitude to locate prey from above; once prey is spotted, they surround it and execute various attack strategies.

Exploration Behavior

Exploration is described by equations that involve a randomly selected individual, the current best solution, the population’s average position, random numbers, and the lower and upper bounds of the search space.

Prey Escape

When prey attempts to flee, hawks adjust their flight strategy based on the prey’s energy level, modeled by equations using escape energy, initial energy, current iteration, and maximum iterations.

Exploitation Behavior

After locating prey, hawks choose among hard surround, soft surround, hard surround with surprise, and soft surround with surprise strategies, reflecting flexible real‑world hunting tactics.

Depending on the prey’s escape energy, the algorithm switches between soft and hard surrounding, updating positions with specific formulas. Additional behaviors such as rapid dive attacks, Lévy‑flight‑based moves, and coordinated encirclement are also defined. If surrounding fails, a random walk is performed.

Case Study: Urban Traffic Signal Optimization

The HHO algorithm is applied to optimize traffic signal timings at an intersection to minimize vehicle waiting time. A target function relates vehicle counts on each approach to green‑light durations, subject to constraints on individual green times (10‑60 s) and total cycle length (≤120 s).

<code>Optimal green‑light configuration (seconds): [31.35168895 28.58712908 25.51190665 34.54725717]
Optimal objective value: 158.56000229652895</code>

Advantages of HHO include a good balance between exploration and exploitation, which helps avoid premature convergence, and a relatively simple mathematical model that is easy to implement and understand without extensive parameter tuning.

Limitations are that the algorithm may still get trapped in local optima on high‑dimensional multimodal problems, and its parameters (e.g., escape energy, jump ability) may require problem‑specific adjustment, increasing the difficulty of parameter selection.