Horse Herd Optimization-Tuned PI Controller for STATCOM-Based Voltage Regulation

Horse Herd Optimization-Tuned PI Controller for STATCOM-Based Voltage Regulation

In modern power systems, maintaining stable voltage profiles under varying load and generation conditions is crucial. One effective approach is using a Static Synchronous Compensator (STATCOM) regulated by finely tuned controllers. This blog explores how Horse Herd Optimization (HHO) is used to fine-tune a PI (Proportional-Integral) controller within a STATCOM for efficient voltage regulation.

Objective Function and Error Definition

The optimization process begins with defining a cost function, also known as the objective function, which aims to minimize multiple types of errors within the STATCOM control loop. These include:

• E1: Error in AC voltage controller

• E2: Error in DC voltage controller

• E3: Error in d-axis current controller

• E4: Error in q-axis current controller

These errors are extracted from the simulation model and form the basis of the optimization routine.

Parameters to Optimize

The PI controller for each section (AC voltage, DC voltage, and current control) contains two tuning parameters: KpK_pKp​ and KiK_iKi​. Thus, a total of six parameters need to be optimized using Horse Herd Optimization. The algorithm is initialized with:

• Lower and upper bounds for each parameter

• Velocity limits for horses

• Initial random positions and velocities of the horses

• Objective function evaluation for each horse’s position

Horse Herd Optimization Mechanism

The HHO algorithm mimics social hierarchy in a horse herd. Horses are categorized based on their cost coefficient (CC) into:

• Alpha horses: Best performing (CC ≤ 0.1)

• Beta horses: Good performance (CC ≤ 0.3)

• Gamma horses: Average performance (CC ≤ 0.6)

• Delta horses: Poor performance (CC > 0.6)

Each category updates its position and velocity according to its role. The global best position is tracked and updated iteratively. Position limits and boundary conditions are enforced at every iteration. The algorithm continues for a specified number of iterations (e.g., 10), after which the best parameter set is finalized.

Overview of the STATCOM Model

The system under study includes:

• A programmable voltage source producing 15 kV

• Applied voltage sags over different time intervals

• A 21 km feeder line

• Multiple loads: 3 MW, 2 MW, and 1 MW with variable load profiles

• A step-down transformer (15 kV to 600 V) connecting the STATCOM

• LC filters to mitigate harmonics

The STATCOM is connected in parallel using a dual-voltage-source-converter (VSC) topology with a DC-link capacitor in between.

Control Architecture

The control structure comprises:

1. AC Voltage Controller

   • Measures terminal voltage

   • Compares it with a 1 pu reference

   • Error is processed through a tuned PI controller to generate reactive current reference (Iq_ref)

2. DC Voltage Controller

   • Measures DC-link voltage

   • Compares it with a 2400 V reference

   • Error processed to generate active current reference (Id_ref)

3. Current Controller

   • Compares actual Id and Iq with their references

   • Processes error through a PI controller

   • Generates modulation signals (Vd, Vq)

These signals are converted into three-phase voltages (Vabc) to control the STATCOM's VSC.

Simulation and Results

Once the optimization is complete:

• The tuned PI parameters for each controller are implemented in the STATCOM model

• The model is simulated with voltage sags and load variations

• The result shows effective voltage regulation, maintaining the terminal voltage close to 1 pu

• The DC-link capacitor voltage remains constant

• The convergence graph indicates rapid improvement of the objective function within a few iterations

Conclusion

Using Horse Herd Optimization for tuning PI controllers in STATCOM systems leads to effective voltage regulation under dynamic load conditions. The multi-loop control strategy, enhanced with intelligent optimization, ensures better reactive power compensation and system stability. This approach can be extended to other FACTS devices and grid-support scenarios where precision tuning is critical.