The Control Analysis of Single Capacity Water Tank Liquid Level System Based on PSO Optimized Fuzzy PID

Zheng‐Hua Tan, Rong-Guei Tsai, Yan-Zuo Chang, Ruei-Yuan Wang, Jen-Wei Weng, Ho-Sheng ChenInternational Journal of Advanced Engineering Management and Science 2025 · DOI

TL;DR

This paper evaluates three distinct strategies for controlling the liquid level in industrial water tanks: classical PID, Fuzzy PID, and Particle Swarm Optimization (PSO)-tuned PID. Through MATLAB/Simulink simulations, the authors demonstrate a clear trade-off: Fuzzy PID offers the fastest response but suffers from high overshoot, while PSO-tuned PID provides unmatched stability and accuracy at the cost of speed.

What Problem Does This Paper Address?

In industrial manufacturing—from chemical processing to food production—maintaining a precise liquid level in a tank is a fundamental requirement. While the Proportional-Integral-Derivative (PID) controller has been the industry workhorse for decades due to its simplicity, it often struggles with nonlinear or time-varying systems where "set-and-forget" parameters don't cut it.

The authors address the challenge of parameter tuning. Manually finding the "Goldilocks" zone for PID gains ($K_p, K_i, K_d$) is time-consuming and often results in sub-optimal performance. By comparing traditional methods against adaptive (Fuzzy) and evolutionary (PSO) algorithms, the paper seeks to provide a roadmap for engineers to choose the right control strategy based on their specific operational priorities.

Methods

The researchers established a mathematical model of a single-capacity water tank to derive its transfer function. They then implemented three control architectures in MATLAB/Simulink:

  1. Classical PID: The baseline model where parameters are fixed. The authors explored how varying $K_p, K_i,$ and $K_d$ individually affected the system's rise time and overshoot.
  2. Fuzzy PID: This controller uses fuzzy logic rules to adjust PID parameters in real-time based on the current error and the rate of change of that error. It’s designed to be "smarter" and more adaptive than a static PID.
  3. PSO-Optimized PID: Particle Swarm Optimization is a computational method that optimizes a problem by iteratively trying to improve a candidate solution. Here, "particles" (potential PID settings) move through the search space to find the global optimum for stability and accuracy, typically minimizing error metrics like ITAE (Integral of Time-weighted Absolute Error).

Datasets

As this is a control theory and simulation-based study, the researchers did not use external datasets. Instead, they generated synthetic data through MATLAB/Simulink simulations. The physical parameters of the water tank (inlet/outlet flow rates and liquid height) served as the environment, and the performance was evaluated against a target liquid level (setpoint) of 200 units.

Key Insights and Ideas

  • The Speed-Stability Paradox: The most striking insight is the inverse relationship between response speed and stability across the three methods. There is no "perfect" controller; there is only the "right" controller for a specific goal.
  • Fuzzy Logic is Aggressive: The Fuzzy PID controller acts like a driver who slams on the gas to reach the speed limit quickly but then has to wobble the steering wheel to stay in the lane. It reaches the target almost immediately but creates significant "overshoot" (going past the target level).
  • PSO is the "Slow and Steady" Winner: The PSO algorithm acts more like a professional chauffeur. It takes the longest to reach the target level, but it does so with zero oscillation and the highest final precision.
  • Static PID is the Middle Ground: Surprisingly, the classical PID performed quite well as a balanced option, sitting between the two advanced methods in terms of both speed and stability.

Key Results

The authors compared the dynamic characteristic curves of all three systems (notably in Figure 14):

  • Response Speed: Fuzzy PID was the fastest, reaching its peak in roughly 10 time units. Classical PID followed at ~15 units, while PSO was the slowest.
  • Overshoot & Oscillation: Fuzzy PID showed the largest overshoot and required about 30 units of time to stop oscillating. Classical PID had small oscillations. PSO-PID had almost zero overshoot and zero oscillation, moving smoothly to the target.
  • Accuracy: PSO control achieved the highest final steady-state accuracy, followed by classical PID. Fuzzy PID occasionally stabilized slightly below the target value.

Limitations and Future Work

The study focuses on a single-capacity tank, which is a relatively simple, first-order system. In the real world, industrial processes often involve "multi-capacity" systems (tanks flowing into other tanks), which introduce higher-order complexities and longer dead times that might challenge these controllers differently.

The authors suggest that while PSO is excellent for accuracy, its slow response might be a hindrance in systems requiring rapid changes. Future research could look into hybrid models—perhaps a PSO-optimized Fuzzy PID—to capture the "lightning" speed of fuzzy logic with the "surgical" precision of swarm optimization. Additionally, testing these models against real-world hardware (rather than just simulation) would be the next logical step to account for sensor noise and mechanical lag.

Summary updated

Abstract

The liquid level automatic control system of a single-capacity water tank is a common industrial process control system, which is widely used in industrial production. Through simulation research, it can effectively ensure the stability of the liquid level of the water tank and meet the requirements of liquid level control in the production process. In this paper, the automatic control system model of a single-volume water tank level is established by MATLAB/Simulink, and its control system is studied and compared. Firstly, the physical modeling of the control system is carried out to obtain its transfer function, which is the basis for subsequent simulation. Then, the influence of Proportional–Integral–Derivative (PID) control parameters on the dynamic characteristic curve of the system is understood, and then the characteristics and application occasions of fuzzy PID control and Particle-Swarm-Optimization (PSO) control are deduced. Finally, the system dynamic characteristic curves of PID control, fuzzy PID control, and PSO control are compared and analyzed. The simulation results show that PID control, fuzzy PID control, and PSO control are suitable for different occasions, and appropriate control strategies can be adopted according to different control system conditions.
BibTeX
@article{tan2025control,
  author = {Zheng‐Hua Tan and Rong-Guei Tsai and Yan-Zuo Chang and Ruei-Yuan Wang and Jen-Wei Weng and Ho-Sheng Chen},
  title = {{The Control Analysis of Single Capacity Water Tank Liquid Level System Based on PSO Optimized Fuzzy PID}},
  journal = {International Journal of Advanced Engineering Management and Science},
  year = {2025},
  doi = {10.22161/ijaems.112.4},
}
Figure