PID CONTROL How PID control developments are improving EV power performance

Related Vendors

ROHM Semiconductor GmbH

Rogers Corporation

dSPACE GmbH

PID control contributes to EV power design – but classic solutions can’t handle EV system nonlinearities, and lack AI’s learning capability. EV designers have responded to these challenges by adopting advanced PID calculus, combining techniques, and adding AI capability. This article compares these approaches, and shows how they can be built into optimal solutions.

In March 2025, a report titled "Electric vehicle market size and share analysis” predicted that the global EV market would grow from USD427.02 billion in 2025 to USD713.07 billion in 2032 – a compound annual growth rate (CAGR) of 7.6 %.¹

Factors driving this growth include increased public awareness of green issues and desire to reduce carbon footprint, and growing international government legislation to make EV ownership more attractive. However, it is also being driven by declining battery prices and significant technological advances.

Individual auto manufacturers can’t control sustainability trends or government legislation, but they can make sure their vehicles have the best and latest EV technology designed in, to maximize their competitive position.

Most improvements relate to reducing size, weight and cost, improving energy efficiency, and boosting battery performance and lifetime. These are critical because they can mitigate capital costs and range anxiety – two of the biggest barriers to EV sales.

And PID control of various types (as we shall see) makes multiple contributions to these improvements. For example, PID controllers help regulate motor speed and torque by adjusting the voltage and current supplied to the motor. This ensures smooth acceleration, deceleration, and stability in driving conditions. Other PID controllers assist in managing battery charging and discharging. They ensure optimal charging rates and prevent overcharging or excessive discharge, improving battery longevity.

Regenerative braking, where an EV’s kinetic energy is converted back to electrical energy stored in the battery, is another key factor. During braking, PID controllers help control the amount of energy fed back into the battery, optimizing efficiency and extending range.

Other examples include thermal management to regulate cooling systems, and DC-DC converter control, to ensure stable power to auxiliary systems and maintain efficiency.

Overall, by dynamically adjusting outputs based on error signals, PID controllers contribute to more stable and efficient EV operation, enhancing performance, range, and reliability. PID technology, which has been used throughout industry for decades, has been adopted by EV designers because it is simple and effective, and offers a fast response with high accuracy. It is also versatile, and suitable for many different applications.

Yet it does have limitations. Tuning can be difficult, especially in dynamic conditions – and poor tuning can cause excessive energy consumption. The derivative term amplifies high-frequency noise, which can lead to instability. Nonlinearities and unpredictable variations in EV dynamics are extremely challenging for PID controllers. Another major disadvantage, highlighted by increasingly pervasive machine learning (ML) and artificial intelligence (AI) technologies, is that they have no learning capabilities, and cannot adapt or learn from past performance.

Accordingly, EV power system control capability – like control systems across most other industries – is being improved through AI techniques. But that doesn’t mean that PID is being entirely replaced; instead, it is evolving in two ways. Firstly, improved algorithms such as fuzzy logic and fractional order PID are being adopted. Secondly AI capability is being added to the PID equation.

Below, we discuss these techniques and how each can improve PID controller performance. Then we look at an example showing how they’re combined into a real EV motor speed control application, while being enhanced with Artificial Intelligence capability.

CONTROL TECHNOLOGY

Why PID control technology continues to thrive

Fuzzy logic

Fuzzy logic is used to enhance PID controllers by translating operator control actions into a rule base².

The term fuzzy refers to things that are not clear or are vague. In the real world we often encounter a situation where we can’t determine whether the state of a variable is true or false; if so, fuzzy logic provides valuable flexibility for reasoning. It is a form of many-valued logic. In this way, we can consider the inaccuracies and uncertainties of any situation.

In the Boolean system truth tables, 1 represents an absolute truth value and 0 represents an absolute false value. But in the fuzzy system, the truth values of variables may be any real number between 0 and 1 (both inclusive). It is employed to handle the concept of partial truth, where the truth value may range between completely true and completely false.

Fuzzy models or sets are mathematical means of representing vagueness and imprecise information. These models have the capability of recognizing, representing, manipulating, interpreting, and utilizing data and information that are vague and lack certainty.

Fuzzy logic system architecture has four main components:

• Fuzzification Module: Fuzzification is the process of assigning the numerical input to fuzzy sets with some degree of membership. (This degree of membership may be anywhere within the interval [0,1].) A fuzzification module transforms the system inputs, which are crisp numbers, into fuzzy sets.

• Knowledge Base: Stores IF-THEN rules provided by experts.

• Inference Engine: Simulates the human reasoning process by making fuzzy inferences on the inputs and IF-THEN rules. IF-THEN rules map input or computed truth values to desired output truth values.

• Defuzzification Module: Transforms the fuzzy set obtained by the inference engine into a crisp value.

While classic PID implementations remain the most widely used control solution across multiple industrial (and other) applications, they are not always the right answer. For example, being linear, they are unsuitable for strongly nonlinear situations; a reheating furnace, for example, has heat transfer properties which may be greatly influenced by operating conditions such as stock material properties, furnace scheduling and throughput rate – so a nonlinear relationship between input and output³.

Additionally, the PID controller’s Integral and Differential gain constants need regular retuning by skilled operators during ongoing processes. By comparison, a fuzzy PID algorithm significantly reduces the need for algorithm re-tuning during slowly-changing processes.

So, PID control is a well-established way of driving a system towards a target position or level, but fuzzy logic offers one way of enhancing it. The technique was developed to handle the fuzziness found in human concepts such as those embedded in the knowledge base of an expert system. Fuzzy logic provides a certain level of artificial intelligence to conventional PID controllers, giving them adaptation to nonlinear, time-varying, and uncertain systems. It has been observed that fuzzy PID controllers perform much better than classical PID controllers.

Regenerative braking – a key element of EV design – is one area in which efficiency can be improved by fuzzy logic control. This possibility has been addressed in an article titled Regenerative Braking of Electric Vehicles Based on Fuzzy Control Strategy, available from scholarly open source publisher MDPI⁴.

The article explains that designing a strategy for distributing regenerative braking force can directly improve EV energy efficiency.

The distribution strategy of the front- and rear-axle braking forces of electric vehicles that possess integrated front-wheel-drive arrangements is based on the Economic Commission of Europe (ECE) regulations, which enable the specification of the total front axle braking force. The motor’s regenerative braking torque model is adjusted to optimize the ratio of motor braking force to the whole front-axle braking force.

The regenerative braking process of electric vehicles is influenced by many factors, such as driving speed and braking intensity, so regenerative braking presents characteristics of nonlinearity, time variability, delay, and incomplete models.

Bearing in mind fuzzy controllers’ improved robustness, adaptability, and fault tolerance, the article considers a fuzzy control strategy to accomplish braking force distribution to the front axle.

A regenerative braking model is created on the Simulink platform using the braking force distribution indicated above, and experiments are run under six specific operating conditions: New European Driving Cycle (NEDC), World Light-Duty Vehicle Test Cycle (WLTC), Federal Test Procedure 72 (FTP-72), Federal Test Procedure 75 (FTP-75), China Light-Duty Vehicle Test Cycle-Passenger (CLTC-P), and New York City Cycle (NYCC).

The findings demonstrate that in six typical cycling road conditions, the energy saving efficiency of electric vehicles has greatly increased, reaching over 15 %. The energy saving efficiency during the WLTC driving condition reaches 25 %, and it rises to 30 % under the FTP-72, FTP-75, and CLTC-P driving conditions. Additionally, under the NYCC road conditions, the energy saving efficiency exceeded 40 %.

Accordingly, these results verify the effectiveness of the regenerative braking control strategy proposed in the paper.

ELECTROMOBILITY

How the rapid growth in e-mobility is driving the power electronics market

Fractional Order PID (FOPID)

Fractional Order PID (FOPID) controllers extend traditional PID types by incorporating fractional calculus, allowing for more flexible and precise control. They use non-integer order differentiation and integration, which can improve system performance in various applications.

While fuzzy logic is used when system behavior is not well-defined or too complex for precise modeling, FOPID is used where system dynamics exhibit long-term memory or anomalous behavior.

Fractional calculus as used within FOPID control allows operations like 0.5th-order differentiation or 1.3th-order integration, providing a broader and more flexible mathematical framework.

A FOPID controller includes the conventional PID algorithm, with its three terms – and associated constants – for Proportional, Integral, and Differential gain. Additionally, though, it has two further tuning parameters, DλD^{\lambda} and DµD^{\mu}, which represent fractional derivatives and integrals. These allow finer control system tuning, which can lead to better stability, faster response times, and improved robustness in dynamic environments.

While these extra tuning parameters allow finer tuning, they also create extra complexity in finding the right parameters for optimum controller performance. However, optimization techniques like Particle Swarm Optimization (PSO) can help find optimal parameter values.

A paper titled ‘Comparative Performance Evaluation of Swarm Intelligence-Based FOPID Controllers for PMSM Speed Control’ describes how FOPID controllers using swarm intelligence tuning can be optimally applied to speed control of Permanent Magnet Synchronous Motors (PMSMs), which are frequently found in EVs as well as other applications⁵.

The study looks at how PMSM motors’ nonlinear and time-varying characteristics pose significant challenges to providing precise and efficient speed control. It then proposes a solution utilizing FOPID controllers, which it believes offer advantages over traditional PID including improved robustness and greater flexibility in handling complex system dynamics. Additionally, the study explores the use of Swarm Intelligence (S.I.) algorithms for the design and tuning of FOPID controllers. Swarm Intelligence algorithms, such as Particle Swarm Optimization (PSO), ant colony optimization (ACO), and Grey Wolf Optimization (GWO), are known for their ability to effectively search and optimize complex parameter spaces.

The study’s key finding is that the ACO-FOPID controller exhibits the best performance in terms of transient response. It achieves a rise time of 0.008978 s, a settling time of 0.01 s, and zero absolute time error (ITAE). These results indicate that the ACO-FOPID controller provides precise and fast speed control for PMSMs, making it a promising solution for practical applications.

This research contributes to the advancement of control systems for PMSMs and showcases the potential of Swarm Intelligence algorithms in optimizing complex control parameters.

Ant colony optimization (ACO)

Ant colony optimization is a meta-heuristic algorithm inspired by ants' behaviour, notably their ability to locate the shortest path between their nest and a food source. In particular, the method was motivated by the power of ants to discover the route that takes them the shortest distance from their colony to a source of food. In ACO, a set of artificial ants cooperate to find suitable solutions to optimization problems by following a pheromone, Trai. The algorithm works as follows: Initialize a set of artificial ants at the starting position. Each chooses the next vertex to visit based on a probabilistic rule that considers the amount of pheromone on the edges and the heuristic information about the desirability of each vertex. After all the ants have completed their tour, the amount of pheromone on the edges is updated based on the quality of the solution found. The pheromone trail is updated by evaporating a certain percentage and depositing new pheromones on the edges of the best solution found so far [24]. ACO’s main advantage is its ability to find suitable solutions in complex, high-dimensional optimization problems with multiple local optima. ACO has been successfully applied to many issues, including the travelling salesperson problem, the quadratic assignment problem, and the job shop scheduling problem. However, ACO can be sensitive to the choice of parameters, such as the pheromone evaporation rate and the balance between pheromone and heuristic information in the decision rule. Therefore, carefully tuning the parameters is often required to perform well.

Further innovation, and AI enhancement

We have seen how fuzzy logic and FOPID techniques have been used to provide diverse improvements to basic PID control. However other strategies are also available to improve adaptability, efficiency, and robustness; these include Model Predictive Control (MPC), Adaptive Control, Sliding Mode Control, (SMC), and H-infinity & Robust Control. Designers compare the various attributes of these strategies with the demands of their particular application to determine which would suit them best.

Yet, while these approaches can be considered as alternatives to fuzzy logic or FOPID, some ongoing research & development is focusing on enhancements, rather than alternatives, to fuzzy/FOPID approaches. These include combining the approaches, and/or adding AI capability to the controller. Let’s take a closer look:

Hybrid fuzzy/FOPID solutions combine the two methods to exploit the strengths of both.

A Fuzzy Fractional Order PID (Fuzzy FO-PID) uses a fuzzy logic controller to determine or adapt the gains or parameters of fractional order PID. The controller has fractional dynamics which are tuned or enhanced by fuzzy rules.

Fractional Fuzzy Logic Controllers have a fuzzy controller structure enhanced with fractional calculus elements, such as fractional membership functions or fractional integration/derivation in decision-making.

Either way, combining fuzzy logic with fractional order control allows:

• Better handling of nonlinearities and uncertainties (via fuzzy logic).

• Better system memory modeling and dynamic response (via fractional order terms).

• Robust and adaptive performance in complex real-world systems.

PID REGULATION

How PID controllers contribute to smart grid functionality

ANFIS and FOPID hybrid – AI enhancement to PID calculus

A paper titled “Electric vehicle speed tracking control using an ANFIS-based fractional order PID controller”⁶ proposes a control scheme which combines FOPID with fuzzy logic enhanced by artificial intelligence (AI). ANFIS means Adaptive Neuro-Fuzzy Inference system, which is a hybrid AI system that combines neural networks and fuzzy logic principles to create a powerful framework for decision-making and pattern recognition.

The paper proposes this hybrid controller to provide efficient speed-tracking control for an EV DC motor.

The optimal controller parameters of the FOPID controller are found via an Ant Colony Optimization (ACO) method. The ANFIS controllers are well trained, tested, and validated using a data set extracted from the fuzzy-based controllers. The performance and accuracy of the ANFIS model are evaluated using statistical parameters such as mean square error (MSE), coefficient of correlation (R), and root mean square error (RMSE). The controller performance, energy consumption, and robustness are tested using the new European drive cycle (NEDC) test.

The paper demonstrates the ANFIS-based controller’s efficacy by comparing its performance with properly tuned fuzzy-based controllers. The proposed controller shows robustness towards external disturbances and offers promising EV speed regulation control.

The reliability and durability of an actuator system play a substantial role in determining the overall efficacy of an EV system. The DC motor is the most commonly used actuator in EVs with wide-ranging applications due to its critical advantages like high reliability, low cost, and easy maintenance.

The comparative results illustrate the superior performance of ANFIS-based FOPID controllers with high prediction and low error rates. A MATLAB- Simulink platform was used for system modeling, controller design, and numerical simulation.

Conclusion

Experts in control systems have different opinions on the future of PID control with AI. Some believe that AI will eventually replace the PID algorithm, while others believe that PID control will continue to be used in many applications⁷.

One thing is clear, however – the use of AI in control systems will only become more prevalent in the coming years. As AI technology improves, it will become more accessible and easier to use, making it a valuable tool for system control.

References

• ¹Electric Vehicle Market Size and YoY Growth Rate, 2025-2032

• ²FUZZY PID CONTROLLERS. Introduction | by Niranjana Madathil | K. R. I. S. S | Medium

• ³HuTanRoach_ATE2018_post_print.pdf

• ⁴Regenerative Braking of Electric Vehicles Based on Fuzzy Control Strategy

• ⁵Comparative Performance Evaluation of Swarm Intelligence-Based FOPID Controllers for PMSM Speed Control | IIETA

• ⁶Electric vehicle speed tracking control using an ANFIS-based fractional order PID controller - ScienceDirect

• ⁷The Future of PID Control: Will AI Take Over?

(ID:50420976)