This section introduces the proposed EMA, which performs several critical functions. The EMA is designed to manage, control, and optimize system performance by addressing various constraints and component behaviors. Power distribution is regulated based on key parameters, such as the state of charge and computational speed. The system utilizes estimated speed values to predict the required power and make informed operational decisions.
3.2. Smart EMA
The smart EMA based on a multi-agent system is designed to optimize the operation and control of HHEVs. This system integrates multiple intelligent agents that interact with one another and with the HHEV’s components to ensure the most efficient and effective use of energy.
- (a)
System control and management
The multi-agent algorithm takes into account several key factors, including the vehicle’s Bat status, hydrogen fuel availability, driving behavior, and environmental conditions. Each agent is assigned specific tasks, such as monitoring battery levels, predicting energy needs, and determining the optimal operating mode for the vehicle. A significant approach to achieving this is through a reinforcement learning (RL) algorithm, which enables agents to learn and adapt their behavior based on feedback from the vehicle’s sensors and control systems. The RL method utilized in this study aims to optimize the operation and control of hydrogen-powered HHEVs by effectively balancing energy consumption and resource utilization. The primary objectives include minimizing hydrogen fuel consumption and reducing overall energy usage for each trip. This entails efficiently distributing power among the vehicle’s components while considering the varying energy consumption rates of both the Bat and hydrogen fuel. By prioritizing Bat power when it is more efficient, the algorithm reduces overall hydrogen consumption. Central to this approach is the integration of a multi-agent system that incorporates RL. This allows agents to adjust their behavior in real time based on sensor and control system feedback, facilitating the optimization of energy management strategies. The goal is to identify the most efficient operating modes based on current conditions and the objectives set by the vehicle operator.
Ultimately, this algorithm seeks to optimize power distribution within HHEVs to ensure safe, reliable, and efficient operation. By balancing electric and hydrogen power sources, the system enhances efficiency, extends the vehicle’s range, and reduces its carbon footprint. The multi-agent framework promotes effective information exchange and decision-making based on individual objectives and overall energy consumption goals, contributing significantly to the efficiency and environmental sustainability of HHEVs [
38].
As shown in
Figure 4, the EMA is a sophisticated tool designed to optimize a vehicle’s energy consumption by accurately predicting the energy required along a given route. It continuously collects real-time speed data from onboard sensors as the vehicle travels along its predetermined path. These measurements are essential for making necessary adjustments, such as controlling the activation of systems like air conditioning, Bat recharging, or auxiliary equipment, at optimal moments to conserve energy. The route is generated by a GPS system that presents multiple possible paths between the starting point and the destination, each characterized by varying road conditions, traffic levels, and terrain. The EMA algorithm calculates the energy flow for each potential route, taking into account factors such as elevation changes, traffic congestion, and vehicle speed, all of which directly impact fuel efficiency and Bat depletion rates. Based on these calculations, the algorithm identifies the route that minimizes energy consumption while ensuring a timely arrival. By integrating real-time data and predictive modeling, the EMA enables dynamic decision-making to optimize energy use, resulting in more sustainable and cost-effective vehicle operation. This process is particularly valuable in hybrid and electric vehicles, where careful management of energy resources like FC, Bat, and SC is essential for maximizing efficiency and range. Additionally,
Figure 5 illustrates the transitions between various system components during the driving cycle. It is evident that equipment can be activated individually, in pairs, or simultaneously, depending on the encountered conditions and constraints. Consequently, the EMA algorithm significantly influences the system’s performance by adjusting its operations based on the current state and requirements. It effectively represents a state diagram of an energy management system for a hybrid electric vehicle, integrating components such as a SC, FC, and Bat.
The system begins in an idle state and transitions to an active state upon receiving a startup signal, leading to a transient event where the SC is first activated to supply power to the load. Depending on energy demand, the system may then shift to the FC state, where the FC either takes over or supplements the power supply. If additional energy is required, the system transitions to the Bat state to ensure continuous power delivery. Transition times (T2, T3, T4, T5, T6) indicate the shifts between these states, determined by specific conditions and energy requirements. The system ultimately proceeds to a shutdown process, concluding in an end state when all operations are completed. This diagram illustrates the dynamic operation of the energy management system, ensuring optimal energy utilization from the SC, FC, and Bat to efficiently meet the vehicle’s power demand.
The flowchart presented in
Figure 6 illustrates the energy management process for a HEV, where various power sources namely, an FC, Bat, and SC are dynamically activated based on real-time conditions to optimize energy consumption. The process begins by assessing the load demand on the vehicle; if the demand is high, the system adjusts its operations accordingly.
Decision-making is guided by the SOC of the FC and Bat, as well as the vehicle’s speed. If the SOC of the FC exceeds a defined threshold (SOCFC) and the vehicle speed is within a specified range (e.g., between 0 and 60 km/h), the fuel cell is activated to meet the energy demand. Conversely, if the FC’s SOC is below the threshold or the speed falls outside this range, the fuel cell is deactivated.
Next, the system evaluates the SOC of the Bat (SOCBat). If the Bat’s SOC is below the threshold, the battery is activated to provide energy to the vehicle. If the SOC is sufficient, the system keeps the battery in standby mode, conserving its energy for future demands.
Additionally, the SC is employed when necessary to manage transient energy surges, ensuring overall system stability and efficiency. Finally, the SOC monitor oversees the charge status of all components, making adjustments to optimize the performance of the HEV. Through this sequence, the system effectively manages the real-time activation and deactivation of energy sources, ensuring that the most energy-efficient configuration is selected at all times.
The EMA’s decision-making process dynamically allocates power from each source to meet Pdemand, adhering to Equation (3):
Priority is given to the Bat and SC when their respective SOC values exceed predefined thresholds, ensuring that transient power demands are met while reducing the strain on the FC. Additionally, the algorithm predicts future power demand (P
predicted) using real-time GPS data, factoring in traffic conditions and terrain variations, as defined by:
This predictive capability enables the EMA to optimize energy allocation proactively.
The optimization objective of the EMA is to minimize hydrogen consumption (Q
used) over the driving cycle while ensuring adequate energy availability. This is achieved by solving the objective function:
In accordance with the constraints SOCFC, SOCBat, SOCSC > SOCmin and Pdemand = PFC + PBat + PSC
- (b)
Power sizing and efficiency
Power sizing is a crucial aspect of designing a HHEV that integrates multiple power sources, such as a proton exchange membrane fuel cell (PEMFC), a Bat, and potentially a SC. Effective power sizing requires a thorough understanding of the energy flow and needs of each component. Here are the key steps to conduct a comprehensive power sizing analysis:
- -
Define Performance Requirements: Determine the vehicle’s key performance metrics, including range, acceleration, and overall power demand. This will form the basis for sizing the power components.
- -
Analyze Drive Cycle: Examine the vehicle’s typical drive cycle, including start-stop events, acceleration, deceleration, and steady-speed driving. This analysis will help establish the power demand profile under various driving conditions.
- -
Select PEMFC System: Choose a PEMFC with a power rating that meets the maximum power demand identified in the drive cycle analysis. Consider factors like efficiency, weight, and volume.
- -
Determine Bat Size: Calculate the Bat size needed to meet peak power demands and support continuous driving. Take into account factors such as depth of discharge, voltage, and capacity required for the desired range.
- -
Assess Supercapacitor Needs: Evaluate the necessity of a supercapacitor based on drive cycle characteristics. SCs are beneficial for handling peak power demands during acceleration and regenerative braking. Size the SC to provide quick power bursts and recover energy during deceleration.
- -
Implement Energy Management Strategy: Develop a strategy to manage the power flow between the PEMFC, Bat, and SC. This strategy should optimize energy use, leveraging the strengths of each component in different driving scenarios.
- -
Evaluate Efficiency: Analyze the efficiency of each component and the overall hybrid system, considering losses during energy conversion, transmission, and storage. Optimize the sizing of each component to enhance the overall efficiency of the hybrid powertrain.
The sizing study for each component involves assessing power requirements and accounting for losses to ensure optimal system performance and achieve a maximum power of 60 kW for the traction motor (refer to
Figure 7). For the FC to deliver a maximum power of 60 kW, it must be sized to provide at least 60 kW of electrical power, factoring in losses from hydrogen to electricity conversion.
To provide accurate vehicle data and assess sizing effectiveness, incorporate the appropriate losses for each component. For achieving 60 kW at the electric motor from the FC, consider losses through the DC/DC converter, bus bar, and DC/AC inverter, in addition to the internal losses of the stack, which are only used for estimating fuel consumption. Similarly, for the battery Bat and SC, internal losses should only be used to estimate their SOC, not for their sizing.
To ensure the FC delivers the required 60 kW to the motor, it is necessary to account for inefficiencies in key components such as the DC/DC converter, the DC/AC inverter, and the internal hydrogen distribution system. Each of these components contributes to the overall efficiency of the power delivery system.
The efficiency of the individual components is as follows:
DC/DC converter efficiency: ηDC/DC = 95%;
DC/AC inverter efficiency: ηDC/AC = 96%;
Hydrogen distribution efficiency: ηH2_dist = 50%.
The total system efficiency of the FC, η
FC, is the product of these three efficiencies:
This means the overall efficiency of the FC system is approximately 45.6%.
To determine the nominal power, the FC needs to generate in order to meet the 60-kW requirement on the motor, we use the relationship:
Therefore, the FC must produce a nominal power output of approximately 73.08 kW to ensure that the motor receives 60 kW, accounting for the losses in the hydrogen distribution, DC/DC conversion, and DC/AC inversion processes.
Similar considerations apply to both the Bat and the SC when determining their sizing for effective power delivery. For the battery, it is essential to account for the DC/DC converter efficiency, which is 95%, as well as the battery’s internal losses, also assumed to be 5%, resulting in an internal efficiency (η
Bat_int) of 95%. These factors ensure that the battery can provide the necessary power by considering the total conversion efficiency from stored energy to usable output. Therefore, the Bat efficiency is expressed as:
For the supercapacitor, assuming the same efficiencies and power nominal as bat:
Although constant efficiency values are assumed for the FC, Bat, and SC, these values are, in reality, complex functions of several parameters, including temperature and state of health. This assumption simplifies the analysis and allows the focus to remain on the overall energy management strategy.
The rationale behind sizing the supercapacitor in a HHEV focuses on its critical functions and purposes. The SC is primarily intended to manage transient events, such as rapid acceleration and regenerative braking, thanks to its high-power density which allows for quick energy absorption and delivery. Additionally, it supports the bat and FC during peak power demands, alleviating stress on these components and thereby improving their durability and efficiency. By providing a reliable power source in situations where traditional power sources might struggle, the SC is crucial for maintaining vehicle reliability under various driving conditions. This strategic sizing, illustrated in detail in the accompanying diagram (see
Figure 8), highlights the SC’s vital role in enhancing system efficiency, safety margins, and overall performance in hybrid electric vehicles.
The HHEV system illustrated in the figure integrates three main ES components: a PEMFC pack, a lithium-ion Bat pack, and an SC pack, each designed to fulfill different power demands within the vehicle. The PEMFC pack, consisting of 1500 cells, generates a total power output of 75 kW, with each cell delivering a nominal power of 50 W, a voltage of 0.7 V, and a current of 71.42 A. This component acts as the primary energy source, supplying continuous power for sustained driving scenarios due to its high energy density. In addition, the lithium-ion Bat pack, comprised of 53 cells, with each cell providing a nominal power of 1.28 kW, a voltage of 3.2 V, and a current of 400 A, contributes 67 kW of power. The Bat is crucial for covering medium power demands, such as during vehicle acceleration or maintaining a constant speed, thus collaborating with the fuel cell to ensure smooth and responsive energy delivery. Furthermore, the SC pack, which consists of 21 cells and delivers 67 kW of power with each cell providing 3.2 kW, a voltage of 2.7 V, and a current of 1200 A, addresses short-term, high-power demands, particularly useful during rapid acceleration or energy recovery in regenerative braking. The overall energy management system ensures that the vehicle receives a combined 60 kW of power, with dynamic allocation from the FC, Bat, and SC depending on real-time driving conditions. In terms of system dynamics, the speed estimator operates at a time step of 1 s, optimizing real-time power distribution. Additionally, the FC’s total capacity is tailored to maximize efficiency, with the lithium-ion Bat and SC pack supporting fast responses and energy recovery. This hybrid system balances energy density with power density, enhancing overall vehicle performance while optimizing energy consumption and reducing emissions, as detailed in
Figure 8.
Table 2 illustrates the sizing characteristics of HHEV components.
The battery capacity in ampere-hours (Ah) is calculated using the formula
For a nominal power (P
nominal) of 1.28 kW and a voltage (V
Bat) of 3.2 V, the Bat capacity (C
Bat) is 400 Ah, meaning it can provide 400 amperes for one hour. The Crate, which indicates the discharge rate relative to the Bat’s capacity, is calculated as:
With a discharge current (I
Discharge) of 400 A and a Bat capacity (C
Bat) of 400 Ah, the Crate is 1, indicating a full discharge in one hour. The Bat’s energy capacity (E
Bat) is 1.28 kWh, calculated as:
In practical applications, the capacity of SCs is typically specified in farads. For example, a SC with a specified capacity of 1200 F directly indicates its energy storage capability, eliminating the need for additional calculations.
The overall system efficiency is assessed based on the chosen operating mode, which depends on the activation status of each component. Thus, the system’s efficiency is determined by the product of the efficiencies of each active component.
Table 3 displays the truth table for the system efficiency, reflecting the activation status of each component as is seen in
Figure 9.
Total system efficiency is a comprehensive metric that assesses how effectively power is distributed among the system’s components.
Table 3 illustrates efficiency calculation considering driving scenario. It is determined by dividing the sum of efficiencies for various pathways by the overall efficiency of each component. This method takes into account both the efficiency of individual components and how well power is managed between them. By evaluating these aspects, total system efficiency provides a thorough evaluation of how effectively the system converts input energy into useful output. It reflects the system’s overall ability to utilize energy efficiently across its different pathways and helps gauge its success in achieving performance goals. It can be expressed as:
where: η
Bat: Bat efficiency (%)
ηFC: FC efficiency (%)
ηSC: SC efficiency (%)