Electric_Vehicle_Braking_by_Fuzzy_Logic_Control

Electric Vehicle Braking by Fuzzy Logic Control

John Paterson, Mike Ramsay

ORTECH International

2395 Speakman Dr.

Mississuaga, Ontario, L5K 1B3, Canada

Abstract: Electrical vehicles are really "energy-management"machines. Energy recovery during braking, at speeds of 80 km/hr down to 8 km/hr, increases efficiency to the high 90's.However, binary switching of braking action from regenerative to mechanical causes a jolt that would catch most drivers by unpleasant surprise. Braking mode switching by fuzzy logic with a linear T-Norm smoothes out the braking curve, but not completely. Using a modified sinusoidal T-Norm is able to

produce a smoother braking curve.

I. INTRODUCTION

The present types of electric and hybrid electric vehicles under development can be considered the third generation of electric vehicles. By that it is meant that the first generation, as early as the 1900's, had a simple DC electric motor, the best available lead acid battery of the times, (1910) and a control switch rheostat for the sped control. These early electric vehicles had better performance in their day than the "infernal combustion" vehicles of that time, but continuous development and raw power soon saw the internal combustion-powered vehicle far exceeding the power and range of the battery-powered electric vehicle . It can be argued, with some truth, that if all the resources spent in developing the internal combustion engine had been spent on electric vehicles and batteries since about 1910, we might have some outstanding electric vehicles today. However that is not how it happened, and we are where we are, with some proven beliefs that continuing investment in hybrids/electrics will be productive. Also, these first generation electric vehicles employed very crude control mechanisms which wasted the vital battery energy, and ultimately put them in disfavour. Primarily, it was the lack of suitable solid-state electronic controls, some mistaken strategies for vehicle operation, and of course the lack of battery technologies with suitable energy and power densities which made them uncompetitive. This interesting history of the early electric vehicles has been well documented by Wouk, Kalberiah, and Landman, Patil and Burba.

During the second world war, there was a great leap forward in technology for electronics and electrical systems, particularly for submarines' electric power drives. In the early 1970's right through to the late 1980's, second generation electric vehicles were being built, this time using efficient power controls and regenerative braking. Efficient power controls, particularly the solid state type, allowed the system to manage power to the motor with passon losses less than 5 %; previously an impossible figure to reach. Regenerative braking allowed the vehicle to use the drive motor as a generator every time the brakes were used, to pump the kinetic energy from braking into the battery, using these same efficient power controls. Surprisingly, regenerative braking saved from 8 4% to as much as 25 5% of the total energy use of the vehicle, depending on the driving cycle and how it was driven . Such savings are necessary in any machine where the energy storage capability is minimal, as is the case with the electric vehicle,

particularly since it can be achieved without the addition of any extra components.

Hybrid vehicles, vehicles with both electric and internal combustion engines, are currently being extensively investigated. The major motivation behind the research at present is the new pollution laws being introduced throughout the United States. The bus industry has been targeted due to its large start-stop driving cycle and the potential to recover energy in the electrical system. However, electric vehicles for consumer use are continually being introduced .

Various methods for retaining energy in electric vehicles have been introduced, however, the use of regenerative braking is the most attractive since it does not require the addition of large extra equipment . Regenerative braking is accepted today as a vital part of the standard architecture for third generation vehicles which now includes the electronics for total energy management on board the vehicle.

Experiments with electric vehicles having a combination of regenarative and physical modes of braking have proven that substantial energy savings are in fact achievable. However there is a critical human factor that is important; it has been found that it is difficult to be sufficiently accurate in switching off the regenerative braking and at the same time to switch on the physical brakes, to the same level of braking. The observation by the passengers of this discontinuous braking is an uncomfortable and unacceptable jolt. Uses of simple rules and conventional binary logic have been defied by this problem. Since there are some definablefuzzy sets of circumstances and groups of rules which apply to this problem, it can in fact be solved very well by the application of a fuzzy controller algorithm implemented in the software of the energy management system.

2. REGENERATIVBER AKING

In using regenerative braking, a significant change is made to the architecture of the vehicle. The brake pedal is sensed for the pressure applied, and a microprocessor makes a comparison between the velocity of the vehicle, and the existing state of charge (SOC) of the battery. The objective of this is to determine energy discharge rates, which vary as the SQUARE of the vehicle speed, and how much energy the battery can absorb and at what power rate. Ultimately, as the vehicle slows down, regeneration becomes useless and the physical brakes have to be applied to stop the vehicle at a particular point. It can be shown, however, that by slowing down from cruising speeds of 80 kmmr to 8 M h r at normal decelerations using regenerative braking, as much as 99 W of the kinetic energy is recovered into the battery, if the battery can take it.

Consider an electric vehicle with a mass of 1500 kg and assume a deceleration of 0.25 m per s2. The total energy then available through braking from 80 kmhr to 0 km/hr, as calculated in equation 1 is:

The graph of this braking curve is shown in Figure 1. The energy available from regenerative braking is:

Therefore 99.9% of the total energy is available from regenerative braking.

These are remarkable figures by any standards, and show the necessity for the architectures of third generation electric vehicle's to be pursued until successful. Successful in these terms means probably an electric vehicle of acceptable emissions range and speed, but considerably less than the range and speed of any conventional vehicle.

Significant changes are made to the power control systems of an electric vehicle to accommodate regenerative braking. In particular, the normal power control electronics must be configured to be switched so that it can act as an energy controller, and possible power rectifier for the output of the generator (the motor acting as a generator). In most cases the high-efficiency motors of today are high speed AC motors, and for simplicity and efficiency reasons their output is usually high voltage AC. Furthermore, the electronics must sense certain conditions in which the battery cannot take the charge, and so the energy needs to be dumped to a resistive load which is very useful in winter as a way of extending the heat in the vehicle. In a hot climate, the electronics control decides that the physical brakes would be used rather than resistive load heating. The above are examples of the kinds of decisions which are continuously being taken in the third-generation electric vehic1e .

Ill. REGENERATIVSEW ITCHING

Regenerative braking is practical only to approximately 8 km/hr at which point the

generation of electricity for retum to the battery is outweighed by the need to decelerate quickly. Regenerative braking is proportional to the vehicle's velocity, therefore allowing the vehicle to coast to a stop rather then coming to a controlled stop. To achieve this braking a physical brake must be applied. To produce a smooth ride the transition from electrical regeneration to physical braking must be done to match the velocity curve of the decelerating vehicle. To achieve this both the electrical and mechanical brakes must be released and applied, respectively, simultaneously. This presents a problem in the mechanical brakes due to unknown and variable timing parameters. These parameters include wear, temperature, and force applied.

These variable parameters combine to form a window of operation for the mechanical braking of the vehicle. This window translates into a window of operation in which the vehicle's velocity may occupy during braking. A sample of this window is sketched in figure 2 for a digital logic operation of switching between the two modes. The lighter shaded region is the window of operation of the mechanical braking and is determined by the parameters outlined above. The darker shaded region is the corresponding window of operation of the braking velocity curve for the vehicle.

To produce a smooth stopping motion the two braking systems must be coupled such that the combined effects of the two do not produce a deceleration change between switching. This change is felt by the operator as either an acceleration or an unpleasant jerking motion. These responses may occur at any velocity since the regenerative braking is controlled by the state of charge of the battery. Once the battery has achieved full charge and no resistive load heating is required, the physical brakes are applied. Under most circumstances, however, the battery will not be at full charge and the switching will occur around 8 km/hr.

4 Fuzzy Logic

The there several variables present in the switching that lend themselves well to the use of fuzzy logic for control. The value of 8 km/hr as the velocity at which the switching is to occur is not fixed and forms the basis for the fuzzy switching. As well the state of charge may also be considered in the fuzzy logic control rules since the velocity at which the switching occurs would be more desirable at either the initial braking motion or at a lower velocity. The initial velocity at which braking occurs could also be used in a fuzzy controller to determine the most beneficial use of the two modes

of braking. The human interface in the form of the pressure applied to the brake pedal will also be considered in the rule set to determine the emergencybraking and the use of both braking methods simultaneously. These variables are shown in table 1.

Since none of these variables can be easily identified as a fixed value, the use fuzzy logic assists the controller in determining the most beneficial use of the two modes to produce the smoothest ride. The smoothest ride is the prime concern of the controller but the governing rules are derived from the necessity to recover the most amount of energy possible. It should be noted that with all vehicle systems operator safety is the first concern.

By changing the switch point from 8 km/hr to 10 km/hr there is only a 0.095% decrease in the amount of energy available for regenerative returns to the battery. If the switch point is reduced to 6 km/hr the change in the amount of energy is increased by only 0.058%. The percentage of energy available versus the velocity at which the brake modes are switched is illustrated in the graph in figure 3. By reducing the switch point the vehicle's will experience a slight coasting effect due to the decreased efficiency of the regenerative braking to maintain a constant deceleration force. Thus minor changes to the switch point are acceptable in terms of energy recovery and management. The advantage of this change is that the physical brakes may be allowed to gently take over from the regenerative brakes. The main advantage is that when the car is operating at lower velocities the controller can determine what switch point is best for the operators comfort and avoid the jerk of a quick changeover to physical brakes.

To assist in making the switch over of modes more continuous in terms of the velocity curve and the operator's comfort, the fuzzy logic membership functions have been determined using a sine function [23]. An example of this is shown in figure 4. This provides the controller a heavier weighting towards the desired position but will allow the variables to effluence the decisions made by the controller. For example, if the operator is braking to negotiate a sharp turn the controller will determine if the battery requires charge and if the present speed will accommodate the amount of charge generated based on the pressure applied tothe brake pedal. All of the variables listed in table 1 would be considered in the rule set.

By exploiting the inherent tolerances in the braking system a fuzzy controller may be used to achieve a more comfortable ride for the vehicle operator. Tolerances in energy recovery were exploited to gain a more acceptable performance. The fuzzy controller model reduces the velocity error but does not eliminate it completely. A sketch of the fuzzy logic errors are illustrated in figure 5. The magnitude of the emors has been reduced and separated to the two conditions where only one of the modes is operating and the other has either not initiated deceleration or has turned off before the controller expected. These errors are due to the time variables of temperature, brake pad wear, brake pad temperature, and the mechanical frictions in the system. It is hoped tbat with the addition of an adaptive control scheme to monitor these time constants and predict them, will reduce the unwanted jerks in the switching of brake modes. By using the tolerances available in both the energy recovery system and the operating conditions a smoother operation of the vehicle is possible. Due to the safety factors involved an extensive investigation will have to be done to ensure vehicle safety at all

times during operation. The main concern at present is the determination of when the vehicle is under Emergency braking and when the excessive pedal force is just a function of the human operator and the current velocity.

ACKNOWLEDGMENT

ORTECH and the authors would like to thank Dr. Gorden Chen from Niagara College for his assistance in the preparation of this paper and his encouragement during the research.