Adaptive Control Technique for Portable Solar Powered EV Charging Adapter to Operate in Remote Location

Every EV (Electric Vehicle) comes with limited energy storing capability. After travelling a certain distance, a charging facility is required to recharge the EV batteries, which is easy to be made available in cities. But, in remote locations, charging service is challenging. Therefore, big countries like USA, Canada, China, Russia, India, Australia, and few Arabian countries are planning to provide pillar top solar panels on remote locations for EV charging in emergency situations. To operate in this situation, a special charging adapter is required to extract maximum power from the panel using the MPPT (Maximum Power Point Tracking) technique, monitor the charging current, and safely complete the charging process. In this paper, a single sensor-based economical charging adapter is presented for EVs to fulfil this objective. Moreover, the Single Input Fuzzy Logic tuned Deterministic Optimization (SIFL-DO) algorithm is proposed to accomplish MPPT operation and battery charging management. Because of its low cost and fast response, the single current sensor-based charging adapter is highly economical. Additionally, the SIFL-DO algorithm has very good condition estimation and decision-making capability, which accurately performs MPPT and charging management. In this work, the capability of the developed adapter with the SIFL-DO algorithm is evaluated on Hardware prototype. Also, comparative studies are performed w.r.t. state-of-the-art techniques. Further to determine the industry’s suitability, the developed technique is tested on European Standard EN50530.


I. INTRODUCTION
E LECTRIC Vehicles (EVs) have shown to be a viable alternative to hydrocarbon automobiles considering they emit no greenhouse gas and do not rely on crude oilproducing countries to determine gas prices. Other than this the entire full energy cycle is more efficient in the case of renewable energy, notably for solar photovoltaic (SPV) energy generation. We can readily observe how electricity rates remain reasonably stable while fuel costs surge throughout every international conflict. Considering all of these factors, electric vehicles are a better substitute to gasoline-powered vehicles. Whereas a significant amount of work is being undertaken to develop efficient electric vehicles and their corresponding infrastructural facilities, electric vehicles primarily use two charging schemes: onboard charging and off-board charging. The charging circuit of the on-board is retained within the automobile itself, and it may be charged whether using AC or DC. The charging circuit is housed within the charging station in the case of the off board, and an EV may indeed be charged using either 1-phase AC, 3-phase AC, or DC power, each of which has advantages as well as disadvantages. Charging an EV using solar energy can be a difficult operation considering solar irradiance and temperature change over time, contrary to an idealistic situation where weather conditions and temperature are constant [1]. As a result, the power generated is indeed not uniform. Various conventional mathematical and neural computing-based optimization approaches and techniques are implemented to assess this and it is known as "Maximum power point tracking (MPPT) of solar arrays [2], and each MPPT algorithm setup has advantages as well as disadvantages. This is implemented in series with the circuit referred to as the DC-DC Buck/Boost converter, whereby the solar panel's power output may very well be regulated by numerous MPP control methods mentioned above by altering the converter's pulse width modulated (PWM) input signal. Using this, solar panels are maintained at their maximum power point (Vmp, Imp), going to charge the EV at the solar PV's maximum power output [3]. Further to that, since solar photovoltaic systems can be broadly classified as gridconnected or remote stand-alone systems, the system's power capacity can differ widely from low-powered operation to high-powered operation, the median generation capacity can differ of these systems is still well above hundreds of kW (kilo-Watts). It is also known that a sizable SPV can produce hundreds of amps of direct current, which also, if connected directly into the Electric vehicle, can cause damage [4].
In literature, different solutions for PV-powered EV chargers are proposed, such as Multifunctional Off-Board EV Charger [5], Bidirectional EV Charger [6], Z-Source based charger [7], Wireless EV Charger [8], Intermittency mitigated EV Charger [9], etc. All these chargers are absolutely fine and suitable for EVs. However, in all these configurations and topologies, the grid is mandatory with PV to operate during the charging process. In the case of the standalone condition, these techniques will not work. Therefore, this paper proposes a novel topology and control to operate in a standalone condition.
Additionally, an EV has a charging current rating that ordinarily ranges from 10% to 15% of the maximum capacity of the EV battery, with most EVs receiving up to 32A for power delivery up to 7.4KWs from a typical charger, where the EV's battery is not continuously cooled throughout this charging period. Consequently, current must be monitored before being supplied into the EV [10], which may be performed by analyzing the solar array's output with sensors [11], [12]. Therefore, although solar PV can be found in even the most sparsely populated areas due to the increased intervention from the government, EV chargers are currently difficult to discover even those in tier 2 or 3 cities, let alone in villages or isolated areas. And hardly any research has been dedicated to the accessibility of charging infrastructure in remote locations where, if an EV user becomes stuck, they would have been unable to use an EV charging station since incorporating complex and costly charging stations for electric may not even be very economical in areas where the number of consumers is rather relatively small, as well as remote locations make it much more difficult for repair or replacement, which arises as a potential hazard which will always be considered as part. These characteristics summarize the fact that this is not the real solution to this particular circumstance. Therefore, in big countries like the USA, Canada, China, Russia, India, Australia, as well as in a few Arabian countries, governments are planning to provide pillar top solar panels in the remote locations for EV charging in an emergency situation (such as shown in Fig. 1). To operate in this situation, a special charging adapter requires, which extracts maximum power from the panel using MPPT technique and monitors charging current and safely completes the charging process.
Therefore, as conclusion, can see the need for a compact, cost-effective charging solution that can be immediately linked to any Photovoltaic system and stored within the car and utilized in cases, for example like these as well as routine everyday charging needs and requirements. Even if certain work is underway in this area as well, wherein multiple sensors are being used to control the power fed into the Electric vehicle based on the capacity of the battery [13], [14], it uses a large number of sensor nodes, making it bulky and expensive, as well as increasing the possibility of electro-magnetic interference in the system. Furthermore, there is unquestionably no system on the market that can accomplish these two objectives, namely effective MPPT operating of solar photovoltaic and current strictly controlled charging of EV.
In this paper, both maximum power point tracking of solar photovoltaic array and controlling the current fed into the electric vehicle to charge the EV Battery safely as per the current limit, Single Input Fuzzy Logic tuned Deterministic Optimization (SIFL-DO) method is proposed. Moreover, for both the operation single sensor is been used to monitor and control the output of the solar PV array and current that is fed into the EV for charging its. Further to through experimental results the proposed method shows the clear advantage it offers in its operation as the step change is not a fixed constant but a variable in nature. Comparison is made with different leading state-of-the-art algorithms, which certainly demonstrates the leading edge of our proposed SIFL-DO methodology. Moreover, efficiency was tested to see the  industry redness of the method based upon the European standard (ES), i.e., EN50530 [15].

II. SYSTEM LAYOUT
The proposed model of the complete charging setup is shown in Fig. 1, where as per the circumstance when EV needs to be charged in any remote location can be directly linked to any available solar panels in that location using our proposed method, which in this case is shown as a complete solution as charging adapter, Fig. 2 represents the circuitry involved in it. Moreover, the EV can be connected along with the adapter to any solar PV even if the solar has no circuitry attached to it. The charging adapter will contain the circuitry for both the MPPT scheme and the charging current control scheme.

A. COMPLETE MODEL SETUP
In Fig. 2, the proposed charging adapter model is given, which shows a solar power monitoring sensor (current sensor), DC-DC Boost converter, and proposed SIFL-DO controller. Moreover, the whole circuitry will be kept inside the vehicle for on-board charging or in a box for off-board charging.  Fig. 3. Where I Ph and I S are the PV current and Reverse Saturation current of the solar cell further V T being the Thermal voltage while V is the output voltage of the module of the solar PV cells connected in series, R s is the equivalent series and parallel resistance of the cell V T = k*T*a/q with 'n' series-connected cells while 'k' is the Boltzmann constant (1.3806503×10ˆ(−23) J/K), "T" is the p-n junction temperature and "a" is ideality constant of Diode. 'q' is the charge (of an electron) [1.60217646×10-19 C].

C. DC-DC BOOST CONVERTER
In below shown Fig. 4 DC-DC Boost Converter is presented, which is to be connected between the Solar PV Array and EV as shown in Fig. 2.
The voltage equation of the Ideal converter is given by; In the practical converter, after taking losses of MOSFET and Diode into account. Output voltage is derived as; For EV, it can be derived as; where D is the Duty Cycle.

III. CONTROL SCHEME
In a single sensor-based EV charging adapter (SS-EVCA), the first objective is maximum power extraction from solar PV array. The second objective is charging current monitoring and keeping in a safe region. In the SS-EVCA scheme, the maximum solar panel power is regulated by maximizing the load transfer power. In Fig. 5, η nm is open circuit cell voltage, R 1 represents electrolyte resistance. R O represents the charge transfer resistance, that models the voltage drop over the electrodeelectrolyte interface due to a load. C O represents double-layer capacitance, that models the effect of charges building up in the electrolyte at the electrode surface. Battery current and voltage are denoted as i(t) and V t (t). Here, during charging, the voltage relationship is as follows.
The current equations are derived as, By using (12) and (8), the V t (t) is derived as, The charging power is derived as, In SS-EVCA, a DC-DC boost converter has been used, so i(t) is derived as, In EV, the load is a lithium-ion battery, so the charging power of the lithium-ion battery is derived as follows. Where, D is the duty cycle of the DC-DC converter. λ is an initial time constant of the capacitor. I pv and V pv are PV current and voltage, respectively From (19), it is clear that the EV charging power is the function of I PV and D. Therefore, the sensed I PV and calculated D values are enough to complete the task in the maximum power extraction process. Here to calculate the optimal values of D, a Deterministic Optimization" (DO) algorithm has been used, where for adaptive behavior a correction factor ω has been used. To generate the value of ω, a single input fuzzy logic (SIFL) controller is used.
Finally charging power is derived as, The objective function is derived as, The working objective of SIFL controller is to generate the values of ω according to the change in I PV . The equation  of change pattern (Ch) is derived as, As inputs, errors and changes in errors are used in the conventional fuzzy controller. Because of these two inputs, the fuzzy rule table takes a large space on two dimensions plane. However, in the proposed SIFL, the rule table takes small space on one dimension plane. Here, as an input, the diagonal distance of rate of change (Ch c ) and change (Ch) coordinate from the origin is used. The diagonal distance (η) calculation and pictorial representation is shown in Fig. 6(a). The minimization of η is the key objective of SIFL. The η is derived as, To fulfil the objective of SIFL, it follows a particular set of rules for the input member function and output member function, and according to the control surface, SIFL decides an optimal action. The input and output membership functions with control surface are shown in Fig. 6(b) & Fig. 6(c), respectively. Moreover, the input and output rule tables for SIFL are given in Table 1 Here, P Charge (t) is the function of D. Therefore, the calculation of D is very important. The operating behavior of the system depends on D and I PV . By using these two parameters, I ref is calculated, which is derived as, When calculated I ref is greater than EV battery nominal charging current rating (I Bnom ), then solar PV panel will operate under protection mode. It means the available solar PV panel power is more than the required power during charging. Moreover, when calculated I ref is less than the EV battery nominal charging current rating (I Bnom ), then solar PV panel will operate under MPPT (Maximum Power Point Tracking) mode. It means that available solar PV panel power is less than or equal to the required power during charging.
where, P PVav is available PV power, P EVr is the power safety rating of EV.
Here, based upon the value of I ref , EV Modes are selected, which are as MPPT mode or protection mode, as shown in (23) and (24). Current is continuously monitored and therefore, accordingly I ref is changed to get the desired operating current.

A. PROTECTION MODE
This condition arises when Eqn. (23) will be satisfied. It is basically, the current I ref is higher than the rated EV charging current. Therefore, in protection mode current is brought down to the rated current. In this situation, by using "Deterministic Optimization" (DO) algorithm, the optimal D(t) is generated using ω. In DO algorithm, P Charge (t), and I PV (t) are used for decision making, and finally, D(t+1) is regulated accordingly.
Equation (25) can be written as; The pseudo-code DO for protection mode of operation is given as follows,

B. MPPT OPERATION
This condition arises when Eqn. (24) will be satisfied. It is basically, the current I ref is lower than the rated EV charging current. Therefore, in protection mode current is brought up to the MPP current. In this situation, by using DO algorithm, the optimal D is generated. DO algorithm related equations and condition for D generation is as follows.
Equation (27) can be written as; The pseudo-code DO for MPPT mode of operation is given as follows, In SS-EVCA controller, using the information of I PV (t), the optimal value of D(t+1) is calculated. The SS-EVCA controller diagram is shown in Fig. 7.

IV. RESULTS AND DISCUSSIONS
For the purpose of testing the proposed approach, a hardware prototype is created, which is shown in Fig. 8. The equipment used for it is Auxiliary power supply (HTC Instruments DC 3005 DC 5A Linear Power Supply), two PLECS -RT Platform (RT Box 1 -Plexim, Processor. Hardware experiments were carried out on three algorithms in particular: Perturb & Observe (P&O) [12], Modified Perturb & Observe (MP&O) [11], and proposed SIFL-DO. Further tests are carried out on these three algorithms.
Furthermore, the solar irradiance range was varied widely and classified into two main categories, Case 1 and Case 2, with comparisons made in both cases on the basis of the characteristics curve developed during the experiments primarily for various Output parameters of the Solar PV array and thus based on the control algorithm and EV Battery charging characteristics of power that was fed into the EV.

A. CASE-1: SOLAR IRRADIANCE VARIATION IN STEP MANNER
In case-1, solar irradiance has been varied in a stepwise manner as shown in Fig. 9 for a period of 5 secs, i.e., from 0.5 secs to 5.5 secs and varies from initially 1000W/m 2 to 500W/m 2 to 900W/m 2 to 600W/m 2 to 800W/m 2 and finally to 700W/m 2 while the change in solar irradiance is in the stepped manner.
Here, output waveforms of solar PV are shown in Fig. 10, where Fig. 10(a) is for P&O [12], followed by Fig. 10(b) is for MP&O [11] and Fig. 10(c) is for our proposed approach SIFL-DO. In Fig. 10(a) and Fig. 10(b), it can be clearly seen that in the steady-state condition, the significantly high number of fluctuations are in PV voltage (V PV ), PV current (I PV ) and PV power (P PV ). This is happening because of the oscillating duty cycle generated by P&O and MP&O. The waveforms of the duty cycle obtained by P&O and MP&O algorithm are also shown in Fig. 10(a) and Fig. 10(b), respectively.
Moreover, during dynamic change conditions, time taken to transition from one mode of operation to another is clearly taking more time in the case of P&O [12] and MP&O [11] due to their rigid and constant nature of step size. While comparing Fig. 10(a), Fig. 10(b) that is P&O and MP&O, respectively, with Fig. 10(c) that is our proposed method SIFL-DO, these unwanted fluctuations are removed significantly in steadystate conditions, and in dynamic change conditions, transition time is much faster which can be clearly noticed between 0.5s to 2.5s. Moreover, in steady-state conditions this fluctuation becomes complete negligible.
The waveforms of EV battery charging are shown in Fig. 11, where Fig. 11(a), Fig. 11(b), and Fig. 11(c) depict the performance of P&O [12], MP&O [11] and proposed SIFL-DO algorithm. In Fig. 11(a) and Fig. 11(b), it can be clearly seen that in steady-state conditions, oscillations are in EV battery charging voltage (V EV ), EV battery charging current (I EV ) and EV battery charging power (P EV ). This is happening because of oscillations in V PV , I PV , and P PV , because of P&O and MP&O. Moreover, during dynamic change conditions, the time taken to transition from one power level to another, it is taking longer time, which disturbs the charging process and reduces the efficiency of the system. However, Fig. 11(c) shows that because of our proposed method SIFL-DO, the oscillations in the steady-state condition is negligible, and in dynamic change condition, transition time is much faster, which enhances the charging process and efficiency of the system.

B. CASE-2: SOLAR IRRADIANCE VARIATION IN RAMP MANNER
In case-2, solar irradiance has been varied in ramp wise manner as shown in Fig. 12 for a time period of 5s, i.e., from 0.5s to 5.5s and varies from initially 1000W/m 2 to 400W/m 2 to 900W/m 2 to 500W/m 2 to 800W/m 2 and finally to 700W/m 2 while the change in solar irradiance is in a ramp manner.
In this condition, the obtained output waveforms of solar PV are shown in Fig. 13, where Fig. 13(a) is for P&O, followed by Fig. 13(b) is for MP&O and Fig. 13(c) is for our proposed approach SIFL-DO. In Fig. 13(a) and Fig. 13(b), it can be clearly seen that in steady-state condition, significantly huge fluctuations are in V PV , I PV and P PV . This is happening because of the oscillating duty cycle generated by P&O and MP&O. The waveforms of D obtained by P&O and MP&O algorithm are also shown in Fig. 13(a) and Fig. 13(b), respectively. Moreover, during dynamic change condition, during shifting from one mode of operation to another, deviation on the tracking track is, in the case of P&O and MP&O, because of their rigid and constant nature of step size. While comparing Fig. 13(a), Fig. 13(b) that is P&O and MP&O, respectively with Fig. 13(c) that is our proposed method SIFL-DO, these unwanted fluctuations are removed significantly in steadystate conditions, and in dynamic change conditions, it is following the change with the slop of ramp. Moreover, during tracking, the fluctuation and deviations are completely negligible. The waveforms of EV battery charging are shown in Fig. 14, where Fig. 14(a), Fig. 14(b), and Fig. 14(c) depict the performance of P&O [12], MP&O [11] and proposed SIFL-DO algorithm. In Fig. 14(a) and Fig. 14(b), it can be clearly seen that in steady-state condition, oscillations are in V EV , I EV and P EV . This is happening because of oscillations in V PV , I PV , and P PV , because of P&O and MP&O. Moreover, during dynamic change condition, the deviation on the tracking track, during transition from one power level to another is clearly visible, which disturbs charging process and reduces the efficiency of the system. However, Fig. 14(c) shows that because of our proposed method SIFL-DO, the oscillations in the steady-state condition is negligible, and in the dynamic change condition, deviation on the tracking track is also negligible, which enhances the charging process and efficiency of the system.

C. CASE-3: EUROPEAN STANDARD EN50530
To conform the industrial suitability of the proposed SIFL-DO algorithm, it has been tested on European Standard (ES) EN50530. Testing is carried out, and efficiency is estimated for both steady-state and dynamic conditions.

1) STATIC EFFICIENCY
The static efficiency calculation procedure in ES En50530 is given in two ways: Firstly, the European efficiency (α Eu ) which is governed by (29) and secondly, the Californian efficiency (α Cl ), which is given by the (30) where % indicates about solar irradiance w.r.t. the base value (1000W/m 2 ), and α is the corresponding efficiency. The test results for both the cases are shown in Fig. 15 and Fig. 16, respectively.

2) DYNAMIC EFFICIENCY
In EN50530 standard, the test is performed in two types of irradiances change condition. Low to Medium irradiances change condition, where it varies in between 100 W/m 2 to 500 W/m 2 , and another is Medium to High irradiances change condition, where it varies in between 300 W/m 2 to 1000 W/m 2 . In both irradiance band, 2 types of irradiances change slops are considered, which has given in Table 2, and efficiency calculation is given in Eqn. (31). Moreover, the obtained waveforms are shown in Fig. 17.
The equation of dynamic efficiency calculation is described as, Where € is a total number of sections of every insolation band. From (31) and Fig. 17, the α dy = 99.22% during Low to Medium irradiances change conditions, and during Medium to High irradiances change conditions, the α dy = 99.30%. The average α dy = 99.26%. The results of all techniques are summarized in Table 3. Here, this obtained dynamic efficiency is greater than 99%. Therefore, according to the EN50530 standard, the proposed SIFL-DO is suitable for industrial application.

V. CONCLUSION
In this paper, a novel charging adapter topology has been proposed, which is based on a single current sensor. Low cost, fast response and high sensitivity to dynamic change, these behaviors of the current sensor make it an accurate and economical charger. Moreover, a novel Single Input Fuzzy Logic tuned Deterministic Optimization (SIFL-DO) algorithm has been proposed to accomplish MPPT operation and battery charging management. SIFL-DO algorithm has very good condition estimation and decision-making capability, which performs MPPT very accurately, monitors the charging process, and keeps it safe in adverse conditions. The capability of developed charging adapter with SIFL-DO algorithm has been evaluated on Hardware prototype, as well as comparative studies have been performed w.r.t. state-ofthe-art techniques, where developed charging scheme shows the dominancy over other techniques. The only limitation of this technique is required comparatively slightly bigger processer to execute this control logics. But, on industrial level, the use of this type of processer is very common. To prove the industrial suitability, the developed technology has been tested on European Standard (ES) EN50530, where it has been successfully satisfied the Standard charitarian. Therefore, the developed technology is an efficient and economical solution to EV charging in remote location using pillar top PV panels.