Design of Landsman Converter with AI based RBFNN Control Strategy for PV Systems

Abstract

Abstract:In this growing industry of electric vehicles (EV), it is important to come up with innovative approaches to the existing problems, such as increasing the efficiency and environmental safety of EV charging systems. It is also seen in this paper a novel auxiliary Landsman converter design for EV charger applications which incorporates Artificial Intelligence for effective Power Factor Correction (PFC). In addition to this advancement, the use of AI within the control strategy of the converter enhances the operational performance and optimization, through maintaining a high power factor while reducing harmonic distortion and improving energy efficiency in charging. In the work presented herein, a complex virtual model of the Landsman converter based on MATLAB/Simulink is developed where AI is used for real-time variation of the converter operation. Such adaptability makes the system effective in correcting power factors and recovering efficient usage even with a variety of load and input voltage levels. The results from the simulations support the adoption of AI integrated into power electronics systems for charging technologies in EV which leads to improved charging systems which are more reliable and smarter.

Specification

Description:2. SYSTEM DESCRIPTION
A Landsman DC-DC converter, which is a high voltage gain. The amount of electricity that can be generated by a SPV system mostly depends on sunlight levels and the temperature of the surrounding environment. Using an effective MPPT controller based on a RBFNN, it is possible to guarantee the greatest amount of energy extraction in various circumstances. The performance of this controller aims to maximize the amount of electricity generated by the solar panels, regardless of variations in the surrounding environment. An illustration of the proposed high gain converter with SPV as an input that is connected to the grid can be seen in Figure 1.

Figure 1: Suggested system with a three-phase grid connection
For the purpose of determining the greatest electricity Point (MPP), which is the operational voltage at which the solar panel is able to extract higher voltage , the controller monitors both the quantities voltage and current. With this controller , PWM generator generates pulses that modify the duty cycle of the proposed converter. This ensures that the output voltage is steady and under control, without any fluctuations. Once this regulated output from the Landsman converter has been obtained, it is then provided to a three-phase Voltage Source Inverter (VSI), which is linked to the three-phase grid by means of an LC filter. In order to provide the grid with a clean and distortion-free AC supply, the inverter is responsible for converting the DC power into AC, and the LC filter is responsible for minimizing harmonics.
With the help of a grid-connected inverter and a proportional-integral (PI) controller, grid voltage synchronization may be accomplished. It is possible to detect mistakes by comparing the reference active power and the actual active power, as well as the reference reactive power and the actual reactive power, respectively. These variations in power are handled by the PI controller. The process begins with the generation of control signals, which are then sent to the PWM generator in order to create pulses. PWM alters the inverter's duty cycle and, in turn, controls the switching operations of the inverter. The inverter's output voltage is reduced in Total Harmonic Distortion (THD) and the inverter is kept in correct synchronization with the grid voltage by performing this procedure.



3. PROPOSED SYSTEM MODELLING
A) PV modeling
Photovoltaic (PV) cells use their PN junction structure to transform sunlight into electrical current. A photovoltaic (PV) panel is made up of several solar cells that are linked in both series and parallel. Figure 2 shows the solar panel's basic equivalent circuitdiagram.
Figure 2: A solar panel's equivalent circuit diagram
The solar panel's output current is,
I_PV=I_ph-I_O [exp((V_PV+I_PV R_s)/(nv_t ))-1]-(V_PV+I_PV R_s)/R_sh (1)
Where,
I_Ph is the current generated by the incident light,
I_d represents the current required for the diode to attain saturation.
V_PV is the photovoltaic output voltage,
v_t is the thermal voltage of the diode,
R_s is the for series resistance
R_sh is the shunt resistance
A Landsman converter is used to elevate the voltage from the photovoltaic system, since it is insufficient to meet the load requirements.
B) Landsman converter
Because of its high voltage gain, the Landsman converter runs in Continuous Conduction Mode (CCM), which guarantees consistent performance despite changes in the amount of light that is available. You can see how the Landsman converter works from the inside out in Figure 3. Figure 4 shows the two main modes of operation of this converter.

Figure 3: Schematic of the Landsman converter
Mode 1 operation
During this mode, the diode D will become reverse biased when the input current I_PV starts to flow. This occurs when the power switch S_1 is turned on. Following that, the capacitor C_1 and the inductors L_1 and L_2 are responsible for storing the energy that is derived from the input voltage.

(a)

(b)
Figure 4: Landsman converter function (a) in Mode 1 and (b) in Mode 2
Mode 2 operation
Turning off switch S_1 in this mode causes diode D to go into a forward biased condition. The energy that was held by the inductor L_1 and the capacitor C_1 begins to be released. So, current goes via capacitor C_2 and load R.
The ripple in the peak-to-peak current of the inductor L_1may be calculated as follows:
〖∆I〗_(L_1 )=∆Φ/L_1 =1/L_1 .1/2.(∆V_(C_1 ))/2.T/2 (2)
The following is the current through capacitor C_1:
I_(C_1 )=I_(L_1 )=C_1 (∆V_(C_1 ))/(1-D)T (3)
In this particular instance, the duty cycle is represented by the symbol D, while the switching time period is indicative by the symbol T. As a result of the voltage ripple across V_C_1,
〖∆V〗_(C_1 )=I_(L_1 )/C_1 (1-D)T (4)
From equation (2), we derive the following result by replacing the value of ∆V_C_1}:
〖∆I〗_(L_1 )=1/L_1 .1/2.I_(L_1 )/〖2C〗_1 (1-D)T.T/2 (5)
〖∆I〗_(L_1 )=1/(8L_1 C_1 ).(I_(L_1 ) (1-D))/f^2 (6)
〖∆I〗_(L_1 )/I_(L_1 ) =1/(8L_1 C_1 ).((1-D))/f^2 (7)
Here, f = 1/Tthis encapsulates the frequency at which the converter switches between states. Here we can see the converter's input-output connection expressed as follows::
I_(L_1 )/I_dc =D/(1-D) (8)
Equation (6) may be modified to provide the following result by plugging in the value of I_L1 from equation (8).
L_1=(DI_dc)/(8〖f^2 C〗_1 〖∆I〗_(L_1 ) ) (9)
In this case, the current flowing out of the converter is represented as asI_dc.
The inductor L_2 value is,
L_2=(DV_dc)/(f〖∆I〗_(L_2 ) ) (10)
Here, V_dcstands for the Landsman converter's output voltage.
C) MPPT controller based on RBFNN
An MPPT controller is crucial for a PV system to optimize the efficiency of its solar panels in different weather situations and with different amounts of sunshine. Here, the MPPT controller based on RBFNN estimates the MPP using the PV panel's output voltage and current.

A feedforward neural network that uses a combination of supervised and unsupervised learning methods is called a RBFNN. There are only three layers to the RBFNN, as shown in Figure 5: the input, hidden, and output layers. The hidden layer applies a non-linear activation function, whereas the output and hidden layers both use radial basis activation functions.
Changes to the weights, activation functions, and connection pattern have a major impact on the RBFNN's performance.

Figure 5: RBFNN integrated powertrain controller
The following list provides a description of the net input and output of the input neuron.
x_i^((1) ) (n)=〖net〗_i^((1)) (11)
y_i^((1) ) (n)=f_i^((1)) [〖net〗_i^((1) ) (n) ]=〖net〗_i^((1) ) (n),i=1,2 (12)
The input layer is denoted by the string x_i^(1), whereas the hidden layer is denoted by the string y_i^(1) in this particular situation. Expressed as net_i^(1), the total contribution from the input layer to a node is a mathematical expression. During the buried layer of the Radial Basis Function Network (RBFN), every node utilizes a Gaussian function as its activation function. This function is responsible for the operation of the node. Similarly to the role that it plays in fuzzy logic systems, this Gaussian function serves as a membership function inside the network.
〖net〗_j^((2) ) (n)=-〖(X-M_j)〗^T ∑_j▒〖(X-M_j)〗 (13)
y_j^((2)) (n)=f_j^((2)) [〖net〗_j^((2)) (n) ]exp[〖net〗_j^((2) ) (n) ],j=1,2,…. (14)
When discussing this context, the standard deviation (sigma) and the mean (mu) of the Gaussian function are used. With just one node k, the output layer generates the linear control signal D..
〖net〗_k^((3))=∑_j▒〖w_j y_j^((2)) 〗 (15)
y_k^((3))=f_k^((3)) [〖net〗_k^((3) ) (n) ]= 〖net〗_k^((3) ) (n) (16)
A connective weight matrix denoted asWis used to link the output and hidden layers. The PWM generator maintains a steady and controlled output voltage free of fluctuations by modulating the duty cycle D of the Landsman converter based on the controller's output.
D) Synchronizing Grid Voltage using PI Controller
In a three-phase grid-connected photovoltaic (PV) system, the main purpose of adopting an RBFNN-based MPPT system is to optimize the power production of the PV system under various operational situations. The Landsman converter can perform DC-DC voltage conversion efficiently with high voltage gain even when operating at low duty cycle. The PI controller is responsible for controlling the operation of the inverter, which guarantees correct synchronization of the grid voltage. MATLAB simulations were used to test the performance of the proposed grid-connected photovoltaic system model. The results of these simulations show that the RBFNN-based MPPT approach provides higher tracking accuracy. In addition, it improves the performance of the photovoltaic system in both dynamic and steady states.

DETAILED DESCRIPTION
Simulating a Landsman converter with both a Proportional-Integral (PI) controller and an extended RBFNNcontroller for PFC involves a multi-step process. This simulation would typically be done in MATLAB/Simulink due to its comprehensive support for electrical systems and control engineering projects. Below is a general guide on how to approach this simulation, highlighting the key steps and considerations for each controller type.

Case1 : with PI control strategy

Fig: Grid Voltage and current
When compared to a traditional DBR-fed charger, the efficiency is better due to the 50% reduction in conduction losses in each cycle of the DBR. The input inductors' continuous conduction mode (CCM) further helps the suggested adjustment by reducing input current ripple.
, Claims:CLAIMS
1. Maximizing the quantity of power that can be produced by a grid-connected three-phase photovoltaic (PV) system under a broad variety of operating situations is the main goal of installing an RBFNN-based MPPT system.

2. The Landsman converter can efficiently convert DC to DC voltages while maintaining a high voltage gain even while running at a low duty cycle. A PI controller regulates activities and keeps the inverter running smoothly, which is responsible for keeping the grid voltage in the correct sync.

3. The proposed concept of a grid-connected photovoltaic system is tested using MATLAB simulations to assess its efficiency. These simulations show that the RBFNN-based MPPT Approach has a much better tracking accuracy.

4. This improvement improves the photovoltaic system's performance in both dynamic and steady-state modes.

Documents