A method of determining the phase shift angle in converters

Abstract

The invention relates to a method of determining the phase shift angle for asymmetric interleaving in DC-DC converters using artificial neural networks. In particular, the invention relates to a method that reduces the fluctuations in the converter circuits using the interleaving method in multiphase DC-DC converters operating under asymmetric operating conditions such as different input voltage and input current.

Description

[0001] A METHOD OF DETERMINING THE PHASE SHIFT ANGLE IN CONVERTERS

[0002] Technical Field

[0003] The invention relates to a method of determining the phase shift angle for asymmetric interleaving in DC-DC converters using artificial neural networks.

[0004] In particular, the invention relates to a method that reduces the fluctuations in the converter circuits using the interleaving method in multiphase DC-DC converters operating under asymmetric operating conditions such as different input voltage and input current.

[0005] State of the Art

[0006] Multiphase DC-DC converters are commonly used in a variety of applications that require energy conversion and power management. In particular, they play an important role in solar energy systems. Solar panels are a naturally variable and fluctuating energy source, which means constantly changing input voltages and currents. Multiphase DC-DC converters are used to provide a stable and efficient power conversion despite these variable conditions. These converters convert the energy from the solar panels into a constant output voltage, making it compatible with energy storage systems or the grid. In other words, in photovoltaic solar energy systems, DC-DC boost converters are used for power conversion by connecting between solar panels and the grid-connected DC-AC converter for maximum power point monitoring. Generally, a DC-DC boost converter is used for each solar panel group in these systems. When this system is considered as DC-DC converters, it can be considered as a multiphase DC-DC converter system with independent inputs and parallelized outputs. The inputs of the converters connected to the independent panel groups cause asymmetric operation. Asymmetric interleaving to be applied in the converter system can minimize the current fluctuation in the common capacitor used in the output and thus the life of the used capacitor can be extended.

[0007] The use of multiphase DC-DC converters in solar energy systems reduces energy losses and improves system performance. In addition, these converters work with high efficiency, allowing solar energy to be used more effectively, thus providing a more sustainable and economical energy management. In addition, the reliability of the system can be increased.

[0008] Current / voltage fluctuations can be minimized by using the interleaving method in the common capacitors at the input or output of the multiphase DC-DC converters where the phases operate under asymmetric conditions. In line with this purpose, the operating conditions of the phases can be detected through the circuit and the waveform of the fluctuation on the capacitor can be obtained analytically and the optimum phase angle can be calculated directly by analytical calculation [1], However, the calculation of trigonometric functions in such methods increases the computational load on digital microcontrollers used for control purposes in DC-DC converters. Instead of direct calculation, the optimum phase shift angle can be determined by using search tables [2], However, in such methods, a large memory must be used to record the searchable tables on microcontrollers. Alternatively, the optimum phase angle can be determined by processing the operating conditions of the phases in real time in the optimization algorithms [3], However, since such algorithms work iteratively, there is a computational load on the digital microcontroller similar to direct calculation methods. In order to apply the interleaving method in multiphase DC-DC converters operating under asymmetric conditions, the optimum phase angle must be calculated instantly. In this respect, the method to be used should provide convenience in terms of account load and memory requirement in order to be realized in digital microcontrollers with limited resources.

[0009] In the current state, there are various patents and utility model applications related to the subject. One of them is the application with the number "US6154090 A". The application relates to the active filter method and the active filter to reduce the residual fluctuation of the current drawn by a load from a network. By actively filtering the upper harmonics of the input current using power factor corrector circuits, a method of reducing the feedback effect on the timely flow of an input current drawn by a load from a network via a rectifier and a plurality of boost converters is disclosed. The method comprises the process steps of dividing a current path of the input current between the rectifier and the load into a plurality of parallel channels, providing a booster converter comprising a booster shock and a booster diode in each channel.

[0010] Application numbered "US6069801A" in the state of the art relates to a method and system that provides power factor correction in switching power conversion. A power converter for receiving power from a source and transmitting power to a load includes the conversion circuit, the load-dependent energy storage circuit, which includes a first side connected to the source and a second side isolated from the first side and connected to the load. The conversion circuit may be adapted to transmit a DC output voltage to the load, and the power converter may further comprise a controller that regulates switching on and off to maintain the DC output voltage at a predetermined value. Application "US6043997A" in the state of the art relates to power conversion, a three-phase boost converter with a primary and an auxiliary stage and a method for reducing the total harmonic distortion at the input of the boost converter. The first, second and third auxiliary reinforcement inductors connected to the corresponding phases of the input include an auxiliary rectifier connected to the first, second and third auxiliary reinforcement inductors and an auxiliary boost switch converter placed between the said auxiliary rectifier and the said output, allowing to induce the corresponding phase currents through the first, second and third auxiliary boost inductors, thus reducing the total harmonic distortion of the input current at the said input of the said three-phase boost.

[0011] In order to reduce fluctuations in multiphase DC-DC converters in the current system, it is generally necessary to control each phase separately and to ensure interphase synchronization. However, this approach is insufficient, especially under asymmetric operating conditions such as different input voltages and currents. Passive filters and phase shift techniques used in traditional methods have difficulty in sufficiently suppressing high-frequency fluctuations, which negatively affects the overall performance of the converter. In addition, these methods often result in more complex circuit designs and higher costs. Therefore, the limitations and inadequacies of traditional methods necessitate the development of more effective and efficient solutions.

[0012] As a result, there is a need for a method that provides the reduction of fluctuations in converter circuits due to the above-mentioned and the inadequacy of existing solutions.

[0013] Brief Description and Objects of the Invention

[0014] The invention relates to a method that reduces the fluctuations in the converter circuits using the interleaving method in multiphase DC-DC converters operating under asymmetric operating conditions such as different input voltage and input current. In traditional multiphase DC-DC converters, the converters in the phases operate under symmetric input voltage / input current. In these converters, phase shifting is applied to the switching signals applied to the phases to minimize the fluctuations using the interleaving method.

[0015] The object of the invention is to reduce the fluctuations in the converters. The interleaving method significantly reduces the total fluctuation by ensuring that the phases operate in a synchronous manner in multiphase systems. This situation allows the converters to operate more stably and efficiently, especially in systems operating under different voltage and current conditions. This method both increases performance and minimizes energy losses in energy conversion processes.

[0016] Another object of the invention is to reduce the thermal stress that may occur on the circuit elements. The invention allows to reduce the thermal stress that may occur on the circuit elements by reducing the fluctuations in the converter circuits to a minimum.

[0017] In the method developed with the invention, the input voltages and currents of the DC-DC converter phases are defined as the input set, and the optimum phase shift angle is defined as the target output. Then, an artificial neural network suitable for this structure was designed. Thus, the optimum phase angle can be obtained instantly according to the voltage and current values perceived through the circuit, the interleaving method can be used and the fluctuations on the circuit elements can be minimized.

[0018] The method developed with the invention does not need trigonometric functions in calculation methods and can determine the optimum angle in a single calculation unlike iterative methods. In this respect, it provides a calculation advantage.

[0019] The method developed with the invention also has much less memory requirement compared to searchable table methods. In this respect, it allows the application of asymmetric interleaving in systems with limited resources.

[0020] Figures

[0021] Figure 1 : A two-phase DC-DC view.

[0022] Figure 2: A view of the current fluctuation of the output common capacitor current according to the phase shift angle to be applied to the switching signals of the phases.

[0023] Figure 3: A view of a two-phase DC-DC boost converter.

[0024] Figure 4: A view of the basic waveforms of the operation of the two-phase converter.

[0025] Figure 5: A view of the change of the average square root value of the common output capacitor current according to the phase shift angle.

[0026] Figure 6: A view of a converter system in which the phase angle is applied to the phases in real time.

[0027] Figure 7: A view of the block diagram to minimize the fluctuations on the circuit elements by determining the optimum phase angle using the artificial neural network and applying it to the circuit in order to apply the interleaving method in multiphase DC-DC converters operating under asymmetric conditions.

[0028] Figure 8: A view of the multilayer sensor. References

[0029] In order to better explain the method developed by this invention, the parts and elements in the figures are numbered and the corresponding of each number is given below:

[0030] 1. Phase 1

[0031] 2. Phase 2

[0032] 3. Phase 1 switching signal

[0033] 4. Phase 2 switching signal

[0034] 5. Digital microcontroller

[0035] 6. Artificial neural network

[0036] 7. Phase 1 input voltage-current signals

[0037] 8. Phase 2 input voltage-current signals

[0038] 9. Phase 2 switching signal with 0optapplied

[0039] Detailed Description of the Invention

[0040] The invention relates to a method of determining the phase shift angle for asymmetric interleaving in DC-DC (direct current-direct current) converters using artificial neural networks. Artificial neural networks (6) can be used to define the relationship between a certain set of inputs and the output corresponding to this set of inputs and to determine the target output by processing a new set of inputs.

[0041] In multiphase DC-DC converters, when the converters in the phases are operated under asymmetric conditions such as different input voltage / input current, the optimum phase angle is not predefined and varies depending on the operating conditions of the phases. In this case, in order to reduce the fluctuations in the circuit, the optimum phase shift angle must be determined dynamically according to the operating conditions of the phases. In this invention, a method for determining the optimum phase shift angle using artificial neural networks (6) has been developed to minimize fluctuations with the interleaving method in multiphase DC- DC converters operating under asymmetric conditions.

[0042] In Figure 1, a two-phase DC-DC is shown, there is a DC-DC converter in each phase. The common output voltage of the converter phases is shown as Vo, and the common output current is shown as Io. The common output capacitor Cois shown with the equivalent series resistance RCO_ESR and the current of this capacitor is shown with Ic0. In the interleaving method, phase shifting is applied to the switching signals applied to these converters and thus the current fluctuations in the common inputs and outputs of the converters are minimized. For the interleaving method, the phase shift angle applied to the switching signal (4) of Phase 2 compared to the Phase 1 switching signal (3) is 0. In multiphase DC-DC converters, when each phase has the same voltage / current, Vin, Ln, operation conditions, that is, in the case of symmetric operation, the optimum phase shift angle required for the interleaving method is obtained as 3607(Number of Phases). The optimum phase shift angle is the angle value that allows the minimum reduction of input and output current and voltage fluctuations. As shown in Figure 2, the current fluctuation of the output common capacitor current can be minimized according to the phase shift angle to be applied to the switching signals of the phases. The average square root value of the capacitor current was used to scale the current fluctuation. The mean square root value of the capacitor current is expressed by ICO,RMS. The optimum phase shift angle value for the two-phase structure is 180°.

[0043] In multiphase DC-DC converters, when the operating conditions are different between the phases, that is, when the phases are operated in asymmetric conditions, the optimum phase shift angle value for interleaving varies according to the operating conditions and is not predefined. An example two-phase DC-DC boost converter is given in Figure 3. For the boost converter in Phase 1 (1), inductor Li, semiconductor switch Qi and diode DIi were used. The converter in Phase 2 (2) consists of inductor L2, semiconductor switch Q2 and diode DI2 circuit elements. The common output voltage of the converters is shown as Vo, and the common output current is shown as Io. The common output capacitor Cois shown with the equivalent series resistance RCO_ESR and the current of this capacitor is shown with Ic0.

[0044] In order to ensure that the input current-input voltage values of the converters are equal to the target values during operation, closed-loop control is required, and pulse width modulation is used to achieve this. While the duty cycle of the Phase 1 switching signal (3) is Di, the duty cycle of the Phase 2 (2) is D2. Di and D2 are independent of each other and are determined by the operating conditions of the phases. For the interleaving method, the phase shift angle applied to the switching signal (4) of Phase 2 compared to the Phase 1 switching signal (3) is 0. In Figure 4, the basic waveforms of the operation of this two-phase converter are given. The switching signal of Qi is shown as Qi:Di, the switching signal of Q2 is shown as Q2:D2, the diode current of DIi is shown as IIDI, the diode current of DI2 is shown as LD2, the output current is shown as Ioand the common output capacitor current is shown as Ico, respectively. For easier demonstration, a= 0 / 360° was used instead of 0. As shown, the waveform of Ico depends on the value of 0. In this converter, the input voltage of the phases Vini, Vin2 and / or the input currents lini, Iin2 may be different from each other. The change of the average square root value of the common output capacitor current according to the phase shift angle applied for an exemplary operation point of this converter (Vini=40V, Vm2=20V, Iini=3A, Iin2=3 A and LI=L2=220UH, C0=47uH) is given in Figure 5. As can be seen, the optimum phase angle cannot be determined by using 360° / (Number of Phases) in this converter. The optimum angle value for this working condition of the asymmetric operation is 120°. The optimum phase angle is indicated by 0opt. Each operation condition leads to a unique optimum phase angle. In this case, when the entire working range of the converter phases is considered, an infinite number of optimum phase angles emerge.

[0045] In order to apply interleaving under asymmetric conditions, the optimum phase angle must be calculated in real time according to the operating conditions of the phases and this phase angle must be applied to the phases in real time. Such a converter system is shown in Figure 6. The calculation of the optimum phase angle can be made on a digital microcontroller (5) that generates the switching signals of the multiphase DC-DC converter. Although there are asymmetric conditions, the current fluctuation of the common output capacitor can be minimized thanks to interleaving and thus the power loss on the equivalent series resistance RCO_ESR of the capacitor given in Equation-1 can be minimized in PC0_ESR. In this way, the life of the capacitor can be prolonged since the operating temperature of the capacitor will be reduced. (Equation- 1)

[0046] An artificial neural network (6) can be used to determine the optimum phase angle required to apply interleaving under asymmetric conditions. The artificial neural network (6) is formed by means of a computer. Artificial neural networks (6) can be used to determine the relationship between a certain set of inputs and the output corresponding to this set of inputs and to determine the target output by processing a new set of inputs. Determining the relationship between inputs and output is called regression, and calculating the output by evaluating random inputs is called feed forward.

[0047] Within the scope of the invention, a method has been developed to minimize the fluctuations on the circuit elements by determining the optimum phase angle using the artificial neural network (6) and applying it to the circuit in order to apply the interleaving method in multiphase DC-DC converters operating under asymmetric conditions. The block diagram of the method is given in Figure 7. In the developed method, the input voltages, and currents of the phases of the DC-DC converter are defined as the input set of the artificial neural network (6), and the optimum phase shift angle is defined as the targeted output of the artificial neural network (6). The input voltages and currents of the phases are detected by the digital microcontroller (5). The artificial neural network (6) is encoded on the digital microcontroller (5) and the feed forward process is performed here. As a result of the feed forward, the optimum phase shift angle is obtained. This optimum phase shift angle is used during the production of the switching signals and is applied to the converter circuit by means of a digital microcontroller (5). Thus, fluctuations are minimized.

[0048] As shown in Figure 8, a multi-layer sensor consisting of 4 input field input layers, 2 intermediate layers containing 8 and 4 nodes and a single layer that gives the optimum phase angle as output was used for the artificial neural network (6). This structure can be briefly defined as 4x8x4xl. The artificial neural network (6) calculates the optimum phase shift angle corresponding to that operation point by using the input voltages and input currents of the DC-DC converters in the phases instantly with the feed forward. In the artificial neural network (6), each layer calculates the weighted sum of the inputs as given in Equation 2. (Equation-2)

[0049] Sj is the output vector of a layer and Xi is the input vector of that layer. Wji and Bjare the weight matrix and deviation matrix specific to the artificial neural network (6). i is the number of inputs in the input vector of the layer, and j is the number of nodes in the layer. A nonlinear activation function is used when calculating the output values of the nodes in the intermediate layers. The corrected linear unit is used for this function and this function is defined in Equation 3. There is no activation function in the output layer. f(s) = !?IGX (0, s) (Equation-3)

[0050] When the structure of the artificial neural network (6) is considered, the number of parameters (elements in the Wji and Bj matrices) that should be kept in memory on the digital microcontroller (5) is 81. When 32-bit memory is used for each parameter for recording the artificial neural network (6) in the digital microcontroller (5), the memory required is 324- Byte. Before leaving the scope of the invention, changes can be made to the artificial neural network (6), or other variations can be used. What is decisive here is how accurately the artificial neural network (6) can calculate the optimum phase angle corresponding to any operating condition. The 4x8x4xl structure shown within the scope of the invention may vary according to the application, and the number of nodes or layers can be adjusted according to the application. If the method proposed within the scope of the invention is transferred for two-phase DC-DC boost converter, more than two DC-DC boost converters or different converter topologies can be adapted. For this purpose, different artificial neural network (6) structures may need to be used. Within the scope of the invention, it is shown that the common output capacitor surge current of the two-phase DC-DC converter whose inputs are shared with independent outputs is reduced to a minimum level. By using the proposed invention, the current or voltage fluctuations of the capacitors or inductors used in the circuit in different applications can be minimized.

[0051] In the converter circuits, the capacitors are used to filter the current / voltage fluctuations caused by the semiconductor switches. Current fluctuations can be minimized by using the interleaving technique in the common capacitors at the input or output of the multiphase DC- DC converters operating under asymmetric conditions. As the current fluctuation level increases, the heat losses on the equivalent series resistors of the capacitors increase and this leads to a decrease in the life of the capacitors. By using the method described in the invention, the fluctuations filtered by the capacitors are reduced. Thus, the life of the capacitors can be extended.

[0052] The developed method estimates the optimum phase shift angle at 28.4ps and is updated every lOOps to obtain fluctuation minimization.

[0053] References

[0054] [1] M. Schuck and R. C. Pilawa-Podgurski, “Ripple minimization through harmonic elimination in asymmetric interleaved multiphase dc-dc converters,” IEEE Transactions on Power Electronics, vol. 30, pp. 7202-7214, 2015.

[0055] [2] M. Schuck, A. D. Ho, and R. C. Pilawa-Podgurski, “Asymmetric interleaving in low- voltage emos power management with multiple supply rails,” IEEE Transactions on Power Electronics, vol. 32, pp. 715-722, 1 2017.

[0056] [3] J.Poon, B. B. Johnson, S. V. Dhople, and S. R. Sanders, “Minimum distortion point tracking,” IEEE Transactions on Power Electronics, vol. 35, no. 10, pp. 11 013-11 025, 2020.

Claims

CLAIMS1. A method for determining the phase shift angle in converters, characterized incomprising steps of;• Measuring the input voltages (Vo) and input currents (Io) of the converter phases with the digital microcontroller (5),• Defining the input voltages (Vo) and currents (Io) as the input set,• Defining the optimum phase shift angle as the target output,• Creating the artificial neural network (6) with a computer according to the input set and target output,• Calculating the optimum phase angle by means of a digital microcontroller (5) that generates the switching signals of the converter,• Determining the optimum phase angle using an artificial neural network (6) encoded on a digital microcontroller (5),• Applying the optimum phase angle to the converter circuit by means of a digital microcontroller (5).

2. The method for determining the phase shift angle according to claim 1, wherein the artificial neural network (6) comprising an input layer with at least four input field, two intermediate layers where one layer containing at least eight and other layer contains four nodes and at least one layer that outpusts the optimal phase angle.

3. The method for determining the phase shift angle according to claim 2, wherein an each layer in the artificial neural network (6) calculates a weighted sum of the inputs using the formula

Images