Adaptive frequency adjustment online optimization efficiency control strategy based on Buck converter

Abstract

The purpose of the present invention is to provide an adaptive frequency adjustment online optimization efficiency control strategy based on Buck converter, which belongs to the field of power electronics technology. The Buck converter can be quickly adjusted to the optimal efficiency operation, improving its convergence performance. The controller automatically tracks and determines the switching frequency of the power converter to achieve the highest power conversion efficiency under various working conditions such as temperature change, component aging, component manufacturing process differences, input voltage changes, load current changes, etc. Based on how to improve the operating efficiency of the Buck converter, an adaptive step control algorithm is proposed to quickly adjust the converter to a high-efficiency operating state, effectively improving the operating efficiency of the converter under different conditions.

Description

Technical Field

[0001] The present invention belongs to the technical field of power electronics, and specifically relates to an online real-time efficiency optimization control strategy and device based on Buck converter. Background Art

[0002] In recent years, DC / DC converters have been used more and more widely in power management. Buck converter is a DC / DC circuit topology. It is widely used in mobile multimedia devices such as mobile phones, GPS, MP3, etc. due to its simple circuit topology, reliability, flexibility and other characteristics. At present, many power management chip manufacturers have launched Buck converters with different current capabilities. Although there are some differences in current capabilities and protection functions of this type of converter, the main framework structure of its circuit is basically the same. Its circuit generally includes a fully controlled switch and a diode switch. The fully controlled switch mainly controls the PWM wave to determine the output voltage value of the converter.

[0003] Because there are many kinds of losses, the efficiency of Buck converter in actual application cannot reach 100%, among which conduction loss, switching power loss and gate drive switching loss are the main ones. There is a parabolic relationship between the total loss of the switching converter and the switching frequency, so there is an optimal operating switching frequency point, at which the converter has the highest operating efficiency. Therefore, how to effectively find the optimal switching frequency of the converter is the key to the efficient operation of the converter.

[0004] How to improve the efficiency of the converter is also a problem that researchers are constantly exploring, so the present invention explores the relationship between the Buck DC converter and the switching frequency, and proposes an adaptive step-size optimization control strategy to improve the efficiency of the converter. The present invention proposes an adaptive step-size high-efficiency control method that can quickly adjust the converter to the optimal efficiency operation and achieve a significant improvement in operating efficiency. Based on how the BUCK converter can improve its operating efficiency, the present invention proposes an adaptive step-size control algorithm, which quickly adjusts the converter to a high-efficiency operating state, effectively improving the operating utilization rate of the converter under different conditions. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a frequency adaptive online optimization control algorithm. By sampling the input current and output current values ​​online in real time, using input current sampling and applying the steepest descent method to quickly adjust the frequency to the optimal switching frequency, the converter can operate with high efficiency. And when the output load changes or the input voltage changes, the optimization algorithm can be quickly started by detecting the change in the output current, and the frequency can be adjusted to the optimal efficiency operating frequency of the new state. The present invention successfully verifies the application of the algorithm to the Buck converter.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In the first aspect, the present invention proposes a real-time efficiency optimization control algorithm based on Buck converter adaptive frequency adjustment. The algorithm has two working states: a normal iteration state for efficiency optimization and an input voltage cutting and output load instantaneous switching state.

[0008] S1: The efficiency optimization iteration state means that the converter, under the same operating state, uses an adaptive frequency adjustment algorithm to find the optimal switching frequency to achieve a high-efficiency operating state of the converter.

[0009] S2: When the converter state changes, the adaptive step frequency adjustment control algorithm is quickly triggered to achieve the purpose of finding the optimal switching frequency in real time at any time.

[0010] The following first introduces the working state of normal iterative optimization efficiency, and then introduces the working state of input voltage and output load switching. The converter achieves real-time efficiency optimization by adjusting the switching frequency, and when the working state of the converter changes after stabilization, the control algorithm can be quickly triggered to run, achieving the purpose of optimizing the optimal efficiency under different operating states of the converter. The online adaptive step size algorithm has the advantages of flexible cycle, fast optimization speed, coordinated accuracy and iteration speed, and good state switching sensitivity.

[0011] S1: The efficiency optimization iteration state refers to the converter, under a certain fixed operating state, finding the optimal switching frequency through the adaptive frequency adjustment algorithm to achieve the high-efficiency operating state of the converter. The specific implementation steps are as follows:

[0012] 1. Given an initial switching frequency f of a converter sw_0 , the PWM wave generated by the initial frequency controls the fully controlled switch MOSFET. After M cycles, the input current I is sampled in The algorithm is not run in the initial cycle, and only an initial sampling operation is performed. The initial current sampling and initial frequency of the algorithm are set to the I in (k-1) and f sw (k-1), fine-tune the frequency by a constant C0, and then enter the next iteration. The current switching frequency is f sw (k) = f sw (k-1)-C0.

[0013] 2. At the current frequency f sw (k) is the switching cycle control MOSFET tube, and the input current I is sampled after M cycles. in Set to the current input current I in (k), the frequency f of the previous M cycles sw(k+1) and the current frequency f sw (k) divided by the optimal frequency range [f sw_low , f sw_high ], convert the order of magnitude to the input current I of the previous M cycles in (k-1) and the current input current I in The above sampling modules all sample the actual values ​​into the DSP through ADC conversion, calculate the results in the DSP and output the control PWM waveform.

[0014] 3. Based on the sampled input current and current switching frequency at the current moment, and the input current sampling and frequency at the previous moment, use the differential equation to calculate the gradient at the current moment The gradient is used to perform the steepest descent method to achieve the purpose of self-adaptation. The gradient function solution is shown in equations (1)-(3):

[0015] ΔI in =I in (k)-I in (k-1) (1)

[0016] Δf=f sw (k)-f sw (k-1) (2)

[0017]

[0018] 4. Compare the calculated value of the discrete gradient function with a smaller constant C1. If it is greater than the constant C1, continue to step (5). Otherwise, do not adjust the frequency. As shown in formula (4), make the next frequency equal to the current frequency and end the iteration.

[0019] f sw (k+1)=f sw (k) (4)

[0020] 5. According to the calculated Calculate the frequency adjustment step f sw_step The frequency adjustment step size is as follows, where the μ parameter in formula (1) is used to weigh the relationship between the speed and accuracy of the entire control algorithm. The μ value is between 0 and 1. The closer the μ value is to 1, the faster the control algorithm converges. The smaller the μ value is, the higher the algorithm convergence accuracy is. The δ parameter in formula (2) is based on the current gradient To adjust the size, If it is greater than a certain constant value C2, δ is taken as a larger parameter value, otherwise, δ is taken as a smaller parameter value. Step length f sw_step The solution of is shown in formula (5);

[0021]

[0022] 6. Use the calculated frequency adjustment step size to adjust the current switching frequency f sw (k) Adjust to get the switching frequency f at the next moment sw (k+1), the output is f sw (k+1) is the frequency control PWM wave. And the current input current sampling and switching frequency are assigned to I in_last and f sw_last , and then go to the second step to iterate until the termination condition of step (4) is met. At this time, the optimal switching frequency of the converter is optimized through adaptive step frequency adjustment.

[0023] S2: Converter input voltage U in and output load R o When the frequency changes, the S1 state can be triggered instantaneously for rapid adjustment. Because the operation cycle of the algorithm is greater than the cycle of the PWM wave, in order to achieve the goal of quickly triggering the frequency optimization control S1 when the converter operation state changes, whether it is synchronized with the start time of the algorithm cycle, so as to meet the real-time requirements, the following control steps are given for this operation scenario.

[0024] 1. Sample the output current value I o When the operating parameters of the converter remain unchanged, the output current is stable. When the content described in S2 occurs, the output current I o It will change, so check whether the value changes. Once it changes, enter step S22. If it does not change, continue to run step S12.

[0025] 2. Detect output current I o The change exceeds a constant value C3, and the value of C3 should be slightly larger than the fluctuation range of the output current sampling. At this time, the S1 running state is triggered to synchronize the initial state change of the algorithm, rather than waiting until the next algorithm cycle to start the frequency optimization. It is realized that the optimization algorithm can be quickly triggered when the converter state changes at any time. Go to step S11 for iteration.

[0026] In a second aspect, the present invention implements a device for adaptive frequency adjustment based on a Buck converter to optimize the optimal switching frequency under different converter operating conditions in real time. The device comprises:

[0027] Buck converter main circuit module: This module consists of an inductor L, a mosfet fully controlled switch, a diode, three output capacitors C1, C2, C3 and the output load in parallel at the output end, and an input power supply U inAnd three voltage stabilizing capacitors C4, C5, and C6 are connected in parallel. The efficiency optimization algorithm is optimized based on the main circuit module.

[0028] Isolation sampling module: It is mainly composed of a sampling constantan wire, an isolation chip, an operational amplifier, and multiple resistors and capacitors for filtering and amplification. It can isolate the power circuit and the signal circuit. It can sample the input current and output current more stably.

[0029] DSP28379D: This module is a digital processor used to process data sampling and algorithm implementation. Matlab was used to generate the C code of the above control algorithm, and the generated C code was burned into the DSP through CCS. After sampling the input current and output current in the DSP, the algorithm was run to output the PWM control signal wave.

[0030] Drive circuit module: The drive circuit module adds a voltage value to the control signal waveform output by the DSP, converting it into an actual control PWM wave that can drive the MOSFET tube. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is the Buck schematic diagram. The algorithm verification is based on the experiment of this circuit.

[0032] Figure 2 This is the circuit mode diagram when the Buck circuit MOSFET is turned on.

[0033] Figure 3 This is the circuit mode diagram when the Buck circuit MOSFET is turned off.

[0034] Figure 4 Flowchart of the algorithm strategy in S1 working state

[0035] Figure 5 Flowchart of the algorithm strategy in S2 working state

[0036] Figure 6 This is the real-time optimization result achieved by the algorithm based on the parameters in Table 1.

[0037] Figure 7 This is the optimal response result of the algorithm based on the conversion of the parameters in Table 1 to the parameter settings in Table 2. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the implementation modes and the accompanying drawings.

[0039] Figure 1 This is the Buck schematic. The module consists of an inductor L, a mosfet fully controlled switch, a diode, an output capacitor C2 and an output load in parallel, and an input power supply Uin In parallel with three voltage-stabilizing capacitors C1, the efficiency optimization algorithm is optimized based on the main circuit module.

[0040] Figure 2 and Figure 3 They are the working modes when the PWM wave controls the MOSFET tube to turn on and off. Figure 2 It shows that when the switch is turned on, the current flows from the power supply U in The positive electrode passes through the S1 switch, L inductor, and load R o Then return to the negative pole of the power supply. Figure 3 It shows that when the switch is turned off, the current flows through the diode D, from the inductor L, the load R o The capacitor C2 and the diode D form a current loop.

[0041] Figure 4 This is a general operation flow chart for running adaptive step frequency modulation optimization. When the main circuit parameters of the Buck circuit do not change, the Buck converter is connected to the power supply and load according to the parameters, and the two ends of the isolation sampling circuit are powered on. The C code is burned into the DSP, the DSP is connected to the driver through a line, the driver is connected to the main circuit, and the output current and input current are connected to the ADC and DSP. Then run, the DSP gives an initial frequency, and then detects the input current after M cycles, calculates the PWM wave frequency of the next algorithm cycle in the DSP, and outputs the control signal, which is driven to control the MOSFET tube, and then iterates continuously until it converges to a small area frequency range.

[0042] Given an initial switching frequency f of the converter sw_0 , the PWM wave generated by the initial frequency controls the fully controlled switch MOSFET. After M cycles, the input current I is sampled in The value of .

[0043] At the current frequency f sw (k) is the switching cycle control MOSFET tube, the upper M cycle frequency f sw (k+1) and the current frequency f sw (k) divided by the optimal frequency range [f sw_low , f sw_high ], convert the order of magnitude to the input current I of the previous M cycles in (k-1) and the current input current I in (k)Consistency.

[0044] Use the difference equation to calculate the gradient at the current moment The gradient is used to perform the steepest descent method to achieve the purpose of self-adaptation. The gradient function solution is shown in equations (1)-(3):

[0045] ΔIin =I in (k)-I in (k-1) (1)

[0046] Δf=f sw (k)-f sw (k-1) (2)

[0047]

[0048] The calculated discrete gradient function value is compared with a smaller constant C1. If it is greater than the constant C1, the process continues to step (5). Otherwise, the frequency is not adjusted. As shown in formula (4), the next frequency is made equal to the current frequency and the iteration ends.

[0049] f sw (k+1)=f sw (k) (1)

[0050] According to the calculated Calculate the frequency adjustment step f sw_step , the frequency adjustment step is shown in (5).

[0051]

[0052] Use the calculated frequency adjustment step size to adjust the current switching frequency f sw (k) Adjust to get the switching frequency f at the next moment sw (k+1), the output is f sw (k+1) is the frequency of the control PWM wave. Continue to sample the input current and iterate.

[0053] Figure 5 This is a flowchart of the optimization algorithm that is triggered immediately when the input voltage and output load of the Buck converter change. When the output load or input voltage changes, the sampled output current Iin changes by more than C3, the algorithm is triggered to start, and the optimization frequency algorithm is immediately driven at this time, regardless of any time in the algorithm cycle M. After the trigger, the input current is sampled immediately, and the PWM frequency is calculated by the algorithm in the DSP, and the PWM control signal is output. After the signal passes through the drive module, it drives the MOSFET switch, and then iterates continuously until it converges to a small area frequency range and fluctuates until the termination condition is reached.

[0054] Figure 6It is the real-time optimization result realized by the algorithm based on the parameters in the table below. The horizontal axis is the frequency in Khz, and the vertical axis is the sampled input current Iin in A. The initial frequency is adjusted to 40Khz, and its parameter settings are consistent with Table 1. The frequency optimization range is limited to 20Khz-140Khz to find the optimal frequency. It can be seen that the algorithm can quickly converge to the optimal frequency, even if it fluctuates to other areas. It can converge quickly. The red point in the figure is the real sampled input current value Iin. Because the sampling is fluctuating, in order to see its current trend more clearly, the result is obtained by unscented Kalman filtering. The result shown by the red point can be seen more clearly. It can be seen that the current extreme point corresponds to the optimal frequency. It can also be found that the frequency quickly converges to 38Khz and oscillates back and forth at 38Khz. The points around the red mark in the figure are the most dense, because the control algorithm oscillates most frequently near this point after converging to 38Khz. The point of 38Khz corresponds to the highest operating efficiency point of the circuit, completing the purpose of efficiency optimization. The adaptive step-size frequency modulation control algorithm was successfully implemented.

[0055]

[0056] Table 1

[0057] Figure 7 This is the optimal response result of the algorithm based on the conversion of the parameters in Table 1 to the parameter settings in Table 2. When the load is reduced to 4.5Ω, U in =15V DC input, output voltage U o Stable at 6.9V. When the algorithm realizes the parameter change, it immediately switches from the previous stable frequency 38Khz to the optimal frequency state, and continues to optimize the optimal operating frequency at the current moment in real time. The frequency limit is 20Khz-170Khz, and it is found that its frequency is constantly increasing. The fast trigger algorithm in S2 mode is completed. The parameter settings after switching are shown in Table 2 below

[0058]

[0059] Table 2.

Claims

1. An adaptive frequency adjustment online efficiency optimization control method based on Buck converter, characterized in that: It includes a fixed duty cycle frequency modulation algorithm, which is an adaptive step frequency modulation algorithm for optimizing efficiency and is verified based on the Buck converter; in the first aspect, an adaptive frequency adjustment real-time efficiency optimization and rapid descent control algorithm based on the Buck converter, the algorithm has two working states: a normal iteration state for optimizing efficiency, and an input voltage switching and output load instantaneous switching state; the adaptive frequency adjustment real-time efficiency optimization and rapid descent control algorithm includes: S1: The efficiency optimization iteration state means that the converter, under the same operating state, uses the adaptive frequency adjustment algorithm to find the optimal switching frequency to achieve the high-efficiency operating state of the converter. The specific implementation steps are as follows: (1) Given an initial switching frequency of the converter, a PWM wave control generated by the initial switching frequency is generated. After a certain period, the value of the input current is sampled. The algorithm is not run in the initial period, and only an operation of sampling the initial value is performed; the initial current sampling and initial frequency of the algorithm are set to the values ​​of the previous period, and the frequency is fine-tuned once to a constant as the current switching frequency, and then the next iteration is entered; (2) Full control is controlled at the current frequency. After a certain period, the sampled input current is set as the current input current. The frequency of the previous cycle and the frequency of the current moment are divided by the maximum value of the optimal frequency range, and the order of magnitude is converted to be consistent with the input current of the previous cycle and the input current of the current moment; (3) Based on the sampled input current and current switching frequency at the current moment and the input current sampling and frequency at the previous moment, the gradient at the current moment is calculated using the differential equation, and the steepest descent method is performed using the gradient to achieve the purpose of self-adaptation; (4) Compare the calculated discrete gradient function value with a smaller constant. If it is greater than the constant, continue to step (5). Otherwise, do not adjust the frequency and end the iteration. (5) calculating the adjustment step size of the frequency according to the calculated gradient; (6) Use the calculated frequency adjustment step size to adjust the current switching frequency to obtain the switching frequency at the next moment, output the control PWM wave with this frequency, and assign the current input current sampling and switching frequency to the value at the previous moment, and then go to the second step to iterate until the termination condition of step (4) is met. At this time, the optimal switching frequency optimization of the converter is completed through adaptive step size frequency adjustment; S2: When the converter state changes, the adaptive step frequency adjustment control algorithm is quickly triggered to achieve the goal of finding the optimal switching frequency in real time at any time; When the input voltage Uin and output load Ro of the converter change, the S1 state can be triggered instantaneously for rapid adjustment, including the following control steps: (1) Sampling the output current value. When the operating parameters of the converter remain unchanged, the output current is stable. When the state described in S2 occurs, the output current Io will change. Therefore, it is detected whether the value changes. If it changes, it enters step S2(2). If it does not change, it continues to run step S1(2); (2) When the output current Io changes by more than a certain constant value C3, which should be slightly larger than the fluctuation range of the output current sampling, the S1 running state is triggered to enable the initial state change of the algorithm, instead of waiting until the next algorithm cycle to start frequency optimization, and then go to step S1(1) for iteration.

2. A device for adaptive frequency adjustment based on Buck converter, characterized in that: The method comprises an adaptive frequency adjustment online efficiency optimization control strategy based on a Buck converter as claimed in claim 1, wherein the device comprises: Buck converter main circuit module: an inductor L, a MOSFET fully controlled switch, a diode, three output capacitors and output load in parallel at the output end, and input power supply and three voltage stabilizing capacitors in parallel at the input end; Isolation sampling module: It consists of a sampling constantan wire, an isolation chip, an operational amplifier, and multiple resistors and capacitors for filtering and amplification; DSP28379D: Digital processor, used to process data sampling and algorithm implementation. Matlab was used to generate C code for the adaptive frequency adjustment, real-time efficiency optimization, and rapid descent control algorithm. The generated code was burned into the DSP through CCS. After sampling the input current and output current in the DSP, the algorithm was run to output the PWM control signal wave. Drive circuit module: The drive circuit module adds a voltage value to the control signal waveform output by the DSP, converting it into an actual control PWM that can drive the MOSFET tube.

Images