LLC converter output voltage optimization method and device based on active disturbance rejection control

Abstract

The invention discloses an LLC converter output voltage optimization method and device based on active disturbance rejection control. The method comprises the following steps: inputting an observed output voltage of a resonant converter to an extended state observer; inputting the reference output voltage into a tracking differentiator, outputting a smooth voltage tracking signal and a voltage change rate estimation value by the tracking differentiator, and subtracting from the output of the extended state observer to obtain a first error signal and a second error signal; inputting the first error signal and the second error signal into a nonlinear state error feedback controller, and outputting an error integral signal by the nonlinear state error feedback controller; subtracting the error integral signal and the output of the extended state observer, and dividing the difference by a stability coefficient to obtain a switching frequency control quantity; parameters in the extended state observer and the nonlinear state error feedback controller and a stability coefficient are obtained through a multi-objective optimization algorithm. By adopting the technical scheme, the anti-disturbance capability of the LLC resonant converter is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power electronics, and in particular to a method and a device for optimizing the output voltage of an LLC converter based on auto-disturbance rejection control. Background Art

[0002] As a DC-DC converter that is easy to implement soft switching, LLC resonant converter has been widely studied due to its wide input voltage range, strong light load working ability and high power density. Compared with other DC-DC converters such as phase-shifted full-bridge converter and series resonant converter, LLC overcomes the problem of difficulty in implementing soft switching under high voltage input and light load operation, which makes it widely used in emerging industries such as data center power supply, electric vehicles, and new energy charging piles.

[0003] The PID control commonly used in industry has the advantages of good stability, convenient parameter adjustment and the ability to eliminate static errors. However, for LLC resonant converters, due to the need to deal with input voltage fluctuations and load jumps, PID control is difficult to quickly eliminate oscillations when disturbances occur. In addition, due to its passive method of eliminating errors based on negative feedback, excessive overshoot may occur, affecting system stability. ADRC is a model-free control method with superior anti-disturbance performance. The prediction of system output using a state observer can pre-calculate the system error, thereby greatly improving control accuracy and shortening control time.

[0004] However, for second-order systems such as LLC, nonlinear active disturbance rejection control requires the adjustment of 11 parameters. The adjustment process is relatively complicated and it is often difficult to obtain accurate results. It is also easy to cause the system to have weak disturbance rejection ability. Summary of the invention

[0005] Purpose of the invention: The present invention provides a method and device for optimizing the output voltage of an LLC converter based on self-anti-disturbance control, aiming to solve the technical problem of weak anti-disturbance capability of LLC resonant converter in the prior art.

[0006] Technical solution: The present invention provides an output voltage optimization method for an LLC converter based on self-disturbance rejection control, comprising: inputting the observed output voltage of the resonant converter into an extended state observer, the output of the extended state observer comprising an output voltage estimate, an output voltage change rate and a total disturbance; the resonant converter comprises an inverter network, a resonant network, a transformer primary side, a transformer secondary side and a rectifier network in a connection order, and the rectifier network outputs voltage to the outside; inputting the reference output voltage into a tracking differentiator, the tracking differentiator outputs a smoothed voltage tracking signal and a voltage change rate estimate, the smoothed voltage tracking signal and the output voltage estimate are subtracted to obtain a first error signal, and the voltage change rate estimate and the output voltage change rate are subtracted to obtain a second error signal; inputting the first error signal and the second error signal into a nonlinear state error feedback controller, and the nonlinear state error feedback controller outputs an error integral signal; the error integral signal and the total disturbance are subtracted and then divided by a stability coefficient to obtain a switching frequency control amount, and the switch tube in the inverter network is controlled based on the switching frequency control amount; the parameters in the extended state observer and the nonlinear state error feedback controller, as well as the stability coefficient, are obtained by a multi-objective optimization algorithm.

[0007] Specifically, the extended state observer adopts the following calculation formula:

[0008]

[0009] Among them, Z1, Z2 and Z3 represent the output voltage estimate, the output voltage change rate and the total disturbance respectively, Z1', Z2' and Z3' represent the differentials of Z1, Z2 and Z3 respectively, y represents the observed output voltage of the resonant converter, ε1 represents the intermediate variable, β1, β2 and β3 represent the first gain, the second gain and the third gain of the extended state observer respectively, the fal function represents the nonlinear function, δ represents the threshold of the input variable of the nonlinear function, b0 represents the stability coefficient, and u represents the switching frequency of the control quantity.

[0010] Specifically, the nonlinear function adopts the following calculation formula:

[0011]

[0012] Wherein, a represents the adjustment coefficient of the extended state observer, x represents the state variable of the extended state observer, and sign function represents the sign function.

[0013] Specifically, the nonlinear state error feedback controller adopts the following calculation formula:

[0014] u0=k1fal(e1,a1,δ)+k2fal(e2,a2,δ),

[0015] Among them, u0 represents the error integration signal, k1 and k2 represent the first proportional coefficient and the second proportional coefficient respectively, e1 and e2 represent the first error signal and the second error signal respectively, and a1 and a2 represent the first adjustment coefficient and the second adjustment coefficient respectively.

[0016] Specifically, the switching frequency control amount is input into the voltage-controlled oscillator to obtain a corresponding frequency signal, and then the frequency signal is input into the PWM signal generator to obtain a square wave signal for controlling the switch tube.

[0017] Specifically, the parameters in the extended state observer and the nonlinear state error feedback controller obtained by the multi-objective optimization algorithm include: the first gain, the second gain and the third gain of the extended state observer, and the first proportional coefficient and the second proportional coefficient of the nonlinear state error feedback controller.

[0018] Specifically, the multi-objective optimization algorithm is used to obtain the value, including: using the grey wolf algorithm to calculate the value.

[0019] Specifically, the fitness function of the grey wolf algorithm includes: calculating the deviation between the target output voltage of the resonant converter and the observed output voltage as the fitness function.

[0020] Specifically, the inverter network is a full-bridge structure, including a first bridge arm and a second bridge arm, the first bridge arm includes a first switch tube and a third switch tube, the second bridge arm includes a second switch tube and a fourth switch tube, the duty cycle of the switch tube is 50%, the first switch tube and the fourth switch tube are controlled by a first control signal, the second switch tube and the third switch tube are controlled by a second control signal, and there is a dead time between the first control signal and the second control signal.

[0021] The present invention also provides an LLC converter output voltage optimization device based on self-disturbance rejection control, comprising: an extended state observer control unit, a tracking differentiator control unit, a nonlinear state error feedback controller control unit, a voltage control unit and a parameter optimization unit, wherein: the extended state observer control unit is used to input the observed output voltage of the resonant converter into the extended state observer, and the output of the extended state observer includes an output voltage estimate, an output voltage change rate and a total disturbance; the resonant converter includes an inverter network, a resonant network, a transformer primary side, a transformer secondary side and a rectifier network in a connection order, and the rectifier network outputs a voltage to the outside; the tracking differentiator control unit is used to input the reference output voltage into the tracking differentiator, and the tracking differentiator outputs a smoothed voltage tracking value. The first error signal is obtained by subtracting the tracking signal from the estimated value of the voltage change rate, the second error signal is obtained by subtracting the estimated value of the voltage change rate from the output voltage change rate; the control unit of the nonlinear state error feedback controller is used to input the first error signal and the second error signal into the nonlinear state error feedback controller, and the nonlinear state error feedback controller outputs an error integral signal; the voltage control unit is used to subtract the error integral signal from the total disturbance and then divide it by the stability coefficient to obtain the switching frequency control amount, and control the switch tube in the inverter network based on the switching frequency control amount; the parameter optimization unit is used to obtain the parameters in the extended state observer and the nonlinear state error feedback controller, as well as the stability coefficient through a multi-objective optimization algorithm.

[0022] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: on the basis of ADAC, the variable parameters required for ADAC are obtained by combining a multi-objective optimization algorithm, and the optimal solution of the variable parameters can be obtained, thereby effectively improving the anti-disturbance capability of the LLC resonant converter. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A control block diagram of the LLC converter output voltage optimization method based on active disturbance rejection control provided by the present invention;

[0024] Figure 2 A schematic diagram of the structure of an active disturbance rejection controller based on the Grey Wolf Algorithm provided by the present invention;

[0025] FIG3( a ) is a small signal response waveform diagram of input and output quantities provided by the present invention;

[0026] FIG3( b ) is a Bode diagram comparing the transfer function after the order reduction by the frequency sweeping method provided by the present invention and the original transfer function;

[0027] Figure 4 Bode diagram of the system before and after the active disturbance rejection control compensation provided by the present invention;

[0028] FIG5( a ) is a schematic diagram of updating the position of a gray wolf provided by the present invention;

[0029] FIG5( b ) is a flow chart of an improved gray wolf optimization algorithm provided by the present invention;

[0030] FIG6( a ) is a parameter self-optimization curve diagram provided by the present invention;

[0031] FIG6( b ) is a fitness function convergence curve diagram provided by the present invention;

[0032] Figure 7 This is a comparison diagram of the output voltage simulation waveforms of the PID control, active disturbance rejection control and optimized gray wolf active disturbance rejection control provided by the present invention. DETAILED DESCRIPTION

[0033] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0034] In a specific implementation, for a second-order system such as an LLC resonant converter, nonlinear auto-disturbance rejection control requires the adjustment of 11 parameters. The adjustment process is relatively complicated and it is often difficult to obtain accurate results. Therefore, the present invention utilizes a multi-objective optimization algorithm (such as the optimized gray wolf algorithm provided by the present invention) to iteratively self-optimize the six key parameters in the nonlinear auto-disturbance rejection control, thereby avoiding the use of empirical parameter adjustment methods or negative feedback control, and finding the optimal solution within the feasible domain of the parameters, thereby reducing errors and further improving the system's anti-disturbance capability, ensuring that the output voltage approaches the target voltage (reference output voltage).

[0035] Refer to FIG. 3( a ), which is a small signal response waveform diagram of input and output quantities provided by the present invention.

[0036] In the specific implementation, the full-bridge LLC resonant converter is modeled by using the EDF extended function description method. The nonlinear terms are approximated by the basic harmonic components defined by the EDF, and the nonlinear sinusoidal quantities such as the inductor current and the capacitor voltage are expressed as the superposition of the sine component and the cosine component with time-varying amplitude. Finally, the transfer function from the switching frequency to the output voltage can be obtained as:

[0037]

[0038] Wherein s represents a complex variable in a complex plane. The 7th-order transfer function is reduced in order by using a transfer function sweep method. A small sinusoidal signal is injected near the static operating point. The switching frequency steady-state value of the full-bridge LLC resonant converter provided by the present invention is approximately 115kHz, and the output voltage steady-state value is 30V. The small signal disturbance amplitude is controlled within 3% to 5% of the steady-state value. Therefore, the injected sinusoidal signal amplitude is set to 5k. 40 points are equidistantly selected between frequencies of 500Hz and 100kHz in the logarithmic space, and finally a small signal response waveform of the input and output can be obtained.

[0039] Refer to FIG. 3( b ), which is a Bode diagram comparing the transfer function after the order reduction by the frequency sweeping method provided by the present invention with the original transfer function.

[0040] In a specific implementation, the original 7th order transfer function G is converted by retaining the dominant zero poles through the transfer function sweep frequency method. p (s) is reduced to the third-order transfer function G Iden , the expression is as follows:

[0041]

[0042] From the Bode diagram, we can see that G Iden With G p (s) It can be well matched in the low and medium frequency bands that are more concerned in controller design, but there will be some deviations in the high frequency band, and the phase lag and gain attenuation of the original function are greater. This verifies that the transfer function derived by the sweep frequency method and the transfer function calculated by the extended function description method have a high degree of match, so the controller design can be carried out according to the reduced transfer function in the future.

[0043] See also Figure 4 , which is the system Bode diagram before and after the active disturbance rejection control compensation provided by the present invention.

[0044] In the specific implementation, the open-loop system before compensation is not stable and has low DC gain, and does not have sufficient phase margin and amplitude margin. However, the closed-loop system after nonlinear self-disturbance rejection control compensation has better stability. Figure 4 It can be seen that the closed-loop system has a phase margin of 78.2°, an amplitude margin of 21.7dB, and a delay margin of 3.07e -5 s, at the crossover frequency fc=44kHz, the amplitude-frequency curve crosses the 0dB line at a slope of -20dB / dec; in the high frequency band, the amplitude-frequency curve with a slope of -40dB / dec drops rapidly, which can fully suppress high-frequency noise and improve the system's stable anti-interference ability.

[0045] See also Figure 1 , which is a control block diagram of the LLC converter output voltage optimization method based on active disturbance rejection control provided by the present invention; refer to Figure 2 , which is a structural diagram of the active disturbance rejection controller based on the grey wolf algorithm provided by the present invention.

[0046] In the embodiment of the present invention, the resonant converter includes an inverter network, a resonant network, a transformer primary side, a transformer secondary side and a rectifier network in the order of connection, and the rectifier network outputs voltage to the outside.

[0047] In a specific implementation, the input voltage of the inverter network is V in, including 4 switch tubes Q1 to Q4, the switch tubes include MOS tubes and corresponding parallel diodes D1 to D4; the middle resonant network part is the resonant inductor L r and L m With the resonant capacitor C r The series structure, together with the excitation inductor, transfers energy from the primary winding N p Transferred to the secondary winding N s ; The rectifier network outputs a voltage V o , including four diodes (two D R1 and two D R2 ), with capacitor C in parallel f and resistor R L .

[0048] In an embodiment of the present invention, the inverter network is a full-bridge structure, including a first bridge arm and a second bridge arm, the first bridge arm includes a first switch tube Q1 and a third switch tube Q3, the second bridge arm includes a second switch tube Q2 and a fourth switch tube Q4, the duty cycle of the switch tube is 50%, the first switch tube Q1 and the fourth switch tube Q4 are controlled by a first control signal, the second switch tube Q2 and the third switch tube Q3 are controlled by a second control signal, and there is a dead time between the first control signal and the second control signal.

[0049] In a specific implementation, the duty cycle is the proportion of the conduction time of the switch tube in a switching cycle. The duty cycle of the switch tube is set to 50% to ensure that the working efficiency of the LLC is maximized while ensuring the smooth completion of the soft switching operation. The role of the dead time is to prevent the first bridge arm and the second bridge arm from being directly connected, thereby protecting the switch tube from damage.

[0050] In a specific implementation, the ADRC includes an extended state observer, a tracking differentiator, and a nonlinear state error feedback controller. Overall, the input V ref Output frequency signal f for inverter network control switch tube s .

[0051] In the embodiment of the present invention, the observed output voltage of the resonant converter is input to the extended state observer, and the output of the extended state observer includes an output voltage estimation value, an output voltage change rate and a total disturbance.

[0052] In practice, the Extended State Observer (ESO) is a tool for estimating state variables in a control system. It extends the traditional state observer to estimate both the internal state of the system and external disturbances or unknown inputs. ESO is particularly useful for improving the robustness of control systems because it can compensate for the effects of these unmodeled dynamics or external disturbances in real time.

[0053] In the specific implementation, the input of the ESO extended state observer includes the observed output voltage of the resonant converter, the product of the switching frequency control amount and the stability coefficient, and the output consists of three parts: voltage estimation (Z1), output voltage change rate (Z2) and total disturbance (Z3) (observation system total disturbance, including internal and external disturbances). The extended state observer is used to observe the real-time effects of unknown disturbances and system models in real time and compensate for them, which can replace the integrator required for linear design. The extended state observer is the core component of the active disturbance rejection control.

[0054] The operation logic of the ESO extended state observer is as follows:

[0055] x3(t)=f(x1,x2,w(t),t),

[0056] Among them, t represents time, x1 to x3 represent the state variables of the system, w(t) represents the external disturbance, and function f represents the total disturbance that combines internal and external disturbances.

[0057] Expanding a new state x3 based on the original system state yields:

[0058]

[0059] Among them, y represents the controlled object.

[0060] In the embodiment of the present invention, an extended state observer is established for this, using the following formula:

[0061]

[0062] Among them, Z1, Z2 and Z3 represent the output voltage estimate, the output voltage change rate and the total disturbance respectively, Z1', Z2' and Z3' represent the differentials of Z1, Z2 and Z3 respectively, y represents the observed output voltage of the resonant converter, ε1 represents the intermediate variable, β1, β2 and β3 represent the first gain, the second gain and the third gain of the extended state observer respectively (corresponding to Z1, Z2 and Z3 respectively), fal function represents the nonlinear function, δ represents the threshold value of the input variable of the nonlinear function, b0 represents the stability coefficient, and u represents the switching frequency of the control quantity.

[0063] In the embodiment of the present invention, the nonlinear function fal is calculated using the following formula:

[0064]

[0065] Where a represents the adjustment coefficient of the extended state observer, x represents the state variable of the extended state observer, and sign function represents the sign function. When a is generally taken as 1 / 2 or 1 / 4, the performance of the ESO extended state observer is optimal.

[0066] In an embodiment of the present invention, the reference output voltage is input to a tracking differentiator, and the tracking differentiator outputs a smoothed voltage tracking signal and a voltage change rate estimate. The smoothed voltage tracking signal and the output voltage estimate are subtracted to obtain a first error signal, and the voltage change rate estimate and the output voltage change rate are subtracted to obtain a second error signal.

[0067] Tracking Differentiator (TD) is a tool for smoothing signals and estimating signal derivatives, and is widely used in nonlinear control, Active Disturbance Rejection Control (ADRC), and other fields. It is mainly used to solve two problems: one is how to extract a clean signal from a noisy measurement signal; the other is how to accurately calculate the derivative of a signal without amplifying the noise.

[0068] In a specific embodiment, the feedback system of the tracking differentiator is as follows:

[0069]

[0070] Among them, r0 and h0 are function control parameters, r0 represents the fast factor, h0 represents the filter factor, v represents the input signal, k represents the sampling time, h represents the sampling period of the controlled object, function fhan represents the fastest comprehensive function of the tracking differentiator, and function fh is the abbreviated representation of function fhan.

[0071] Through the above system model of tracking differentiator, x1 can track the input signal v quickly and without overshoot, and x2 can track the differential signal in the process.

[0072] In the embodiment of the present invention, the first error signal and the second error signal are input into a nonlinear state error feedback controller, and the nonlinear state error feedback controller outputs an error integral signal;

[0073] Nonlinear State Error Feedback (NLSEF) is a method for designing nonlinear controllers, which is particularly suitable for dealing with systems with complex dynamic characteristics and uncertainties. NLSEF enhances the robustness and performance of the system by introducing nonlinear terms, enabling it to better cope with nonlinear behavior, external disturbances and model uncertainties.

[0074] The nonlinear state error feedback control rate can play the role of a nonlinear integrator, thereby converting the system into a simple integrator series standard type, as shown in the following formula:

[0075] u0=k1fal(e1,a1,δ)+k2fal(e2,a2,δ),

[0076] Among them, u0 represents the error integration signal, k1 and k2 represent the first proportional coefficient and the second proportional coefficient respectively, e1 and e2 represent the first error signal and the second error signal respectively, and a1 and a2 represent the first adjustment coefficient and the second adjustment coefficient respectively.

[0077] In a specific implementation, 0<a1<1<a2.

[0078] In the embodiment of the present invention, the switching frequency control amount is input to the voltage controlled oscillator VCO to obtain a corresponding frequency signal, and then the frequency signal is input to the PWM signal generator to obtain a square wave signal for controlling the switch tube.

[0079] In the embodiment of the present invention, the six key parameters β1, β2, β3, k1, k2 and b0 in the nonlinear active disturbance rejection controller are optimized by optimizing the gray wolf algorithm, and the iterative self-optimization of the parameters is achieved until the optimal key parameters are found.

[0080] Refer to FIG. 5( a ), which is a schematic diagram of the gray wolf position update provided by the present invention; refer to FIG. 5( b ), which is a flow chart of the improved gray wolf optimization algorithm provided by the present invention.

[0081] In the embodiment of the present invention, the fitness function of the gray wolf algorithm is determined, and the target output voltage (reference output voltage) V of the resonant converter is calculated. ref The deviation from the observed output voltage is used as a fitness function.

[0082] In the specific implementation, based on the method of gray wolf group predation in the gray wolf optimization algorithm, the population parameters are initialized first to determine the initial position and hierarchy of the gray wolf individuals, and the three gray wolf individuals with the best fitness function values ​​are named a, b and c wolves in turn, and the corresponding position information is X a , X b and X c , the rest are all e wolves. The hunting behavior of gray wolves is completed under the leadership of three alpha wolves. The hunting behavior mainly includes three parts: tracking, chasing and approaching prey; chasing, surrounding and harassing prey; attacking prey. The behavior of gray wolves encircling prey is defined as follows:

[0083] D”=|C”·X” p (t)-X”(t)|

[0084] X”(t+1)=X” p (t)-A”·D”,

[0085] Where D” represents the distance between the individual and the prey, X” (t+1) is the position update formula of the gray wolf, t is the current number of iterations, A” and C” are coefficient vectors, X p" and X" are the position vector of the prey and the position vector of the gray wolf respectively. The calculation formulas of A" and C" are as follows:

[0086]

[0087] Where J” represents the convergence factor, which decreases linearly from 2 to 0 with the number of iterations, and the modulos of r1” and r2” are random numbers between [0, 1].

[0088] In the decision space of the optimization problem, since the best solution is not known, it is necessary to use the three optimal solutions (three alpha wolves) to help the gray wolf individuals hunt, determine the location of the prey, and update their own positions according to the positions of the optimal gray wolf individuals, gradually approaching the prey. The mathematical model of the gray wolf individual tracking prey can be established by the following formula:

[0089]

[0090] Among them, D a ”、D b ” and D c " respectively represent the distances between a, b and c and the rest of the individuals, X a ”、X b ” and X c " "represent the current positions of a, b and c respectively, C1", C2" and C3" represent random vectors respectively, and X" represents the current position vector of the gray wolf individual.

[0091]

[0092]

[0093] Among them, X1”, X2” and X3” respectively represent the step length and direction of the gray wolf individual moving towards a, b and c, and define the final position of the individual, A1, A2 and A3 respectively represent the weight coefficients of the gray wolf individual moving towards a, b and c, and X”(t+1) represents the position of the gray wolf individual at the next moment. The swarm intelligence optimization algorithm is used to update the search speed and moving position of the gray wolf individual to avoid the population from stagnating in the optimization in advance. The algorithm for the search speed and moving position of the gray wolf individual is as follows:

[0094] v i,j (t+1)=wv i,j (t)+c1r1[p i,j -x i,j (t)]+c2r2[ω1X1+ω2X2+ω3X3-x i,j (t)]

[0095] x i,j (t+1)=x i,j (t)+vi,j (t+1){i=1,2,…n; j=1,2,N},

[0096] in:

[0097] Where n is the number of particles; N is the dimension of the search space, and p i,j represents the optimal position that the gray wolf has experienced, v i,j represents the search speed of the individual gray wolf, w represents the weight coefficient, ω1 to ω3 are inertia weight coefficients, c1r1(p i,j -x i,j (t)) represents the gray wolf's "cognitive" behavior, c2r2(ω1X1+ω2X2+ω3X3-x i,j (t)) represents the "social" behavior that the gray wolf learns from wolves a, b, and c respectively. Finally, it is determined whether the maximum number of iterations has been reached. If not, the above steps are continued; if reached, the loop is stopped and the optimal parameter solution is selected.

[0098] Refer to FIG. 6( a ), which is a parameter self-optimization curve provided by the present invention; refer to FIG. 6( b ), which is a fitness function convergence curve provided by the present invention.

[0099] In the specific implementation, considering the amount of computation and the computation rate, the maximum number of iterations of the present invention is set to 10 times. FIG6(a) and FIG6(b) respectively show the self-optimization curves of the six key parameters β1, β2, β3, k1, k2 and b0 and the convergence curve of the fitness function. It can be seen from FIG6(a) that Beta1 finally stabilizes at 4.12*10 after multiple iterations. 5 After iterative optimization, Beta2 finally stabilized at 7.3*10 10 , Beta3 finally stabilized at 2.6*10 15 , the optimal parameter values ​​of k1, k2 and b0 are 1.42*10 4 , 4.9 and 5.6*10 6 In addition, it can be seen that the optimal parameter values ​​of all parameters have been determined after the fifth iteration of optimization, so the maximum number of iterations is set to 10 to meet the optimization requirements. As can be seen from Figure 6(b), the convergence curve calculated according to the fitness function finally converges to 0 at the fifth iteration and tends to be stable, which can prove that the optimization method provided by the present invention has better convergence and stability.

[0100] See also Figure 7 , which is a comparison diagram of output voltage simulation waveforms of PID control, auto-disturbance rejection control and optimized grey wolf auto-disturbance rejection control provided by the present invention.

[0101] In the embodiment of the present invention, the input voltage is set to 360V, the output voltage is expected to be 30V, the transformer ratio is 12:1, the load resistance is set to 0.96Ω, and the resonant frequency is 115kHz. It can be seen from the figure that the output voltage overshoot of PID control and auto-disturbance rejection control is 7.65V, and the overshoot of auto-disturbance rejection control improved by the optimized gray wolf algorithm is 6.62V, which is reduced by 13.46%; the output power adjustment time of auto-disturbance rejection control and auto-disturbance rejection control improved by the optimized gray wolf algorithm is 5ms, which is significantly better than PID control; in addition, the load fluctuation condition is simulated at 0.01s, the load is doubled, and the dynamic characteristics of PID control are obviously inferior to the other two control methods. The overshoot of auto-disturbance rejection control is 1V, and the overshoot of auto-disturbance rejection control improved by the optimized gray wolf algorithm is 0.8V, which is reduced by 20%, and the adjustment time is also optimized. It can be seen that the auto-disturbance rejection control improved by the optimized gray wolf algorithm has good steady-state and dynamic performance, and the anti-disturbance ability is further improved.

[0102] The present invention also provides an LLC converter output voltage optimization device based on self-disturbance rejection control, comprising: an extended state observer control unit, a tracking differentiator control unit, a nonlinear state error feedback controller control unit, a voltage control unit and a parameter optimization unit, wherein: the extended state observer control unit is used to input the observed output voltage of the resonant converter into the extended state observer, and the output of the extended state observer includes an output voltage estimate, an output voltage change rate and a total disturbance; the resonant converter includes an inverter network, a resonant network, a transformer primary side, a transformer secondary side and a rectifier network in a connection order, and the rectifier network outputs a voltage to the outside; the tracking differentiator control unit is used to input the reference output voltage into the tracking differentiator, and the tracking differentiator outputs a smoothed voltage tracking value. The first error signal is obtained by subtracting the tracking signal from the estimated value of the voltage change rate, the second error signal is obtained by subtracting the estimated value of the voltage change rate from the output voltage change rate; the control unit of the nonlinear state error feedback controller is used to input the first error signal and the second error signal into the nonlinear state error feedback controller, and the nonlinear state error feedback controller outputs an error integral signal; the voltage control unit is used to subtract the error integral signal from the total disturbance and then divide it by the stability coefficient to obtain the switching frequency control amount, and control the switch tube in the inverter network based on the switching frequency control amount; the parameter optimization unit is used to obtain the parameters in the extended state observer and the nonlinear state error feedback controller, as well as the stability coefficient through a multi-objective optimization algorithm.

[0103] In an embodiment of the present invention, a unit for executing methods, steps or functions in an LLC converter output voltage optimization device based on active disturbance rejection control may refer to the LLC converter output voltage optimization method based on active disturbance rejection control provided by the present invention for the methods, steps or functions executed.

Claims

1. A method for optimizing the output voltage of an LLC converter based on active disturbance rejection control, characterized in that: include: The observed output voltage of the resonant converter is input into an extended state observer, and the output of the extended state observer includes an output voltage estimate, an output voltage change rate, and a total disturbance; The resonant converter includes, in order of connection, an inverter network, a resonant network, a transformer primary side, a transformer secondary side and a rectifier network, and the rectifier network outputs voltage to the outside; The reference output voltage is input to the tracking differentiator, the tracking differentiator outputs a smooth voltage tracking signal and a voltage change rate estimation value, the smooth voltage tracking signal and the output voltage estimation value are subtracted to obtain a first error signal, and the voltage change rate estimation value and the output voltage change rate are subtracted to obtain a second error signal; Inputting the first error signal and the second error signal into a nonlinear state error feedback controller, the nonlinear state error feedback controller outputting an error integral signal; The error integral signal and the total disturbance are subtracted and then divided by the stability coefficient to obtain the switching frequency control quantity, and the switch tube in the inverter network is controlled based on the switching frequency control quantity; the parameters in the extended state observer and the nonlinear state error feedback controller, as well as the stability coefficient, are obtained through a multi-objective optimization algorithm.

2. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 1, characterized in that: The extended state observer adopts the following calculation formula: Among them, Z1, Z2 and Z3 represent the output voltage estimation value, output voltage change rate and total disturbance respectively, Z1 ’ 、Z2 ’ and Z3 ’ Represent the differentials of Z1, Z2 and Z3 respectively, y represents the observed output voltage of the resonant converter, ε1 represents the intermediate variable, β1, β2 and β3 represent the first gain, second gain and third gain of the extended state observer respectively, fal function represents the nonlinear function, δ represents the threshold of the input variable of the nonlinear function, b0 represents the stability coefficient, and u represents the switching frequency of the control quantity.

3. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 2, characterized in that: The nonlinear function is calculated using the following formula: Wherein, a represents the adjustment coefficient of the extended state observer, x represents the state variable of the extended state observer, and sign function represents the sign function.

4. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 3 is characterized in that: The nonlinear state error feedback controller adopts the following calculation formula: u0=k1fal(e1,a1,δ)+k2fal(e2,a2,δ), Among them, u0 represents the error integration signal, k1 and k2 represent the first proportional coefficient and the second proportional coefficient respectively, e1 and e2 represent the first error signal and the second error signal respectively, and a1 and a2 represent the first adjustment coefficient and the second adjustment coefficient respectively.

5. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 4 is characterized in that: The controlling of the switch tube in the inverter network based on the switching frequency control amount includes: The switching frequency control amount is input into the voltage-controlled oscillator to obtain the corresponding frequency signal, and then the frequency signal is input into the PWM signal generator to obtain a square wave signal for controlling the switch tube.

6. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 5, characterized in that: The parameters of the extended state observer and nonlinear state error feedback controller obtained by the multi-objective optimization algorithm include: A first gain, a second gain and a third gain of the extended state observer, and a first proportional coefficient and a second proportional coefficient of the nonlinear state error feedback controller.

7. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 6, characterized in that: The method is obtained by a multi-objective optimization algorithm, including: Calculated using the Grey Wolf Algorithm.

8. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 7, characterized in that: The fitness function of the gray wolf algorithm includes: The deviation between the target output voltage and the observed output voltage of the resonant converter is calculated as the fitness function.

9. The LLC converter output voltage optimization method based on active disturbance rejection control according to claim 1, characterized in that: The inverter network is a full-bridge structure, including a first bridge arm and a second bridge arm, the first bridge arm includes a first switch tube and a third switch tube, the second bridge arm includes a second switch tube and a fourth switch tube, the duty cycle of the switch tube is 50%, the first switch tube and the fourth switch tube are controlled by a first control signal, the second switch tube and the third switch tube are controlled by a second control signal, and there is a dead time between the first control signal and the second control signal.

10. An LLC converter output voltage optimization device based on active disturbance rejection control, characterized in that: include: An extended state observer control unit, a tracking differentiator control unit, a nonlinear state error feedback controller control unit, a voltage control unit and a parameter optimization unit, wherein: The extended state observer control unit is used to input the observed output voltage of the resonant converter into the extended state observer, and the output of the extended state observer includes an output voltage estimate, an output voltage change rate, and a total disturbance; the resonant converter includes an inverter network, a resonant network, a transformer primary side, a transformer secondary side, and a rectifier network in a connection order, and the rectifier network outputs a voltage to the outside; The tracking differentiator control unit is used to input the reference output voltage into the tracking differentiator, the tracking differentiator outputs a smooth voltage tracking signal and a voltage change rate estimation value, the smooth voltage tracking signal and the output voltage estimation value are subtracted to obtain a first error signal, and the voltage change rate estimation value and the output voltage change rate are subtracted to obtain a second error signal; The nonlinear state error feedback controller control unit is used to input the first error signal and the second error signal into the nonlinear state error feedback controller, and the nonlinear state error feedback controller outputs an error integral signal; The voltage control unit is used to obtain a switching frequency control amount by taking the difference between the error integral signal and the total disturbance and then dividing it by the stability coefficient, and controlling the switch tube in the inverter network based on the switching frequency control amount; The parameter optimization unit is used to obtain the parameters in the extended state observer and the nonlinear state error feedback controller, as well as the stability coefficient through a multi-objective optimization algorithm.

Images