Adaptive selection method of virtual inductor for resisting pulse load disturbance

Abstract

The invention discloses a method for adaptively selecting a virtual inductor for resisting pulse load disturbance, comprising the following steps: establishing a main circuit topology structure of a three-phase three-wire voltage type rectifier with a pulse load; obtaining a mathematical model in an abc coordinate system according to the obtained main circuit topology structure, and then obtaining a mathematical model in a d-q coordinate system through equivalent transformation; converting the obtained mathematical model into an Euler-Lagrangian model; constructing a virtual energy storage matrix to be injected into the Euler-Lagrangian model to improve the convergence speed of the energy function and design a control law; setting the virtual inductance value to the value of the virtual inductor. L n Selection and load pulse frequency f PL associated so that L n Adaptive selection function f ( f PL ). The present invention will L n Selection and load pulse frequency f PL Related, proposed L n Adaptive selection function f ( f PL ), while meeting the DC bus performance requirements, ensuring the system stability margin.

Description

Technical Field

[0001] The invention relates to the technical field of rectifier control, and in particular to a virtual inductor adaptive selection method for resisting pulse load disturbance. Background Art

[0002] With the development of energy storage and power electronic conversion technology, high-power pulse weapon systems such as radar, electromagnetic catapult, and electromagnetic gun are widely used in national defense and military. Such loads have strong dynamic changes, and their load ends also require high-quality DC power supply. For example, most new radar equipment currently requires that the maximum fluctuation rate of the rectifier DC side output voltage δmax ≤ 5%, so the rectifier with pulse load needs to have strong robustness and the ability to resist load disturbances.

[0003] One of the solutions is to construct a virtual energy storage matrix to improve the convergence speed of the pulse load system switching between different steady states, thereby further reducing the impact of the pulse load on the DC voltage fluctuation rate. In the constructed virtual energy storage matrix, increase the virtual inductance L n , can further improve the rapidity of the convergence of the system energy function, and select the appropriate L n Can reduce DC bus u dc However, in actual use, it is found that L n The simple method of selecting a larger fixed value is only applicable to low-frequency conditions of pulse loads; when the load pulse frequency is high, a larger L n The value will affect the rapidity of the control algorithm response and introduce burrs to the system electrical parameters. In this case, L should be lowered. n value; therefore, the virtual inductance L in the pulse-proof load n How to adaptively select is an urgent problem to be solved. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a virtual inductor adaptive selection method for resisting pulse load disturbance.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] The method for adaptively selecting a virtual inductor for resisting pulse load disturbance comprises the following steps:

[0007] Establish the main circuit topology of three-phase three-wire voltage rectifier with pulse load;

[0008] The mathematical model in the abc coordinate system is obtained according to the main circuit topology structure, and then the mathematical model in the dq coordinate system is obtained through equivalent transformation;

[0009] Convert the obtained mathematical model into an Euler-Lagrangian model;

[0010] Construct a virtual energy storage matrix and inject it into the Euler-Lagrangian model to improve the convergence speed of the energy function and design the control law;

[0011] Set the virtual inductance value L n The selection of load pulse frequency f PL associated so that L n Adaptive selection function f(f PL ).

[0012] Further: the establishment of a three-phase three-wire voltage type rectifier with a pulse load main circuit topology structure includes:

[0013] Three-phase grid voltage a 、e b 、e c , rectifier AC side current i a 、i b 、i c , L is the input filter inductance of the rectifier, and the equivalent input resistance is r s ; Rectifier DC side capacitor C, capacitor voltage u dc is the rectifier DC output voltage; the load includes the conventional load R c and pulse load R PL And pulse switch S PL , where the pulse switch S PL By switching on and off, different frequencies f are generated PL and different duty cycle D PL The periodic pulse power.

[0014] Further: the mathematical model in the abc coordinate system is obtained according to the obtained main circuit topology structure, and then the mathematical model in the dq coordinate system is obtained through equivalent transformation:

[0015]

[0016] In the formula, S dq is the switching function S abc Component on the dq axis.

[0017] Further: the conversion of the obtained mathematical model into the Euler-Lagrangian model comprises:

[0018]

[0019] Where:

[0020]

[0021] Among them, M T is a positive definite diagonal matrix; J T is an antisymmetric matrix; RT is a symmetric positive definite matrix; u represents the energy exchange between the system and the outside world;

[0022] Let x e =xx * , where x * is the given instruction of each state variable, and the error energy function H is taken e (x) = x e T M T x e / 2(positive definite), design the control law u so that

[0023]

[0024] Further: the constructed virtual energy storage matrix is ​​injected into the Euler-Lagrangian model to improve the convergence speed of the energy function, and the designed control law includes:

[0025] Construct a virtual energy storage matrix as follows:

[0026]

[0027] Visible M N Based on the traditional interconnected and damping distributed passive control (IDA-PBC) design method, MN is injected into the control law, and

[0028]

[0029] Substituting into formula (3) we get

[0030]

[0031] Known x e =xx * , x * is the instruction of x, Represents the rate of change of x: when xx * >0,x * Will produce Guide x to become smaller; otherwise, xx * <0, then x * produce Guide x to become larger; therefore This proves that:

[0032] (Negative determination)

[0033] Substituting equation (2) into equation (6), we get the control law:

[0034]

[0035] The above formula can be expanded to obtain the passive control law of injecting virtual energy storage:

[0036]

[0037] Further: the virtual inductance value L in the virtual energy storage n The selection of load pulse frequency f PL Associated, construct the virtual inductance value L n Adaptive selection function f(f PL )include:

[0038] Sampling value f PL =[0.5,1,2,5,8,10,20,30,40,50] and L n =[1:1:10], fitting adaptive empirical function L n =f(f PL ):

[0039]

[0040] Among them, f PL0 is the minimum pulse frequency, L n0 f PL0 The corresponding prior value, a is the fitting coefficient.

[0041] Compared with the prior art, the present invention has the following technical advances:

[0042] The present invention will L n The selection of load pulse frequency f PL Associated, constructed L n Adaptive selection function f(f PL ), so that L n The value can adapt to the working conditions of different pulse frequencies of pulse loads, that is, under low-frequency conditions, the adaptive function f(f PL ) n Selecting a larger value can produce rigid suppression for large disturbances of the DC bus voltage under low-frequency conditions, ensuring that the system has sufficient stability margin; under high-frequency conditions, f(f PL ) Adaptively lower L n The value can not only improve the rapidity of the DC bus voltage response to high-frequency disturbances and reduce the introduction of burrs, but also suppress small-amplitude high-frequency disturbances of the DC bus voltage. n Adaptive selection function f(f PL ) has improved the flexibility of the passive control technology of virtual energy storage, and its rounding algorithm has also simplified the complexity in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0044] In the attached picture:

[0045] Figure 1 is a flow chart of the present invention;

[0046] Figure 2 The present invention is a main circuit diagram of a three-phase three-wire voltage type rectifier with a pulse load;

[0047] Figure 3 Three current inner loop control structure diagrams of the present invention;

[0048] Figure 4 The overall control strategy diagram of the pulse load rectifier of the VESI-PBC of the present invention;

[0049] Figure 5 This is the overall control structure diagram of the rectifier with pulse load of the present invention;

[0050] Figure 6 It is the parameter root locus diagram of the system function of the present invention;

[0051] Figure 7 For the present invention dc Dynamic performance parameters vary with L n The law of change;

[0052] Figure 8 is the adaptive empirical function L of the present invention n =f(f PL )picture;

[0053] Fig. 9 The different pulse frequencies f of the present invention PL Lower DC bus voltage u dc Waveform graph;

[0054] Fig.10 The different pulse frequencies f of the present invention PL The inner loop current i d With the command i d * Waveform diagram of

[0055] Fig.11 The different pulse frequencies f of the present invention PL Under this condition, the rectifier output current i dc With the load current i load The waveform of the capacitor current i C With DC voltage u dc Waveform graph;

[0056] Fig.12 The present invention fixes L n Value and variable L n =f(f PL ) dc Simulation waveform comparison;

[0057] Fig.13 This is the waveform diagram actually measured by the test system of the present invention. DETAILED DESCRIPTION

[0058] The following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0059] like Figure 1 , Figure 2 and Figure 3 As shown, the virtual inductor adaptive selection method for resisting pulse load disturbance includes:

[0060] like Figure 1 As shown, the present application discloses a passive control method for injecting virtual energy storage, comprising the following steps:

[0061] S01: Establish the main circuit topology of three-phase three-wire voltage type rectifier with pulse load;

[0062] S02: obtaining a mathematical model in the abc coordinate system according to the obtained main circuit topology structure, and then obtaining a mathematical model in the dq coordinate system through equivalent transformation;

[0063] S03: Convert the obtained mathematical model into an Euler-Lagrangian model;

[0064] S04: Construct a virtual energy storage matrix and inject it into the Euler-Lagrangian model to improve the convergence speed of the energy function and design the control law.

[0065] S05: Set the virtual inductance value L n The selection of load pulse frequency f PL associated so that L n Adaptive selection function f(f PL ).

[0066] Specifically, S01 includes:

[0067] Figure 2 This is the main circuit diagram of the three-phase three-wire voltage type rectifier with pulse load of the present invention. a 、e b 、e c is the three-phase grid voltage (three-phase AC power supply voltage), i a 、i b 、i cis the rectifier AC side current; L is the rectifier input filter inductance, and the equivalent input resistance is r s ; C is the rectifier DC side capacitor, capacitor voltage u dc is the rectifier DC output voltage; the load includes the conventional load R c (static load) and pulse load R PL (Dynamic load), in order to simulate the actual working conditions of radar equipment, the pulse switch S PL By switching on and off, different frequencies f are generated PL and different duty cycle D PL The periodic pulse power.

[0068] As the control switch of the PWM rectifier, S a,b,c The switching frequency is much higher than the grid voltage frequency, usually several kHz. When the switch is turned on, When the switch is off, Pulse switch S PL It has nothing to do with the rectifier control and is determined by the load device itself (S PL =1; when turned off, S PL =0), the frequency range is 0.1Hz~1000Hz.

[0069] Specifically, S02 includes:

[0070] according to Figure 2 The rectifier with pulse load topology can obtain the mathematical model in the abc coordinate system, and then through equivalent transformation, the mathematical model in the dq coordinate system is obtained as follows:

[0071]

[0072] In the formula, S dq is the switching function S abc Component on the dq axis.

[0073] Specifically, S03 includes:

[0074] The above PWM rectifier mathematical model is converted into the Euler-Lagrange (EL) model:

[0075]

[0076] Where:

[0077]

[0078] Among them, M T is a positive definite diagonal matrix, representing the energy storage passive devices in the system; J T is an antisymmetric matrix, reflecting the system coupling relationship and interconnection structure; R TIt is a symmetric positive definite matrix composed of dissipative elements such as system resistance and load; u represents the energy exchange between the system and the external (grid / power supply).

[0079] The design method of traditional interconnected and damped distributed passive control (IDA-PBC) is: Let x e =xx * (x * is the given instruction of each state variable), take the error energy function H e (x) = x e T M T x e / 2(positive definite), design the control law u so that

[0080]

[0081] According to Lyapunov's second judgment method, find the Lyapunov function, that is, H e (x) positive definite, if If negative, the system can be considered asymptotically stable. It can be seen that the control law u designed by IDA-PBC makes the system state variables x quickly converge to the steady-state x under the premise of ensuring system stability. * .

[0082] On this basis, the virtual damping injection method (VDI-PBC) is further adopted to make the error energy function of passive control converge faster. That is, the virtual damping matrix R is constructed. a =diag{r a ,r a ,1 / r a}(r a >0), let

[0083]

[0084] Where R d =R T +R a , substituting it into formula (3) we get

[0085]

[0086] Visible energy function H e (x) is exponentially stable, and its convergence rate is determined by R a Substituting equation (2) into equation (4), we get the control law

[0087]

[0088] Expanding this formula, we can get the command voltage V on the AC side of the PWM rectifier under the VDI-PBC algorithm: d 、Vq :

[0089]

[0090] By injecting a virtual damping matrix R a , so that the error function H of passive control e (x) has a faster convergence speed, and r a The larger the value, the faster the convergence speed. a It will cause over-modulation in the system, resulting in current distortion and affecting the stability and dynamic performance of the system.

[0091] Specifically, S04 includes:

[0092] Under the strong dynamic impact of pulse load, in order to further improve the rapidity of system convergence to steady state, this application constructs a virtual energy storage matrix based on VDI-PBC as follows:

[0093]

[0094] Visible M N Zhengding, order

[0095]

[0096] Substituting into formula (3) we get

[0097]

[0098] Known x e =xx * , x * is the instruction of x, Represents the rate of change of x: when xx * >0,x * Will produce Guide x to become smaller; otherwise, xx * <0, then x * produce Guide x to become larger; therefore This proves

[0099] Negative determination

[0100] The above formula shows that the system error energy function H e (x) converges exponentially, and the convergence speed is higher than that obtained by the existing damping injection passive control method. Increase M N The value of the item in L n and C n , which can further improve the error energy function H e (x), that is, to increase the speed at which the system enters a steady state.

[0101] It can be seen that formula (7) can make the system error energy function H e (x) converges exponentially, and the convergence speed is higher than that of the damping injection method. Its convergence speed is also determined by R a and M N Decide.

[0102] Substituting equation (2) into equation (6), we get the control law:

[0103]

[0104] With VDI-PBC control law VDI In comparison, M N Injecting energy storage device matrix, the energy storage matrix coefficient changes from L to L+L n , C becomes C+C n Therefore, the passive control law of injecting virtual energy storage can be obtained by expanding the above formula:

[0105]

[0106] In order to facilitate discussion and analysis, (L+L n ), written directly as L n From the derivation process (7), we can see that in the VESI-PBC algorithm, increasing L n , which can further improve the rapidity of convergence of the system energy function.

[0107] The voltage outer loop + current inner loop double closed loop structure is the most commonly used control strategy for voltage type rectifiers. The command voltage V of the current inner loop PI control d 、V q for

[0108]

[0109] Comparing equations (5), (9) and (10), the three control algorithms all use active current feedback control and all need to undergo current feedforward decoupling. Their inner loop control structures are as follows: Figure 3 shown.

[0110] Generally speaking, the system internal resistance r s Small (<1Ω), in order to make the passive controller converge faster, r a The value is larger (>10). Therefore, the internal resistance r s It can be almost ignored. By comparing (a) and (b), we can see that r a The role of is similar to the proportion link of PI; compared with (b) and (c), it can be seen that VDI-PBC has no L n s item, take L n = 0; therefore, the two control structures of PBC can be unified into a structure Figure 3 (c), i.e. L n =0 is VDI-PBC; L n When ≠0, it is VESI-PBC.

[0111] The voltage outer loop adopts PI control, and the DC voltage u dc Feedback, through the outer loop PI controller for the current inner loop G i (s) Generate active current command i d * The overall control strategy of the pulse load rectifier based on the VESI-PBC algorithm is as follows: Figure 4 As shown. Among them, due to the reactive current command i q * =0, its dynamic changes and trends are all 0, (L n s+r s )i q * =0, so VDI-PBC can be used directly; if i q * ≠0, the VESI-PBC algorithm shown by the dotted line can be used.

[0112] Based on formula (1), the correlation between the internal parameters of the pulse load rectifier is as follows:

[0113]

[0114] correspond Figure 4 The actual control mechanism of the VESI-PBC algorithm is analyzed by using equation (11). Firstly, it is clear that the control goal of the system in this application is to improve the ability of the DC bus voltage to resist load disturbances. The direct performance requirement is: reduce the DC voltage u dc According to formula ①, the capacitor current i C The change of is as small as possible; and formula ② shows that i C The fluctuation comes from the periodic pulse load current i load The rectifier DC side output current i dc The difference between dc As close as possible to the load current i load ; Formula ③ shows that i dc is the controlled quantity i d Therefore, the essence of load disturbance control is to control the controllable quantity i d Control to get as close to the disturbance source i as possible load waveform.

[0115] contrast Figure 3 Three inner loop control structures and equations (9) and (10):

[0116] (1) The inner loop PI control object is (i d * -i d ), its control effect is only in i d to i d * tracking, so the dual PI pair u dc The control is limited to the outer loop PI. Due to the buffer capacitor C, u dc The dynamic change lags behind the pulse current i load , thus the response speed is limited;

[0117] (2) PBC control in addition to (i d * -i d ) will also play a role d * Introducing inner loop control, especially VESI-PBC also introduces i d * s item At the same time, formula (9) shows that due to the generation of i d The controlled quantity V d Item 4 Implementation i d to i d * Tracking, Response 3 d * Dynamic changes of , item 2 Advance Response d * The changing trend of dc The dynamic change trend of the load current is the disturbance caused by the load current. Therefore, compared with PI control, PBC control has a faster response speed to load disturbances and dynamic changes of DC voltage.

[0118] Specifically, S05 includes:

[0119] Next, from the perspective of control theory, we will further discuss the introduction of the energy storage matrix M. N After L n The impact of the value on the system stability and dynamic performance.

[0120] based on Figure 3 (c) Write the current inner loop transfer function of PBC control

[0121]

[0122] It can be seen that L n =0 is VDI-PBC; L n ≠0 is VESI-PBC. According to the mathematical model formula (1), the overall control structure of the rectifier with pulse load is obtained as follows: Figure 5shown.

[0123] It can be seen that when the pulse switch S PL When switching between 0 / 1, the DC voltage u dc To the instruction u dc * The transfer functions are G sysI (s) and G sysII (s). Let G1(s) = 1.5G PI (s)G i (s), G2(s)=1 / Cs, and the system transfer function is

[0124]

[0125] Let K Load =1 / R c +S PL / R PL This parameter represents the load capacity, which is determined by the normal load capacity (1 / R c ) and pulse load (1 / R PL ) consists of two parts, K Load The larger it is, the heavier the load. PL The process of controlling the pulse load to increase or decrease load follows the periodic variation law of pulse power, which is determined by the pulse frequency f PL and duty cycle D PL Two parameters characterize it.

[0126] Substituting equation (12) into equation (13), we can get the system function G sys (s), the derivation is shown in Appendix A.1, let L n = 0 and 1, we can discuss the load parameter K under VDI-PBC and VESI-PBC algorithms. Load The root locus from 0 to ∞, that is, the law of change of system stability with load, is as follows Figure 6 (a) and (b) are shown; Figure (c) is K Load =0.25, VESI-PBC algorithm parameter L n The root locus from 0→∞. Other parameters are set according to the simulation parameter values ​​in the following text.

[0127] Depend on Figure 6 It can be seen that:

[0128] (1) The zero poles are always located in the left half plane, and the system is in a stable state, but the main pole p1 is closer to the zero point, and the stability margin is not high. As can be seen from Figures (a) and (b), with the increase of K Load As the load increases, p1 approaches z1 (zero point) and the stability margin decreases; that is, the heavier the load, the lower the stability margin.

[0129] (2) Comparison with Ln =0 and L n =1, that is, the zero-pole position of Figure (a) (b), the main pole p1 of both remains unchanged, and the secondary pole p2 decreases (-66→-25), that is to say, compared with VDI-PBC, the stability margin of VESI-PBC algorithm is slightly reduced.

[0130] (3) Comparison (a) and (b), L n =0 when K Load Within a certain range, the root locus deviates from the horizontal axis and the system enters an underdamped state; n = 1, the root loci are all located on the horizontal axis, and the system is always in an overdamped state. n It can improve the rigidity and anti-interference ability of the system.

[0131] Figure (c) shows that, for a fixed K Load , with L n As the main pole p1 approaches the zero point, L n The larger the S, the lower the stability margin. PL Stability during switching, L n The value cannot be too large.

[0132] The pulse load operation process is actually a continuous periodic load increase and decrease process, and its control structure periodically switches between two states (sysI and sysII), such as Figure 5 As shown; the system function G sys The load parameter of (s) is K Load1 =1 / R c (S PL =0) and K Load2 =1 / R c +1 / R PL (S PL =1) jumps back and forth, such as Figure 6 (b) is shown by the dotted line. Figure 5 Write down the DC voltage u dc The whole process expression.

[0133] U dc (s)=G0(s)U dc * (s)+G r (s)I load (s)

[0134] The first term in the formula is u dc For instruction u dc * The control objective is to convert u dcStable at the command value, only produces dynamic response at startup; the second item is the load current i load To u dc The disturbance, that is, u dc Dynamic load changes (S PL The second item is the main dynamic response of the pulse load operation process, and it is a continuous and periodic dynamic process. This application mainly discusses the dynamic performance of the second item during the pulse load switching process.

[0135] Let the DC voltage fluctuation amplitude Δu dc =u dc -u dc * , the fluctuation amplitude ΔU of the pulse load switching process is obtained by the derivation of Appendix A.3 dc (s) and written in second-order standard form

[0136]

[0137] Let the DC voltage u dc Volatility δ(t) = ±Δu dc (t) / u dc * , from which we can get the time domain expression of |δ(t)|:

[0138]

[0139] Among them, the coefficient Damping ratio Time constant

[0140] The above parameters determine the dynamic performance of |δ(t)| and are related to L n For each parameter, plot the L n The changing rules are as follows Figure 7 As shown (L n =0, it is the VDI-PBC scheme).

[0141] Maximum volatility δ max (|Δu dc_max | / u dc * ) and adjustment time t s (The time for the fluctuation rate δ to adjust to σ) is an important parameter to characterize the dynamic performance of DC voltage. Figure 7 It can be seen that the A value is very small (10 -3 Level), can be ignored, and the maximum volatility δ can be derived max and its peak time t p Similarly, since T2 is much smaller than T1, let |δ(t s)|≤σ, the adjustment time t s expression:

[0142]

[0143] From the expression and Figure 7 Analysis shows that:

[0144] (1) Formula (14) shows that the volatility δ is related to R PL Related, R PL The smaller the max The larger the load, the greater the fluctuation. The following analysis conclusions are based on the same load (R PL =1Ω).

[0145] (2) When L n = 0, that is, under the VDI-PBC algorithm, B = 1 / C, Visible delta max With r a Not relevant. Figure 7 Medium Delta max The trend chart shows the maximum fluctuation rate of VDI-PBC δ max Up to 55%, while increasing r a , cannot effectively reduce δ max .

[0146] (3) When L n ≠0, that is, under the VESI-PBC algorithm, the main gain coefficient B (the secondary gain A value is very small) increases with L n Increase or decrease, δ max Also with L n It can be seen that increasing L n , which can effectively reduce the DC voltage fluctuation rate.

[0147] (4) Figure 7 Both equation (15) show that δ max and t s The change law is opposite, increasing L n , the main time constant T1 also increases (the secondary time constant T2 is very small), and the adjustment time t s Also with L n Increase and increase.

[0148] Therefore, in the VESI-PBC algorithm, the appropriate L n Can reduce u dc However, in order to ensure a certain response speed, L n Don't take too big.

[0149] Formula (15) only considers u dcThe full response process of pulse load addition and reduction, but does not take into account the pulse period T PL and duty cycle D PL When the loading time t load =T PL ×D PL , or load shedding time t unload =T PL ×(1-D PL ) is greater than the adjustment time t s When t load (t unload ) <t s When the system has not yet reached the steady-state value, it needs to enter the next transient state (load / load reduction) from the transient state (load reduction / load). Due to the periodicity, the system needs to switch repeatedly between the two transient states to form a certain degree of steady state, which is called "transient steady state". Especially when t load (t unload ) <t p When |-δ(t load )| and |δ(t unload )|<δ max , and we know that: T PL The smaller the f PL The larger it is, the smaller the volatility δ is.

[0150] Based on the above conclusions: Increase L n , although it can effectively reduce the rectifier DC side voltage u dc The fluctuation range of f will be reduced, but it will sacrifice the stability margin and response speed of the system to a certain extent. PL The larger the value, the faster the load disturbance speed. Therefore, in order to meet the DC bus performance requirements and ensure the stability margin of the system, it is possible to consider the pulse frequency f PL From the adaptive value L n .

[0151] The specific approach is: based on simulation experiments, the value f can be sampled PL =[0.5,1,2,5,8,10,20,30,40,50] and L n =[1:1:10], based on the DC bus performance requirements, for example, δ max ≤5% is the constraint target, thus fitting the adaptive empirical function L n =f(f PL ):

[0152]

[0153] Among them, f PL0 is the minimum pulse frequency, L n0 f PL0The corresponding prior value, a is the fitting coefficient. L n The value can be a decimal, but for ease of operation, it is generally an integer. int{} is the rounding function (rounding can be selected). If the DC voltage fluctuation rate δ is required max If there is a certain margin, int{} can be rounded up; if you want to ensure a certain stability margin, you can round down. Corresponding to the simulation parameters in the following text, the adaptive empirical function is drawn as follows Figure 8 As shown, where L n0 =10, a=0.25, f PL0 =0.5Hz.

[0154] In actual experiments, it is not necessary to select too dense sampling values, and a piecewise function can be used. For example, f PL It is divided into three sections: low, medium and high frequency.

[0155] Based on the rectifier structure with pulse load, and Figure 6 Control strategy, build its Matlab / Simulink model. The simulation parameters are as follows: ① Source: Use a diesel generator with a rated power of P s =250kW, rated frequency f s =50Hz; ②Rectifier: AC side filter inductance L = 2.16mH, AC side equivalent resistance r s =0.3Ω(estimated value );DC side command voltage u dc * =700V, DC side energy storage capacitor C = 20mF, power device IGBT switching frequency f PWM =30kHz; ③ Load end: conventional load R c =10Ω(static power 49kW), pulse load R PL =4Ω (pulse power 122.5kW), the pulse peak power is 2.5 times the static power, the pulse frequency f PL and duty cycle D PL Adjustable; ④Control strategy: The voltage outer loop adopts PI control, k p_U =1, k i_U =8; the current inner loop uses PI, VDI-PBC (r a =50) and VESI-PBC (L n =1, 5 and 10) three algorithms are compared. In order to compare with VDI-PBC, PI takes k p_I =1, k i_I =50.

[0156] In order to verify the applicability of the algorithm to pulse powers of different frequencies, the load pulse frequency f is taken PL=0.5Hz→1Hz→5Hz→10Hz→50Hz→100Hz. Existing research shows that the f of pulse load equipment PL The lower the pulse duty cycle D PL =50%, the lower the system stability. Therefore, the simulation and actual test process of this application mainly focuses on the low and medium frequency bands. The new radar equipment requires the maximum fluctuation rate of DC voltage δ max ≤5%, according to formula (15) we can get L n ≥9.2.

[0157] Figure 4 The circled positions mark the observation points of the internal parameters of the simulation test, and the parameters (DC voltage u dc , active current command i d * , actual active current i d , rectifier DC side output current i dc , capacitor current i C )The simulation waveform is as follows Figure 9-11 As shown. Calculate u dc Maximum volatility ±δ max As shown in Table 1, the bold part is the volatility that does not meet the performance indicators.

[0158] Table 1

[0159]

[0160] in accordance with Figure 9-11 According to Table 1, the following conclusions can be drawn:

[0161] (1) The simulation test waveforms and data are all under the same outer loop control, only the inner loop control algorithm is different. dc The regulation of fluctuations is limited and can only be reduced by increasing the energy storage of the DC side capacitor C.

[0162] (2) Pulse frequency f PL The lower the pulse load is, the greater the impact of the pulse load on the system. Inner loop PI or traditional PBC control can only be controlled at f PL ≥50Hz to reach δ max ≤5% performance requirement.

[0163] (3) From Fig. 9 As can be seen from Table 1, VESI-PBC can effectively reduce the maximum fluctuation rate of DC bus voltage δ max , and L n The larger the value, the max The smaller it is, the stiffer the system is.

[0164] (4) Through Figure 10-11By observing the internal parameters of the system, we can further observe the effect of VESI-PBC on u dc Volatility suppression mechanism:

[0165] ① As the above conclusion: the role of the inner ring PI is only in i d to i d * tracking. Fig.10 It can be seen that under the PI algorithm, d Follow me d * , under the same PI parameters, its tracking effect is not as good as VDI-PBC. Of course, by adjusting the optimization k p_I and k i_I Better tracking accuracy can be achieved.

[0166] ②But if Fig.11 As shown, even if the inner ring PI increases i d to i d * The tracking accuracy of i dc Unable to quickly follow the pulse current i load The instantaneous impact of C is larger, causing u dc Larger fluctuations.

[0167] ③ Fig.10 It shows that under the control of VESI-PBC d Ahead of i d * , and L n The bigger, i d The closer it is to a square wave, the corresponding i dc (See Fig.11 ) has high rapidity and can quickly follow the pulse current at the moment of impact. load , get a smaller i C , which effectively reduces u dc It can be seen that the essence of VESI-PBC is to introduce Control dc Fast Track load , thus achieving the purpose of resisting load disturbance, which is consistent with the above theoretical analysis conclusion.

[0168] (5) However, the problem with VESI-PBC is that Not only has it introduced d * The normal dynamic changes also amplify other interference currents, making i d 、i dc and i C Generates burr noise, and L nThe larger the value, the larger the burr; at the same time, f PL The higher the load, the more severe the dynamic change of the load, and the more glitches there are. However, the DC side capacitor C can effectively filter out these high-frequency glitches.

[0169] Based on conclusion (5), L n The value is selected using an adaptive function. Taking this simulation as an example, Figure 8 The fitted f(.) parameters. Fig.12 The following figure shows the fixed value L used under the VESI-PBC algorithm. n =10 and variable L n =f(f PL ), DC voltage u dc Comparison of simulation waveforms.

[0170] It can be seen from the figure that when L n When f is reduced, the waveform burr can be effectively reduced, but the fluctuation rate increases significantly; in this simulation case, when f PL After ≥30Hz, VDI-PBC (L n =0) can meet the performance requirements.

[0171] The following is the actual experiment:

[0172] The experimental platform mainly includes an AC power supply, a three-phase rectifier, a pulse load and a control cabinet, including: the three-phase AC power supply adopts a 250kW diesel generator set; the rectifier uses TMS320F2812 as the core control circuit board, its AC side filter inductor is 2.16mH, the DC side energy storage capacitor is 20mF / 800V, the DC voltage control is 700V, and the power device switching frequency is 30kHz; the DC static load is 10Ω; the pulse load is 8Ω, the pulse duty cycle is 50%, and the frequency is adjustable; the measured voltage and current data are obtained using Agilent oscilloscope recorder MSO6102A.

[0173] The inner loop control adopts PI control, VDI-PBC and VESI-PBC control algorithms respectively, and compares different pulse frequencies f PL Lower DC side voltage u dc , AC side voltage u a and current i a Waveform.

[0174] In fact, the PI control after optimizing parameters is similar to the VDI-PBC control (r a =50, optimal test, r a The dynamic waveforms of the control algorithm are similar, so only the measured waveforms under PI control and VESI-PBC control algorithms are compared. Among them, VESI-PBC control uses variable L n =f(fPL ):f PL =2Hz, L n =10; f PL =10Hz, L n =2; f PL ≥30Hz, L n = 0. Next Fig.13 The control effects of PI and VESI-PBC algorithms at different pulse frequencies are compared, and the index values ​​are calculated as shown in Table 2:

[0175] Table 2

[0176]

[0177] According to the measured waveform and Table 2, the following conclusions can be drawn:

[0178] (1) Since the power supply is a diesel generator set, the system capacity is effective, so the AC voltage u abc The amplitude will fluctuate due to the impact of pulse power. The voltage fluctuation rate of phase a in Table 2 is U a_r is the effective value; by comparison, it can be found that the fluctuation rate δ on the AC side ua and the DC side fluctuation rate δ udc have the same trend, namely, δ udc The smaller δ is, the ua The smaller.

[0179] (2) In contrast, the VESI-PBC algorithm can effectively reduce u dc The fluctuation range of L n When u is larger, dc There will be some burrs, see Fig.13 (b); when it is reduced, the burr phenomenon is weakened;

[0180] (3) From the measured data, it can be seen that although the VESI-PBC algorithm reduces the fluctuation rate of the AC and DC sides, the AC current i a The distortion rate of the load current i is increased; this is because the VESI-PBC algorithm responds quickly to the load current i at the moment of the pulse. load , constructing an i close to the pulse current dc This will naturally lead to distortion of the AC current. It can be considered that the VESI-PBC algorithm sacrifices the distortion rate of the AC current to achieve the purpose of reducing the voltage fluctuation rate on the DC side.

[0181] Aiming at the performance requirements of pulse power load equipment such as radar on the DC side voltage and the periodic dynamic characteristics of pulse load, this application establishes and analyzes the equivalent topological structure and mathematical model of a three-phase rectifier with pulse load, and proposes a passive control with virtual energy storage injection (VESI-PBC) algorithm based on the design of traditional passive control algorithm: on the one hand, with the goal of improving the convergence speed of the energy function, the design process of the VESI-PBC algorithm is derived under the premise of ensuring the stability of the system; on the other hand, from the perspective of control principle, the introduction of the energy storage matrix M in the VESI-PBC algorithm is discussed. N After that, increase the virtual inductance value L n Impact on the stability and dynamic performance of the system.

[0182] Both theoretical analysis and experiments indicate that in the VESI-PBC algorithm, increasing L n The value can effectively reduce the fluctuation amplitude of DC voltage and improve the ability of DC bus to resist load interference, but it has a certain impact on the system stability margin and increases the current burr. Therefore, based on the influence of pulse load on the system, this application sets L n The selection of load pulse frequency f PL Related, proposed L n Adaptive selection function f(f PL ).

[0183] Based on simulation and actual experiments, this paper verifies the effectiveness of VESI-PBC algorithm in reducing DC bus voltage fluctuation rate by comparing with PI and existing VDI-PBC control, and points out that the algorithm controls i d Advance response dc , which actually responds to the load current i load , thus having a strong ability to resist load disturbances, and is suitable for situations where the pulse load has a high power level and strong dynamic characteristics.

[0184] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of protection of the claims of the present invention.

Claims

1. A virtual inductor adaptive selection method for resisting pulse load disturbance, characterized in that: The steps include: Establish the main circuit topology of three-phase three-wire voltage type rectifier with pulse load; The mathematical model in the abc coordinate system is obtained according to the main circuit topology structure, and then the mathematical model in the dq coordinate system is obtained through equivalent transformation; Convert the obtained mathematical model into an Euler-Lagrangian model; Construct a virtual energy storage matrix and inject it into the Euler-Lagrangian model to improve the convergence speed of the energy function and design the control law; The virtual inductance value L in the virtual energy storage n The selection of load pulse frequency f PL associated so that the virtual inductance value L n Adapt to different frequency characteristics of pulse load, construct L n Adaptive selection function f(f PL ); The virtual inductance value L in the virtual energy storage n The selection of load pulse frequency f PL Associated, construct the virtual inductance value L n Adaptive selection function f(f PL )include: Sampling value f PL =[0.5,1,2,5,8,10,20,30,40,50] and L n =[1:1:10], fitting adaptive empirical function L n =f(f PL ): Among them, f PL0 is the minimum pulse frequency, L n0 f PL0 The corresponding prior value, a is the fitting coefficient.

2. The method for adaptively selecting virtual inductance for resisting pulse load disturbance according to claim 1, characterized in that: The main circuit topology structure of the three-phase three-wire voltage type rectifier with pulse load is established as follows: Three-phase grid voltage a 、e b 、e c , rectifier AC side current i a 、i b 、i c , L is the input filter inductance of the rectifier, and the equivalent input resistance is r s ; Rectifier DC side capacitor C, capacitor voltage u dc is the rectifier DC output voltage; the load includes the conventional load R c and pulse load R PL And pulse switch S PL , where the pulse switch S PL By switching on and off, different frequencies f are generated PL and different duty cycle D PL The periodic pulse power.

3. The method for adaptively selecting virtual inductance for resisting pulse load disturbance according to claim 2, characterized in that: The mathematical model in the abc coordinate system is obtained according to the obtained main circuit topology structure, and then the mathematical model in the dq coordinate system is obtained through equivalent transformation: In the formula, S dq is the switching function S abc Component on the dq axis.

4. The method for adaptively selecting virtual inductance for resisting pulse load disturbance according to claim 3 is characterized in that: The conversion of the obtained mathematical model into the Euler-Lagrangian model comprises: Where: Among them, M T is a positive definite diagonal matrix; J T is an antisymmetric matrix; R T is a symmetric positive definite matrix; u represents the energy exchange between the system and the outside world; Let x e =xx * , where x * is the given instruction of each state variable, and the error energy function H is taken e (x) = x e T M T x e / 2(positive definite), design the control law u so that 5. The method for adaptively selecting virtual inductance for resisting pulse load disturbance according to claim 4 is characterized in that: The virtual energy storage matrix is ​​constructed and injected into the Euler-Lagrangian model to improve the convergence speed of the energy function. The control law is designed to include: Construct a virtual energy storage matrix as follows: Visible M N Based on the traditional interconnection and damping distributed passive control (IDA-PBC) design method, M N Inject the control law, let Substituting into formula (3) we get Known x e =xx * , x * is the instruction of x, Represents the rate of change of x: when xx * >0,x * Will produce Guide x to become smaller; otherwise, xx * <0, then x * produce Guide x to become larger; therefore This proves that: Substituting equation (2) into equation (6), we get the control law: The above formula can be expanded to obtain the passive control law of injecting virtual energy storage:

Images