Bonding graph model prediction control method of full-bridge inverter

Abstract

The invention relates to a bonding graph model prediction control method for a full-bridge inverter, and the method comprises the steps: splitting the bridge inverter into a combination form of buck and boost circuits, and building a bonding graph model of the bridge inverter; and based on the bonding graph model of the bridge type inverter, realizing model prediction control of the bridge type inverter. The method is suitable for a single-phase inverter and a three-phase inverter, the switching state of each phase of bridge arm does not affect the switching states of the other two phases of bridge arms, each phase of bridge arm can work independently, and three-phase independent control and decoupling of each phase of current are achieved; switching vectors do not need to be traversed, the calculated amount is small, and especially for an inverter system under digital control, a control system is simplified; the provided bonding graph model can be used for modeling a power switch device, the established model has no error, the control precision based on the model is high, the bonding graph model can be further applied to circuits with switch devices such as a switch power supply and a power electronic converter, and the application field of the bonding graph is expanded.

Description

Technical Field

[0001] The present invention relates to the technical field of full-bridge inverter model predictive control, and in particular to a bond graph model predictive control method for a full-bridge inverter. Background Art

[0002] Existing full-bridge inverter model predictive control is based on the inverter's bridge arm output relationship model and traverses the switch vector for prediction. This control method is not suitable for single-phase inverters. The inverter output current ripple is large and coordinate changes are required. The calculation amount is large and the requirements for the controller are high.

[0003] For single-phase inverters, it is difficult to establish the state equation using existing modeling methods. The transfer function established by Laplace transform must be discretized in model predictive control. Different discretization methods produce different and inaccurate state equations, resulting in poor accuracy of model predictive control.

[0004] Bond graph is a common method for establishing state equations with high accuracy. It has been widely used in the field of mechanical dynamic modeling. However, in the field of power electronics, due to the presence of power switching devices in the circuit, the traditional bond graph modeling method cannot model the power switching devices. Therefore, establishing state equations based on bond graphs can only be applied to traditional analog circuits. For circuits with switching devices such as switching power supplies and power electronic converters, it is impossible to establish the circuit state equations through bond graph models. Summary of the invention

[0005] In order to solve the problems in the prior art that it is difficult to establish the inverter state equation, and the calculation amount is large and the control accuracy is low, the purpose of the present invention is to provide a bond graph model predictive control method for a full-bridge inverter that realizes three-phase independent control and decoupling of each phase current, does not need to traverse the switching vector, has a small amount of calculation, and simplifies the control system.

[0006] To achieve the above object, the present invention adopts the following technical solution: a bond graph model predictive control method for a full-bridge inverter, the method comprising the following steps in order:

[0007] (1) Split the bridge inverter into a combination of buck and boost circuits, and establish a bond graph model of the bridge inverter;

[0008] (2) Based on the bond graph model of the bridge inverter, the model predictive control of the bridge inverter is realized.

[0009] The step (1) specifically comprises the following steps in order:

[0010] (1a) Split the bridge inverter into a combination of buck and boost circuits:

[0011] The bridge arm is split into a forward buck circuit and a reverse boost circuit. For the forward buck circuit, its input is the potential difference u between points P and N. PN , the output is the potential difference u between points A and N AN , for the reverse boost circuit, its input is u AN , the output is u PN Since the forward buck and reverse boost circuits operate simultaneously in CCM mode, the inductor L 1-1 , L 1-2 It is equivalent to parallel connection, so the equivalent inductance of the bridge inverter is L / 2. The switch S and the diode D are equivalent to the switches that are alternately turned on, which are named SD bonding. There are two forms: buck and boost. For these two forms of SD bonding, the relationship is:

[0012]

[0013] u1=swu6+(1-sw)u5

[0014] Where, u1, u5, u6 are the voltages of the first, fifth, and sixth keys respectively, i1, i5, i6 are the currents of the first, fifth, and sixth keys respectively; sw is the driving signal of switch S;

[0015] (1b) The bonding graph model of the bridge inverter is established based on the SD bonding in the form of buck and boost.

[0016] The step (2) specifically comprises the following steps in order:

[0017] (2a) Establish the state equation of the bridge inverter system:

[0018] The 0 node is a common potential junction, which is used to connect variables with equal voltage in the circuit. For the bottom 0 node, we have:

[0019] u2=u 11 =u 13 =u 14

[0020] i 14 =i2+i 11 -i 13

[0021] Node 1 is a common current node, which is used to connect variables with equal current in the circuit. For the bottom node 1, we have:

[0022] i 13 =i 15 =i 16

[0023] u 15 =u13 -u 16

[0024] Establish the state equation, and the input variable u(t) is:

[0025] u(t)=sw(t)

[0026] Where sw(t) is the driving signal of switch S at time t;

[0027] The state variable x(t) is:

[0028]

[0029] Where q is the capacitor charge and the capacitor current Capacitor voltage The capacitor voltage p is the inductor flux, and the inductor voltage Inductor Current Then the first differential of the state variable is:

[0030]

[0031] In the formula, i 14 、u 15 , u3 and u 10 That is i Cf 、u L2 、u L1-1 and u L1-2 ;

[0032] u 14 、i 15 , i3 and i 10 It is expressed as:

[0033]

[0034] On the load side, there is u 16 =Ri 16 =Ri 13 =Ri 15 , for the bottom node, we have:

[0035]

[0036] For the bottom 0 node, we have:

[0037]

[0038] in, but:

[0039]

[0040] The above formula is the first state variable differential term i14 That is, the expression for the current of the 14th bond;

[0041] For the bottom 0 node, due to u 13 =u 14 ,but

[0042]

[0043] The above formula is the second state variable differential term u 15 That is, the expression for the current of the 15th bond;

[0044] For the leftmost node, we have

[0045] u3=-u2-swu6

[0046] in, u6=u 12 =U in , then the above formula is converted to:

[0047]

[0048] The above formula is the expression of the third state variable differential term u3, which is the voltage of the third key;

[0049] For the rightmost node, we have:

[0050] u 10 =-u 11 -swu7

[0051] in, u7=u 12 =U in , then the above formula is converted to:

[0052]

[0053] The above formula is the fourth state variable differential term u 10 That is, the expression of the voltage of the 10th key. Combining the above formulas, we can get:

[0054]

[0055] Written in matrix form:

[0056]

[0057] in:

[0058]

[0059] In the formula, i2, i3, i 10 、i 11 、i13 、i 14 、i 15 、i 16 are the currents of the second, third, tenth, eleventh, thirteenth, fourteenth, fifteenth, and sixteenth keys respectively; u2, u6, u7, u 11 、u 12 、u 13 、u 14 、u 16 are the voltages of the second, sixth, seventh, eleventh, twelfth, thirteenth, fourteenth and sixteenth buttons respectively; sw is the driving signal of switch S; R, C, L are the equivalent load of the circuit, the output filter capacitor and the boost inductor respectively; Uin is the input voltage of the BOOST circuit; A and B are the coefficient matrices; q 14 is the capacitance C f The charge, p 15 ,p3,p 10 They are inductors L2 and L 1-1 , L 1-2 The magnetic flux, L 1-1 and L 1-2 are equivalent inductances, C f is the capacitor in the LCL filter, L2 is the inductance on the AC side of the LCL filter, i cf is the current flowing through capacitor C f The current, U L2 , U L1-1 , U L1-2 They are inductors L2 and L 1-1 , L 1-2 The terminal voltage of

[0060] (2b) Discretize the state equation, determine the output equation and cost function, and the discretization can be obtained:

[0061]

[0062] Output equation

[0063] Then the cost function J is set to:

[0064]

[0065] J is the error integral form;

[0066] Where Ts is the calculation period;

[0067] (2c) Perform model predictive control:

[0068] Sampling u 14 、i 15 、i3、i 10 , calculate q 14 、p15 ,p3,p 10 According to formula (1), the state variables when sw is 0 and 1 are predicted respectively, so as to calculate p 15 Substitute the predicted value into formula (2) and take the sw corresponding to the minimum J to achieve model predictive control.

[0069] It can be seen from the above technical scheme that the beneficial effects of the present invention are as follows: first, the present invention is applicable to both single-phase and three-phase inverters, the switching state of each phase bridge arm does not affect the switching state of the other two phase bridge arms, and each phase bridge arm can work independently, realizing three-phase independent control and decoupling of each phase current; second, the present invention only determines whether the switching state of the power switch tube in the next calculation cycle is open or closed through the state equation, that is, the input quantity of the state equation is only one switch signal, and its value is only 0 and 1. There is no need to traverse the switch vector, and the calculation amount is small, especially for the inverter system under digital control, the data processing requirements of the main controller are low, and the control system is simplified; third, the proposed bond graph model can model the power switching device, the constructed model has no error, and the control accuracy based on the model is high. It can be further applied to circuits with switching devices such as switching power supplies and power electronic converters, expanding the application field of the bond graph and providing a new method for circuit control with switching devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is the combination relationship between the single-phase bridge arm of the bridge inverter and the buck and boost circuits;

[0071] Figure 2 yes Figure 1 The bridge arm equivalent circuit plus the circuit diagram of the LCL filter;

[0072] Figure 3 It is a schematic diagram of SD bonding;

[0073] Figure 4 It is the bonding diagram of the single-phase bridge arm of the full-bridge inverter;

[0074] Figure 5 It is a flow chart of the method for model predictive control. DETAILED DESCRIPTION

[0075] A bond graph model predictive control method for a full-bridge inverter, the method comprising the following steps in order:

[0076] (1) Split the bridge inverter into a combination of buck and boost circuits, and establish a bond graph model of the bridge inverter;

[0077] (2) Based on the bond graph model of the bridge inverter, the model predictive control of the bridge inverter is realized.

[0078] The step (1) specifically comprises the following steps in order:

[0079] (1a) Split the bridge inverter into a combination of buck and boost circuits:

[0080] The bridge arm is split into a forward buck circuit and a reverse boost circuit. For the forward buck circuit, its input is the potential difference u between points P and N. PN , the output is the potential difference u between points A and N AN , for the reverse boost circuit, its input is u AN , the output is u PN Since the forward buck and reverse boost circuits operate simultaneously in CCM mode, the inductor L 1-1 , L 1-2 It is equivalent to parallel connection, so the equivalent inductance of the bridge inverter is L / 2. Figure 1 , Figure 2 As can be seen from the figure, the switch S and the diode D are equivalent to switches that are turned on alternately, which are named SD bonds. There are two forms: buck and boost. For these two forms of SD bonds, they can be described using a unified representation method, such as Figure 3 As shown, the relationship is:

[0081]

[0082] u1=swu6+(1-sw)u5

[0083] Where, u1, u5, u6 are the voltages of the first, fifth, and sixth keys respectively, i1, i5, i6 are the currents of the first, fifth, and sixth keys respectively; sw is the driving signal of switch S;

[0084] (1b) The bond graph model of the bridge inverter is established based on the SD bonding in the buck and boost forms. The bond graph of the single-phase bridge arm of the bridge inverter can be obtained as follows: Figure 4 As shown, the control signals of the two SD bonds are complementary. Figure 1 The two L in the figure, let the L of the buck bridge arm be L 1-1 , the L of the boost bridge arm is L 1-2 Since the bridge inverter uses an LCL filter, L 1-1 and L 1-2 Equivalent to the inductance on the inverter side.

[0085] The step (2) specifically comprises the following steps in order:

[0086] (2a) Establish the state equation of the bridge inverter system:

[0087] The 0 node is a common potential junction, which is used to connect variables with equal voltage in the circuit. Figure 4 The bottom 0 node has:

[0088] u2=u 11 =u 13 =u 14

[0089] i 14 =i2+i 11 -i 13

[0090] Node 1 is a common current node, which is used to connect variables with equal current in the circuit. For the bottom node 1, we have:

[0091] i 13 =i 15 =i 16

[0092] u 15 =u 13 -u 16

[0093] according to Figure 4 Establish the state equation, and the input variable u(t) is:

[0094] u(t)=sw(t)

[0095] Where sw(t) is the driving signal of switch S at time t;

[0096] The state variable x(t) is:

[0097]

[0098] Where q is the capacitor charge and the capacitor current Capacitor voltage The capacitor voltage p is the inductor flux, and the inductor voltage Inductor Current Then the first differential of the state variable is:

[0099]

[0100] In the formula, i 14 、u 15 , u3 and u 10 That is i Cf 、u L2 、u L1-1 and u L1-2 ;

[0101] u 14 、i 15 , i3 and i 10 It is expressed as:

[0102]

[0103] On the load side, there is u 16 =Ri 16 =Ri 13 =Ri 15 , for the bottom node, we have:

[0104]

[0105] For the bottom 0 node, we have:

[0106]

[0107] in, but:

[0108]

[0109] The above formula is the first state variable differential term i 14 That is, the expression for the current of the 14th bond;

[0110] For the bottom 0 node, due to u 13 =u 14 ,but

[0111]

[0112] The above formula is the second state variable differential term u 15 That is, the expression for the current of the 15th bond;

[0113] For the leftmost node, we have

[0114] u3=-u2-swu6

[0115] in, u6=u 12 =U in , then the above formula is converted to:

[0116]

[0117] The above formula is the expression of the third state variable differential term u3, which is the voltage of the third key;

[0118] For the rightmost node, we have:

[0119] u 10 =-u 11 -swu7

[0120] in, u7=u 12 =U in, then the above formula is converted to:

[0121]

[0122] The above formula is the fourth state variable differential term u 10 That is, the expression of the voltage of the 10th key. Combining the above formulas, we can get:

[0123]

[0124] Written in matrix form:

[0125]

[0126] in:

[0127]

[0128] In the formula, i2, i3, i 10 、i 11 、i 13 、i 14 、i 15 、i 16 are the currents of the second, third, tenth, eleventh, thirteenth, fourteenth, fifteenth, and sixteenth keys respectively; u2, u6, u7, u 11 、u 12 、u 13 、u 14 、u 16 are the voltages of the second, sixth, seventh, eleventh, twelfth, thirteenth, fourteenth and sixteenth buttons respectively; sw is the driving signal of switch S; R, C, L are the equivalent load of the circuit, the output filter capacitor and the boost inductor respectively; Uin is the input voltage of the BOOST circuit; A and B are the coefficient matrices; q 14 is the capacitance C f The charge, p 15 ,p3,p 10 They are inductors L2 and L 1-1 , L 1-2 The magnetic flux, L 1-1 and L 1-2 are equivalent inductances, C f is the capacitor in the LCL filter, L2 is the inductance on the AC side of the LCL filter, i cf is the current flowing through capacitor C f The current, U L2 , U L1-1 , U L1-2 They are inductors L2 and L 1-1 , L 1-2 The terminal voltage of

[0129] (2b) Discretize the state equation, determine the output equation and cost function, and the discretization can be obtained:

[0130]

[0131] Output equation

[0132] Then the cost function J is set to:

[0133]

[0134] J is the error integral form;

[0135] Where Ts is the calculation period;

[0136] (2c) Perform model predictive control:

[0137] like Figure 5 As shown, the sampling u 14 、i 15 、i3、i 10 , calculate q 14 、p 15 ,p3,p 10 According to formula (1), the state variables when sw is 0 and 1 are predicted respectively, so as to calculate p 15 Substitute the predicted value into formula (2) and take the sw corresponding to the minimum J to achieve model predictive control.

[0138] In summary, the present invention is applicable to both single-phase and three-phase inverters, the switching state of each phase bridge arm does not affect the switching state of the other two phase bridge arms, and each phase bridge arm can work independently, realizing three-phase independent control and decoupling of each phase current; the present invention only determines whether the switching state of the power switch tube is open or closed in the next calculation cycle through the state equation, that is, the input quantity of the state equation is only one switching signal, and its value is only 0 and 1. There is no need to traverse the switch vector, and the calculation amount is small, especially for the inverter system under digital control, the data processing requirements for the main controller are low, and the control system is simplified; the proposed bond graph model can model the power switching device, the constructed model has no error, and the control accuracy based on the model is high, which can be further applied to circuits with switching devices such as switching power supplies and power electronic converters, expanding the application field of the bond graph and providing a new method for circuit control with switching devices.

Claims

1. A bond graph model predictive control method for a full-bridge inverter, characterized in that: The method comprises the following steps in order: (1) Split the bridge inverter into a combination of buck and boost circuits, and establish a bond graph model of the bridge inverter; (2) Based on the bond graph model of the bridge inverter, the model predictive control of the bridge inverter is realized.

2. The bond graph model predictive control method for a full-bridge inverter according to claim 1, characterized in that: The step (1) specifically comprises the following steps in order: (1a) Split the bridge inverter into a combination of buck and boost circuits: The bridge arm is split into a forward buck circuit and a reverse boost circuit. For the forward buck circuit, its input is the potential difference u between points P and N. PN , the output is the potential difference u between points A and N AN , for the reverse boost circuit, its input is u AN , the output is u PN Since the forward buck and reverse boost circuits operate simultaneously in CCM mode, the inductor L 1-1 , L 1-2 It is equivalent to parallel connection, so the equivalent inductance of the bridge inverter is L / 2. The switch S and the diode D are equivalent to the switches that are alternately turned on, which are named SD bonding. There are two forms: buck and boost. For these two forms of SD bonding, the relationship is: u1=swu6+(1-sw)u5 Where, u1, u5, u6 are the voltages of the first, fifth, and sixth keys respectively, i1, i5, i6 are the currents of the first, fifth, and sixth keys respectively; sw is the driving signal of switch S; (1b) The bonding graph model of the bridge inverter is established based on the SD bonding in the form of buck and boost.

3. The bond graph model predictive control method for a full-bridge inverter according to claim 1, characterized in that: The step (2) specifically comprises the following steps in order: (2a) Establish the state equation of the bridge inverter system: The 0 node is a common potential junction, which is used to connect variables with equal voltage in the circuit. For the bottom 0 node, we have: u2=u 11 =u 13 =u 14 i 14 =i2+i 11 -i 13 Node 1 is a common current node, which is used to connect variables with equal current in the circuit. For the bottom node 1, we have: i 13 =i 15 =i 16 in 15 =in 13 -in 16 Establish the state equation, and the input variable u(t) is: u(t)=sw(t) Where sw(t) is the driving signal of switch S at time t; The state variable x(t) is: Where q is the capacitor charge and the capacitor current Capacitor voltage The capacitor voltage p is the inductor flux, and the inductor voltage Inductor Current Then the first differential of the state variable is: In the formula, i 14 、u 15 , u3 and u 10 That is i Cf 、u L2 、u L1-1 and u L1-2 ; u 14 、i 15 , i3 and i 10 It is expressed as: On the load side, there is u 16 =Ri 16 =Ri 13 =Ri 15 , for the bottom node, we have: For the bottom 0 node, we have: in, but: The above formula is the first state variable differential term i 14 That is the expression of the current of the 14th bond; for the bottom 0 node, due to u 13 =u 14 ,but The above formula is the second state variable differential term u 15 That is the expression of the current of the 15th bond; for the leftmost node, there is u3=-u2-swu6 in, u6=u 12 =U in , then the above formula is converted to: The above formula is the expression of the third state variable differential term u3, which is the voltage of the third key; For the rightmost node, we have: in 10 =-in 11 -swu7 in, u7=u 12 =U in , then the above formula is converted to: The above formula is the fourth state variable differential term u 10 That is, the expression of the voltage of the 10th key. Combining the above formulas, we can get: Written in matrix form: in: In the formula, i2, i3, i 10 、i 11 、i 13 、i 14 、i 15 、i 16 are the currents of the second, third, tenth, eleventh, thirteenth, fourteenth, fifteenth, and sixteenth keys respectively; u2, u6, u7, u 11 、u 12 、u 13 、u 14 、u 16 are the voltages of the second, sixth, seventh, eleventh, twelfth, thirteenth, fourteenth and sixteenth buttons respectively; sw is the driving signal of switch S; R, C, L are the equivalent load of the circuit, the output filter capacitor and the boost inductor respectively; Uin is the input voltage of the BOOST circuit; A and B are the coefficient matrices; q 14 is the capacitance C f The charge, p 15 ,p3,p 10 They are inductors L2 and L 1-1 , L 1-2 The magnetic flux, L 1-1 and L 1-2 are equivalent inductances, C f is the capacitor in the LCL filter, L2 is the inductance on the AC side of the LCL filter, i cf is the current flowing through capacitor C f The current, U L2 , U L1-1 , U L1-2 They are inductors L2 and L 1-1 , L 1-2 The terminal voltage of (2b) Discretize the state equation, determine the output equation and cost function, and the discretization can be obtained: Output equation Then the cost function J is set to: J is the error integral form; Where Ts is the calculation period; (2c) Perform model predictive control: Sampling u 14 、i 15 、i3、i 10 , calculate q 14 、p 15 ,p3,p 10 According to formula (1), the state variables when sw is 0 and 1 are predicted respectively, so as to calculate p 15 Substitute the predicted value into formula (2) and take the sw corresponding to the minimum J to achieve model predictive control.

Images