JP6653645B2 - Control device for DC / DC converter - Google Patents

Description

  The present invention relates to a DC / DC converter control device.

  It is conceivable to estimate a reactor current in a DC / DC converter that boosts a DC input voltage to a predetermined DC voltage and outputs the boosted DC voltage. At this time, if the reactor current can be estimated without using a current sensor, the number of current sensors can be reduced. It is also conceivable to control the DC / DC converter using model predictive control.

  Non-Patent Document 1 describes a technique for controlling a DC / DC converter including a reactor and a switching element using model predictive control using an observer and PID control. In this technique, a reactor current, which is a current flowing through the reactor, is estimated by an observer.

Takashi Nakagawa, Yasushi Liu, Tamotsu Katane, "Model Predictive Control of Boost DC-DC Converter and Its Implementation," IEEJ National Convention 4-003, IEEE Japan, 2013, P.S. 4-5

  By the way, as a DC / DC converter, a first switching element on the positive bus side is provided so as to be able to both increase and output a DC input voltage to a predetermined DC voltage and output the same. , And a second switching element on the negative electrode bus side in some cases. In this case, one end of the reactor is connected to a connection point between the first switching element and the second switching element. When the technique described in Non-Patent Document 1 is applied to control of a DC / DC converter including two switching elements and a reactor, a voltage error may occur due to a dead time during which the two switching elements are simultaneously turned off. There is. This voltage error may degrade transient response performance. Further, due to the dead time described above, there is a possibility that the reactor current and the output voltage of the DC / DC converter cannot be accurately calculated.

  An object of the DC / DC converter control device of the present invention is to efficiently and accurately estimate a reactor current in a DC / DC converter including two switching elements and a reactor, and to reduce a rise time of a transient response. This is to realize control for achieving both reduction of the overshoot amount.

  One aspect of the present invention is a DC / DC including a first switching element and a second switching element, and a reactor having both ends connected to a connection point between the first switching element and the second switching element and a DC power supply. A controller for a converter, comprising: a PI controller for performing a proportional-integral calculation of a deviation between a command value and a measurement value of an output voltage of the DC / DC converter; and a next duty ratio for controlling an output voltage of the DC / DC converter. And an on-time of the first switching element, an on-time of the second switching element, and a dead time, based on the duty ratio of the first switching element and the triangular wave comparison PWM. A switching time calculator, a measured value of a DC power supply voltage and an output voltage, and an output voltage of the DC / DC converter. And an observer for estimating a reactor current and the output voltage using the ON time of the first switching element, the ON time of the second switching element, the dead time time obtained by the switching time calculator, and a state equation. Wherein the model predictive controller uses, as an inductance of the reactor used in a state equation, an inductance having a dependency on a reactor current estimated value, and when the first switching element is on, and (2) A value related to the predicted value of the reactor current or the output voltage and the output of the PI controller when another state equation is used when the switching element is on and the duty ratio is changed. Is calculated as an evaluation function, and when the evaluation function is minimized, The duty ratio is determined as the next duty ratio that is a DC / DC converter of the controller.

  According to the present invention, in a DC / DC converter including two switching elements and a reactor, it is possible to efficiently and accurately estimate a reactor current, and to achieve both a reduction in a rise time of a transient response and a reduction in an overshoot amount. The desired control can be realized.

FIG. 1 is a diagram illustrating a basic configuration of a motor drive device including a DC / DC converter control device according to an embodiment of the present invention. FIG. 3 is a diagram illustrating a control function configuration of a control device of the DC / DC converter according to the embodiment of the present invention. FIG. 3 is a diagram illustrating an entire configuration of an observer illustrated in FIG. 2. FIG. 4 is a diagram illustrating an example of triangular wave comparison pulse width modulation according to the embodiment of the present invention. FIG. 4 is a diagram illustrating a configuration of first to fourth observers illustrated in FIG. 3. FIG. 4 is a diagram illustrating a configuration of a fifth observer illustrated in FIG. 3. FIG. 4 is a diagram illustrating a configuration of second and fourth observers illustrated in FIG. 3. It is a figure showing an example of current dependence in inductance of a reactor in an embodiment of the invention. FIG. 3 is a diagram illustrating a schematic configuration of a model prediction controller illustrated in FIG. 2. FIG. 10 is a diagram illustrating a detailed configuration of a model prediction controller illustrated in FIG. 9. It is a figure which shows another example of the control function structure of the control apparatus of the DC / DC converter in embodiment of this invention. FIG. 12 is a diagram illustrating a schematic configuration of a model prediction controller in the configuration of FIG. 11. FIG. 13 is a diagram illustrating a detailed configuration of a model prediction controller illustrated in FIG. 12.

<First embodiment>
FIG. 1 shows a basic configuration of a motor drive device 100 including a DC / DC converter control device 30 according to the embodiment. The motor drive device 100 includes a DC power supply 10, a DC / DC converter 11, a low-voltage capacitor 17, a high-voltage capacitor 18, and a load 104. The DC / DC converter 11 has a reactor 12, a first switching element 14, and a second switching element 16. The first switching element 14 corresponds to an upper switching element, and the second switching element 16 corresponds to a lower switching element. The load 104 has an inverter 105 and a motor 106 connected to the inverter 105 and driven by the inverter 105. The motor 106 is a three-phase motor driven by U-phase, V-phase, and W-phase three-phase AC currents.

  One end of a reactor 12 is connected to a positive electrode of the DC power supply 10, and a connection point C of one end of a first switching element 14 and one end of a second switching element 16 is connected to the other end of the reactor 12. The other end of the first switching element 14 is connected via a positive electrode bus 19 to the positive electrode side of an inverter 105 constituting the load 104. The other end of second switching element 16 is connected via a negative electrode bus 20 to the negative electrode of DC power supply 10 and the negative electrode side of inverter 105. The low-voltage side capacitor 17 is connected between one end of the reactor 12 and the positive electrode of the DC power supply 10 and the negative electrode bus 20 on the input side of the DC / DC converter 11, and is used for smoothing the voltage. The high-voltage capacitor 18 is connected between the positive bus 19 and the negative bus 20 on the output side of the DC / DC converter 11, and is used for smoothing the voltage.

  In the embodiment, the first switching element 14 and the second switching element 16 are NPN transistors. The first switching element 14 has a collector on the positive bus 19 side and an emitter on the reactor 12 side. The second switching element 16 has a collector on the reactor 12 side and an emitter on the negative electrode bus 20 side. A freewheeling diode is connected to each of the first switching element 14 and the second switching element 16 in parallel.

In the DC / DC converter 11, the first switching element 14 is turned off and the second switching element 16 is turned on, so that a reactor current i _L flows from the positive electrode to the negative electrode of the DC power supply 10 via the reactor 12. . As a result, energy is stored in the reactor 12. Next, by turning off second switching element 16, reactor current i _L is cut off, and a voltage higher than the voltage of DC power supply 10 (power supply voltage v _b ) is generated at the end of reactor 12. Then, a current corresponding thereto is charged high pressure side capacitor 18 flows toward the positive bus line 19 the capacitor voltage v _c and rises. The capacitor voltage v _c is applied to the load 104. When the first switching element 14 is turned on, a reactor current i _L flows from the high-voltage side capacitor 18 to the positive electrode of the DC power supply 10. Thus, the capacitor voltage v _c is reduced. The output voltage of the DC / DC converter 11, namely capacitor voltage v _c is determined by the duty ratio indicating the ON percentage of the first switching element 14 for one period of the carrier signal.

  In the DC / DC converter 11, the ON / OFF state of each of the switching elements 14 and 16 is controlled by the control device 30. FIG. 2 is a diagram illustrating a control function configuration of the control device 30 of the DC / DC converter 11 in the embodiment.

The control device 30 includes a subtractor 31, a PI controller 32, an observer 34, a model prediction controller (MPC) 50, and a limiter 60. As shown in FIG. 1, the control unit 30, the measured value of the capacitor voltage v _c is inputted from the voltage sensor 21, the measured value of the supply voltage v _b is input from the voltage sensor 22. Further, the control device 30 outputs the U-phase coil current i _u among the coil currents i _u , i _v , and i _w that are the output current of the inverter 105 and flow through the U-phase, V-phase, and W-phase stator coils of the motor 106. And the measured value of the W-phase coil current i _w are input by the current sensor 24. Since the sum of the three-phase coil currents is 0, control device 30 obtains the remaining V-phase coil current _iv by calculation from the measured values of U-phase coil current i _u and W-phase coil current i _w . An angle sensor 26 such as a resolver is attached to the motor 106. The angle sensor 26 measures the rotation angle θ of the rotor constituting the motor 106 and outputs the measured value to the control device 30. The control device 30 calculates the rotation speed of the rotor constituting the motor 106 from the measured value of the rotation angle θ.

As shown in FIG. 2, the subtracter 31 calculates a deviation between the command value v _c ^* and the measured value v _c of the capacitor voltage (v _c ^* -v _c). The deviation is input to the PI controller 32. Here, the command value v _c ^* is the capacitor voltage is set based on the required output P of the motor 106, that is, set as the capacitor voltage value necessary to drive the motor at the required output P. For example, when the motor drive device 100 (FIG. 1) is mounted on the vehicle, the required output P is calculated from a measured value of the operation amount of the accelerator pedal, a calculated value of the rotation speed of the rotor, and the like.

The PI controller 32 calculates the reactor current command value i _L ^* by performing a PI operation, which is a proportional integral operation, on the input deviation. This command value i _L ^* is input to a model prediction controller 50 described later.

The observer 34, the capacitor voltage measurement v _c, the power supply voltage measured value v _b, and the calculated value i _m of the DC / DC converter output current as the load current is input. Output current calculation value i _m, for example is calculated from the motor power and the capacitor voltage measurements v _c consumed by the motor 106. The motor power is calculated, for example, from the measured values of the three-phase coil currents i _u , i _v , i _{w and} the measured values of a voltage sensor (not shown) that measures the voltage between the different two-phase lines. The motor power may be calculated from the required output P.

Observer 34, as will be described in detail later, the capacitor voltage measurement v _c, the power supply voltage measured value v _b, the output current calculation value i _m, and the reactor current from the current duty ratio d (k) i _L and capacitor voltage v Estimate _c . In the following, the estimated values in the figure are indicated by superscript wavy lines (tilde). Also, in the following description, a value used in the current control cycle is denoted by (k), and a value used in the next control cycle is denoted by (k + 1). The control device 30 controls the switching of the DC / DC converter 11 for each control cycle.

Model predictive controller 50, the reactor current command value i _L ^*, a power supply voltage measurement v _b, the output current calculation value i _m, reactor current estimated value i _L (tilde), the capacitor voltage estimated value v _c (tilde) and the current From the duty ratio d (k), the next duty ratio d (k + 1) before the restriction is calculated. The current duty ratio d (k) is calculated as d (k + 1) in the previous control cycle, and is a duty ratio used for switching the switching elements 14 and 16 in the current control cycle. The duty ratio is determined by using a value represented by 10 bits or a predetermined conversion formula from a value represented by 10 bits so that the duty ratio calculated by the model prediction controller 50 in FIG. Use the value converted in.

  The next duty ratio d (k + 1) is used for switching the switching elements 14 and 16 in the next control cycle, and is a duty ratio for controlling the output voltage of the DC / DC converter 11. The duty ratio before the restriction is input to the limiter 60. The limiter 60 limits the input duty ratio to fall between a preset upper limit and lower limit. For example, the output taken from the DC power supply 10 can be increased as the duty ratio decreases in a certain range, that is, as the step-up rate is changed in a direction of increasing the boosting rate. The output drops. For this reason, the duty ratio calculated and output by the model prediction controller 50 is limited so as to be in a range of a preset upper limit and a lower limit so as to particularly limit the lower limit. Control device 30 controls DC / DC converter 11 using the next duty ratio d (k + 1) output from limiter 60. Hereinafter, the configurations of the observer 34 and the model prediction controller 50 will be described in detail.

FIG. 3 is a diagram showing the overall configuration of the observer 34 shown in FIG. Observer 34 estimates reactor current i _L. The observer 34 includes a switching time calculator 35, a first observer 36, a second observer 37, a third observer 38, a fourth observer 39, and a fifth observer 40.

When the switching time calculator 35 receives an input of the set value of the duty ratio of the DC / DC converter 11 from the outside, the set times ΔT ₁ and ΔT ₃ which are the switching times of the first switching element 14 and the second switching element 16. , ΔT ₅ and dead time periods ΔT ₂ , ΔT ₄ . First to fifth observer 36-40, time _{ΔT 1, ΔT 2, ΔT 3} , ΔT 4, and [Delta] T _5, the capacitor voltage measurement v _c, and the power supply voltage measured value v _b, and the output current calculation value i _m The estimated reactor current value i _L (tilde) and the estimated capacitor voltage value v _c (tilde) are calculated based on With reference to FIG. 4, the calculation of the switching time in the switching time calculator 35 will be described.

The switching time calculator 35 calculates the ON time of the first switching element 14, the ON time of the second switching element 16, and the dead time based on the duty ratio of the first switching element 14 and the triangular wave comparison PWM. Specifically, the switching time calculator 35 outputs a triangular wave having a predetermined period (shown by a thick solid line in FIG. 4), a triangular wave delayed by the dead time (a triangular wave delayed by a dead time), and a duty ratio d (FIG. 4). ) To determine a pulse width modulation signal (PWM signal). The triangular wave delayed by the dead time is indicated by a dotted line in FIG. The pulse width modulation signal is for controlling the ON / OFF state of the first switching element (upper element) 14 and the second switching element (lower element) 16. Here, the time from the lower point of the triangular wave until the first switching element 14 is switched from the on-state to the off-state, that is, the time during which the triangular wave has a duty ratio d or less is defined as a set time ΔT ₁ . Further, after the set time ΔT _{1 has} elapsed, a period from when the first switching element 14 is turned off to when the second switching element 16 is turned on is referred to as a dead time time ΔT ₂ . The dead time ΔT ₂ corresponds to a time lag between the triangular wave at the duty ratio d and the triangular wave delayed by the dead time. After the dead time time ΔT _{2 has} elapsed, the time during which the second switching element 16 is in the ON state, that is, the time when the triangular wave exceeds the duty ratio is defined as the set time ΔT ₃ . Further, after the set time ΔT _{3 has} elapsed, a period from when the second switching element 16 is turned off to when the first switching element 14 is turned on is referred to as a dead time time ΔT ₄ . The dead time ΔT ₄ corresponds to a time difference between the triangular wave at the duty ratio d and the triangular wave delayed by the dead time. Further, after the dead time time ΔT _{4 has} elapsed, the time from when the first switching element 14 is turned on to when the second switching element 16 is turned off, that is, until the lower point of the triangular wave is reached is set to the set time ΔT ₅ And The switching time calculator 35 calculates a PWM signal so that the ON time of the first and second switching elements 14 and 16 according to the input duty ratio is obtained, and sets the corresponding set times ΔT ₁ , ΔT ₃ and ΔT _5. And dead time periods ΔT ₂ and ΔT ₄ .

The first switching element 14 and the second switching element 16 of the DC / DC converter 11 are controlled by a PWM signal determined by set times ΔT ₁ , ΔT ₃ , ΔT ₅ and dead time times ΔT ₂ , ΔT ₄ . Further, the switching time calculator 35 outputs the set times ΔT ₁ , ΔT ₃ , ΔT ₅ and the dead time times ΔT ₂ , ΔT ₄ to first to fifth observers 36 to 40 described later, respectively.

Returning to FIG. 3, the observer 34, the power supply voltage v _b is sampled in synchronization with the cycle of the carrier signal, receiving an input of the measurement value of the output current i _m and capacitor voltage v _c (corresponding to the output voltage). Then, the observer 34 estimates the reactor current at the set time ΔT ₁ , dead time time ΔT ₂ , set time ΔT ₃ , dead time time ΔT ₄ and set time ΔT ₅ set by the switching time calculator 35 from the input. Find the value i _L (tilde). In the embodiment, receives an input of the measurement value of the supply voltage is sampled at a controlled timing in synchronism with the bottom point of the triangular wave v _b, the output current i _m and capacitor voltage v _c in FIG. 4, the reactor current i for each control timing _Find an estimate of _L.

When the first switching element 14 is in the ON state, the state equation of the DC / DC converter 11 is represented by Expression (1). Here, r _L indicates a resistance component of the reactor 12. Note that, in the following, the time differential value is indicated by adding a superscript point (dot).

When the bilinear transformation is applied to Expression (1) to discretize it, Expression (2) can be expressed.

Further, coefficients A, B and C can be defined as in equation (3) with respect to equation (2).

On the other hand, when the second switching element 16 is in the ON state, the state equation of the DC / DC converter 11 is represented by Expression (4).

When bilinear transformation is applied to Expression (4) for discretization, Expression (5) can be expressed.

In addition, coefficients A, B, and C can be defined as in equation (6) with respect to equation (5).

  Since the state equations in the control device 30 are expressed as Expressions (1) to (6), the first observer 36, the second observer 37, the third observer 38, and the fourth observer 39 are configured as shown in FIG. Configuration. Further, the fifth observer 40 has a configuration as shown in FIG.

On the other hand, as shown in FIG. 7, the second observer 37 that estimates the dead time time ΔT ₂ includes two observers 41 and 42 and a changeover switch 44 therein. Observer 41 is the observer the first switching element 14 performs estimation of the reactor current i _L and capacitor voltage v _c using the state equation of the on state. Observer 42 is the observer second switching element 16 is an estimate of the reactor current i _L and capacitor voltage v _c using the state equation of the on state. When the estimated value of the input reactor current i _L is a positive current (a direction flowing out of the positive electrode of the DC power supply 10), the first switching element 14 is equivalent to the ON state. Thus, to estimate the reactor current i _L and capacitor voltage v _c using an observer 41 switches the changeover switch 44 to the observer 41 side. When the estimated value of the input reactor current i _L is a negative current (in the direction of flowing into the positive electrode of the DC power supply 10), the second switching element 16 is on and the first switching element 14 is off. Thus, the changeover switch 44 is switched to the observer 42 side, and the reactor current i _L and the capacitor voltage vc are estimated using the observer 42. The fourth observer 39 that estimates the dead time ΔT ₄ may have the same configuration.

Further, when using the state equations of equations (1) to (6), the inductance of reactor 12 actually changes with reactor current i _L. Thus, the preset relational expression, the estimate i _L (tilde estimate i _L (tilde), or the reactor current estimated in the previous observer 36 to 39 of the previous reactor current estimated by the fifth observer 40 ) Is used to calculate the inductance L. In the calculation using the state equations of the equations (1) to (6), the inductance L calculated using the estimated value i _L (tilde) of the reactor current is used. Thereby, controllability can be further improved. FIG. 8 shows an example of the current dependence of the inductance of the reactor in the embodiment. In FIG. 8, the inductance is normalized by setting the inductance when the reactor current is 0 A to 1. The reactor current is shown by normalizing the maximum current to 1. As described above, as the inductance L of the reactor used in the state equation, an inductance having a dependency on the reactor current estimated value is used, and is changed according to the reactor current.

Returning to FIG. 3, the first observer 36 has a time ΔT ₁ during which the first switching element 14 is in the on state from the lower point of the triangular wave until the first switching element 14 is switched from the on state to the off state. Is entered. The first observer 36 receives an input of time [Delta] T _1, T the time [Delta] T ₁ state equation with the first switching element 14 is turned on (Equation (2) and Equation (3)), v _c (t ) to assign a measurement of the capacitor voltage v _c. The first observer 36, the state equation (Equation (2) and Equation (3)) measurements of the power supply voltage v _b to v _b (t) of the i _m (t) the measured value of the output current i _m substitute. Further, in order to obtain the inductance L of the state equation, the estimated value i _L (tilde) of the reactor current i _L previously estimated by the fifth observer 40 is used. Thus, the first observer 36, and outputs the estimated estimation value of the reactor current i _L i _L (tilde) ([Delta] T ₁₎ and capacitor voltage v estimated value of _c v _c (tilde) (ΔT _1).

To the second observer 37, a dead time time ΔT ₂ from the time when the first switching element 14 is turned off to the time when the second switching element 16 is turned on after the lapse of the time ΔT ₁ is input. When receiving the input of the time ΔT ₂ , the second observer 37 receives _{one of} the observers 41 and 42 according to the direction of the estimated value i _L (tilde) (ΔT ₁ ) of the reactor current iL input from the first observer 36. Select When the observer 41 is selected, the time ΔT ₂ in the state equation (Equations (2) and (3)) when the first switching element 14 is in the ON state, and the first observer 36 in the state equation (v _c (t)). substituting the estimated value v _c of the capacitor voltage v _c (tilde) (ΔT _1). Moreover, the observer 41 substitutes the measured value of the output current i _m to the state equation (Equation (2) and Equation (3)) measurements of the power supply voltage v _b to v _b (t) of, i _m (t) . Further, in order to obtain the inductance L of the state equation, an estimated value i _L (tilde) of the reactor current i _L estimated by the first observer 36 is used. Thus, the second observer 37, and outputs the estimated reactor current i _L of the estimated value i _L (tilde) ([Delta] T ₂₎ and the estimated value v _c of the capacitor voltage vc (tilde) (ΔT _2).

On the other hand, if the observer 42 is selected, the first observer on the state equation with the second switching element 16 is turned on (Equation (5) and Equation (6)) time T of the [Delta] T _2, v _c (t) estimated value of the capacitor voltage v _c in 36 v substituting _c (tilde) (ΔT _1). Moreover, the observer 42 substitutes the measured value of the output current i _m to the state equation (Equation (5) and Equation (6)) measurements of the power supply voltage v _b to v _b (t) of, i _m (t) . Further, in order to obtain the inductance L of the state equation, an estimated value i _L (tilde) of the reactor current i _L estimated by the first observer 36 is used. Thus, the second observer 37, and outputs the estimated reactor current i _L of the estimated value i _L (tilde) ([Delta] T ₂₎ and capacitor voltage v estimated value of _c v _c (tilde) (ΔT _2).

To the third observer 38, a time ΔT ₃ during which the second switching element 16 is in the ON state after the dead time time ΔT _{2 has} elapsed is input. The third observer 38 receives an input of time [Delta] T _3, the equation of state when the second switching element 16 is turned on (Equation (5) and Equation (6)) time T of the [Delta] T _3, v _c (t ) to assign an estimate v _c of the capacitor voltage v _c in the second observer 37 (tilde) ([Delta] T _2). The third observer 38, the state equation (Equation (5) and Equation (6)) measurements of the power supply voltage v _b to v _b (t) of the i _m (t) the measured value of the output current i _m substitute. Further, in order to obtain the inductance L of the state equation, an estimated value i _L (tilde) of the reactor current i _L estimated by the second observer 37 is used. Thus, the third observer 38 estimates i _L (tilde) ([Delta] T ₃₎ of the reactor current i _L and outputs the estimated estimation value v _c of the capacitor voltage v _c (tilde) ([Delta] T _3).

The dead time period ΔT ₄ from when the second switching element 16 is turned off to when the first switching element 14 is turned on after the lapse of the time ΔT ₃ is input to the fourth observer 39. When the fourth observer 39 receives the input of the time ΔT ₄ , any one of the observers 41 and 42 according to the direction of the estimated value i _L (tilde) (ΔT ₃ ) of the reactor current i _L input from the third observer 38. Or choose. When the observer 41 is selected, the time ΔT ₄ in the state equation (Equations (2) and (3)) when the first switching element 14 is in the ON state, and the third observer 38 in the state equation v _c (t). substituting the estimated value v _c of the capacitor voltage v _c (tilde) (ΔT _3). Moreover, the observer 41 substitutes the measured value of the output current i _m to the state equation (Equation (2) and Equation (3)) measurements of the power supply voltage v _b to v _b (t) of, i _m (t) . Further, in order to obtain the inductance L of the state equation, the estimated value i _L (tilde) of the reactor current i _L estimated by the third observer 38 is used. Thus, the fourth observer 39, and outputs the estimated estimation value of the reactor current i _L i _L (tilde) ([Delta] T ₄₎ and capacitor voltage v estimated value of _c v _c (tilde) ([Delta] T _4).

On the other hand, if the observer 42 is selected, the third observer equation of state when the second switching element 16 is turned on (Equation (5) and Equation (6)) time T of the [Delta] T _4, v _c (t) estimated value of the capacitor voltage v _c in 38 v substituting _c (tilde) (ΔT _3). Moreover, the observer 42 substitutes the measured value of the output current i _m to the state equation (Equation (5) and Equation (6)) measurements of the power supply voltage v _b to v _b (t) of, i _m (t) . Further, in order to obtain the inductance L of the state equation, the estimated value i _L (tilde) of the reactor current i _L estimated by the third observer 38 is used. Thus, the fourth observer 39, and outputs the estimated estimation value of the reactor current i _L i _L (tilde) ([Delta] T ₄₎ and capacitor voltage v estimated value of _c v _c (tilde) ([Delta] T _4).

The fifth observer 40, after the dead time period [Delta] T ₄ has elapsed, the first switching element 14 and the second switching element 16 is in the ON state time [Delta] T ₅ until it is in the OFF state is input. The fifth observer 40 receives an input of time [Delta] T _5, T the time [Delta] T ₅ of the state equation with the first switching element 14 is turned on (Equation (2) and Equation (3)), v _c (t ) to assign the estimated value of the capacitor voltage v _c in the fourth observer 39 v _c (tilde) ([Delta] T _4). A fifth observer 40, the state equation (Equation (2) and Equation (3)) measurements of the power supply voltage v _b to v _b (t) of the i _m (t) the measured value of the output current i _m substitute. Further, in order to obtain the inductance L of the state equation, the estimated value i _L (tilde) of the reactor current i _L estimated by the fourth observer 39 is used. Further feeding back the previous estimate v _c (tilde) value obtained by multiplying the observer gain k _1, k ₂ to ([Delta] T ₅₎ of the capacitor voltage v _c in the fifth observer 40 last. Thus, the fifth observer 40, and outputs the estimated estimation value of the new reactor current i _L i _L (tilde) ([Delta] T ₅₎ and capacitor voltage v estimated value of _c v _c (tilde) ([Delta] T ₅₎ .

Such treatment is possible to estimate the estimated value i _L of the reactor current i _L at each time ΔT ₁ ~ΔT ₅ (tilde) (T) and the capacitor voltage v estimated value of _c v _c (tilde) (T) it can. Estimate v _c of the estimated value i _L (tilde) and capacitor voltage v _c of the reactor current i _L estimated by the fifth observer 40 (tilde) is input to the model predictive controller 50 (FIG. 2).

  FIG. 9 shows a schematic configuration of the model prediction controller 50 shown in FIG. FIG. 10 shows a detailed configuration of the model prediction controller 50 shown in FIG. As shown in FIGS. 9 and 10, the model prediction controller 50 has a plurality of switching time calculators 51 and a prediction control main unit (MPC main unit) 52. In FIG. 9, only one switching time calculator 51 is shown for simplification. In FIG. 10, the parts other than the switching time calculator 51 constitute the prediction control main unit 52. The prediction control main unit 52 includes a plurality of adders 53, a plurality of prediction calculators 54, a plurality of evaluation function calculators 55, and an optimum duty ratio setting unit 56.

The function of each switching time calculator 51 is the same as that of the switching time calculator 35 shown in FIG. Specifically, the switching time calculator 51 calculates the PWM signal so that the ON time of the first switching element 14 and the second switching element 16 according to the input duty ratio is obtained, and the corresponding time ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , and ΔT ₅ are set.

  At this time, as shown in FIG. 10, different duty ratios are input to the plurality of switching time calculators 51. First, based on the current duty ratio d (k), a plurality of adders 53 add a predetermined value or subtract a predetermined value, and obtain a plurality of duty ratios different by a predetermined value. Is input to the switching time calculator 51. The duty ratio input to the switching time calculator 51 corresponds to a temporary duty ratio.

  In the model prediction controller 50 shown in FIG. 10, the duty ratio is represented by a value of 0 to 1023, which is a number that can be represented by 10 bits. In this case, when the duty ratio is 1023, it means that the first switching element 14 is turned on in one control cycle. When the duty ratio is 0, it means that the first switching element 14 is turned off in one control cycle and the second switching element 16 is turned on in one control cycle.

  In the following, the duty ratio at which the evaluation function is minimized in the range from (d (k) -64) to (d (k) +64) is defined as the next duty ratio based on the current duty ratio d (k). A case where the ratio is determined as d (k + 1) will be described. The reason for setting the width to the reference value d (k) to 64 as the range in which the duty ratio is changed is that when the dead time is 5 μs and the PWM control cycle is 100 μs, the ratio of the dead time in one control cycle is 10 bits. This is to make the value larger than the value represented by. Specifically, since 5 (μs) / 100 (μs) × 1023 = 51, the width is set to 64 so as to be a numerical value larger than 51 and a small numerical value so as to reduce the operation load as much as possible. .

The evaluation function is the square of the deviation between the predicted value of the reactor current and a value related to the output of the PI controller 32. In the following, the evaluation function J [d (k) + a] is defined by a reactor current prediction value predicted by the model prediction controller 50 and a reactor current command value output from the PI controller 32, as shown in Expression (7). Let it be the square of the deviation from i _L ^* .

  In Expression (7), a is an arbitrary integer. For example, in the example described below, a is set to any one integer from -64 to +64.

Hereinafter, the switching time calculator 51 to which the current duty ratio d (k) is input as the duty ratio among the plurality of switching time calculators 51 is referred to as a switching time calculator A0. A switching time calculator 51 to which d (k) + i is input as a duty ratio is referred to as a switching time calculator Ai, and a switching time calculator 51 to which d (k) -i is input as a duty ratio is referred to as a switching time calculator. It is designated as vessel-Ai. i is any one of the values 1 to 64. Since the current duty ratio d (k) is input to the switching time calculator A0, the times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ and ΔT ₅ are set as in the case of the observer 34 shown in FIG. I do. The set time is input to the corresponding prediction calculator 54.

  The same number as the switching time calculators 51 is set for the plurality of prediction calculators 54. The prediction calculator 54 corresponding to the switching time calculator A0 is referred to as a prediction calculator B0. Further, the prediction calculator 54 corresponding to the switching time calculator Ai is denoted by Bi, and the prediction calculator 54 corresponding to the switching time calculator -Ai is denoted by -Bi.

Since the duty ratio d (k) + i is input to the switching time calculator Ai, the switching time calculator Ai sets the times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , ΔT ₅ according to the duty ratio. The set time is input to the corresponding prediction operation unit Bi.

Since the switching time calculator-A1 receives the duty ratio d (k) -i, the switching time calculator-A1 sets the times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , ΔT ₅ according to the duty ratio. The set time is input to the corresponding prediction operation unit -Bi.

The prediction calculator 54 calculates a predicted value of the reactor current according to the input set times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , and ΔT ₅ . Note that, in the following, the predicted values in the figure are indicated by superscript angle lines (hats).

The prediction operation unit B0 receives the set times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , ΔT ₅ according to the duty ratio d (k). Prediction calculator B0, the time settings entered, in accordance with the output current calculation value i _m, the power supply voltage measured value v _b, reactor current estimated value i _L (tilde), and capacitor voltage estimated value v _c (tilde), Similar to the observer 34 in FIG. 3, the reactor current prediction value at the time ΔT ₅ is calculated.

Hereinafter, of the reactor current predicted values, the reactor current predicted value based on the current duty ratio d (k) is indicated as i _L (hat) [d (k)]. The reactor current predicted value based on the duty ratio d (k) + i is indicated as i _L (hat) [d (k) + i], and the reactor current predicted value based on the duty ratio d (k) −i is represented by i _L (Hat) [d (k) -i].

The calculated reactor current predicted values i _L (hat) [d (k)], i _L (hat) [d (k) + i], and i _L (hat) [d (k) −i] are calculated by multiple evaluations. Each of the function calculators 55 is input to the corresponding evaluation function calculator 55.

  The same number of the plurality of evaluation function computing units 55 as the number of the prediction computing units 54 are set. The evaluation function calculator 55 corresponding to the prediction calculator B0 is referred to as an evaluation function calculator C0. The evaluation function operation unit 55 corresponding to the prediction operation unit Bi is indicated by Ci, and the evaluation function operation unit 55 corresponding to the prediction operation unit -Bi is indicated by -Ci.

The evaluation function calculator 55 inputs the predicted reactor current value i _L (hat) [d (k)], i _L (hat) [d (k) + i], i _L (hat) [d (k) − i] and the reactor current command value iL ^* , the evaluation function J [d (k) + a] of the above equation (7) is calculated. The calculated evaluation function J [d (k) + a] is input to the optimum duty ratio setting unit 56.

  The optimum duty ratio setting unit 56 receives the evaluation function J [d (k) + a] from all the evaluation function calculators 55 and selects the smallest evaluation function from the input evaluation functions. Then, the optimum duty ratio setting section 56 outputs the duty ratio (duty ratio before restriction) d (k + 1) used in the calculation of the selected evaluation function to the limiter 60 (FIG. 2) as the optimum duty ratio. The limiter 60 limits the optimal duty ratio to the optimal upper and lower limits as described above. The control device 30 generates a switching signal (SW signal) for controlling the first switching element 14 and the second switching element 16 of the DC / DC converter 11 according to the optimum duty ratio after the restriction, and Output to converter 11. The switching of the first switching element 14 and the second switching element 16 of the DC / DC converter 11 is controlled by the input switching signal. As a result, the power supply voltage is boosted and output to the load 104. Alternatively, the voltage generated at the load 104 is stepped down and output to the DC power supply 10, and the DC power supply 10 is charged. For example, when the motor drive device 100 is mounted on a vehicle and is used to drive wheels, the motor 106 generates regenerative power during braking of the vehicle, and the voltage generated by the regenerative power is applied to the DC / DC converter 11. The voltage is reduced and supplied to the DC power supply 10.

According to the control apparatus 30 of the above-described DC / DC converter, the DC / DC converter 11, without using a current sensor for measuring the reactor current i _L, the reactor current i _L efficiently and can be accurately estimated. In addition, since the optimum duty ratio can be set from the predicted calculation value of the reactor current when the duty ratio is changed, it is possible to realize control that can achieve both a reduction in the rise time of the transient response and a reduction in the amount of overshoot.

In a DC / DC converter including two switching elements, a dead time occurs. In the embodiment, the DC / DC converter can be accurately controlled in consideration of a reactor current and an output voltage in the dead time. On the other hand, in the configuration described in Non-Patent Document 1, since there is only one switching element and there is no free wheel diode, there is no need to provide a dead time for switching. By simply combining the configuration of Non-Patent Document 1 with the configuration of the DC / DC converter including two switching elements, it is not possible to accurately calculate the average current value of the reactor calculated by the observer due to the dead time. For this reason, the performance of the transient response deteriorates. In the evaluation function of the expression (3) in Non-Patent Document 1, d∞ uses the battery voltage E, and changes more slowly than the reactor current. Further, (3) the ││i [k + 1] -Ir││ ² used in the evaluation function of equation only selected becomes close duty ratio from the average value of the current of the reactor current. Thus, in the configuration of Non-Patent Document 1, the transient response is hardly improved. In embodiments, in a configuration in which a dead time is provided by the connection of the return diode comprises two switching elements 14 and 16, the error of the estimated value of the reactor current i _L and capacitor voltage v _c can be reduced.

  Note that the method of creating the switching time of the triangular wave comparison PWM is not limited to the method using the triangular wave delayed by the dead time shown in FIG. The embodiment is applicable as long as the ON time and the dead time of the first switching element 14 and the second switching element 16 can be calculated.

<Second embodiment>
FIG. 11 is a diagram illustrating another example of the control function configuration of the DC / DC converter control device 30 according to the embodiment. FIG. 12 is a diagram showing a schematic configuration of the model prediction controller 50 in the configuration of FIG. FIG. 13 is a diagram showing a detailed configuration of the model prediction controller 50 shown in FIG.

In the configuration of another example shown in FIG. 11, in the configuration shown in FIGS. 1 to 10, the PI controller 32 subtracts the deviation (vc * −vc) between the command value vc * of the capacitor voltage and the measured value vc. 31 is input. Then, the PI controller 32 calculates a capacitor voltage command change value e by performing a PI operation, which is a proportional integral operation, on the input deviation, and outputs it to the second adder 57. The second adder 57, the capacitor voltage command value v _c ^*, by adding the capacitor voltage command change value e output from the PI controller 32 to calculate the corrected capacitor voltage instruction value v _c ^**. Corrected capacitor voltage instruction value v _c ^** is as a new capacitor voltage command value, is output from the second adder 57 to the model predictive controller 50. The capacitor voltage command change value e may be positive, negative, or zero. Corrected capacitor voltage instruction value v _c ^** is input to the model predictive controller 50. The configuration of the observer 34 shown in FIG. 11 is the same as that of the observer 34 shown in FIGS. 1 to 10.

Model predictive controller 50 is corrected capacitor voltage instruction value v _c ^**, the power supply voltage measured value v _b, the output current calculation value i _m, reactor current estimated value i _L (tilde), the capacitor voltage estimated value v _c (tilde ) And the current duty ratio d (k), the next duty ratio before the restriction is calculated. The calculated next duty ratio before the restriction is output to the limiter 60.

12, as shown in FIG. 13, the prediction control body portion 52 constituting a model predictive controller 50 is corrected capacitor voltage instruction value v _c ^**, the power supply voltage measured value v _b, the output current calculation value i _m , A reactor current estimated value i _L (tilde) and a capacitor voltage estimated value v _c (tilde) are input. Further, the prediction control main unit 52 receives from the switching time calculator 51 the set times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , ΔT ₅ according to the current duty ratio d (k).

More specifically, as shown in FIG. 13, the plurality of prediction calculators 54 constituting the prediction control main body unit 52 receive, from the corresponding switching time calculator 51, the set times ΔT ₁ , ΔT ₂ , ΔT ₃ , ΔT ₄ , and ΔT ₅ are input. Prediction operator 54, the time settings entered, in accordance with the output current calculation value i _m, the power supply voltage measured value v _b, reactor current estimated value i _L (tilde), and capacitor voltage estimated value v _c (tilde), Similarly to the observer 34 of FIG. 3, the capacitor voltage predicted value at the time ΔT ₅ is calculated.

In the following, of the capacitor voltage prediction values, the capacitor voltage prediction value based on the current duty ratio d (k) is indicated as _vc (hat) [d (k)]. Also, a predicted capacitor voltage value based on the duty ratio d (k) + i is denoted by _vc (hat) [d (k) + i], and a predicted capacitor voltage value based on the duty ratio d (k) -i is denoted by _vc (Hat) [d (k) -i].

Calculated capacitor voltage estimated value v _c (hat) [d (k)], v c ( hat) [d (k) + i ], v c ( hat) [d (k) -i], the plurality of evaluation Each of the function calculators 55 is input to the corresponding evaluation function calculator 55.

Evaluation function calculator 55, the input capacitor voltage estimated value v _c (hat) [d (k)], v c ( hat) [d (k) + i ], v c ( hat) [d (k) - i] and the corrected capacitor voltage command value v _c ^** , an evaluation function J [d (k) + a] is calculated as shown in Expression (8).

In Expression (8), a is an arbitrary integer. For example, in the example shown in FIG. 13, a is set to any one integer from -64 to +64. The calculated evaluation function J [d (k) + a] is input to the optimum duty ratio setting unit 56. The evaluation function J [d (k) + a ] , the prediction value v _c of the capacitor voltage (hat) [d (k)], v c ( hat) [d (k) + i ], v c ( hat) [d ( k) -i] and the square of the deviation between the value related to the output of the PI controller 32.

  The optimum duty ratio setting unit 56 receives the evaluation function J [d (k) + a] from all the evaluation function calculators 55 and selects the smallest evaluation function from the input evaluation functions. Then, the optimum duty ratio setting unit 56 outputs the duty ratio (duty ratio before restriction) d (k + 1) used in the calculation of the selected evaluation function to the limiter 60 (FIG. 11) as the optimum duty ratio. The limiter 60 limits the optimal duty ratio d (k + 1) to an optimal upper limit and lower limit. Control device 30 controls the switching of DC / DC converter 11 (FIG. 1) according to the limited optimum duty ratio.

In the case of the control device 30 of the above-described DC / DC converter, without using a current sensor for measuring the reactor current i _L, the reactor current i _L efficiently, and it can be accurately estimated. In addition, the optimum duty ratio can be set from the predicted calculated value of the capacitor voltage when the duty ratio is changed, and control can be realized in which both the rise time of the transient response and the overshoot amount can be reduced. Other configurations and operations are the same as the configurations in FIGS. 1 to 10.

  In the above embodiment, the equations (1) and (4) are discretized using the bilinear transformation. However, the present invention is not limited to this. The zero-order hold, the forward difference, the backward Discretization may be performed using the difference.

  As described above, according to the embodiment of the present invention, the estimation error of the reactor current can be improved by using an observer to which a different state equation is applied between the ON state of the first switching element 14 and the ON state of the second switching element 16. Can be suppressed. In addition, when the estimated value of the reactor current during the dead time is positive and negative, the state equation in which the first switching element 14 is on and the state equation in which the second switching element 16 is on are applied to the reactor current, respectively. An estimation error can be suppressed. Further, by feeding back the estimated value of the capacitor voltage, the estimation accuracy of the reactor current can be improved. Further, it is possible to realize control capable of achieving both a reduction in the rise time of the transient response and a reduction in the amount of overshoot.

  In the embodiment, the same dimension observer is used, but a minimum dimension observer may be applied.

  Reference Signs List 10 DC power supply, 11 DC / DC converter, 12 reactor, 14 first switching element, 16 second switching element, 17 low voltage side capacitor, 18 high voltage side capacitor, 19 positive electrode bus, 20 negative electrode bus, 21, 22 voltage sensor, 24 Current sensor, 26 angle sensor, 30 controller, 31 subtractor, 32 PI controller, 34 observer, 35 switching time calculator, 36 first observer, 37 second observer, 38 third observer, 39 fourth observer, 40 Fifth observer, 41, 42 observer, 44 changeover switch, 50 model predictive controller, 51 switching time calculator, 52 predictive control main body, 53 adder, 54 predictive calculator, 55 evaluation function calculator, 56 optimal duty ratio Setting unit, 57 second adder, 6 0 limiter, 100 motor drive, 104 load, 105 inverter, 106 motor.

Claims (5)

A control device for a DC / DC converter including a first switching element and a second switching element, and a reactor having both ends connected to a connection point of the first switching element and the second switching element and a DC power supply,
A PI controller that performs a proportional-plus-integral operation on a deviation between a command value and a measured value of the output voltage of the DC / DC converter;
A model prediction controller that outputs a next duty ratio for controlling an output voltage of the DC / DC converter;
A switching time calculator that calculates an ON time of the first switching element, an ON time of the second switching element, and a dead time based on the duty ratio of the first switching element and the triangular wave comparison PWM;
The measured values of the DC power supply voltage and the output voltage, the output current of the DC / DC converter, the ON time of the first switching element, the ON time of the second switching element, and the dead time obtained by the switching time calculator. Using a state equation and an observer for estimating a reactor current and the output voltage,
The model predictive controller,
As an inductance of the reactor used in the equation of state, an inductance having a dependency on a reactor current estimated value is used, and different states are used when the first switching element is on and when the second switching element is on. Using an equation and calculating the square of the deviation between the predicted value of the reactor current or the output voltage and the value related to the output of the PI controller when the duty ratio is changed, as an evaluation function. A DC / DC converter control device that determines the duty ratio when the evaluation function is minimized as the next duty ratio. The control device for a DC / DC converter according to claim 1,
The output of the PI controller is a reactor current command value,
The model predictive controller,
In the case where the duty ratio is changed, the square of the deviation between the reactor current predicted value and the reactor current command value is calculated as the evaluation function, and the duty ratio when the evaluation function is minimized is calculated. A DC / DC converter control device that determines the duty ratio for the next time. The control device for a DC / DC converter according to claim 1,
The output of the PI controller is a change value of the output voltage command,
An adder that adds a change value of the output voltage command to the output voltage command value,
The model predictive controller,
When the duty ratio is changed, the square of the deviation between the predicted value of the output voltage and the output of the adder is calculated as the evaluation function, and the duty ratio when the evaluation function is minimized is calculated as A DC / DC converter control device that determines the duty ratio for the next time. The control device for a DC / DC converter according to claim 2 or 3,
A control device for a DC / DC converter, wherein the next duty ratio is limited by optimal upper and lower limits, and the output voltage is controlled by the limited next duty ratio. The control device for a DC / DC converter according to any one of claims 1 to 4,
The model predictive controller calculates the on-time of the first switching element, the on-time of the second switching element, and the dead time, which are used when calculating the predicted value of the reactor current or the output voltage. A plurality of predicted values of the reactor current or a plurality of predicted values of the output voltage are set according to the plurality of provisional duty ratios obtained by adding or subtracting a predetermined value from the duty ratio. A control device for a DC / DC converter to calculate.

Images