Method for dead time compensation in an inverter for electric drive

Abstract

Method for dead-time compensation in a converter (9) with at least one phase for the power supply of an electric drive (8), wherein the converter (9) has a control loop (1) for the current of the at least one phase and an observer (12) operating in parallel to the control loop (1), by which an observer actual value (13) for the current is calculated on the basis of an observer model (11) without the occurrence of a dead time, and to compensate the disturbance variable (10) caused by the dead time in the control loop (1), a compensation value for the control is formed from a control deviation (14) multiplied by a gain coefficient value (K), wherein the control deviation (14) consists of the difference between the actual current value of the control loop (1) and the observer actual value (13) of the current, characterized in that the gain coefficient value (K) depends on the control deviation (14) in a characteristic curve (16).

Description

[0001] The invention relates to a method for dead-time compensation in a converter for an electric drive. In particular, the invention relates to a method for dead-time compensation in a converter with at least one phase for the power supply of an electric drive, wherein the converter has a control loop for the current of the at least one phase and an observer operating in parallel to the control loop, by which an observer actual value for the current is calculated based on an observer model without the occurrence of a dead time, and to compensate for the disturbance caused by the dead time in the control loop, a compensation value for the control is formed from a control deviation multiplied by a gain coefficient value, wherein the control deviation consists of the difference between the actual current value of the control loop and the observer actual value of the current.

[0002] Furthermore, the invention relates to an inverter for an electric drive and a vehicle with at least one electric drive.

[0003] From DE 10 2007 032 484 A1, a control method with multi-channel feedback is known in which a feedback variable determined by sampling a controlled variable at a sampling frequency and another feedback variable determined by averaging the controlled variable over a period of time are each compared separately with a reference variable. Each of the control deviations determined in this way is fed to a separate controller, which generates an individual controller output variable from each. The controller output variable generated from the individual controller output variables is used to adjust the controlled variable so that the controlled variable follows the reference variable.

[0004] Electric drives with AC motors, or with multiple phases in the case of three-phase motors, utilize frequency converters. These converters generate single-phase AC or multi-phase three-phase AC from a DC link or a DC power source such as an electric vehicle's traction battery. Controlling the electric drives requires regulating the frequency, voltage, and current of the generated AC for each phase. This is typically achieved for each phase using two switchable half-bridges, which connect the positive and negative terminals of the DC power supply to a single output. Pulse-width modulation (PWM) is usually employed for this purpose. A transistor or FET, for example, can be used as the switching element in each half-bridge.

[0005] In such a converter, a dead time occurs during a zero crossing of the output current between the switching element of one pole's half-bridge being switched off and the switching element of the other pole's half-bridge being switched on. This is necessary to prevent a short circuit of the supply voltage across the two switching elements of the two half-bridges if their switching times overlap.

[0006] However, the resulting dead time leads to a number of disadvantages, such as undesirable distortion of the phase voltages and phase currents of the drive, a reduction in the efficiency of the drives controlled by the inverter, and instabilities in the drive control at low speeds. It also results in higher thermal loads and greater torque ripple in the controlled drives.

[0007] Electric drives are widely used in industrial trucks, for example as drive motors for the travel system, as well as actuators for electric steering systems and drives for hydraulic systems or work devices such as lifting masts. In these applications, the resulting decrease in efficiency is particularly detrimental in battery-electric industrial trucks that draw their power solely from a traction battery, as it reduces the range and maximum possible operating time. Furthermore, the instability at low speeds and the increased torque ripple of the driven motors are especially problematic in industrial trucks, since these drives are often operated at very low speeds, for example, the travel drive during slow maneuvering to pick up a load or a steering drive.

[0008] According to the state of the art, dead-time compensation is known to compensate for undesirable voltage drops in the inverter due to dead time. A known method involves adding or subtracting a certain amount from the inverter's output voltage, depending on the operating point. The values ​​to be added or subtracted can, for example, be obtained by extrapolating the measured current values.

[0009] However, a disadvantage of this method for dead-time compensation is that it results in an increase in noise, since the measurement noise, with amplification, affects the pulse width of the pulse width modulation.

[0010] Furthermore, it is known according to the state of the art to provide a Luenberger observer for the phase current of the controlled drive, so that the dead time compensation is no longer based on the actual interpolated current but on a target current calculated by the observer model.

[0011] Both dead-time compensation by adding or subtracting fixed values ​​and dead-time compensation using a model or observer of the powertrain employ differential terms or constant coefficients with very large numerical values. Ideally, these coefficients should be infinitely large to minimize or prevent the control deviation between the model and the physical powertrain. However, both approaches lead to system instabilities in the case of small deviations, because the impact of disturbances is maximized in such cases.

[0012] The present invention is therefore based on the objective of providing a method for dead time compensation in an inverter for an electric drive, an inverter for an electric drive and a forklift truck with such an electric drive, with which the aforementioned disadvantages are avoided and the most accurate possible dead time compensation is made possible.

[0013] This problem is solved by a method for dead-time compensation in an inverter for an electric drive with the features of claim 1, an inverter for an electric drive with the features of claim 7, and a vehicle with the features of claim 9. Advantageous embodiments of the invention are solved in the dependent claims.

[0014] The problem is solved according to the invention in that, in a method for dead-time compensation in a converter with at least one phase for the power supply of an electric drive, wherein the converter has a control loop for the current of the at least one phase and an observer operating in parallel to the control loop, by which an observer actual value for the current is calculated on the basis of an observer model without the occurrence of a dead time, and to compensate for the disturbance caused by the dead time in the control loop, a compensation value for the control is formed from a control deviation multiplied by a gain coefficient value, wherein the control deviation consists of the difference between the actual current value of the control loop and the observer actual value of the current, according to the invention the gain coefficient value depends on the control deviation in a characteristic curve.

[0015] This method adaptively scales the compensation based on the control deviation between the observer model and the actual current of the physical drive train. This allows for flexible amplification and enables extensive distortion correction of the phase current of an electric motor, such as an AC motor. Highly efficient and effective distortion correction of the phase voltages and phase currents of the electric drive can be achieved, even with the dead times of inverters encountered in practice. Furthermore, the method can be implemented cost-effectively using software, as inverters typically already have sufficiently powerful control computers for controlling and modulating the signal.Furthermore, the efficiency of the drive is improved, as there are no losses due to harmonics that do not contribute to the drive power, and the control of the electric drive is more stable at low speeds. The thermal load on the electric drive is also reduced, resulting in lower torque ripple.

[0016] In an advantageous embodiment of the method, the gain coefficient value can depend on at least one further parameter and the characteristic curve can be a characteristic map.

[0017] This allows for a very flexible consideration of other conditions as well.

[0018] Another advantageous parameter is the frequency of at least one phase.

[0019] If the gain coefficient value also depends on the frequency of the controlled phase, this achieves a dependence of the dead-time compensation on the speed of the electric drive and, in particular, the behavior at low speeds can be further optimized.

[0020] In a favorable embodiment of the method according to the invention, the characteristic curve forms a small gain coefficient value for small control deviations and a large gain coefficient value for large control deviations.

[0021] The increase in the gain coefficient value is advantageous in the case of large control deviations, as it is progressive.

[0022] Such nonlinear behavior makes it possible to minimize the control deviation between the model and the physical drivetrain without causing instabilities in the system in the case of small deviations.

[0023] In one embodiment of the method, the converter can have several phases and the gain coefficient value can form a matrix.

[0024] The task is also solved by an inverter with at least one phase for the power supply of an electric drive, wherein the inverter has a control loop for the current of the at least one phase and a control computer for controlling the control loop, by the control computer carrying out a previously described procedure.

[0025] The inverter according to the invention has the advantages already described above.

[0026] The inverter can have multiple phases.

[0027] The task can also be solved by a vehicle with at least one electric drive, in particular a drive system, controlled via a previously described inverter.

[0028] Especially with an electric drive as the driving mechanism, the reduced power loss and better controllability at low speeds have a particularly positive effect, as, for example, the range of a battery-electric vehicle is increased and maneuverability at low speeds is improved.

[0029] In an advantageous embodiment of the invention, the vehicle is a forklift truck.

[0030] Particular advantages arise with industrial trucks such as forklifts, reach trucks, stacker cranes, and pallet trucks, as these vehicles utilize a variety of electric drives. These can include, for example, a drive system, a hydraulic drive, a direct drive for work devices, or even electric steering. For all these drives, industrial trucks benefit from improved controllability at low speeds, and battery-electric industrial trucks, in particular, benefit from increased maximum operating time and range on a single charged traction battery.

[0031] It is also conceivable to use the inventive method and the inventive converter in general for compensating disturbances in automation systems if these disturbances can be described by linear differential equations.

[0032] Further advantages and details of the invention are explained in more detail with reference to the exemplary embodiments shown in the schematic figures. Here, Fig. 1. Schematic representation of the setup of a controlled system with a parallel observer model according to the state of the art. Fig. 2. Schematic representation of the setup of a controlled system with a parallel observer model according to the state of the art with dead-time compensation, Fig. 3 schematically an embodiment of the method according to the invention, Fig. 4 an excerpt of the scheme of Fig. 3 in detail, Fig. 5 a diagram of the phase current of an electric drive without dead-time compensation and Fig. 6 a diagram of the phase current of an electric drive with dead-time compensation according to the embodiment of the Fig. 3.

[0033] The Fig. Figure 1 schematically shows the setup of a control system 1 with a parallel observer 2 according to the state of the art. Observer 2 corresponds to a Luenberger observer 3. A control deviation between the actual value y of the control system 1 and an observer's actual value ŷ, which is calculated by observer 2 using an observer model, forms a control deviation or error feedback 4, which is multiplied by a gain coefficient value k. Both the input state values ​​u and the output state values ​​y are used for the reconstruction of the states.

[0034] The Fig. Figure 2 schematically shows the setup of a controlled system 1 with a parallel observer model 2 according to the state of the art with dead-time compensation. A control deviation between the actual value y of the controlled system 1 and an observer actual value ŷ, which is calculated by the observer 2 using an observer model, forms a control deviation or error feedback 4, which is multiplied by a gain coefficient value K and used in the controlled system 1 to compensate for a disturbance caused by the dead time, introduced by the vector st and the matrix P. A and  correspond to the system matrices of the controlled system 1 and the observer 2, respectively, which each also have input matrices B and B̂ and output matrices C and Ĉ in the schematic representation. To achieve complete compensation of the disturbance caused by the dead time, the following must be satisfied: K(x̂-x) + P * st = 0.The assumption is that the output matrix C can be considered the identity matrix in the case of vector values ​​for x and a multiphase converter, for example, if interactions between the individual phases can be neglected. This is best achieved if an inverse matrix K is used. -1 In its diagonalized form, it possesses eigenvalues ​​that approach infinity. However, this leads to the problems of an unstable system described above for small control values ​​or small control deviations.

[0035] The Fig. Figure 3 schematically shows an embodiment of the method according to the invention. A current controller 5 receives an input value via feedback 6 from an actual current value I_actual of an electric machine 7 as an electric drive 8, together with a target current value I_target. A disturbance variable 10 acts in a converter 9 due to the dead time. In an observer 12, an observer actual value of the current 13 is generated from a model 11 of the electric drive 8. The difference of this observer actual value from the feedback 6 of the actual current value I_actual forms a control deviation 14 and serves as the input value U_in for an adaptive dead time compensation 15. An output value U_out of the adaptive dead time compensation 15 serves as a correction value for the current controller 5.

[0036] The Fig. Figure 4 shows an excerpt of the scheme of Fig. 3. In detail, the adaptive dead-time compensation 15. A matrix K represents the gain coefficient value and has characteristic values ​​for the gain of the control deviation values ​​14 as eigenvalues. These characteristic values ​​depend on a characteristic curve 16, where the gain values ​​increase progressively with increasing control deviation 14. An element-wise multiplication 17 is then performed to generate the output variable U_out.

[0037] The Fig. Figure 5 shows a diagram of the phase current of an electric drive with conventional dead-time compensation over time. The current waveform exhibits a slight step and deviation from a sinusoidal waveform when passing through the zero line, resulting from disturbances caused by the dead time.

[0038] The Fig. Figure 6 shows a diagram of the phase current of an electric drive with dead-time compensation according to the embodiment of the Fig. 3 over time. compared to the representation in the Fig. 5 shows a significantly better compensation of errors due to the dead time of the inverter.

Claims

[1] Method for dead-time compensation in an inverter (9) with at least one phase for the power supply of an electric drive (8), wherein the inverter (9) has a control loop (1) for the current of the at least one phase and an observer (12) operating in parallel to the control loop (1), by which an observer actual value (13) for the current is calculated based on an observer model (11) without the occurrence of a dead time, and to compensate for the disturbance variable (10) caused by the dead time in the control loop (1), a compensation value for the control is formed from a control deviation (14) multiplied by a gain coefficient value (K), wherein the control deviation (14) consists of the difference between the actual current value of the control loop (1) and the observer actual value (13) of the current, characterized by , that the gain coefficient value (K) in a characteristic curve (16) depends on the control deviation (14). [2] Method according to claim 1, characterized by , that the gain coefficient value (K) depends on at least one other parameter and the characteristic curve (16) is a characteristic map. [3] Method according to claim 2, characterized by , that another parameter is the frequency of at least one phase. [4] Method according to any one of claims 1 to 3, characterized by , that the characteristic curve (16) results in a small gain coefficient value (K) for small control deviations (14) and a large gain coefficient value (K) for large control deviations (14). [5] Method according to claim 4, characterized by , that the increase in the gain coefficient value (K) is progressive with large control deviation (14). [6] Method according to any one of claims 1 to 5, characterized by , that the converter (9) has multiple phases and the gain coefficient value (K) forms a matrix. [7] Inverter with at least one phase for supplying power to an electric drive (8), wherein the inverter (9) has a control loop (1) for the current of the at least one phase and a control computer for controlling the control loop (1), characterized by that the control computer performs a method according to one of claims 1 to 6. [8] Inverter according to claim 7, characterized by , that the converter (9) has multiple phases. [9] Vehicle with at least one electric drive (8) controlled via an inverter (9) according to claim 7 or 8. [10] Vehicle according to claim 9, characterized by that the vehicle is a forklift truck.

Images