Method for controlling a converter connected to an electric machine
By using an LC filter and an optimized voltage pulse sequence in the DC-AC converter, the problems of motor terminal overvoltage and inverter losses were solved, achieving optimized voltage control and reduced losses.
Patent Information
- Application Number
- CN201910112569.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-02-13
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2039-02-13
AI Technical Summary
Existing technologies for controlling DC-AC converters suffer from leakage current, inverter losses, and motor stator winding overvoltage problems caused by large voltage variations, which are particularly difficult to solve effectively when long cable connections are involved.
By using an LC filter and an optimized voltage pulse sequence control method, incremental and decremental voltage pulses are generated to limit overvoltage across motor terminals, avoid the use of damping resistors, and optimize switching operations to reduce losses and voltage variations.
It effectively limits overvoltage at motor terminals and voltage variations at cable inputs, reduces inverter losses, and optimizes voltage control of cable connections.
Smart Images

Figure CN110165967B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method for controlling a converter connected to an electric machine and a control system suitable for implementing the method. Background Art
[0002] DC-AC converters (also known as inverters) are designed to control electric motors. It is known that when switching the inverter's transistors, large voltage variations (dv / dt) can occur. This can cause leakage currents, losses in the inverter, and overvoltages in the windings of the motor's stator, especially if the machine is connected to the converter via long cables.
[0003] To overcome these problems, one solution consists in placing a dv / dt filter comprising a number of passive components, namely an inductor, a capacitor and a damping resistor, at the output of the inverter. However, this solution requires the use of resistors which dissipate power.
[0004] Another solution involves using a converter with at least three voltage levels. The problem then lies in generating a first voltage pulse to half the voltage (E / 2) and another pulse with an amplitude of E / 2 to reach voltage E. The second pulse is generated with a delay twice the propagation time of the cable. However, this sequence cannot be easily implemented using a two-stage converter.
[0005] In order to monitor the voltage variation (dv / dt) on each switch operation in a two-stage converter, Juhamatii Korhonen, Juha-Pekka et al., Proceedings of the 35th Annual Conference of the IEEE Industrial Electronics Society, IECON 2009 Juho Tyser, Pertti Silventoinen, Hannu Sarén, and Kimmo Rauma's publication, "Control of an Inverter Output-Active Du / Dt Filtering Method," IECON 2009, proposes generating a sequence of multiple consecutive pulses at each switching transition of the inverter's power switches. This solution avoids the use of damping resistors. However, this solution is not always satisfactory when it comes to achieving the desired voltage at the cable input and, therefore, at the cable output (i.e., across the motor terminals).
[0006] An object of the present invention is therefore to provide a method for optimizing the control applied to a converter, which method makes it possible to limit overvoltages across the terminals of an electric machine.
[0007] The solution of the invention may also be suitable for minimizing the losses due to the switches in the inverter and / or for minimizing the variations in the voltage at the input of the cable. Summary of the Invention
[0008] This object is achieved by a method for controlling a converter connected to a motor via a cable via a filter, the cable comprising at least two conductors and exhibiting an impedance, the filter being an LC filter and comprising at least one inductor and one capacitor for each conductor of the cable, the converter comprising at least two switching arms, each switching arm comprising at least two power switches, the power switches being switched on and off to chop a DC voltage having a value E so as to transmit it at at least two voltage levels, the switching arms being controlled so as to generate voltage pulses on the output phases connected thereto, the voltage pulses potentially being increasing voltage pulses and comprising rising vertical voltage edges or decreasing voltage pulses and comprising falling vertical voltage edges,
[0009] The method comprises determining at least one sequence of a plurality of voltage pulses forming a square wave signal to be applied to each conductor at each switching transition of two power switches of one and the same switching leg, thereby minimizing an overvoltage level across terminals of a motor (M), the sequence comprising a number 2N of consecutive pulses, N being greater than or equal to 1, each pulse being defined by a unique rank n, the sequence being generated such that each increasing voltage pulse of rank n exhibits a pulse width that is the same as a pulse width of a decreasing voltage pulse of rank equal to 2N+1-n, and the method comprising the steps of:
[0010] - determining the number of consecutive pulses of said sequence;
[0011] - determining the width of each pulse of the sequence suitable for minimizing the level of overvoltage across the terminals of the electric machine (M) while satisfying the following relationship:
[0012] t N =t p +t r / 2
[0013] in:
[0014] -t N The moment in time corresponding to the end of a pulse of rank N;
[0015] -t p corresponds to the propagation time of the cable;
[0016] -t r Corresponds to the rise time for the voltage appearing at the input of the cable to reach 50% of said value E of the DC voltage.
[0017] According to one feature, the method comprises the step of determining the overvoltage level based on the relationship:
[0018]
[0019] in:
[0020] -Int Vo corresponding to said overvoltage level resulting from calculation of the integral of the voltage Vo across the terminals of the electric machine;
[0021] -L corresponds to the inductance of the filter's inductor;
[0022] -C corresponds to the capacitance of the filter capacitor;
[0023] -Zc corresponds to the characteristic impedance of the cable.
[0024] According to another feature, the method comprises at least one step of comparing the overvoltage level determined for the pulse sequence with a threshold value.
[0025] According to another feature, the method comprises a step of increasing the number of pulses in the sequence as soon as the obtained overvoltage level exceeds the threshold value.
[0026] According to another feature, the method comprises the step of determining the rise time for the voltage at the input of the cable to reach 50% of said value E of DC voltage according to the following relationship:
[0027] t r =f(L,C,Zc,d1,d2…d N )
[0028] in:
[0029] -L corresponds to the inductance of the filter's inductor;
[0030] -C corresponds to the capacitance of the filter capacitor;
[0031] -Zc corresponds to the characteristic impedance of the cable;
[0032] -d1, d2...d N The width of each pulse corresponds to rank n of the sequence of pulses.
[0033] According to another feature, the method comprises the steps of inputting or identifying the following parameters:
[0034] -L, corresponds to the inductance of the filter's inductor;
[0035] -C, corresponds to the capacitance of the filter capacitor;
[0036] -Zc, corresponding to the characteristic impedance of the cable;
[0037] -tp, corresponds to the propagation time of the cable.
[0038] The invention also relates to a system for controlling a converter connected to an electric machine by a cable via a filter, the cable comprising at least two conductors and exhibiting an impedance, the filter being an LC filter and comprising at least one inductor and one capacitor for each conductor of the cable, the converter comprising at least two switching arms, each switching arm comprising at least two power switches, the power switches being switched on and off to chop a DC voltage having a value E for delivery at at least two voltage levels, the switching arms being controlled so as to generate voltage pulses on the output phases connected thereto, the voltage pulses potentially being increasing voltage pulses and comprising a rising vertical voltage edge or a decreasing voltage pulse and including a falling vertical voltage edge, the system being adapted to: determine at least one sequence of a plurality of voltage pulses forming a square wave signal to be applied to each conductor at each switch transition of two power switches of one and the same switching leg, thereby minimizing an overvoltage level across the terminals of the motor (M), the sequence comprising a number 2N of consecutive pulses, N being greater than or equal to 1, each pulse being defined by a unique rank n, the sequence being generated such that each increasing voltage pulse of rank n exhibits a pulse width that is the same as a pulse width of a decreasing voltage pulse of rank equal to 2N+1-n, and the system being further adapted to:
[0039] - determining the number of consecutive pulses of said sequence;
[0040] - determining the width of each pulse of the sequence suitable for minimizing the level of overvoltage across the terminals of the electric machine (M) while satisfying the following relationship:
[0041] t N =t p +t r / 2
[0042] in:
[0043] -t N The moment in time corresponding to the end of a pulse of rank N;
[0044] -t p corresponds to the propagation time of the cable;
[0045] -t r Corresponds to the rise time for the voltage appearing at the input of the cable to reach 50% of said value E of the DC voltage.
[0046] According to one feature, the system is adapted to determine the overvoltage level based on the following relationship:
[0047]
[0048] in:
[0049] -Int Vo corresponding to said overvoltage level resulting from calculation of the integral of the voltage Vo across the terminals of the electric machine;
[0050] -L corresponds to the inductance of the filter's inductor;
[0051] -C corresponds to the capacitance of the filter capacitor;
[0052] -Zc corresponds to the characteristic impedance of the cable.
[0053] According to another feature, the system is suitable for comparing the overvoltage level determined for the pulse sequence with a threshold value.
[0054] According to another feature, the system is suitable for increasing the number of pulses in the sequence as long as the obtained overvoltage level exceeds the threshold value.
[0055] According to another feature, the system is suitable for determining the rise time for the voltage at the input of the cable (Vi) to reach 50% of said value E of DC voltage based on the following relationship:
[0056] t r =f(L,C,Zc,d1,d2…d N )
[0057] in:
[0058] -L corresponds to the inductance of the filter's inductor;
[0059] -C corresponds to the capacitance of the filter capacitor;
[0060] -Zc corresponds to the characteristic impedance of the cable;
[0061] -d1, d2...d N The width of each pulse corresponding to rank n of the pulse sequence.
[0062] According to another feature, the system comprises means for inputting or identifying the following parameters:
[0063] -L, corresponds to the inductance of the filter's inductor;
[0064] -C, corresponds to the capacitance of the filter capacitor;
[0065] -Zc, corresponding to the characteristic impedance of the cable;
[0066] -tp, corresponds to the propagation time of the cable.
[0067] The solution described in the prior publication cited above achieves pulses of always the same width. To achieve this, it relies on the characteristics of the filter (the values of the inductance and capacitance of the capacitors). Since the cable is connected in parallel with the capacitors of the resonant filter, the input impedance (Zi) of the cable is parallel to the capacitors of the filter.
[0068] In this case, a first solution is to choose a capacitor that exhibits a very large capacitance, so that a low impedance relative to the impedance of the cable can be applied and the impedance of the cable can therefore be neglected. This makes it easier to determine the pulse width for the chosen mode (number of pulses and pulse type) by knowing only the value of the inductance and the value of the capacitance of the capacitor. The resulting voltage variation (dv / dt) is set to the propagation time of the cable (t p ) to limit the overvoltage at the ends of the cable and therefore across the terminals of the motor. However, this solution has certain limitations:
[0069] 1) Beyond a given cable length, the input impedance Zi becomes non-negligible (Zi depends on the propagation time t p impedance);
[0070] 2) The capacitance of the filter capacitor becomes very high, resulting in very large peak currents, greater than the peak current that a single cable can carry. The losses in the converter's power switches become very high.
[0071] Another approach involves considering the cable's elements (t p , Zc) (Zc is the characteristic impedance of the cable, equivalent to resistance) and an LC resonant filter. The capacitance value of the resonant filter's capacitor can be selected independently of the cable. The impedance of the filter's capacitor can be higher than the input impedance Zi of the cable connected to the filter. Peak currents are reduced, and therefore, the losses generated in the converter's power components are lower compared to the solutions proposed above. However, this freedom in selecting the capacitor's capacitance value is only possible under the condition that a specific optimized pulse sequence is applied. Therefore, the issue lies in the optimization sequence generated by the method of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0072] Other features and advantages will become apparent from the following detailed description, which is provided with reference to the accompanying drawings, in which:
[0073] - Figure 1 shows an arrangement comprising an inverter-type converter connected to an electrical machine (e.g. an electric motor);
[0074] - Figure 2A A single phase of the converter is shown in a simplified manner, said phase being connected to the motor, and Figure 2B Show the equivalent circuit diagram;
[0075] - Figure 3 Several diagrams illustrating the principles of the present invention are shown schematically;
[0076] - Figure 4 schematically illustrates a diagram representing an optimized pulse sequence of pulses generated according to the present invention;
[0077] - Figures 5A to 5C Several diagrams illustrating the principles of the present invention are shown schematically;
[0078] - Figure 6 Operational simulations illustrating the principles of the present invention;
[0079] - Figure 7 An example of an algorithm executed to determine an optimized pulse sequence is shown. DETAILED DESCRIPTION
[0080] The invention is applicable to a control arrangement comprising in particular a converter connected to an electric machine. Without limitation, the electric machine may be an electric motor.
[0081] The converter is a DC-AC converter, ie it is an inverter.
[0082] The inverter 200 may be incorporated into a variable speed drive.
[0083] Inverter 200 receives a DC voltage E as input. Without limitation, the DC voltage is delivered by a DC bus connected to a rectifier stage at the input and designed to convert the AC voltage provided by the grid into a DC voltage. The DC bus comprises two power lines and bus capacitors Cb1 and Cb2 connected between the two power lines to stabilize the DC voltage.
[0084] The inverter 200 may have two or three output phases U, V, W to which the motor is connected. Figure 1 , an inverter 200 with three output phases is shown.
[0085] The inverter 200 includes a plurality of switching arms connected in parallel between two power lines of a bus.
[0086] The inverter 200 may have two voltage levels or more than two voltage levels.
[0087] In an inverter with two voltage levels, each switching leg includes two power switches T1U, T2U, T1V, T2V, T1W, T2W. A unique output phase is connected to the midpoint between the two switches.
[0088] With two voltage levels, each switching arm can deliver a voltage equal to 0 or a voltage equal to E.
[0089] Without limitation, the power switch may be an IGBT, a MOSFET, or a JFET transistor.
[0090] The output phases of the inverter 200 are each connected to a distinct conductor of a cable Cx intended to connect the output of the inverter to the motor M. This cable Cx may in particular have a length Lg of several meters. It comprises certain characteristics (in particular, length Lg, characteristic impedance Zc and propagation time tp).
[0091] Recall that in cables, we define the input impedance Zi, which characterizes the frequency characteristics of the cable independent of its length, and the characteristic impedance Zc of the cable. This impedance is defined by the following expression:
[0092]
[0093] Where Lc is the inductance for a given length of cable and Cc is the capacitance for the same given length of cable between two phases.
[0094] For times less than 2*tp (the propagation time of the cable), the impedance Zi is close to the impedance Zc.
[0095] The control arrangement comprises a filter F connected to the output of the inverter 200. This filter is an LC filter and advantageously does not include a damping resistor. For each output phase, the filter comprises an inductor Lf (of inductance L) and a capacitor Cf (of capacitance C) connected in series on the output phase, the capacitor Cf being connected to the inductor and the power line of the bus.
[0096] The control arrangement further comprises a control system 100 intended to generate control commands to be applied to the power switches of the inverter 200 and to send said control commands to the units for driving the power switches. The control system 100 comprises at least one microprocessor and memory storage means.
[0097] As is conventional, the control system 100 is intended to generate control commands for the power switches of the switching arms.
[0098] To simplify, Figure 2A An arrangement with a single phase connected to a motor M is shown.
[0099] On each switching arm, two switches are controlled in a complementary manner, ie when one of the switches is in the closed state, the other switch is in the open state, and vice versa.
[0100] As control is applied to the switching arm with two levels, the DC voltage is chopped between the two voltage levels. The first voltage level is equal to 0V, and the second voltage level is equal to a DC voltage equal to E.
[0101] Controlling the switching arm allows the generation of a voltage pulse on the output phase connected to it. The voltage pulse can be:
[0102] - an increasing voltage pulse and comprising a rising vertical voltage edge between 0 V and a value E, followed by a plateau of non-zero duration at the value E;
[0103] a decreasing voltage pulse and comprising a falling vertical voltage edge between the value E and the value 0V, followed by a plateau of non-zero duration at the value 0V.
[0104] Each pulse is defined by a start time and an end time, separated by a duration forming the width (d) of the pulse. As will be seen below, this width can vary from one pulse to the next.
[0105] As is conventional, control commands for the switches of the inverter 200 are generated by pulse width modulation (hereinafter, PWM). The control commands applied to the two switches of one and the same switching arm allow the generation of voltage pulses as outputs ( Figure 3 ).
[0106] refer to Figure 2A , define the following quantities:
[0107] -E corresponds to the DC voltage delivered as input to the inverter 200;
[0108] Vin corresponds to the voltage delivered on each output phase of the inverter 200. This voltage is made up of successive voltage pulses obtained by complementary switching of the two power switches of the switching leg in question.
[0109] - Vi corresponds to the input voltage of the cable downstream of the filter and therefore across the terminals of the capacitor Cf of the filter;
[0110] Vo corresponds to the output voltage of the cable (ie across the terminals of the motor M connected to the cable Cx). The cable may in particular have a certain length between its input and output, which will affect the operation of the arrangement;
[0111] -Ir corresponds to the resonant current flowing through the capacitor Cf of the filter F;
[0112] -IL corresponds to the current flowing through the inductor Lf of the filter F.
[0113] By using an inverter with more than two levels, it is known that the overvoltage across the terminals of the motor can be substantially reduced. A first pulse is generated at half voltage (E / 2), followed by another pulse of amplitude E / 2, which passes through the cable with twice the propagation time t pThe delay of 0 is superimposed on the first pulse (E / 2→E). This waveform is easily generated using an inverter with three voltage levels (0, E / 2 and E).
[0114] The principle of the invention is based in particular on providing a sequence of pulses that makes it possible to approximate this waveform of the voltage Vi in order to limit overvoltage oscillations at the end of the cable Cx connected to the motor. The solution of the invention is applicable regardless of the number of levels of the inverter, but it will be readily understood that it is particularly advantageous for inverters that can generate only two voltage levels.
[0115] In order to obtain a variation of the voltage Vi obeying the profiles 0, E / 2 and E, a sequence of consecutive pulses characterized by the following relationship will be used:
[0116] t N =t p +t r / 2
[0117] Among them, t N is the time defining the axis of symmetry of the sequence (see below), t p is the propagation time of the cable (a known characteristic), and t r It is the rise time of voltage Vi to half voltage (E / 2).
[0118] According to one aspect of the invention, the problem is to optimize the control applied to the inverter 200 to achieve one or more of the following goals:
[0119] - Minimize overvoltages and voltage oscillations across the terminals of the motor;
[0120] - Minimize the losses in the converter, i.e. the amplitude of the resonant current Ir affected each time the power switch is switched on;
[0121] - Minimize the variations (dv / dt) of the voltage Vi at the input of the cable.
[0122] refer to Figure 3 The principle of the present invention specifically includes generating a sequence of multiple consecutive pulses at each switching transition of the two power switches of the same switching arm. When turning on and off the power switches of the switching arm, a sequence of multiple consecutive pulses is generated instead of each rising edge and each falling edge of the voltage pulse generated in each chopping period T_mli. The term "switching transition" should be understood to mean that one of the two switches of the switching arm transitions from the open state to the closed state, and the other of the two switches of the switching arm transitions from the closed state to the open state (complementary control). For inverters with more than two voltage levels, this principle will be applied in the same way.
[0123] Each train of pulses takes the form of a square wave signal comprising a plurality of consecutive pulses, ie an increasing pulse followed by a decreasing pulse.
[0124] According to one particular aspect of the invention, the control system is dedicated to determining the optimized pulse sequence to be applied at each switching transition of the power switches of a switching leg when chopping the voltage over a chopping period T_mli.
[0125] The optimized pulse sequence determined by the system comprises an optimized number of consecutive pulses, and each pulse is characterized by a determined width that is also optimized with respect to the target.
[0126] Optimizing the pulse sequence makes it possible to limit overvoltages and oscillations of the voltage Vo across the terminals of the electric machine during the transitions.
[0127] refer to Figure 4 , the optimized pulse sequence therefore includes the following general features:
[0128] - it comprises a number 2N of successive pulses;
[0129] - an increasing pulse followed by a decreasing pulse, and vice versa;
[0130] - Each pulse is defined by a rank n, where n ranges from 1 to 2N;
[0131] Each pulse of rank n has t n-1 The start time and t n A pulse of rank 1 therefore has a start time defined by t0 and an end time defined by t1;
[0132] Each pulse of rank n has d n Indicates the pulse width; the pulse width d n Corresponding to time t n-1 With time t n the time that has passed between
[0133] -The pulse width of rank n is defined by the following relationship:
[0134]
[0135] in:
[0136] -θ n Defines the conduction angle for rank n. The value of this angle depends on the number of pulses in the sequence and the sequence of pulses applied. For example, for Figure 5B The defined sequence sum is equal to N of 2, and the conduction angle is equal to:
[0137] θ1=0.506rd, θ2=1.318rd, θ3=1.318rd, θ4=0.506rd
[0138] And for Figure 5A For a defined sequence and N equal to 2, the conduction angles will be equal to each other.
[0139]
[0140] -Fres corresponds to Figure 2B The cable impedance Zi is associated with the first resonant frequency of the filter.
[0141] The frequency Fres is the resonant frequency of the LC filter associated with the cable. The first method is to use Figure 2B Fres is determined by the equivalent circuit of the switching circuit. This circuit shows the configuration for switching one phase, with the other two phases of the converter connected via a bus. The three capacitors, denoted by the reference symbol C_cx, represent the total capacitance of the cable between each phase. The equivalent inductance of the filter is L, and the equivalent capacitance of the filter is C.
[0142] Time t1 and t2...t n is determined relative to the pulse width and is derived from the following relationship.
[0143] Time t0 is the start of the sequence.
[0144] -t n =t n-1 +d n (2)
[0145] -The pulse widths are determined independently;
[0146] - time t N corresponds to the median time of the series;
[0147] - in the same sequence, a pulse (respectively, decreasing or increasing) pulse corresponding to each (increasing or decreasing) pulse of rank n is symmetrical with respect to amplitude and with respect to duration and is defined by a rank equal to 2N+1-n; time t N An axis of symmetry is thus formed between the pulse located upstream and the pulse located downstream;
[0148] - The sequence consists of an equal number of increasing and decreasing pulses.
[0149] To determine the optimized pulse sequence, the control system 100 applies an algorithm that makes it possible to determine both the number of pulses in the sequence and the width of each pulse.
[0150] The algorithm run by system 100 is specifically based on the goal to be achieved.
[0151] Without limitation, the target may be selected from:
[0152] Minimize overvoltage and oscillations across the motor terminals, which can be expressed by the following relationship:
[0153]
[0154] in:
[0155] -Int Vo corresponds to the integral of the voltage Vo across the terminals of the electric machine;
[0156] Minimize the losses in the converter, i.e. the magnitude of the resonant current (Ir) each time the transistor is switched (Irsw). This goal can be expressed by the following relationship:
[0157]
[0158] in:
[0159] -Int Irsw The integral of the losses generated by the switch corresponds to each time the switch is switched on and off.
[0160] Minimize the ground current, that is, minimize the dv / dt of the voltage Vi. This goal can be expressed by the following relationship:
[0161]
[0162] in:
[0163] -Int dv / dt Corresponds to the integral of the variation of the voltage Vi at the input of the cable.
[0164] Advantageously, the objective pursued consists of minimizing overvoltages and oscillations across the terminals of the electric machine.This objective will be defined by a threshold value, denoted T_OverV, below which the overvoltage level must be located for the sequence to be considered valid.
[0165] Without limitation, for example Figure 7 The algorithm operation is schematically shown in FIG. The algorithm is described below by using the example of the main objective of reducing the overvoltage across the terminals of the motor when a sequence of multiple consecutive pulses is applied during each switching transition on the switching arms of the inverter. The objective is thus defined by the relationship (3) already defined above:
[0166]
[0167] in:
[0168] -Int Vo Corresponds to the integral of the voltage Vo across the terminals of the motor.
[0169] This expression defines the overvoltage level represented by OverV such that:
[0170] OverV=Int Vo (6)
[0171] Note that if the objective is different and is chosen from one of the two other objectives defined above, the algorithms will have an equivalent structure.
[0172] In order to run the algorithms, the control system comprises at least one control unit including a microprocessor and memory storage means.
[0173] According to a particular aspect of the present invention, the above relationship (3) is used to determine the switching times t1, t2, t3 . . . t 2N .
[0174] More precisely, the switching time is determined by satisfying the constraints set by equation (3). To achieve this, the following argument is applied, starting specifically with the transfer function of the cable between voltage Vi and voltage Vo. The cable can be modeled by the parameters Zi(stp), which is the input impedance of the cable, and T2(stp), which is the transfer function of the cable, both expressed as Laplace transforms.
[0175]
[0176]
[0177] The Laplace transfer function f(s) of function f(t) is defined as:
[0178]
[0179] Here, s is the Laplace transform operator and t is time.
[0180] Derive the variation of Vo with input voltage Vin in the Laplace domain:
[0181]
[0182] The voltage Vin is decomposed into 2N voltage steps (stepn) of amplitude 1 and -1, the sign of the amplitude is superimposed by the rank n, and the switching time.
[0183]
[0184] In order to meet the minimum overvoltage ripple, the value Int defined by equation (3) Vo Minimize:
[0185]
[0186]
[0187] The term h(t) is the inverse function of the Laplace transform of G(s).
[0188]
[0189] G(s) is defined by the following equation, which is the transfer function seen by the inverter output.
[0190]
[0191] Used to make Int Vo A necessary condition for minimization and thus for voltage oscillation is that the system of equations is satisfied with 2N variables.
[0192]
[0193]
[0194] Therefore, for N=3:
[0195] By virtue of the symmetry of the pulse, time t n They are related to each other through the following relationships:
[0196] t3=t N , t6=Zt N , t S =Zt N -t1, t4 = Zt N -t2
[0197] Through the above relationships, the system can be simplified into three equations.
[0198]
[0199] Use the above equations to determine t1, t2, t3, ... t 2N (Here, N=3).
[0200] Without limitation, reference Figure 7 , the algorithm for controlling the operation of the system includes the following steps:
[0201] Step E1
[0202] The first step involves identifying the characteristics of the setup. These characteristics are specifically:
[0203] - the propagation time t of the cable connecting the inverter to the machine p ;
[0204] - Characteristic impedance Zc of the cable;
[0205] - the inductance L of the filter inductor Lf;
[0206] - capacitance C of the filter capacitor Cf;
[0207] - A DC voltage E delivered as input to the inverter 200.
[0208] By way of example, the identification operation may be achieved by inputting various parameters or by automatic recognition, for example, in a learning step performed when starting the setup.
[0209] Step E2
[0210] It is a matter of recording an acceptable overvoltage threshold value T_OverV for the motor M. By way of example, this threshold value T_OverV can be input using a suitable interface.
[0211] Step E3
[0212] This marks the beginning of the sequence of determined pulses. To this end, the algorithm arbitrarily starts with a number of pulses defined by N=2 (i.e. 2N pulses in total) and Figure 5B The sequence defined in [1] minimizes the switching losses. The sequence can be started with different numbers of pulses. In any case, the algorithm allows convergence with respect to the target sequence including the optimized number N.
[0213] Step E4
[0214] In this step, the system thus determines the time t1 corresponding to the end time of the first pulse (n=1) and the time t2 corresponding to the end time of the second pulse (n=2).
[0215] The pulse width d is calculated based on the following reproduced equation (1) n :
[0216]
[0217] Among them, relative to Figure 5B , θ1=0.506rd, θ2=1.318rd, θ3=1.318rd, θ4=0.506rd.
[0218] Time t1 and t2...t n The width of the relative pulse is determined and derived from the relationship (2) reproduced below: Time t0 is the start of the sequence.
[0219] t n =t n-1 +d n (2)
[0220] The pulses of ranks 3 and 4 are symmetric with respect to amplitude and with respect to duration for pulses of ranks 2 and 1, respectively (according to the following relation (7): d n =d 2N+1-n )
[0221] Step E5
[0222] On the basis of the preceding data, it then becomes a matter of theoretically determining the overvoltage level OverV that will be generated by the previously established pulse sequence. The overvoltage level OverV is given by the relationship (6) above.
[0223] Step E6
[0224] This is a comparison step for the purpose of determining whether the overvoltage value OverV determined in step E5 is lower than or equal to a threshold value T_OverV pre-stored in the system memory.
[0225] Step E7
[0226] If the overvoltage value OverV is lower than or equal to the threshold voltage T_OverV, the pulse sequence is considered valid and the sequence can be terminated by determining the width of each pulse (especially the width of those symmetrical pulses) using the following relationship:
[0227] d n =d 2N+1-n
[0228] Step E8
[0229] If the overvoltage value OverV is higher than the threshold value T_OverV, the width of the pulse is affected. Arbitrarily, the width of the pulse of rank 1 is increased and the width of the pulse of rank 2 is reduced. The time t1 is shifted by a certain value (e.g. equal to t p / 100) to increase the width of the rank 1 pulse, and time t2 is shifted by the same value (t p / 100) to reduce the pulse width of rank 2.
[0230] Step E9
[0231] This is a test step for determining whether two pulses of rank 1 and rank 2 can be combined into a single pulse. To do so, the test consists in determining whether t2 = t1.
[0232] If the time t2 does not merge with t1 , the method returns to step E5 defined above to optimize the pulse width for N=2 until an acceptable overvoltage level is reached.
[0233] Step E10
[0234] After step E9, if time t2 is merged with time t1, the system checks the case where N=1.
[0235] Step E11
[0236] It is a matter of determining the overvoltage level OverV that may occur for N=1, the pulse of rank 1 having a width defined by the time t1 determined above.
[0237] Step E12
[0238] This is a new test performed on the overvoltage level OverV calculated in the previous step with respect to the pre-stored threshold value T_OverV.
[0239] Step E13
[0240] If the overvoltage level OverV is lower than or equal to the threshold value T_OverV, a pulse sequence of N=1 is confirmed, wherein a pulse of rank 1 is defined by the time t1.
[0241] Step E14
[0242] If the overvoltage level OverV is higher than the threshold value T_OverV, the system concludes that the number of pulses in the sequence must be increased.
[0243] The system therefore sets the number N to 3.
[0244] Step E15
[0245] The system runs a loop to determine the appropriate number of pulses. The loop first includes: determining the time t1, t2 to t N and therefore the pulse sequence, allowing the determined goal to be achieved.
[0246] Without limitation, the objectives are:
[0247] Minimize overvoltage and oscillations across the terminals of the motor, which can be expressed by the following relationship (3):
[0248]
[0249] in:
[0250] -Int Vo Corresponds to the integral of the voltage Vo across the terminals of the motor.
[0251] The system determines each time t in the pulse train n (ie, time t1, t2, t3 (when N=3, equal to t N )), allowing to achieve the target goal, i.e. to obtain a minimum overvoltage level.
[0252] Step E16
[0253] The system then determines the duration tr The duration t r corresponds to the time it takes for the voltage Vi at the input of the cable to reach the value E / 2. In the case of a three-phase converter, the voltage Vi corresponds to the composite voltage between the two phases. This duration t r The pulse width d1 to d2 is determined based on the characteristic impedance Zc of the components of the LC filter and the cable and the time calculated in step E17 above. N .
[0254] therefore:
[0255] t r =f(L, C, Zc, d1, d2...d N ) (8)
[0256] in:
[0257] d n =t n -t n-1
[0258] It should be noted that for sequence calculations the data must satisfy the following two relations in all cases:
[0259] t N =t p +t r / 2 (9)
[0260] and
[0261] t r <2t p (10)
[0262] If the equation (9) is satisfied, that is, if the voltage Vo at time t N Approaching E / 2, the sequence defined by the number N of increasing pulses and N of decreasing pulses is optimized.
[0263] In the definition of t N In the definition, the duration of the sequence (2*t N ). Alternatively, the slope of the change in voltage Vi over time (dv / dt) can also be determined by the following relationship:
[0264]
[0265] Step E17
[0266] It is a matter of determining the overvoltage level OverV that can be generated for an N-defined pulse sequence (in the first cycle, N is at least equal to 3).
[0267] Step E18
[0268] Once the sequence has been constructed, the problem is to verify that the overvoltage level generated is satisfactory. This step therefore consists in verifying that the overvoltage level obtained by means of the sequence is lower than or equal to a threshold value T_OverV.
[0269] Step E19
[0270] If the obtained overvoltage level OverV is still above the threshold T_OverV, a new sequence must be calculated. This new sequence will be generated by incrementing the value of N by one unit (N=N+1). As long as the condition regarding the obtained overvoltage level remains unfulfilled, the new sequence will be calculated by increasing the number of pulses defined by N by one unit on each cycle.
[0271] Step E20
[0272] If the level of overvoltage OverV determined based on the latest generated sequence is lower than or equal to the threshold T_OverV, the system can terminate the sequence by: determining the width d of each pulse in the sequence using the following relationship n :
[0273] d n =t n -t n-1
[0274] And consider the following relationship to ensure that with respect to time t N Symmetry of:
[0275] d n =d 2N+1-n
[0276] Step E21
[0277] This step is the end of the algorithm. In this step, the system has determined the optimal pulse sequence allowing the target objective to be achieved in step E17.
[0278] Figures 5A to 5C Various diagrams are shown to allow a better understanding of the principles of the invention.
[0279] Figure 5A A sequence of N=2 pulses is shown, where all pulses are of the same width. In this sequence, the resonant current Ir remains the same from the beginning to the end of the sequence, which is one of the goals of the sequence. However, the sequence of pulses is not optimized because the resonant current Ir is not zero during the switching transitions, resulting in switching losses.
[0280] exist Figure 5BIn , since the resonant current is zero at each switching transition, the applied sequence is optimized with respect to the previous sequence. This is possible because the width of the pulses in the sequence is optimized with the goal of reducing the switching losses defined by the above relation (4).
[0281] exist Figure 5C The proposed sequence of pulses makes it possible to minimize the overvoltage on the terminals of the motor during each switching transition on the switching arm. The rise time taken for the voltage Vi at the input of the cable to reach the value E / 2 defines the duration t r Furthermore, the sequence is characterized by the voltage Vo (dashed line) across the terminals of the machine M at time t N Close to the value E / 2, defines the symmetry of the sequence.
[0282] Figure 6 A simulation is shown which makes it possible to demonstrate the attenuation of the overvoltage effect when the sequence of applied pulses is optimized. The voltage Vo and voltage Vi curves obtained both have mean values close to the ideal solution (dashed lines).
[0283] As is apparent from the above, the control method of the present invention and the system for its implementation have a certain number of advantages, including:
[0284] - they make it possible to determine a sequence of pulses optimized with respect to a target, whatever the size of the filter F;
[0285] Even if the inverters used are two-level inverters, they allow the overvoltages across the terminals of the electric machine during switching operations to be limited as much as possible.
Claims
1. Method for controlling a converter connected to a motor through a cable (Cx) via a filter (F), the cable (Cx) comprising at least two conductors and exhibiting an impedance, the filter being an LC filter and comprising for each conductor of the cable at least one inductor (Lf) and one capacitor (Cf), the converter comprising at least two switching arms, each switching arm comprising at least two power switches, the power switches being switched on and off so as to chop a DC voltage having a value E so as to transmit it at at least two voltage levels, the switching arms being controlled so as to generate voltage pulses on the output phases connected thereto, the voltage pulses being potentially increasing voltage pulses and comprising rising vertical voltage edges or decreasing voltage pulses and comprising falling vertical voltage edges, the method being characterized in that it comprises: Determine at least one sequence of a plurality of voltage pulses forming a square wave signal to be applied to each conductor at each switch transition of two power switches of the same switching arm, thereby minimizing the overvoltage level across the terminals of the motor (M), the sequence comprising a number 2N of consecutive pulses, N being greater than or equal to 1, each pulse being defined by a unique rank n, the sequence being generated such that each increasing voltage pulse of rank n exhibits the same pulse width as the pulse width of a decreasing voltage pulse of rank equal to 2N+1-n, and in that it comprises the following steps: - determining the number of consecutive pulses of said sequence; - determining the width of each pulse of the sequence suitable for minimizing the level of overvoltage across the terminals of the electric machine (M) while satisfying the following relationship: t N =t p +t r / 2 in: -t N The moment in time corresponding to the end of a pulse of rank N; -t p corresponds to the propagation time of the cable; -t r corresponds to the rise time for the voltage present at the input of the cable (Vi) to reach 50% of said value E of DC voltage, The method is further characterized in that it comprises the step of increasing the number of pulses in the sequence once the overvoltage level obtained exceeds a threshold value.
2. The method according to claim 1, characterized in that It comprises the steps of determining the overvoltage level based on the following relationship: in: -Int Vo corresponds to said overvoltage level resulting from calculation of the integration of the voltage Vo across the terminals of the electric machine; -L corresponds to the inductance of the filter's inductor; -C corresponds to the capacitance of the capacitor of the filter; -Zc corresponds to the characteristic impedance of the cable.
3. The method according to claim 2, characterized in that It comprises at least one step of comparing the overvoltage level determined for the pulse sequence with the threshold value.
4. The method according to claim 1, characterized in that It comprises the step of determining the rise time for the voltage at the input of the cable (Vi) to reach 50% of said value E of the DC voltage based on the following relationship: t r =f(L,C,Zc,d1,d2…d N ) in: -L corresponds to the inductance of the filter's inductor; -C corresponds to the capacitance of the capacitor of the filter; -Zc corresponds to the characteristic impedance of the cable; -d1, d2...d N The width of each pulse corresponds to the rank n of the sequence of pulses.
5. The method according to claim 2, characterized in that It includes the steps of inputting or identifying the following parameters: -L, corresponds to the inductance of the filter's inductor; -C, corresponds to the capacitance of the capacitor of the filter; -Zc, corresponding to the characteristic impedance of the cable; -tp, corresponds to the propagation time of the cable.
6. System for controlling a converter connected to an electric machine through a cable (Cx) via a filter (F), the cable (Cx) comprising at least two conductors and exhibiting an impedance, the filter being an LC filter and comprising, for each conductor of the cable, at least one inductor (Lf) and one capacitor (Cf), the converter comprising at least two switching arms, each switching arm comprising at least two power switches, the power switches being switched on and off so as to chop a DC voltage having a value E so as to transmit it at at least two voltage levels, the switching arms being controlled so as to generate voltage pulses on the output phases connected thereto, the voltage pulses being potentially increasing voltage pulses and includes a rising vertical voltage edge or a decreasing voltage pulse and includes a falling vertical voltage edge, the system being characterized in that it is suitable for: determining at least one sequence of a plurality of voltage pulses forming a square wave signal to be applied to each conductor at each switch transition of two power switches of the same switching arm, thereby minimizing the overvoltage level across the terminals of the motor (M), the sequence comprising a number 2N of consecutive pulses, N being greater than or equal to 1, each pulse being defined by a unique rank n, the sequence being generated so that each increasing voltage pulse of rank n exhibits a pulse width that is the same as the pulse width of a decreasing voltage pulse of rank equal to 2N+1-n, and in that it is further suitable for: - determining the number of consecutive pulses of said sequence; - determining the width of each pulse of the sequence suitable for minimizing the level of overvoltage across the terminals of the electric machine (M) while satisfying the following relationship: t N =t p +t r / 2 in: -t N The moment in time corresponding to the end of a pulse of rank N; -t p corresponds to the propagation time of the cable; -t r corresponds to the rise time for the voltage present at the input of the cable (Vi) to reach 50% of said value E of DC voltage, The system is further characterized in that it is adapted to increase the number of pulses in the sequence as soon as the obtained overvoltage level exceeds a threshold value.
7. The system according to claim 6, characterized in that It comprises the steps of determining the overvoltage level based on the following relationship: in: -Int Vo corresponds to said overvoltage level resulting from calculation of the integration of the voltage Vo across the terminals of the electric machine; -L corresponds to the inductance of the filter's inductor; -C corresponds to the capacitance of the capacitor of the filter; -Zc corresponds to the characteristic impedance of the cable.
8. The system according to claim 7, characterized in that It is suitable for comparing the overvoltage level determined for the pulse sequence with the threshold value.
9. System according to one of claims 6 to 8, characterized in that It is suitable for: The rise time for the voltage at the input of the cable (Vi) to reach 50% of said value E of the DC voltage is determined based on the following relationship: t r =f(L,C,Zc,d1,d2…d N ) in: -L corresponds to the inductance of the filter's inductor; -C corresponds to the capacitance of the capacitor of the filter; -Zc corresponds to the characteristic impedance of the cable; -d1, d2...d N The width of each pulse corresponds to the rank n of the sequence of pulses.
10. The system according to claim 7, characterized in that It includes modules for inputting or for identifying the following parameters: -L, corresponds to the inductance of the filter's inductor; -C, corresponds to the capacitance of the capacitor of the filter; -Zc, corresponding to the characteristic impedance of the cable; -tp, corresponds to the propagation time of the cable.
