Electric Power Conversion Equipment

This invention relates to power conversion equipment which converts a direct current into an alternating current or vice versa, and in particular it relates to an apparatus for controlling a pulse width modulation (PWM) inverter.

An example of prior art inverter modulation methods has been disclosed in an article entitled "Evaluating the Recent Inverter Control Methods" which appeared in the "Denkisha no Kagaku", pp 14, FIG. 1, April 1993, published by the Denkisha Kenkyuukai.

As illustrated in FIG. 2 herein, in a traction motor inverter a constant control is provided to maintain a constant ratio between the magnitude and the fundamental frequency of an output voltage when the fundamental frequency of the output voltage is low (which region will be referred to as a variable-voltage variable-frequency region), whereas a frequency control is provided while maintaining a maximum voltage value in a region in which the output voltage fundamental frequency increases such that the magnitude of its output voltage becomes a maximum (which region will be referred to as a constant-voltage variable-frequency region). Since in the variable-voltage variable-frequency region the output voltage is regulated by pulse width modulation control, a multi-pulse mode is utilized in which a half cycle of its output voltage comprises a plurality of voltage pulses. On the other hand, in the constant voltage variable-frequency region, in order to maximize its voltage usage rate and to minimize the size of the equipment, a one pulse mode is utilized in which a half cycle of its output voltage comprises a single broad pulse.

A prior art inverter which uses GTO thyristors as its switching elements (referred to as a GTO inverter hereinafter), has adopted a multi-pulse mode according to a pulse count switching method which decreases the number of pulses to be included in one cycle thereof with increasing output voltage fundamental frequency, as shown in FIG. 3. This is due to an upper limitation of several hundred Hz imposed on a switching frequency in the GTO thyristor. In this method, however, a magnetic noise accompanies tone variation, thereby generating a cacophonic noise due to discontinuity between switching frequencies at the time the pulse count switching is employed.

Further, there is another problem associated with the GTO thyristor in that a voltage jump of approximately 10% occurs between a three-pulse mode output voltage (containing three pulses in a half cycle of its output voltage and a one-pulse mode output voltage) containing one pulse, depending on the limitation of a minimum OFF time in the GTO thyristor, thereby causing a fluctuation to take place in the torque generated by the traction motor at the time of switching from the three-pulse mode to the one-pulse mode.

A principal object of the invention is to provide a two-level inverter apparatus which is capable of controlling the magnitude of its output voltage from zero to a maximum value through a combination of a multi-pulse mode and a one-pulse mode, which substantially eliminates the discontinuity between the switching frequencies, thereby reducing the cacophonic magnetic noise accompanying tone variations, and which enables continuous control over substantially the entire region of the output voltage, by minimizing a gap in the output voltages between the multi-pulse mode and the one-pulse mode.

The aforementioned main object of the invention is achieved by the inverter apparatus according to the invention, which comprises: a multi-pulse generator for generating a gate control signal for outputting a bipolar modulation voltage which is controlled by pulse width modulation having a uniform pulse generation cycle over any cycle of its output voltage fundamental wave, and a gate control signal for outputting an overmodulation voltage the pulse width of which is widened in the vicinity of a peak of the output voltage fundamental wave relative to that in the vicinity of a zero cross point thereof; a one-pulse generator for generating a single pulse having the same polarity as that of the output voltage fundamental wave; and a selection circuit for selecting the output from either the multi-pulse generation means or the one pulse generation means, dependent on a condition of the number of pulses contained in one cycle of the output fundamental wave, a magnitude of an output voltage, a modulation ratio, or an output voltage fundamental frequency.

The multi-pulse generator is arranged so that a pulse generation cycle of a pulse width modulated portion on an output voltage waveform can be set independently of the output voltage fundamental frequency. Thus, its switching frequency at the time of bipolar modulation can be maintained constant, and its switching frequency at the time of overmodulation can be caused gradually to approach a predetermined switching frequency set for at the time of one-pulse, thereby eliminating discontinuity between switching frequencies.

Further, by controlling both the output voltage at which changes between the multi-pulse mode and the one-pulse mode are enabled, and the phase relative to the output voltage fundamental wave, it is possible to achieve the change between the modes smoothly, without causing fluctuations in the current and torque that the motor generates.

Other objects, advantages and novel features of the present invention will become apparent from the following detailed description of the invention when considered in conjunction with the accompanying drawings.

FIG. 1 is a schematic diagram illustrating one embodiment of the invention;

FIG. 2 is a graphic representation of operational characteristics of a traction motor inverter of the invention;

FIG. 3 is a graphic illustration of the operation of a prior art inverter modulation method;

FIG. 4 is a graphic representation of operational characteristics of an inverter of the invention;

FIG. 5 is a schematic diagram illustrating one example of a configuration of a multi-pulse generation means of the invention;

FIG. 6 shows a modulation waveform, a carrier waveform, and a switching function in a bipolar mode;

FIG. 7 shows a modulation waveform, a carrier waveform, and a switching function in an overmodulation mode;

FIG. 8 is a diagram of a fundamental waveform of an output voltage and a switching function in one pulse mode associated thereto;

FIG. 9 is a graphic illustration of a beat phenomenon;

FIG. 10 is a diagram indicating a correlation between the average pulse number in the vicinity of the zero cross point of the output voltage fundamental wave, and the current pulsation;

FIGS. 11(a) and 11(b) show that the transient current fluctuation immediately after mode change varies according to the shift phase;

FIG. 12 is a diagram showing the mode change timing in FIG. 11(a);

FIG. 13 is a diagram showing the mode change timing in FIG. 11(b);

FIG. 14 is a diagram showing the relationship between the shift phase and the transient current fluctuation immediately after mode change;

FIG. 15 is a diagram showing a shift phase set allowable region;

FIG. 16 shows an example of the PWM mode select means of the invention; and

FIG. 17 is a diagram showing the relationship between the output voltage fundamental frequency of the inverter and the switching frequency according to the invention.

Preferred embodiments of the present invention will be described in the following with reference to FIGS. 1 to 17.

A diagram which illustrates a PWM mode of inverter operation according to the invention is shown in FIG. 4. The inverter is operated in a bipolar mode in its low output voltage range, in an overmodulation mode in its high output voltage range, and in a one pulse mode in its maximum output voltage range respectively.

FIG. 1 is a schematic diagram of one embodiment of the invention, wherein numeral 6 is an induction motor, 5 is a two-level three-phase PWM inverter which drives the induction motor, 9 is an inverter power source from an aerial direct current supply, and 7, 8 denote respectively a filter reactor and a capacitor provided on the DC input side of the inverter.

A multi-pulse generator 2, a one-pulse generator 3 and a PWM mode select means 4 respectively generate inverter control signals Su, Sv, Sw (the subscript x representing any phase of u, v, w) in response to an output voltage command E.sup.* and a frequency command Fi.sup.*. The frequency command Fi.sup.* is integrated in integrator unit 1, to generate an output voltage fundamental wave phase .theta.x of each phase. The inverter control signals S1.sub.x, S2.sub.x and S.sub.x are characterized by a switching function which is defined to take 1 when the positive branch of the inverter is ON, and take 0 when the negative branch thereof is ON.

An example of the multi-pulse generator 2 is shown in FIG. 5 (which indicates only one phase component of its contents). Here, switching functions in the bipolar mode and the overmodulation mode are generated by the same generator. An output voltage command-to-modulation ratio convertor 21 generates a signal indicative of a modulation ratio A, (that is, the amplitude of a modulation wave relative to that of the carrier), from an output voltage command E*. Assuming a carrier amplitude of 1, the ratio takes on a value of 0.ltoreq.A.ltoreq.1 in the bipolar mode, and that A>1 in the overmodulation mode. So that the magnitude of the output voltage fundamental wave coincides with the voltage command the relationship between E* and A is defined by equation 1 in the bipolar mode, and by equation 2 in the overmodulation mode. ##EQU1## When the sinusoidal wave generator 22 generates an output y=sin (x), sin.theta.x of the output voltage fundamental wave phase .theta.x (equivalent to modulation phase) is obtained. The latter is multiplied by the modulation ratio A to produce a modulation wave a.sub.x, which is input to a switching function computer 24, together with a carrier frequency Fc (which is equivalent to the switching frequency in the bipolar mode). The switching function computer 24 generates a carrier of triangular pulses having an amplitude of 1 and a frequency of Fc, which is compared with the value of the modulation wave a.sub.x to generate a switching function S1.sub.x. Alternatively, the switching function may be obtained without using such triangular pulses, through computation of the modulation wave ax and its pulse intervals.

FIGS. 6 and 7 show examples of waveforms of the switching functions according to the bipolar mode and the overmodulation mode, obtained through comparison with the triangular pulses.

In the inverter equipment according to the present invention, devices such as IGBTs, large capacity transistors or the like which can be operated at switching frequencies of several kHz are utilized as a switching element (hereinafter referred to as an IGBT inverter in general), and the modulation wave and the carrier are arranged to be asynchronous in the multi-pulse mode of operation. Thereby, the switching frequency is maintained constant in the bipolar mode, and gradually approaches a predetermined switching frequency in the one pulse mode which will be described later.

Since the modulation wave and the carrier wave are asynchronous in the multi-pulse mode, it is necessary for the carrier to have a substantially higher frequency (preferably approximately ten times higher, based on experience) than the modulation frequency.

An example of a switching function waveform which is generated by the one pulse generator of FIG. 1 is shown in FIG. 8. When the sign of the fundamental wave of output voltage is positive, irrespective of its amplitude, the value of switching function S2.sub.x is 1, and when the sign is negative the value of S2.sub.x is 0.

Next, an arrangement for combining the multi-pulse mode and the one pulse mode so that the high output voltage range can be controlled smoothly will be described in the following.

A prior art overmodulation method is discussed in a publication entitled "Overmodulation control method for voltage-type three-phase PWM inverters" which appeared in the National Convention Proceedings No. 106, Heisei 3, Industrial Application Division, the Japan Society of Electric Engineers. According to this prior art, a six-step inverter operation is overextended to increase the modulation ratio in the overmodulation mode, thereby covering one pulse mode operation. However, when the one pulse mode is achieved by extending the overmodulation mode (that is, the overmodulation is realized by excessively increasing the modulation ratio) certain problems occur as follows.

Once problem is that the point at which the overmodulation mode is changed to the one pulse mode depends on its switching frequency. Thus, it cannot be set at discretion.

Another problem is that in the case where a modulation wave and a carrier wave are asynchronous in the overmodulation mode, pulses of the modulation wave in the vicinity of the zero-cross point may or may not be generated in the vicinity of a change between the overmodulation mode, and the one-pulse mode due to the influence of turn-on and turn-off time of the devices. As a result, the positive and the negative output voltages are balanced, thereby causing a beat phenomenon to be introduced by which a low frequency pulsation is superimposed on a load current of the inverter.

A third problem is that with reference to FIG. 7, an output voltage waveform (which is equivalent to the waveform of a switching function to be described later) can be divided into two portions: one having a uniform pulse interval (equidistance pulse), that is, having a uniform pulse generation cycle in the vicinity of the zero crossing of the modulation wave; and the other portion having a single broad pulse width component with its center corresponding to a peak of the modulation wave. Therefore, a switch-over from the overmodulation mode to the one-pulse mode may occur in any portion having an equidistance pulse in the overmodulation mode. In this instance, load current in the inverter fluctuates, which may result in a breakdown of switching elements due to overcurrents or a substantial fluctuation in motor torque.

In order to solve these problems, it is necessary to specify a particular shift voltage (hereinafter referred to as a shift voltage) at which a shift between the overmodulation and the one-pulse mode is allowed, and a particular phase (hereinafter referred to as a shift phase) in the output voltage fundamental wave at which the shift should be executed.

First, determination of the shift voltage will be described below.

It is desirable for the shift voltage to be set at a value in close proximity to the output voltage in the one-pulse mode operation (that is, close to 100%), because fluctuation in motor torque during mode shifting is minimized when the difference between its output voltage and its maximum value available is minimum.

In the asynchronous PWM, however, individual voltage pulses contained in one cycle of the output voltage fundamental wave have a different pulse width according to each cycle, and when the number of pulses decreases in the vicinity of the zero crossing point of the output voltage fundamental wave as the output voltage in the overmodulation mode approaches 100%, its adverse effect becomes significant, causing an unbalance between the positive and negative polarities of output voltages, and eventually causing a beat phenomenon in the load current of the inverter. An example of such is shown in FIG. 9.

FIG. 10 shows an example of the relationship between the average pulse number in the vicinity of the zero crossing point of the output voltage fundamental wave and current pulsation due to the beat phenomenon. A portion of the modulation wave where its absolute value is under 1.0 corresponds to an equidistance pulse region as shown in FIG. 7, thereby, an average pulse number can be obtained by equation 3. Further, the current pulsation ratio is defined by equation 4. As can been seen from FIG. 10 unless at least one pulse is secured in the vicinity of zero cross point of the output voltage fundamental wave, the low frequency pulsation in the inverter load current due to the beat phenomenon becomes extremely great. ##EQU2##

Thereby, an appropriate shift voltage is preferably set at a value which ensures at least one voltage pulse in the vicinity of the zero cross point of the output voltage fundamental wave. Since this value depends on the output fundamental frequency Fi* and the carrier frequency Fc in the multi-pulse mode, it may be derived from calculation of these values. Alternatively, this value may be calculated and preset beforehand from an upper limit of the output voltage fundamental frequency Fi*.

Next, the manner in which the shift phase is controlled will be described.

Dependent on the phase of the output voltage fundamental wave at the time when a change between the overmodulation mode and the one-pulse mode occurs, the manner in which transient fluctuations appear in a load current in the inverter and in a torque generated immediately after the mode change differs. An example of such current fluctuations is shown in FIG. 11. FIG. 11(a) shows a case where all the three phases were changed in batch at the 0.degree. point of the U-phase of the output voltage fundamental wave (see FIG. 12), and a transient current fluctuation was observed immediately after the mode change. In contrast, FIG. 11(b) shows another case where all the three phases were changed in batch at 90.degree. of the U-phase of the output voltage fundamental wave (FIG. 13), and there was observed almost no transient current fluctuation.

FIG. 14, illustrates the relationships between the shift phase (U-phase taken as a reference) of the output voltage fundamental wave when shifting all the three phases in batch from the overmodulation mode to the one-pulse mode and transient current fluctuation, where the current fluctuation rate is defined by equation 5.

In FIG. 14, at every 60.degree. of phase of the output voltage fundamental wave a large current fluctuation rate appears. This occurs where a mode change between the overmodulation and the one-pulse modes takes place while one of the three phases is in the equidistance pulse region in the overmodulation mode, thereby allowing mixed presence of both the modes, which in turn causes an increased transient unbalancing of the output voltages among three phases, and a resultant large transient current fluctuation. Therefore, according to the invention, such transient fluctuations in the current and torque can be successfully suppressed by assuring that the shift phase occurs in a portion where all the phases are in a wide pulse region and in the overmodulation mode as illustrated in FIG. 15.

In order to provide for batch mode changes of all three phases from the overmodulation mode to the one-pulse mode, it is necessary to secure a region where the output voltages in the overmodulation mode for all the three phases coincide with a wide pulse region. For this purpose, it must be arranged that at an intersection of modulation waves of any two of the three phases (that is, at the 30.degree., 90.degree., 150.degree., 210.degree., 270.degree., or 330.degree. points of the reference U-phase modulation wave), the absolute value of the modulation wave is greater than 1.0. For example, in respect of 30.degree., where au=Asin30.degree.>1 (and thus A>2), and where a correlation between the modulation ratio A and the output voltage E* in the overmodulation mode can be obtained by equation 2, it must hold that E*>95.6%. Therefore, in order for a batch mode change for the three phases to be executed between the overmodulation and the one-pulse mode, the shift voltage must be set at a value which is greater than 95.6%, and has at least one voltage pulse in the vicinity of the zero dross point of the output voltage fundamental wave in the overmodulation mode.

An example of a PWM mode select means 4 for implementing the above-mentioned shift voltage and shift phase controls of the invention is shown in FIG. 16. A mode select command generation means 42 compares a shift voltage Ec (set in a shift voltage setting unit 41) and a voltage command E*, and outputs a mode select command Mc indicating of which of the multipulse mode and the one-pulse mode is to be selected.

The mode select command Mc has been described hereinabove as being obtained on the basis of the output voltage command E*. However, since the output voltage command E* corresponds uniquely to the modulation ratio A, a particular modulation ratio Ac corresponding to the shift voltage may be preset, which is then compared with the modulation ratio A to generate the mode select command Mc.

Further, also in the variable voltage variable frequency region, since the output voltage command and the output voltage fundamental frequency correspond uniquely with each other, it may be arranged such that a particular output voltage fundamental frequency Fic which corresponds to the shift voltage may be preset beforehand. This output voltage fundamental frequency is then compared with the frequency command Fi* to generate the mode select command Mc.

A shift phase control means 44 reads Mc, and in case a mode change is required, compares phase .theta.x of the output voltage fundamental wave and a shift phase .theta.c set in a shift phase set means 43; then if .theta.x=.theta.c, it changes its mode select signal M. Dependent on the mode select signal M, mode select switches 45, 46, 47 select either one of output S1.sub.x of the multipulse generation means and output S2.sub.x of the one-pulse generation means, then determine its switching function S.sub.x.

The shift phase may also be controlled by taking absolute values of the modulation wave at respective phases. If all exceed 1.0 (which indicates that all the three phases lie in a wide pulse region in the overmodulation mode), a change of output from between the multipulse generation means and the one-pulse generation means is enabled at this instance.

Thereby, a two-level inverter according to the invention has been provided which features the advantage that a gap in the output voltages at the time of mode change between the multipulse mode and the one-pulse mode can be reduced as small as 1-2% in comparison with approximately 10% in the conventional GTO inverter, thereby allowing continuous control of the output voltage from zero to its maximum value. In addition, a smooth change can be ensured between the multipulse mode and the one-pulse mode operation without causing any fluctuations in the current and the generated torque.

FIG. 17 illustrates the relationship between the output voltage fundamental frequency and the switching frequency, where there exist no such large discontinuities such as observed in FIG. 3 in respect of the conventional inverter modulation method. Thus, discontinuous tone change occurring due to the magnetic noise is eliminated.

In the inverter equipment which is capable of controlling its output voltage from zero to a maximum voltage by use of the multi-pulse mode and the one-pulse mode in conjunction according to the invention, discontinuous changes in the magnetically-induced noise can be eliminated, and a substantially continuous control over the full range of its output voltage can be accomplished as well.

Although the invention has been described and illustrated in detail, it is to be clearly understood that the same is by way of illustration and example, and is not to be taken by way of limitation. The spirit and scope of the present invention are to be limited only by the terms of the appended claims.