^{1}

^{2}

^{3}

^{1}

^{2}

^{3}

Speed sensorless control schemes have potential benefits for industrial applications because they contribute to reducing process cost and they avoid using fragile sensors as encoders or resolvers in hostile environment. In addition, simplicity, reliability, and fast response of control structures to signal commands are much-needed features. In this paper, a new Speed Sensorless Direct Control (SSDC) technique allowing the achievement of these objectives is proposed. This technique combines Field Oriented Control (FOC) and Direct Torque Control (DTC) properties in the same approach. The estimated speed is reached only according to the measured current and voltage of the stator. DTC is extended to speed sensorless direct control with any notable modification. The proposed scheme is implemented to the induction machine benchmark and evaluated in real time under various possible scenarios of use. Experimental results show that the proposed SSDC has interesting capabilities to conduct induction motor in real time operation with good accuracy.

The variable speed control of electric drives has benefited in recent years from significant methodological and technological advances. In fact, advances in digital signal processors such as DSP and dSPACE kits and the development of power components today make it possible to implement very complex algorithms with a short computation time. Several control algorithms are developed, tested, and industrialized. Field oriented control (FOC) [

As is known, the available FOC structures require a speed sensor and the supply voltage is generated by an inverter governed by the Space Vector Pulse Width Modulation (SVPWM) strategy which is switching period generally in the order of 100 to 200 microseconds. By modulation principle, the switching times of the inverter IGBTs are variable in this range and can therefore reach very low values. The waveform of the current is then sufficiently smooth. The torque and speed rising times in the closed loop operating points are about the rotor time constant [

Conventionally, DTC structures are speed sensorless controls. In contrast to the FOC method, the DTC technique eliminates PI controllers, transform matrices, current regulators, and PWM stage. Inverter IGBTs are in fact controlled with constant switching time that lies in general in the range of 25 to 50 microseconds. Torque and stator flux magnitude are regulated by adequately selecting a voltage vector among those available on the inverter [

FOC and DTC are well studied and compared in various publications [

The proposed algorithm for estimating and controlling the speed of the induction machine benchmark is experimentally verified in real time running on Dspace DS1104. Practical results prove that the DTC scheme can be extended to realize a reliable and efficient speed sensorless approach. This approach uses some expressions and properties of FOC. The developed method is here labelled Speed Sensorless Direct Control (SSDC).

The paper is organized as follows. Section

The dynamic model of induction machine can be formulated according to d-q-axis components in various reference frames. Let us first consider Concordia’s stationary reference frame. The usual following set of equations is used [

In normal steady state, stator and rotor electrical variables have sinusoidal wave forms with the same frequency and different magnitudes that depend on the considered operating point. These quantities vary in transient regime according to the adopted control. Various expressions can be used to calculate electromagnetic torque. The most used relation is the following where

In what follows, it will be very helpful to express stator and rotor flux vectors as functions of their magnitudes and angles:

Let us now consider a reference frame rotating with an angle

Transition from Concordia’s variables to Park’s ones is realized by the rotation operator

According to a specific topic, a suitable Park reference frame is selected. In practical point of view, there are two very interesting cases. The first one concerns a rotating reference frame hose d-axis is aligned to stator flux vector

In the case of a stator flux reference frame, angle

The most important conclusion that can be derived from this model is the fact that stator flux magnitude

For a rotor flux reference frame, angle

This model implies that rotor flux magnitude

In variable AC drives, a three-phase voltage source inverter feeds the induction machine. For a lossless-admitted inverter, the output voltage is strictly defined by the dc bus voltage

Figure

Inverter voltage vectors in Concordia’s reference frame.

Any proposed control scheme is developed to find out different solutions with two major objectives, accurate and quick control of the motor regime on one hand and reduction of the complexity and the cost of the algorithm on the other hand. Direct torque control (DTC) scheme developed and presented by I. Takahashi [

First, let us note that induction machine benchmark achieves the demanded load if adequate stator flux magnitude and frequency are realised. Stator flux vector is therefore a key variable in any emphasised control. It is observed by Takahashi that stator flux control can be achieved in a simple way if the motor is fed by a three phase inverter. According to (

Furthermore, if the inverter command is maintained during a time interval

According to this relation, selecting a null

Consequently, with appropriate sequence of the inverter command, the stator flux vector can be driven along any trajectory with a predefined average speed. The most appropriate flux trajectory for electrical machines control is naturally the circular path in the d-q plane. A very simple way to obtain this trajectory is to use a hysteresis comparator to control the flux vector. In order to maintain the flux within the hysteresis band, the motor should be currently fed by a suitable voltage vector with a constant switching period

On the other hand, accelerating or decelerating stator flux rotation is naturally linked to electromagnetic torque variation. Therefore, variation of angle

Stator and rotor machine circuits have different time responses. It is well known that stator variables vary more rapidly as compared to rotor ones. Stator quantities define a fast mode while rotor quantities correspond to a slow mode. Therefore, if flux magnitude is first controlled, a negative deviation of

Note finally that the effect of a particular voltage vector on torque and flux magnitude depends on the position

General structure of a DTC scheme.

The widely used DTC algorithm is built with six centered sectors for

Let also

Flux and torque errors codification.

Flux and torque errors | Hysteresis index |
---|---|

| |

| |

| |

| |

| |

In the sense of this codification, Table

A Takahashi switching table.

| 1 | 1 | 1 | 2 | 2 | 2 |

| ||||||

| 1 | 2 | 3 | 1 | 2 | 3 |

| ||||||

| | | | | | |

| ||||||

| | | | | | |

| ||||||

| | | | | | |

| ||||||

| | | | | | |

| ||||||

| | | | | | |

| ||||||

| | | | | | |

It has seen in the previous section that electromagnetic torque varies in the same sense as stator flux vector angle

Therefore, one can use the same DTC algorithm previously defined to control the motor speed with minor modification. This modification should not affect the switching table and should only affect its codification. First, it is necessary to change the input corresponding to electromagnetic torque

Flux and speed errors codification for two cases of loads.

Flux and speed errors | Hysteresis index for Load 1 | Hysteresis index for load 2 |
---|---|---|

| | |

| | |

| | |

| | |

| | |

Now, the problem is how to estimate motor electrical rotor speed

As outlined in Section

In the previous equation,

It is also important to note that error between actual rotor speed sensed for example by a tachometer or encoder and estimated speed defined by (

Schematic diagram of rotorique speed estimation.

As a result of this development, Figure

General Structure of the proposed SSDC control.

We note here that this structure can be used as direct voltage control structure by replacing the input electrical rotor speed

The previous developed algorithm is implemented and tested in an experimental environment. Photo of Figure

a dSpace DS1104 controller board with TMS320F240 slave processor and ADC interface board CP1104. The DS1104 board is installed in Intel(R) Pentium(R) D CPU 3.4 GHz PC for software development and results visualization,

a four poles induction motor loaded with a dc generator. Actual rotor speed is sensed from a tachometer coupled to machine shaft. Motor parameters and rated values are given below:

Rs = 7.5 Ω, Rr = 6.5 Ω, Ls = Lr = 0.354 H, M = 0.34 H, p = 2, F = 50 Hz, Wn = 293.22 rad/s, Pn = 1 KW,

a three-phase VSI whose dc bus voltage is generated by a rectifier connected to 400 V 50 Hz AC electrical sources,

a four channels 150 MHz digital oscilloscope and two channels 20 MHz analog oscilloscope support the experiment for online recording and visualization,

current and voltage sensors LEM-based and calibrated so that the analogue obtained signals should remain in the range 0 to ±10 V required by dSpace converters.

Photo of one side of the experimental setup.

The benchmark includes many other devices such as standard measurement equipment, analogue signal filters, analogues circuits defining d-q voltage, and currents components.

To evaluate the performance of the proposed SSDC algorithm, it was implemented in Simulink dSPACE DS1104 environment. The implemented structure is composed by four blocs: voltage and current acquisition, flux and speed estimation, switching table and command signals. A constant switching period

This case verifies the algorithm for a motor start-up scenario under no load condition. Stator flux and electrical rotor speed commands are set to

Figure

Predicted and actual speed (rad/s) versus time (s).

Instantaneous waveform of measured stator current.

Instantaneous waveform of stator flux.

From the no load steady state operating point previously described, three steps of motor load are executed by closing the dc generator armature circuit on successive three decreasing resistances. The reference values and hysteresis bands are those of Case

Figures

Estimated and actual speed during load changing.

Stator flux magnitude during load changing.

Stator power evolution during load changing.

Torque evolution during load changing.

Rotor slip pulsation evolution during load changing.

During simulation and practical tests, it was discovered that the speed hysteresis band could be reduced to a very high precision level. In fact, the proposed control scheme works very well even if one imposes a severe speed hysteresis band around 1 to 2%. This performance cannot be realized in the standard DTC algorithm with this tolerance for the torque. Ripple range of machine electromagnetic torque is very sensitive to the switching period and dc bus voltage of the VSI. A severe torque hysteresis band exhibits a dramatic full down of the DTC algorithm. The same phenomenon occurs if one proceeds to filter the torque signal. For the proposed algorithm, filtering speed estimate is possible.

To illustrate these observations, speed tolerance was reduced to 2% (5.86 rad/sec) while keeping other data of Case

To confirm the promising accuracy of developed approach, another program that controls stator flux magnitude and frequency by PI regulators under SPWM modulation technique was used also in real time. Reference values of

Estimated and actual motor speed with 2% hysteresis band.

Evolution of rotor slip pulsation with 2% hysteresis band.

Actual electrical machine speed in unregulated scenario.

It is obvious that the developed algorithm can be used for direct control of stator electrical frequency. The input

Voltage output in DVC case.

A direct speed estimation and control scheme has been developed theoretically and practically validated in this work. The proposed approach combines properties of FOC and DTC techniques. It extends DTC structure to speed sensorless direct control. The idea is focused on how to use standard Takahashi switching table to control induction machine speed. Software implementation needs no major modification of the standard DTC scheme. All necessary developments are presented and commented. Real time practical results obtained by the dSpace DS1104 board have good performances. In fact, the error between estimated and measured actual speed does not exceed 2.4%. The proposed technique can provide interesting benefit for industrial applications because it avoids using fragile speed sensors in hostile environment. Furthermore, the method can work as direct voltage control structure by controlling stator electrical frequency.

Admittedly, control without a speed sensor is nowadays a hot topic, especially in terms of the constraints of adaptation of the parameters; adaptation that does not reduce the dynamic stability of the algorithm. This is our perspective subject.

The data used to support the findings of this study are included within the article.

The authors declare that they have no conflicts of interest.