TWI431300B - Method for parameter identification of induction machine - Google Patents

Description

Induction machine parameter estimation method

The invention relates to a method for estimating the parameters of an induction machine, in particular to a method for estimating the parameters of an induction machine with a low-voltage unloaded starting state.

Induction motors are widely used in the industrial field. Their application level has been provided by pure power, and gradually extended to precise control (rotation speed, torque and positioning). In order to improve the control accuracy of the induction motor, one of the relevant parameters of the induction motor is Indispensable information. The method for obtaining the relevant parameters can be estimated by using the equivalent model of the induction motor, wherein the commonly used equivalent model can be divided into two types: a steady state equivalent model and a dynamic equivalent model.

In the steady-state equivalent model, the impedance parameter of the relevant parameter can be obtained by standard tests, for example, through the stall test, the no-load start test, and the DC test on the stator side, the impedance parameter of the steady-state equivalent model can be obtained. Under the steady-state equivalent model architecture, the user can analyze the operating conditions at steady state, but the transient characteristics of the steady-state equivalent model are caused by the change of the unloaded load and the transient consideration of the signal change at the power supply end. The description above is a bit lacking.

In the dynamic equivalent model, transient effects such as voltage, current, stator frequency, and load variation are considered. Common estimation methods include spectrum analysis technology, reference model adaptation technology and optimization technology. Although the dynamic equivalent model obtained by the related art can describe the transient characteristic more completely, in the estimation process of the relevant parameter, an additional control circuit such as a converter must be used, thereby exciting the response and causing the estimation. The required implementation hardware is increased.

In addition, whether it is a steady-state equivalent model or a dynamic equivalent model, in the analysis In the estimation process, only the impedance parameters in the relevant parameters can be obtained, and one of the related parameters (moment of inertia and friction coefficient) cannot be obtained. In view of this, it is necessary to have a set of induction machine parameter estimation methods that take into account low hardware cost, overestimation of the accuracy and complete parameters, in order to meet different needs.

The main object of the present invention is to provide an induction machine parameter estimation method, which has a low hardware cost burden and a simple architecture.

Another object of the present invention is to provide an induction machine parameter estimation method, which can estimate one of the relevant parameters of the induction machine.

Another object of the present invention is to provide an induction machine parameter estimation method, which can obtain relatively complete related parameters.

In order to achieve the foregoing object, the method for estimating the parameter of the induction machine of the present invention comprises: a step of lowering the rate of increase of the speed of rotation, which is to reduce the input voltage of one of the induction machines of the unloaded start, and to reduce the rate of increase of the speed of starting the induction machine; An impedance slip curve obtaining step is to measure and record a time varying voltage, a time varying current and a time varying speed of the induction machine from a start to a steady state speed, and obtain a time domain analysis Obtaining an impedance slip curve; an initial value obtaining step is to use the DC test of the stator side to match the impedance slip curve, and calculate the initial value of the impedance parameter by the least square method; an impedance parameter optimization step, Based on the initial value of the impedance parameter, a minimum root mean square method is combined with a particle swarm algorithm to optimize the impedance parameter; and a mechanical parameter calculation step is used to determine the impedance parameter. It is substituted into the steady-state equivalent circuit, and the mechanical parameters are obtained by using the time-varying voltage, the speed signal and the mechanical equation from the start-up to the steady-state speed.

The sensor parameter estimation method of the present invention, wherein the input voltage is reduced to a minimum starting voltage of the induction machine to be tested.

The sensor parameter estimation method of the present invention, wherein the minimum root mean square method is used to define an objective function, and the objective function has a number of gradient amounts.

The sensor parameter estimation method of the present invention, wherein the number gradient amount is used as a correction component of a correction amount of the particle group algorithm.

The above and other objects, features and advantages of the present invention will become more <RTIgt; The "gradient amount" in the present invention refers to a first derivative of the objective function.

The term "no load" as used in the present invention means that the power output end of the induction machine is not connected to any transmission member, and is operated in an unloaded state.

The "slip" as used in the present invention refers to a ratio of a slip difference to a synchronous speed of one of the rotating magnetic fields of the stator, wherein the slip value is the difference between the synchronous speed and the actual speed.

Referring to FIG. 1 , the sensor parameter estimation method of the present invention comprises: a rotation speed increase rate reduction step S1, an impedance slip rate curve determination step S2, an initial value determination step S3, and an impedance parameter. Optimizing step S4 and a mechanical parameter calculation step S5, and recording, by a measuring instrument, a time varying voltage during start-up to steady state speed when the induction machine is started at low voltage and no load, The time-varying current and the time-varying speed are further estimated by a computing processor such as a computer.

The rotation speed increase rate lowering step S1 is to reduce the input voltage of the induction machine that is started without load, and to reduce the rate of increase of the rotation speed of the induction machine. Under the normal starting state of the induction machine, the dynamic equivalent model of the induction machine has different characteristics from the steady-state equivalent model. More specifically, under the stator reference model, the dynamic equivalent model can be expressed as:

Where i _qs , i _ds is the stator current, i _qr , i _dr is the rotor current, v _qs , v _ds is the stator voltage, v _qr , v _dr is the rotor voltage, R _s is the stator resistance, R _r is the rotor resistance, L _m is the mutual inductance, L _s is the stator inductance, L _r is the rotor inductance, ω _r is the rotor speed, and p is the differential factor.

In equation (1), the partial terms of the impedance matrix are dependent on the rotational speed as a function of the rotational speed. In the case of rotational speed variations, the equations are nonlinear. If the speed is fixed or zero, the equation group is linear. When the input voltage of the induction machine is lowered, and the induction machine is in a low-voltage no-load starting state, the speed increase rate of the induction machine can be changed within a unit time, that is, dω _r /dt 0 , the transient effect caused by the change of the rotational speed is minimized, and the transient effect can be neglected, and the steady-state equivalent model can be used to estimate the impedance parameter of the induction machine without additional control circuit. The steady-state equivalent model approximated by the dynamic equivalent model can be as shown in FIG. 2, and the steady-state equivalent circuit can be expressed as a function of the rotational speed and has an equivalent resistance and an equivalent reactance. The effective resistance and the equivalent reactance are:

Wherein, R _s is stator resistance, R _r is rotor equivalent resistance, X _m is magnetic excitation reactance, X _s is stator equivalent reactance, X _r is rotor equivalent reactance, s ( n ) is slip ratio , ω _s is the synchronous speed and ω _r ( n ) is the rotational speed, where n =0, 1, 2 Λ N -1, which is the number of samples of one of the positive integers.

In this embodiment, the bottom limit of the input voltage to be reduced is not limited herein, and may be adjusted according to different specifications of the induction machine, preferably the input voltage is lowered to the minimum start of the induction machine to be tested. The voltage allows the speed of the induction machine to rise slowly at startup.

The impedance slip curve is determined by step S2, and the measuring instrument is used to measure and record a time varying voltage, a time varying current and a time varying speed of the induction machine from the start to the steady state speed, and by a time domain analysis Obtain an impedance slip curve.

When the induction machine has lowered the starting voltage, so that the speed of the induction machine can be slowly increased at the time of starting, the impedance parameter of the induction machine can be described by a steady-state equivalent model, and has the equivalent resistance and the same Effective reactance. In order to accurately obtain the relationship between the equivalent resistance and the equivalent reactance and the slip ratio, the measuring instrument can be used to record the time-varying voltage and the time-varying current of the induction machine from the start to the steady-state speed, and the speed is changed in time. The time-varying voltage and the time-varying current have the same sampling scale in time. Where the time varying voltage and time varying electricity The stream can analyze the amplitude and phase of each sampling scale through time domain analysis, and thereby obtain the primary side equivalent resistance and the equivalent reactance corresponding to each sampling point; the slip rate can be obtained by each sampling point Obtained, and thereby obtain the corresponding slip rate of each sampling point. The impedance slip curve can be obtained under the condition of having the same sampling scale. In the present embodiment, the impedance slip curve is as shown in FIG.

The initial value obtaining step S3 is to calculate the initial value of the impedance parameter by the least square method by using the DC test on the stator side and the impedance slip ratio curve. The optimization of the impedance parameters must use a set of initial values as the starting point for the search, and a reasonable and reasonable initial value will help to improve the convergence efficiency. In this embodiment, when the induction machine is operated at a rated current, the measuring instrument is used to measure the operating voltage and the operating current of the induction machine, and the stator resistance R _{s is} obtained, and if the stator winding is Y connected Type, then the equation is:

Where V _{DC is} the operating voltage and I _{DC is} the operating current.

When the stator resistance When known, the operating voltage, operating current, and rotational speed can be used to calculate the s ( n ) and the R ( s ( n )) through the stator resistance. Substituting into equation (2) can be obtained:

among them

In the least squares method, the sum of squared errors is:

The necessary conditions for minimization are:

After finishing, you can get:

When the slip ratios s ( n ) and Q ( s ( n )) are known, substituting equation (8) can calculate K ₁ and K ₂ . In the impedance parameter, since the magneto-impact reactance X _m >> X _s and X _m >> X _r , X _r = X _s is assumed in equation (3) and substituted into equation (3):

among them,

Similarly, the sum of squared errors is:

Where K ( n )=(1+( s ( n )) ² K ₁ ), the same reason:

Here K ₅ = K ₄ / K ₃ = , the initial value of other impedance parameters is:

X _s = X _r = K ₃ - X _m (14)

The initial value of the impedance parameter can be determined through the above calculation process, and the initial value of the impedance parameter can be used as the basis for the subsequent optimization of the impedance parameter to obtain the preferred impedance parameter.

The impedance parameter optimization step S4 is performed by a minimum root mean square method (LMS) combined with a Particle Swarm Optimization (PSO) algorithm based on the initial value of the impedance parameter. The parameter is optimized to search for the wide-area optimal solution of the impedance parameter. First, the part of the minimum root mean square method is described in detail. The minimum root mean square method can handle the minimization search of multivariable nonlinear functions. The impedance slip curve obtained in step S2 is expressed as R ( s ( n )), X ( s ( n )), and s ( n ), n =0, 1, 2... N -1. The LMS can express its objective functions ER and EX as:

among them,

M ( n )=( R _r / s ( n )) ² +( X _m + X _r ) ² (20)

The objective functions ER and EX are both functions having a minimum value, and at the minimum value, the magnitude of the gradient of the two functions for each impedance parameter is zero. The present invention searches for the minimum value of the objective function ER with R _s , R _{r , and} searches for the minimum value of the objective function EX with X _s , X _r and X _m , and the error amount of the resistance can be expressed as:

Therefore, the number gradient of the objective function ER and the minimum value of the objective function EX can be expressed as:

In the present invention, the minimum root mean square method can enhance the local search ability, and then cooperate with the particle swarm algorithm to complete the optimization of the wide area search. Here, for the purpose of overall optimization, the adaptation function can be defined as:

FIT = ER + EX (28)

Since the present invention provides a reasonable set of initial values in step S3, in the concept of group in the particle swarm algorithm, the particles need only be scattered in a random number in the parameter space near the initial value, so as to be wide for the appropriate region. Domain search, each particle can be represented in the parameter space as:

The particle position update method of the particle swarm algorithm is as follows:

among them, Represents the updated particle position, Represents the particle position before the update, Represents one of the particle position update corrections. Wherein, the correction amount has a gradient amount, and the gradient amount is used as a correction component of the correction amount, and the correction amount is as follows:

Among them, C ₁ is the cognitive parameter of the particle's own experience, and C ₂ is the cognitive parameter of the particle group experience. Is the best position of the particle itself in the k- time iteration process, It is the optimal position in the k- time iteration process in the group, ▽ E _{i is} the gradient amount of the objective function ER and EX , which is the equations (23) to (27), α ^k is the acceleration factor, and iter _max is the setting The number of iterations, α _max is the set maximum acceleration factor, α _min is the set minimum acceleration factor, and rand is a random value between 0 and 1.

In this embodiment, the C ₁ = C ₂ =1 is set, the α _max = 0.5 is set, the α _min = 0.3 is set, and when the number of updates of the particle position reaches the number of iterations, the impedance is ended. Parameter optimization step S4. More specifically, in the process of impedance parameter search, due to the characteristics of the particle swarm algorithm, the process of parameter optimization avoids the problem of convergence. For the particle swarm algorithm, it has the characteristics of searching for the best solution in the wide area. However, when the particle is in the vicinity of the optimal solution, it may miss the best because of its poor proximity search ability. Solution, so the gradient amount is added to the particle swarm algorithm, and the gradient amount is used as the correction component of the correction amount of the particle swarm algorithm, so that the optimization algorithm can simultaneously have wide-area and adjacent search. Ability to find the best impedance parameter.

The mechanical parameter calculation step S5 is performed by substituting the impedance parameter into the steady-state equivalent circuit, and determining a mechanical parameter with a time-varying voltage, a rotational speed signal and a mechanical equation initiated during the steady-state rotational speed. Please refer to Figure 2, in more detail, the impedance parameter is substituted into the steady-state equivalent circuit, and the time-varying air gap power P _AG can be calculated with the time-varying voltage and speed from the start-up to the steady-state speed. The air gap power can be expressed as:

A mechanical power P _m can be expressed as:

Wherein the P _RCL is the equivalent copper loss of the rotor circuit.

An electromagnetic torque T _e can be expressed as:

In the dynamic equivalent model, the mechanical equation of the induction machine can be expressed as:

Where ω _m is the mechanical speed of the rotor, ie ω _m = 2 ω _r /P. P is the number of poles of the induction machine, J is the moment of inertia, and B is the coefficient of friction. Under the condition of no-load starting, let T _Load =0, and the above formula can be obtained:

Under the condition of no-load starting, when the rotor speed is gradually accelerated from the stationary state to the steady-state speed, it is considered that the steady-state condition is fixed, that is, dω _m / dt =0. Substituting this condition into equation (36) gives the coefficient of friction B as:

When the coefficient of friction is known, the moment of inertia J can be obtained by the following equation:

The average value of the moment of inertia J can be obtained from the range in which the rotational speed is accelerated from the stationary state to the steady-state rotational speed.

By operating the induction machine in a low-voltage, no-load start-up state, the dynamic characteristics of the induction machine can be approximated to a steady-state characteristic per unit time, so that the impedance parameter is obtained without an additional control circuit. By measuring the time-varying voltage and the time-varying current during the steady-state speed of the induction machine, the impedance slip curve of the induction machine can be obtained, and a reasonable set of calculations can be calculated by the least square method. The initial value of the impedance parameter. Then, based on the initial value, the particle swarm optimization algorithm is combined with the least root mean square method to search for the optimal solution, and the correction amount of the particle swarm algorithm includes the gradient component as the correction component of the correction amount. This impedance parameter is used to obtain the best solution. And in the low-voltage no-load starting state, with the optimized impedance parameter, the more accurate mechanical parameters can be obtained step by step.

The method for estimating the parameter of the induction machine of the invention reduces the starting voltage so that the induction machine is in a low-voltage unloaded starting state for subsequent correlation parameter estimation, which does not require complicated control circuit and saves extra hardware cost. The effect.

The sensor parameter estimation method of the invention uses a gradient amount combined with a particle swarm algorithm to perform an optimal search of the impedance parameters, and has the effect of optimizing the impedance parameters.

The method for estimating the parameter of the induction machine of the invention reduces the starting voltage so that the induction machine is in a low-voltage no-load starting state, which is helpful for obtaining a mechanical parameter and has the effect of obtaining relatively complete related parameters.

While the invention has been described in connection with the preferred embodiments described above, it is not intended to limit the scope of the invention. The technical scope of the invention is protected, and therefore the scope of the invention is defined by the scope of the appended claims.

[this invention]

S1. . . Speed increase rate reduction step

S2. . . Impedance slip curve determination step

S3. . . Initial value step

S4. . . Impedance parameter optimization step

S5. . . Mechanical parameter calculation step

Figure 1: Flow chart of the method for estimating the parameters of the induction machine of the present invention.

Figure 2: Steady-state equivalent circuit diagram of the sensor parameter estimation method of the present invention.

Fig. 3 is a graph showing the impedance slip rate of the sensor parameter estimation method of the present invention.

S1. . . Speed increase rate reduction step

S2. . . Impedance slip curve determination step

S3. . . Initial value step

S4. . . Impedance parameter optimization step

S5. . . Mechanical parameter calculation step

Claims (4)

An induction machine parameter estimation method comprises: a speed increase rate lowering step, which reduces an input voltage of one of the induction machines without load start, so that the speed increase rate of the induction machine is reduced; and an impedance slip rate curve is obtained. In the step, a measuring instrument is used to measure and record a time-varying voltage, a time-varying current and a time-varying speed of the induction machine from a start-up to a steady-state speed, and an impedance slip curve is obtained by a time domain analysis; An initial value obtaining step is to use the DC test of the stator side to match the impedance slip curve, and calculate the initial value of the impedance parameter by the least square method; an impedance parameter optimization step is performed by the initial value of the impedance parameter Based on a minimum rms method combined with a particle swarm algorithm to optimize the impedance parameter; and a mechanical parameter calculation step, the impedance parameter is substituted into the steady-state equivalent circuit, and the collocation is initiated The mechanical parameters are determined by time varying voltage, speed signal and mechanical equation during steady state speed. The method for estimating an inductive machine parameter according to claim 1, wherein the input voltage is reduced to a minimum starting voltage of the induction machine to be tested. The sensor parameter estimation method according to claim 1, wherein the minimum root mean square method is used to define an objective function, and the objective function has a number of gradients. The method for estimating an inductive machine parameter according to claim 3, wherein the number gradient is used as a correction amount of the particle swarm algorithm Positive component.