阅读说明：本技术 一种逆变器死区电压补偿模型的构建方法及应用 (Construction method and application of inverter dead zone voltage compensation model ) 是由 童乔凌 刘涛 闵闰 于 2021-06-09 设计创作，主要内容包括：本发明公开了一种逆变器死区电压补偿模型的构建方法及其应用,包括：S1、通过e指数函数拟合的方式,分别拟合各相补偿电压与其相电流、逆变器功率回路中其相所对应的硬件参数以及逆变器的最大补偿电压之间的非线性关系,得到各相补偿电压方程；S2、基于死区补偿电压与各相补偿电压之间的线性变换关系,结合各相补偿电压方程,得到死区补偿电压与各相电流、逆变器功率回路中各相所对应的硬件参数以及逆变器的最大补偿电压之间的非线性关系,记为逆变器死区电压补偿模型。本发明不需要用到逆变器的电气参数,也避免了传统补偿算法的补偿误差和繁琐的理论模型计算,可以获得更好的补偿效果,能够实时的、自适应的对逆变器的非线性效应进行精确补偿。(The invention discloses a construction method and application of an inverter dead zone voltage compensation model, wherein the construction method comprises the following steps: s1, fitting nonlinear relations between each phase of compensation voltage and phase current thereof, hardware parameters corresponding to each phase of the inverter power loop and the maximum compensation voltage of the inverter respectively in an e exponential function fitting mode to obtain a compensation voltage equation of each phase; and S2, obtaining a nonlinear relation between the dead zone compensation voltage and each phase current, hardware parameters corresponding to each phase in the inverter power loop and the maximum compensation voltage of the inverter by combining a compensation voltage equation of each phase based on the linear transformation relation between the dead zone compensation voltage and each phase compensation voltage, and recording the nonlinear relation as an inverter dead zone voltage compensation model. The method does not need to use the electrical parameters of the inverter, avoids the compensation error of the traditional compensation algorithm and the complex theoretical model calculation, can obtain better compensation effect, and can carry out accurate compensation on the nonlinear effect of the inverter in real time and in a self-adaptive manner.)

1. A method for constructing an inverter dead zone voltage compensation model is characterized by comprising the following steps:

s1, fitting nonlinear relations between each phase of compensation voltage and phase current thereof, hardware parameters corresponding to each phase of the inverter power loop and the maximum compensation voltage of the inverter respectively in an e exponential function fitting mode to obtain a compensation voltage equation of each phase;

and S2, obtaining a nonlinear relation between the dead zone compensation voltage and each phase current, hardware parameters corresponding to each phase in an inverter power loop and the maximum compensation voltage of the inverter by combining the compensation voltage equation of each phase based on the linear transformation relation between the dead zone compensation voltage and the compensation voltage of each phase, and recording the nonlinear relation as an inverter dead zone voltage compensation model.

2. The method for constructing the inverter dead zone voltage compensation model according to claim 1, wherein the m-th phase compensation voltage equation is as follows:

wherein m is a or b or c; Δ V_mCompensating voltage for the mth phase; v_dIs the maximum compensation voltage of the inverter; w is a_mThe hardware parameters are corresponding to the mth phase in the power loop of the inverter; i.e. i_mIs the m phase current.

3. The method for constructing the inverter dead zone voltage compensation model according to claim 1, wherein the inverter dead zone voltage compensation model comprises the inverter dead zone voltage compensation model in an α β coordinate system, a dq coordinate system or a stationary three-phase coordinate system;

the cost function of the inverter dead zone voltage compensation model is as follows:

wherein, V_refIs a reference voltage of the inverter; v_sThe compensated output voltage of the inverter.

4. The method for constructing the inverter dead zone voltage compensation model according to claim 3, wherein the inverter dead zone voltage compensation model in the α β coordinate system is:

wherein, is Δ V_αCompensating voltage for dead zone of the alpha shaft of the inverter; Δ V_βThe voltage is compensated for the dead band of the inverter beta shaft.

5. The method for constructing the inverter dead zone voltage compensation model according to claim 3, wherein for the inverter dead zone voltage compensation model in the α β coordinate system:

wherein, V_αIs the actual output voltage of the inverter alpha shaft; v_βIs the actual output voltage of the inverter beta shaft; t is_fIs the filter time constant; s is the laplace constant.

6. An inverter nonlinear effect compensation method is characterized by comprising the following steps:

calculating the maximum compensation voltage of the inverter based on the step voltage output by the inverter when the motor is static and the corresponding steady-state feedback current;

substituting the maximum compensation voltage of the inverter into the inverter dead zone voltage compensation model constructed by the inverter dead zone voltage compensation model construction method according to any one of claims 1 to 5 for identification to obtain hardware parameters corresponding to each phase in the inverter power loop, and further obtain dead zone compensation voltage to compensate the nonlinear effect of the inverter.

7. The inverter nonlinear effect compensation method of claim 6, wherein the inverter dead-time voltage compensation model is identified by minimizing a cost function of the inverter dead-time voltage compensation model.

8. The inverter nonlinear effect compensation method according to claim 7, wherein a neural network algorithm is adopted to reversely update the hardware parameters corresponding to each phase in the inverter power loop to be identified so as to minimize a cost function of the inverter dead-zone voltage compensation model.

9. The method for compensating nonlinear effect of inverter according to claim 8, wherein when minimizing the cost function of the dead-zone voltage compensation model of the inverter in the α β coordinate system, the identification recurrence formula of the hardware parameter corresponding to each phase in the power loop of the inverter is:

w_a(k)＝w_a(k-1)+Δw_a

w_b(k)＝w_b(k-1)+Δw_b

w_c(k)＝w_c(k-1)+Δw_c

wherein, Δ w_m(k-1) is the variation of the hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; eta is the learning rate; v_error(k-1) is the generation obtained in the k-1 st iterationA value of a valence function; v_dIs the maximum compensation voltage of the inverter; w is a_m(k-1) is a hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; i.e. i_mIs the m phase current; v_αrCompensating the voltage of the alpha shaft of the inverter; v_βrCompensating the voltage for the beta axis of the inverter; m is a or b or c.

10. The inverter nonlinear effect compensation method according to any one of claims 6 to 9, wherein the maximum compensation voltage of the inverter is:

wherein, V₁And V₂The step voltage is output by the inverter when the motor is static; i is₁And I₂Is the steady state feedback current of the inverter when the motor is static.

Technical Field

The invention belongs to the field of motor control, and particularly relates to a construction method and application of a dead-zone voltage compensation model of an inverter.

Background

With the development of science and technology, more and more power electronic products are applied to daily production and life of people, and an inverter is an important application in the power electronic industry and mainly used for converting direct current into alternating current. However, in order to avoid the insertion of dead time into the upper and lower tubes of the same bridge arm of the three-phase inverter, and the nonlinear effects of turn-on/off delay, tube voltage drop and the like of the power device, the distortion of the actual output voltage and the reference voltage is easily caused, the distortion of three-phase current is caused, and the control performance of the motor is affected, so how to reduce the distortion of the current and the voltage becomes a technical problem to be solved urgently.

In the prior art, the nonlinear effect of the inverter is usually compensated to reduce the distortion of current and voltage, and most of the common inverter dead-zone compensation methods at present are based on a simple sign function model, a trapezoidal compensation model and a lookup table compensation model based on offline measurement to compensate; however, hardware parameters of different inverters are different, the above model needs to perform complicated offline calibration for different hardware parameters, is complex to operate, has limitations, and cannot compensate the nonlinear effect of the inverter in real time and in a self-adaptive manner for different inverters. In addition, the model is simple, is obtained by approximately constructing the relation between the dead zone compensation voltage and the phase current, has low accuracy, and cannot realize the accurate compensation of the nonlinear effect of the inverter.

Disclosure of Invention

Aiming at the defects or the improvement requirements of the prior art, the invention provides a construction method of an inverter dead zone voltage compensation model and application thereof, which are used for solving the technical problem that the prior art can not adaptively and accurately compensate the nonlinear effect of the inverter.

In order to achieve the above object, in a first aspect, the present invention provides a method for constructing an inverter dead zone voltage compensation model, including the following steps:

s1, fitting nonlinear relations between each phase of compensation voltage and phase current thereof, hardware parameters corresponding to each phase of the inverter power loop and the maximum compensation voltage of the inverter respectively in an e exponential function fitting mode to obtain a compensation voltage equation of each phase;

and S2, obtaining a nonlinear relation between the dead zone compensation voltage and each phase current, hardware parameters corresponding to each phase in the inverter power loop and the maximum compensation voltage of the inverter by combining a compensation voltage equation of each phase based on the linear transformation relation between the dead zone compensation voltage and each phase compensation voltage, and recording the nonlinear relation as an inverter dead zone voltage compensation model.

Further preferably, the m-th phase compensation voltage equation is:

wherein m is a or b or c; Δ V_mCompensating voltage for the mth phase; v_dIs the maximum compensation voltage of the inverter; w is a_mThe hardware parameters are corresponding to the mth phase in the power loop of the inverter; i.e. i_mIs the m phase current.

Further preferably, the cost function of the inverter dead-zone voltage compensation model is:

wherein, V_refIs a reference voltage of the inverter; v_sThe compensated output voltage of the inverter.

The inverter dead zone voltage compensation model comprises an inverter dead zone voltage compensation model under an alpha beta coordinate system, a dq coordinate system or a static three-phase coordinate system.

Further preferably, the inverter dead-zone voltage compensation model in the α β coordinate system is:

wherein, is Δ V_αCompensating voltage for dead zone of the alpha shaft of the inverter; Δ V_βFor the dead of inverter beta-shaftThe zone compensates for the voltage.

Further preferably, for the inverter dead-zone voltage compensation model in the α β coordinate system:

V_s ²＝(V_α-ΔV_α)²+(V_β-ΔV_β)²

wherein, V_αIs the actual output voltage of the inverter alpha shaft; v_βIs the actual output voltage of the inverter beta shaft; t is_fIs the filter time constant; s is the laplace constant.

In a second aspect, the present invention provides a method for compensating nonlinear effect of an inverter, comprising the following steps:

calculating the maximum compensation voltage of the inverter based on the step voltage output by the inverter when the motor is static and the corresponding steady-state feedback current;

the maximum compensation voltage of the inverter is substituted into the inverter dead zone voltage compensation model provided by the first aspect of the invention for identification, so as to obtain hardware parameters corresponding to each phase in the inverter power loop, and further obtain dead zone compensation voltage, so as to compensate the nonlinear effect of the inverter.

Further preferably, the inverter dead-time voltage compensation model is identified by minimizing a cost function of the inverter dead-time voltage compensation model.

Further preferably, the hardware parameters corresponding to each phase in the inverter power loop to be identified are reversely updated by adopting a neural network algorithm so as to minimize a cost function of the inverter dead-zone voltage compensation model.

Further preferably, when the cost function of the inverter dead-zone voltage compensation model in the α β coordinate system is minimized, the identification recurrence formula of the hardware parameter corresponding to each phase in the inverter power loop is as follows:

w_a(k)＝w_a(k-1)+Δw_a

w_b(k)＝w_b(k-1)+Δw_b

w_c(k)＝w_c(k-1)+Δw_c

wherein, Δ w_m(k-1) is the variation of the hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; eta is the learning rate; v_error(k-1) is the cost function value obtained under the k-1 iteration; v_dIs the maximum compensation voltage of the inverter; w is a_m(k-1) is a hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; i.e. i_mIs the m phase current; v_αrCompensating the voltage of the alpha shaft of the inverter; v_βrCompensating the voltage for the beta axis of the inverter; m is a or b or c.

Further preferably, the maximum compensation voltage of the inverter is:

wherein, V₁And V₂The step voltage is output by the inverter when the motor is static; i is₁And I₂Is the steady state feedback current of the inverter when the motor is static.

Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:

1. the invention provides a method for constructing an inverter dead zone voltage compensation model, which considers that corresponding hardware parameters in an inverter power loop are different, and respectively fits nonlinear relations among the compensation voltage of each phase and the phase current thereof, the corresponding hardware parameters in the inverter power loop and the maximum compensation voltage of the inverter in an e-exponential function fitting mode, so as to construct the inverter dead zone voltage compensation model, wherein the e-exponential can obtain a smoother transition process, better meet the physical process of capacitor charging and discharging, and can adaptively and accurately compensate the nonlinear effect of the inverter.

2. The invention provides a nonlinear effect compensation method of an inverter, which substitutes the maximum compensation voltage of the inverter into the inverter dead zone voltage compensation model for identification to respectively obtain the corresponding hardware parameters of each phase in the power loop of the inverter, does not need to use the electrical parameters of the inverter, avoids the compensation error of the traditional compensation algorithm and the complex theoretical model calculation, and can obtain better compensation effect; and for different inverter parameters, the invention can carry out self-adaptive compensation, avoids repeated calibration work and can carry out real-time and self-adaptive accurate compensation on the nonlinear effect of the inverter.

3. The inverter nonlinear effect compensation method provided by the invention is characterized in that a cost function of an inverter dead-zone voltage compensation model is constructed on the basis of the principle that a compensated voltage vector is a constant amplitude vector, and the inverter dead-zone voltage compensation model is dynamically identified by minimizing the cost function; because the amplitude of the compensated voltage vector changes with the rotating speed and the current, the method can track the amplitude change in a self-reference mode, and has real-time performance.

4. The nonlinear effect compensation method for the inverter provided by the invention adopts a neural network algorithm to reversely update hardware parameters corresponding to each phase in the power loop of the inverter to be identified so as to minimize the cost function of the dead-zone voltage compensation model of the inverter. The calculation amount of the neural network algorithm is small, and the e exponential function used for constructing the inverter dead zone voltage compensation model is an activation function commonly used in the neural network, and is more suitable for identifying hardware parameters corresponding to each phase.

5. Because the difference of parasitic capacitance resistance parameters of the inverter power loop causes the hardware parameters of each phase to be inconsistent, the inverter nonlinear effect compensation method provided by the invention considers the difference of the corresponding hardware parameters in the inverter power loop, and respectively identifies the hardware parameters of each phase, so that the method is more accurate.

6. The inverter nonlinear effect compensation method provided by the invention is automatically completed in a digital control chip, does not need manual adjustment, is simple and convenient, has small calculation amount, does not occupy more calculation resources, and is convenient for transplantation and application among different systems.

Drawings

Fig. 1 is a flowchart of a method for constructing an inverter dead-zone voltage compensation model according to embodiment 1 of the present invention;

fig. 2 is a flowchart of a method for compensating for nonlinear effects of an inverter according to embodiment 2 of the present invention;

fig. 3 is a control flow chart of a method for compensating for nonlinear effects of an inverter according to embodiment 2 of the present invention;

fig. 4 is a graph of an identification result of the inverter nonlinear effect compensation method provided in embodiment 2 of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

Examples 1,

A method for constructing an inverter dead zone voltage compensation model, as shown in FIG. 1, includes the following steps:

s1, fitting nonlinear relations between each phase of compensation voltage and phase current thereof, hardware parameters corresponding to each phase of the inverter power loop and the maximum compensation voltage of the inverter respectively in an e exponential function fitting mode to obtain a compensation voltage equation of each phase;

specifically, considering that different corresponding hardware parameters in the inverter power loop are different, the compensation voltage equations of all phases are obtained through fitting in an e exponential function fitting mode;

wherein, the m-th phase compensation voltage equation is as follows:

wherein m is a or b or c; Δ V_mCompensating voltage for the mth phase; v_dIs the maximum compensation voltage of the inverter; w is a_mThe hardware parameters are corresponding to the mth phase in the power loop of the inverter; i.e. i_mIs the m phase current.

S2, obtaining a nonlinear relation between the dead zone compensation voltage and each phase current, hardware parameters corresponding to each phase in an inverter power loop and the maximum compensation voltage of the inverter based on the linear transformation relation between the dead zone compensation voltage and each phase compensation voltage and by combining a compensation voltage equation of each phase, and recording the nonlinear relation as an inverter dead zone voltage compensation model; the cost function of the inverter dead zone voltage compensation model is as follows:

V_refis a reference voltage of the inverter; v_sThe compensated output voltage of the inverter.

Specifically, an inverter dead-zone voltage compensation model can be constructed based on a linear transformation relation between dead-zone compensation voltage and each phase compensation voltage in an alpha beta coordinate system, a dq coordinate system or a static three-phase coordinate system; namely, the inverter dead zone voltage compensation model comprises an inverter dead zone voltage compensation model under an alpha beta coordinate system, a dq coordinate system or a static three-phase coordinate system; the three can be mutually converted, and the identification effect is equivalent.

In the present embodiment, details are described by taking an inverter dead-zone voltage compensation model in an α β coordinate system as an example:

the inverter dead zone voltage compensation model under the alpha beta coordinate system is as follows:

wherein, is Δ V_αCompensating voltage for dead zone of the alpha shaft of the inverter; Δ V_βThe voltage is compensated for the dead band of the inverter beta shaft.

The cost function of the inverter dead zone voltage compensation model in the alpha beta coordinate system is as follows:

V_s ²＝(V_α-ΔV_α)²+(V_β-ΔV_β)²

wherein, V_αIs the actual output voltage of the inverter alpha shaft; v_βIs the actual output voltage of the inverter beta shaft; t is_fIs the filter time constant; s is the laplace constant.

Examples 2,

A compensation method for nonlinear effect of an inverter comprises a maximum compensation voltage identification unit and an identification unit. The maximum compensation voltage identification unit is used for identifying the maximum compensation voltage according to two different step voltages output when the motor is static. The identification unit is used for identifying the inverter dead zone voltage compensation model obtained in the embodiment 1 when the motor runs, so that the self-adaptive compensation of the nonlinear effect of the inverter is realized. Specifically, when the motor runs, the identification unit adopts equivalent dead zone compensation voltage to describe voltage distortion caused by the nonlinear effect of the inverter, carries out online identification on an inverter dead zone voltage compensation model, and then compensates the model according to the identified model, thereby eliminating the distortion of output voltage and current caused by the nonlinear effect of the inverter.

As shown in fig. 2, the specific steps are as follows:

calculating the maximum compensation voltage of the inverter based on the step voltage output by the inverter when the motor is static and the corresponding steady-state feedback current, so as to provide basic parameters for subsequent model identification; wherein, the maximum compensation voltage of the inverter is:

V₁and V₂The step voltage is output by the inverter when the motor is static; i is₁And I₂Is the steady state feedback current of the inverter when the motor is static.

And substituting the maximum compensation voltage of the inverter into the inverter dead zone voltage compensation model obtained in the embodiment 1 for identification to obtain hardware parameters corresponding to each phase in the power loop of the inverter, and further obtaining the dead zone compensation voltage to compensate the nonlinear effect of the inverter. Specifically, the inverter dead-time voltage compensation model is identified by minimizing a cost function of the inverter dead-time voltage compensation model. Preferably, the hardware parameters corresponding to each phase in the inverter power loop to be identified are reversely updated by adopting a neural network algorithm so as to minimize the cost function of the inverter dead-zone voltage compensation model. The calculation amount of the neural network algorithm is small, and the e exponential function used for constructing the inverter dead zone voltage compensation model is an activation function commonly used in the neural network, and is more suitable for identifying hardware parameters corresponding to each phase.

The following details are given by taking an inverter dead-zone voltage compensation model in an α β coordinate system as an example:

as shown in fig. 3, which is a control flow chart of a nonlinear effect compensation method for an inverter, the dead-zone voltage compensation model of the inverter constructed in embodiment 1 is corresponding to a neural network structure;

dead zone compensation voltage is obtained based on the inverter dead zone voltage compensation model, and the dead zone compensation voltage of the alpha axis of the inverter is recorded as delta V in the embodiment_αThe dead-zone compensation voltage of the inverter beta shaft is recorded as delta V_β。

Adding a dead-zone compensation voltage to a conventional alpha-beta shaft output reference voltage to obtain a new reference voltage to control the output of the inverter, wherein specifically, the actual output voltage of the inverter is as follows:

wherein, V_αIs the actual output voltage of the inverter alpha shaft; v_βIs the actual output voltage of the inverter beta shaft; v_αrCompensating the voltage of the alpha shaft of the inverter; v_βrAnd compensating the voltage of the beta shaft of the inverter.

After compensation, the vector amplitude of the output voltage is calculated by utilizing the characteristics of the original compensated output reference voltage. Through a designed neural network structure, hardware parameters (corresponding to the weight w of the neural network) corresponding to each phase in an inverter power loop in the inverter dead-zone voltage compensation model are corrected step by step on line so as to minimize a cost function of the inverter dead-zone voltage compensation model, and therefore dynamic identification of the inverter dead-zone voltage compensation model is realized: specifically, in the training process, the identification recurrence formula of the hardware parameter corresponding to each phase in the inverter power loop is as follows:

w_a(k)＝w_a(k-1)+Δw_a

w_b(k)＝w_b(k-1)+Δw_b

w_c(k)＝w_c(k-1)+Δw_c

wherein, Δ w_m(k-1) is the variation of the hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; eta is the learning rate, and the value in the embodiment is 0.02; v_error(k-1) is the cost function value obtained under the k-1 iteration; v_dIs the maximum compensation voltage of the inverter; w is a_m(k-1) is a hardware parameter corresponding to the mth phase in the inverter power loop under the k-1 th iteration; i.e. i_mIs the m phase current; v_αrCompensating the voltage of the alpha shaft of the inverter; v_βrCompensating the voltage for the beta axis of the inverter; m is a or b or c.

Wherein the maximum compensation voltage V of the inverter_dThe calculation method of (2) is as follows: when the motor is static, the inverter outputs two voltage vectors V₁、V₂And respectively obtaining a steady-state feedback current I₁、I₂And calculating the maximum compensation voltage (also called maximum dead-time compensation voltage) of the inverter according to the obtained voltage vector and the steady-state feedback current. Wherein, in order to realize and simplify the steady state, the voltage vector V is output under an alpha beta coordinate system₁(V_α＝0，V_β＝V₁) Obtaining a feedback current I₁＝I_β1；V_αIs the actual output voltage of the inverter alpha shaft; v_βIs the actual output voltage of the inverter beta shaft.

The above process is based on the principle that the compensated voltage vector is a constant amplitude vector to identify the dynamic model, because the amplitude of the compensated voltage vector changes with the rotation speed and the currentTherefore, the neural network algorithm adopts a self-reference mode to realize the tracking of the amplitude change. The identification result diagram of the inverter nonlinear effect compensation method obtained in this embodiment is shown in fig. 4, and it can be seen from the diagram that the hardware parameters corresponding to each phase in the inverter power loop are continuously updated in the identification process and finally tend to be stable, and at this time, the compensated voltage vector approaches the reference voltage V of the inverter_ref. It should be noted that the nonlinear effect compensation method in the present invention is not limited to the α β coordinate system, and can be implemented in other coordinate systems, such as dq coordinate system and stationary three-phase coordinate system.

Based on the method, when the motor runs, the dead zone voltage compensation model is gradually identified according to the constructed neural network structure, and dead zone voltage compensation is simultaneously carried out, so that voltage and current distortion caused by the nonlinear effect of the inverter is eliminated. After the identification is completed, hardware parameters corresponding to each phase in the power loop of the inverter are stable, a self-adaptive inverter dead zone voltage compensation model can be obtained, and online compensation is performed according to the identified inverter dead zone voltage compensation model, so that voltage and current distortion is eliminated. The invention can identify the nonlinear effects of different inverters by adjusting the weight, and can adaptively realize the compensation of the nonlinear effects of different inverters.

The related technical scheme is the same as embodiment 1, and is not described herein.

It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.