CN104009697B

CN104009697B - Substation inspection robot uses the method for mixing observation device detection positional information

Info

Publication number: CN104009697B
Application number: CN201410230665.1A
Authority: CN
Inventors: 余海涛; 孟高军; 胡敏强; 黄磊
Original assignee: Southeast University
Current assignee: Southeast University
Priority date: 2014-05-28
Filing date: 2014-05-28
Publication date: 2016-08-17
Anticipated expiration: 2034-05-28
Also published as: CN104009697A

Abstract

The invention discloses a kind of method that substation inspection robot uses mixing observation device detection positional information, initial position detection closed-loop control system, hybrid observer, model reference normalization algorithm, fuzzy controller, saliency compensating controller and hybrid observer are combined, with rotor-position and the velocity information of accurate and effective detection substation inspection robot.The inventive method can be accurate and effective the detection position of substation inspection robot and velocity information.

Description

A method for substation inspection robot to detect position information using hybrid observation device

技术领域technical field

本发明涉及一种变电站巡视机器人采用混合观测装置检测位置信息的方法，将初始位置检测闭环控制系统、模型参考归一化算法、模糊控制器、凸极效应补偿控制器和混合观测器结合在一起，以准确、有效的检测变电站巡视机器人的转子位置和速度信息。The invention relates to a method for a substation inspection robot to detect position information by using a hybrid observation device, which combines an initial position detection closed-loop control system, a model reference normalization algorithm, a fuzzy controller, a salient pole effect compensation controller and a hybrid observer , to accurately and effectively detect the rotor position and speed information of the substation inspection robot.

背景技术Background technique

目前，在各种结构的机器人系统中，由于采用永磁同步电机(PMSM)的方案效率较高，因此这种方案具有着重要的地位。特别是在电力机器人和小型机器人中，PMSM由于这些优点而得到了更多的应用。但是，一般情况下，机器人的驱动电机采用机械式位置传感器来检测电机的转速和动子位置，如光电编码器和旋转变压器。然而，机械式传感器的存在带来了很多弊端：1)电机与控制器之间的连接元件增多，坑干扰能力变差，降低了系统可靠性；2)加大了电机空间尺寸和体积，减少了功率密度，增加了系统的硬件成本和维护成本；3)在高温与强腐蚀环境中，将使传感器性能变差、甚至失效，导致电机驱动系统无法正常工作。At present, in robot systems of various structures, the scheme of using permanent magnet synchronous motor (PMSM) has a high efficiency, so this scheme plays an important role. Especially in electric robots and small robots, PMSMs have been more used due to these advantages. However, in general, the driving motor of the robot uses a mechanical position sensor to detect the speed of the motor and the position of the mover, such as a photoelectric encoder and a resolver. However, the existence of mechanical sensors has brought many disadvantages: 1) The number of connection elements between the motor and the controller increases, the pit interference ability becomes worse, and the system reliability is reduced; 2) The space size and volume of the motor are increased, reducing the The power density is increased, and the hardware cost and maintenance cost of the system are increased; 3) In a high temperature and strong corrosive environment, the performance of the sensor will deteriorate or even fail, resulting in the failure of the motor drive system to work normally.

而能否对转子初始位置进行准确估计是永磁同步发电机高性能控制策略(矢量控制或直接转矩)和无位置传感器运行实现的前提条件，也是关系到机器人是否顺利起动，以及能否实现最大转矩起动的关键问题；因此，转子初始位置检测一直是工程技术界研究的热点和难点问题之一，尤其在电力机器人中，因为机器人所进行的操作为基本为高压操作，检测的线路非常危险，如果转子初始位置检测不准确，会造成电力机器人的反响转动，结果有可能因为误操作破坏整个电力线路和机器人，严重的甚至会造成高压短路。Whether the initial position of the rotor can be accurately estimated is a prerequisite for the high-performance control strategy (vector control or direct torque) of the permanent magnet synchronous generator and the realization of position sensorless operation. It is also related to whether the robot can start smoothly and whether it can realize The key issue of maximum torque starting; therefore, the detection of the initial rotor position has always been one of the hot and difficult issues in the engineering and technology circles, especially in electric robots, because the operation of the robot is basically a high-voltage operation, and the detection line is very Dangerous, if the detection of the initial position of the rotor is inaccurate, it will cause the electric robot to reverberate and rotate. As a result, the entire power line and the robot may be damaged due to misoperation, and even cause a high-voltage short circuit in severe cases.

传感器的核心是控制系统能对转子的实时位置和速度进行准确的估算，常用无传感器的控制方法可分为2类：The core of the sensor is that the control system can accurately estimate the real-time position and speed of the rotor. The commonly used sensorless control methods can be divided into two categories:

1、采用电机理想模型的开环计算法，如直接计算法、反电动势积分法等，基于开环的计算方法简单直接，动态性能较好；但计算时依赖电机参数，而电机运行时参数总处于变化之中，这样势必会影响转子位置估计的准确性；并且在电机速度很低时，反电动势非常小，容易和各种干扰信号掺杂在一起，信噪比变低，使得反电势难于检测；所以这种方法并不适合用于电机静止或低速时无传感器位置估算；1. The open-loop calculation method of the ideal model of the motor is used, such as the direct calculation method, the back electromotive force integration method, etc. The calculation method based on the open loop is simple and direct, and the dynamic performance is better; but the calculation depends on the motor parameters, and the total parameters of the motor are running. is changing, which will inevitably affect the accuracy of rotor position estimation; and when the motor speed is very low, the back electromotive force is very small, it is easy to be mixed with various interference signals, and the signal-to-noise ratio becomes low, making the back electromotive force difficult to detection; so this method is not suitable for sensorless position estimation when the motor is stationary or at low speed;

2、基于外部高频信号注入的转子位置辨识方案，如旋转高频电压注入法、旋转高频电压注入法和旋转高频电流注入法；高频信号注入法是通过给电机三相绕组注入高频信号(电压或电流信号)，依靠电机转子自身的凸极性或由于饱和导致的凸极效应，使高频信号产生的磁场受到转子凸极的调制作用，因此高频信号中将带有转子位置信息，再将高频信号从定子电流或电压中解调出来就能提取出电机转子的位置信息；这种方法依靠外加激励信号，并不依赖于转速，但估算转子位置所需要的时间较长，位置量更新频率不高，所以高频信号注入法在电机静止和低速时有更好的估算效果。2. The rotor position identification scheme based on external high-frequency signal injection, such as rotating high-frequency voltage injection method, rotating high-frequency voltage injection method and rotating high-frequency current injection method; the high-frequency signal injection method is to inject high Frequency signal (voltage or current signal), depending on the saliency of the motor rotor itself or the saliency effect caused by saturation, the magnetic field generated by the high-frequency signal is modulated by the salient pole of the rotor, so the high-frequency signal will contain rotor position information, and then demodulate the high-frequency signal from the stator current or voltage to extract the position information of the motor rotor; this method relies on an external excitation signal and does not depend on the speed, but the time required to estimate the rotor position is relatively short Long, the update frequency of the position value is not high, so the high-frequency signal injection method has a better estimation effect when the motor is stationary and at low speed.

发明内容Contents of the invention

发明目的：为了克服现有技术中存在的不足，本发明提供一种变电站巡视机器人采用混合观测装置检测位置信息的方法，能够使永磁同步电机在全速下都能够完美稳定地运行。Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a method for a substation inspection robot to detect position information using a hybrid observation device, which can make the permanent magnet synchronous motor run perfectly and stably at full speed.

技术方案：为实现上述目的，本发明采用的技术方案为：Technical scheme: in order to achieve the above object, the technical scheme adopted in the present invention is:

变电站巡视机器人采用混合观测装置检测位置信息的方法，将初始位置检测闭环控制系统、模型参考归一化算法、模糊控制器、凸极效应补偿控制器和混合观测器结合在一起，以准确、有效的检测变电站巡视机器人的转子位置和速度信息，具体包括如下步骤：The substation inspection robot adopts the method of detecting position information by hybrid observation device, and combines the initial position detection closed-loop control system, model reference normalization algorithm, fuzzy controller, saliency effect compensation controller and hybrid observer together to achieve accurate and effective The detection of the rotor position and speed information of the substation inspection robot includes the following steps:

(1)在PMSM转子初始位置检测时采用初始位置跟踪闭环控制系统，首先在电流开环的情况下，在两相旋转坐标系下注入高频电压信号，通过构建转子位置跟踪闭环系统，估计出PMSM转子初始位置；(1) When detecting the initial position of the PMSM rotor, the initial position tracking closed-loop control system is adopted. Inject a high-frequency voltage signal into the coordinate system, and estimate the initial position of the PMSM rotor by constructing a rotor position tracking closed-loop system;

(2)变电站巡视机器人运行后，采用混合观测器对机器人位置信息进行实时在线检测，所述混合观察器的结构为：采用高频注入观测器和滑模观测器相结合的方式，并通过模糊控制器处理检测到的PMSM转子初始位置；(2) After the substation inspection robot is running, a hybrid observer is used to detect the position information of the robot in real time. The structure of the hybrid observer is: a combination of a high-frequency injection observer and a sliding The controller processes the detected initial rotor position of the PMSM;

高频注入观测器的检测原理为：通过软件锁相环实现对负序高频电流的相位的跟踪，从而获取矢量角误差，同时采用PI调节器调节矢量角的误差使之趋于零，使PMSM转子位置的估计值收敛于真实值θ_r；The detection principle of the high-frequency injection observer is as follows: the phase tracking of the negative-sequence high-frequency current is realized through the software phase-locked loop, so as to obtain the vector angle error, and at the same time, the PI regulator is used to adjust the error of the vector angle to make it tend to zero, so that Estimation of PMSM rotor position converges to the true value θ _r ;

滑模观测器的检测原理为：首先采用滑模观测器对扩展反电动势EEMF进行估算，随后通过检测电流和观测电流之间的误差构成滑模面对扩展反电动势EEMF进行观测，以获得转子位置检测值 The detection principle of the sliding mode observer is as follows: firstly, the sliding mode observer is used to estimate the extended back electromotive force EEMF, and then the error between the detected current and the observed current constitutes a sliding mode surface to observe the extended back electromotive force EEMF to obtain the rotor position detection value

模糊控制器用于替代传统的加权算法，将高频注入观测器和滑模观测器的检测误差ε和误差变化率dε作为模糊控制器的输入，PMSM转子位置信息作为模糊控制器的输出；The fuzzy controller is used to replace the traditional weighting algorithm, and the detection error ε and error change rate dε of the high-frequency injection observer and sliding mode observer are used as the input of the fuzzy controller, and the PMSM rotor position information as the output of the fuzzy controller;

(3)通过模型参考归一化算法得到PMSM的运行速度并将运行速度反馈到混合观测器中的模糊控制器中，调节混合因子a(e)，运行速度可改变混合因子a(e)函数的形状，根据不同的运行速度调整高频注入观测器和滑模观测器对控制输出的影响，获得不同的控制特性；(3) Obtain the running speed of PMSM through the model reference normalization algorithm and will run at speed Feedback to the fuzzy controller in the mixing observer, adjust the mixing factor a(e), the running speed The shape of the mixing factor a(e) function can be changed according to different operating speeds Adjust the influence of high-frequency injection observer and sliding mode observer on the control output to obtain different control characteristics;

(4)为了解决PMSM在低速下运行，凸极效应所带来的误差，采用SVPWM技术控制，每一次PWM周期中都有三种线性无关的V_m、V_n和V_l电压矢量，其中每一种电压矢量，对应着不同的响应电流值变i_αβm、i_αβn和i_αβl，根据前后两次线性无关的电压矢量而产生的电流响应值计算出电感矩阵，从而计算出交、直轴电感，并将交、直轴电感反馈到高频注入观测器和滑模观测器，对凸极效应进行补偿。(4) In order to solve the error caused by the saliency effect of PMSM running at low speed, SVPWM technology is used for control. There are three linearly independent voltage vectors of V _m , V _n and V _l in each PWM cycle, each of which The voltage vectors correspond to different response current values i _αβm , i _αβn and i _αβl , and the inductance matrix is calculated according to the current response values generated by the two linearly independent voltage vectors before and after, so as to calculate the cross-axis and direct-axis inductance, And the alternating and direct axis inductances are fed back to the high frequency injection observer and the sliding mode observer to compensate the saliency effect.

有益效果：本发明提供的变电站巡视机器人采用混合观测装置检测位置信息的方法，将初始位置检测闭环控制系统、混合观测器、模型参考归一化算法、模糊控制器、凸极效应补偿控制器和混合观测器结合在一起，具有如下特点：1、采用无位置传感器技术节约了硬件成本和维修成体，同时提高了系统的抗干扰性和鲁棒性；2、采用初始位置检测闭环控制系统，能够非常准确地检测电机的转子初始位置，可以实现机器人的顺利启动，并且可以实现最大转矩启动；3、采用混合观测器检测位置信息的方法，可以实现全速度周期的位置信息的实时在线检测，同时提高了变电站巡视机器人系统的稳定性和精确度；4、采用归一化算法估算机器人运行速度，可避免对角度微分得到速度而引入的测量噪音误差。Beneficial effects: the substation inspection robot provided by the present invention adopts the method of detecting position information by the hybrid observation device, and combines the initial position detection closed-loop control system, hybrid observer, model reference normalization algorithm, fuzzy controller, saliency effect compensation controller and The combination of hybrid observers has the following characteristics: 1. The use of position sensorless technology saves hardware costs and maintenance costs, and at the same time improves the anti-interference and robustness of the system; 2. Adopts the initial position detection closed-loop control system, which can Very accurate detection of the initial position of the rotor of the motor can realize the smooth start of the robot and the maximum torque start; 3. The method of detecting position information by using a hybrid observer can realize real-time online detection of position information in the full speed cycle, At the same time, the stability and accuracy of the substation inspection robot system are improved; 4. The normalization algorithm is used to estimate the running speed of the robot, which can avoid the measurement noise error introduced by the angle differential to obtain the speed.

附图说明Description of drawings

图1为旋转式高频电压注入法原理图；Figure 1 is a schematic diagram of the rotary high-frequency voltage injection method;

图2为带有滑模观测器的扩展反电动势检测法原理图；Fig. 2 is the schematic diagram of the extended back EMF detection method with a sliding mode observer;

图3为基于SVPWM的电压矢量与电流响应变化图；Fig. 3 is a voltage vector and current response change diagram based on SVPWM;

图4为带有模糊控制器的混合观测器；Figure 4 is a hybrid observer with a fuzzy controller;

图5为模糊控制器的原理图；Fig. 5 is the schematic diagram of fuzzy controller;

图6为混合函数因子曲线图；Fig. 6 is a mixing function factor curve diagram;

图7为归一化算法的转速估计器框图；Fig. 7 is the block diagram of the rotating speed estimator of normalization algorithm;

图8为初始位置跟踪闭环控制系统。Figure 8 is the initial position tracking closed-loop control system.

具体实施方式detailed description

下面结合附图对本发明作更进一步的说明。The present invention will be further described below in conjunction with the accompanying drawings.

下面结合本发明的设计思想做出进一步的分析和说明。Further analysis and description will be made below in conjunction with the design idea of the present invention.

变电站巡视机器人启动后，采用混合观测器对机器人位置信息进行实时在线检测，其中高频注入观测器的检测原理为：After the substation inspection robot is started, the hybrid observer is used to detect the position information of the robot in real time online. The detection principle of the high-frequency injection observer is as follows:

旋转高频电压信号注入的原理如图1所示，设旋转高频电压信号的角频率为ω_i、幅值为v_si，则旋转高频电压信号表示为：The principle of rotating high-frequency voltage signal injection is shown in Figure 1. Suppose the angular frequency of the rotating high-frequency voltage signal is ω _i and the amplitude is v _si , then the rotating high-frequency voltage signal Expressed as:

${v v}_{q q d d s the s i i}^{s the s} = = [\begin{matrix} {ν ν}_{q q s the s i i}^{s the s} \\ {ν ν}_{d d s the s i i}^{s the s} \end{matrix}] = = {ν ν}_{s the s i i} [\begin{matrix} c c o o s the s (({ω ω}_{i i} t t)) \\ - - s the s i i n no (({ω ω}_{i i} t t)) \end{matrix}] = = {ν ν}_{s the s i i} {e e}^{{jω jω}_{i i} t t} - - - - - - ((11))$

其中：为高频电压的q轴分量；为高频电压的d轴分量。in: is the q-axis component of the high-frequency voltage; is the d-axis component of the high-frequency voltage.

旋转高频电压信号激励下的三相逆变器输出端电机的直流响应为将经过带同滤波器BPF滤波后，得到dq轴高频电流为：The DC response of the motor at the output end of the three-phase inverter under the excitation of the rotating high-frequency voltage signal is Will After being filtered with the same filter BPF, the dq axis high frequency current is obtained for:

${i i}_{q q d d i i} = = {i i}_{d d q q s the s__i i p p}^{s the s} + + {i i}_{d d q q s the s__i i n no}^{s the s} = = {i i}_{i i p p} {e e}^{j j (({w w}_{r r} ((t t)) - - π π / / 22))} + + {i i}_{i i n no} {e e}^{j j ((22 {θ θ}_{r r} - - {w w}_{r r} ((t t)) + + π π / / 22))} - - - - - - ((22))$

式中正、负相序电流分量的幅值分别为：The amplitudes of the positive and negative phase sequence current components in the formula are:

$\{\begin{matrix} {i i}_{i i p p} = = [[\frac{Σ Σ L L}{{ΣL ΣL}^{22} - - {ΔL Δ L}^{22}}]] \frac{{U u}_{s the s i i}}{{w w}_{i i}} \\ {i i}_{i i n no} = = [[\frac{Δ Δ L L}{{ΣL ΣL}^{22} - - {ΔL ΔL}^{22}}]] \frac{{U u}_{s the s i i}}{{w w}_{i i}} \end{matrix}$

其中，ΣL＝(L_d+L_q)/2为d、q轴电感的平均值，ΔL＝(L_d-L_q)/2为d、q轴电感的半差电感；为正相序电流分量；为负相序电流分量；i_ip为正相序电流直流分量；i_in为负相序电流直流分量；w_r(t)为PMSM转子速度；w_i(t)为注入高频电压后反映的PMSM转子位置角速度，即低速段位置时PMSM转子的位置角；由于只有负相序电流分量包含PMSM转子的位置角信息θ_r，因此首先通过高通滤波器SFF将正相序电流分量滤除，再让负相序电流分量先乘以得到后，再乘以后，可得矢量角误差ε为：Among them, ΣL=(L _d +L _q )/2 is the average value of d and q axis inductance, ΔL=(L _d -L _q )/2 is the half-difference inductance of d and q axis inductance; is the positive phase sequence current component; is the negative phase sequence current component; i _ip is the positive phase sequence current DC component; i _in is the negative phase sequence current DC component; _w _r (t) is the PMSM rotor speed; PMSM rotor position angular velocity, that is, the position angle of the PMSM rotor at the position of the low-speed section; since only the negative phase sequence current component contains the position angle information θ _r of the PMSM rotor, so the positive phase sequence current component is first passed through the high-pass filter SFF filter out, and then let the negative phase sequence current component multiply by get After that, multiply by After that, the vector angle error ε can be obtained as:

其中：为负相序电流q轴分量；为负相序电流d轴分量；为PMSM转子位置的估计值，θ_r为PMSM转子的真实值；同时采用PI调节器调节矢量角的误差使之趋于零，就可以使PMSM转子位置的估计值收敛于真实值θ_r，对作时间微分，就可以估计PMSM转子角速度 in: is the q-axis component of the negative phase sequence current; is the d-axis component of the negative phase sequence current; is the estimated value of the PMSM rotor position, θ _r is the real value of the PMSM rotor; at the same time, the PI regulator is used to adjust the error of the vector angle to make it tend to zero, so that the estimated value of the PMSM rotor position converges to the true value θ _r , for By time differentiation, the PMSM rotor angular velocity can be estimated

当变电站巡视机器人处于运行状态后，采用滑模观测器来获取PMSM转子位置信息，结构框图如图2所示，在d-q旋转坐标系中PMSM的电压方程为：When the substation inspection robot is in the running state, the sliding mode observer is used to obtain the rotor position information of the PMSM. The structural block diagram is shown in Figure 2. The voltage equation of the PMSM in the d-q rotating coordinate system is:

$[\begin{matrix} {u u}_{d d} \\ {u u}_{q q} \end{matrix}] = = [\begin{matrix} R R + + {DL DL}_{d d} & - - {w w}_{r r} {L L}_{d d} \\ {w w}_{r r} {L L}_{d d} & R R + + {DL DL}_{q q} \end{matrix}] [\begin{matrix} {i i}_{d d} \\ {i i}_{q q} \end{matrix}] + + [\begin{matrix} 00 \\ {w w}_{r r} {K K}_{E E.} \end{matrix}] - - - - - - ((44))$

其中：[u_d u_q]^T为旋转坐标系下电压；[i_d i_q]^T为旋转坐标系下电流；R为定子电阻；D为微分算子；w_r为转子角速度(电角度)；K_E为反电势常数；L_d为d轴电感；L_q为q轴电感。Among them: [u _d u _q ] ^T is the voltage in the rotating coordinate system; [i _d i _q ] ^T is the current in the rotating coordinate system; R is the stator resistance; D is the differential operator; w _r is the rotor angular velocity (electrical angle) ; K _E is the back electromotive force constant; L _d is the d-axis inductance; L _q is the q-axis inductance.

将公式(4)变换到α-β静止坐标系下，得到：Transform the formula (4) into the α-β static coordinate system to get:

$[\begin{matrix} {u u}_{α α} \\ {u u}_{β β} \end{matrix}] = = [\begin{matrix} R R + + {DL DL}_{α α} & - - {w w}_{r r} {L L}_{α α β β} \\ {w w}_{r r} {L L}_{α α β β} & R R + + {DL DL}_{β β} \end{matrix}] [\begin{matrix} {i i}_{α α} \\ {i i}_{β β} \end{matrix}] + + {w w}_{r r} {K K}_{E E.} [\begin{matrix} - - {sinθ sinθ}_{r r} \\ {cosθ cosθ}_{r r} \end{matrix}] - - - - - - ((55))$

其中：[u_α u_β]^T为旋转坐标系下电压；[i_α i_β]^T为旋转坐标系下电流；L_α＝L_o+L₁cos2θ_r；L_αβ＝L₁sin2θ_r；L_β＝L_o-L₁cos2θ；L_o＝(L_d+L_q)/2；L₁＝(L_d-L_q)/2；θ_r为海浪发电系统PMSM在运行时的PMSM位置角。Among them: [u _α u _β ] ^T is the voltage in the rotating coordinate system; [i _α i _β ] ^T is the current in the rotating coordinate system; L _α ＝L _o +L ₁ cos2θ _r ; L _αβ ＝L ₁ sin2θ _r ; L _β =L _o -L ₁ cos2θ; L _o =(L _d +L _q )/2; L ₁ =(L _d -L _q )/2; θ _r is the PMSM position angle of the wave power generation system PMSM during operation.

公式(4)中包含有θ_r、2θ_r项，其中2θ_r将给后期的计算带来很大的难度，因此，可以通过适当的变换使其消除，从公式(5)中可以看出：电感矩阵的不对称是2θ_r的出现的主要原因，因而，将d-q轴下的PMSM的电压方程(4)重写为：Formula (4) contains the terms θ _r and 2θ _r , among which 2θ _r will bring great difficulty to the later calculation, so it can be eliminated through appropriate transformation, as can be seen from formula (5): The asymmetry of the inductance matrix is the main reason for the appearance of 2θ _r , therefore, the voltage equation (4) of the PMSM under the dq axis is rewritten as:

$[\begin{matrix} {u u}_{d d} \\ {u u}_{q q} \end{matrix}] = = [\begin{matrix} R R + + {DL DL}_{d d} & - - {w w}_{r r} {L L}_{q q} \\ {w w}_{r r} {L L}_{q q} & R R + + {DL DL}_{d d} \end{matrix}] [\begin{matrix} {i i}_{d d} \\ {i i}_{q q} \end{matrix}] + + [\begin{matrix} 00 \\ {w w}_{r r} {K K}_{E E.} + + (({L L}_{d d} - - {L L}_{q q})) (({w w}_{r r} {i i}_{d d} - - {di di}_{q q} / / d d t t)) \end{matrix}] - - - - - - ((66))$

公式(6)变换到α-β静止坐标系下，得：Formula (6) is transformed into the α-β stationary coordinate system, and we get:

$[\begin{matrix} {u u}_{α α} \\ {u u}_{β β} \end{matrix}] = = [\begin{matrix} R R + + {DL DL}_{d d} & {w w}_{r r} (({L L}_{d d} - - {L L}_{q q})) \\ - - {w w}_{r r} (({L L}_{d d} - - {L L}_{q q})) & R R + + {DL DL}_{d d} \end{matrix}] [\begin{matrix} {i i}_{α α} \\ {i i}_{β β} \end{matrix}] + + [[(({w w}_{r r} {K K}_{E E.} + + (({L L}_{d d} - - {L L}_{q q})) (({w w}_{r r} {i i}_{d d} - - \frac{{di di}_{q q}}{d d t t}))]] [\begin{matrix} - - {sinθ sinθ}_{r r} \\ {cosθ cosθ}_{r r} \end{matrix}] - - - - - - ((77))$

为了便于使用滑模观测器对反电动势进行观测，将电压方程(4)改写成电流的状态方程形式：In order to use the sliding mode observer to observe the back electromotive force, the voltage equation (4) is rewritten into the form of the state equation of the current:

$\frac{d d}{d d t t} [\begin{matrix} {i i}_{α α} \\ {i i}_{β β} \end{matrix}] = = A A \cdot &Center Dot; [\begin{matrix} {i i}_{α α} \\ {i i}_{β β} \end{matrix}] + + \frac{11}{{L L}_{d d}} [\begin{matrix} {u u}_{α α} \\ {u u}_{β β} \end{matrix}] + + \frac{E E.}{{L L}_{d d}} [\begin{matrix} {sinθ sinθ}_{m m} \\ - - {cosθ cosθ}_{m m} \end{matrix}] - - - - - - ((88))$

其中：in:

$A A = = \frac{11}{{L L}_{d d}} [\begin{matrix} - - R R & - - {ω ω}_{r r} (({L L}_{d d} - - {L L}_{q q})) \\ {ω ω}_{r r} (({L L}_{d d} - - {L L}_{q q})) & - - R R \end{matrix}]$

反电动势 $E = [\begin{matrix} E_{α} \\ E_{β} \end{matrix}] = [(w_{r} K_{E} + (L_{d} - L_{q}) (w_{r} i_{d} - \frac{{di}_{q}}{d t})] [\begin{matrix} - {sinθ}_{r} \\ {cosθ}_{r} \end{matrix}]$ Counter electromotive force $E. = [\begin{matrix} {E.}_{α} \\ {E.}_{β} \end{matrix}] = [(w_{r} K_{E.} + (L_{d} - L_{q}) (w_{r} i_{d} - \frac{{di}_{q}}{d t})] [\begin{matrix} - {sinθ}_{r} \\ {cosθ}_{r} \end{matrix}]$

构造如下的滑模观测器：Construct the following sliding mode observer:

$\frac{d d}{d d t t} [\begin{matrix} {\overset{^^}{i i}}_{α α} \\ {\overset{^^}{i i}}_{β β} \end{matrix}] = = A A \cdot &Center Dot; [\begin{matrix} {\overset{^^}{i i}}_{α α} \\ {\overset{^^}{i i}}_{β β} \end{matrix}] + + \frac{11}{{L L}_{d d}} [\begin{matrix} {u u}_{α α} \\ {u u}_{β β} \end{matrix}] + + \frac{{Z Z}_{α α β β}}{{L L}_{d d}} - - - - - - ((99))$

其中： ${[\begin{matrix} {\hat{i}}_{α} & {\hat{i}}_{β} \end{matrix}]}^{T}$ 为定子α和β轴电流观测值， $Z_{α β} = [h sgn (\hat{i} - i_{α}), h sgn (\hat{i} - i_{β})],$ h为滑膜增益，sgn是符号函数。in: ${[\begin{matrix} {\hat{i}}_{α} & {\hat{i}}_{β} \end{matrix}]}^{T}$ are the observed values of stator α and β axis currents, $Z_{α β} = [h sgn (\hat{i} - i_{α}), h sgn (\hat{i} - i_{β})],$ h is the synovial gain and sgn is the sign function.

公式(9)减去公式(8)，得到电流观测误差的状态方程为：Subtracting formula (8) from formula (9), the state equation of the current observation error is obtained as:

当满如下条件时，滑模观测器进入滑模状态：When the following conditions are met, the sliding mode observer enters the sliding mode state:

$[[{i i}_{α α} - - {\overset{^^}{i i}}_{α α},, {i i}_{β β} - - {\overset{^^}{i i}}_{β β}]] \frac{d d}{t t} [\begin{matrix} {i i}_{α α} - - {\overset{^^}{i i}}_{α α} \\ {i i}_{β β} - - {\overset{^^}{i i}}_{β β} \end{matrix}] < < 00 - - - - - - ((1111))$

若滑模增益k足够大，系统进入滑膜状态，有：If the sliding mode gain k is large enough, the system enters the sliding film state, as follows:

$\frac{d d}{t t} [\begin{matrix} {i i}_{α α} - - {\overset{^^}{i i}}_{α α} \\ {i i}_{β β} - - {\overset{^^}{i i}}_{β β} \end{matrix}] = = [\begin{matrix} {i i}_{α α} - - {\overset{^^}{i i}}_{α α} \\ {i i}_{β β} - - {\overset{^^}{i i}}_{β β} \end{matrix}] = = 00 - - - - - - ((1212))$

将上式(12)带入到公式(10)，得到：Putting the above formula (12) into formula (10), we get:

Z＝E (13)Z=E (13)

其中Z中包含有不连续高频信号，因此为去除不连续高频信号，将其通入低通滤波器后得到等价控制量，即：Among them, Z contains discontinuous high-frequency signals, so in order to remove discontinuous high-frequency signals, the equivalent control amount is obtained after passing them through a low-pass filter, namely:

$[\begin{matrix} {Z Z}_{α α} \\ {Z Z}_{β β} \end{matrix}] = = [\begin{matrix} {E E.}_{α α} \\ {E E.}_{β β} \end{matrix}] = = [[(({w w}_{r r} {K K}_{E E.} + + (({L L}_{d d} - - {L L}_{q q})) (({w w}_{r r} {i i}_{d d} - - \frac{{di di}_{q q}}{d d t t}))]] [\begin{matrix} - - {sinθ sinθ}_{r r} \\ {cosθ cosθ}_{r r} \end{matrix}] - - - - - - ((1414))$

由公式(14)，可以得到PMSG在高速运行时的转子位置角 From formula (14), the rotor position angle of the PMSG at high speed can be obtained

${\overset{^^}{θ θ}}_{e e r r} = = a a r r c c t t a a n no ((- - \frac{{E E.}_{α α}}{{E E.}_{β β}})) - - - - - - ((1515))$

为了避免电机凸极效应的影响，利用机器人系统采用SVPWM控制技术，实时对d、q轴电感进行在线辨识，构建凸极效应补偿控制器对混合观测器进行补偿，其d、q轴电感的求解原理如下：In order to avoid the influence of motor saliency effect, the robot system adopts SVPWM control technology to carry out online identification of d and q axis inductance in real time, build a saliency effect compensation controller to compensate the hybrid observer, and solve the d and q axis inductance The principle is as follows:

图3为基于SVPWM的电压矢量与电流响应变化图，为将检测到的电流峰值作为电流响应的变化量，电感矩阵可以表示为：Figure 3 is a graph of voltage vector and current response changes based on SVPWM. In order to use the detected current peak as the change in current response, the inductance matrix can be expressed as:

$\begin{matrix} [\begin{matrix} {L L}_{1111} & {L L}_{1212} \\ {L L}_{21 twenty one} & {L L}_{22 twenty two} \end{matrix}] = = [\begin{matrix} {u u}_{α α 11} - - {Ri Ri}_{α α 11} & {u u}_{α α 22} - - {Ri Ri}_{α α 22} \\ {u u}_{β β 11} - - {Ri Ri}_{β β 11} & {u u}_{β β 22} - - {Ri Ri}_{β β 22} \end{matrix}] \\ {[\begin{matrix} \frac{{i i}_{α α 11}}{Δ Δ t t} & \frac{{i i}_{α α 22}}{Δ Δ t t} \\ \frac{{i i}_{β β 11}}{Δ Δ t t} & \frac{{i i}_{β β 22}}{Δ Δ t t} \end{matrix}]}^{- - 11} / / H h ((t t)) = = \frac{Δ Δ t t}{H h ((t t)) (({i i}_{α α 11} {i i}_{β β 22} - - {i i}_{α α 22} {i i}_{β β 11}))} \\ [\begin{matrix} {u u}_{α α 11} - - {Ri Ri}_{α α 11} & {u u}_{α α 11} - - {Ri Ri}_{α α 22} \\ {u u}_{β β 11} - - {Ri Ri}_{β β 11} & {u u}_{β β 22} - - {Ri Ri}_{β β 22} \end{matrix}] [\begin{matrix} {i i}_{β β 22} & - - {i i}_{α α 22} \\ - - {i i}_{β β 11} & {i i}_{α α 11} \end{matrix}] \end{matrix} - - - - - - ((1616))$

其中R为定子电阻，Δt为SVPWM下的连续两次电压矢量施加的时间间隔，i_α1、i_β1、i_α2、i_β2分别为αβ坐标系下1、2次电压矢量的电流响应值，可以看出，公式(16)可以获得PMSM的电感参数，从而求出d、q轴电感，如公式(17)和(18)所示in R is the stator resistance, Δt is the time interval between two consecutive voltage vector applications under SVPWM, i _α1 , i _β1 , i _α2 , and i _β2 are the current response values of the first and second voltage vectors in the αβ coordinate system, respectively, which can be seen Then, formula (16) can obtain the inductance parameter of PMSM, so as to obtain the d and q axis inductance, as shown in formulas (17) and (18)

L_d＝L₁+L₂＝[L₁₁+L₂₂+(L₁₁-L₂₂)/cos2θ_r]/2 (17)L _d =L ₁ +L ₂ =[L ₁₁ +L ₂₂ +(L ₁₁ -L ₂₂ )/cos2θ _r ]/2 (17)

L_q＝L₁-L₂＝[L₁₁+L₂₂-(L₁₁-L₂₂)/cos2θ_r]/2 (18)L _q =L ₁ -L ₂ =[L ₁₁ +L ₂₂ -(L ₁₁ -L ₂₂ )/cos2θ _r ]/2 (18)

对于滑模观测器，将公式(17)和(18)所求的L_d和L_q，和检测到的i_α、i_β与w_r一起代入公式(7)即可获得：For the sliding mode observer, substituting the L _d and L _q calculated by the formulas (17) and (18), together with the detected i _α , i _β and w _r into the formula (7) can be obtained:

${\overset{^^}{θ θ}}_{h h r r} = = a a r r c c t t a a n no ((- - \frac{{E E.}_{α α}}{{E E.}_{β β}})) = = a a r r c c t t a a n no ((- - \frac{{u u}_{α α} - - {K K}_{11}}{{u u}_{β β} - - {K K}_{22}})) - - - - - - ((1919))$

其中K₁、K₂为计算获得的系数值，因此加入凸极效应跟踪观测器，直接利用检测到的α、β轴电压进行计算，很大程度上减少了运算的复杂性。Among them, K ₁ and K ₂ are calculated coefficient values. Therefore, the salient pole effect tracking observer is added, and the detected α and β axis voltages are directly used for calculation, which greatly reduces the complexity of the calculation.

对于高频信号注入观测器，将公式(17)和(18)把实时辨识的电感L_d和L_q反馈给位置估算模块PI调节器，可以获得更加准确的估算信息。For the high-frequency signal injection observer, formulas (17) and (18) can be used to feed back the real-time identified inductance L _d and L _q to the PI regulator of the position estimation module, so that more accurate estimation information can be obtained.

混合观测装置的主要由模糊控制器组成，代替传统的加权算法。将跟踪误差ε和误差变化率dε作为模糊控制器的输入，变电站巡视机器人的位置信息作为模糊控制器的输出，模糊控制器具体结构图如图4所示，其具体原理如下：The hybrid observation device is mainly composed of a fuzzy controller, which replaces the traditional weighting algorithm. Taking the tracking error ε and error change rate dε as the input of the fuzzy controller, the position information of the substation inspection robot As the output of the fuzzy controller, the specific structure diagram of the fuzzy controller is shown in Figure 4, and its specific principles are as follows:

首先射跟踪误差ε和误差变化率dε定义为：Firstly, the tracking error ε and error change rate dε are defined as:

$\{\begin{matrix} ϵ ϵ ((k k)) = = {\overset{^^}{θ θ}}_{e e r r} ((k k)) - - {\overset{^^}{θ θ}}_{h h r r} ((k k)) \\ d d ϵ ϵ ((k k)) = = {\overset{^^}{θ θ}}_{e e r r} ((k k)) - - {\overset{^^}{θ θ}}_{e e r r} ((k k - - 11)) \end{matrix} - - - - - - ((2020))$

其图4中参考模型的传递函数为：The transfer function of the reference model in Figure 4 is:

$\frac{{\overset{^^}{θ θ}}_{e e r r} ((s the s))}{{\overset{^^}{θ θ}}_{h h r r} ((s the s))} = = \frac{{ω ω}_{n no}^{22}}{{s the s}^{22} + + 22 {ζω ζω}_{n no} s the s + + {ω ω}_{n no}^{22}} - - - - - - ((21 twenty one))$

根据变电站机器人响应快速且稳态误差为零的系统。选择阻尼系数ζ＝1，则在上述约束下间隔时间Δt和自然频率ω_n之间的关系如下：According to the substation robot system with fast response and zero steady-state error. If the damping coefficient ζ=1 is selected, the relationship between the interval time Δt and the natural frequency ω _n under the above constraints is as follows:

$((11 - - {ω ω}_{n no} Δ Δ t t)) {e e}^{- - {ω ω}_{n no} {t t}_{r r}} = = 0.1 0.1 - - - - - - ((22 twenty two))$

只要Δt给定，就可以求出ω_n。并可以得到传递函数(21)的离散形式：As long as Δt is given, ω _n can be obtained. And the discrete form of the transfer function (21) can be obtained:

$\frac{{\overset{^^}{θ θ}}_{e e r r} (({z z}^{- - 11}))}{{\overset{^^}{θ θ}}_{h h r r} (({z z}^{- - 11}))} = = \frac{a a ((e e)) ((11 + + {z z}^{- - 11} + + {z z}^{- - 22}))}{[[11 - - a a ((e e))]] ((11 + + {z z}^{- - 11} + + {z z}^{- - 22}))} - - - - - - ((23 twenty three))$

差分方程为：The difference equation is:

${\overset{^^}{θ θ}}_{e e r r} ((k k)) = = ((11 - - a a ((e e)))) [[- - {\overset{^^}{θ θ}}_{h h r r} ((k k - - 11)) - - {\overset{^^}{θ θ}}_{h h r r} ((k k - - 22))]] - - a a ((e e)) [[{\overset{^^}{θ θ}}_{e e r r} ((k k)) + + {\overset{^^}{θ θ}}_{e e r r} ((k k - - 11)) + + {\overset{^^}{θ θ}}_{e e r r} ((k k - - 22))]] - - - - - - ((24 twenty four))$

其实a(e)为混合因子，校正单元采用了逐步下降法，通过最小化参考模型输出与高频信号注入观测器检测值差值的平方来校正模糊控制器的模糊参数：In fact, a(e) is a mixing factor, and the correction unit adopts the step-by-step descent method, by minimizing the reference model output Injection observer detection value with high frequency signal The square of the difference is used to correct the fuzzy parameters of the fuzzy controller:

$J J ((k k + + 11)) = = \frac{11}{22} ϵ ϵ {((k k + + 11))}^{22} = = \frac{11}{22} {[[{\overset{^^}{θ θ}}_{e e r r} ((k k + + 11)) - - {\overset{^^}{θ θ}}_{h h r r} ((k k + + 11))]]}^{22} - - - - - - ((2525))$

${Δc Δ c}_{m m,, n no} ((k k + + 11)) &Proportional; &Proportional; - - \frac{\partial \partial J J ((k k + + 11))}{\partial \partial {c c}_{m m,, n no} ((k k))} = = - - α α \frac{\partial \partial J J ((k k + + 11))}{\partial \partial {c c}_{m m,, n no} ((k k))} - - - - - - ((2626))$

经过以上的过程加工，其输出量即为变电站巡视机器人的位置信息 After the above process processing, the output is the position information of the substation inspection robot

关于公式(23)和(24)中混合因子a(e)的确定为也为本专利的重点之一，将a(e)构造成：The determination of the mixing factor a(e) in formulas (23) and (24) is also one of the key points of this patent, and a(e) is constructed as:

$\{\begin{matrix} a a ((e e)) = = 00 & {\overset{^^}{w w}}_{r r} < < {w w}_{11} \\ a a ((e e)) = = \frac{exp exp (({\overset{^^}{w w}}_{r r} ρ ρ)) - - exp exp (({w w}_{11} ρ ρ))}{exp exp (({w w}_{22} ρ ρ)) - - exp exp (({w w}_{11} ρ ρ))} & {w w}_{11} < < {\overset{^^}{w w}}_{r r} \leq \leq {w w}_{22} \\ a a ((e e)) = = 11 & {\overset{^^}{w w}}_{r r} > > {w w}_{22} \end{matrix} - - - - - - ((2727))$

公式中：w₁和w₂分别为转速的分界点，当时，只有高频电压观测器单独作用，当完全实施滑模观测器检测，在w₁和w₂之间时，高频电压观测器和滑模观测器共同检测。其中图6为混合函数因子曲线图。In the formula: w ₁ and w ₂ are the dividing points of the speed respectively, when When , only the high-frequency voltage observer works alone, when The sliding mode observer is fully implemented, and between w ₁ and w ₂ , the high frequency voltage observer and the sliding mode observer are jointly detected. Figure 6 is a graph of the mixing function factor.

公式(27)中采用模型参考归一化算法得到，根据公式：In formula (27) It is obtained by using the model reference normalization algorithm, according to the formula:

${\overset{^^}{θ θ}}_{h h r r} = = a a r r c c t t a a n no ((- - \frac{{E E.}_{α α}}{{E E.}_{β β}})) = = a a r r c c t t a a n no ((- - \frac{{u u}_{α α} - - {K K}_{11}}{{u u}_{β β} - - {K K}_{22}})) - - - - - - ((2828))$

机器人PMSM估算速度可用模型参考归一化算法得到，估计算法的结构如图7所示，这种方法可避免对角度微分得到速度而引入的测量噪音误差。Robotic PMSM estimated velocity It can be obtained by the model reference normalization algorithm, and the structure of the estimation algorithm is shown in Figure 7. This method can avoid the measurement noise error introduced by the angle differential to obtain the velocity.

首先，对检测到的α、β轴的反电动势进行归一化处理First, normalize the detected back electromotive force of the α and β axes

${E E.}_{n no} = = \frac{11}{\sqrt{{E E.}_{α α}^{22} + + {E E.}_{β β}^{22}}} [\begin{matrix} {λ λ}_{α α} \\ {λ λ}_{β β} \end{matrix}] - - - - - - ((2929))$

对于α、β轴的反电动势，速度w_r变化要慢得多，可以看作常量，从而有：For the back electromotive force of the α and β axes, the speed w _r changes much more slowly and can be regarded as a constant, thus:

dE/dt＝w_rJE (30)dE/dt=w _r JE (30)

其中 $J = [\begin{matrix} 0 & - 1 \\ 1 & 0 \end{matrix}] .$ in $J = [\begin{matrix} 0 & - 1 \\ 1 & 0 \end{matrix}] .$

以上作为归一化算法的参考模型，可调模型定义为：The above is the reference model of the normalization algorithm, and the adjustable model is defined as:

$d d \overset{^^}{E E.} / / d d t t = = {\overset{^^}{w w}}_{r r} J J \overset{^^}{E E.} + + G G ((\overset{^^}{E E.} - - E E.)) - - - - - - ((3131))$

其中：为可调模型的输出；为估计速度；G为反馈回路增益，其目的是为了使可调模型收敛。in: is the output of the adjustable model; For the estimated speed; G is the feedback loop gain, its purpose is to make the adjustable model converge.

当速度估算存在误差时，将导致归一化的反电势产生误差这一误差与可调模型的输出一起，得到下面的定律：When there is an error in the velocity estimate, it will result in an error in the normalized back EMF This error is related to the adjustable model output Together, we get the following law:

其实k_p和k_i为PI调节器的调节参数，可调模型收敛后，收敛到零，从而估算的速度最终收敛到实际速度w_r。In fact, k _p and _ki are the adjustment parameters of the PI regulator. After the adjustable model converges, converges to zero, thus estimating the speed of Eventually it converges to the actual speed w _r .

初始位置检测原理：Initial position detection principle:

当估计的转子初始位置与实际的初始位置足够相近时，即Δθ≈0，存在sin(2Δθ)≈2Δθ。图8中，对ε_Δθ信号积分后得到转子位置初次估计值，可表达如下：When the estimated rotor initial position is close enough to the actual initial position, that is, Δθ≈0, there exists sin(2Δθ)≈2Δθ. In Fig. 8, the initial estimated value of the rotor position is obtained after integrating the ε _Δθ signal, which can be expressed as follows:

$\overset{^^}{θ θ} = = {ϵ ϵ}_{Δ Δ θ θ} \cdot &Center Dot; \frac{{k k}_{i i}}{s the s} = = {i i}_{i i n no} s the s i i n no 22 {Δθ Δθ}_{r r} \cdot &Center Dot; \frac{{k k}_{i i}}{s the s} \approx \approx \frac{{kk kk}_{i i}}{s the s} {Δθ Δθ}_{r r} = = 22 {kk kk}_{i i} \cdot \cdot \frac{11}{s the s} \cdot \cdot {Δθ Δθ}_{r r} - - - - - - ((3333))$

式中，k_i为积分增益，k_i>0。In the formula, _ki is the integral gain, and _ki >0.

可将公式(33)等效为初始位置跟踪闭环控制系统，其结构如图4所示，在初始位置检测过程中，转子位置初始估计系统是为了实现转子位置估计值对实际值的准确跟踪，而电机的实际转子初始位置是固定不变得，它可以看作为初次估计系统的阶跃信号给定，它为I型系统，理论上可实现估计初始位置对实际位置的无静差跟踪。该系统的闭环传递函数可写成Φ(s)＝2i_ink_i/(s+2i_ink_i)，为保证该系统为负反馈稳定系统，闭环传递函数的极点必须位于复平面左半平面，必须满足i_in>0，也保证了系统开环增益为正值。The formula (33) can be equivalent to the initial position tracking closed-loop control system, and its structure is shown in Figure 4. During the initial position detection process, the rotor position initial estimation system is to realize the accurate tracking of the rotor position estimated value to the actual value. The actual rotor initial position of the motor is fixed, and it can be regarded as the step signal given by the initial estimation system. It is an I-type system, and theoretically, it can realize the static error-free tracking of the estimated initial position to the actual position. The closed-loop transfer function of the system can be written as Φ(s)=2i _in _ki /(s+2i _in _ki ), in order to ensure that the system is a negative feedback stable system, the pole of the closed-loop transfer function must be located in the left half plane of the complex plane, It must satisfy i _in >0, which also ensures that the system open-loop gain is a positive value.

以实际转子初始位置θ_r∈(π/2,π)为例，此时，Δθ_r∈(0,π/2)rad，由公式(3)可知对ε_Δθ进行PI调节后得到位置估计值将从初始值0rad逐渐变大，而转子的实际位置θ_r不变，则Δθ_r会变小；通过图4所示的闭环调节系统，最终Δθ_r会收敛到0rad并保持稳定；根据可以得到转子位置初始估计值的 Taking the actual rotor initial position θ _r ∈ (π/2, π) as an example, at this time, Δθ _r ∈ (0, π/2) rad, it can be known from formula (3) The position estimate is obtained after PI adjustment of ε _Δθ will gradually increase from the initial value 0rad, but the actual position of the rotor θr remains unchanged, then _Δθr will become smaller; through the closed-loop adjustment system shown in Figure 4, _Δθr will eventually converge to _0rad and remain stable; according to An initial estimate of the rotor position can be obtained from

以上所述仅是本发明的优选实施方式，应当指出：对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made. It should be regarded as the protection scope of the present invention.

Claims

1. the method that substation inspection robot uses mixing observation device detection positional information, it is characterised in that: will just Beginning position detection closed-loop control system, model reference normalization algorithm, fuzzy controller, saliency compensating controller and Hybrid observer combines, with rotor-position and the velocity information of accurate and effective detection substation inspection robot, Specifically include following steps:

(1) initial position closed-loop tracking control system is used when PMSM initial position of rotor detects, first at electric current In the case of open loop, rotate in two-phaseInject high-frequency voltage signal under coordinate system, follow the tracks of by building rotor-position Closed-loop system, estimates PMSM initial position of rotor；

(2), after substation inspection robot runs, use hybrid observer that robot location's information is carried out real-time online Detection, the structure of described mixing viewer is: use the mode that high frequency injection observer and sliding mode observer combine, and The PMSM initial position of rotor detected is processed by fuzzy controller；

High frequency injects the Cleaning Principle of observer: by software phase-lock loop realize to the phase place of negative phase-sequence high frequency electric with Track, thus obtain azimuth error, use the error of pi regulator regulation azimuth to be allowed to go to zero simultaneously, make PMSM The estimate of rotor-positionConverge on actual value θ_r；

The Cleaning Principle of sliding mode observer is: estimate extension counter electromotive force EEMF initially with sliding mode observer, Constitute sliding-mode surface by the error between detection electric current and observation electric current subsequently extension counter electromotive force EEMF is observed, To obtain rotor-position detected value

High frequency, for substituting traditional weighting algorithm, is injected the detection error of observer and sliding mode observer by fuzzy controller ε and error rate d ε is as the input of fuzzy controller, PMSM rotor position informationAs fuzzy controller Output；

(3) speed of service of PMSM is obtained by model reference normalization algorithmAnd by the speed of serviceFeedback In fuzzy controller in hybrid observer, regulation hybrid cytokine a (e), the speed of serviceHybrid cytokine a (e) can be changed The shape of function, according to the different speeds of serviceAdjust high frequency and inject observer and the sliding mode observer shadow to controlling output Ring, it is thus achieved that different control characteristics；

(4) run under the low speed to solve PMSM, the error that saliency is brought, use SVPWM technology Control, PWM cycle has the V of three kinds of linear independences each time_m、V_nAnd V_lVoltage vector, each of which voltage Vector, correspond to different response current values and becomes i_αβm、i_αβnAnd i_αβl, according to the voltage vector of twice linear independence of front and back And the current-responsive value produced calculates inductance matrix, thus calculate cross, straight axle inductance, and by anti-for cross, straight axle inductance It is fed to high frequency and injects observer and sliding mode observer, saliency is compensated.