Engine starting motor position-free control method with continuous estimation angle in starting process

Through the embedded processor module and multi-stage phase-locked loop structure, continuous angle estimation and control of the aircraft engine starting process is realized, and the loop bandwidth reduction of high-frequency injection method and I/F method in high-temperature and high-speed environments is solved. The switching point angle is intermittent, providing stable and high-precision start control.

CN119051516APending Publication Date: 2024-11-29GUIZHOU AEROSPACE LINQUAN MOTOR CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411167200.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing high-frequency injection method and I/F method have problems such as reducing loop bandwidth and intermittent switching point angle during the start of the aircraft engine, and it is difficult to effectively control the starting process of the turbine engine in high temperature, high speed and high vibration environments.

Method used

Adopt the embedded processor-based functional module, including I/F generator, angular frequency estimation, angular frequency weighted superposition, position estimation, reference current calculation, current loop calculation and eαβ calculation module, through multi-stage phase-locked loop and maximum rotation Moment and current ratio control realizes continuous angle estimation and control of the starting process.

Benefits of technology

This method overcomes the impact of high-frequency injection method on loop bandwidth and the intermittent switching point angle of I/F method, provides more stable dynamic characteristics and high-precision position estimation, and is suitable for permanent magnets with magnetoresistive torque. Motor-started turbocharged aircraft engines without the need for position sensors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119051516A_ABST
    Figure CN119051516A_ABST
Patent Text Reader

Abstract

An engine starting motor position-free control method with continuous estimation angles in the starting process is composed of an I / F generator module, an angular frequency estimation phase-locked loop module, an angular frequency weighted stacking module, a position estimation module, a reference current calculation module, a current loop calculation module, an e alpha beta calculation module and the like which are realized based on an embedded processor. During starting, angular frequencies generated by the I / F generator and the angular frequency estimation phase-locked loop are weighted through angular frequency weighting, the weighted angular frequencies are input into the position estimation module to serve as angular frequency offset, and then the angular frequencies and the rotor position are obtained through the position estimation phase-locked loop. According to the starting control method, angle estimation in the starting process is continuous all the time, an angle switching process does not exist, and the starting success rate can be increased. In the starting process, before the rotating speed of the engine reaches the ignition rotating speed, the output current reference is the maximum current capable of being output by the controller, and after the engine reaches the ignition rotating speed, the output current is gradually reduced according to the characteristics of the engine.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of motor control, and specifically relates to a method for position-free control of an engine starting motor with continuous estimated angle during the starting process, and is particularly suitable for the application field of position-free motor control of starting a turbine aircraft engine using a permanent magnet motor with reluctance torque. Background Art

[0002] Aircraft engine starter motors operate in a high temperature, high speed, high vibration, and highly corrosive environment. Position sensors such as photoelectric encoders, Hall sensors, and resolvers cannot be used in this extreme environment to obtain the rotor position required for motor control. A position-free control method is required to start the engine.

[0003] According to the scope of application of the position-free control method, the position-free control method is usually divided into two categories: the position-free control method applicable to medium and high speed and the position-free control method applicable to zero speed and low speed. The position-free control method applicable to medium and high speed is based on the motor model and observes the motor speed and rotor position through the motor back electromotive force or magnetic flux. The control methods used for zero speed and low speed are mainly high-frequency injection method and I / F method. The high-frequency injection method injects high-frequency voltage or high-frequency current into the motor and obtains the motor speed and rotor position by extracting high-frequency signal feedback. The I / F method is an open-loop control method. It does not need to obtain accurate rotor position information during control. The rotating magnetic field generated by the stator current is used to pull the rotor into synchronization. After reaching the position where the rotor position can be obtained through the motor model, the angle is switched to the real angle for control. The position used by the I / F method at the start is not the real rotor position, and it cannot be started with heavy load. Therefore, the I / F method is often used in occasions with pump load characteristics.

[0004] The torque-speed characteristics of a turbine engine at startup are similar to those of a pump load, and its resistance torque includes a fixed base value T a and Bn which is proportional to the square of the speed 2 The starting torque required is small when the speed is low. As the speed increases, the torque required to start the engine increases at the square of the speed. Assume that the rated operating speed of the turbine engine is n N , ignition speed is n f , the starting speed is n2; the general ignition speed is n f =(0.08-0.15)n N , starting and dragging speed n2 = (0.3-0.45) n N .

[0005] Normally, the starting motor can be observed by back EMF to obtain an angular frequency and an angular speed no higher than 0.4n f , the angular frequency and the rotational speed can be reliably observed to be no higher than 0.6n f .

[0006] For engine starting applications, the high frequency injection method and I / F method have the following shortcomings:

[0007] 1. The high-frequency injection method uses a low-pass filter (LPF) to filter out the injected high-frequency information, which will filter out the high-frequency useful information of the loop control, making the loop bandwidth very low, affecting the normal loop control;

[0008] 2. From the perspective of load characteristics, the I / F method is suitable for engine starting, but the I / F method has a switching point. The open-loop angle given by the I / F method is not equal to the angle estimated by the back-EMF position-free detection method. There is an angle discontinuity at the switching point, which can easily lead to switching failure. Summary of the invention

[0009] The object of the present invention is to provide a method for position-free control of an engine starter motor with continuous estimated angle during the starting process in view of the deficiencies of the above-mentioned technology.

[0010] The present invention is achieved through the following technical solutions.

[0011] The present invention provides a method for controlling an engine starter motor without position during a starting process with a continuous estimation angle, using a functional module based on an embedded processor, including an I / F generator, an angular frequency estimation module, an angular frequency weighted superposition module, a position estimation module, a reference current calculation module, a current loop calculation module, an αβ The calculation module specifically comprises the following steps:

[0012] Step 1. The I / F generator outputs a current reference I ref To the reference current calculation module, the I / F generator module outputs the weighting coefficient K and the reference angular frequency ω ref To the angular frequency weighted superposition module;

[0013] Step 2: The angular frequency estimation module outputs the initial estimated angular frequency ω1 to the angular frequency weighted superposition module, and the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 To the position estimation module and e αβ Computation module;

[0014] Step 3. The angular frequency weighted superposition module outputs the complementary weighting coefficient K N To the position estimation module;

[0015] Step 4. The position estimation module outputs the accurate estimated angular frequency ω to the I / F generator module, and the position estimation module outputs the estimated position θ to the current loop calculation module.

[0016] Step 5. The reference current calculation module outputs the d-axis current reference i dref and q-axis current reference i qrefTo the current loop calculation module;

[0017] Step 6. The current loop calculation module outputs 6 drive signals G to the full-controlled full bridge;

[0018] Step 7.e αβ Calculation module input A phase current i a 、B phase current i b 、A phase voltage u a and B phase voltage u b , e αβ The calculation module outputs the αβ axis current i αβ To the current loop calculation module, the αβ axis current i αβ Including α-axis current i α and the β-axis current i β ;e αβ The calculation module outputs the normalized αβ axis to estimate the back EMF voltage e αβ To the angular frequency estimation and position estimation modules, normalize the αβ axis to estimate the back EMF voltage e αβ Including the normalized α axis to estimate the back EMF voltage e α and normalized β axis to estimate the back EMF voltage e β .

[0019] Furthermore, the I / F generator module in step 1 is used to generate an output current reference I ref , weighting coefficient K and reference angular frequency ω ref , including the following steps:

[0020] Step 11. Perform the starter motor rotor pre-positioning operation and set the current reference I ref Set to the maximum output current I max 0.1 times, so that the weighting coefficient K = 1, according to the speed n = 60r / min, according to the reference angular frequency ω ref =(2π / 60)·n·pSet the reference angular frequency ω ref , p is the number of pole pairs of the starting motor, and after running for 2s, it enters the I / F method acceleration starting stage;

[0021] Step 12. Perform the I / F method to accelerate the start operation. Within 100ms, the current reference I ref By 0.1I max Increase to I max , then maintain at I ref , so that the weighting coefficient K = 1, according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π)·ωref / pCalculate the reference angular frequency ω ref Instantaneous acceleration, k m is the margin coefficient, take k m =0.6~0.8,ω m is the mechanical angular frequency of the starter motor, T(I) is the function of the output torque T of the starter motor relative to the current I of the starter motor, T(I) is determined by the inherent characteristics of the starter motor, T L (n) is the aircraft engine drag moment T L As a function of the motor speed n, J is the moment of inertia of the aircraft generator, T L (n) and J are determined by the inherent characteristics of the aircraft engine, so that the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, when n = (60 / 2π) · ω ref / pThe calculated speed n reaches the aircraft engine ignition speed n f When it is 0.4 times of the starting value, it enters the transition start stage;

[0022] Step 13. Perform transition start operation to make the current reference I ref =I max , according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π0)·ω ref / pCalculate the reference angular frequency ω ref The instantaneous acceleration makes the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, according to n = (60 / 2π) · ω ref / p calculates the speed n, when the speed n is 0.4n f Increase to 0.6n f When the speed n reaches 0.6n, the weighting coefficient K decreases linearly from 1 to 0. f When , the back-EMF estimation method starts.

[0023] Step 14. Perform the back-EMF estimation method start operation, set the weighting coefficient K = 0, and accurately estimate the angular frequency ω output by the position estimation module, so that ω ref =ω, calculate the speed n according to n = (60 / 2π)·ω / p, at 0.6n f <n<n f When the current reference I ref =I max , so that the starting motor works at constant torque, when n is changed from n f When the starting speed n2 is reached, the current reference Iref by I max Decrease, so that the starting motor works at constant power. If n>n2, it enters the starting completion stage;

[0024] Step 15. Perform the start-up operation to make the current reference I ref =0, weighting coefficient K=0.

[0025] Further, when the angular frequency estimation module outputs the initial estimated angular frequency ω1 in step 2, the angular frequency estimation module uses a phase-locked loop to generate the initial estimated angular frequency ω1, including the following steps:

[0026] Step 21. Estimate the back EMF voltage e based on the normalized αβ axis αβ and the previous cycle angle θ3 of the phase-locked loop output, according to e rr1 =e α cos(θ3)+e β s in(θ3) calculates the error value e rr1 ;

[0027] Step 22. Add 0 to the error value e rr1 Subtract to obtain the regulator input e rrf ;

[0028] Step 23. rrf Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p1 It is 3 to 5 times the angular frequency corresponding to the starting disengagement speed n2, that is, k p1 =(3~5)×(2π / 60)·n2·p, where p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω f =k p1 e rrf ;

[0029] Step 24. Set the compensation angular frequency △ω f and the bias angular frequency ω mid The sum gives the angular frequency ω f , take the bias angular frequency ω mid It is 1 / 2 times of the angular frequency corresponding to the starting disengagement speed n2, that is, ω mid =1 / 2×(2π / 60)·n2·p;

[0030] Step 25. Set the angular frequency ω f Perform integration operation to obtain the angle θ2 of this cycle;

[0031] Step 26. Set the angular frequency ω f After the low-pass filter LPF processing, the initial estimated angular frequency ω1 is generated, and the cut-off frequency of the low-pass filter is set to 30 Hz.

[0032] Furthermore, the angular frequency weighted superposition module realizes the I / F generator module outputting the reference angular frequency ω according to the weighted coefficient K=1. ref The weighted superposition of the initial estimated angular frequency ω1 output by the angular frequency estimation module comprises the following steps:

[0033] Step 31. Press K N =1-K to calculate the complementary weighting coefficient K N ;

[0034] Step 32. Press ω ref1 =Kω ref +K N ω1 calculates the bias angular frequency ω ref1 .

[0035] Further, in step 4, the position estimation module outputs the accurate estimated angular frequency ω to the I / F generator module, and the position estimation module outputs the estimated position θ to the current loop calculation module, and a phase-locked loop is used to generate the accurate estimated angular frequency ω and the estimated position θ, including the following steps:

[0036] Step 41. Estimate the back EMF voltage e based on the normalized αβ axis αβ and the estimated position θ1 of the previous cycle output by the phase-locked loop, according to e rr2 =e α cos(θ1)+e β sin(θ1) calculates the error value e rr2 ;

[0037] Step 42. Add 0 to the error value e rr2 Subtract to obtain the regulator input e rra ;

[0038] Step 43. rra Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p =(20~50)p, p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω a =k p e rra ;

[0039] Step 44: Output the bias angular frequency ω according to the angular frequency weighted superposition module ref1 and complementary weighting coefficient K N , according to ω=K N △ω a +ω ref1 Calculate and accurately estimate the angular frequency ω;

[0040] Step 45: Integrate the precisely estimated angular frequency ω to obtain the estimated position θ of this period.

[0041] Furthermore, the d-axis current reference i of the reference current calculation module in step 5 dref and q-axis current reference i qref The calculation includes the following steps:

[0042] Step 51. According to the current reference I ref ,according to Calculate the d-axis current reference i dref , ψ f , L d and L q They represent the permanent magnet flux, d-axis inductance, and q-axis inductance of the starting motor respectively;

[0043] Step 52. According to the current reference I ref and the d-axis current reference i dref ,according to Calculate the q-axis current reference i qref .

[0044] Furthermore, the step 6 also includes a process in which the current loop calculation module is used to generate 6-way drive signals G of the full-controlled full bridge, including the following steps:

[0045] Step 61. According to the αβ axis current i αβ , perform Park transformation to obtain the d-axis current i d and q-axis current i q ;

[0046] Step 62: Based on the d-axis current reference i dref and the d-axis current i d , change i dref Subtract i d , and then process the difference by proportional integral regulator to obtain the d-axis voltage u d ;

[0047] Step 63. Based on the q-axis current reference i qref and q-axis current i q , change i qref Subtract i q , and then the difference is processed by proportional integral regulator to obtain the q-axis voltage u q ;

[0048] Step 64: According to the d-axis voltage u d and q-axis voltage u q , perform Park inverse transformation to obtain the α-axis voltage u α and β axis voltage u β ;

[0049] Step 65. According to the α-axis voltage u α and β axis voltage u β, perform space vector pulse width modulation SVPWM operation to obtain the driving signal G.

[0050] Furthermore, in step 7, αβ The calculation module calculates the normalized αβ axis estimated back EMF voltage e of the starting motor αβ , including the following steps:

[0051] Step 71. According to the A phase current i a and B phase current i b , perform Clark transformation to obtain the αβ axis current i αβ ;

[0052] Step 72: According to the A phase voltage u a and B phase voltage u b , perform Clark transformation to obtain the αβ axis voltage u αβ0 ;

[0053] Step 73. According to the αβ axis current i αβ 、αβ axis voltage u αβ0 , the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 , d-axis inductance L d , q-axis inductance L q and phase resistance R, according to Calculate the αβ axis to estimate the back EMF voltage e αβ0 ;

[0054] Step 74. Estimate the back EMF voltage e based on the αβ axis αβ0 ,according to Normalize to obtain the normalized αβ axis estimated back EMF voltage e αβ .

[0055] The beneficial effects of the present invention are:

[0056] 1. The proposed method for position-free control of the engine starting motor with continuous estimated angle during the starting process is suitable for the application field of position-free motor control of starting turbine aircraft engines using permanent magnet motors with reluctance torque.

[0057] 2. The proposed method of position-free control of the engine starting motor with continuous angle estimation during the starting process overcomes the deficiency that the position sensor cannot be installed in high temperature, high speed, high vibration and strong corrosion environments.

[0058] 3. Compared with the high-frequency injection method without position estimation, it overcomes the impact of adding a low-pass filter to filter out the injection of high frequency on the loop bandwidth.

[0059] 4. Compared with the I / F method, there is no position estimation, which overcomes the shortcomings of the I / F method without position estimation and the angle discontinuity at the switching point of the back-EMF method position estimation.

[0060] 5. Compared with the traditional back-EMF position estimation method which adopts a single-stage phase-locked loop structure, the proposed two-stage phase-locked loop structure has the first-stage phase-locked loop responsible for locking the frequency to ensure adaptability to a wide speed range, and the second-stage phase-locked loop responsible for locking the phase to ensure accurate phase locking, which can take into account both dynamic characteristics and observation accuracy requirements.

[0061] 6. The regulator in the phase-locked loop only uses simple proportional regulation, and does not use integral regulation, so there is no adverse effect caused by de-integral saturation.

[0062] 7. The maximum torque current ratio control is adopted to determine, which can effectively reduce the armature winding current and reduce copper loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A block diagram of a method for controlling the position of an engine starter motor without a continuous angle estimation during the starting process;

[0064] Figure 2 This is the working principle diagram of the I / F engine module;

[0065] Figure 3 It is the main waveform diagram of each stage of starting control;

[0066] Figure 4 This is the working principle diagram of the angular frequency estimation module;

[0067] Figure 5 This is the working principle diagram of the frequency weighted superposition module;

[0068] Figure 6 This is the working principle diagram of the position estimation module;

[0069] Figure 7 This is the working principle diagram of the reference current calculation module;

[0070] Figure 8 This is the working principle diagram of the current loop module;

[0071] Fig. 9 for e αβ Schematic diagram of the working principle of the computing module. DETAILED DESCRIPTION

[0072] The technical solution of the present invention is further described below, but the scope of protection claimed is not limited to the description.

[0073] This embodiment is a method for controlling the engine starter motor without position with continuous estimation angle during the dynamic process. The controlled motor is a permanent magnet starter generator with reluctance torque, 2 pairs of poles, d-axis inductance Ld and q-axis inductance Lq are 60μH and 90μH respectively, the rated power generation speed is 36000r / min, the no-load phase voltage is 60V at 36000r / min, and the power generation power is 40kW. When starting, the output torque is required to be greater than 20N·m below 5000r / min, and the output power is required to be greater than 10kW at 5000-9000r / min. The corresponding engine ignition speed ns is 5700r / min, and the starting disengagement speed n2=9000r / min.

[0074] Figure 1 The figure shows the block diagram of the method for controlling the engine starter motor without position in which the starting angle is continuously estimated. The functional modules implemented based on the embedded processor include I / F generator, angular frequency estimation, angular frequency weighted superposition, position estimation, reference current calculation, current loop calculation, e αβ Calculation; specifically includes the following steps:

[0075] Step 1. The I / F generator outputs a current reference I ref To the reference current calculation module, the I / F generator module outputs the weighting coefficient K and the reference angular frequency ω ref To the angular frequency weighted superposition module;

[0076] Step 2: The angular frequency estimation module outputs the initial estimated angular frequency ω1 to the angular frequency weighted superposition module, and the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 To the position estimation module and e αβ Computation module;

[0077] Step 3. The angular frequency weighted superposition module outputs the complementary weighting coefficient K N To the position estimation module;

[0078] Step 4. The position estimation module outputs the accurate estimated angular frequency ω to the I / F generator module, and the position estimation module outputs the estimated position θ to the current loop calculation module.

[0079] Step 5. The reference current calculation module outputs the d-axis current reference i dref and q-axis current reference i qref To the current loop calculation module;

[0080] Step 6. The current loop calculation module outputs 6 drive signals G to the full-controlled full bridge;

[0081] Step 7.e αβ Calculation module input A phase current i a 、B phase current i b 、A phase voltage ua and B phase voltage u b , e αβ The calculation module outputs the αβ axis current i αβ To the current loop calculation module, the αβ axis current i αβ Including α-axis current i α and the β-axis current i β ;e αβ The calculation module outputs the normalized αβ axis to estimate the back EMF voltage e αβ To the angular frequency estimation and position estimation modules, normalize the αβ axis to estimate the back EMF voltage e αβ Including the normalized α axis to estimate the back EMF voltage e α and normalized β axis to estimate the back EMF voltage e β .

[0082] Figure 2 The following is the working principle diagram of the I / F engine module. Figure 3 The following are the main waveforms of each stage of starting control. Figure 2 and Figure 3 The I / F generator module is used to generate an output current reference I ref , weighting coefficient K and reference angular frequency ω ref , including the following steps:

[0083] Step 11. Perform the starter motor rotor pre-positioning operation and set the current reference I ref Set to the maximum output current I max 0.1 times, so that the weighting coefficient K = 1, according to the speed n = 60r / min, according to the reference angular frequency ω ref =(2π / 60)·n·pSet the reference angular frequency ω ref , p is the number of pole pairs of the starting motor, and after running for 2s, it enters the I / F method acceleration starting stage;

[0084] Step 12. Perform the I / F method to accelerate the start operation. Within 100ms, the current reference I ref By 0.1I max Increase to I max , then maintain at I ref , so that the weighting coefficient K = 1, according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π)·ω ref / pCalculate the reference angular frequency ω ref Instantaneous acceleration, k m is the margin coefficient, take km =0.6~0.8,ω m is the mechanical angular frequency of the starter motor, T(I) is the function of the output torque T of the starter motor relative to the current I of the starter motor, T(I) is determined by the inherent characteristics of the starter motor, T L (n) is the aircraft engine drag moment T L As a function of the motor speed n, J is the moment of inertia of the aircraft generator, T L (n) and J are determined by the inherent characteristics of the aircraft engine, so that the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, when n = (60 / 2π) · ω ref / pThe calculated speed n reaches the aircraft engine ignition speed n f When it is 0.4 times of the starting value, it enters the transition start stage;

[0085] Step 13. Perform transition start operation to make the current reference I ref =I max , according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π0)·ω ref / pCalculate the reference angular frequency ω ref The instantaneous acceleration makes the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, according to n = (60 / 2π) · ω ref / p calculates the speed n, when the speed n is 0.4n f Increase to 0.6n f When the speed n reaches 0.6n, the weighting coefficient K decreases linearly from 1 to 0. f When , the back-EMF estimation method starts.

[0086] Step 14. Perform the back-EMF estimation method start operation, set the weighting coefficient K = 0, and accurately estimate the angular frequency ω output by the position estimation module, so that ω ref =ω, calculate the speed n according to n = (60 / 2π)·ω / p, at 0.6n f <n<n f When the current reference I ref =I max , so that the starting motor works at constant torque, when n is changed from n f When the starting speed n2 is reached, the current reference I ref by I maxDecrease, so that the starting motor works at constant power. If n>n2, it enters the starting completion stage. If the starting fails, n<0.6n f Then return to the transition start stage;

[0087] Step 15. Perform the start-up operation to make the current reference I ref =0, weighting coefficient K=0.

[0088] In step 11, since the starting resistance torque of the turbine engine is small at low speed, there is only static motor resistance torque, and the current reference I ref Set to 0.1I max This is to prevent excessive starting current from causing rotor vibration and affecting the pre-positioning effect of the starter motor rotor. The speed of the starter motor rotor pre-positioning effect stage should not be set too high, otherwise it is easy to cause rotor pre-positioning failure; the speed of the starter motor rotor pre-positioning should not be too low, otherwise the time to ensure that the pre-positioning completes two revolutions will exceed 2s, which will significantly affect the completion speed of the starter motor rotor pre-positioning stage.

[0089] In step 12, set the current reference I ref By 0.1I max Increase to I max , is to prevent the current reference from changing too fast and causing starting failure. For the margin coefficient k m From the perspective of ensuring the success of starting, the instantaneous acceleration cannot exceed dω m / dt=[T(I)-T L (n)] / J, from the perspective of ensuring the starting speed, the greater the instantaneous acceleration, the better; therefore, considering the compatibility of preventing the engine torque fluctuation from causing the starting failure and increasing the starting speed as much as possible, k is taken. m =0.6-0.8.

[0090] In step 13, the speed n is increased from 0.4n f Increase to 0.6n f The linear reduction of the weighting coefficient K from 1 to 0 is to make the angular frequency weighted superposition module output the bias angular frequency ω ref1 By n<0.4n f The given value of the I / F method is smoothly switched to n>0.4n f The following depends on the initial estimated angular frequency ω1 output by the angular frequency estimation module.

[0091] In steps 11 to 13, the starter motor speed n is generated by the I / F method open loop setting. In the transition start stage of step 13, the angular frequency estimation module can normally calculate the starter motor position according to the back electromotive force, laying the foundation for entering the back electromotive force estimation method start stage in step 14. In step 14, the starter motor speed n is determined by the angular frequency estimation module.

[0092] Figure 4 The figure shows the working principle of the angular frequency estimation module. The angular frequency estimation module uses a phase-locked loop to generate an initial estimated angular frequency ω1, including the following steps:

[0093] Step 21. Estimate the back EMF voltage e based on the normalized αβ axis αβ and the previous cycle angle θ3 of the phase-locked loop output, according to e rr1 =e α cos(θ3)+e β s in(θ3) calculates the error value e rr1 ;

[0094] Step 22. Add 0 to the error value e rr1 Subtract to obtain the regulator input e rrf ;

[0095] Step 23. rrf Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p1 It is 3 to 5 times the angular frequency corresponding to the starting disengagement speed n2, that is, k p1 =(3~5)×(2π / 60)·n2·p, where p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω f =k p1 e rrf ;

[0096] Step 24. Set the compensation angular frequency △ω f and the bias angular frequency ω mid The sum gives the angular frequency ω f , take the bias angular frequency ω mid It is 1 / 2 times of the angular frequency corresponding to the starting disengagement speed n2, that is, ω mid =1 / 2×(2π / 60)·n2·p;

[0097] Step 25. Set the angular frequency ω f Perform integration operation to obtain the angle θ2 of this cycle;

[0098] Step 26. Set the angular frequency ω f After the low-pass filter LPF processing, the initial estimated angular frequency ω1 is generated, and the cut-off frequency of the low-pass filter is set to 30 Hz.

[0099] In step 23, the regulator uses proportional regulation with steady-state error instead of proportional integral regulation without steady-state error because when the phase-locked loop is not working in a good locked state, the integral is saturated and desaturation is difficult. rr1 The range is -1 to +1. In order to ensure that the phase-locked loop can cover the entire starting speed range, the proportional adjustment coefficient kp1 At least it needs to be greater than the angular frequency corresponding to the starting disengagement speed n2, so the proportional adjustment coefficient k is taken p1 The starting speed n2 corresponds to 3 to 5 times the angular frequency. p1 The adjustment coefficient k is 3 to 5 times higher than the corresponding angular frequency of the starting disengagement speed n2. p1 The angular frequency ω of the phase-locked loop output f Vibrate heavily.

[0100] In step 24, take the offset angular frequency ω mid The 1 / 2 times of the angular frequency corresponding to the starting disengagement speed n2 is to minimize the adjustment range of the phase-locked loop, and the adjustment range is only -1 / 2×(2π / 60)·n2·p to 1 / 2×(2π / 60)·n2·p. At the same time, when the starting motor speed is low, the back EMF voltage is also small, and the offset angular frequency ω mid A positive given value is added to the phase-locked loop to increase the probability of the phase-locked loop output angle θ2 increasing in a positive direction and entering a steady state, thereby effectively suppressing the phase-locked loop from operating in an undesirable state where the angle θ2 decreases in a negative direction.

[0101] In step 26, due to the proportional adjustment coefficient k of the phase-locked loop p1 Large, resulting in an angular frequency ω f The jump is large, so a low-pass filter is added to improve the stability of the initial estimated angular frequency ω1.

[0102] Figure 5 The figure shows the working principle of the frequency weighted superposition module. The angular frequency weighted superposition module realizes the I / F generator module outputs the reference angular frequency ω according to the weighting coefficient K=1. ref The weighted superposition of the initial estimated angular frequency ω1 output by the angular frequency estimation module comprises the following steps:

[0103] Step 31. Press K N =1-K to calculate the complementary weighting coefficient K N ;

[0104] Step 32. Press ω ref1 =Kω ref +K N ω1 calculates the bias angular frequency ω ref1 .

[0105] Figure 6 The working principle diagram of the position estimation module is shown. The position estimation module uses a phase-locked loop to generate accurate estimated angular frequency ω and estimated position θ, including the following steps:

[0106] Step 41. Estimate the back EMF voltage e based on the normalized αβ axis αβand the estimated position θ1 of the previous cycle output by the phase-locked loop, according to e rr2 =e α cos(θ1)+e β sin(θ1) calculates the error value e rr2 ;

[0107] Step 41. Add 0 to the error value e rr2 Subtract to obtain the regulator input e rra ;

[0108] Step 43. rra Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p =(20~50)p, p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω a =k p e rra ;

[0109] Step 44: Output the bias angular frequency ω according to the angular frequency weighted superposition module ref1 and complementary weighting coefficient K N , according to ω=K N △ω a +ω ref1 Calculate and accurately estimate the angular frequency ω;

[0110] Step 45: Integrate the precisely estimated angular frequency ω to obtain the estimated position θ of this period.

[0111] With respect to the angular frequency estimation phase-locked loop, the adjustment coefficient k used by the position estimation phase-locked loop in step 43 is p The phase-locked loop angular frequency adjustment range is small, so the output accurate estimated angular frequency ω is relatively stable, which can ensure that the estimated position θ output by the integrator is also relatively stable.

[0112] In step 44, due to the addition of the offset angular frequency ω ref1 , effectively reducing the adjustment range of the phase-locked loop, which can ensure high accuracy of the estimated position θ. N , when the speed n increases from low to high, the accurate estimation of the angular frequency ω can be smoothly transitioned from the given value determined by the I / F method to the estimation determined by the back-electric method, which can effectively suppress the influence of the low speed accuracy of the position estimation phase-locked loop on the accurate estimation of the angular frequency ω.

[0113] In step 45, the bias angular frequency ω determined by the weighted I / F method and the back EMF method is ref1The estimated position θ is generated after the integrator, and the continuous smoothing effect of the integrator is used to ensure the continuity of the estimated position θ when switching from the I / F method to the back-EMF method. Compared with the I / F method without position estimation, it overcomes the shortcomings of the I / F method without position estimation and the angle discontinuity at the switching point of the back-EMF method position estimation.

[0114] Figure 4 Angular frequency estimation module and Figure 6 The phase-locked loop regulator in the position estimation module only uses simple proportional regulation without using integral regulation, so there is no adverse effect caused by integral saturation.

[0115] Figure 7 The reference current calculation module is used to calculate the d-axis current reference i dref and q-axis current reference i qref The calculation includes the following steps:

[0116] Step 51. According to the current reference I ref ,according to Calculate the d-axis current reference i dref , ψ f , L d and L q They represent the permanent magnet flux, d-axis inductance, and q-axis inductance of the starting motor respectively;

[0117] Step 52. According to the current reference I ref and the d-axis current reference i dref ,according to Calculate the q-axis current reference i qref .

[0118] d-axis current reference i dref and q-axis current reference i qref Determined by the maximum torque current ratio control, it can effectively reduce the armature winding current and copper loss. At the same time, it can also better match the permanent magnet motor with reluctance torque.

[0119] Figure 8 The current loop calculation module is used to generate 6-way drive signals G of the full-controlled full bridge, including the following steps:

[0120] Step 61. According to the αβ axis current i αβ , perform Park transformation to obtain the d-axis current i d and q-axis current i q ;

[0121] Step 61. Based on the d-axis current reference i dref and the d-axis current i d , change i dref Subtract i d, and then process the difference by proportional integral regulator to obtain the d-axis voltage u d ;

[0122] Step 61. Based on the q-axis current reference i qref and q-axis current i q , change i qref Subtract i q , and then the difference is processed by proportional integral regulator to obtain the q-axis voltage u q ;

[0123] Step 61. According to the d-axis voltage u d and q-axis voltage u q , perform Park inverse transformation to obtain the α-axis voltage u α and β axis voltage u β ;

[0124] Step 61. According to the α-axis voltage u α and β axis voltage u β , perform space vector pulse width modulation SVPWM operation to obtain the driving signal G.

[0125] Fig. 9 Shown is e αβ The working principle diagram of the computing module. αβ The calculation module is used to calculate the normalized αβ axis estimated back EMF voltage e of the starting motor αβ , including the following steps:

[0126] Step 71. According to the A phase current i a and B phase current i b , perform Clark transformation to obtain the αβ axis current i αβ ;

[0127] Step 72: According to the A phase voltage u a and B phase voltage u b , perform Clark transformation to obtain the αβ axis voltage u αβ0 ;

[0128] Step 73. According to the αβ axis current i αβ 、αβ axis voltage u αβ0 , the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 , d-axis inductance L d , q-axis inductance L q and phase resistance R, according to Calculate the αβ axis to estimate the back EMF voltage e αβ0 ;

[0129] Step 74. Estimate the back EMF voltage e based on the αβ axis αβ0 ,according to Normalize to obtain the normalized αβ axis estimated back EMF voltage e αβ .

[0130] In step 73, the back EMF calculation takes into account the influence of the reluctance torque, and can better match the permanent magnet motor with reluctance torque.

[0131] In step 74, the back-EMF voltage is normalized, so that the error value of the phase-locked loop in the angular frequency estimation and position estimation can be controlled within the range of -1 to 1, effectively reducing the difficulty of parameter setting in the phase-locked loop.

[0132] The proposed back-EMF position estimation uses a two-stage phase-locked loop structure. The first stage is the first-stage phase-locked loop in the angular frequency estimation module, and the second stage is the second-stage phase-locked loop in the position estimation. The first-stage phase-locked loop has a high bandwidth, which enables it to be used in a wide speed range; since the second-stage phase-locked loop uses the angular frequency ω output by the first-stage phase-locked loop ref1 Therefore, the bandwidth of the second-stage phase-locked loop can be much smaller than that of the first-stage phase-locked loop, so that the angular frequency and position obtained by the second-stage phase-locked loop are more stable, ensuring the accuracy of position estimation. Compared with the traditional back-EMF method position estimation using a single-stage phase-locked loop structure, the proposed two-stage phase-locked loop structure, the first-stage phase-locked loop is responsible for locking the frequency to ensure adaptability to a wide speed range, and the second-stage phase-locked loop is responsible for locking the phase to ensure accurate phase locking, which can take into account both dynamic characteristics and observation accuracy requirements.

[0133] The proposed method for controlling the engine starting motor without position during the starting process with a continuous angle estimation overcomes the disadvantage that the position sensor cannot be installed in high temperature, high speed, high vibration and strong corrosion environments. Compared with the high-frequency injection method without position estimation, it overcomes the effect of adding a low-pass filter to filter out the high frequency injection on the loop bandwidth. The proposed method for controlling the engine starting motor without position during the starting process with a continuous angle estimation is suitable for the application field of position-free motor control that uses a permanent magnet motor with reluctance torque to start a turbine aircraft engine.

Claims

1. A method for controlling the engine starter motor without position during the starting process with a continuous angle estimation, using functional modules based on an embedded processor, including an I / F generator, an angular frequency estimation module, an angular frequency weighted superposition module, a position estimation module, a reference current calculation module, a current loop calculation module, and an αβ The calculation module specifically comprises the following steps: Step 1. The I / F generator outputs a current reference I ref To the reference current calculation module, the I / F generator module outputs the weighting coefficient K and the reference angular frequency ω ref To the angular frequency weighted superposition module; Step 2: The angular frequency estimation module outputs the initial estimated angular frequency ω1 to the angular frequency weighted superposition module, and the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 To the position estimation module and e αβ Computation module; Step 3. The angular frequency weighted superposition module outputs the complementary weighting coefficient K N To the position estimation module; Step 4. The position estimation module outputs the accurate estimated angular frequency ω to the I / F generator module, and the position estimation module outputs the estimated position θ to the current loop calculation module. Step 5. The reference current calculation module outputs the d-axis current reference i dref and q-axis current reference i qref To the current loop calculation module; Step 6. The current loop calculation module outputs 6 drive signals G to the full-controlled full bridge; Step 7.e αβ Calculation module input A phase current i a 、B phase current i b 、A phase voltage u a and B phase voltage u b , e αβ The calculation module outputs the αβ axis current i αβ To the current loop calculation module, the αβ axis current i αβ Including α-axis current i α and the β-axis current i β ;e αβ The calculation module outputs the normalized αβ axis to estimate the back EMF voltage e αβ To the angular frequency estimation and position estimation modules, normalize the αβ axis to estimate the back EMF voltage e αβ Including the normalized α axis to estimate the back EMF voltage e α and normalized β axis to estimate the back EMF voltage e β .

2. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process according to claim 1, characterized in that: The I / F generator module in step 1 is used to generate an output current reference I ref , weighting coefficient K and reference angular frequency ω ref , including the following steps: Step 11. Perform the starter motor rotor pre-positioning operation and set the current reference I ref Set to the maximum output current I max 0.1 times, so that the weighting coefficient K = 1, according to the speed n = 60r / min, according to the reference angular frequency ω ref =(2π / 60)·n·pSet the reference angular frequency ω ref , p is the number of pole pairs of the starting motor, and after running for 2s, it enters the I / F method acceleration starting stage; Step 12. Perform the I / F method to accelerate the start operation. Within 100ms, the current reference I ref By 0.1I max Increase to I max , then maintain at I ref , so that the weighting coefficient K = 1, according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π)·ω ref / pCalculate the reference angular frequency ω ref Instantaneous acceleration, k m is the margin coefficient, take k m =0.6~0.8,ω m is the mechanical angular frequency of the starter motor, T(I) is the function of the output torque T of the starter motor relative to the current I of the starter motor, T(I) is determined by the inherent characteristics of the starter motor, T L (n) is the aircraft engine drag moment T L As a function of the motor speed n, J is the moment of inertia of the aircraft generator, T L (n) and J are determined by the inherent characteristics of the aircraft engine, so that the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, when n = (60 / 2π) · ω ref / pThe calculated speed n reaches the aircraft engine ignition speed n f When it is 0.4 times of the starting value, it enters the transition start stage; Step 13. Perform transition start operation to make the current reference I ref =I max , according to dω m / dt=k m [T(I)-T L (n)] / J,ω m =ω ref / p、I=I ref and n = (60 / 2π0)·ω ref / pCalculate the reference angular frequency ω ref The instantaneous acceleration makes the reference angular frequency ω ref According to the calculated instantaneous acceleration increase, according to n = (60 / 2π) · ω ref / p calculates the speed n, when the speed n is 0.4n f Increase to 0.6n f When the speed n reaches 0.6n, the weighting coefficient K decreases linearly from 1 to 0. f When , the back-EMF estimation method starts. Step 14. Perform the back-EMF estimation method start operation, set the weighting coefficient K = 0, and accurately estimate the angular frequency ω output by the position estimation module, so that ω ref =ω, calculate the speed n according to n = (60 / 2π)·ω / p, at 0.6n f <n<n f When the current reference I ref =I max , so that the starting motor works at constant torque, when n is changed from n f When the starting speed n2 is reached, the current reference I ref by I max Decrease, so that the starting motor works at constant power. If n>n2, it enters the starting completion stage; Step 15. Perform the start-up operation to make the current reference I ref =0, weighting coefficient K=0.

3. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process as claimed in claim 1, characterized in that: When the angular frequency estimation module outputs the initial estimated angular frequency ω1 in step 2, the angular frequency estimation module uses a phase-locked loop to generate the initial estimated angular frequency ω1, including the following steps: Step 21. Estimate the back EMF voltage e based on the normalized αβ axis αβ and the previous cycle angle θ3 of the phase-locked loop output, according to e rr1 =e α cos(θ3)+e β sin(θ3) calculates the error value e rr1 ; Step 22. Add 0 to the error value e rr1 Subtract to obtain the regulator input e rrf ; Step 23. rrf Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p1 It is 3 to 5 times the angular frequency corresponding to the starting disengagement speed n2, that is, k p1 =(3~5)×(2π / 60)·n2·p, where p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω f =k p1 e rrf ; Step 24. Set the compensation angular frequency △ω f and the bias angular frequency ω mid The sum gives the angular frequency ω f , take the bias angular frequency ω mid It is 1 / 2 times of the angular frequency corresponding to the starting disengagement speed n2, that is, ω mid =1 / 2×(2π / 60)·n2·p; Step 25. Set the angular frequency ω f Perform integration operation to obtain the angle θ2 of this cycle; Step 26. Set the angular frequency ω f After the low-pass filter LPF processing, the initial estimated angular frequency ω1 is generated, and the cut-off frequency of the low-pass filter is set to 30 Hz.

4. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process as claimed in claim 1, characterized in that: The angular frequency weighted superposition module realizes the output of the reference angular frequency ω by the I / F generator module according to the weighting coefficient K=1. ref The weighted superposition of the initial estimated angular frequency ω1 output by the angular frequency estimation module comprises the following steps: Step 31. Press K N =1-K to calculate the complementary weighting coefficient K N ; Step 32. Press ω ref1 =Kω ref +K N ω1 calculates the bias angular frequency ω ref1 .

5. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process according to claim 1, characterized in that: In step 4, the position estimation module outputs the accurate estimated angular frequency ω to the I / F generator module, and the position estimation module outputs the estimated position θ to the current loop calculation module, and a phase-locked loop is used to generate the accurate estimated angular frequency ω and the estimated position θ, including the following steps: Step 41. Estimate the back EMF voltage e based on the normalized αβ axis αβ and the estimated position θ1 of the previous cycle output by the phase-locked loop, according to e rr2 =e α cos(θ1)+e β sin(θ1) calculates the error value e rr2 ; Step 42. Add 0 to the error value e rr2 Subtract to obtain the regulator input e rra ; Step 43. rra Perform regulator processing, the regulator adopts proportional regulation, and takes the proportional regulation coefficient k p =(20~50)p, p is the number of pole pairs of the starting motor, so that the compensation angular frequency △ω a =k p e rra ; Step 44: Output the bias angular frequency ω according to the angular frequency weighted superposition module ref1 and complementary weighting coefficient K N , according to ω=K N △ω a +ω ref1 Calculate and accurately estimate the angular frequency ω; Step 45: Integrate the precisely estimated angular frequency ω to obtain the estimated position θ of this period.

6. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process as claimed in claim 1, characterized in that: The d-axis current reference i of the reference current calculation module in step 5 dref and q-axis current reference i qref The calculation includes the following steps: Step 51. According to the current reference I ref ,according to Calculate the d-axis current reference i dref , ψ f , L d and L q They represent the permanent magnet flux, d-axis inductance, and q-axis inductance of the starting motor respectively; Step 52. According to the current reference I ref and the d-axis current reference i dref ,according to Calculate the q-axis current reference i qref .

7. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process according to claim 1, characterized in that: The step 6 also includes a process in which the current loop calculation module is used to generate 6-way drive signals G of the full-controlled full bridge, including the following steps: Step 61. According to the αβ axis current i αβ , perform Park transformation to obtain the d-axis current i d and q-axis current i q ; Step 62: Based on the d-axis current reference i dref and the d-axis current i d , change i dref Subtract i d , and then process the difference by proportional integral regulator to obtain the d-axis voltage u d ; Step 63. Based on the q-axis current reference i qref and q-axis current i q , change i qref Subtract i q , and then the difference is processed by proportional integral regulator to obtain the q-axis voltage u q ; Step 64: According to the d-axis voltage u d and q-axis voltage u q , perform Park inverse transformation to obtain the α-axis voltage u α and β axis voltage u β ; Step 65. According to the α-axis voltage u α and β-axis voltage u β , perform space vector pulse width modulation SVPWM operation to obtain the driving signal G.

8. The method for controlling the engine starter motor without position with continuous estimated angle during the starting process according to claim 1, characterized in that: In step 7, αβ The calculation module calculates the normalized αβ axis estimated back EMF voltage e of the starting motor αβ , including the following steps: Step 71. According to the A phase current i a and B phase current i b , perform Clark transformation to obtain the αβ axis current i αβ ; Step 72: According to the A phase voltage u a and B phase voltage u b , perform Clark transformation to obtain the αβ axis voltage u αβ0 ; Step 73. According to the αβ axis current i αβ 、αβ axis voltage u αβ0 , the angular frequency weighted superposition module outputs the bias angular frequency ω ref1 , d-axis inductance L d , q-axis inductance L q and phase resistance R, according to Calculate the αβ axis to estimate the back EMF voltage e αβ0 ; Step 74. Estimate the back EMF voltage e based on the αβ axis αβ0 ,according to Normalize to obtain the normalized αβ axis estimated back EMF voltage e αβ .