JP4211133B2 - Sensorless control system for permanent magnet synchronous motor - Google Patents

Description

【００１５】
ここで、ｖα、ｖβは電機子電圧のα、β軸成分、ｉα、ｉβは電機子電流のα、β軸成分、Ｒは電機子抵抗、Ｌ_d、Ｌ_qはｄ，ｑ軸インダクタンス成分、ｐは微分演算子を表す。
【００１６】
上式に示すように、突極形モータでは、停止時においてもｄ，ｑ軸インダクタンスの違いにより、位置角の情報が２θ_reの三角関数の形で存在する。しかし、界磁の極性を含む絶対的な位置（ｄ軸方向）は、上式に示すように、一般的な数学モデルからでは推定することができない。また、円筒形モータは、一般的にＬ_d＝Ｌ_q＝Ｌと見なすことができるため、上式は次式のようになり、この数学モデルからでは停止時における位置角情報が得られない。
【００１７】
【数２】

【００１８】
本発明は、停止時の磁極位置を推定するため、ＰＭモータにおける電機子鉄心の磁化特性の非直線性を利用し、この非直線性で磁極位置による電流応答の違いから磁極位置を推定するものである。
【００１９】
図１１の（ａ）〜（ｃ）は、ＰＭモータのｕ，ｖ，ｗの各相にそれぞれ正負の一定電圧（定格の７５％）を時間２００μｓだけ印加したときの各相の電流値を、回転子の位置角１０°毎に測定した結果を示す。同図の（ａ）に示すＩ_u ⁺は、正の電圧をｕ相に、負の電圧をｖ，ｗ相に印加したときのｕ相電流値を示す。また、Ｉ_u ^-は、負の電圧をｕ相に、正の電圧をｖ，ｗ相に印加したときのｕ相電流値を示す。同様に、（ｂ）及び（ｃ）に示すＩ_v ⁺，Ｉ_v ^-及びＩ_w ⁺，Ｉ_w ^-は、各相に電圧を印加したときのｖ，ｗ相の電流値を示す。なお、負の電流値Ｉ_u ^-、Ｉ_v ^-、Ｉ_w ^-は、大きさを比較しやすくするため絶対値で表している。
【００２０】
上記の図１１の（ａ）に示される測定結果から、ｕ相巻線付近にある界磁の極性がＮであるθ_re＝０付近では、増磁方向に電圧を印加した場合、すなわち正の電圧を印加した場合は、磁束が飽和し、インダクタンスが減少するため、電流Ｉ_u ⁺の時間変化は大きくなる。逆に、負の電圧を印加した場合は、インダクタンスは増加あるいは変化しないため、電流Ｉ_u ^-の時間変化は正の電圧を印加した場合に比較して小さくなる。また、ｕ相巻線付近にある界磁の極性がＳであるθ_re＝１８０°付近では、電流Ｉ_u ⁺に比較して電流Ｉ_u ^-の時間変化が大きくなる。
【００２１】
図１１の（ｄ）には、各相の電流値の差ΔＩ_u（＝Ｉ_u ⁺＋Ｉ_u ^-）、ΔＩ_v（＝Ｉ_v ⁺＋Ｉ_v ^-）、ΔＩ_w（＝Ｉ_w ⁺＋Ｉ_w ^-）を示す。同図の結果から、ΔＩ_uはθ_re＝０°と１８０°付近で正負に０.１アンペア程度の差があり、同様にΔＩ_vはθ_re＝１２０°と３００°付近で差があり、ΔＩ_wはθ_re＝２４０°と６０°付近で差がある。すなわち、各相共に、電流値の差ΔＩ_u、ΔＩ_v、ΔＩ_wには極軸付近において増磁方向と減磁方向の電流値に差がある。
【００２２】
これら各相の電流値の差は、印加される電圧に対して電機子鉄心の磁化特性の非直線性によって磁束が飽和してインダクタンスが減少するために現れる。したがって、極軸付近での増磁方向と減磁方向とでの電流値の差が大きくなることを利用し、この電流値の差が最も大きくなる電圧ベクトル位置を検出することで停止時の磁極位置を推定することができる。このことから、本発明は、以下の構成を特徴とする。
【００２３】
（第１の発明）
永久磁石式同期電動機の磁極位置に応じて該電動機の電機子巻線に供給する電圧角度を制御する永久磁石式同期電動機のセンサレス制御システムにおいて、
前記電動機の停止状態で、一定間隔で角度を変えた正負の一定電圧の電圧ベクトルを該電動機に印加し、各正負の電圧ベクトルの印加に対する電機子巻線の電流を検出し、この検出された電流値をそのときに印加された電圧ベクトルの角度でｄ−ｑ座標変換して回転子に固定した座標のｄ軸電流値として検出し、正負の電圧ベクトルの印加に対する前記ｄ軸電流値の差が最も大きくなる前記電圧ベクトルの位置を磁極位置として推定する磁極位置推定装置を備えたことを特徴とする。
【００２４】
（第２の発明）
上記の第１の発明において、推定される磁極位置の精度は、電圧ベクトルの角度間隔で決まり、この角度間隔を小さくするほど精度が高くなる反面、電圧ベクトルの出力回数が多くなって推定処理に時間がかかる。そこで、本発明は磁極位置の推定を大まかな推定と細かな推定の複数段階に分けることで推定処理回数及び時間を短縮するもので、以下の構成を特徴とする。
【００２５】
前記磁極位置推定装置は、
前記電動機の停止状態で、一定間隔で大まかに角度を変えた正負の一定電圧の電圧ベクトルを該電動機に印加したときに検出される前記ｄ軸電流値の差が最も大きくなる該電圧ベクトルの位置を凡その磁極位置として推定する第１の推定手段と、
前記第１の推定手段で検出した電圧ベクトル位置を中心にして、一定間隔で細かに角度を変えた電圧ベクトルを該電動機に印加したときの前記ｄ軸電流値の差が最も大きくなる該電圧ベクトルの位置を磁極位置と推定する処理を繰り返して最終的な磁極位置を推定する第２の推定手段とを備えたことを特徴とする。
【００２６】
（第３の発明）
前記の第１の推定手段において、実際の磁極位置θ_reが９０°または−９０°付近の場合、それぞれのｄ軸電流値Ｉ_d1，Ｉ_d7にほとんど違いが現れないことがある。この例を下記表１にｄ軸電流値Ｉ_d1，Ｉ_d7及びその絶対値の差｜Ｉ_d1−Ｉ_d7｜を示す。
【００２７】
【表１】

【００２８】
理論的には、−９０°＜θ_re＜９０°の場合はＩ_d1＞Ｉ_d7、θ_re＝９０°または−９０°の場合はＩ_d1＝Ｉ_d7、その他の場合はＩ_d1＜Ｉ_d7となるはずであるが、上記の表１に示すように「ばらつき」があり、その差も非常に小さくなっている。
【００２９】
これは、磁極位置が９０°と−９０°付近においては、磁気飽和の影響が小さく、インダクタンスの変化が少ないため、ｄ軸電流値にほとんど違いが現れず、その結果として電流検出器の精度やノイズ等の影響によって誤差が生じていると思われる。
【００３０】
そこで、本発明は、上記のｄ軸電流値Ｉ_d1，Ｉ_d7の絶対値の差｜Ｉ_d1−Ｉ_d7｜が誤差範囲（例えば、上記の表では０.０３アンペア）以下の場合には第１の推定手段における凡その磁極位置の推定に代えて、以下の推定を行う。
【００３１】
まず、図１２の９０°と−９０°の電圧ベクトルｖ₄とｖ₁₀をモータに印加し、この電圧印加に対して、前記の推定１と同様に検出されるｄ軸電流Ｉ_dの大きい方の電圧ベクトル位置を推定磁極位置θ_M1、その電流をＩ_dmaxとする。
【００３２】
これにより、実際の磁極位置θ_reを図１２の電圧ベクトルに対してθ_M1±１５°の精度で推定することができるが、表１に示すように、磁極位置θ_reが９０°±１５°と−９０°±１５°付近においても絶対値の差｜Ｉ_d1−Ｉ_d7｜が小さな値となる。
【００３３】
そこで、さらに、図１３に示すように、推定磁極位置θ_M1を基準にして、±１５°の位置の電圧ベクトルｖ₁₃とｖ₁₄をモータに印加し、この印加に対するｄ軸電流の絶対値の差を比較することで推定磁極位置θ_M1を更新する。この更新した推定磁極位置θ_M1について第２の推定手段による磁極位置の絞り込みを行う。以上のことから、本発明は、以下の構成を特徴とする。
【００３４】
前記磁極位置推定装置は、
前記第１の推定手段で推定された磁極位置が９０°付近または−９０°付近の場合、推定された磁極位置を基準にしてその前後に一定の角度差をもつ電圧ベクトルを前記電動機に印加したときの前記ｄ軸電流値の差が最も大きくなる該電圧ベクトルの位置を磁極位置として更新する手段を備えたことを特徴とする。
【００３５】
【発明の実施の形態】
図１は、本発明の実施形態を示すシステム構成図である。ＰＭモータ（永久磁石式同期電動機）１は、ＰＷＭインバータ２によって電機子巻線に電流が供給され、同期運転される。ＰＷＭインバータ２は、ドライブ回路３から主回路スイッチのオン・オフ制御信号が印加される。
【００３６】
電流検出器５Ｕ，５Ｖは、ＰＭモータ１に供給されるＵ，Ｖ相の電流を検出する。
【００３７】
コントローラ４は、推定した磁極位置に応じてドライブ回路３へのドライブ信号の角度を制御し、電流検出器５Ｕ，５Ｖからの検出電流からＰＭモータ１の巻線電流を制御することでトルク制御を行う。
【００３８】
ＰＭモータ１に結合されるロータリーエンコーダ６は、回転子位置変動の観測用に設けられたもので、実際のセンサレス制御システムでは設けられない。
【００３９】
ここで、本実施形態では、停止状態のＰＭモータ１の磁極位置を推定する手段として、コントローラ４には、電圧ベクトル発生手段１１と、電流検出手段１２と、第１及び第２の推定手段１３、１４及びこれら手段を制御する制御手段１５を設ける。
【００４０】
電圧ベクトル発生手段１１は、制御手段１５の制御の基に、図１２に示すような３０°毎に角度を変えた正負の電圧ベクトルを順次発生し、さらには推定磁極位置θ_Mを基準にしてその前後で角度を変化させた電圧ベクトルを順次発生する。これら電圧ベクトルは、ＰＷＭインバータ２によって停止状態のＰＭモータ１に角度を変えた電圧として印加される。
【００４１】
電流検出手段１２は、ＰＭモータ１に正負の電圧が印加されることで電機子巻線に流れる前記の電流Ｉ_n ⁺，Ｉ_n ^-を検出する。
【００４２】
第１の推定手段１３と第２の推定手段１４は、その協同で、モータ１の停止状態で、一定間隔で角度を変えた正負の一定電圧の電圧ベクトルをモータ１に印加し、各正負の電圧ベクトルの印加に対する回転子に固定した座標のｄ軸電流値を検出し、該電流値の差が最も大きくなる電圧ベクトルの位置を磁極位置として推定する。
【００４３】
このうち、第１の推定手段１３は、電動機の停止状態で、一定間隔で大まかに角度を変えた正負の一定電圧の電圧ベクトルを該電動機に印加し、各電圧ベクトルの印加に対する回転子に固定した座標のｄ軸電流値を検出し、該電流値の差が最も大きくなる前記電圧ベクトルの位置を凡その磁極位置として推定する。この推定処理を以下に詳細に説明する。
【００４４】
最初に、電流検出手段１２で検出する電流Ｉ_n ⁺，Ｉ_n ^-を電圧ベクトルの角度でｄ−ｑ座標に変換し、２つのｄ軸電流値Ｉ_dを検出し、そのうちの大きい方の電圧ベクトル位置を推定磁極位置θ_M1、その時の電流値Ｉ_dを最大ｄ軸電流値Ｉ_dmaxとする。この推定は、増磁側か減磁側かを判定し、磁極位置が正負の電圧ベクトルの何れの側にあるかを推定する。
【００４５】
この後、推定磁極位置θ_M1になる電圧ベクトル位置の前後何れの方向に実際の磁極位置θ_reがあるかを推定磁極位置θ_M1と最大ｄ軸電流値Ｉ_dmaxを用いて推定する。この推定処理手順を図２に示す。
【００４６】
図２において、Ｉ_dnは、図１２における電圧ベクトルｎを出力したときのｄ軸電流値である。また、ｉ，ｊは推定終了に用いる判定係数である。図２の処理は、まず、ｉ＝１，ｊ＝１，ｎ＝０を初期設定し（Ｓ１）、印加しようとする電圧ベクトルｖ_nとしてｎ＝θ_M1／３０°＋２ｊを決定する（Ｓ２）。つまり、Ｉ_dmaxになった電圧ベクトルとは３０°切り替えた電圧ベクトルｖ_nを決定する。この決定において、ｎ≦０になる場合は、ｎ＝ｎ＋１２にする。
【００４７】
次に、処理Ｓ２で決定した電圧ベクトルｖ_nを出力し（Ｓ３）、この電圧ベクトルに対して電流検出手段１２で検出されるｄ軸電流Ｉ_dnがＩ_dn＞Ｉ_dmaxか否かを判定する（Ｓ４）。
【００４８】
この判定で、Ｉ_dn＞Ｉ_dmaxの場合、その電圧ベクトルｖ_nの角度（ｎ−１）×３０°を推定磁極位置θ_M1、そのときの電流Ｉ_dnをＩ_dmaxとして更新し、またｉ＝０とする（Ｓ５）。
【００４９】
前記判定で、Ｉ_dn≦Ｉ_dmaxの場合、ｉ＝ｊ＝１か否かを判定し（Ｓ６）、ｉ＝ｊ＝１であればｊ＝０にし、ｉ＝ｊ＝１でなければ推定処理を終了する（Ｓ７）。
【００５０】
処理Ｓ５またＳ７の終了で、ｎ＝４またはｎ＝１０でない限り、処理Ｓ２に戻り、印加する電圧ベクトルｖ_nを変えて電流Ｉ_dnとＩ_dmaxの比較を行う（Ｓ８）。
【００５１】
以上の推定処理は、例えば、推定磁極位置θ_M1と最大ｄ軸電流値Ｉ_dmaxが得られた電圧ベクトルが図１２のｖ₁であるとき、処理Ｓ２〜Ｓ５とＳ８の繰り返しではＩ_dn＞Ｉ_dmaxが続く限り電圧ベクトルｖ₂、さらにｖ₃まで切り替えると共に推定磁極位置θ_M1と最大ｄ軸電流値Ｉ_dmaxを更新して最も近い電圧ベクトル位置の検索と更新を繰り返す。そして、これら処理でＩ_dn≦Ｉ_dmaxになったとき、電圧ベクトルの検索を停止し、このときの電圧ベクトルｖ_nを実際の磁極位置θ_reに最も近い凡その磁極位置として推定する。
【００５２】
以上の処理により、実際の磁極位置θ_reに最も近い電圧ベクトルｎの角度が推定磁極位置θ_M1となる。これら第１の推定手段では、電圧ベクトルを３０°毎に出力する場合であるため、磁極位置θ_reは、θ_M1±１５°の精度で推定することができる。
【００５３】
しかし、前記のように、磁極位置θ_reが９０°または−９０°付近の場合、それぞれのｄ軸電流値Ｉ_d1，Ｉ_d7にほとんど違いが現れないことがある。このため、第１の推定手段１３では、磁極位置θ_reが９０°または−９０°付近の場合、図１３に示すように、推定磁極位置θ_M1を基準にして、一定の角度差（例えば±１５°）の位置の電圧ベクトルｖ₁₃，ｖ₁₄をモータに印加し、この印加に対するｄ軸電流の絶対値の差を比較することで推定磁極位置θ_M1を更新する。図１３の場合では磁極位置θ_reを±７.５°の精度で推定することができる。
【００５４】
図１に戻って、第２の推定手段１４は、第１の推定手段で検出した電圧ベクトル位置を中心にして、一定間隔で細かに角度を変えた電圧ベクトルをモータ１に印加し、これら電圧ベクトルの印加に対するｄ軸電流値の差が最も大きくなる電圧ベクトルの位置を磁極位置と推定する処理を繰り返して最終的な磁極位置を推定する。
【００５５】
第１の推定手段１３の例では、実際の磁極位置θ_reは、θ_M1±１５°以内の精度で推定されているので、第２の推定手段１４では、まず、図３の（ａ）に示すように、θ_M1±７.５°の角度にした電圧ベクトルｖ₁’ｖ₂’をそれぞれモータ１に印加し、第１の推定手段１３と同様に、電圧ベクトルｎを印加したときのｄ軸電流Ｉ_dnを比較して新たな推定磁極位置θ_M1を決定する。さらに、この決定の基に、（ｂ）に示すように、θ_M1±３.７５°の角度になる電圧ベクトルｖ₃’，ｖ₄’をモータ１に印加してθ_M21を求める。同様に、θ_M21を基に（ｃ）に示すように電圧ベクトルｖ₅’，ｖ₆’によるθ_M22を求める。これら処理を繰り返すことで、推定磁極位置θ_Mを実際の磁極位置θ_reに高い精度で一致させた推定ができる。
【００５６】
なお、以上までの推定装置は、例えば、電圧ベクトルは定格の７５％の大きさで２００μｓ間だけ出力し、各電圧ベクトルの出力後はすべてのゲート信号を６００μｓ間だけ非導通にして各相電流を零にし、同じ状態から次の電圧ベクトルを出力する。
【００５７】
以上のとおり、本実施形態では、停止状態のモータ１に印加する電圧ベクトルに対する電流応答により、磁極位置を推定するため、モータ定数の設定誤差や変動の影響を受けることなく確実に磁極位置を推定できる。
【００５８】
本発明の推定装置を検証するための実験を行った。なお、ＰＭモータ１は、定格容量４００Ｗ、最大定格電圧１５２Ｖ、最大定格電流１.９Ａ、同期速度５０Ｈ_Z、最大回転数３０００ｒｐｍ、磁極は４極のものを使用し、ロータリーエンコーダ６には分解能４０００ｐ／ｒのものを使用して磁極位置推定を行った。
【００５９】
図４〜図９は、実験結果を示し、図４は実際の磁極位置θ_re＝５４°で推定を行ったときのｄ軸電流Ｉ_dnと、その最大値Ｉ_dmaxの変化を示す。Ｉ_dnは、電圧ベクトルｎを印加したときのｄ軸電流である。まず、第１の推定手段による最初の推定によって、電流Ｉ_n ⁺，Ｉ_n ^-として電圧ベクトルｖ₁とｖ₇が求められ、実際の磁極位置θ_re＝５４°であるため、ｄ軸電流Ｉ_d1＞Ｉ_d7となる。そのため、電圧ベクトルｖ₁の角度０°が推定磁極位置θ_M1となり、Ｉ_d1がＩ_dmaxとなる。
【００６０】
第１の推定手段１３での次の推定（図２の処理）では、最初に電圧ベクトルｖ₂を印加し、図４ではＩ_d2＞Ｉ_dmaxであるため、θ_M1は３０°に更新される。さらに、電圧ベクトルｖ₃を印加し、Ｉ_d3＞Ｉ_dmaxであるため、θ_M1は６０°に更新される。最後に、電圧ベクトルｖ₄が印加されるが、Ｉ_d4＜Ｉ_dmaxであるため、θ_M1は更新されずに第１の推定手段による処理を終了する。このとき、推定磁極位置θ_M1は実際の磁極位置θ_re＝５４°に最も近い電圧ベクトルｖ₃の角度６０°になる。
【００６１】
次に、第２の推定手段１４では、前記の図３の電圧ベクトルｖ₁’〜ｖ₆’の順次切り替えた印加により、最終的な推定結果θ_M2には電圧ベクトルｖ₁’の角度５２.５°となった。このときの各相電流値を２００μｓ間隔でサンプルした結果を図５に示す。同図に示すように、電圧ベクトルの印加後、電流値は直ちに減少し、いずれの電圧ベクトルもほぼ同じ電流零の状態から印加されている。
【００６２】
図６は、実際の磁極位置θ_re＝３６°でのＩ_dnとＩ_dmaxの変化を示す。第１の推定手段１３においては、電圧ベクトルｖ₁₂の角度−３０°がθ_M1となり、第２の推定手段では電圧ベクトルｖ₄’を印加したときのｄ軸電流値Ｉ_d4がＩ_dmaxとなり、θ_M2は−３３.７５°という結果を得ることができた。
【００６３】
同様に、図７と図８にそれぞれ実際の磁極位置θ_re＝９９°と−８１°でのＩ_dnとＩ_dmaxの変化を示す。どちらの場合も、それぞれ実際の磁極位置θ_reが９０°と−９０°近辺にあって、それぞれのｄ軸電流値Ｉ_dにほとんど違いが現れない特殊な条件のため、図２のように、推定された磁極位置θ_M1を基準にしてその前後に一定の角度差をもつ電圧ベクトルを印加した推定を行い、それぞれθ_M1には１０５°−７５°、θ_M2には１０１.２５°−８６.２５°と推定された。
【００６４】
図９は、実際の磁極位置θ_reを０°から９°おきに３６０°まで、計４１点にしたときのそれぞれの推定結果を示す。この結果は、平均誤差４.１３°、最大誤差１６.８８°の精度で磁極位置を推定することができた。また、推定に要する時間は、最大で９.１ｍｓ、最小で７.４ｍｓであった。また、推定中の巻線電流値は、最大で定格の８０％程度となるが、いずれの場合も回転子の変位は認められなかった。
【００６５】
【発明の効果】
以上のとおり、本発明によれば、電機子鉄心の磁化特性の非線形性を利用し、永久磁石式同期電動機の磁極位置を推定するようにしたため、停止状態での磁極位置を高い精度で確実に推定できる効果がある。
【００６６】
また、第１の推定処理で凡その磁極位置を推定し、第２の推定処理で磁極位置を細かに推定するため、推定処理回数を減らして推定時間を短縮できる。
【００６７】
また、磁極位置が９０°または−９０°付近になるときも磁極位置を確実に推定できる。
【図面の簡単な説明】
【図１】本発明の実施形態を示すシステム構成図。
【図２】実施形態における第１の推定手段での推定処理手順図。
【図３】実施形態における第２の推定手段での推定磁極位置θ_M1、θ_M21、θ_M22を基準とした電圧ベクトル図。
【図４】本発明に基づいた実験結果を示す位置推定特性（θ_re＝５４°）。
【図５】本発明に基づいた実験結果を示す各相電流値の変化（θ_re＝５４°）。
【図６】本発明に基づいた実験結果を示す位置推定特性（θ_re＝３６°）。
【図７】本発明に基づいた実験結果を示す位置推定特性（θ_re＝９９°）。
【図８】本発明に基づいた実験結果を示す位置推定特性（θ_re＝−８１°）。
【図９】本発明に基づいた実験結果を示す磁極位置推定特性。
【図１０】ＰＭモータモデル
【図１１】本発明を原理的に説明するための電圧印加時の各相電流値。
【図１２】本発明に係る電圧ベクトルｎ（１〜１２）。
【図１３】推定磁極位置θ_M1を基準とした電圧ベクトル図。
【符号の説明】
１…ＰＭモータ
２…ＰＷＭインバータ
３…ドライブ回路
４…コントローラ
５Ｕ，５Ｖ…電流検出器
６…ロータリーエンコーダ
１１…電圧ベクトル発生手段
１２…電流検出手段
１３…第１の推定手段
１４…第２の推定手段
１５…制御手段[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a sensorless control system for a permanent magnet type synchronous motor such as a PM motor or a brushless motor, and more particularly to a magnetic pole position estimation device that detects a magnetic pole position when the motor is stopped without using a position sensor.
[0002]
[Prior art]
Since a control device for a permanent magnet type synchronous motor performs torque control by passing a current through an armature winding in accordance with the magnetic pole position of a rotor, a position sensor such as an encoder or a resolver is generally provided. However, these position sensors cause an increase in size and cost of the motor control system, and there are problems in terms of reliability and environmental resistance. Therefore, researches on sensorless control methods that eliminate the need for position sensors have been actively conducted.
[0003]
Currently proposed methods for estimating the magnetic pole position when the motor is stopped can be broadly classified as follows.
[0004]
(1) A method of estimating using the relationship between voltage and current on two-axis coordinates using the fact that the inductance of the salient pole motor changes with the magnetic pole position.
[0005]
(2) A method of estimating by using voltage information when an alternating current is injected by utilizing the fact that the inductance of the salient pole motor changes depending on the magnetic pole position.
[0006]
(3) A method of estimating by using the peak current value at the time of applying a pulse voltage by utilizing the fact that the inductance of the salient pole motor changes depending on the magnetic pole position.
[0007]
(4) A method of estimating from a current vector locus when a high frequency voltage is applied by utilizing the fact that the inductance of the salient pole motor changes depending on the magnetic pole position.
[0008]
(5) A method in which a conductive nonmagnetic material is pasted on the rotor surface of a cylindrical motor to estimate the magnetic pole position when stopped.
[0009]
[Problems to be solved by the invention]
Most conventional sensorless control methods are based on the motor model equation, and obtain information on the magnetic pole position and rotor speed by utilizing the fact that the inductance of the salient-pole motor changes depending on the magnetic pole position.
[0010]
However, in the case of a cylindrical motor, there is generally little difference between the inductance values of the d-axis and the q-axis, so the estimation from the model formula cannot estimate the magnetic pole position or the like when the motor is started or stopped, When the motor starts, the torque decreases or temporarily involves a large reverse rotation.
[0011]
An object of the present invention is to provide a sensorless control system for an electric motor that can reliably and accurately estimate the magnetic pole position when the permanent magnet synchronous motor is stopped.
[0012]
[Means for Solving the Problems]
(Principle of the invention)
FIG. 10 shows a PM motor model of a two-pole machine. The stator winding axis is the α-β coordinate, the axis fixed to the rotor is the dq coordinate, and the d axis is in the direction of the magnetic flux generated by the field. Further, an axis estimated by the estimating device side is d _e -q _e coordinates, the actual position angle theta _re, and Δθ the angle difference between the estimated position angle theta _M, d-axis and d _e axis.
[0013]
The voltage equation on the α-β coordinate axis of the stator winding in the above general motor model is as follows.
[0014]
[Expression 1]

[0015]
Where vα and vβ are the armature voltage α and β-axis components, iα and iβ are the armature current α and β-axis components, R is the armature resistance, L _d and L _q are the d and q-axis inductance components, p represents a differential operator.
[0016]
As shown in the above equation, in the salient pole type motor, the position angle information exists in the form of a 2θ _re trigonometric function due to the difference in d and q axis inductances even when stopped. However, the absolute position including the field polarity (d-axis direction) cannot be estimated from a general mathematical model, as shown in the above equation. In addition, since the cylindrical motor can be generally regarded as L _d = L _q = L, the above equation becomes the following equation, and position angle information at the time of stopping cannot be obtained from this mathematical model.
[0017]
[Expression 2]

[0018]
The present invention uses the non-linearity of the magnetization characteristics of the armature core in the PM motor to estimate the magnetic pole position at the time of stop, and estimates the magnetic pole position from the difference in current response due to this non-linearity. It is.
[0019]
(A)-(c) of FIG. 11 shows the current value of each phase when a positive and negative constant voltage (75% of the rating) is applied to each phase of u, v, w of the PM motor for a time of 200 μs. The result measured for every 10 degrees of rotor position angles is shown. I _u ⁺ shown in (a) of FIG. 4 indicates a u-phase current value when a positive voltage is applied to the u phase and a negative voltage is applied to the v and w phases. I _u ⁻ represents a u-phase current value when a negative voltage is applied to the u phase and a positive voltage is applied to the v and w phases. Similarly, I _v ⁺ , I _v ⁻ and I _w ⁺ , I _w ⁻ shown in (b) and (c) indicate current values of the v and w phases when a voltage is applied to each phase. Note that the negative current values I _u ⁻ , I _v ⁻ , and I _w ⁻ are expressed as absolute values for easy comparison of the magnitudes.
[0020]
From the measurement results shown in FIG. 11 (a) above, in the vicinity of θ _re = 0 where the polarity of the field near the u-phase winding is N, when a voltage is applied in the direction of magnetization, that is, positive When a voltage is applied, the magnetic flux is saturated and the inductance decreases, so that the time change of the current I _u ⁺ increases. On the other hand, when a negative voltage is applied, the inductance does not increase or change, so that the time change of the current I _u ⁻ is smaller than that when a positive voltage is applied. Also, in the vicinity of θ _re = 180 ° where the field polarity in the vicinity of the u-phase winding is S, the time change of the current I _u ⁻ is larger than the current I _u ⁺ .
[0021]
In FIG. 11 (d), the current value differences ΔI _u (= I _u ⁺ + I _u ⁻ ), ΔI _v (= I _v ⁺ + I _v ⁻ ), ΔI _w (= I _w ⁺ + I _w ^{−) are shown.} ). From the results shown in the figure, ΔI _u has a difference of about 0.1 amperes in the vicinity of θ _re = 0 ° and 180 °, and similarly, ΔI _v has a difference in the vicinity of θ _re = 120 ° and 300 °. ΔI _w has a difference between θ _re = 240 ° and around 60 °. That is, in each phase, there is a difference in the current values in the magnetizing direction and the demagnetizing direction in the vicinity of the polar axis in the current value differences ΔI _u , ΔI _v , ΔI _w .
[0022]
The difference in the current value of each phase appears because the magnetic flux is saturated and the inductance decreases due to the nonlinearity of the magnetization characteristics of the armature core with respect to the applied voltage. Therefore, by utilizing the fact that the difference in current value between the magnetizing direction and the demagnetizing direction in the vicinity of the polar axis is large, by detecting the voltage vector position where this current value difference is the largest, The position can be estimated. Accordingly, the present invention is characterized by the following configuration.
[0023]
(First invention)
In the sensorless control system of a permanent magnet type synchronous motor that controls the voltage angle supplied to the armature winding of the motor according to the magnetic pole position of the permanent magnet type synchronous motor,
When the motor is stopped, a voltage vector of positive and negative constant voltages whose angles are changed at regular intervals is applied to the motor, and the current of the armature winding is detected with respect to the application of each positive and negative voltage vector . The current value is detected as the d-axis current value of the coordinates fixed to the rotor by dq coordinate conversion at the angle of the voltage vector applied at that time, and the difference of the d-axis current value with respect to the application of the positive and negative voltage vectors. A magnetic pole position estimation device is provided that estimates the position of the voltage vector that has the largest value as the magnetic pole position.
[0024]
(Second invention)
In the first aspect of the invention, the accuracy of the estimated magnetic pole position is determined by the angular interval of the voltage vector, and the accuracy increases as the angular interval is reduced, but the number of output of the voltage vector increases and the estimation process is performed. take time. Therefore, the present invention reduces the number of times of estimation processing and time by dividing the estimation of the magnetic pole position into a plurality of stages of rough estimation and fine estimation, and has the following configuration.
[0025]
The magnetic pole position estimation device is
The position of the voltage vector at which the difference in the d-axis current value detected when the voltage vector of a positive and negative constant voltage whose angle is roughly changed at regular intervals is applied to the motor is the largest when the motor is stopped. First estimating means for estimating as a magnetic pole position,
The voltage vector having the largest difference in d-axis current value when a voltage vector whose angle is finely changed at regular intervals around the voltage vector position detected by the first estimating means is applied to the motor. And a second estimating means for estimating the final magnetic pole position by repeating the process of estimating the position of the magnetic pole position as the magnetic pole position.
[0026]
(Third invention)
In the first estimating means, when the actual magnetic pole position θ _re is 90 ° or near −90 °, there _may be little difference between the d-axis current values I _d1 and I _d7 . This example is shown in Table 1 below with the d-axis current values I _d1 and I _d7 and the difference between the absolute values | I _d1 −I _d7 |.
[0027]
[Table 1]

[0028]
Theoretically, if −90 ° <θ _re <90 °, I _d1 > I _d7 , if θ _re = 90 ° or −90 °, I _d1 = I _d7 , otherwise I _d1 <I _d7. However, as shown in Table 1 above, there is “variation”, and the difference is very small.
[0029]
This is because when the magnetic pole position is around 90 ° and −90 °, the influence of magnetic saturation is small and the change in inductance is small, so there is almost no difference in the d-axis current value. It seems that an error is caused by the influence of noise.
[0030]
Therefore, the present invention is the first when the difference | I _d1 −I _d7 | of the absolute values of the d-axis current values I _d1 and I _d7 is less than the error range (for example, 0.03 amperes in the above table). The following estimation is performed instead of the estimation of the magnetic pole position in the first estimation means.
[0031]
First, the 90 ° and −90 ° voltage vectors v ₄ and v ₁₀ in FIG. 12 are applied to the motor, and the larger d-axis current I _d detected in the same manner as in the above estimation 1 for this voltage application. _{Is the} estimated magnetic pole position θ _M1 , and its current is I _dmax .
[0032]
As a result, the actual magnetic pole position θ _re can be estimated with an accuracy of θ _M1 ± 15 ° with respect to the voltage vector of FIG. 12. However, as shown in Table 1, the magnetic pole position θ _re is 90 ° ± 15 °. Even in the vicinity of −90 ° ± 15 °, the difference | I _d1 −I _d7 | becomes a small value.
[0033]
Therefore, as shown in FIG. 13, voltage vectors v ₁₃ and v ₁₄ at ± 15 ° positions are applied to the motor with reference to the estimated magnetic pole position θ _M1, and the absolute value of the d-axis current with respect to this application is calculated. The estimated magnetic pole position θ _M1 is updated by comparing the difference. The magnetic pole position is narrowed down by the second estimating means for the updated estimated magnetic pole position θ _M1 . From the above, the present invention is characterized by the following configurations.
[0034]
The magnetic pole position estimation device is
When the magnetic pole position estimated by the first estimating means is near 90 ° or near −90 °, a voltage vector having a certain angular difference before and after the estimated magnetic pole position is applied to the motor. Means for updating the position of the voltage vector at which the difference in d-axis current value is greatest as the magnetic pole position.
[0035]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a system configuration diagram showing an embodiment of the present invention. The PM motor (permanent magnet type synchronous motor) 1 is synchronously operated by supplying current to the armature winding by the PWM inverter 2. The PWM inverter 2 receives a main circuit switch on / off control signal from the drive circuit 3.
[0036]
The current detectors 5U and 5V detect U and V phase currents supplied to the PM motor 1.
[0037]
The controller 4 controls the angle of the drive signal to the drive circuit 3 according to the estimated magnetic pole position, and controls torque by controlling the winding current of the PM motor 1 from the detected current from the current detectors 5U and 5V. Do.
[0038]
The rotary encoder 6 coupled to the PM motor 1 is provided for observing the rotor position variation, and is not provided in an actual sensorless control system.
[0039]
Here, in this embodiment, as means for estimating the magnetic pole position of the PM motor 1 in the stopped state, the controller 4 includes a voltage vector generation means 11, a current detection means 12, and first and second estimation means 13. , 14 and control means 15 for controlling these means.
[0040]
Based on the control of the control means 15, the voltage vector generation means 11 sequentially generates positive and negative voltage vectors whose angles are changed every 30 ° as shown in FIG. 12, and further, with the estimated magnetic pole position θ _M as a reference. A voltage vector whose angle is changed before and after is sequentially generated. These voltage vectors are applied to the PM motor 1 in a stopped state by the PWM inverter 2 as voltages having different angles.
[0041]
The current detection means 12 detects the currents I _n ⁺ and I _n ⁻ that flow through the armature winding when positive and negative voltages are applied to the PM motor 1.
[0042]
The first estimation means 13 and the second estimation means 14 cooperate to apply a voltage vector of positive and negative constant voltages whose angles are changed at regular intervals to the motor 1 while the motor 1 is stopped. A d-axis current value of coordinates fixed to the rotor with respect to application of the voltage vector is detected, and the position of the voltage vector at which the difference between the current values is the largest is estimated as the magnetic pole position.
[0043]
Of these, the first estimation means 13 applies a voltage vector of positive and negative constant voltages whose angles are roughly changed at regular intervals to the motor in a stopped state of the motor, and is fixed to the rotor for the application of each voltage vector. The d-axis current value of the coordinate is detected, and the position of the voltage vector at which the difference between the current values is the largest is estimated as the approximate magnetic pole position. This estimation process will be described in detail below.
[0044]
First, the current detector 12 current detected by I _n ^+, I _n ^- into a d-q coordinates at an angle of a voltage vector, detects two d-axis current value I _d, the larger the voltage of them The vector position is the estimated magnetic pole position θ _M1 , and the current value I _d at that time is the maximum d-axis current value I _dmax . In this estimation, it is determined whether the magnetism side is demagnetized or demagnetized, and it is estimated which side of the positive or negative voltage vector the magnetic pole position is.
[0045]
Thereafter, it estimated using or around either direction of the voltage vector position where the estimated magnetic pole position theta _M1 is the actual magnetic pole position theta _re estimated magnetic pole position theta _M1 and the maximum d-axis current value I _dmax. This estimation processing procedure is shown in FIG.
[0046]
In FIG. 2, I _dn is a d-axis current value when the voltage vector n in FIG. 12 is output. I and j are determination coefficients used for the end of estimation. In the process of FIG. 2, first, i = 1, j = 1, and n = 0 are initially set (S1), and n = θ _M1 / 30 ° + 2j is determined as the voltage vector v _n to be applied (S2). . In other words, the voltage vector becomes I _dmax determines the voltage vector v _n of switching 30 °. In this determination, if n ≦ 0, n = n + 12.
[0047]
Then, the output voltage vector v _n determined by the processing S2 (S3), d-axis current I _dn detected by the current detection unit 12 with respect to this voltage vector is determined whether I _dn> I _dmax (S4).
[0048]
In this determination, if the I _dn> I _dmax, the voltage vector v _n of the angle (n-1) × 30 ° the estimated magnetic pole position theta _M1, and updates the current I _dn at that time as I _dmax, also i = Set to 0 (S5).
[0049]
In the above determination, if I _dn ≦ I _dmax , it is determined whether i = j = 1 (S6). If i = j = 1, j = 0 is set. If i = j = 1, the estimation process is not performed. Is finished (S7).
[0050]
At the end of step S5 also S7, unless n = 4 or n = 10, returns to the process S2, performs the comparison of the current I _dn and I _dmax by changing the voltage vector v _n to be applied (S8).
[0051]
In the above estimation process, for example, when the estimated magnetic pole position θ _M1 and the voltage vector from which the maximum d-axis current value I _dmax is obtained are v ₁ in FIG. 12, I _dn > I in the repetition of the processes S2 to S5 and S8. _As long as _dmax continues, the voltage vector v _{2 is} switched to v ₃ , and the estimated magnetic pole position θ _M1 and the maximum d-axis current value I _dmax are updated to repeat the search and update of the closest voltage vector position. When I _dn ≦ I _dmax in these processes, the search for the voltage vector is stopped, and the voltage vector v _n at this time is estimated as the approximate magnetic pole position closest to the actual magnetic pole position θ _re .
[0052]
With the above processing, the angle of the voltage vector n closest to the actual magnetic pole position θ _re becomes the estimated magnetic pole position θ _M1 . Since these first estimation means output voltage vectors every 30 °, the magnetic pole position θ _re can be estimated with an accuracy of θ _M1 ± 15 °.
[0053]
However, as described above, when the magnetic pole position θ _re is around 90 ° or −90 °, there _may be little difference between the d-axis current values I _d1 and I _d7 . For this reason, in the first estimating means 13, when the magnetic pole position θ _re is near 90 ° or −90 °, as shown in FIG. 13, a certain angular difference (for example, _{±±) is used} with reference to the estimated magnetic pole position θ _M1. The estimated magnetic pole position θ _M1 is updated by applying voltage vectors v ₁₃ and v ₁₄ at a position of 15 ° to the motor and comparing the difference in absolute value of the d-axis current with respect to this application. In the case of FIG. 13, the magnetic pole position θ _re can be estimated with an accuracy of ± 7.5 °.
[0054]
Returning to FIG. 1, the second estimation means 14 applies to the motor 1 a voltage vector whose angle is finely changed at regular intervals around the voltage vector position detected by the first estimation means. The final magnetic pole position is estimated by repeating the process of estimating the position of the voltage vector where the difference in d-axis current value with respect to the application of the vector is the largest as the magnetic pole position.
[0055]
In the example of the first estimating means 13, the actual magnetic pole position θ _re is estimated with an accuracy within θ _M1 ± 15 °. Therefore, in the second estimating means 14, first, as shown in FIG. As shown, when the voltage vector v ₁ 'v ₂ ' at an angle of θ _M1 ± 7.5 ° is applied to the motor 1 and the voltage vector n is applied as in the first estimating means 13, d The shaft current I _dn is compared to determine a new estimated magnetic pole position θ _M1 . Further, based on this determination, as shown in (b), voltage vectors v ₃ ′ and v ₄ ′ having an angle of θ _M1 ± 3.75 ° are applied to the motor 1 to obtain θ _M21 . Similarly, θ _M22 based on voltage vectors v ₅ ′ and v ₆ ′ is obtained based on θ _M21 as shown in (c). By repeating these processes, it is possible to estimate the estimated magnetic pole position θ _{M so as} to coincide with the actual magnetic pole position θ _re with high accuracy.
[0056]
In the estimation device described above, for example, the voltage vector is output at a magnitude of 75% of the rating for only 200 μs, and after the output of each voltage vector, all the gate signals are made non-conductive for 600 μs, and each phase current is output. Is set to zero, and the next voltage vector is output from the same state.
[0057]
As described above, in the present embodiment, the magnetic pole position is estimated based on the current response to the voltage vector applied to the motor 1 in the stopped state. Therefore, the magnetic pole position is reliably estimated without being affected by the setting error or fluctuation of the motor constant. it can.
[0058]
An experiment was conducted to verify the estimation apparatus of the present invention. Incidentally, PM motor 1, rated capacity 400W, maximum rated voltage 152V, the maximum rated current 1.9A, synchronous speed 50H _Z, the maximum rotational speed 3000 rpm, pole Use a four-pole, resolution 4000p the rotary encoder 6 The magnetic pole position was estimated using the / r one.
[0059]
4 to 9 show experimental results, and FIG. 4 shows changes in the d-axis current I _dn and its maximum value I _dmax when estimation is performed at the actual magnetic pole position θ _re = 54 °. I _dn is a d-axis current when the voltage vector n is applied. First, the initial estimated by the first estimation means, the current I _n ^+, I _n ^- the voltage vector v ₁ and v ₇ obtained as, for the actual magnetic pole position θ _re = 54 °, d-axis current I _d1 > I _d7 . Therefore, the angle 0 ° of the voltage vector v ₁ is the estimated magnetic pole position θ _M1 and I _d1 is I _dmax .
[0060]
In the next estimation by the first estimation means 13 (the process of FIG. 2), the voltage vector v ₂ is first applied. In FIG. 4, since I _d2 > I _dmax , θ _M1 is updated to 30 °. . Furthermore, since the voltage vector v ₃ is applied and I _d3 > I _dmax , θ _M1 is updated to 60 °. Finally, the voltage vector v ₄ is applied, but since I _d4 <I _dmax , θ _M1 is not updated and the processing by the first estimating means is terminated. At this time, the estimated magnetic pole position θ _M1 becomes the angle 60 ° of the voltage vector v ₃ closest to the actual magnetic pole position θ _re = 54 °.
[0061]
Then, the angle 52 of the the second estimation means 14, by sequentially switching the application of the voltage vector v ₁ of FIG. 3 'to v _6', the voltage vector v ₁ in the final estimation result theta _M2 '. It was 5 °. FIG. 5 shows the result of sampling the phase current values at this time at intervals of 200 μs. As shown in the figure, after the voltage vector is applied, the current value immediately decreases, and all the voltage vectors are applied from substantially the same current zero state.
[0062]
FIG. 6 shows changes in I _dn and I _dmax at the actual magnetic pole position θ _re = 36 °. In the first estimation means 13, the angle −30 ° of the voltage vector v ₁₂ is θ _M1 , and in the second estimation means, the d-axis current value I _d4 when the voltage vector v ₄ ′ is applied becomes I _dmax , As a result, θ _M2 was −33.75 °.
[0063]
Similarly, FIGS. 7 and 8 show changes in I _dn and I _dmax at actual magnetic pole positions θ _re = 99 ° and −81 °, respectively. In both cases, the actual magnetic pole position θ _re is in the vicinity of 90 ° and −90 °, respectively, and the d-axis current value I _d is not significantly different. Therefore, as shown in FIG. Estimation is performed by applying a voltage vector having a certain angle difference before and after the estimated magnetic pole position θ _M1 as a reference, and θ _M1 is 105 ° -75 ° and θ _M2 is 101.25 ° -86, respectively. Estimated to be .25 °.
[0064]
FIG. 9 shows respective estimation results when the actual magnetic pole position θ _{re is set} to 41 points in total from 0 ° to 360 ° every 9 °. As a result, the magnetic pole position could be estimated with an accuracy of an average error of 4.13 ° and a maximum error of 16.88 °. In addition, the time required for estimation was 9.1 ms at the maximum and 7.4 ms at the minimum. The estimated winding current value is about 80% of the maximum rating, but no rotor displacement was observed in either case.
[0065]
【The invention's effect】
As described above, according to the present invention, since the magnetic pole position of the permanent magnet synchronous motor is estimated using the non-linearity of the magnetization characteristics of the armature core, the magnetic pole position in the stopped state can be reliably and accurately detected. There is an effect that can be estimated.
[0066]
In addition, since the approximate magnetic pole position is estimated in the first estimation process and the magnetic pole position is estimated in detail in the second estimation process, the estimation time can be reduced by reducing the number of estimation processes.
[0067]
Further, the magnetic pole position can be reliably estimated even when the magnetic pole position is 90 ° or near −90 °.
[Brief description of the drawings]
FIG. 1 is a system configuration diagram showing an embodiment of the present invention.
FIG. 2 is an estimation processing procedure diagram of first estimation means in the embodiment.
FIG. 3 is a voltage vector diagram based on estimated magnetic pole positions θ _M1 , θ _M21 , and θ _M22 in the second estimating means in the embodiment.
FIG. 4 is a position estimation characteristic (θ _re = 54 °) showing an experimental result based on the present invention.
FIG. 5 shows a change in current value of each phase showing an experimental result based on the present invention (θ _re = 54 °).
FIG. 6 is a position estimation characteristic (θ _re = 36 °) showing an experimental result based on the present invention.
FIG. 7 is a position estimation characteristic (θ _re = 99 °) showing an experimental result based on the present invention.
FIG. 8 is a position estimation characteristic (θ _re = −81 °) showing experimental results based on the present invention.
FIG. 9 shows magnetic pole position estimation characteristics showing experimental results based on the present invention.
FIG. 10 shows a PM motor model. FIG. 11 shows each phase current value when a voltage is applied to explain the present invention in principle.
FIG. 12 shows a voltage vector n (1-12) according to the present invention.
FIG. 13 is a voltage vector diagram based on the estimated magnetic pole position θ _M1 .
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... PM motor 2 ... PWM inverter 3 ... Drive circuit 4 ... Controller 5U, 5V ... Current detector 6 ... Rotary encoder 11 ... Voltage vector generation means 12 ... Current detection means 13 ... First estimation means 14 ... Second estimation Means 15 ... Control means

Claims (3)

In the sensorless control system of a permanent magnet type synchronous motor that controls the voltage angle supplied to the armature winding of the motor according to the magnetic pole position of the permanent magnet type synchronous motor,
When the motor is stopped, a voltage vector of positive and negative constant voltages whose angles are changed at regular intervals is applied to the motor, and the current of the armature winding is detected with respect to the application of each positive and negative voltage vector . The current value is detected as the d-axis current value of the coordinates fixed to the rotor by dq coordinate conversion at the angle of the voltage vector applied at that time, and the difference of the d-axis current value with respect to the application of the positive and negative voltage vectors. A sensorless control system for a permanent magnet type synchronous motor, comprising: a magnetic pole position estimation device that estimates the position of the voltage vector that has the largest value as a magnetic pole position. The magnetic pole position estimation device is
The position of the voltage vector at which the difference in the d-axis current value detected when the voltage vector of a positive and negative constant voltage whose angle is roughly changed at regular intervals is applied to the motor is the largest when the motor is stopped. First estimating means for estimating as a magnetic pole position,
The voltage vector having the largest difference in d-axis current value when a voltage vector whose angle is finely changed at regular intervals around the voltage vector position detected by the first estimating means is applied to the motor. 2. The sensorless control system for a permanent magnet type synchronous motor according to claim 1, further comprising: a second estimation unit that estimates a final magnetic pole position by repeating the process of estimating the position of the magnetic pole position as a magnetic pole position. The magnetic pole position estimation device is
When the magnetic pole position estimated by the first estimating means is near 90 ° or near −90 °, a voltage vector having a certain angular difference before and after the estimated magnetic pole position is applied to the motor. 3. A sensorless control system for a permanent magnet type synchronous motor according to claim 1, further comprising means for updating the position of the voltage vector at which the difference in the d-axis current value is greatest as the magnetic pole position. .

Images