JP3676133B2 - Frequency characteristics identification method for power system - Google Patents

Description

【０００７】
ここで、τは相関関数演算の際の仮想的な時間パラメータであり、ｘ^* （ｔ）はｘ（ｔ）の共役関数である。
【０００８】
ステップＳ２で同定用信号ｘ（ｔ）を電力系統に注入し、ｙ（ｔ）を検出する（ステップＳ３）。
【０００９】
次にステップＳ４において、下記式（２）の演算を行い、ｘ（ｔ）とｙ（ｔ）との相関関数〔Ｒ_xy（τ）〕を求める。
【００１０】
【数２】

【００１１】
ここで、ｙ^* （ｔ）はｙ（ｔ）の共役関数である。
【００１２】
さらにステップＳ５において、前記式（１），式（２）それぞれをフーリエ変換し、この変換結果を自己スペクトル密度および相互スペクトル密度として、ステップＳ６へ移る。
【００１３】
ステップＳ６では相互スペクトル密度／自己スペクトル密度の演算を行い、この演算結果を図４（イ）に示した電力系統の周波数特性Ｇ（ｊω）の同定値としている。
【００１４】
上述のステップＳ１〜Ｓ６による電力系統の周波数特性同定方法はスペクトル解析法と称され、比較的ノイズに強いため、微弱な同定用信号で前記周波数特性を同定でき、フーリエ変換を用いて複数の周波数点における特性を一度に求められるので、短時間に前記周波数特性を同定できるなどの特長を有している。
【００１５】
【発明が解決しようとする課題】
図５は、アクティブフィルタにより電力系統の高調波成分を抑制する概念的な機器構成図である。
【００１６】
図５において、１ａは上位の電力系統、１ｂはこの発明の対象となる下位の電力系統、２は下位の電力系統１ｂの負荷としての高調波電流発生源、４は電圧整合用の変圧器、５は上位の電力系統１ａと下位の電力系統１ｂとの間の電流を検出する電流検出器、６は一般負荷、７はＬＣフィルタ、１０はアクティブフィルタ、１４はアクティブフィルタ１０の出力電流を検出する電流検出器、２０，３０，４０のいずれかは下位の電力系統１ｂの周波数特性を導出し、この導出値をアクティブフィルタ１０へ伝達するための同定回路であり、同定回路２０は同定用信号指令値〔ｘ^* （ｔ）〕をアクティブフィルタ１０へ指令し、この指令に基づいてアクティブフィルタ１０の出力電流が〔ｘ（ｔ）〕になるとして、前述の図４（ロ）に示したフローチャートの動作を行う。
【００１７】
この同定回路２０では、電力系統１ｂの周波数特性Ｇ（ｊω）＝〔電流検出器５の検出値ｙ（ｔ）／電流検出器１４の検出値ｘ（ｔ）〕の演算を行っている。
【００１８】
図５からも明らかなように、アクティブフィルタ１０以外の他の機器が全て稼働している状態で下位の電力系統１ｂの周波数特性を同定する必要があるが、このとき、前記他の機器の内には高調波電流発生源２があるために、図６のブロック図に示す如く、この高調波電流〔ｈ（ｔ）〕が外乱となり、同定回路２０が導出した同定値には外乱誤差が含まれることとなる。
【００１９】
すなわち、電流検出器５の検出値ｙ（ｔ）は、電流検出器１４の検出値ｘ（ｔ）＋高調波電流発生源２の高調波電流ｈ（ｔ）に系統の周波数特性Ｇ（ｊω）が作用したものとなる。
なお、言うまでもないことであるが、前記電流検出器５および１４の検出値ｙ（ｔ）およびｘ（ｔ）は高調波電流成分のみを示している。
【００２０】
この発明の目的は、上位の電力系統と下位の電力系統との間の検出値から前記同定用信号による成分のみを抽出しつつ、電力系統の周波数特性同定方法を提供することにある。
【００２１】
【課題を解決するための手段】
この第１の発明は、発振周波数の異なる複数の正弦波信号を合成した同定用信号に基づいて電力系統の周波数特性を同定する際に、 前記同定用信号を構成する個々の正弦波信号の位相を前回より２π／ｎ〔ｒａｄ〕（ｎ：複数）ずつ変位させつつ電力系統にｎ回注入し、この注入された同定用信号に基づく前記電力系統の応答から、ｎ個の同定用信号と応答との相関関数を求め、求めたｎ個の相関関数の加算平均値をフーリエ変換して相互スペクトル密度を求め、この相互スペクトル密度を前記同定用信号の自己スペクトル密度で除算演算した値を前記周波数特性の同定値としたことを特徴とする。
【００２２】
第２の発明は、発振周波数の異なる複数の正弦波信号を合成した同定用信号に基づいて電力系統の周波数特性を同定する際に、それぞれの前記正弦波信号の振幅を互いに等しくし、該正弦波信号の位相を前回より２π／ｎ〔ｒａｄ〕（ｎ：複数）ずつ変位させつつ電力系統にｎ回注入し、この注入された同定用信号に基づく前記電力系統の応答から、ｎ個の同定用信号と応答との相関関数を求め、求めたｎ個の相関関数の加算平均値をフーリエ変換した値を前記周波数特性の同定値としたことを特徴とする。
【００２３】
第３の発明は前記第１又は第２の発明において、前記同定用信号は、電力系統と連系運転する半導体電力変換装置に付加的に指令される同定用信号指令値に基づき該電力変換装置で生成することを特徴とする。
【００２４】
第４の発明は前記第３の発明において、前記同定用信号を半導体電力変換装置の出力電流又は出力電圧とし、前記応答を上位の電力系統と前記電力系統との間の電流又は電圧として前記周波数特性の同定値を演算することを特徴とする。
【００２５】
第５の発明は前記第３の発明において、前記同定用信号を前記同定用信号指令値とし、前記応答を上位の電力系統と前記電力系統との間の電流又は電圧として前記周波数特性の同定値を演算することを特徴とする。
【００２６】
この発明は、以下に記載の作用に着目してなされたものである。
【００２７】
例えば、ｃｏｓ波形信号で、一方をＡｃｏｓ（ωｔ＋φ₁ ）とし、他方をＢｃｏｓ（ωｔ＋φ₂ ）とすると、両者の間の相関関数は式（３）で表される。
【００２８】
【数３】
（ＡＢ／２）ｃｏｓ｛ωｔ＋（φ₂ −φ₁ ）｝ …（３）
すなわち、上記相関関数には前記両者間の位相差成分が現れる。
【００２９】
そこで、前記Ａｃｏｓ（ωｔ＋φ₁ ）を同定用信号を構成する前記正弦波信号の一部とし、前記Ｂｃｏｓ（ωｔ＋φ₂ ）をその応答分のみとした場合に、前記φ₁ を前回より２π／ｎ〔ｒａｄ〕，（ｎ：複数）ずつ変位させつつ電力系統に注入して得られるｎ組の上記式（３）で示した相関関数それぞれの〔φ₂ −φ₁ 〕の値が同じになることから、上述の如くφ₁ を変位させてもそれぞれの同定用信号に対応する相関関数値は変わらない。
【００３０】
しかしながら、図６に示した高調波電流ｈ（ｔ）に起因する分も含んだ応答のうち、該ｈ（ｔ）による応答分は同定用信号による応答分とは無関係なある固定された位相にあるため、得られる前記式（３）で示した相関関数における〔φ₂ −φ₁ 〕が、上述の如くφ₁ を変位させたときにそれぞれ異なり、得られた相関関数それぞれのベクトル的な加算平均値が零となり、その結果、前記ｈ（ｔ）による応答分を排除することができる。
【００３１】
【発明の実施の形態】
図１は、この発明の第１の実施例を示すフローチャートであり、図５に示した同定回路３０における動作を示している。
【００３２】
同定用信号指令値〔ｘ^* （ｔ）〕をアクティブフィルタ１０へ指令したときに、この指令に基づいてアクティブフィルタ１０が同定用信号〔ｘ（ｔ）〕なる電流を発生する。このとき、ｘ（ｔ）は複数のｃｏｓ波電流の合成したものとすると、ｘ（ｔ）は式（４）で表される。
【００３３】
【数４】

【００３４】
先ず、ステップＳ１１では上記式（４）の相関関数Ｒ_xx（τ）を演算し、この演算結果は、式（５）で表される。
【００３５】
【数５】

【００３６】
次に、ステップＳ１２では式（５）をフーリエ変換して得られ、式（６）で表される自己スペクトル密度Ｓ_xx（ｊω）を求めている。
【００３７】
【数６】

【００３８】
ステップＳ１３では式（４）で示したｘ（ｔ）のθ₁ ，θ₂ ，・・・が初期値の状態（すなわち、前記ｎ＝ｎとして）で、図５に示した下位の電力系統１ｂへ注入する。
【００３９】
この電力系統１ｂには高調波電流発生源２が動作中であり、この高調波電流が下記式（７）で表される外乱〔ｈ（ｔ）〕となる。
【００４０】
【数７】

【００４１】
すなわち、ステップＳ１４で検出した〔ｙ（ｔ）〕は、式（８）で表される。
【００４２】
【数８】

【００４３】
ここで、Ｋ_k は周波数ω_k における電力系統１ｂのゲイン、ψ_k は周波数ω_k における位相遅れを示し、右辺の第１群は同定用信号ｘ（ｔ）による応答を示し、また、右辺の第２群は外乱ｈ（ｔ）による応答を示している。
【００４４】
ステップＳ１５では、前記式（４）で示したｘ（ｔ）と、式（８）で示したｙ（ｔ）とを前記式（２）の定義に従った相関関数Ｒ_xy（ｔ）を式（９）に示す如く求めている。
【００４５】
【数９】

【００４６】
上述のステップＳ１３からステップＳ１５の動作をステップＳ１６とステップＳ１７のシーケンスにより、同定用信号ｘ（ｔ）を構成する個々の正弦波信号の位相を前回より２π／ｎ〔ｒａｄ〕，（ｎ：複数）ずつ変位させつつ電力系統に注入し、従って、それぞれのｘ（ｔ）とｙ（ｔ）とによるｎ個の式（９）で表される相関関数が得られるので、ステップＳ１８へ移る。
【００４７】
ステップＳ１８では得られたｎ個の相関関数の加算平均値を求め、この加算平均値Ｒ_xy（ｔ）のバーは式（１０）で表される。
【００４８】
【数１０】

【００４９】
この式から明らかなように、外乱ｈ（ｔ）に対する項が消去されている。
【００５０】
ステップＳ１９において式（１０）をフーリエ変換し、式（１１）で示される相互スペクトル密度Ｓ_xy（ｊω）を求めている。
【００５１】
【数１１】

【００５２】
さらに、ステップＳ２０ではステップＳ１９で求めた相互スペクトル密度を、ステップ１２で求めた自己スペクトル密度で除算演算して、式（１２）で示す如く、それぞれの周波数における図５に示した下位の電力系統１ｂの周波数特性Ｇ（ｊω）の演算値としている。
【００５３】
【数１２】

【００５４】
図２は、この発明の第２の実施例を示すフローチャートであり、図５に示した同定回路４０における動作を示している。
【００５５】
同定用信号指令値〔ｘ^* （ｔ）〕をアクティブフィルタ１０へ指令したときに、この指令に基づいてアクティブフィルタ１０が同定用信号〔ｘ（ｔ）〕なる電流を発生する。このとき、ｘ（ｔ）は複数のｃｏｓ波電流の合成し、それぞれのｃｏｓ波電流の振幅が「１」となるように規格化すると、ｘ（ｔ）は式（１３）で表される。
【００５６】
【数１３】

【００５７】
先ず、ステップＳ３１では式（１３）で示したｘ（ｔ）のθ₁ ，θ₂ ，・・・が初期値の状態（すなわち、前記ｎ＝ｎとして）で、図５に示した下位の電力系統１ｂへ注入する。
【００５８】
この電力系統１ｂには高調波電流発生源２が動作中であり、この高調波電流が前記式（７）で表される外乱〔ｈ（ｔ）〕となる。
【００５９】
すなわちステップＳ３２で検出した〔ｙ（ｔ）〕は、式（１４）で表される。
【００６０】
【数１４】

【００６１】
ここで、Ｋ_k は周波数ω_k における電力系統１ｂのゲイン、ψ_k は周波数ω_k における位相遅れを示し、右辺の第１群は同定用信号ｘ（ｔ）による応答を示し、また、右辺の第２群は外乱ｈ（ｔ）による応答を示している。
【００６２】
ステップＳ３３では、前記式（１３）で示したｘ（ｔ）と、式（１４）で示したｙ（ｔ）とを前記式（２）の定義に従った相関関数Ｒ_xy（ｔ）を式（１５）に示す如く求めている。
【００６３】
【数１５】

【００６４】
上述のステップＳ３１からステップＳ３３の動作をステップＳ３４とステップＳ３５のシーケンスにより、同定用信号ｘ（ｔ）を構成する個々の正弦波信号の位相を前回より２π／ｎ〔ｒａｄ〕，（ｎ：複数）ずつ変位させつつ電力系統に注入し、従って、それぞれのｘ（ｔ）とｙ（ｔ）とによるｎ個の式（１５）で表される相関関数が得られるので、ステップＳ３６へ移る。
【００６５】
ステップＳ３６では得られたｎ個の相関関数の加算平均値を求め、この加算平均値Ｒ_xy（ｔ）のバーは式（１６）で表される。
【００６６】
【数１６】

【００６７】
この式から明らかなように、外乱ｈ（ｔ）に対する項が消去されている。
【００６８】
ステップＳ３７において式（１６）をフーリエ変換し、さらにステップＳ３８では式（１７）で示す如く、それぞれの周波数における図５に示した下位の電力系統１ｂの周波数特性Ｇ（ｊω）の演算値としている。
【００６９】
【数１７】

【００７０】
図３は、この発明の第３の実施例を示す電力系統における機器構成図であり、この実施例において図５に示した構成と異なる点は、同定回路４０または同定回路４０に代えて、同定回路５０を備えていることである。
【００７１】
すなわち、同定回路５０ではアクティブフィルタ１０に入力される同定用信号指令値〔ｘ^* （ｔ）〕に基づき下位の電力系統１ａの周波数特性の同定を行うようにしており、その際の演算処理は、図１または図２に示したフローチャートと同じでよい。
【００７２】
なお、上述の実施例では同定用信号として一般的な電流を用い、電力系統の電流を検出して行う例について説明したが、同定用信号として電圧を用い、電力系統の電圧を検出して、この発明の同定方法を行ってもよい。
【００７３】
【発明の効果】
この発明によれば、上位の電力系統と下位の電力系統との間の検出値から該電力系統をスペクトル解析法に基づく同定用信号による成分のみを抽出できるので、この電力系統の周波数特性を同定の際の外乱としての高調波電流発生源の影響を除去できる。
【図面の簡単な説明】
【図１】この発明の第１の実施例を示すフローチャート
【図２】この発明の第２の実施例を示すフローチャート
【図３】この発明の第３の実施例を示す機器構成図
【図４】従来例を示すフローチャート
【図５】電力系統の機器構成図
【図６】図５のブロック図
【符号の説明】
１ａ…上位の電力系統、１ｂ…下位の電力系統、２…高調波電流発生源、４…変圧器、５…電流検出器、６…負荷、７…ＬＣフィルタ、１０…アクティブフィルタ、１４…電流検出器、２０，３０，４０，５０…同定回路。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for identifying a frequency characteristic of a power system via a semiconductor power conversion device or the like that is connected to the power system.
[0002]
[Prior art]
Along with the widespread use of semiconductor power converters typified by inverters for driving motors, deterioration of the power environment such as an increase in harmonic distortion of the power system becomes a problem, and in order to correct this deterioration, the system is connected to the power system. For example, an active filter is installed as a semiconductor power conversion device.
[0003]
In this active filter, it is necessary to identify the frequency characteristics of the power system in order to better correct the above-described deterioration of the power environment, and an identification signal [x (t )] To identify the frequency characteristic [G (jω)] based on the block diagram of FIG. 4 (a) and the flowchart of FIG. 4 (b). To do.
[0004]
In FIG. 4A, the response y (t) based on the identification signal injected into the power system is the voltage or current between the illustrated power system and the upper power system.
[0005]
As shown in FIG. 4B, in step S1, the following equation (1) is calculated as the correlation function [R _xx (τ)] of x (t).
[0006]
[Expression 1]

[0007]
Here, τ is a virtual time parameter in the correlation function calculation, and x ^* (t) is a conjugate function of x (t).
[0008]
In step S2, the identification signal x (t) is injected into the power system, and y (t) is detected (step S3).
[0009]
Next, in step S4, the following equation (2) is calculated to obtain a correlation function [R _xy (τ)] between x (t) and y (t).
[0010]
[Expression 2]

[0011]
Here, y ^* (t) is a conjugate function of y (t).
[0012]
Further, in step S5, each of the equations (1) and (2) is subjected to Fourier transform, and the transformation results are set as a self-spectral density and a cross-spectral density, and the process proceeds to step S6.
[0013]
In step S6, the mutual spectral density / self-spectral density is calculated, and the calculation result is used as the identification value of the frequency characteristic G (jω) of the power system shown in FIG.
[0014]
The power system frequency characteristic identification method in steps S1 to S6 described above is called a spectrum analysis method, and is relatively resistant to noise. Therefore, the frequency characteristic can be identified with a weak identification signal, and a plurality of frequencies can be identified using Fourier transform. Since the characteristics at a point can be obtained at a time, the frequency characteristics can be identified in a short time.
[0015]
[Problems to be solved by the invention]
FIG. 5 is a conceptual device configuration diagram that suppresses harmonic components of the power system using an active filter.
[0016]
In FIG. 5, 1a is a high-order power system, 1b is a low-order power system to which the present invention is applied, 2 is a harmonic current generation source as a load of the low-order power system 1b, 4 is a voltage matching transformer, 5 is a current detector for detecting a current between the upper power system 1a and the lower power system 1b, 6 is a general load, 7 is an LC filter, 10 is an active filter, and 14 is an output current of the active filter 10. One of current detectors 20, 30, and 40 is an identification circuit for deriving the frequency characteristic of the lower power system 1b and transmitting the derived value to the active filter 10. The identification circuit 20 is an identification signal. command value [x ^* (t)] and the command to the active filter 10, as the output current of the active filter 10 is [x (t)] on the basis of this instruction, as shown in the aforementioned FIG. 4 (b) full It performs an operation of over chart.
[0017]
In the identification circuit 20, the frequency characteristic G (jω) of the power system 1b = [detection value y (t) of the current detector 5 / detection value x (t) of the current detector 14] is calculated.
[0018]
As is clear from FIG. 5, it is necessary to identify the frequency characteristics of the lower power system 1b in a state where all the devices other than the active filter 10 are operating. Since there is a harmonic current generation source 2, this harmonic current [h (t)] becomes a disturbance as shown in the block diagram of FIG. 6, and the identification value derived by the identification circuit 20 includes a disturbance error. Will be.
[0019]
That is, the detection value y (t) of the current detector 5 is equal to the detection value x (t) of the current detector 14 + the harmonic current h (t) of the harmonic current source 2 and the frequency characteristic G (jω) of the system. Will act.
Needless to say, the detection values y (t) and x (t) of the current detectors 5 and 14 indicate only harmonic current components.
[0020]
An object of the present invention is to provide a frequency characteristic identification method for a power system while extracting only the component based on the identification signal from detection values between a higher power system and a lower power system.
[0021]
[Means for Solving the Problems]
In the first invention, when identifying frequency characteristics of a power system based on an identification signal obtained by synthesizing a plurality of sine wave signals having different oscillation frequencies, the phase of each sine wave signal constituting the identification signal is determined. Is injected n times into the power system while being displaced by 2π / n [rad ] ( n: plural) from the previous time, and from the response of the power system based on the injected identification signal, n identification signals and responses the correlation function between, obtained the n an average value of the correlation function calculated the cross spectral density by Fourier transform, a value obtained by dividing calculating the cross spectral density self spectral density of the identification signal wherein the frequency A characteristic identification value is used.
[0022]
In the second invention, when identifying the frequency characteristics of the power system based on the identification signal obtained by synthesizing a plurality of sine wave signals having different oscillation frequencies, the amplitudes of the sine wave signals are made equal to each other. The wave signal is phase-shifted by 2π / n [rad ] ( n: plural) from the previous time and injected into the power system n times. From the response of the power system based on the injected identification signal, n identifications are made. A correlation function between a signal for use and a response is obtained, and a value obtained by Fourier-transforming the obtained average of n correlation functions is used as an identification value of the frequency characteristic.
[0023]
A third invention is the power converter according to the first or second invention, wherein the identification signal is based on an identification signal command value additionally commanded to a semiconductor power converter that is connected to a power system. It is generated by.
[0024]
In a fourth aspect based on the third aspect, the identification signal is an output current or output voltage of a semiconductor power converter, and the response is a current or voltage between a higher-order power system and the power system. The characteristic identification value is calculated.
[0025]
A fifth invention is the identification value of the frequency characteristic according to the third invention, wherein the identification signal is the identification signal command value, and the response is a current or voltage between a higher-order power system and the power system. Is calculated.
[0026]
The present invention has been made paying attention to the following actions.
[0027]
For example, if a cos waveform signal, one is Acos (ωt + φ ₁ ) and the other is Bcos (ωt + φ ₂ ), the correlation function between them is expressed by equation (3).
[0028]
[Equation 3]
(AB / 2) cos {ωt + (φ ₂ −φ ₁ )} (3)
That is, a phase difference component between the two appears in the correlation function.
[0029]
Therefore, when Acos (ωt + φ ₁ ) is a part of the sine wave signal constituting the identification signal and Bcos (ωt + φ ₂ ) is only the response, φ ₁ is 2π / n [ rad], (n: plural), and the [φ ₂ −φ ₁ ] values of the n correlation functions expressed by the above equation (3) obtained by injecting the power system while being displaced are the same. As described above, even if φ ₁ is displaced, the correlation function value corresponding to each identification signal does not change.
[0030]
However, among the responses including the component due to the harmonic current h (t) shown in FIG. 6, the response component due to h (t) has a fixed phase that is unrelated to the response component due to the identification signal. Therefore, [φ ₂ −φ ₁ ] in the correlation function shown in the above-described equation (3) is different when φ ₁ is displaced as described above, and the vector-like addition of each of the obtained correlation functions The average value becomes zero, and as a result, the response due to the h (t) can be eliminated.
[0031]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a flow chart showing a first embodiment of the present invention, and shows the operation of the identification circuit 30 shown in FIG.
[0032]
When the identification signal command value [x ^* (t)] is commanded to the active filter 10, the active filter 10 generates a current as the identification signal [x (t)] based on this command. At this time, assuming that x (t) is a combination of a plurality of cos wave currents, x (t) is expressed by Expression (4).
[0033]
[Expression 4]

[0034]
First, in step S11, the correlation function R _xx (τ) of the above equation (4) is calculated, and the calculation result is expressed by equation (5).
[0035]
[Equation 5]

[0036]
Next, in step S12, the self-spectral density S _xx (jω) obtained by Fourier transform of the equation (5) and represented by the equation (6) is obtained.
[0037]
[Formula 6]

[0038]
In step S13, θ ₁ , θ ₂ ,... Of x (t) shown in Expression (4) are in an initial value state (that is, n = n), and the lower power system 1b shown in FIG. Inject into.
[0039]
The harmonic current generating source 2 is operating in the power system 1b, and this harmonic current becomes a disturbance [h (t)] expressed by the following formula (7).
[0040]
[Expression 7]

[0041]
That is, [y (t)] detected in step S14 is expressed by equation (8).
[0042]
[Equation 8]

[0043]
Here, K _k is the gain of the power system 1b at frequency omega _k, [psi _k represents the phase delay at frequency omega _k, the first group on the right side shows the response by identification signal x (t), also on the right side The second group shows the response due to the disturbance h (t).
[0044]
In step S15, x (t) shown in the equation (4) and y (t) shown in the equation (8) are converted into a correlation function R _xy (t) according to the definition of the equation (2). It is calculated as shown in (9).
[0045]
[Equation 9]

[0046]
The operation from step S13 to step S15 is performed by the sequence of steps S16 and S17, and the phase of each sine wave signal constituting the identification signal x (t) is 2π / n [rad], (n: plural) from the previous time. ) And is injected into the power system while being displaced. Therefore, n correlation functions represented by the equation (9) based on the respective x (t) and y (t) are obtained, and the process proceeds to step S18.
[0047]
In step S18, an average value of n correlation functions obtained is obtained, and the bar of the average value R _xy (t) is expressed by equation (10).
[0048]
[Expression 10]

[0049]
As is apparent from this equation, the term for the disturbance h (t) is eliminated.
[0050]
In step S19, the equation (10) is Fourier transformed to obtain the cross spectral density S _xy (jω) represented by the equation (11).
[0051]
[Expression 11]

[0052]
Further, in step S20, the cross-spectral density obtained in step S19 is divided by the self-spectral density obtained in step 12, and the lower power system shown in FIG. The calculated value is the frequency characteristic G (jω) of 1b.
[0053]
[Expression 12]

[0054]
FIG. 2 is a flow chart showing a second embodiment of the present invention, and shows the operation of the identification circuit 40 shown in FIG.
[0055]
When the identification signal command value [x ^* (t)] is commanded to the active filter 10, the active filter 10 generates a current as the identification signal [x (t)] based on this command. At this time, when x (t) is a combination of a plurality of cos wave currents and normalized so that the amplitude of each of the cos wave currents is “1”, x (t) is expressed by Expression (13).
[0056]
[Formula 13]

[0057]
First, in step S31, θ ₁ , θ ₂ ,... Of x (t) shown in Expression (13) are in an initial value state (ie, n = n), and the lower power shown in FIG. Inject into system 1b.
[0058]
In this power system 1b, the harmonic current generation source 2 is operating, and this harmonic current becomes the disturbance [h (t)] expressed by the above equation (7).
[0059]
That is, [y (t)] detected in step S32 is expressed by equation (14).
[0060]
[Expression 14]

[0061]
Here, K _k is the gain of the power system 1b at frequency omega _k, [psi _k represents the phase delay at frequency omega _k, the first group on the right side shows the response by identification signal x (t), also on the right side The second group shows the response due to the disturbance h (t).
[0062]
In step S33, the correlation function R _xy (t) according to the definition of the equation (2) is converted from the x (t) expressed by the equation (13) and y (t) expressed by the equation (14). It is calculated as shown in (15).
[0063]
[Expression 15]

[0064]
The operation from step S31 to step S33 is performed by the sequence of step S34 and step S35, and the phase of each sine wave signal constituting the identification signal x (t) is 2π / n [rad], (n: plural) from the previous time. ) And is injected into the power system while being displaced. Therefore, n correlation functions represented by the equation (15) based on the respective x (t) and y (t) are obtained, and the process proceeds to step S36.
[0065]
In step S36, an average value of the n correlation functions obtained is obtained, and the bar of the average value R _xy (t) is expressed by equation (16).
[0066]
[Expression 16]

[0067]
As is apparent from this equation, the term for the disturbance h (t) is eliminated.
[0068]
In step S37, the equation (16) is Fourier-transformed, and in step S38, as shown in the equation (17), the calculated value of the frequency characteristic G (jω) of the lower power system 1b shown in FIG. .
[0069]
[Expression 17]

[0070]
FIG. 3 is an equipment configuration diagram in the power system showing the third embodiment of the present invention. In this embodiment, the difference from the configuration shown in FIG. 5 is that the identification circuit 40 or the identification circuit 40 is replaced with an identification circuit. The circuit 50 is provided.
[0071]
That is, the identification circuit 50 identifies the frequency characteristics of the lower power system 1a based on the identification signal command value [x ^* (t)] input to the active filter 10, and the calculation process at that time is as follows. The flowchart may be the same as that shown in FIG. 1 or FIG.
[0072]
In the above-described embodiment, an example in which a general current is used as the identification signal and the current of the power system is detected has been described, but the voltage is used as the identification signal to detect the voltage of the power system, You may perform the identification method of this invention.
[0073]
【The invention's effect】
According to the present invention, it is possible to extract only the component based on the identification signal based on the spectrum analysis method from the detection value between the upper power system and the lower power system, so that the frequency characteristic of the power system is identified. It is possible to eliminate the influence of the harmonic current source as a disturbance at the time .
[Brief description of the drawings]
FIG. 1 is a flow chart showing a first embodiment of the present invention. FIG. 2 is a flow chart showing a second embodiment of the present invention. FIG. 3 is an equipment configuration diagram showing the third embodiment of the present invention. ] Flow chart showing a conventional example [FIG. 5] Device configuration diagram of the power system [FIG. 6] Block diagram of FIG.
DESCRIPTION OF SYMBOLS 1a ... Upper power system, 1b ... Lower power system, 2 ... Harmonic current generation source, 4 ... Transformer, 5 ... Current detector, 6 ... Load, 7 ... LC filter, 10 ... Active filter, 14 ... Current Detector, 20, 30, 40, 50 ... identification circuit.

Claims (5)

When identifying the frequency characteristics of the power system based on the identification signal that combines multiple sinusoidal signals with different oscillation frequencies,
The phase of each sine wave signal constituting the identification signal is injected n times into the power system while shifting the phase by 2π / n [rad ] ( n: plural) from the previous time, and based on the injected identification signal From the power system response, obtain a correlation function between n identification signals and responses, and perform a Fourier transform on the obtained average value of the n correlation functions to obtain a cross spectral density. A frequency characteristic identification method for a power system, characterized in that a value obtained by dividing by a self-spectral density of a signal for use is used as an identification value of the frequency characteristic. When identifying the frequency characteristics of the power system based on the identification signal that combines multiple sinusoidal signals with different oscillation frequencies,
The amplitude of each sine wave signal is made equal to each other, and the phase of the sine wave signal is injected n times into the power system while being displaced by 2π / n [rad ] ( n: plural) from the previous time, and this injected identification A correlation function between n identification signals and responses is obtained from the response of the power system based on the signal for use, and a value obtained by performing Fourier transform on the obtained average value of the n correlation functions is obtained as an identification value of the frequency characteristic. A method for identifying frequency characteristics of an electric power system. In the power system frequency characteristic identification method according to claim 1 or 2,
The identification signal is generated by the power conversion device based on an identification signal command value additionally commanded to a semiconductor power conversion device interconnected with the power system. . In the frequency characteristic identification method of the electric power system according to claim 3,
The identification signal of the frequency characteristic is calculated by using the identification signal as an output current or output voltage of a semiconductor power converter, and using the response as a current or voltage between a higher-order power system and the power system. Frequency characteristic identification method for power system. In the frequency characteristic identification method of the electric power system according to claim 3,
The frequency of the power system, wherein the identification signal is used as the identification signal command value, and the identification value of the frequency characteristic is calculated using the response as a current or voltage between a higher-order power system and the power system. Characteristic identification method.

Images