Distributed state fusion estimation method of wind driven generator
阅读说明:本技术 一种风力发电机的分布式状态融合估计方法 (Distributed state fusion estimation method of wind driven generator ) 是由 陈博 杨盛伟 王如生 于 2020-06-30 设计创作,主要内容包括:一种针对噪声统计特性未知情况下风力发电机的分布式状态融合估计方法,提出双馈感应发电机的数学模型;建立发电机的状态空间模型;设计噪声统计特性未知的分布式融合估计方法;设置电机参数,求解两个优化问题得到局部最优增益以及分布式加权融合矩阵;迭代更新得到发电机状态的融合估计。本发明针对噪声统计特性无法准确获取时,提出了一种新的分布式融合估计方法来处理风力发电机的状态估计问题,且具有较高的估计精度。(A distributed state fusion estimation method for a wind driven generator under the condition of unknown noise statistical characteristics provides a mathematical model of a doubly-fed induction generator; establishing a state space model of the generator; designing a distributed fusion estimation method with unknown noise statistical characteristics; setting motor parameters, and solving two optimization problems to obtain local optimal gain and a distributed weighted fusion matrix; and carrying out iterative updating to obtain a fusion estimation of the generator state. The invention provides a novel distributed fusion estimation method for solving the problem of state estimation of a wind driven generator when the noise statistical characteristics cannot be accurately obtained, and the method has high estimation precision.)
1. A method for distributed state fusion estimation of a wind turbine, the method comprising the steps of:
the method comprises the following steps: a mathematical model of the doubly-fed induction generator is established, and the process is as follows:
the mathematical model of the doubly-fed induction generator is a high-order, nonlinear and strongly coupled multivariable system, and in order to establish the mathematical model, the following assumptions are made: 1.1. neglecting space harmonic wave, the magnetic potential is distributed along the circumference of the air gap according to sine; 1.2. neglecting magnetic circuit saturation, the self inductance and mutual inductance of each winding are linear; 1.3. the influence of frequency and temperature change on the resistance of the winding is not considered; 1.4. when the stator current flows to the motor, the current value is a negative value; 1.5. using a synchronous rotation two-phase dq coordinate system derivation equation, wherein the d axis and the q axis are different by 90 degrees, and under the assumption, a double-fed induction motor mathematical model is derived, and the expression is as follows:
ψds=-Lsids+Lmidr
ψdr=-Lsidr-Lmids
ψqr=-Lsiqr-Lmiqs
wherein idsAnd idrIs the component of the stator current and rotor current on the d-axis; i.e. iqsAnd iqrIs the component of the stator current and rotor current on the q-axis; v. ofdsAnd vdrIs the component of the stator voltage and the rotor voltage on the d-axis; v. ofqsAnd vqrIs the component of the stator voltage and the rotor voltage on the q-axis; omegab,ωs,ωrThe angular velocities of the base stage, the stator and the rotor respectively; l ismEquivalent mutual inductance coefficients of the stator and the rotor between the same phase winding axis of the generator are obtained; l iss,LrEquivalent inductances of the stator and the rotor in dq rotation coordinate system respectively; rs,RrRespectively the resistance of the stator and the rotor; psids,ψqs,ψdr,ψqrRespectively are flux linkages of the stator and the rotor under a dq coordinate system;
step two: establishing a state space model of the generator, wherein the process is as follows:
selecting the current x ═ i of the generatordsiqsidriqr]TAs the state variable, the voltage u ═ vdsvqsvdrvqr]TAs input variables, a linear state space model is obtained, the expression of which is
Where γ is a coefficient matrix of process noise; w (t) is unknown bounded process noise, wT(t)w(t)<Is unknown, according to a mathematical model of a doubly-fed induction generator, A and B are
Whereins=(ωs-ωr)/ωb;
Discretizing the linear state space model (1)
Wherein A isd=eAT,
in addition, the sensor measurement information includes stator voltage and rotor voltage, etc. for yi(t) is the measurement output of the ith sensor, and the observation equation is defined as
yi(t)=Cix(t)+γivi(t)(i=1,2,...,L) (3)
Wherein C isiIs the measurement matrix of the i-th sensor, gammaiCoefficient matrix of measurement noise, v, for sensor ii(t) is the unknown bounded process noise of sensor i,
step three: the distributed fusion estimation method comprises the following processes:
aiming at (2) - (3), a distributed fusion estimation method with unknown noise statistical characteristics is designed and comprises the following steps:
3.1, prediction: is provided withPredicting the status of the ith sensor
3.2, estimation: is provided with
3.3, fusion: is provided with
Wherein the optimum gain in (5)Distributed weighted fusion matrix omega in sum (6)i(t) (i ═ 1, 2.., L) was obtained by solving the following two convex optimization problems, first defined
Optimum gain
weighted fusion matrix omegai(t) (i ═ 1, 2.., L) was obtained by solving the following convex optimization problem:
wherein
Technical Field
The method mainly aims at the state monitoring of the wind driven generator, and estimates the state of the wind driven generator through a multi-sensor fusion technology.
Background
With the increasing severity of the environmental problems in the world, the international society has paid more and more attention to the problems of energy safety, ecological pollution, climate abnormality and the like, so that the use of fossil energy is reduced, and the development and utilization of renewable energy are accelerated, which has become a common consensus in all countries in the world. Among them, wind power generation plays an important role in the development of renewable energy, and has been applied on a large scale on a global scale. The wind power generation is a process of converting wind kinetic energy into mechanical kinetic energy and then converting the mechanical kinetic energy into electric energy, and the wind power generation has the working principle that the wind power is used for pushing a windmill blade to rotate, and then the speed of the windmill blade is increased through a speed increasing machine to drive a generator to generate electricity. Typically, various sensors are embedded within the wind turbine for measuring the condition of the electrical machine for monitoring and maintenance purposes.
Currently, there have been a number of research efforts to propose various estimation algorithms and models to estimate the state of the generator. The dynamic state estimation method based on Kalman filtering solves the problem of state estimation of the wind turbine model under the condition of unknown wind speed. And the extended Kalman filter adjusts a feedforward feedback optimal controller, and is used for tracking the power point of the permanent magnet synchronous generator under the interference of multiple wind speeds. It should be noted that kalman filter, extended kalman filter, unscented kalman filter, etc. are commonly used to process white noise with known covariance. However, in practical wind power systems, the generator may be subject to a variety of unknown disturbances, resulting in noise statistics that are not readily available with precision. Meanwhile, the estimation method only considers the situation of a single sensor, but with the development of information technology, the estimation precision of the single sensor cannot meet the requirement of the estimation precision of the system. Therefore, the technology provides a multi-sensor distributed fusion estimation method with unknown noise statistical characteristics, and the method is used for estimating the state of the wind driven generator.
Disclosure of Invention
In order to solve the problem that the noise statistical characteristics are known and improve the state estimation precision, the invention provides a distributed fusion estimation method with unknown noise statistical characteristics.
In order to achieve the purpose, the invention provides the following technical scheme:
a distributed state fusion estimation method of a wind driven generator comprises the following steps:
the method comprises the following steps: a mathematical model of the doubly-fed induction generator is established, and the process is as follows:
the mathematical model of the doubly-fed induction generator is a high-order, nonlinear and strongly coupled multivariable system, and in order to establish the mathematical model, the following assumptions are made: 1.1. neglecting space harmonic wave, the magnetic potential is distributed along the circumference of the air gap according to sine; 1.2. neglecting magnetic circuit saturation, the self inductance and mutual inductance of each winding are linear; 1.3. the influence of frequency and temperature change on the resistance of the winding is not considered; 1.4. when the stator current flows to the motor, the current value is a negative value; 1.5. using a synchronous rotation two-phase dq coordinate system derivation equation, wherein the d axis and the q axis are different by 90 degrees, and under the assumption, a double-fed induction motor mathematical model is derived, and the expression is as follows:
ψds=-Lsids+Lmidr
ψqs=-Lsiqs+Lmiqr
ψdr=-Lsidr-Lmids
ψqr=-Lsiqr-Lmiqs
wherein idsAnd idrIs the component of the stator current and rotor current on the d-axis; i.e. iqsAnd iqrIs the component of the stator current and rotor current on the q-axis; v. ofdsAnd vdrIs the component of the stator voltage and the rotor voltage on the d-axis; v. ofqsAnd vqrIs the component of the stator voltage and the rotor voltage on the q-axis; omegab,ωs,ωrThe angular velocities of the base stage, the stator and the rotor respectively; l ismFor the stator and rotor in a generatorEquivalent mutual inductance coefficients between the same phase winding axes; l iss,LrEquivalent inductances of the stator and the rotor in dq rotation coordinate system respectively; rs,RrRespectively the resistance of the stator and the rotor; psids,ψqs,ψdr,ψqrRespectively are flux linkages of the stator and the rotor under a dq coordinate system;
step two: establishing a state space model of the generator, wherein the process is as follows:
selecting the current x ═ i of the generatordsiqsidriqr]TAs the state variable, the voltage u ═ vdsvqsvdrvqr]TAs input variables, a linear state space model is obtained, the expression of which is
Where γ is a coefficient matrix of process noise; w (t) is unknown bounded process noise, wT(t)w(t)<Is unknown, according to a mathematical model of a doubly-fed induction generator, A and B are
Whereins=(ωs-ωr)/ωb;
Discretizing the linear state space model (1)
x(t+1)=Adx(t)+Bdu(t)+γw(t) (2)
WhereinT is a sampling period;
in addition, the sensor measuresThe information includes stator voltage, rotor voltage, etc. let yi(t) is the measurement output of the ith sensor, and the observation equation is defined as
yi(t)=Cix(t)+γivi(t)(i=1,2,...,L) (3)
Wherein C isiIs the measurement matrix of the i-th sensor, gammaiCoefficient matrix of measurement noise, v, for sensor ii(t) is the unknown bounded process noise of sensor i,κ is unknown; each sensor can send the measured data to a control center in real time so that maintenance personnel can monitor the state of the generator remotely;
step three: the distributed fusion estimation method comprises the following processes:
aiming at (2) - (3), a distributed fusion estimation method with unknown noise statistical characteristics is designed and comprises the following steps:
3.1, prediction: is provided withPredicting the status of the ith sensor
3.2, estimation: is provided withEstimating the state of the ith sensor
3.3, fusion: is provided with
Estimating for fusion states
Wherein the optimum gain in (5)Distributed weighted fusion matrix omega in sum (6)i(t) (i ═ 1, 2.., L) was obtained by solving the following two convex optimization problems, first defined
Optimum gainObtained by solving the following convex optimization problem:
weighted fusion matrix omegai(t) (
wherein
The invention has the following beneficial effects: a distributed fusion estimation method of unknown noise statistical information is provided for a state model of a wind driven generator. By solving the two convex optimization problems, the optimal gain and weighting fusion matrix can be obtained, and compared with the existing estimation method, the method can better estimate the state of the wind driven generator under the condition that the noise statistical information cannot be accurately obtained, so that a worker can accurately monitor and maintain the generator.
Drawings
FIG. 1 is a flow chart of an embodiment of the method of the present invention.
Fig. 2 is a block diagram of a distributed fusion architecture.
FIG. 3 is a diagram of a true state trajectory and a fused estimated state trajectory of a wind turbine.
Fig. 4 shows the distributed fusion estimation error.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1 to 4, a method for estimating distributed state fusion of a wind turbine includes the following steps:
the method comprises the following steps: a mathematical model of the doubly-fed induction generator is established, and the process is as follows:
the mathematical model of the doubly-fed induction generator is a high-order, nonlinear and strongly coupled multivariable system, and in order to establish the mathematical model, the following assumptions are made: 1.1. neglecting space harmonic wave, the magnetic potential is distributed along the circumference of the air gap according to sine; 1.2. neglecting magnetic circuit saturation, the self inductance and mutual inductance of each winding are linear; 1.3. the influence of frequency and temperature change on the resistance of the winding is not considered; 1.4. when the stator current flows to the motor, the current value is a negative value; 1.5. the equation is derived using a synchronously rotating two-phase dq coordinate system, with the d-axis and q-axis being 90 ° apart. Under the assumption, a mathematical model of the doubly-fed induction motor is developed, and the expression is as follows:
ψds=-Lsids+Lmidr
ψqs=-Lsiqs+Lmiqr
ψdr=-Lsidr-Lmids
ψqr=-Lsiqr-Lmiqs
wherein idsAnd idrIs the component of the stator current and rotor current on the d-axis; i.e. iqsAnd iqrIs the component of the stator current and rotor current on the q-axis; v. ofdsAnd vdrIs the component of the stator voltage and the rotor voltage on the d-axis; v. ofqsAnd vqrIs the component of the stator voltage and the rotor voltage on the q-axis; omegab,ωs,ωrThe angular velocities of the base stage, the stator and the rotor respectively; l ismEquivalent mutual inductance coefficients of the stator and the rotor between the same phase winding axis of the generator are obtained; l iss,LrEquivalent inductances of the stator and the rotor in dq rotation coordinate system respectively; rs,RrRespectively the resistance of the stator and the rotor; psids,ψqs,ψdr,ψqrRespectively are flux linkages of the stator and the rotor under a dq coordinate system;
step two: establishing a state space model of the generator, wherein the process is as follows:
selecting the current x ═ i of the generatordsiqsidriqr]TAs the state variable, the voltage u ═ vdsvqsvdrvqr]TAs input variables, a linear state space model is obtained, the expression of which is
Where γ is a coefficient matrix of process noise; w (t) is unknown bounded process noise, wT(t)w(t)<Is unknown, according to a mathematical model of a doubly-fed induction generator, A and B are
Whereins=(ωs-ωr)/ωb;
Discretizing the linear state space model (1)
x(t+1)=Adx(t)+Bdu(t)+γw(t) (2)
Wherein A isd=eAT,
T is a sampling period;in addition, the sensor measurement information includes stator voltage and rotor voltage, etc. for yi(t) is the measurement output of the ith sensor, and the observation equation is defined as
yi(t)=Cix(t)+γivi(t)(i=1,2,...,L) (3)
Wherein C isiIs the measurement matrix of the i-th sensor, gammaiCoefficient matrix of measurement noise, v, for sensor ii(t) is the unknown bounded process noise of sensor i,
κ is unknown; each sensor can send the measured data to a control center in real time so that maintenance personnel can monitor the state of the generator remotely;step three: the distributed fusion estimation method comprises the following processes:
aiming at (2) - (3), a distributed fusion estimation method with unknown noise statistical characteristics is designed and comprises the following steps:
3.1, prediction: is provided withPredicting the status of the ith sensor
3.2, estimation: is provided with
Estimating the state of the ith sensor
3.3, fusion: is provided withEstimating for fusion states
Wherein the optimum gain in (5)
Distributed weighted fusion matrix omega in sum (6)i(t) (i ═ 1, 2.., L) was obtained by solving the following two convex optimization problems, first defined
Optimum gainObtained by solving the following convex optimization problem:
weighted fusion matrix omegai(t) (i ═ 1, 2.., L) was obtained by solving the following convex optimization problem:
wherein
To verify the effectiveness of the method designed by the present invention, the following example was used for verification.
Setting basic parameters of the motor: f. ofb=50Hz,fs=50Hz,fr=40Hz,Rs=0.004Ω,Rr=0.005Ω,Ls=0.09Ω,Lr=0.08Ω,LbSubstituting 3.95 omega into the state space model of the wind driven generator. The input parameters are set as follows: stator d-axis voltage vds0.02V, q-axis voltage Vqs0.99V; rotor d-axis voltage vdr0.02V, q-axis voltage Vdr0.206V. Four sensors are adopted to measure four current components of the generator, and the measurement matrixes are respectively C1=[1 0 0 0]、C2=[0 1 0 0]、C3=[0 0 1 0]、C4=[0 0 0 1]. The coefficient matrixes of the process noise and the observation noise are unit matrixes, and the assumed bounded noise is
Wherein p is1(t)∈[0,1],p2(t)∈[0,1],p3(t)∈[0,1],p4(t)∈[0,1],pv1(t)∈[0,1],pv2(t)∈[0,1],pv3(t)∈[0,1],pv4(t)∈[0,1]
Solving (8) by LMI toolkit to obtain optimal gainSolving (9) to obtain a weighted fusion matrix omegai(t) (i ═ 1, 2.., L), according to solutionAnd Ωi(t) (i ═ 1, 2.., L), local state estimates can be calculated by combining equations (5) and (6)And fusion of the estimates
The real track and the distributed fusion estimation track of the drawing system are shown in fig. 3, and it can be seen from the figure that the fusion estimation method provided by the invention can better estimate the state of the starting motor. In addition, the error of the actual state and the fusion estimation state is calculated, as shown in fig. 4, the estimation error is small, which shows that the fusion estimation algorithm of the present invention has high estimation accuracy. The experimental result shows that when the statistical characteristics of the noise cannot be accurately obtained, the state of the wind driven generator can be accurately estimated by the method. Therefore, the distributed fusion estimation method designed by the invention is more suitable for the actual wind power generation system.