CN111614086A - Filtering estimation and prediction estimation method for multi-state variables of power system - Google Patents

Filtering estimation and prediction estimation method for multi-state variables of power system Download PDF

Info

Publication number
CN111614086A
CN111614086A CN202010522875.3A CN202010522875A CN111614086A CN 111614086 A CN111614086 A CN 111614086A CN 202010522875 A CN202010522875 A CN 202010522875A CN 111614086 A CN111614086 A CN 111614086A
Authority
CN
China
Prior art keywords
estimation
state
model
time
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202010522875.3A
Other languages
Chinese (zh)
Inventor
李晓
张竹青
张玉华
孟华
田彦彦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Henan Boteli Intellectual Property Services Co ltd
Original Assignee
Henan Boteli Intellectual Property Services Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Henan Boteli Intellectual Property Services Co ltd filed Critical Henan Boteli Intellectual Property Services Co ltd
Priority to CN202010522875.3A priority Critical patent/CN111614086A/en
Publication of CN111614086A publication Critical patent/CN111614086A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • H02J3/003Load forecast, e.g. methods or systems for forecasting future load demand
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06Energy or water supply
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • H02J3/004Generation forecast, e.g. methods or systems for forecasting future energy generation
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28Arrangements for balancing of the load in a network by storage of energy
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/381Dispersed generators
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2300/00Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/20The dispersed energy generation being of renewable origin
    • H02J2300/22The renewable source being solar energy
    • H02J2300/24The renewable source being solar energy of photovoltaic origin
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02EREDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00Energy generation through renewable energy sources
    • Y02E10/50Photovoltaic [PV] energy
    • Y02E10/56Power conversion systems, e.g. maximum power point trackers

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Economics (AREA)
  • Power Engineering (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Water Supply & Treatment (AREA)
  • Public Health (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Feedback Control In General (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the technical field of filtering estimation and prediction estimation of state variables and discloses a filtering estimation and prediction estimation method for multi-state variables of a power system, which comprises the following steps: step 1: establishing a time point estimation method based on a Kalman filter; step 2: establishing a Kalman filter-based vector block state estimation method; and step 3: establishing block state real-time smooth estimation, filtering estimation and prediction estimation based on Kalman filtering; the model established by the method is more reasonable, accords with the actual situation, has higher actual use value, and effectively improves the working stability of the distributed power grid; the goal of minimizing the estimated error is achieved.

Description

Filtering estimation and prediction estimation method for multi-state variables of power system
Technical Field
The invention relates to the technical field of filtering estimation and prediction estimation of state variables, in particular to a filtering estimation and prediction estimation method for multi-state variables of a power system.
Background
Energy is an indispensable factor on which humans depend for survival and development. However, with global population growth and rapid economic development, humans are experiencing an unprecedented energy and environmental crisis. Because the traditional energy can not meet the urgent needs of human society and economic development, the solar photovoltaic power generation is taken as a clean and renewable new energy source with huge energy in the same living, and is gradually favored by people after wind power generation. The reason for this is that it is inexhaustible, safe and pollution-free, and is not limited by the resource distribution region. The sunlight irradiates the ground generally, solar energy is available everywhere, solar photovoltaic power generation is particularly suitable for remote areas without electricity, and the construction of a long-distance power grid and the electric energy loss on a power transmission line can be reduced.
The existing microgrid system in China mainly takes solar energy and wind energy as main materials, strong uncertainty exists, and the electricity consumption of a user also has certain uncertainty, and the uncertainty can be described by using a non-stationary random process. Particularly, in a microgrid system composed of distributed photovoltaic power generation and multiple loads, the changes of the power generation state, the energy storage state and the multiple loads are non-stationary processes. Therefore, for real-time estimation, especially prediction estimation, of these variables, a non-stationary model of multiple input multiple output should be constructed. However, most of the prediction methods commonly used in the prior art are methods based on statistical regression, and not only lack a state model, but also lack real-time recursion. Especially, when the non-stationarity is strong, the estimation and prediction effects are poor.
The problem is addressed herein by means of Kalman filtering with powerful capability to handle non-stationary processes. Kalman filtering is an optimal recursive filtering method proposed by american scholars Kalman (r.e.kalman) and buche (r.s.bucy) in 1960. The method considers the statistical characteristics of the measured value and the observed value, and is not only suitable for a stable random process, but also more importantly suitable for a non-stable random process. Kalman filtering is a method for optimally estimating the state of a system by using a linear system state equation and inputting and outputting observation data through the system. Kalman filtering does not require that both signal and noise are assumptions of a stationary process. The linear Kalman filtering is a Chinese linear optimal filter and has the excellent characteristics of real-time recursive estimation under the condition that modeling errors are white noise and performance indexes based on mean square errors. Therefore, since the Kalman filtering theory appeared, it has been applied to many departments such as communication systems, power systems, aerospace, environmental pollution control, industrial control, radar signal processing, etc., and has achieved many successful results. Point-to-point smoothing estimation, filtering estimation and prediction estimation can be realized by using Kalman filtering. Because the power utilization and power generation processes of the power system have strong non-stationary characteristics and obvious periodicity, the state variables in the period have strong coupling and relevance. In this project, the period is described as a state block, which may be a day, a week, a month or even a year, for which block state estimation is performed, which is a semi-real time estimation process. In order to achieve real-time estimation, a measurement equation is skillfully rewritten so as to realize real-time smoothing, filtering and prediction estimation of each state in a block.
Kalman filtering is an optimal recursive filtering method proposed by american scholars Kalman (r.e.kalman) and buche (r.s.bucy) in 1960. The method considers the statistical characteristics of the measured value and the observed value, and is not only suitable for a stable random process, but also more importantly suitable for a non-stable random process. Kalman filtering is a method for optimally estimating the state of a system by using a linear system state equation and inputting and outputting observation data through the system. Kalman filtering does not require that both signal and noise are assumptions of a stationary process. The linear Kalman filtering is a Chinese linear optimal filter and has the excellent characteristics of real-time recursive estimation under the condition that modeling errors are white noise and performance indexes based on mean square errors. Therefore, since the Kalman filtering theory appeared, it has been applied to many departments such as communication systems, power systems, aerospace, environmental pollution control, industrial control, radar signal processing, etc., and has achieved many successful results. Because the power utilization and power generation processes of the power system have strong non-stationary characteristics and obvious periodicity, the state variables in the period have strong coupling and relevance. In view of the excellent characteristics of Kalman filtering in the flight and stationary process, how to provide a Kalman filtering-based filtering estimation and prediction estimation method for multi-state variables of a power system is a technical problem to be solved by technical personnel in the field.
Disclosure of Invention
The invention provides a filtering estimation and prediction estimation method for multi-state variables of a power system, aiming at the problem that the estimation and prediction effects in the non-stationary process are poor in a micro-grid system composed of distributed photovoltaic power generation and multiple loads in the prior art.
In order to solve the technical problems, the invention adopts the following technical scheme:
a filtering estimation and prediction estimation method for multi-state variables of a power system is designed, and comprises the following steps:
step 1: time point estimation method based on Kalman filter
1.1, establishing a time series dynamic model of the real-time change of the multi-state vector along with a time point as a state equation of a Kalman filter, and taking the real-time change of the multi-state variable along with the time point in the power system as an input quantity:
x(k+1)=A(k+1,k)x(k)+w(k) (1)
wherein, in the formula (1), x (k) is an observation vector, a (k +1, k) is a state transition matrix, and w (k) is modeling noise; the corresponding observation equation is shown in equation (2):
y(k+1)=C(k+1)x(k+1)+ν(k+1) (2)
wherein, C (k) in the formula (2) is an observation matrix, and v (k) is observation noise;
1.2 establishing Kalman Filter based on time point estimation method
1.2.1 first, it is assumed that an optimal estimate for the kth state x (k) has been obtained
Figure BDA0002537236320000031
And the corresponding estimation error covariance matrix P (k | k):
Figure BDA0002537236320000032
Figure BDA0002537236320000033
1.2.2 then, based on
Figure BDA0002537236320000034
And P (k | k), and further obtaining a predicted value
Figure BDA0002537236320000035
And prediction estimation error covariance P (k +1| k):
Figure BDA0002537236320000041
P(k+1|k)=A(k+1,k)P(k|k)AT(k+1,k)+Qw(k) (6)
1.2.3 next, the optimal gain matrix K (K +1) is calculated:
K(k+1)=A(k+1,k)P(k+1,k)HT(k+1)[H(k+1)P(k+1,k)HT(k+1)]-1(7)
1.2.4 obtaining the optimal estimated value of the state x (k +1)
Figure BDA0002537236320000042
And estimated error covariance P (k +1| k + 1):
Figure BDA0002537236320000043
Figure BDA0002537236320000044
P(k+1|k+1)=[I-K(k+1)H(k+1)P(k+1,k)](10)
1.2.5, continuing to loop the process of 1.2.2-1.2.4 until the filtering is finished;
step 2: method for establishing vector block state estimation based on Kalman filter
2.1 firstly, regarding the whole variable in a time period as a vector block, establishing a dynamic system model of the multi-state vector block changing in time period as a state equation of a Kalman filter, and regarding the real-time change quantity of the multi-state vector block along with a time point in the power system as an input quantity:
X(m+1)=A(m+1,M)X(m,M)+W(m)w(m,M) (11)
wherein, in the formula (11), X (M, M) is an observation vector, a (M +1, M) is a state matrix, and w (M) w (M, M) is modeling noise; the corresponding observation equation is shown in equation (12):
Y(m+1)=HX(m+1)+V(m+1) (12)
wherein, in the formula (12), H is an observation matrix, and V (m +1) is observation noise;
the Kalman filter model as shown in equations (3) - (10) can obtain the estimated value of each state vector block
Figure BDA0002537236320000045
And an estimation error covariance matrix P (m + m):
Figure BDA0002537236320000051
Figure BDA0002537236320000052
2.2 establishing Kalman Filter model based on vector Block State estimation method
2.2.1 first, assume that the optimal estimate of the m-th vector chunk state has been obtained
Figure BDA0002537236320000053
And corresponding estimation error covariance matrix
Figure BDA0002537236320000054
Figure BDA0002537236320000055
Figure BDA0002537236320000056
2.2.2 based on
Figure BDA0002537236320000057
And
Figure BDA0002537236320000058
further obtaining the predicted value
Figure BDA0002537236320000059
And prediction estimation error covariance
Figure BDA00025372363200000510
Figure BDA00025372363200000511
Figure BDA00025372363200000512
2.2.3 calculate optimal gain array:
Figure BDA00025372363200000513
2.2.4 obtaining the optimal estimation value of the state block X (m +1)
Figure BDA00025372363200000514
And estimate error covariance
Figure BDA00025372363200000515
Figure BDA00025372363200000516
Figure BDA00025372363200000517
P(m+1|m+1)=[I-K(m+1)H(m+1)P(m+1,m)](22)
2.2.5, continuing to loop the process of 2.2.2-2.2.4 until the filtering is finished;
and step 3: establishing Kalman filtering based block state real-time smooth estimation, filtering estimation and prediction estimation
3.1 first, formula (12) in step 2 is rewritten as follows:
y(m+1,k)=H(m+1,k)X(m+1)+ν(m+1,k),m=1,2,…;k=1,2,…,M (23)
3.2, a real-time smooth estimation model is established, and real-time updating estimation is carried out on states x (m +1,1), x (m +1,2), L and x (m +1, k-1) before the current time k, wherein the model is as follows:
Figure BDA0002537236320000061
3.3, establishing a real-time filtering estimation model, and updating and estimating the state x (m +1, k) at the current k moment in real time, wherein the model is as follows:
Figure BDA0002537236320000062
3.4 establishing a real-time prediction estimation model for the state after the current k moment
x (M +1, k +1), x (M +1, k +2), L, x (M +1, M) are used for real-time prediction estimation, and the model is as follows:
Figure BDA0002537236320000063
3.5, taking a formula (11) as a state equation, and taking a formula (23) as a state model of a measurement equation to carry out a Kalman filter model;
3.6 building a Kalman Filter model in the Multi-State vector Block
3.6.1 first, assume that the optimal estimate for the mth time instant in the mth state block X (M) is known
Figure BDA0002537236320000064
Sum estimation error covariance matrix
Figure BDA0002537236320000065
Figure BDA0002537236320000066
Figure BDA0002537236320000067
3.6.2 based on
Figure BDA0002537236320000068
And
Figure BDA0002537236320000069
further obtaining a prediction value and a prediction estimation error covariance of the (m +1) th block state:
Figure BDA00025372363200000610
Figure BDA0002537236320000071
3.6.3 further results in an intra-block recursive filter:
Figure BDA0002537236320000072
3.6.4 loops through the 3.6.2-3.6.3 process until the filtering is complete.
Further, the modeling noise w (k) and the measurement noise v (k) satisfy the following statistical characteristics:
Figure BDA0002537236320000073
wherein Q (k) and R (k) are both error variance matrices, and the error variance matrices Q (k) and R (k) are symmetric non-negative definite matrices and symmetric positive definite matrices, respectively, S (k) ∈ Rn×m
Further, the above formula (11) and formula (12)
Figure BDA0002537236320000074
Figure BDA0002537236320000075
V(m+1)~N[0,R(m+1)],
R(m+1)=diag{R(m+1,1),R(m+1,2),…,R(m+1,M)},
Figure BDA0002537236320000076
Figure BDA0002537236320000081
Further, the measurement noise v (m, k) in the formula (23) has a statistical characteristic E { ν (m, k) } of 0; e { v (m, k) vT(m,j)}=R(m,k)kjm=1,2,…,k,j=1,2,…,M。
Further, in the formula (39)
Figure BDA0002537236320000082
Figure BDA0002537236320000083
Figure BDA0002537236320000084
Further, before the first step, a sensor model of each state variable needs to be established:
yi(t)=hi(x(t))+νi(t),i=1,2,…,N (40)
where N represents the number of sensors, X (t) represents a model of a state variable, hi(. cndot.) is a continuous function;
then, after the equation (40) is discretized by the equal period T, a discrete observation equation is obtained, wherein the discrete observation equation is
yi(kT)=hi(x(kT),kT)+νi(kT),i=1,2,…,N (41)
Wherein h isiDenotes the measurement function of the corresponding sensor i, determined by the properties of the sensor, vi(kT) denotes measurement error; next, the following sequence model of the state variable x (kT) with time is established
x((k+1)T)=f(x(kT),kT)+w(kT) (42)
Where f (x), (k), k ═ a (k +1, k) x (k) is a state transfer function, and w (kt) represents a modeling error.
Further, the multi-state variables include state variables of the distributed power generation units and state variables of each load of the power utilization.
Further, the time period in step 2 includes one hour, one day, one week, one month, or one year.
The invention provides a filtering estimation and prediction estimation method for multi-state variables of a power system, which has the beneficial effects that:
(1) the method comprises the steps of constructing a sequence model of each state variable changing along with time, considering the non-stationarity characteristic of the constructed model, converting the sequence model into a Kalman filtering framework for processing, and respectively establishing a real-time point estimation method of the state, a semi-real-time estimation method of the block state, a real-time block smoothing estimation method, a block filtering estimation method, a block prediction estimation method and other estimation methods by establishing a point filtering model, a block state model and a prediction model of relevant state variables; the method has strong applicability to non-stationary random processes, and effectively solves the problem of poor prediction results in the prior art;
(2) the method comprises the steps of constructing a recursion model of state variables changing along with time on the basis of a Kalman filtering framework, considering the specific characteristics of a non-stable process of the microgrid, constructing a distributed multi-load unit, considering the input and output dynamic models of natural factors such as temperature, humidity, wind speed and illumination radiation, analyzing the dynamic characteristics of the multi-load under different properties and the statistical characteristics of data generated by the multi-load change of an external unit, and constructing a prediction control model according to the statistical characteristics; compared with the prior art, the prediction estimation method considers more input and output influence factors, and the established model is more reasonable, conforms to the actual situation, has higher actual use value, and effectively improves the working stability of the distributed power grid;
(3) according to the method, the characteristics of minimized estimated error and strong robustness of estimated estimation are realized during the prediction processing process according to the characteristics of the non-stationary process of the microgrid.
Drawings
The invention will be further described in detail with reference to examples of embodiments shown in the drawings to which, however, the invention is not restricted.
FIG. 1 is a graph showing the comparison of predicted temperature and actual temperature in an embodiment of the present invention;
FIG. 2 is an error curve of the comparison of the predicted temperature and the actual temperature in the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
The invention relates to a filtering estimation and prediction estimation method for multi-state variables of a power system, which comprises the following steps:
in this embodiment, temperature variation prediction is used as an input quantity of the method, and selected data is derived from historical data (source website: https:// close. dest.com.cn /) published by a meteorological station of qinghua university in 2017 in month 6, wherein a state transition matrix a (k +1, k) is shown in table 1, an initial state X (0) is 20.3, an initial estimation error covariance p (0) is 1, a modeling error w (k) is subject to normal distribution with a mean value of 0 and a standard deviation of 0.08, and an observation noise v (k) is subject to normal distribution with a mean value of 0 and a standard deviation of 0.2.
TABLE 1 State transition matrix A (k +1, k)
A(2,1) 0.9557 A(10,9) 1.076 A(18,17) 0.9869
A(3,2) 1.0103 A(11,10) 1.1576 A(19,18) 0.9767
A(4,3) 0.9643 A(12,11) 1.169 A(20,19) 0.9694
A(5,4) 0.9471 A(13,12) 1.1004 A(21,20) 0.9614
A(6,5) 0.9609 A(14,13) 1.0401 A(22,21) 0.9708
A(7,6) 0.9767 A(15,14) 1.0281 A(23,22) 0.985
A(8,7) 0.9821 A(16,15) 1.0273 A(24,23) 0.9656
A(9,8) 1.0364 A(17,16) 1.0133 A(25,24) 0.996
Firstly, establishing a sensor model of each state variable:
yi(t)=hi(x(t))+νi(t),i=1,2,…,N (40)
wherein N represents the number of temperature sensors, X (t) represents a model of a temperature state variable, hi(. cndot.) is a continuous function;
then, after the equation (40) is discretized by the equal period T, a discrete observation equation is obtained, wherein the discrete observation equation is
yi(kT)=hi(x(kT),kT)+νi(kT),i=1,2,…,N (41)
Wherein h isiDenotes the measurement function of the corresponding temperature sensor i, which is determined by the properties of the temperature sensor vi(kT) denotes measurement error; next, the following sequence model of the state variable x (kT) with time is established
x((k+1)T)=f(x(kT),kT)+w(kT) (42)
Where f (x), (k), k ═ a (k +1, k) x (k) is a state transfer function, and w (kt) represents a modeling error.
Step 1: time point estimation method based on Kalman filter
1.1, establishing a time series dynamic model of the real-time change of the multi-state temperature vector along with the time point as a state equation of a Kalman filter, and taking the real-time change of the multi-state temperature variable along with the time point in the power system as an input quantity:
x(k+1)=A(k+1,k)x(k)+w(k) (1)
wherein, in the formula (1), x (k) is an observation vector, a (k +1, k) is a state transition matrix, and w (k) is modeling noise; the corresponding observation equation is shown in equation (2):
y(k+1)=C(k+1)x(k+1)+ν(k+1) (2)
wherein, C (k) in the formula (2) is an observation matrix, and v (k) is observation noise; the modeling noise w (k) and the measurement noise v (k) meet the following statistical characteristics:
Figure BDA0002537236320000111
wherein Q (k) and R (k) are both error variance matrices, and the error variance matrices Q (k) and R (k) are symmetric non-negative definite matrices and symmetric positive definite matrices, respectively, S (k) ∈ Rn ×m
1.2 establishing Kalman Filter based on time point estimation method
1.2.1 first, it is assumed that an optimal estimate for the kth state x (k) has been obtained
Figure BDA0002537236320000129
And the corresponding estimation error covariance matrix P (k | k):
Figure BDA0002537236320000121
Figure BDA0002537236320000122
1.2.2 then, based on
Figure BDA0002537236320000123
And P (k | k), and further obtaining a predicted value
Figure BDA0002537236320000124
And prediction estimation error covariance P (k +1| k):
Figure BDA0002537236320000125
P(k+1|k)=A(k+1,k)P(k|k)AT(k+1,k)+Qw(k) (6)
1.2.3 next, the optimal gain matrix K (K +1) is calculated:
K(k+1)=A(k+1,k)P(k+1,k)HT(k+1)[H(k+1)P(k+1,k)HT(k+1)]-1(7)
1.2.4 obtaining the optimal estimated value of the state x (k +1)
Figure BDA0002537236320000126
And estimated error covariance P (k +1| k + 1):
Figure BDA0002537236320000127
Figure BDA0002537236320000128
P(k+1|k+1)=[I-K(k+1)H(k+1)P(k+1,k)](10)
1.2.5, continuing to loop the process of 1.2.2-1.2.4 until the filtering is finished;
step 2: method for establishing temperature vector block state estimation based on Kalman filter
2.1, firstly, regarding the temperature variable in a time period as a vector block as a whole, establishing a dynamic system model of a multi-state vector block for time period change as a state equation of a Kalman filter, wherein the multi-state variable comprises a state variable of a distributed power generation unit and a state variable of each load of electricity consumption, and in the embodiment, the multi-state variable refers to a multi-state temperature variable comprising a temperature state quantity of a distributed power generation power supply and a temperature state variable of each load of electricity consumption; the time period in step 2 includes one day, one week, one month or one year, and the time period selected in this embodiment is one hour.
Taking the real-time change quantity of the multi-state vector blocks in the power system along with the time points as input quantity:
X(m+1)=A(m+1,M)X(m,M)+W(m)w(m,M) (11)
wherein, in the formula (11), X (M, M) is an observation vector, a (M +1, M) is a state matrix, and w (M) w (M, M) is modeling noise; the corresponding observation equation is shown in equation (12):
Y(m+1)=HX(m+1)+V(m+1) (12)
wherein, in the formula (12), H is an observation matrix, and V (m +1) is observation noise; in the formula (11) and the formula (12):
Figure BDA0002537236320000131
Figure BDA0002537236320000132
V(m+1)~N[0,R(m+1)],
R(m+1)=diag{R(m+1,1),R(m+1,2),…,R(m+1,M)},
Figure BDA0002537236320000133
Figure BDA0002537236320000134
the Kalman filter model as shown in equations (3) - (10) can obtain the estimated value of each state vector block
Figure BDA0002537236320000141
And the estimation error covariance matrix P (m | m):
Figure BDA0002537236320000142
Figure BDA0002537236320000143
2.2 establishing Kalman Filter model based on vector Block State estimation method
2.2.1 first, assume that the optimal estimate of the m-th vector chunk state has been obtained
Figure BDA0002537236320000144
And corresponding estimation error covariance matrix
Figure BDA0002537236320000145
Figure BDA0002537236320000146
Figure BDA0002537236320000147
2.2.2 based on
Figure BDA0002537236320000148
And
Figure BDA0002537236320000149
further obtaining the predicted value
Figure BDA00025372363200001410
And prediction estimation error covariance
Figure BDA00025372363200001411
Figure BDA00025372363200001412
Figure BDA00025372363200001413
2.2.3 calculate optimal gain array:
Figure BDA00025372363200001414
2.2.4 obtaining the optimal estimation value of the state block X (m +1)
Figure BDA00025372363200001415
And estimate error covariance
Figure BDA00025372363200001416
Figure BDA00025372363200001417
Figure BDA00025372363200001418
P(m+1|m+1)=[I-K(m+1)H(m+1)P(m+1,m)](22)
2.2.5, continuing to loop the process of 2.2.2-2.2.4 until the filtering is finished;
and step 3: establishing Kalman filtering based block state real-time smooth estimation, filtering estimation and prediction estimation
3.1 first, formula (12) in step 2 is rewritten as follows:
y(m+1,k)=H(m+1,k)X(m+1)+ν(m+1,k),m=1,2,…;k=1,2,…,M (23)
the measurement noise v (m, k) in the formula (23) has a statistical characteristic E { ν (m, k) } of 0; e { v (m, k) vT(m,j)}=R(m,k)kjm=1,2,…,k,j=1,2,…,M。
3.2, a real-time smooth estimation model is established, and real-time updating estimation is carried out on states x (m +1,1), x (m +1,2), L and x (m +1, k-1) before the current time k, wherein the model is as follows:
Figure BDA0002537236320000151
3.3, establishing a real-time filtering estimation model, and updating and estimating the state x (m +1, k) at the current k moment in real time, wherein the model is as follows:
Figure BDA0002537236320000152
3.4 establishing a real-time prediction estimation model for the state after the current k moment
x (M +1, k +1), x (M +1, k +2), L, x (M +1, M) are used for real-time prediction estimation, and the model is as follows:
Figure BDA0002537236320000153
3.5, taking a formula (11) as a state equation, and taking a formula (23) as a state model of a measurement equation to carry out a Kalman filter model;
3.6 building a Kalman Filter model in the Multi-State vector Block
3.6.1 first, assume that the optimal estimate for the mth time instant in the mth state block X (M) is known
Figure BDA0002537236320000154
Sum estimation error covariance matrix
Figure BDA0002537236320000155
Figure BDA0002537236320000156
Figure BDA0002537236320000161
3.6.2 based on
Figure BDA0002537236320000162
And
Figure BDA0002537236320000163
further obtaining a prediction value and a prediction estimation error covariance of the (m +1) th block state:
Figure BDA0002537236320000164
Figure BDA0002537236320000165
3.6.3 further results in an intra-block recursive filter:
Figure BDA0002537236320000166
in the formula (39):
Figure BDA0002537236320000167
Figure BDA0002537236320000168
Figure BDA0002537236320000169
3.6.4 loops through the 3.6.2-3.6.3 process until the filtering is complete.
The prediction results obtained in the above embodiment are shown in fig. 1 and attached table 2; 1,8,16,24 indicate the estimation and prediction of the state from the first, eighth, sixteenth and twenty-fourth points, respectively.
TABLE 2 estimation and prediction values from points
Figure BDA00025372363200001610
Figure BDA0002537236320000171
According to the prediction and the real results of the attached figures 1 and the table 2, the relative error between the real value and the predicted value of each point is calculated, and an error curve of the attached figure 2 and a prediction error numerical value of the table 3 are formed.
TABLE 3 prediction error
Figure BDA0002537236320000172
Figure BDA0002537236320000181
As can be seen from the error curves shown in fig. 2 and the prediction error values in table 3, the relative error between the prediction result and the true value is small (the relative error is between 0.003349 and 1.185067), the goodness of fit of the data is high, and the degree of fit is high; therefore, the actual estimation and prediction effects are ideal.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (8)

1. A filtering estimation and prediction estimation method for multi-state variables of a power system is characterized by comprising the following steps:
step 1: time point estimation method based on Kalman filter
1.1, establishing a time series dynamic model of the real-time change of the multi-state vector along with a time point as a state equation of a Kalman filter, and taking the real-time change of the multi-state variable along with the time point in the power system as an input quantity:
x(k+1)=A(k+1,k)x(k)+w(k) (1)
wherein, in the formula (1), x (k) is an observation vector, a (k +1, k) is a state transition matrix, and w (k) is modeling noise; the corresponding observation equation is shown in equation (2):
y(k+1)=C(k+1)x(k+1)+ν(k+1) (2)
wherein, C (k) in the formula (2) is an observation matrix, and v (k) is observation noise;
1.2 establishing Kalman Filter based on time point estimation method
1.2.1 first, it is assumed that an optimal estimate for the kth state x (k) has been obtained
Figure FDA0002537236310000017
And the corresponding estimation error covariance matrix P (k | k):
Figure FDA0002537236310000011
Figure FDA0002537236310000012
1.2.2 then, based on
Figure FDA0002537236310000013
And P (k | k), and further obtaining a predicted value
Figure FDA0002537236310000014
And prediction estimation error covariance P (k +1| k):
Figure FDA0002537236310000015
P(k+1|k)=A(k+1,k)P(k|k)AT(k+1,k)+Qw(k) (6)
1.2.3 next, the optimal gain matrix K (K +1) is calculated:
K(k+1)=A(k+1,k)P(k+1,k)HT(k+1)[H(k+1)P(k+1,k)HT(k+1)]-1(7)
1.2.4 obtaining the optimal estimated value of the state x (k +1)
Figure FDA0002537236310000016
And estimated error covariance P (k +1| k + 1):
Figure FDA0002537236310000021
Figure FDA0002537236310000022
P(k+1|k+1)=[I-K(k+1)H(k+1)P(k+1,k)](10)
1.2.5, continuing to loop the process of 1.2.2-1.2.4 until the filtering is finished;
step 2: method for establishing vector block state estimation based on Kalman filter
2.1 firstly, regarding the whole variable in a time period as a vector block, establishing a dynamic system model of the multi-state vector block changing in time period as a state equation of a Kalman filter, and regarding the real-time change quantity of the multi-state vector block along with a time point in the power system as an input quantity:
X(m+1)=A(m+1,M)X(m,M)+W(m)w(m,M) (11)
wherein, in the formula (11), X (M, M) is an observation vector, a (M +1, M) is a state matrix, and w (M) w (M, M) is modeling noise; the corresponding observation equation is shown in equation (12):
Y(m+1)=HX(m+1)+V(m+1) (12)
wherein, in the formula (12), H is an observation matrix, and V (m +1) is observation noise; the Kalman filter model as shown in equations (3) - (10) can obtain the estimated value of each state vector block
Figure FDA0002537236310000023
And the estimation error covariance matrix P (m | m):
Figure FDA0002537236310000024
Figure FDA0002537236310000025
2.2 establishing Kalman Filter model based on vector Block State estimation method
2.2.1 first, assume that the optimal estimate of the m-th vector chunk state has been obtained
Figure FDA0002537236310000026
And corresponding estimation error covariance matrix
Figure FDA0002537236310000027
Figure FDA0002537236310000028
Figure FDA0002537236310000031
2.2.2 based on
Figure FDA0002537236310000032
And
Figure FDA0002537236310000033
further obtaining the predicted value
Figure FDA0002537236310000034
And prediction estimation error covariance
Figure FDA0002537236310000035
Figure FDA0002537236310000036
Figure FDA0002537236310000037
2.2.3 calculate optimal gain array:
Figure FDA0002537236310000038
2.2.4 obtaining the optimal estimation value of the state block X (m +1)
Figure FDA0002537236310000039
And estimate error covariance
Figure FDA00025372363100000310
Figure FDA00025372363100000311
Figure FDA00025372363100000312
P(m+1|m+1)=[I-K(m+1)H(m+1)P(m+1,m)](22)
2.2.5, continuing to loop the process of 2.2.2-2.2.4 until the filtering is finished;
and step 3: establishing Kalman filtering based block state real-time smooth estimation, filtering estimation and prediction estimation
3.1 first, formula (12) in step 2 is rewritten as follows:
y(m+1,k)=H(m+1,k)X(m+1)+ν(m+1,k),m=1,2,…;k=1,2,…,M (23)
3.2, a real-time smooth estimation model is established, and real-time updating estimation is carried out on states x (m +1,1), x (m +1,2), L and x (m +1, k-1) before the current time k, wherein the model is as follows:
Figure FDA00025372363100000313
3.3, establishing a real-time filtering estimation model, and updating and estimating the state x (m +1, k) at the current k moment in real time, wherein the model is as follows:
Figure FDA0002537236310000041
3.4 establish a real-time prediction estimation model, and carry out real-time prediction estimation on states x (M +1, k +1), x (M +1, k +2), L, x (M +1, M) after the current k moment, wherein the model is as follows:
Figure FDA0002537236310000042
3.5, taking a formula (11) as a state equation, and taking a formula (23) as a state model of a measurement equation to carry out a Kalman filter model;
3.6 building a Kalman Filter model in the Multi-State vector Block
3.6.1 first, assume that the optimal estimate for the mth time instant in the mth state block X (M) is known
Figure FDA0002537236310000043
Sum estimation error covariance matrix
Figure FDA0002537236310000044
Figure FDA0002537236310000045
Figure FDA0002537236310000046
3.6.2 based on
Figure FDA0002537236310000047
And
Figure FDA0002537236310000048
further obtaining a prediction value and a prediction estimation error covariance of the (m +1) th block state:
Figure FDA0002537236310000049
Figure FDA00025372363100000410
3.6.3 further results in an intra-block recursive filter:
Figure FDA00025372363100000411
3.6.4 loops through the 3.6.2-3.6.3 process until the filtering is complete.
2. The method for filtering estimation and prediction estimation of state variables of an electric power system according to claim 1, characterized in that the modeling noise w (k) and the measurement noise v (k) satisfy the following statistical characteristics:
Figure FDA0002537236310000051
wherein Q (k) and R (k) are both error variance matrices, and the error variance matrices Q (k) and R (k) are symmetric non-negative definite matrices and symmetric positive definite matrices, respectively, S (k) ∈ Rn×m
3. The method for filter estimation and prediction estimation of state variables of power system according to claim 1, wherein the equations (11) and (12) are shown in the following formulas
Figure FDA0002537236310000052
Figure FDA0002537236310000053
V(m+1)~N[0,R(m+1)],
R(m+1)=diag{R(m+1,1),R(m+1,2),…,R(m+1,M)},
Figure FDA0002537236310000054
Figure FDA0002537236310000055
4. The method of claim 1The method for filtering estimation and prediction estimation of state variables of the power system is characterized in that measurement noise v (m, k) in the formula (23) has the following statistical characteristic E { ν (m, k) } 0; e { v (m, k) vT(m,j)}=R(m,k)kjm=1,2,…,k,j=1,2,…,M。
5. The method for filter estimation and prediction estimation of state variables of power system according to claim 1, characterized in that in the formula (39)
Figure FDA0002537236310000061
Figure FDA0002537236310000062
Figure FDA0002537236310000063
6. The method for filtering estimation and prediction estimation of state variables of an electric power system according to claim 1, wherein before the step 1, a sensor model of each state variable is established:
yi(t)=hi(x(t))+νi(t),i=1,2,…,N (40)
where N represents the number of sensors, X (t) represents a model of a state variable, hi(. cndot.) is a continuous function;
then, after the equation (40) is discretized by the equal period T, a discrete observation equation is obtained, wherein the discrete observation equation is
yi(kT)=hi(x(kT),kT)+νi(kT),i=1,2,…,N (41)
Wherein h isiDenotes the measurement function of the corresponding sensor i, determined by the properties of the sensor, vi(kT) denotes measurement error; next, the time-dependent order of the state variables x (kT) is established as followsColumn model
x((k+1)T)=f(x(kT),kT)+w(kT) (42)
Where f (x), (k), k ═ a (k +1, k) x (k) is a state transfer function, and w (kt) represents a modeling error.
7. The method for filter estimation and prediction estimation of state variables of an electric power system according to claim 1, characterized in that the multi-state variables comprise state variables of distributed power generation units and state variables of loads using electricity.
8. The method for filtering estimation and prediction estimation facing to state variables of power system according to claim 1, wherein the time period in step 2 includes one hour, one day, one week, one month or one year.
CN202010522875.3A 2020-06-12 2020-06-12 Filtering estimation and prediction estimation method for multi-state variables of power system Withdrawn CN111614086A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010522875.3A CN111614086A (en) 2020-06-12 2020-06-12 Filtering estimation and prediction estimation method for multi-state variables of power system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010522875.3A CN111614086A (en) 2020-06-12 2020-06-12 Filtering estimation and prediction estimation method for multi-state variables of power system

Publications (1)

Publication Number Publication Date
CN111614086A true CN111614086A (en) 2020-09-01

Family

ID=72202542

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010522875.3A Withdrawn CN111614086A (en) 2020-06-12 2020-06-12 Filtering estimation and prediction estimation method for multi-state variables of power system

Country Status (1)

Country Link
CN (1) CN111614086A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113345532A (en) * 2021-06-01 2021-09-03 江南大学 Method for estimating rapid state of polyphenylene oxide production process

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113345532A (en) * 2021-06-01 2021-09-03 江南大学 Method for estimating rapid state of polyphenylene oxide production process
CN113345532B (en) * 2021-06-01 2022-07-15 江南大学 Method for quickly estimating state in production process of polyphenyl ether

Similar Documents

Publication Publication Date Title
Hu et al. A hybrid model based on CNN and Bi-LSTM for urban water demand prediction
CN111008728A (en) Method for predicting short-term output of distributed photovoltaic power generation system
CN104318334B (en) A kind of Short-Term Load Forecasting Method based on correlation FARIMA models long
CN108306303A (en) A kind of consideration load growth and new energy are contributed random voltage stability assessment method
CN102184453A (en) Wind power combination predicting method based on fuzzy neural network and support vector machine
CN107623337B (en) A kind of energy management method for micro-grid
CN110380444B (en) Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula
CN102799948B (en) A kind of parallel networking type photovoltaic power station power generation system output power Forecasting Methodology
CN104463356A (en) Photovoltaic power generation power prediction method based on multi-dimension information artificial neural network algorithm
CN106910144A (en) Based on timesharing carve it is actual with can coefficient heavy construction by when energy consumption on-line prediction method
CN110212551B (en) Micro-grid reactive power automatic control method based on convolutional neural network
CN103986193B (en) A kind of method that maximum wind grid connection capacity obtains
CN112508279A (en) Regional distributed photovoltaic prediction method and system based on spatial correlation
Qijun et al. Photovoltaic power prediction based on principal component analysis and Support Vector Machine
CN114662751B (en) Garden multifunctional short-term load forecasting and optimizing method based on LSTM
CN116384583A (en) Photovoltaic power prediction method based on multiple neural networks
CN111614086A (en) Filtering estimation and prediction estimation method for multi-state variables of power system
CN105207255B (en) A kind of power system peak regulation computational methods suitable for wind power output
CN105354761B (en) Safety and efficiency evaluation method and system for accessing wind power into power grid
Shiralievich et al. Learning algorithm of artificial neural network factor forecasting power consumption of users
CN109345809A (en) The distributed optimization method of solar energy radio acquisition system
Zhang et al. Research on intelligent load forecast in power system dispatching automation
Hua et al. Theory study and application of the BP-ANN method for power grid short-term load forecasting
Zuo et al. Short-term load forecasting for community battery systems based on temporal convolutional networks
CN113673141A (en) Energy router modeling and optimization control method based on data driving

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20200901

WW01 Invention patent application withdrawn after publication