New energy power grid state estimation method based on deep Kalman filtering

Abstract

The invention discloses a new energy power grid state estimation method based on deep Kalman filtering, and the method comprises the steps: obtaining a priori estimation value of the operation state of a system at the current moment through employing a long-short-term memory neural network and the operation state of a power system at the last moment; meanwhile, according to a system measurement equation based on an alternating current power flow model, a system measurement estimated value at the current moment is obtained; and according to a Kalman gain coefficient of deep learning-based recurrent neural network learning, in combination with the prior estimation value of the system operation state at the current moment and the measurement estimation value and the actual measurement value of the system at the current moment, obtaining a posterior estimation value of the system operation state at the current moment.

Description

Technical Field

[0001] The present invention relates to a new energy power grid state estimation method based on deep Kalman filter, and belongs to the field of power system state estimation. Background Art

[0002] Power system state estimation is used to estimate and monitor the real-time operating state of a power system, and is a core component in an Energy Management System (EMS). It receives incomplete and noisy measurement data (raw data) describing the system operating state, including node voltage magnitudes, node injection powers, etc., and provides the accurate operating state of the system, namely the node voltage magnitude and the node voltage phase angle, based on a given estimation criterion. It is the basis for the EMS to achieve functions such as economic dispatch, optimal power flow, and voltage control of the power system.

[0003] Widely used power system state estimation methods include Weight Least Squares (WLS), Weight Least Absolute Values (WLAV), etc. WLS can obtain an unbiased optimal estimate on the premise that the system measurement noise follows a Gaussian distribution. However, when there are bad data in the system measurements, the estimation accuracy of WLS will be affected. WLAV can effectively identify the bad data in the system measurements and ensure the accurate estimation of the system operating state in the presence of bad data. Both belong to the state estimation methods for the system operating state at a single time section and do not consider the correlation between the system states at adjacent times.

[0004] With the wide access of distributed new energy units on the power generation side and the increase in the proportion of flexible loads on the load side, the operating characteristics of the power system have become increasingly complex. Due to the strong volatility and randomness of the output of new energy units and the randomness of the charging demand of electric vehicle loads on the load side, the system operating state changes more frequently. To achieve accurate tracking of the system real-time operating state, a state estimation method based on Kalman filter has been proposed, which considers the evolution relationship of the system operating state over time. For example, the power system state estimation method based on Extended Kalman Filter (EKF) and the state estimation method based on Unscented Kalman Filter (UKF). With the development of deep learning technology, the power system real-time state estimation method based on deep neural network has been studied and applied. However, there are still the following deficiencies in the current power system state estimation: (1) The model-based state estimation method relies on the accurate modeling of the system state space model, including the system state transition matrix, measurement noise distribution, etc. However, it is difficult to accurately characterize and provide these factors for the new power system; (2) The data-based state estimation method lacks the construction of system prior knowledge and often has a large number of parameters to be trained, which limits its interpretability and scalability. SUMMARY OF THE INVENTION

[0005] In view of this, the present invention proposes a new energy grid state estimation method based on deep Kalman filtering to overcome the defects and deficiencies of the above-mentioned existing technologies.

[0006] A new energy grid state estimation method based on deep Kalman filtering includes: obtaining a prior estimate of the system operating state at the current moment by using a long short-term memory neural network and the operating state of the power system at the previous moment; meanwhile, obtaining a measurement estimate of the system at the current moment according to the system measurement equation based on the AC power flow model; and obtaining a posterior estimate of the system operating state at the current moment according to the Kalman gain coefficient learned by a recurrent neural network based on deep learning, combining the prior estimate of the system operating state at the current moment, the measurement estimate of the system at the current moment, and the actual measurement value.

[0007] Further, the steps of calculating the prior estimate of the system operating state at the current moment and the measurement estimate of the system at the current moment include: Constructing a state transition equation by using the system state transition matrix at the previous moment, the posterior estimate of the system operating state at the previous moment, and the input gain term at the previous moment, and constructing a system measurement estimate equation by using the measurement equation and measurement noise to describe the state space model of the power system; Using a deep neural network based on a long short-term memory neural network to learn the state transition function; meanwhile, considering the strong randomness and volatility of the distributed new energy output on the system operating state, taking the change amount of the active power injected into the node connecting the distributed new energy at the current moment as one of the inputs of the long short-term memory neural network, obtaining a new state transition equation, and further obtaining the prior estimate of the system operating state at the current moment and the measurement estimate of the system at the current moment.

[0008] Further, the state space model of the power system is described as: ;

[0009] Wherein, represents t the prior estimate of the system operating state at the current moment, i.e., the current moment; represents the system state transition matrix, which is a diagonal matrix; representst The posteriori estimate value of the system operating state at time - 1, i.e., the previous moment; Represents the input gain term at the previous moment, which is a parameter vector; Represents the system measurement estimate value at the current moment, including the node voltage magnitude , the active power injected at the node , the reactive power injected at the node , the active power of the branch and the reactive power of the branch ; Represents the measurement noise; Represents the measurement equation.

[0010] Furthermore, the measurement equation is: ;

[0011] Wherein, Represents the set of nodes directly connected to node i , including node i itself; Represents t the prior estimate value of the phase - angle difference of branch at time g ij , b ij respectively represent the conductance and susceptance of branch , g si , b si respectively represent the ground conductance and susceptance of node i , , .

[0012] Furthermore, the diagonal elements in the system state - transition matrix and the elements in the parameter vector are obtained by the Holt two - parameter linear exponential smoothing method, and the solution process is: ; ; ; ; ; ;

[0013] Wherein, and ​​respectively represent horizontal term smoothing and trend term smoothing, and respectively represent the horizontal smoothing factor and the trend smoothing factor.

[0014] Furthermore, the new state transition equation is: , ;

[0015] wherein, f LSTM (·) represents the state transition function, represents the node i at t the change in the active power injection at time, represents the set of nodes connected to the new energy units; Furthermore, the prior estimate value t of the system operating state at time and t the system measurement estimate value at time .

[0016] Furthermore, in the Kalman filter framework, the calculation process of the Kalman gain coefficient is: ; ; ;

[0017] wherein, and respectively represent the covariance matrices of the process noise and the measurement noise ; and respectively represent the state error covariance matrix and the measurement residual error covariance matrix; represents the Jacobian matrix.

[0018] Furthermore, the input data of the recurrent neural network for learning the Kalman gain coefficient includes the following four categories: ; ;

[0019] ;

[0020] ;

[0021] wherein, represents t the actual measurement value of the system at time.

[0022] Further, denote the Kalman gain coefficient obtained through deep learning as , then t the posteriori estimate value of the system operating state at .

[0023] The present invention also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the foregoing method can be implemented.

[0024] The beneficial effects of the technical solution of the present invention are reflected in: 1) The foregoing technical solution of the present invention proposes a digital-analog hybrid-driven power system state estimation method that integrates deep learning and Kalman filtering. Compared with the traditional model-based state estimation method, this method does not need to assume a specific distribution form of the measurement noise, and directly learns the Kalman filter gain through the data-driven module, thus avoiding the calculation process of the Jacobian matrix and the Hessian matrix, and reducing the dependence on the model structure and parameter setting; 2) The present invention constructs the state transition relationship of the system through a deep neural network, explicitly introduces the influence of the volatility of distributed new energy output on the system state evolution, so that the constructed state estimation model has stronger adaptability and prediction ability in the scenario of massive distributed new energy access, and improves the estimation accuracy and engineering practicability; 3) Compared with the pure learning-based state estimation method, the present invention integrates the update framework of Kalman filtering in the overall structure, so that the estimated state results have stronger physical interpretability, which is beneficial to meeting the actual application requirements in power system operation and scheduling. Description of the Drawings

[0025] Figure 1 is a schematic diagram of the execution process of the new energy grid state estimation method based on deep Kalman filtering according to an embodiment of the present invention.

[0026] Figure 2 are the predicted value and the true value of the voltage amplitude of node 2 when the 15-node power system is taken as an example in an embodiment of the present invention.

[0027] Figure 3 are the predicted value and the true value of the voltage phase angle of node 2 when the 15-node power system is taken as an example in an embodiment of the present invention.

[0028] Figure 4 is the overall algorithm flow of the method in an embodiment of the present invention. Detailed Embodiments

[0029] The present invention will be further described below in conjunction with the drawings, specific embodiments and examples. The purpose of providing the examples is only for illustration, not for any limitation.

[0030] An embodiment of the present invention proposes a digital-analog hybrid-driven power system state estimation method that combines deep learning and Kalman filtering. Under the framework of the classical Kalman filtering, this method realizes the state estimation of the power system with the background of distributed new energy penetration by combining deep learning technology. The logic of this method is as follows: First, in the prediction stage, the prior estimated value of the system operating state is obtained through a long short-term memory neural network based on deep learning. At the same time, the system measurement estimated value is obtained from the model-based AC power flow equation combined with the prior estimated value of the system operating state obtained above. In the filtering stage, the Kalman gain coefficient is learned through a recurrent neural network based on deep learning. Furthermore, combined with the prior estimated value of the system operating state obtained above, the posterior estimated value of the system operating state is obtained.

[0031] Figure 1 is a schematic diagram of the execution process of the new energy grid state estimation method based on deep Kalman filtering in an embodiment of the present invention, where represents the posterior estimated value of the system operating state at the initial moment, represents moving forward to the previous moment, and the remaining variables will be explained in the subsequent formula descriptions.

[0032] Please refer to Figure 1 , the new energy grid state estimation method based on deep Kalman filtering proposed in an embodiment of the present invention is roughly divided into the following two stages: Stage 1: Calculate t the prior estimated value of the system operating state and the system measurement estimated value at time (the current moment).

[0033] Using the long short-term memory neural network (LSTM) and the operating state of the power system at the previous moment ( t -1 moment), the prior estimated value of the system operating state at the current moment ( t moment) is obtained; at the same time, according to the system measurement equation based on the AC power flow model, the system measurement estimated value at the current moment is obtained. The specific process is as follows: First, construct a state transition equation using the system state transition matrix at the previous moment, the posterior estimated value of the system operating state at the previous moment, and the input gain term at the previous moment, and construct a system measurement estimated equation using the measurement equation and the measurement noise. Use these two equations to construct a state space model describing the power system, as follows: ;

[0034] where, represents t the prior estimated value of the system operating state at time; represents the system state transition matrix, which is a diagonal matrix; representst The posteriori estimated value of the system operating state at time - 1; Represents the input gain term at the previous time, which is a parameter vector; Represents the system measurement estimated value at the current time, including the node voltage magnitude , the active power injected at the node , the reactive power injected at the node , the active power of the branch and the reactive power of the branch ; Represents the measurement noise; Represents the measurement equation, expressed as: ;

[0035] Where, Represents the set of nodes directly connected to node i , including node i itself; Represents t the prior estimated value of the phase - angle difference of branch at time g ij , b ij respectively represent the conductance and susceptance of branch , g si , b si respectively represent the ground conductance and susceptance of node i , , .

[0036] The diagonal element in the system state transition matrix and the element in the parameter vector are obtained by the Holt two - parameter linear exponential smoothing method. The solution process is as follows: ;

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] Where, and respectively represent horizontal term smoothing and trend term smoothing, and respectively represent the horizontal smoothing factor and the trend smoothing factor.

[0043] The state transition equation obtained by the Holt two-parameter linear exponential smoothing method mentioned above is applicable to scenarios with small system load fluctuations. However, the present invention is more applicable to scenarios with large fluctuations in new energy output on the power generation side of the system.

[0044] In the context of distributed new energy penetration, in order to accurately predict the evolution of the system operating state over time, a deep neural network based on long short-term memory neural network is used to learn the state transition function f LSTM . At the same time, considering the impact of the strong randomness and volatility of distributed new energy output on the system operating state, the change in the active power injection of the node connecting distributed new energy at the current moment d t is used as one of the inputs of the LSTM to obtain a new state transition equation as follows: , ;

[0045] wherein, represents the change in the active power injection of node i at t moment, represents the set of nodes connected with new energy units.

[0046] Furthermore, the prior estimated value t of the system operating state at moment and the measured estimated value t of the system at moment can be obtained: .

[0047] Phase 2, calculate t the posterior estimated value of the system operating state at

[0048] According to the Kalman gain coefficient learned by the recurrent neural network based on deep learning, combined with the prior estimated value t of the system operating state at moment and the measured estimated value t of the system at , calculate t the posterior estimated value of the system operating state at moment. The specific process is as follows: In the Kalman filter framework, the calculation process of the Kalman gain coefficient ;

[0049] ;

[0050] ;

[0051] Wherein, and respectively represent the covariance matrices of process noise and measurement noise ; and respectively represent the state error covariance matrix and the measurement residual error covariance matrix; represents the Jacobian matrix.

[0052] Therefore, when designing the recurrent neural network for learning the Kalman gain coefficient , its input data includes the following four categories: ;

[0053] ;

[0054] ;

[0055] ;

[0056] Wherein, represents t the actual measurement value of the system at time

[0057] Denote the Kalman gain coefficient obtained through deep learning as . Therefore, in the filtering stage, the posterior estimate value t of the system operating state at time can be calculated.

[0058] The overall algorithm flow of the method in the embodiment of the present invention is as shown in Figure 4 and mainly includes: 1. Initialization: The trained long short-term memory neural network (LSTM) and recurrent neural network (RNN) structures; 2. Input: The measurement equation of the power system , the initial state variable , the measurement value (the node voltage amplitude U, the node injection power and the branch power ); 3. Prediction step of the Kalman filter framework: ; 4. Update step of the Kalman filter framework: 1) Calculate the Kalman gain coefficient using the trained recurrent neural network (RNN). ; 2) ; 5. Output: .

[0059] The present invention is proven to be applicable to the real-time operation state tracking of power systems, can achieve accurate estimation of the operation state of power systems, and has the characteristics of wide compatibility and strong adaptability. Taking a 15-node power system as an example, the present invention is used to track the real-time operation state of the power system. The mean squared error (MSE) and mean absolute error (MAE) of the node voltage amplitude and node voltage phase angle are used as evaluation indexes, which are defined as: 1) Node voltage amplitude: ;

[0060] ;

[0061] 2) Node voltage phase angle: ;

[0062] ;

[0063] Among them, represents the number of system nodes, represents the total number of tracking moments. and represent the estimated values of the node voltage amplitude and node voltage phase angle of node at time obtained by the present invention, and represent the true values of the node voltage amplitude and node voltage phase angle of node at time . The calculation results show that the MSE and MAE of the node voltage amplitude are and respectively, and the MSE and MAE of the node voltage phase angle are and respectively.

[0064] Meanwhile, taking node 2 as an example, the comparison between the estimated values and the true values of the voltage amplitude and phase angle of node 2 is given, see Figure 2 and Figure 3 . From the results in Figure 2 and Figure 3 , it can be seen that the estimated values obtained by the present invention are very close to the true values. The above results prove that the present invention can achieve accurate tracking of the real-time operation state of power systems.

[0065] Another embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the foregoing embodiments can be implemented. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes several instructions for causing a computer device (such as a PC, a server or a network device, etc.) to execute the various steps of the method of the present invention to implement the functions of the present invention.

[0066] The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several equivalent substitutions or obvious modifications can be made, and if the performance or use is the same, they should all be regarded as belonging to the protection scope of the present invention.

Claims

1. A new energy grid state estimation method based on deep Kalman filtering, characterized in that: include: By using the long short-term memory neural network and the operating state of the power system at the previous moment, the prior estimate of the system operating state at the current moment is obtained; At the same time, according to the system measurement equation based on the AC power flow model, the estimated value of the system measurement at the current moment is obtained; According to the Kalman gain coefficient learned by the recurrent neural network based on deep learning, the a priori estimate of the system operating state at the current moment and the system measurement estimate and actual measurement value at the current moment are combined to obtain the a posteriori estimate of the system operating state at the current moment.

2. The new energy grid state estimation method according to claim 1, characterized in that: The step of calculating the a priori estimated value of the system operating state at the current moment and the system measurement estimated value at the current moment comprises: The state transfer equation is constructed using the system state transfer matrix at the previous moment, the a posteriori estimate of the system operating state at the previous moment, and the input gain term at the previous moment, and the system measurement estimation equation is constructed using the measurement equation and the measurement noise to describe the state space model of the power system; A deep neural network based on long short-term memory neural network is used to learn the state transfer function. At the same time, considering the impact of the strong randomness and volatility of distributed renewable energy output on the system operation state, the change in active power injected into the nodes connected to the distributed renewable energy at the current moment is used as one of the inputs of the long short-term memory neural network to obtain a new state transfer equation, and then the a priori estimate of the system operation state at the current moment and the system measurement estimate at the current moment are obtained.

3. The new energy grid state estimation method according to claim 2, characterized in that: The state space model of the power system is described as: ； in, represent t The time is the a priori estimate of the system operation status at the current time; Represents the system state transfer matrix, which is a diagonal matrix; represent t -1 moment is the a posteriori estimate of the system operating status at the previous moment; Represents the input gain term at the previous moment, which is a parameter vector; Represents the estimated value of the system measurement at the current moment, including the node voltage amplitude , node injected active power , node injected reactive power , Branch active power and branch reactive power ; represents the measurement noise; Represents the measurement equation.

4. The new energy grid state estimation method according to claim 3, characterized in that: The measurement equation for: ； in, Representatives and Nodes i The set of directly connected nodes, including nodes i itself; represent t Time Branch The a priori estimate of the phase angle difference, g ij , b ij Represents branches The conductance and susceptance of g si , b si Respectively represent nodes i The conductance and susceptance to ground, , .

5. The new energy grid state estimation method according to claim 3, characterized in that: System state transition matrix The diagonal elements in and parameter vector Elements in It is obtained by Holt's two-parameter linear exponential smoothing method, and the solution process is: ； ； ； ； ； ； in, and Represents horizontal term smoothing and trend term smoothing, and Represent the horizontal smoothing factor and trend smoothing factor respectively.

6. The new energy grid state estimation method according to claim 3, characterized in that: The new state transfer equation is: ， ； in, f LSTM (·) represents the state transition function, Representative Node i exist t The change in injected active power at each moment, Represents the set of nodes connected to new energy units; Then seek t A priori estimate of the system operating status at the moment and t System measurement estimate at time : 。 7. The new energy grid state estimation method according to any one of claims 3 to 6, characterized in that: In the Kalman filter framework, the Kalman gain coefficient The calculation process is: ； ； ； in, and Represents the process noise and measurement noise The covariance matrix of and Represent the state error covariance matrix and the measurement residual error covariance matrix respectively; Represents the Jacobian matrix.

8. The new energy grid state estimation method according to claim 7, characterized in that: According to the calculation process of the Kalman gain coefficient in the Kalman filter framework, the input data of the recurrent neural network for learning the Kalman gain coefficient includes the following four categories: ； ； ； ； in, represent t The actual measurement value of the system at that moment.

9. The new energy grid state estimation method according to claim 8, characterized in that: The Kalman gain coefficient obtained by deep learning is recorded as ,therefore, t The a posteriori estimate of the system operating state at time .

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 can be implemented.

Images