Dynamic Stability Prediction Control Method for DC Microgrid Based on Fuzzy Observer

Abstract

The present invention specifically relates to a dynamic stability prediction control method for a DC microgrid based on a fuzzy observer, including: establishing a dynamic model of the DC microgrid; establishing a T-S fuzzy model of the DC microgrid based on the dynamic model of the DC microgrid and combining the sector nonlinear method; constructing a corresponding fuzzy observer and a fuzzy predictive controller based on the T-S fuzzy model of the DC microgrid; inputting the grid signal of the DC microgrid at the current moment into the fuzzy observer to output the corresponding system state; then inputting the system state into the fuzzy predictive controller to output the corresponding future control input; finally, solving the future grid control quantity based on the future control input, and further realizing the control of the DC microgrid based on the future grid control quantity. The present invention adapts to the uncertainty and complexity of the DC microgrid through a fuzzy observer and a dual-network delay compensation structure, and can enhance the robustness to network delay and greatly reduce the computational burden.

Description

Technical Field

[0001] The present invention relates to the technical field of DC microgrid control, and particularly to a dynamic stability prediction control method for a DC microgrid based on a fuzzy observer. Background Art

[0002] Microgrid technology represents the development trend of future distributed energy supply systems. Among them, a DC microgrid is a microgrid composed of DC, which is an important part of future intelligent power distribution and utilization systems, and is of great significance for promoting energy conservation and emission reduction and achieving sustainable energy development. Compared with an AC microgrid, a DC microgrid can more efficiently and reliably accommodate distributed renewable energy generation systems such as wind and light, energy storage units, electric vehicles, and other DC electrical loads.

[0003] The advantages of DC microgrids, such as high efficiency and robustness, simple control, and natural interfaces for distributed generators and electrical loads, make them the focus of current research. On this basis, the dynamic stability prediction control method for DC microgrids is of great significance for the safe and stable operation of DC microgrids. However, the nonlinear optimization solver of existing dynamic stability prediction control methods has a very high computational burden, and under current network conditions, it cannot cope with random data packet loss and network-induced delays that may be caused by network imperfections.

[0004] Currently, control based on the T-S (Takagi-Sugeno) fuzzy model is relatively common in the control of complex nonlinear systems and has achieved effective control effects. In particular, fuzzy control gain theories based on techniques such as parallel compensation (PDC) and linear matrix inequality (LMI) have enabled linear control theory to be widely applied to the control of nonlinear systems. However, due to the uncertainty of constant power supply loads in DC microgrids and the complexity of DC microgrid systems, existing control theory methods based on the T-S fuzzy model have not been directly successfully applied to actual DC microgrid systems. Therefore, how to design a method that can adapt to the uncertainty and complexity of DC microgrids is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] Aiming at the deficiencies of the above-mentioned prior art, the technical problem to be solved by the present invention is: how to provide a dynamic stability prediction control method for a DC microgrid based on a fuzzy observer, so as to be able to adapt to the uncertainty and complexity of the DC microgrid, thereby improving the effect of DC microgrid dynamic stability prediction.

[0006] To solve the above technical problem, the present invention adopts the following technical solutions:

[0007] A dynamic stability prediction control method for a DC microgrid based on a fuzzy observer, comprising:

[0008] S1: Establish the dynamic model of the DC microgrid;

[0009] S2: Based on the dynamic model of the DC microgrid, establish the T-S fuzzy model of the DC microgrid by combining the sector nonlinear method;

[0010] S3: Construct the corresponding fuzzy observer and fuzzy predictive controller based on the T-S fuzzy model of the DC microgrid;

[0011] S4: Input the grid signal of the DC microgrid at the current moment into the fuzzy observer to output the corresponding system state; then input the system state into the fuzzy predictive controller to output the corresponding future control input; finally, solve the future grid control quantity based on the future control input, and further realize the control of the DC microgrid based on the future grid control quantity.

[0012] Preferably, in step S1, the dynamic model of the DC microgrid is represented by the following formula:

[0013]

[0014] Among them,

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025] In the formula: represents the fitting value of the state space vector; h represents the fitting value of the function; represents the voltage fitting value of the DC power supply; represents the transpose of the Qth CPL state; h Q represents the h function of the Qth CPL state; r s represents the resistance of the filter connected to the DC power supply; denotes the fitted value of the voltage of the capacitor in the j-th CPL; v C0,j denotes the initial voltage of the capacitor in the j-th CPL; x s =[i L,s v C,s T , i L,s and v C,s respectively denote the inductor current and capacitor voltage in the DC microgrid energy storage system; x j =[i L,j v C,j T , i L,j and v C,j respectively denote the inductor current and capacitor voltage in the j-th constant power load of the DC microgrid; s = Q + 1, Q represents the number of constant power load systems in the DC microgrid energy storage system; A, B, C, D denote undetermined matrix coefficients; a, b, c, d denote undetermined coefficients; L s denotes the inductor of the j-th filter connected to the DC source; i es denotes the injection current of the energy storage system; V dc denotes the voltage of the DC power supply; C s denotes the capacitor of the j-th filter connected to the DC source; denotes DC parameters; denotes AC parameters; denotes a vector; ·(t) is related to an event; · T denotes the transpose; · s is related to the DC power supply; · e is related to the energy storage system; · Q denotes the Q-th; · j denotes the j-th.

[0026] Preferably, in step S2, the T-S fuzzy model of the DC microgrid is represented by the following formula:

[0027]

[0028] Wherein,

[0029]

[0030]

[0031]

[0032] In the formula: denotes the equivalent T-S fuzzy representation of state x; denotes the fitted value of the input vector; ​​Denote the undetermined coefficient \(A\) equivalent to the T-S fuzzy model; Denote the voltage fitting value of the DC power supply; \(d\) Q Denote the undetermined coefficient \(d\) corresponding to the \(Q\)th CPL; \(U\) k1,1 Denote the input quantity of the 1st CPL after normalization; \(U\) kQ,Q Denote the input quantity of the \(Q\)th CPL after normalization; \(x\) represents the state space; \(k\) 1 and \(k\) 2 and \(k\) Q Denote \(i\) after binary decomposition; \(u\) represents the input vector; \(m\) i Denote the normalized membership function of the \(j\)th rule; \(i = 1 + k\) 1 ×2 0 + \(k\) 2 ×2 1 +…+ \(k\) Q ×2 Q-1 ∈{1,...,2 Q} represents the number of linear models \(j\in\{1,...,q\}, k\) j ∈{0,1} represents the \(j\)th bit in the \(Q\)-bit binary representation of \(i - 1\); Denote the superposition of fuzzy levels from 1 to \(Q\).

[0033] Preferably, in step S3, the fuzzy observer is represented by the following formula:

[0034]

[0035] \(L = P\) 1 -1 \(M\);

[0036] In the formula: Denote the output vector matrix; \(C\) represents the undetermined coefficient matrix of the output equation; \(M\) represents the observer gain coefficient; Denote the state of the observer; and respectively denote the state vector and output vector of the fuzzy observer; \(L\) represents the gain of the fuzzy observer; Denote the estimation error; \(dt\) represents the known time-varying delay from the network sensor to the controller calculated by the network delay compensator; \(P1\) represents the Lyapunov matrix.

[0037] Preferably, in step S3, the fuzzy predictive controller is represented by the following formula:

[0038]

[0039]

[0040] \(X = \Psi+\Theta U\);

[0041] U = (Θ T ΔΘ + Λ)Θ T Δ(Ψ - W);

[0042]

[0043]

[0044]

[0045]

[0046] where: h represents a function, y is the output vector; x represents the state space; u represents the input vector; E h represents the undetermined coefficient matrix of the discrete TS fuzzy system; C xk represents the undetermined coefficient matrix of the output equation; represents estimating the system state at the k - dk + 1 moment based on the available information at the k - dk moment; both represent the undetermined coefficients in the Ψ matrix; x ok represents the state of the observer after discretization; · k represents the sequence after time discretization, corresponding to different moments; Ψ represents the matrix containing all past information; U represents the future control input matrix; X represents the system state matrix; Θ represents the undetermined coefficient matrix; Λ and Δ represent matrix operation symbols; represents estimating the system state at the k moment based on the available information at the j moment.

[0047] Preferably, when designing the future control input matrix U of the fuzzy predictive controller, the following cost function needs to be considered:

[0048]

[0049] Transform the cost function into vector form J = (X - W) T Δ(X - W) + U T ΛU;

[0050] Δ = diag{δ 0 , δ 1 ,..., δ τ}

[0051] Λ = diag{λ 0 , λ 1 ,..., λ τ};

[0052] W = [w k w k+2 … w k+τ T ; ​

[0053]

[0054] H = Θ T ΔΘ + Λ ≥ 0;

[0055] K = (Ψ - W) T ΔΘ;

[0056] G = (Ψ - W) T Δ(Ψ - W);

[0057] Where: J represents the cost function; w k+j represents the function of future reference; the coefficient δ j and λ j are respectively used to determine the weights of future behavior and control input effort, which are control design parameters and represent the importance of tracking error or control input energy; λ j represents the weight of the cost function; W represents the future reference of the system; represents estimating the system state at k - d k time according to the available information at k - d k+1 time; H, K, G are constructed discriminant matrices.

[0058] Preferably, in step S3, a first network delay compensator for calculating the STC delay is provided between the input of the fuzzy observer and the DC microgrid, and a second network delay compensator for calculating the CTA delay is provided between the fuzzy predictive controller and the DC microgrid.

[0059] Preferably, in step S4, it specifically includes the following steps:

[0060] S401: Calculate the STC delay d k at k time through the first network delay compensator, and obtain the control output of the DC microgrid at k - d k time Then, obtain the control input at k - d k time through the future control input matrix U of the fuzzy predictive controller

[0061] S402: Use the control input k and the control output at k - d time as the input of the fuzzy observer, and output the system state at k - d k time through the fuzzy observer

[0062] S403: Use the system state as the input of the fuzzy predictive controller, and output the corresponding future control input u k, and update the future control input matrix U for the future;

[0063] S404: Obtain the CTA delay τ through the second network delay compensator k , and then determine k + τ based on the updated future control input matrix U k The future control input at the moment As the future grid control quantity of the DC microgrid.

[0064] Preferably, in step S403, the control input Control output And the upper limit of the CTA delay are brought into the formula of the fuzzy predictive controller to calculate the corresponding matrices Ψ and Θ; then, through the system state Update the system state matrix X, and calculate and update the future control input matrix U in combination with the following formula;

[0065] X = Ψ + ΘU;

[0066] In the formula: Ψ represents the matrix containing all past information; U represents the future control input matrix; X represents the system state matrix; Θ represents the matrix of undetermined coefficients.

[0067] Preferably, if the first network delay compensator or the second network delay compensator fails to obtain data, the STC delay d k Or the CTA delay τ k Increment by one.

[0068] The dynamic stability predictive control method for a DC microgrid based on a fuzzy observer in the present invention has the following beneficial effects:

[0069] In the present invention, the grid signals of the DC microgrid at the current moment are input into the fuzzy observer to obtain the system state, then the system state is input into the fuzzy predictive controller to obtain the future control input, and finally, based on the future control input, the future grid control quantity of the DC microgrid is solved, enabling the estimation of the states of constant power loads (CPLs) and power buffers according to measurement information through a fuzzy observer independent of time delay, which can greatly enhance the applicability of the dynamic stability control of the DC microgrid, that is, the fuzzy observer and the fuzzy predictive controller can well adapt to the uncertainties of the DC microgrid, can be used for the dynamic stability comprehensive predictive control of the power buffer of the DC microgrid containing a 5G network, and is of great significance for the safe and stable operation of the DC microgrid, thereby improving the effect of the dynamic stability prediction of the DC microgrid.

[0070] 5G technology enables data exchange on a millisecond time scale. Fast information exchange is a technical prerequisite for the predictive control method proposed in this patent. However, due to high data exchange on the communication channel, random packet loss and network-induced delays may occur. Therefore, in the present invention, a first network delay compensator for calculating STC delay is provided between the input of the fuzzy observer and the DC microgrid, and a second network delay compensator for calculating CTA delay is provided between the fuzzy predictive controller and the DC microgrid. The double-network delay compensator structure fully considers the possible delays and data packet losses in the imperfect 5G network, has robustness to network delays, and can significantly reduce the computational burden. Furthermore, it can effectively adapt to the complexity of the DC microgrid, thereby improving the effect of DC microgrid dynamic stability prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to make the objectives, technical solutions, and advantages of the invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings, where:

[0072] Figure 1 is the logic block diagram of the dynamic stability prediction control method for the DC microgrid;

[0073] Figure 2 (a) is the schematic diagram of the DC microgrid, Figure 2 (b) is the circuit diagram of the DC microgrid;

[0074] Figure 3 is the network predictive controller with a double-network delay compensator structure;

[0075] Figure 4 is the flowchart of the operation of the network predictive controller. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0076] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. The components of the embodiments of the present invention usually described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0077] It should be noted that similar reference numerals and letters denote similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the figures, or the orientation or positional relationship in which the inventive product is customarily placed during use. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance. In addition, terms such as "horizontal" and "vertical" do not mean that the components are required to be absolutely horizontal or hanging vertically, but may be slightly inclined. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but may be slightly inclined. In the description of the present invention, it should also be noted that unless otherwise clearly specified and defined, the terms "set", "installed", "connected", "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0078] The following will be further described in detail through specific embodiments:

[0079] Embodiment:

[0080] A dynamic stability prediction control method for a DC microgrid based on a fuzzy observer is disclosed in this embodiment.

[0081] As Figure 1 shown, the dynamic stability prediction control method for a DC microgrid based on a fuzzy observer includes:

[0082] S1: Establish a dynamic model of the DC microgrid of the DC microgrid;

[0083] S2: Based on the dynamic model of the DC microgrid, establish a T-S fuzzy model of the DC microgrid in combination with the sector nonlinear method;

[0084] S3: Construct a corresponding fuzzy observer and a fuzzy predictive controller based on the T-S fuzzy model of the DC microgrid;

[0085] In this embodiment, a first network delay compensator for calculating STC delay is provided between the input of the fuzzy observer and the DC microgrid, and a second network delay compensator for calculating CTA delay is provided between the fuzzy predictive controller and the DC microgrid, thus forming a network predictive controller with a dual network delay compensator structure.

[0086] S4: Input the grid signal of the DC microgrid at the current moment into the fuzzy observer to output the corresponding system state; then input the system state into the fuzzy predictive controller to output the corresponding future control input; finally, solve the future grid control quantity based on the future control input, and further realize the control of the DC microgrid based on the future grid control quantity.

[0087] In this embodiment, realizing the control of the DC microgrid based on the future grid control quantity means using the future grid control quantity as the control input of the DC microgrid.

[0088] In the present invention, the grid signal of the DC microgrid at the current moment is input into the fuzzy observer to obtain the system state, then the system state is input into the fuzzy predictive controller to obtain the future control input, and finally the future grid control quantity of the DC microgrid is solved based on the future control input, enabling the estimation of the states of the constant power load (CPL) and the power buffer according to the measurement information through a fuzzy observer independent of time delay, greatly enhancing the applicability of the dynamic stability control of the DC microgrid, that is, the fuzzy observer and the fuzzy predictive controller can well adapt to the uncertainties of the DC microgrid, can be used for the dynamic stability comprehensive prediction control of the power buffer of the DC microgrid containing a 5G network, and is of great significance for the safe and stable operation of the DC microgrid, thereby improving the effect of the dynamic stability prediction of the DC microgrid.

[0089] 5G technology enables data exchange on a millisecond time scale, and fast information exchange is the technical premise of the predictive control method proposed in this patent. However, due to the high data exchange on the communication channel, random packet loss and network-induced delays may occur. Therefore, in the present invention, a first network delay compensator for calculating STC delay is provided between the input of the fuzzy observer and the DC microgrid, and a second network delay compensator for calculating CTA delay is provided between the fuzzy predictive controller and the DC microgrid. The dual network delay compensator structure fully considers the possible delays and data packet losses in the imperfect 5G network, is robust to network delays, and can greatly reduce the computational burden, thereby being able to effectively adapt to the complexity of the DC microgrid and improving the effect of the dynamic stability prediction of the DC microgrid.

[0090] In this embodiment:

[0091] TS represents Takagi–Sugeno.

[0092] NDC represents the Network Delay Compensator.

[0093] CPL represents the Constant Power Load.

[0094] SiL represents Software-in-the-Loop.

[0095] MPC represents the Model Predictive Controller.

[0096] LMI represents the Linear Matrix Inequality.

[0097] STC represents Sensor-to-Controller.

[0098] CTA represents Controller-to-Actuator.

[0099] NPC represents the Network Predictive Controller.

[0100] 1. DC Microgrid Dynamic Model

[0101] Combined Figure 2 it can be known that the entire DC microgrid can be decoupled into Q + 1 subsystems: Q constant power load systems (CPL) and one energy storage system (ESS).

[0102] The modeling of the constant power load is:

[0103] For j = {1, 2, 3,..., Q} and s = Q + 1, where x j = [i L,j v C,j T ;

[0104] i L,j and v C,j are respectively the inductor current and capacitor voltage in the j-th constant power load, and:

[0105]

[0106] where P j is the power of the j-th constant power load, and it is set as a known constant.

[0107] The source subsystem can be written as:

[0108]

[0109] x s = [i L,s v C,s T and i L,s and v C,s are respectively the inductor current and capacitor voltage in the ESS, where:

[0110] ​​

[0111] To make full use of the TS fuzzy model and obtain the equivalent fuzzy representation of (1) and (3), it is necessary to apply a coordinate transformation near the operating point to construct a system with the equilibrium point as the origin. By taking the energy storage current i es as the control input, the entire DC microgrid model can be written in the following form:

[0112]

[0113] where:

[0114] and

[0115]

[0116]

[0117] It can be seen from equations (5) and (6) that the entire microgrid system includes Q non-linear terms (i.e., h j ).

[0118] In other words, each constant power load is modeled by the non-linear term P j / V c,j . Therefore, the number of non-linear terms is the same as the number of constant power loads.

[0119] In summary, the dynamic model of the DC microgrid is expressed by the following formula:

[0120]

[0121] where,

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132] In the formula: represents the fitted value of the state - space vector; h represents the fitted value of the function; represents the fitted value of the voltage of the DC power supply; represents the transpose of the Q - th CPL state; h Q represents the h - function of the Q - th CPL state; r s represents the resistance of the filter connected to the DC power supply; represents the fitted value of the voltage of the capacitor in the j - th CPL; v C0,j represents the initial voltage of the capacitor in the j - th CPL; x s = [i L,s v C,s T , i L,s and v C,s respectively represent the inductor current and capacitor voltage in the DC micro - grid energy storage system; x j = [i L,j v C,j T , i L,j and v C,j respectively represent the inductor current and capacitor voltage in the j - th constant - power load of the DC micro - grid; s = Q + 1, Q represents the number of constant - power load systems in the DC micro - grid energy storage system; A, B, C, D represent undetermined matrix coefficients; a, b, c, d represent undetermined coefficients; L s represents the inductance of the j - th filter connected to the DC source; i es represents the injected current of the energy storage system; V dc represents the voltage of the DC power supply; C s represents the capacitance of the j - th filter connected to the DC source; represents DC parameters; represents AC parameters; represents a vector; ·(t) represents being related to an event; · T represents transpose; · s represents being related to the DC power supply; · e represents being related to the energy storage system; · Q represents the Q - th; · j represents the j - th.

[0133] 2. T - S Fuzzy Model

[0134] The present invention adopts the sector - nonlinear method to systematically calculate the equivalent Takagi - Sugeno fuzzy model.

[0135] ​​First, each non - linear term of the model (5) is surrounded by two linear sectors; then, through the slopes of these sectors, the fuzzy membership functions and the system matrix are calculated. The j - th constant - power load is locally stable in this region:

[0136]

[0137] where w 1,j and w 2,j are positive scalars, which can be obtained by the linear matrix inequality method. For the region (7), there is where the lower and upper bounds of U 0,j and U 1,j are:

[0138]

[0139] Adopting the sector non - linear method, there is:

[0140]

[0141] Solving formula (9), the fuzzy grades and

[0142]

[0143] Substituting equation (9) into equation (5), the equivalent TS fuzzy model can be obtained:

[0144]

[0145] where and is the control input and the normalized membership function m i , and it can be obtained that:

[0146]

[0147] where i = 1 + k 1 ×2 0 +k 2 ×2 1 +…+k Q ×2 Q-1 ∈{1,...,2 Q} is the number of linear models, j∈{1,...,q}, k j ∈{0,1} is the j - th bit in the Q - bit binary representation of i - 1. In addition

[0148] (11) The continuous - time dynamic model can be discretized into the following form:

[0149]

[0150] wherein and E d is the perturbed input after discretization operation.

[0151] The Euler method is a simple and effective method for deriving discrete-time systems. Based on the Euler method:

[0152]

[0153] where T is the discretization constant. Based on the systems represented by (11) and (13) respectively.

[0154] In summary, the T-S fuzzy model of the DC microgrid is represented by the following formula:

[0155]

[0156] wherein,

[0157]

[0158]

[0159]

[0160] In the formula: represents the equivalent T-S fuzzy representation of the state x; represents the fitted value of the input vector; represents the undetermined coefficient A equivalent to the T-S fuzzy model; represents the voltage fitted value of the DC power supply; d Q represents the undetermined coefficient d corresponding to the Qth CPL; U k1,1 represents the input quantity of the first CPL after normalization; U kQ,Q represents the input quantity of the Qth CPL after normalization; x represents the state space;; k 1 、k 2 、k Q represents i after binary decomposition; u represents the input vector; m i represents the normalized membership function of the jth rule; i = 1 + k 1 ×2 0 + k 2 ×2 1 +…+ k Q ×2 Q-1 ∈ {1,..., 2 Q} represents the number of linear models j ∈ {1,..., q}, k j∈{0,1} represents the j-th bit in the Q-bit binary representation of i - 1; represents the superposition of the fuzzy levels from 1 to Q.

[0161] 3. Network Prediction Controller

[0162] First of all, wireless communication continues to play an important role in the modernization of the power system. Compared with wired communication, wireless communication saves costs and has the advantages of deployment convenience and flexibility. As a technology that has been verified in the commercial mobile environment, 5G can meet many key requirements of the power grid. 5G is a cost-effective alternative to the newly deployed fiber optics with high costs and the proprietary microwave with insufficient bandwidth. It also performs better than RF mesh networks in terms of latency and is superior to narrowband power line carriers in terms of flexibility and signal-to-noise ratio.

[0163] The advantages of 5G network technology can be summarized as follows:

[0164] 1) Public operator networks, including 2G / 3G / 4G networks, cannot meet the service requirements due to insufficient resources such as poor privacy and high rental costs.

[0165] 2) Due to the limitations of transmission distance and sensitivity to geographical environment, short-distance wireless communication technologies (2G / 3G / 4G) are only applicable to simple service applications in a small range and cannot meet the full-service requirements of distribution automation.

[0166] 3) 5G network technology is a more secure communication technology than 2G / 3G / 4G networks.

[0167] Complex large-scale distributed DC microgrids are usually connected via Ethernet and / or wireless connections, as Figure 3 shown. In the 5G scenario, data is exchanged on a millisecond time scale, making the dynamic stability of the microgrid possible. However, due to the high data exchange on the communication channel, random data packet loss and network-induced delays may occur. Figure 3 Shown is the novel network prediction controller based on the TS fuzzy model designed by the present invention. To compensate for network transmission delays, network delay compensators are placed in the controller-to-actuator channel and the sensor-to-controller channel. In addition, in an actual DC microgrid, multiple sensors need to be added and the voltages and currents of all capacitors and inductors need to be collected to measure all CPL states, which requires too high costs. Therefore, the present invention adds a fuzzy observer to estimate the CPL (constant power load) information based on the available measurement data. Finally, a stable control input is designed using the fuzzy MPC controller.

[0168] The specific details of each module will be given below:

[0169] 1) Network Delay Compensation

[0170] The present invention defaults that an imperfect network will have latency problems.

[0171] To propose a robust network control method, two NDCs are used to evaluate the network latency τ k and d k , and the network latency is compensated by using a timestamp scheme.

[0172] In this scheme, the transmitted information and the starting transmission time are packed and sent through the network. Therefore, the NDC can calculate the latency by subtracting the reception time from the starting transmission time. In addition, since the network latencies τ k and d k are random, the NDC can receive data discontinuously. Therefore, the NDC selects the latest timestamp data from all the received data at a specific moment. For example, at time k, there are two timestamp data packets k1 and k2 (k1 > k2), and the NDC retains k1 and calculates the network latency as k - k1 ≥ 0. In addition, for a specific time, if no data is received or data with a timestamp is received earlier, the NDC retains the latest received data and increases the network latency by one level. Therefore, the NDC can handle a limited sequence of data loss.

[0173] 2) Fuzzy delay-independent observer

[0174] In practical applications, a fuzzy observer needs to be designed for model (5) so as to obtain all the states of (5) by measuring partial states. In this embodiment, the problem of designing an observer independent of fuzzy delay for the TS fuzzy model (11) is discussed, and its time-delay output is:

[0175]

[0176] dt is the known time-varying network sensor-to-controller delay calculated by NDC1. The following observer structure is proposed:

[0177]

[0178] where and are the state and output vectors of the observer respectively, and L is the observer gain. The defined estimation error whose dynamic expression can be derived as:

[0179]

[0180] h(.) is defined in formula (5). The goal is to design the observer gain matrix L such that the system (17) is asymptotically stable (i.e., ). Therefore, Consider the Lyapunov–Krasovskii functional:

[0181]

[0182] where \(P_1\) and \(P_2\) are symmetric positive definite matrices. By neglecting the time derivative of \(\tau\) t we have the following:

[0183]

[0184] By substituting (17) into (19), we get:

[0185]

[0186] Since is a continuous function satisfying \(h(0)=0\), it is locally Lipchitz continuous and there exists a positive Lipchitz coefficient \(\gamma\) such that:

[0187]

[0188] Therefore, equation (20) can be written as:

[0189]

[0190] Therefore, we have:

[0191]

[0192]

[0193] The negativity of equation (24) is due to:

[0194]

[0195] Applying the Schur complement and defining \(M = P\) 1 \(L\), we have:

[0196]

[0197] By solving the linear matrix inequalities subject to \(P_1>0\) and \(P_2>0\), the observer gain matrix is calculated as:

[0198] \(L = P\) 1 -1 \(M\) (26)

[0199] Satisfying the conditions \(P_1>0\), \(P_2>0\) and equation (26), the dynamic stability of the observer error is guaranteed by the Lyapunov stability theory.

[0200] 3) Fuzzy model predictive controller

[0201] The controller proposed by the present invention is established based on the following assumptions: network delay d k and the data loss sequence in the CTA channel is limited by a given known τ. In an actual NCS protocol, if a data packet is not received at a certain transmission time, it means that the data packet is lost. To prevent the networked system from becoming an open loop, it is assumed that the number of consecutive lost packets is finite. In addition, with 5G communication technology, information can be exchanged quickly. In summary, the present invention considers that the given known network delay and data loss have upper limits.

[0202] Considering discrete time with an output, from the TS fuzzy system (13), we can obtain:

[0203]

[0204] Based on the information available at the controller side (i.e., the system state estimated by the observer and the STC delay dk calculated by NDC1 and the model (28)), the prediction of k + τ can be calculated as:

[0205]

[0206] Since there is no control over the past states until and the value of d k is calculated by NDC1, by deleting the corresponding rows of the past states, we can obtain:

[0207] X = Ψ + ΘU (29)

[0208] where:

[0209]

[0210]

[0211]

[0212] It should be noted that Ψ contains all the available information belonging to the past, U contains the unknown future control inputs, and MPC should be used for design. From (30), it can be concluded that the dimension of Ψ depends on the upper limit of the CTA delay τ, which is constant. Therefore, the dimension of the vector Ψ is constant. However, if the STC delay dk changes, the elements of Ψ will change. To design the vector U, the following cost function needs to be considered:

[0213]

[0214] where w k+j is a function of the future reference, and the coefficients δ j and λ jDetermine the weights for future behavior and control input effort separately. δ j and λ j are control design parameters that represent the importance of tracking error or control input energy. To find the solution that minimizes (31) analytically, it is necessary to rewrite it in vector representation as follows:

[0215] J = (X - W) T Δ(X - W) + U T ΛU (31)

[0216] where Δ = diag{δ 0 , δ 1 ,..., δ τ} and Λ = diag{λ 0 , λ 1 ,..., λ τ} where diag{.} represents a diagonal matrix, W = [w k w k+2 … w k+τ T . Substituting (30) into (32), we have:

[0217] J = (Ψ + ΘU - W) T Δ(Ψ + ΘU - W) + U T ΛU × U T Θ T ΔΘU + (Ψ - W) T ΔΘU + U T Θ T Δ(Ψ - W)) + (Ψ - W) T Δ(Ψ - W) + U T ΛU = U T HU + KU + U T K T + G (32)

[0218] where:

[0219] H = Θ T ΔΘ + Λ ≥ 0, K = (Ψ - W) T ΔΘ

[0220] G = (Ψ - W) T Δ(Ψ - W)

[0221] The optimization problem is to minimize J with respect to U. This quadratic problem has an analytical solution:

[0222] U = (Θ T ΔΘ + Λ)Θ T Δ(Ψ - W) (33)

[0223] ​The solution for the control input is given by Equation (34), which theoretically minimizes the cost function (31).

[0224] Remark 1: (Dimensions of the matrix in Equation (34) and computational burden for calculating the control input): Since there is only one control input for the considered DC microgrid, Θ T ΔΘ + Λ ∈ R τ×τ . Therefore, the dimensions of the matrix Θ T ΔΘ + Λ are a function of the upper bound of the CTA link delay and are independent of the number of system states. Thus, for a more complex DC microgrid with multiple CPLs, the computational burden of the matrix inversion calculation (Θ T ΔΘ + Λ) -1 does not increase significantly. On the other hand, the dimension of the matrix Θ ∈ R nτ×τ depends on the number of system states and the CTA delay. The matrix Ψ is composed of another matrix with dimensions nτ × d k . Therefore, by increasing the STC delay, the computational amount of the matrix Ψ increases because matrix multiplication is required Therefore, the computational burden of the method proposed in the present invention mainly depends on τ, n, and d k .

[0225] Remark 2: (Design procedure of the proposed controller): The procedure for the proposed controller and the closed-loop DC microgrid system is as Figure 4 shown. In each instance of the controller design, NDC1 checks whether data has been received. It selects the latest timestamp data from the sensors and measures the STC delay d k . On the other hand, if NDC1 has not received any new data, it retains the latest data and increments the delay d k by 1. Then, the observer (16) uses the last available output provided by NDC1 and its corresponding control input The latter can be obtained from the MPC. The output of the observer is the best estimate of the system state. Based on the estimated available y (k-dk) and the upper bound of the CTA delay τ, the matrices Ψ and Θ defined in (30) are calculated. Thus, the sequence U of future control inputs is obtained from (34). The timestamp packet of U is transmitted over the network to the actuator. Then, NDC2 selects the latest timestamp data from the controller, evaluates the CTA delay τ k , and applies the control input to the system. If NDC1 and / or NDC2 do not receive any data in some cases, they use the latest data and increment the delay d k and / or τ k by one.

[0226] Remark 3: (Applicability of the method proposed in this invention to different DC microgrid topologies and loads): This invention considers a class of loads that consume constant power. The reason is that such loads have non-linear properties and are introduced in different practical applications, including more electric vehicles, airplanes, and ships. However, the proposed method is applicable to different DC microgrid topologies and other loads with linear and non-linear behaviors. For any linear or non-linear DC microgrid with a given state-space representation, an equivalent TS fuzzy model can be systematically derived. Note that for linear dynamics, the TS fuzzy model will have only one rule and the same system matrix as the linear dynamics. Then, based on Figure 4 the design procedure provided in

[0227] In summary, the fuzzy observer is represented by the following formula:

[0228]

[0229] L = P 1 -1 M;

[0230] In the formula: represents the output vector matrix; C represents the undetermined coefficient matrix of the output equation; M represents the observer gain coefficient; represents the state of the observer; and represent the state vector and output vector of the fuzzy observer respectively; L represents the gain of the fuzzy observer; represents the estimation error; dt represents the known time-varying network sensor-to-controller delay calculated by the network delay compensator; P1 represents the Lyapunov matrix.

[0231] The fuzzy predictive controller is represented by the following formula:

[0232]

[0233]

[0234] X = Ψ + ΘU;

[0235] U = (Θ T ΔΘ + Λ)Θ T Δ(Ψ - W);

[0236]

[0237]

[0238]

[0239]

[0240] In the formula: h represents a function, y is the output vector; x represents the state space; u represents the input vector; E h represents the undetermined coefficient matrix of the discrete TS fuzzy system; C xk represents the undetermined coefficient matrix of the output equation; represents estimating the system state at the k - dk+1 moment based on the available information at the k - dk moment; both represent the undetermined coefficients in the Ψ matrix; x ok represents the state of the observer after discretization; · k represents the sequence after time discretization, corresponding to different moments; Ψ represents the matrix containing all past information; U represents the future control input matrix; X represents the system state matrix; Θ represents the undetermined coefficient matrix; Λ and Δ represent matrix operation symbols; represents estimating the system state at the k moment based on the available information at the j moment.

[0241] When designing the future control input matrix U of the fuzzy predictive controller, the following cost function needs to be considered:

[0242]

[0243] Transform the cost function into vector form J = (X - W) T Δ(X - W) + U T ΛU;

[0244] Δ = diag{δ 0 , δ 1 ,..., δ τ}

[0245] Λ = diag{λ 0 , λ 1 ,..., λ τ};

[0246] W = [w k w k+2 …w k+τ T ;

[0247]

[0248] H = Θ T ΔΘ + Λ ≥ 0;

[0249] K = (Ψ - W) T ΔΘ;

[0250] G = (Ψ - W) T Δ(Ψ - W);

[0251] ​Where: J represents the cost function; w k+j represents the function of future reference; the coefficient δ j and λ j are respectively used to determine the weights of future behavior and control input effort, which are control design parameters and represent the importance of tracking error or control input energy; λ j represents the weight of the cost function; W represents the future reference of the system; represents estimating the system state at k - d k moment according to the available information at k - d k+1 moment; H, K, and G are constructed discrimination matrices.

[0252] Specifically, it includes the following steps:

[0253] S401: Calculate the STC delay d at k moment through the first network delay compensator k , and obtain the control output of the DC microgrid at k - d k moment Then, obtain the control input at k - d k moment through the future control input matrix U of the fuzzy prediction controller

[0254] S402: Take the control input k and control output at k - d moment as the input of the fuzzy observer, and output the system state at k - d k moment through the fuzzy observer

[0255] S403: Take the system state as the input of the fuzzy prediction controller, and output the corresponding future control input u k through the fuzzy prediction controller, and update the future control input matrix U;

[0256] S404: Obtain the CTA delay τ k through the second network delay compensator, and further determine the future control input at k + τ k moment based on the updated future control input matrix U as the future grid control quantity of the DC microgrid.

[0257] In the specific implementation process, in step S403, bring the control input control output and the upper limit of the CTA delay into the formula of the fuzzy prediction controller, calculate the corresponding matrices Ψ and Θ; then update the system state matrix X through the system state , and calculate and update the future control input matrix U in combination with the following formula;

[0258] X = Ψ + ΘU;

[0259] Where: Ψ represents a matrix containing all past information; U represents a future control input matrix; X represents a system state matrix; Θ represents a matrix of undetermined coefficients.

[0260] In the specific implementation process, if the first network delay compensator or the second network delay compensator fails to obtain data, then the STC delay d k or the CTA delay τ k is incremented by one.

[0261] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Those of ordinary skill in the art should understand that any modifications or equivalent replacements made to the technical solutions of the present invention without departing from the spirit and scope of the present technical solutions shall be covered by the scope of the claims of the present invention.

Claims

1. A dynamic stability prediction control method for a DC microgrid based on a fuzzy observer, characterized in that, it includes: S1: Establish a dynamic model of the DC microgrid; S2: Based on the dynamic model of the DC microgrid, combine the sector nonlinear method to establish a T-S fuzzy model of the DC microgrid; S3: Construct a corresponding fuzzy observer and a fuzzy predictive controller based on the T-S fuzzy model of the DC microgrid; In step S3, the fuzzy predictive controller is expressed by the following formula: X = Ψ + ΘU; U = (Θ T ΔΘ + Λ)Θ T Δ(Ψ - W); where: h represents a function; y is the output vector; x represents the state space; u represents the input vector; E h represents the undetermined coefficient matrix of the discrete TS fuzzy system; C xk represents the undetermined coefficient matrix of the output equation; represents estimating the system state at time k-d k using the available information at time k-d k+1 ; both represent the undetermined coefficients in the Ψ matrix; x ok represents the state of the observer after discretization; · k represents the sequence after time discretization, corresponding to different times; Ψ represents the matrix containing all past information; U represents the future control input matrix; X represents the system state matrix; Θ represents the undetermined coefficient matrix; Λ and Δ represent the operation symbols of matrices; In step S3, a first network delay compensator for calculating the STC delay is set between the input of the fuzzy observer and the DC microgrid, and a second network delay compensator for calculating the CTA delay is set between the fuzzy predictive controller and the DC microgrid; S4: Input the grid signal of the DC microgrid at the current moment into the fuzzy observer to output the corresponding system state; then input the system state into the fuzzy predictive controller to output the corresponding future control input; finally, solve the future grid control quantity based on the future control input, and further realize the control of the DC microgrid based on the future grid control quantity; In step S4, it specifically includes the following steps: S401: Calculate the STC delay d at time k through the first network delay compensator k , and obtain the control output of the DC microgrid at time k - d k Then, obtain the control input at time k - d through the future control input matrix U of the fuzzy predictive controller k ​​ S402: Take the control input and control output at the k-d k moment as the inputs of the fuzzy observer, and output the system state at the k-d moment through the fuzzy observer k ​​ S403: Take the system state as the input of the fuzzy predictive controller, and output the corresponding future control input u k through the fuzzy predictive controller, and update the future control input matrix U; S404: Obtain the CTA delay τ through the second network delay compensator k , and then determine k + τ based on the updated future control input matrix U k for the future control input at time as the future grid control quantity of the DC microgrid.

2. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 1, characterized in that: In step S1, the dynamic model of the DC microgrid is expressed by the following formula: Among them, Wherein: represents the fitted value of the state - space vector; h represents a function; represents the fitted value of the voltage of the DC power supply; represents the transpose of the Q - th CPL state; h Q represents the h - function of the Q - th CPL state; r s represents the resistance of the filter connected to the DC power supply; represents the fitted value of the voltage of the capacitor in the j - th CPL; v C0,j represents the initial voltage of the capacitor in the j - th CPL; x s = [i L,s v C,s T , i L,s and v C,s respectively represent the inductor current and capacitor voltage in the DC micro - grid energy storage system; x j = [i L,j v C,j T , and respectively represent the inductor current and capacitor voltage in the j - th constant - power load of the DC micro - grid; s = Q + 1, Q represents the number of constant - power load systems in the DC micro - grid energy storage system; A, B, C, D represent undetermined matrix coefficients; a, b, c, d represent undetermined coefficients; L s represents the inductance of the filter connected to the DC source; i es represents the injection current of the energy storage system; V dc represents the voltage of the DC power supply; C s represents the capacitance of the filter connected to the DC source; represents DC parameters; represents AC parameters; represents a vector; ·(t) represents being related to an event; · T represents transpose; · s represents being related to the DC power supply; · e represents being related to the energy storage system; · Q represents the Q - th; · j represents the j - th.​​ 3. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 2, characterized in that: In step S2, the T-S fuzzy model of the DC microgrid is expressed by the following formula: Among them, In the formula: represents the equivalent TS fuzzy representation of state x; represents the fitting value of the input vector; represents the undetermined coefficient equivalent to the T-S fuzzy model; represents the voltage fitting value of the DC power supply; d Q represents the undetermined coefficient corresponding to the Qth CPL; U k1,1 represents the input quantity of the 1st CPL after normalization; U kQ,Q represents the input quantity of the Qth CPL after normalization; x represents the state space; k 1 、k 2 、k Q represent i after binary decomposition; u represents the input vector; m i represents the normalized membership function of the ith rule; i = 1 + k 1 ×2 0 +k 2 ×2 1 +…+k p ×2 p-1 ∈{1,…,2 Q} represents the number of linear models j 1 ∈{1,...,q}, k j1 ∈{0,1} represents the jth 1 bit in the Q-bit binary representation of i - 1; represents the superposition of fuzzy levels from 1 to Q.

4. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 3, characterized in that: In step S3, the fuzzy observer is expressed by the following formula: L = P 1 -1 M; In the formula: represents the output vector matrix; C represents the undetermined coefficient matrix of the output equation; M represents the observer gain coefficient; represents the state of the observer; and respectively represent the state vector and output vector of the fuzzy observer; L represents the gain of the fuzzy observer; dt represents the known time-varying delay from the network sensor to the controller calculated by the network delay compensator; P1 represents the Lyapunov matrix.

5. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 1, characterized in that: When designing the future control input matrix U of the fuzzy predictive controller, the following cost function needs to be considered: Transform the cost function into vector form \(J=(X - W)\) T \(\Delta(X - W)+U\) T \(\Lambda U\); Δ = diag{δ 0 , δ 1 ,..., δ τ}A Λ = diag{λ 0 , λ 1 ,..., λ}; W = [w k w k+2 … w k+τ T ;​ H = Θ T ΔΘ + Λ ≥ 0; K = (Ψ - W) T ΔΘ; G = (Ψ - W) T Δ(Ψ - W); Where: J represents the cost function; represents the function of future reference; the coefficients and are respectively used to determine the weights of future behavior and control input effort, which are control design parameters and represent the importance of tracking error or control input energy; W represents the future reference of the system; represents estimating the system state at k-d k moment according to the available information at k-d k+1 moment; H, K, G are constructed discriminant matrices.

6. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 1, characterized in that: In step S403, the control input control output and the upper limit of the CTA delay are substituted into the formula of the fuzzy prediction controller to calculate the corresponding matrices Ψ and Θ; furthermore, through the system state update the system state matrix X, and calculate and update the future control input matrix U in combination with the following formula; X = Ψ + ΘU; In the formula: Ψ represents a matrix containing all past information; U represents a future control input matrix; X represents a system state matrix; Θ represents a matrix of undetermined coefficients.

7. The dynamic stability prediction control method for a DC microgrid based on a fuzzy observer according to claim 1, characterized in that: If the first network delay compensator or the second network delay compensator fails to obtain data, the STC delay d k or the CTA delay τ k is incremented by one.

Images