Robust two-level control method for islanded microgrid based on communication network and random switching topology

Abstract

The present invention relates to a robust secondary control method for an isolated microgrid based on a communication network and a random switching topology, including: constructing an asynchronous distributed secondary controller for an isolated microgrid system based on a hierarchical control principle; introducing an event trigger mechanism as network communication to construct a networked asynchronous distributed secondary controller; establishing a global frequency error system and a global power level error system; deriving the stability conditions of the global frequency error system and the global power level error system using a Lyapunov stability method; and testing and verifying the networked distributed secondary controller. The present invention can reduce the impact of link failures on the system, not only can achieve the desired control objectives, but also can effectively reduce some network constraints. In the secondary control layer, a sliding mode control is additionally designed, focusing on resisting external interference and matching uncertainty in the transmission channel. Technically, the sliding mode control can be designed simultaneously with the secondary control, and is easy to implement.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid control, and in particular to a robust secondary control method for an island microgrid based on a communication network and a random switching topology. Background Art

[0002] Renewable energy, environmental protection and power supply security are current issues in the development of the power system, which have promoted the rapid development of microgrid systems in recent years. Generally speaking, there are two operating modes of microgrid systems: grid-connected and islanded. When a large power grid fails, the grid-connected microgrid will be out of the control of the large power grid and maintain the power demand of key loads in island mode to ensure the reliability of its power supply.

[0003] Hierarchical control is the most common control strategy in the current isolated microgrid. Under the idea of ​​hierarchical control, the control of the isolated microgrid is divided into a multi-layer architecture. For the primary control layer, droop control is generally used to complete the initial adjustment of voltage and frequency, as well as the distribution of power output. Secondary control plays a vital role in improving control accuracy. It is mainly used to compensate for the deviation caused by primary control. For the secondary control layer of the microgrid, compared with the traditional centralized control, the distributed control idea has higher flexibility and security, and has become the mainstream of more research.

[0004] Distributed control has a good tolerance for failures of each unit in an isolated microgrid, but the frequent occurrence of such phenomena will have a certain impact on the stability and performance of the microgrid system. It is worth mentioning that the communication topology in the microgrid is prone to change under the constraints of the actual environment, while most existing studies believe that the communication topology is fixed or switched in any form. In addition, the external environment always affects the operation of the microgrid system to a greater or lesser extent. Therefore, how to improve the robustness of the microgrid operation is also full of research significance. Sliding mode control is an effective method, but the sliding mode control introduced in existing studies has a certain complexity, and its stability analysis process is relatively cumbersome.

[0005] In addition, microgrids are important applications of cyber-physical systems, and a large number of research results have shown that event triggering mechanisms can effectively reduce the communication burden while maintaining satisfactory performance. The current main research direction is how to improve the traditional event triggering mechanism while ensuring good control effects and further reduce the communication burden. In addition, network security is another key issue that needs to be considered in information transmission. Network attacks can cause communication information to be stolen or tampered with, resulting in system performance degradation or even system crashes. It can be said that improving communication efficiency and network attack phenomena are issues that need to be considered in the current microgrid secondary control, and the existing research results still have room for improvement in the above-mentioned issues, such as the single description of the microgrid communication topology and the complexity of the designed sliding mode control. Summary of the invention

[0006] In order to solve the problem that the current research on secondary control of isolated microgrids needs to improve the description of each DG communication topology and the control robustness, the purpose of the present invention is to provide a robust secondary control method for isolated microgrids based on communication networks and random switching topologies, which can reduce the impact of link failures on the system, effectively reduce some network constraints, and enhance the robustness of the power grid during operation.

[0007] To achieve the above object, the present invention adopts the following technical solution: a robust two-level control method for an island microgrid based on a communication network and a random switching topology, the method comprising the following steps in order:

[0008] (1) Construct an asynchronous distributed two-level controller for the island microgrid system based on the hierarchical control principle;

[0009] (2) Introducing an event trigger mechanism as network communication, considering the impact of false data injection attacks in information transmission, and building a networked asynchronous distributed secondary controller for the isolated microgrid system;

[0010] (3) Based on the networked asynchronous distributed two-level controller, the frequency error and power level error are defined, the sliding mode control is designed, and the global frequency error system and the global power level error system are established;

[0011] (4) Using the Lyapunov stability method, derive the stability conditions of the global frequency error system and the global power level error system;

[0012] (5) Test and verify the networked distributed secondary controller.

[0013] The step (1) specifically refers to: in the isolated AC microgrid model, the communication network relationship between each distributed power source and its neighbors is described by a topology graph model. in and Represents the node set and edge set respectively. If there is an edge defined as ε mn =(m,n)∈ε, then node n is called the neighbor of node m. Denote all neighbors of node i, Defined as a graph The adjacency matrix of mm =1, if a mn =1, then ε mn ∈ε, otherwise a mn =0; Representation diagram The degree matrix of Note: The Laplace matrix of ; the path is the sequence of edges in the graph. To ensure the convergence of the consistency algorithm, the graph It is connected, and there is a path between any nodes;

[0014] Let {θ(t)} be a right-continuous Markov process defined in a standard probability space, whose values ​​are contained in the finite set In the example, the transfer rate matrix of {θ(t)} is Defined as:

[0015]

[0016] Among them, κ>0,lim Λ＞0 (o(Λ) / Λ)=0,π rq It is expressed as the transition rate from mode r to mode q, and satisfies

[0017] Introducing λ(t) as an asynchronous mode, the relationship between θ(t) and λ(t) is given by the conditional probability describe;

[0018] π rs =Pr{λ(t)=s∣θ(t)=r} (2)

[0019] in, for and

[0020] For the primary control layer of the island microgrid, the droop control strategy is:

[0021]

[0022] Where, m is the mth distributed power source; ω m and V m Represent the frequency and voltage amplitude respectively; and Represent the nominal values ​​of frequency and voltage respectively; P m and Q m Represent active output and reactive output respectively; α m and β m represents the droop coefficient;

[0023] For frequency control, remember Indicates auxiliary frequency control input; the nominal frequency value is:

[0024]

[0025] The power balance equation is:

[0026]

[0027] in, Indicates the total actual output power of all distributed power sources; Represents the overall load demand; P loss Represents the total real power loss, which is given by the formula The power loss is estimated, where is the power loss rate, which is 5% to 7%. Each distributed power source reasonably allocates the corresponding real power according to the maximum power limit it can generate and the actual load demand. The reference real power value is expressed as: in, represents the maximum generated power of the mth distributed generation; Represents the desired power allocation level; introduces a local variable And define For auxiliary power control input, the new frequency nominal value is calculated as follows:

[0028]

[0029] Under any initial frequency and power value, frequency recovery and accurate active power distribution of all distributed power sources can be achieved:

[0030]

[0031] Among them, ω ref and For a given reference frequency and desired power utilization level, according to the topology diagram and Markov process, the asynchronous distributed two-level controller constructed based on the hierarchical control principle is:

[0032]

[0033] Among them, ω m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Indicates the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t).

[0034] The step (2) specifically refers to:

[0035] For frequency control, the triggering time of the mth distributed generation satisfies the following rules:

[0036]

[0037] Among them, d m is a positive integer; t k is the sampling time, is the sampling time of the frequency data of the mth distributed power source; l is the sampling period; Θ ω is the event trigger weight matrix; Indicates the frequency data currently sampled by the mth distributed power source Frequency data with latest transmission The error between

[0038]

[0039]

[0040] Among them, ω ref represents the reference frequency signal; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, The latest transmission frequency of the neighbor of the mth distributed generation satisfies

[0041]

[0042] Among them, ζ, ψ, ε represent the basic threshold parameters for event triggering; tanh(·) is used to adjust the threshold value in real time;

[0043] Get the event trigger mechanism for the frequency signal:

[0044]

[0045] The power signal and the frequency signal select the same transmission channel to obtain the event trigger mechanism of the power signal:

[0046]

[0047] Divide the trigger interval: Define the error function:

[0048]

[0049] Considering the impact of false data injection attacks in information transmission, false data injection attacks are based on nonlinear functions. The data information is falsely injected, and the frequency signal transmission data received by the mth distributed power source from its neighbor n under the false data injection attack is:

[0050]

[0051] Among them, μ(t)∈{0,1} is a Bernoulli random variable, which is used to indicate whether the false data injection attack occurs in the network transmission channel;

[0052] Similarly, for power signal transmission, after false data injection attack The power signal transmission data received by the mth distributed power source from its neighbor n is:

[0053]

[0054] Therefore, combined with the event trigger mechanism, a networked asynchronous distributed secondary controller based on false data injection attack is constructed for the isolated microgrid system, including and

[0055]

[0056] in: It is a networked asynchronous distributed secondary frequency controller. It is a networked asynchronous distributed two-stage power controller; m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Represents the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; ω ref and γ refFor a given reference frequency and desired power utilization level; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and It is the transmission data of the frequency signal and power level signal of the mth distributed power source under the event trigger mechanism; and is the transmission data of the frequency signal and power level signal received by the mth distributed power source under the false data injection attack from its neighbor nth distributed power source; t k is the sampling time, and l is the sampling period.

[0057] The step (3) specifically refers to:

[0058] Define the frequency error vector The frequency error system is constructed as:

[0059]

[0060]

[0061] in:

[0062]

[0063]

[0064]

[0065] Among them, v ω (t) is the external disturbance signal, y ω (t) is the output signal, is satisfied The norm of a bounded nonlinear function is are known parameters, is the adjacency matrix, is the Laplacian matrix, represents the gain of the secondary controller;

[0066] For sliding mode control, construct the sliding surface:

[0067]

[0068] Among them, c ω are known parameters;

[0069] Establish asynchronous sliding mode control law:

[0070]

[0071] is the sliding mode controller gain, is the robust term gain;

[0072] Substituting the asynchronous sliding mode control law into the frequency error system, we get the global frequency error system:

[0073]

[0074] Similarly, define the power level error The global power level error system is obtained as:

[0075]

[0076] in, and are the frequency error and the power level error; and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and is false data; μ(t) is Bernoulli distribution; and is the error function in the event triggering mechanism; t k is the sampling time; l is the sampling period.

[0077] The step (4) specifically refers to: using the Lyapunov stability method to perform stability analysis on the global frequency error system and the global power level error system:

[0078] Conclusion 1: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix A r , the Laplacian matrix as well as Then, given the attack parameters μ and G ω , event trigger parameter ζ m and ψ m , the upper bound of nonlinear function Output parameters in equation H Infinity Performance Level γ ω , the identity matrix I, if there exists a positive definite matrix P rω , event trigger matrix Θ ω >0, secondary controller gain and the sliding mode controller gain Make any and The following inequality (22) holds true, then the constructed asynchronous robust two-stage frequency controller has an external disturbance v ω (t) and nonlinear The frequency recovery of each distributed power source under any initial conditions is achieved at the same time:

[0079]

[0080] in:

[0081]

[0082]

[0083]

[0084]

[0085] Conclusion 2: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix Laplacian Matrix as well as Then, given the attack parameters μ and Event trigger parameter m and ψ m , the upper bound of nonlinear function Output parameters in equation H Infinity Performance Level The identity matrix I, if there exists a positive definite matrix Event Trigger Matrix Secondary controller gain and the sliding mode controller gain Make any and The following inequality (23) holds true, then the constructed asynchronous robust two-stage power controller has the following characteristics in the presence of external disturbances: and nonlinear The accurate real power recovery of each distributed power source under any initial conditions is achieved at the same time:

[0086]

[0087] in:

[0088]

[0089]

[0090]

[0091]

[0092] It can be seen from the above technical scheme that the beneficial effects of the present invention are: First, since the communication topology of the isolated microgrid often exhibits a probabilistic switching phenomenon based on statistical data in practice, the present invention attempts to characterize this switching topology with a Markov process. Based on this idea, if the microgrid system can consider in advance the impact of changes in its communication topology, the impact of link failures on the system can be reduced; second, based on the multi-agent consistency protocol, the present invention creates an asynchronous distributed secondary controller within the network framework. Its asynchronous idea aims to solve the challenge of accurately acquiring the mode. From the control effect point of view, the designed secondary control can not only achieve the desired control goal, but also effectively reduce some network constraints, such as communication load and the impact of network attacks; third, in order to enhance the robustness of the power grid during operation, the present invention additionally designs a sliding mode control in the secondary control layer, focusing on resisting external interference and matching uncertainty in the transmission channel. Technically speaking, the sliding mode control can be designed simultaneously with the secondary control and is easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 is a flow chart of the introduced event triggering mechanism;

[0094] Figure 2 It is a schematic diagram of the test island microgrid system;

[0095] Figure 3 It is a switching communication topology diagram with two modes;

[0096] Figure 4 is the modal evolution diagram of the communication topology in the microgrid system;

[0097] Figure 5 is the modal evolution diagram of the asynchronous distributed two-stage controller;

[0098] Figure 6 It is a set of false data injection attack sequence diagrams based on Bernoulli distribution;

[0099] Figure 7 It is a schematic diagram for verifying the frequency value of the networked asynchronous distributed two-level controller in three stages;

[0100] Figure 8 It is a schematic diagram to verify the power level values ​​of the networked asynchronous distributed two-stage controller in three stages;

[0101] Fig. 9 It is a schematic diagram for verifying the output power value of the networked asynchronous distributed two-stage controller in three stages;

[0102] Fig.10 is a schematic diagram of the frequency error and power level error values ​​when sliding mode control is not applied;

[0103] Fig.11 is a schematic diagram of the frequency error and power level error values ​​when sliding mode control is applied;

[0104] Fig.12 It is a schematic diagram of the release time and interval changes of the dynamic event trigger mechanism. DETAILED DESCRIPTION

[0105] A robust two-level control method for an islanded microgrid based on a communication network and a random switching topology, the method comprising the following steps in order:

[0106] The method comprises the following steps in order:

[0107] (1) Construct an asynchronous distributed two-level controller for the island microgrid system based on the hierarchical control principle;

[0108] (2) Introducing an event trigger mechanism as network communication, considering the impact of false data injection attacks in information transmission, and building a networked asynchronous distributed secondary controller for the isolated microgrid system;

[0109] (3) Based on the networked asynchronous distributed two-level controller, the frequency error and power level error are defined, the sliding mode control is designed, and the global frequency error system and the global power level error system are established;

[0110] (4) Using the Lyapunov stability method, derive the stability conditions of the global frequency error system and the global power level error system;

[0111] (5) Test and verify the networked distributed secondary controller.

[0112] The step (1) specifically refers to: in the isolated AC microgrid model, the communication network relationship between each distributed power source and its neighbors is described by a topology graph model. in and Represents the node set and edge set respectively. If there is an edge defined as ε mn =(m,n)∈ε, then node n is called the neighbor of node m. Denote all neighbors of node i, Defined as a graph The adjacency matrix of mm =1, if a mn=1, then ε mn ∈ε, otherwise a mn =0; Representation diagram The degree matrix of Note: The Laplace matrix of ; the path is the sequence of edges in the graph. To ensure the convergence of the consistency algorithm, the graph It is connected, and there is a path between any nodes;

[0113] Generally speaking, information exchange between distributed power sources is achieved through their respective topological structures. This topology may change as the environment changes. Consider a situation where there is an interconnection between node A and node B, and at a certain time point, the line from A to B is cut off due to the external environment or its own structure. In this case, the fault detection device itself may troubleshoot through the safety device, but this may take some time to repair. And where the fault occurs, there is still a possibility of continued failure even after repair. In this way, through a period of experimental observation, combined with statistical data, a standard Markov model can be established to describe the above process.

[0114] Let {θ(t)} be a right-continuous Markov process defined in a standard probability space, whose values ​​are contained in the finite set In the example, the transfer rate matrix of {θ(t)} is Defined as:

[0115]

[0116] Among them, κ>0,lim Λ＞0 (o(Λ) / Λ)=0,π rq It is expressed as the transition rate from mode r to mode q, and satisfies

[0117] In addition, it is challenging to obtain real-time modal information in practice. Therefore, when building a controller for a Markov system, the controller may not be able to obtain the modal information of the current system in a timely manner. In view of this, the present invention adopts the idea of ​​asynchronous control in the controller design. The controller can estimate the modal evolution of the system based on conditional probability, so that the expected controller mode always maintains asynchronous properties and further maintains an indirect association with the system.

[0118] Introducing λ(t) as an asynchronous mode, the relationship between θ(t) and λ(t) is given by the conditional probability describe;

[0119] π rs=Pr{λ(t)=s∣θ(t)=r} (2)

[0120] in, for and

[0121] For the primary control layer of the island microgrid, the droop control strategy is:

[0122]

[0123] Where, m is the mth distributed power source; ω m and V m Represent the frequency and voltage amplitude respectively; and V0 m Represent the nominal values ​​of frequency and voltage respectively; P m and Q m Represent active output and reactive output respectively; α m and β m represents the droop coefficient;

[0124] For frequency control, remember Indicates auxiliary frequency control input; the nominal frequency value is:

[0125]

[0126] The power balance equation is:

[0127]

[0128] in, Indicates the total actual output power of all distributed power sources; Represents the overall load demand; P loss Represents the total real power loss, which is given by the formula The power loss is estimated, where is the power loss rate, which is 5% to 7%. Each distributed power source reasonably allocates the corresponding real power according to the maximum power limit it can generate and the actual load demand. The reference real power value is expressed as: in, represents the maximum generated power of the mth distributed generation; Represents the desired power allocation level; introduces a local variable And define For auxiliary power control input, the new frequency nominal value is calculated as follows:

[0129]

[0130] Under any initial frequency and power value, frequency recovery and accurate active power distribution of all distributed power sources can be achieved:

[0131]

[0132] Among them, ω ref and For a given reference frequency and desired power utilization level, according to the topology diagram and Markov process, the asynchronous distributed two-level controller constructed based on the hierarchical control principle is:

[0133]

[0134] Among them, ω m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Indicates the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t).

[0135] like Figure 1 As shown, the step (2) specifically refers to:

[0136] For frequency control, the triggering time of the mth distributed generation satisfies the following rules:

[0137]

[0138] Among them, d m is a positive integer; t k is the sampling time, is the sampling time of the frequency data of the mth distributed power source; l is the sampling period; Θ ω is the event trigger weight matrix; Indicates the frequency data currently sampled by the mth distributed power source Frequency data with latest transmission The error between

[0139]

[0140]

[0141] Among them, ω ref represents the reference frequency signal; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, The latest transmission frequency of the neighbor of the mth distributed generation satisfies

[0142] Different from the classic event trigger mechanism, the trigger condition adopted by the present invention has a dynamic threshold value, and the size of the threshold value can be changed in real time.

[0143]

[0144] Among them, ζ, ψ, ε represent the basic threshold parameters for event triggering; tanh(·) is used to adjust the threshold value in real time; since this dynamic threshold contains the state of the system, it can be modified according to the change of the state, which is determined by the properties of the hyperbolic tangent function tanh(·). Then, combining the above conditions, we can get the event triggering mechanism for frequency signals:

[0145]

[0146] The power signal and the frequency signal select the same transmission channel to obtain the event trigger mechanism of the power signal:

[0147]

[0148] Divide the trigger interval: Define the error function:

[0149]

[0150] Considering the impact of false data injection attacks in information transmission, false data injection attacks are based on nonlinear functions. The data information is falsely injected, and the frequency signal transmission data received by the mth distributed power source from its neighbor n under the false data injection attack is:

[0151]

[0152] Among them, μ(t)∈{0,1} is a Bernoulli random variable, which is used to indicate whether the false data injection attack occurs in the network transmission channel;

[0153] Similarly, for power signal transmission, after false data injection attack The power signal transmission data received by the mth distributed power source from its neighbor n is:

[0154]

[0155] Therefore, combined with the event trigger mechanism, a networked asynchronous distributed secondary controller based on false data injection attack is constructed for the isolated microgrid system, including and

[0156]

[0157] in: It is a networked asynchronous distributed secondary frequency controller. It is a networked asynchronous distributed two-stage power controller; m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Represents the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; ω ref and γ ref For a given reference frequency and desired power utilization level; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and It is the transmission data of the frequency signal and power level signal of the mth distributed power source under the event trigger mechanism; and is the transmission data of the frequency signal and power level signal received by the mth distributed power source under the false data injection attack from its neighbor nth distributed power source; t k is the sampling time, and l is the sampling period.

[0158] The step (3) specifically refers to:

[0159] Define the frequency error vector The frequency error system is constructed as:

[0160]

[0161]

[0162] in:

[0163]

[0164]

[0165]

[0166] Among them, v ω (t) is the external disturbance signal, y ω (t) is the output signal, is satisfied The norm of a bounded nonlinear function is are known parameters, is the adjacency matrix, is the Laplacian matrix, represents the gain of the secondary controller;

[0167] For sliding mode control, construct the sliding surface:

[0168]

[0169] Among them, c ω are known parameters;

[0170] Establish asynchronous sliding mode control law:

[0171]

[0172] is the sliding mode controller gain, is the robust term gain;

[0173] Substituting the asynchronous sliding mode control law into the frequency error system, we get the global frequency error system:

[0174]

[0175] Similarly, define the power level error The global power level error system is obtained as:

[0176]

[0177] in, and are the frequency error and the power level error; and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and is false data; μ(t) is Bernoulli distribution; and is the error function in the event triggering mechanism; t k is the sampling time; l is the sampling period.

[0178] Considering the actual working environment of the microgrid, there are often environmental factors such as nonlinear factors and the influence of external disturbances in the system. For these reasons, a nonlinear term and an external disturbance are considered in the frequency error system.

[0179] The step (4) specifically refers to: using the Lyapunov stability method to perform stability analysis on the global frequency error system and the global power level error system:

[0180] Conclusion 1: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix Laplacian Matrix as well as Then, given the attack parameters μ and G ω , event trigger parameter ζ m and ψ m , the upper bound of nonlinear function Output parameters in equation H Infinity Performance Level γ ω , the identity matrix I, if there exists a positive definite matrix P r ω , event trigger matrix Θ ω >0, secondary controller gain and the sliding mode controller gain Make any and The following inequality (22) holds true, then the constructed asynchronous robust two-stage frequency controller has an external disturbance v ω (t) and nonlinear The frequency recovery of each distributed power source under any initial conditions is achieved at the same time:

[0181]

[0182] in:

[0183]

[0184]

[0185]

[0186]

[0187] Conclusion 2: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix A r , the Laplacian matrix as well as Then, given the attack parameters μ and Event trigger parameter m and ψ m , the upper bound of nonlinear function Output parameters in equation H Infinity Performance Level The identity matrix I, if there exists a positive definite matrix Event Trigger Matrix Secondary controller gain and the sliding mode controller gain Make any and The following inequality (23) holds true, then the constructed asynchronous robust two-stage power controller has an external disturbance and nonlinear The accurate real power recovery of each distributed power source under any initial conditions is achieved at the same time:

[0188]

[0189] in:

[0190]

[0191]

[0192]

[0193]

[0194] Based on an improved IEEE bus test system, a set of actual system parameters are given, and simulation tests are carried out using MATLAB software to verify that it can be used for secondary control of isolated island microgrid systems and has the expected control performance.

[0195] The following combination Figures 1 to 12 The present invention is further described.

[0196] Embodiment 1

[0197] The method of the present invention is simulated and analyzed as follows. In order to verify the effectiveness of the present invention, an improved IEEE 34 bus test system is used as the modeling object. Figure 2 As shown, the microgrid system is connected to 5 distributed power sources and 3 loads. In addition, the present invention considers that the island microgrid system under study has two switching structures: Figure 3 Different topological connections for the studied microgrid system, Figure 4 , Figure 5 A set of possible modal evolution processes are given. Based on the given topological structure and the relevant knowledge of graph theory, the matrix and Here, the present invention assumes that both the distributed power sources DG1 and DG2 can obtain the frequency reference value ω in both topologies. ref and the desired power utilization level That is, b1=b2=1.

[0198] In addition, consider that the microgrid system has the following parameters: the frequency reference value is f ref =50HZ, the initial frequency of each distributed power source is f(0)=49.5HZ (ω=2πf); the power values ​​of the three connected loads are set to: The maximum power limit that distributed power can generate They are: 5MW, 8MW, 10MW, 12MW and 9MW respectively; the initial active power output P m (0) is: 1MW, 1MW, 1.5MW, 1.5MW and 2MW. Therefore, some necessary parameters can be calculated: the expected power utilization level is Initial power utilization level are 0.1905, 0.1190, 0.1429, 0.1190 and 0.2116; the droop coefficient is calculated as where f max =51HZ and f min =49HZ.

[0199] like Figure 3 As shown, the present invention assumes that the information communication of each distributed power source has a switching topology. Here, a two-mode example is used to demonstrate it. The two mode distributions include a normal operating state and a state for a fault condition, for example, mode 1 is a normal mode and mode 2 is a fault mode. Figure 3 The one-way communication between DG 1 and DG 5 (mode 2) can be considered as a fault, and then the protection device can repair it within a period of time. This is a typical switching situation. The probability of this switching can be fully predicted using statistical data. Here, if the transfer matrix is ​​established Then we can use Markov process to describe it. In addition, considering the asynchronous control designed, we introduce a conditional probability matrix To describe the relationship between controller modes and system modes. Figure 4 , 5 A set of possible system modes and controller mode evolution processes in this embodiment are given.

[0200]

[0201]

[0202] In addition, through the matrix convex optimization technique, the following controller gains can ensure that Conclusion 1 and Conclusion 2 are feasible, And the attack function satisfy The event trigger parameter is set to m =0.008,ψ m =10 and ε m = 1. In addition, Figure 6 The occurrence of false data injection attack is also simulated, where "1" indicates the presence of an attack signal.

[0203] Some simulation results will be given later to more directly illustrate the effectiveness of the proposed method. The entire simulation process lasts 60 seconds and is divided into the following three stages.

[0204] Phase 1: verify the effectiveness of the present invention, activating the secondary control at t = 0s;

[0205] Phase 2: Verify the fault tolerance of the present invention to load changes, and set the required power of load 2 at t = 20s Changed to 6MW;

[0206] Phase 3: Verify the plug-and-play capability of the present invention. DG6 is connected to the microgrid system at t = 40s. The known parameters are set as: f 6 (0) = 49.5 Hz, P 6 (0) = 3KW, and And it is assumed that DG6 can obtain the reference frequency and the desired power allocation level, that is, b6=1.

[0207] Figure 7 , Figure 8 , Fig. 9 The simulation results of three stages are shown. First, at t = 0s, for a given initial value of unbalance, the output frequency ω of each DG is i It asymptotically recovers to the given reference value and achieves accurate active power distribution according to the required power distribution level, which means that the proposed distributed control scheme has a good ability to compensate for deviations.

[0208] Then, at t=20s, due to the power demand of load 2 With the increase of 6MW, the frequency, power allocation level and active power output of each DG change accordingly. However, under the action of the designed controller, the frequency amplitude still asymptotically recovers to the given reference value after a short fluctuation, and the accurate active power allocation of the second stage is realized according to the new power allocation level, which means that the proposed distributed two-level control scheme has good ability to resist load changes.

[0209] Finally, at t = 40s, it can be seen that the connection of DG6 causes fluctuations in frequency and active power output, but the designed controller quickly responds to these changes, the frequency still converges to the given reference value, and the power allocation level and active power value converge to the new reference value, completing the third stage of accurate frequency recovery and active power allocation, further demonstrating that the proposed method has good plug-and-play capability.

[0210] In addition, the present invention will also test the effectiveness of the designed sliding mode control. According to the system framework, additional sliding mode control is added to the secondary control layer to enhance the robustness of the system operation. Due to external interference and uncertainty, it is difficult for the closed-loop system to obtain accurate convergence. Here, it is first assumed that the form of the matching uncertainty is satisfy The form of the external disturbance selected is Gaussian noise

[0211] Set relevant parameters as well as

[0212]

[0213]

[0214] In addition, in order to reduce the impact of chattering, a continuous saturation function is used in the actual simulation Instead of the discontinuous saturation function sgn(·), the following sliding mode control law is designed:

[0215]

[0216]

[0217] The final simulation results are as follows Fig.10 , Fig.11 As shown. Fig.10 It can be seen that under the influence of external disturbances and uncertain states, the convergence of the closed-loop system fluctuates. Generally speaking, for a hierarchical control structure, an island microgrid should form a stable operating state after applying secondary control. However, the actual external environment will affect the system, which is why we use sliding mode control. Fig.11It can be seen that the fluctuation is eliminated after applying the sliding mode control, and the convergence of the error state is well corrected. In summary, the additional sliding mode control adopted by the present invention improves the robustness of the system, and it is also a method that is easy to implement.

[0218] Finally, the present invention can effectively reduce the communication burden between DGs. Fig.12 The transmission interval and transmission instant of each DG event trigger are shown. In order to better illustrate the effect of the event trigger mechanism, the following formula is used to calculate the trigger rate:

[0219] Trigger rate = number of transmitted data / number of sampled data

[0220] The sampling period is selected as l = 0.04s, and the total number of samples is During the entire simulation process, the trigger rate of each DG is calculated as:

[0221]

[0222] This means that a large amount of unnecessary data is not transmitted, which greatly reduces the communication burden. In addition, the event trigger mechanism used in the present invention has significant advantages over the traditional mechanism. Let ψ m =0, the event trigger mechanism used in the present invention can be converted into a traditional fixed threshold method. Under this mechanism, the trigger rate of each DG is calculated as:

[0223]

[0224] The following table clearly shows the specific comparison results. It can be seen that in this embodiment, the event triggering method used in the present invention has further improved the communication quantity of each DG under different sampling periods compared with the classic event triggering method, which also reflects the superiority of the present invention from the side.

[0225] Transmission rate (present invention) DG1 DG2 DG3 DG4 DG5 DG6 l＝0.01 5.7％ 6.1％ 6.9％ 6.2％ 6.0％ 6.1％ l＝0.02 10.2％ 11.8％ 12.2％ 12.8％ 12.2％ 12.6％ l＝0.04 20.4％ 22.2％ 25.2％ 24.5％ 23.3％ 24.2％ l＝0.08 40.6％ 43.9％ 48.8％ 48.6％ 45.7％ 48.6％

[0226]

[0227]

[0228] In summary, the present invention can reduce the impact of link failures on the system; the present invention creates an asynchronous distributed secondary controller within the network framework, and its asynchronous idea is to solve the challenge of accurately acquiring the mode. From the control effect point of view, the designed secondary control can not only achieve the desired control target, but also effectively reduce some network constraints, such as communication load and the impact of network attacks; in order to enhance the robustness of the power grid during operation, the present invention additionally designs sliding mode control in the secondary control layer, focusing on resisting external interference and matching uncertainty in the transmission channel. Technically speaking, the sliding mode control can be designed simultaneously with the secondary control and is easy to implement.

[0229] Obviously, the above-mentioned implementation examples of the present invention are only examples for clearly explaining the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made on the basis of the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.

Claims

1. A robust two-level control method for an island microgrid based on a communication network and a random switching topology, characterized in that: The method comprises the following steps in order: (1) Construct an asynchronous distributed two-level controller for the island microgrid system based on the hierarchical control principle; (2) An event-triggered mechanism is introduced as network communication, the impact of false data injection attacks in information transmission is considered, and a networked asynchronous distributed secondary controller is constructed for the isolated microgrid system; (3) Based on the networked asynchronous distributed two-level controller, the frequency error and power level error are defined, the sliding mode control is designed, and the global frequency error system and the global power level error system are established; (4) Using the Lyapunov stability method, derive the stability conditions of the global frequency error system and the global power level error system; (5) Test and verify the networked distributed secondary controller; The step (2) specifically refers to: For frequency control, the triggering time of the mth distributed generation satisfies the following rules: Among them, d m is a positive integer; t k is the sampling time, is the sampling time of the frequency data of the mth distributed power source; l is the sampling period; Θ ω is the event trigger weight matrix; Indicates the frequency data currently sampled by the mth distributed power source Frequency data with latest transmission The error between Among them, ω ref represents a reference frequency signal; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, The latest transmission frequency of the neighbor of the mth distributed generation satisfies Among them, ζ, ψ, ε represent the basic threshold parameters for event triggering; tanh(·) is used to adjust the threshold value in real time; Get the event trigger mechanism for the frequency signal: The power signal and the frequency signal select the same transmission channel to obtain the event trigger mechanism of the power signal: Divide the trigger interval: Define the error function: Considering the impact of false data injection attacks in information transmission, false data injection attacks are based on nonlinear functions. The data information is falsely injected, and the frequency signal transmission data received by the mth distributed power source from its neighbor n under the false data injection attack is: Among them, μ(t)∈{0,1} is a Bernoulli random variable, which is used to indicate whether the false data injection attack occurs in the network transmission channel; Similarly, for power signal transmission, after false data injection attack The power signal transmission data received by the mth distributed power source from its neighbor n is: Therefore, combined with the event trigger mechanism, a networked asynchronous distributed secondary controller based on false data injection attack is constructed for the isolated microgrid system, including and in: It is a networked asynchronous distributed secondary frequency controller. It is a networked asynchronous distributed two-stage power controller; m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Represents the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; ω ref and γ ref For a given reference frequency and desired power utilization level; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and It is the transmission data of the frequency signal and power level signal of the mth distributed power source under the event trigger mechanism; and is the transmission data of the frequency signal and power level signal received by the mth distributed power source under the false data injection attack from its neighbor nth distributed power source; t k is the sampling time, l is the sampling period; The step (3) specifically refers to: Define the frequency error vector The frequency error system is constructed as: in: Among them, v ω (t) is the external disturbance signal, y ω (t) is the output signal, is satisfied The norm of the bounded nonlinear function, l ω are known parameters, is the adjacency matrix, is the Laplacian matrix, represents the gain of the secondary controller; For sliding mode control, construct the sliding surface: Among them, c ω are known parameters; Establish asynchronous sliding mode control law: is the sliding mode controller gain, is the robust term gain; Substituting the asynchronous sliding mode control law into the frequency error system, we get the global frequency error system: Similarly, define the power level error The global power level error system is obtained as: in, and are the frequency error and the power level error; and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t); and is false data; μ(t) is Bernoulli distribution; and is the error function in the event triggering mechanism; t k is the sampling time; l is the sampling period.

2. The robust two-level control method for an island microgrid based on a communication network and a random switching topology according to claim 1, characterized in that: The step (1) specifically refers to: in the isolated AC microgrid model, the communication network relationship between each distributed power source and its neighbors is described by a topology graph model. in and Represents the node set and edge set respectively. If there is an edge defined as ε mn =(m,n)∈ε, then node n is called the neighbor of node m. Denote all neighbors of node i, Defined as a graph The adjacency matrix of mm =1, if a mn =1, then ε mn ∈ε, otherwise a mn =0; Representation diagram The degree matrix of Note: The Laplace matrix of ; the path is the sequence of edges in the graph. To ensure the convergence of the consistency algorithm, the graph It is connected, and there is a path between any nodes; Let {θ(t)} be a right-continuous Markov process defined in a standard probability space, whose values ​​are contained in the finite set In the example, the transfer rate matrix of {θ(t)} is Defined as: Among them, κ>0,lim Λ＞0 (o(Λ) / Λ)=0,π rq It is expressed as the transition rate from mode r to mode q, and satisfies Introducing λ(t) as an asynchronous mode, the relationship between θ(t) and λ(t) is given by the conditional probability describe; π rs =Pr{λ(t)=s∣θ(t)=r} (2) in, for and For the primary control layer of the island microgrid, the droop control strategy is: Where, m is the mth distributed power source; ω m and V m Represent the frequency and voltage amplitude respectively; and Represent the nominal values ​​of frequency and voltage respectively; P m and Q m Represent active output and reactive output respectively; α m and β m represents the droop coefficient; For frequency control, remember Indicates auxiliary frequency control input; the nominal frequency value is: The power balance equation is: in, Indicates the total actual output power of all distributed power sources; Represents the overall load demand; P loss Represents the total real power loss, which is given by the formula The power loss is estimated, where is the power loss rate, which is 5% to 7%. Each distributed power source reasonably allocates the corresponding real power according to the maximum power limit it can generate and the actual load demand. The reference real power value is expressed as: in, represents the maximum generated power of the mth distributed generation; Represents the desired power allocation level; introduces a local variable And define For auxiliary power control input, the new frequency nominal value is calculated as follows: Under any initial frequency and power value, frequency recovery and accurate active power distribution of all distributed power sources can be achieved: Among them, ω ref and For a given reference frequency and desired power utilization level, according to the topology diagram and Markov process, the asynchronous distributed two-level controller constructed based on the hierarchical control principle is: Among them, ω m (t) and Represents the frequency value and power level value of the mth distributed power source; ω n (t) and Indicates the frequency and power level value of the nth distributed generation that is the neighbor of the mth distributed generation; Indicates whether the mth distributed power source has a neighboring nth distributed power source. If so, otherwise, Indicates whether the mth distributed generation can obtain reference information. If it can, otherwise, and represents the gain of the secondary controller, the subscript r represents the Markov process θ(t), and s represents the asynchronous mode λ(t).

3. The robust two-level control method for an island microgrid based on a communication network and a random switching topology according to claim 1, characterized in that: The step (4) specifically refers to: using the Lyapunov stability method to perform stability analysis on the global frequency error system and the global power level error system: Conclusion 1: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix Laplacian Matrix as well as Then, given the attack parameters μ and G ω , event trigger parameter ζ m and ψ m , the upper bound of nonlinear function The parameter l in the output equation ω , H infinite performance level γ ω , the identity matrix I, if there exists a positive definite matrix Event trigger matrix Θ ω >0, secondary controller gain and the sliding mode controller gain Make any and The following inequality (22) holds true, then the constructed asynchronous robust two-stage frequency controller has an external disturbance v ω (t) and nonlinear The frequency recovery of each distributed power source under any initial conditions is achieved at the same time: in: Conclusion 2: The island microgrid system with Markov switching topology connection is regarded as a directed graph Given its adjacency matrix Laplacian Matrix as well as Then, given the attack parameters μ and Event trigger parameter m and ψ m , the upper bound of nonlinear function Output parameters in equation H Infinity Performance Level The identity matrix I, if there exists a positive definite matrix Event Trigger Matrix Secondary controller gain and the sliding mode controller gain Make any and The following inequality (23) holds true, then the constructed asynchronous robust two-stage power controller has the following characteristics in the presence of external disturbances: and nonlinear The accurate real power recovery of each distributed power source under any initial conditions is achieved at the same time: in:

Images