A method for establishing a discrete-continuous hybrid system model of active distribution network based on finite state machine

Abstract

The present invention discloses a method for establishing a discrete-continuous hybrid system model of an active distribution network based on a finite state machine. The method comprises the following steps: establishing a finite state machine model according to different states and state transition rules in the physical system of the active distribution network; obtaining the instantaneous voltage, current, frequency and sampling period T of different nodes through sensors; s ; Find different sampling periods T s The greatest common factor ΔT of the active power distribution network is then used to determine the synchronization time within a finite time period. A discrete-continuous hybrid system model is established based on the greatest common factor ΔT and the synchronization time. This constructed discrete-continuous hybrid system clearly depicts the complex dynamic characteristics of the physical spatial cross-information within the active power distribution network, providing methodological support for analyzing the impact of cyberattacks on active power distribution networks, designing security defense strategies, and assessing security risks.

Description

Technical Field

[0001] The present invention belongs to the field of active power distribution network system modeling, and in particular relates to a method for establishing a discrete-continuous hybrid system model of an active power distribution network based on a finite state machine. Background Art

[0002] The active power distribution network system deeply integrates various elements of multi-dimensional, heterogeneous time, space, continuous and discrete states to form a complex network that integrates multiple complex and collaborative functions such as sensing and data acquisition, reliable communication, high-performance data processing, and intelligent control. This type of system is conducive to improving the ability to understand and utilize the world, especially the ability to control and change the world. However, the existing technical operation level is still immature and has not formed a complete technical system. Even the theoretical research based on this problem is still in its infancy. The present invention intends to use an integrated system modeling method as a starting point for carrying out such work and construct a theoretical model framework foundation for studying such problems.

[0003] In active distribution network systems, a large number of devices and parameters present a "discrete-continuous" cross-spatiotemporal state during operation due to differences in sampling periods and information interaction frequencies. Discrete computational processes and continuous physical processes coexist in the system, making it difficult to characterize the fusion characteristics between the two with a single model. Summary of the Invention

[0004] To maximize the inclusion of system characteristics in the system model, address the unification issue between discrete and continuous modeling, and provide a theoretical basis for analyzing the impact of cyberattacks on active distribution networks, designing security defense strategies, and assessing security risks, the present invention aims to provide a method for establishing a discrete-continuous hybrid system model for active distribution networks based on a finite state machine.

[0005] To achieve the above objectives, the present invention intends to adopt the following technical solutions:

[0006] A method for establishing a discrete-continuous hybrid system model of an active distribution network based on a finite state machine comprises the following steps:

[0007] Step S1: Establish a finite state machine model M = (Q, Σ, Δ, q0, F) according to different states and state transition rules in the active distribution network physical system;

[0008] Step S2: Obtain the instantaneous voltage, current, frequency and sampling period T of different nodes through sensors s ;

[0009] Step S3: Find different sampling periods T s The greatest common factor ΔT of , and then the synchronization time within a limited time period is obtained from the greatest common factor ΔT;

[0010] Step S4: Establish a discrete-continuous hybrid system model based on the greatest common factor ΔT and the synchronization time.

[0011] Furthermore, step S1 specifically includes: the active distribution network physical system includes distributed power sources, and the distributed power sources include photovoltaic power generation modules, battery power generation modules, and wind power generation modules. The finite state machine model is M = (Q, Σ, Δ, q0, F), specifically:

[0012] Q: a finite set of internal states;

[0013] Σ: trigger event input set;

[0014] Δ:Q×Σ→Q;

[0015] q0∈Q is the initial state;

[0016] is the terminal state;

[0017] The distributed power supply includes a photovoltaic power generation module, a battery power generation module, and a wind power generation module. The initial and final states of the power generation module are both off-state. When the module is powered off, it will return to the off-state regardless of its current state. Module failures include power generation equipment failures and grid failures. In the photovoltaic power generation module, the finite state set of module variables can be described as Q1 = {q 11 ,q 12 ,q 13 ,q 14 ,q 15 ,q 16 ,q 17}, where q 11 In the power-off state, q 12 In the self-test state, q 13 For the preparation state, q 14 For sleep state, q 15 is working state, q 16 is the restricted state, q 17 It is a fault state; the triggering events Σ for state transition include: power on, power off, self-diagnosis and detection states.

[0018] When the condition V is met out ≥V start , P<P stop (where V out is the output voltage, V start is the starting voltage, P is the system output power, P stop To stop power), the module reaches the maximum power, fault, fault removal, clock delay signal.

[0019] The finite state machine set Q2 of battery power generation = {q 21 ,q 22 ,q 23 ,q 24 ,q 25 ,q 26}Include: Stop state q 21 , running status q 22 , minimum charge state q 23 , maximum charge state q 24 , charging operation status q 25 , fault state q 26 ; The triggering events Σ for state transition include: charging, discharging, power failure, fault, and fault removal.

[0020] The finite state machine set Q3 of wind power generation = {q 31 ,q 32 ,q 33 ,q 34 ,q 35}Include: Stop state q 31 , running status q 32 ,MPPT operating statusq 33 , constant power operation state q 34 , fault state q 35 The triggering events Σ for state transition include: power failure, fault, fault removal, wind speed v greater than the set wind speed value v s , the output power P tracks the optimal power P ropt , wind speed v is greater than the rated wind speed value v m .

[0021] Furthermore, in step S2, voltage and current sensors collect various parameter information, including state information and event information, from the physical object. This information includes both discrete and continuous states. While physical objects in physical systems exhibit continuity, information systems use discrete binary representations to abstractly describe computational objects. This results in different sampling periods for different objects.

[0022] Furthermore, in step S3, there are a large number of asynchronous sampling points and a small number of synchronous sampling points in the sampling information of the active distribution network system. s , take the greatest common factor ΔT, and use the initial and final moments of the cycle as synchronous sampling points, also known as synchronization moments

[0023] Furthermore, the system model established in step S4 presents a system state that is a mixture of discrete and continuous systems, and is represented by the following equation:

[0024]

[0025] Y j (k+1)=F j (x j ,k)+B j (u j (k))+D j (ω j (k)),j∈S2 (2)

[0026] S1∪S2=S:={1,2,…,n} (3)

[0027] Among them F i (x i ,t),i∈S1 represents the fast-changing system state, which appears as a continuous system state (S1 is the time set of the continuous system), F j (x j ,k),j∈S2 represents the slowly changing system state (S2 is the time set of the discrete system), which is presented as the discrete system state, B i (u i (t)) and B j (u j (k)) represent the control inputs of the continuous system and discrete system respectively, and D i (ω i (t)) and D j (ω j (k)) denote the external perturbations in the continuous and discrete components, respectively.

[0028] In response to deception attacks and replay attacks that occur in active distribution networks, the impact of network attacks is defined as an uncertain variable Δ, and this component with random dynamic changes is integrated into the "discrete-continuous" hybrid active distribution network system to form an integrated system model:

[0029]

[0030] Y j (k+1)=F j (x j ,k)+B j (u j (k))+D j (ω j (k))+Δ j (t),j∈S2 (5)

[0031] S1∪S2=S:={1,2,…,nh} (6)

[0032] where x i (t),u i (t),ω i (t) and x j(t),u j (t),ω j (t) are the system state vector, output vector and external disturbance of the system respectively, and the corresponding matrices A, B, and D are system matrices of appropriate dimensions, Δ i (t) and Δ j (t) represents the components in the continuous and discrete systems respectively, which are dynamically changing and random.

[0033] Design corresponding security defense strategies for different network attacks to achieve system security control. When the communication network in the distribution network is attacked by deception or replay attacks, the fast-changing signals with short sampling periods and high sampling frequencies are mainly used. In order to avoid large signal mutations, the proximity principle is adopted for switching. When the load in the distribution network is attacked by dynamic loads, the attacker obtains the system frequency by embedding sensors in the power system, and then causes the grid frequency to deviate from the standard frequency through attacks to destroy the stability of the grid. For this, a frequency deviation range ω is set. s When it exceeds this range, the switching system switches to the corresponding power system state equation, solves the non-salient pole configuration optimization problem through the optimization method, and finds the minimum load that needs to be protected, thereby ensuring system stability.

[0034] For power systems with dynamic load change attacks, the load is divided into safety loads and fragile loads Total load (k represents the kth bus), where is the value in formula (4) is u(t). Based on the collected information, the state equation of the continuous system is established:

[0035]

[0036] where x i (t)=[δ θ ω] T ,δ is the voltage phase angle at the generator bus,ω is the rotor angular frequency deviation at the generator bus,θ is the voltage phase angle at all load buses. Then the coordinated descent method is used to solve the minimum load that needs to be protected, that is, And satisfy 0≤P LP ≤P LV , S is the location of all potential attacker sensors on the bus, P LP and the positive semidefinite matrix X s is the variable that needs to be found.

[0037] Compared with the prior art, the present invention can achieve the following optimization effects:

[0038] The present invention provides a method for establishing a discrete-continuous hybrid system model of an active distribution network based on a finite state machine. First, a finite state machine model is established according to different states and state transition rules in the physical system of the active distribution network. The instantaneous voltage, current, frequency and sampling period T of different nodes are obtained by sensors. s ; Find different sampling periods T s The greatest common factor ΔT of the active power distribution network is calculated, and the synchronization time within a finite time period is derived from the greatest common factor ΔT. A discrete-continuous hybrid system model is established based on the greatest common factor ΔT and the synchronization time. The discrete-continuous hybrid system constructed by the present invention can clearly depict the complex and dynamic characteristics of information, physical space intersections, and dynamic changes in active power distribution network systems, providing a theoretical basis for analyzing the impact of cyber attacks in active power distribution networks, designing security defense strategies, and assessing security risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a schematic diagram of an active power distribution network of the present invention;

[0040] Figure 2 It is a flow chart of a method for establishing a discrete-continuous hybrid system model of an active power distribution network based on a finite state machine of the present invention;

[0041] Figure 3 It is a schematic diagram of the state of the "discrete-continuous" hybrid system of the present invention. DETAILED DESCRIPTION

[0042] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings and embodiments. Obviously, the embodiments described are only part of the embodiments of the present invention, rather than all of the embodiments. The following embodiments are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0043] Figure 1 Figure 1 is a schematic diagram of an active power distribution network, which includes a master control system 1, a communication network 2, and an active power distribution system 3. The master control system 1 dispatches and protects the active power distribution system 3 via the communication network 2. The active power distribution system 3 includes a distributed power source 31 and a load 32. The distributed power source 31 includes a photovoltaic power generation module 311, a battery power generation module 312, and a wind power generation module 313. The photovoltaic power generation module 311 includes a controller 3111, an actuator 3112, and a sensor 3113; the battery power generation module 312 includes a controller 3121, an actuator 3122, and a sensor 3123; and the wind power generation module 313 includes a controller 3131, an actuator 3132, and a sensor 3133. Figure 1In the figure, only three subsystems, namely the photovoltaic power generation module 311, the battery power generation module 312 and the wind power generation module 313, are taken as examples. They adopt distributed connection to supply power to the load 32. The present invention is not limited to this, and the distributed power source 31 may also include other subsystems. Taking the photovoltaic power generation module 311 as an example, when the communication network 2 is attacked, the controller 3111 issues a control command, and the actuator 3112 is the local execution unit of the subsystem to execute the command. After the command is executed, the sensor 3113 collects the current system information in the subsystem and transmits the information to the communication network 2, which is then transmitted to the main control system 1 for feedback, thereby realizing the scheduling and protection of the active distribution network 3.

[0044] Due to the different sampling periods of the three module subsystems, many parameters present a "discrete-continuous" cross-spatiotemporal state. The discrete calculation process and the continuous physical process coexist in the system, making it difficult to use a single model to describe the fusion characteristics between the two. Figure 1 The main control system 1 is integrated in Figure 2 The method for establishing a discrete-continuous hybrid system model of an active distribution network based on a finite state machine shown in FIG. 1 includes the following steps:

[0045] Step S1: Establish a finite state machine model M = (Q, Σ, Δ, q1, F) according to different states and state transition rules in the active distribution network physical system;

[0046] Furthermore, step S1 specifically includes: establishing a finite state machine model based on the different states and state transition rules in the physical system of the active power distribution network. The role of the finite state machine (FSM) is to ensure a switching mechanism between multiple finite states (for example, a mechanism for changing the sampling period at different times determined by the event trigger mechanism), and thus to switch between the set modes of the state of the collected data. At the same time, the states will change with the changes in the transition conditions (the main impact is the change of the corresponding data collection period). When the corresponding set (state) in the state machine is transferred (switched), the corresponding sampling period also changes. Therefore, the main purpose is to establish such a data collection and utilization mechanism for use in step S2.

[0047] The physical system of the active distribution network includes a distributed power source 31, which includes a photovoltaic power generation module 311, a battery power generation module 312, and a wind power generation module 313. The finite state machine model is M = (Q, Σ, Δ, q1, F), specifically:

[0048] Q: a finite set of internal states;

[0049] Σ: trigger event input set;

[0050] Δ:Q×Σ→Q;

[0051] q1∈Q is the initial state;

[0052] is the terminal state;

[0053] Among them, the initial state and the end state of the power generation module are both the power-off state. When the module is powered off, it will return to the power-off state regardless of its state; module failure includes power generation equipment failure and grid failure.

[0054] The finite state set Q1 of the photovoltaic power generation module 311 = {q 11 ,q 12 ,q 13 ,q 14 ,q 15 ,q 16 ,q 17}, where q 11 In the power-off state, q 12 In the self-test state, q 13 For the preparation state, q 14 For sleep state, q 15 is working state, q 16 is the restricted state, q 17 It is a fault state; the triggering events Σ for state transition include: power on, power off, self-diagnosis and detection states.

[0055] When the condition V is met out ≥V start , P<P stop (where V out is the output voltage, V start is the starting voltage, P is the system output power, P stop To stop power), the module reaches the maximum power, the module fails, the fault is removed, and the module clock delay signal.

[0056] Table 1 Photovoltaic power generation module 311 state transition set Q1

[0057]

[0058]

[0059] in, To prevent this from happening, q 11 The power-off state is the initial state. The module will jump to q after power is turned on. 12 Self-check status, jump to q after self-check is completed 13 In the ready state, if the output voltage V of the photovoltaic matrix is ​​greater than the set starting voltage V s , then jump to q 15Working state. In the working state, if the module output power P is less than the stop power P p , then return to q 13 Ready state; if the module parameters are abnormal, jump to q 16 Restricted state, returns to working state after normal. 16 In the limiting state, the module output power P is less than the stop power P p , then jump to q 13 Ready state.

[0060] The finite state machine set Q2 of the battery power generation module 312 = {q 21 ,q 22 ,q 23 ,q 24 ,q 25 ,q 26}Includes: Power-off state q 21 , running status q 22 , minimum charge state q 23 , maximum charge state q 24 , charging operation status q 25 , fault state q 26 The triggering events for state transition include: charging, discharging, power failure, module failure, and fault resolution.

[0061] Table 2 Battery power generation module 312 state transition set Q2

[0062]

[0063] in, To prevent this from happening, q 21 The power-off state is the initial state, and the module will jump to q when it is discharged. 22 Running state, jump to q when discharged to the minimum charge 23 Minimum charge state; when the module is in the running state and the minimum charge state, the charging will jump to the charging running state; when the module is charged to the maximum charge in the running state, it will jump to Q 24 Maximum state of charge.

[0064] The finite state machine set Q3 of the wind power generation module 313 = {q 31 ,q 32 ,q 33 ,q 34 ,q 35}Include: Stop state q 31 , running status q 32 ,MPPT operating statusq 33 , constant power operation state q 34 , fault state q 35The triggering events Σ for state transition include: power failure, fault, fault removal, wind speed v greater than the set working wind speed value v s , the output power P tracks the optimal power P ropt , wind speed v is greater than the rated wind speed value v m .

[0065] Table 3 Wind power generation module 313 state transition set Q3

[0066]

[0067] Among them, the stop state is the initial state. When the wind speed v is greater than the set wind speed value v s Enters the running state; when the output power P tracks the optimal power P ropt When the wind speed v is greater than the rated wind speed value v, the system switches from the running state to the MPPT running state; m When , it enters the constant power operation state from the MPPT operation state.

[0068] Step S2: Obtain the instantaneous voltage, current, frequency and sampling period T of different nodes through sensors s ;

[0069] Furthermore, step S2 specifically includes: obtaining the instantaneous voltage and current frequency and state sampling period T of different nodes through voltage and current sensors s The acquired information has discrete states and continuous states. Physical objects in physical systems exhibit continuity characteristics, while information systems use discrete binary to abstractly describe computing objects, which results in different sampling periods for different objects.

[0070] Step S3: Find different sampling periods T s The greatest common factor ΔT of , and then the synchronization time within a limited time period is obtained from the greatest common factor ΔT;

[0071] Furthermore, step S3 specifically includes finding different sampling periods T s The greatest common factor ΔT is used to obtain the synchronization time within a limited time period. Since there are a large number of asynchronous sampling points and a small number of synchronous sampling points in the sampling information of the active distribution network system, the greatest common factor ΔT is taken according to different sampling cycles, and the initial and end times of the cycle are used as the synchronization sampling points, also known as the synchronization time. At the same time Figure 3 As shown, X1(t), X2(t), and X3(t) represent the output currents of the photovoltaic power generation module, the battery power generation module, and the wind power generation module, respectively. Figure 1The data is obtained by sampling sensors 3113, 3123, and 3133 in the distributed generation system. The sampling periods of these three modules are different, making them difficult to characterize with a single model. Therefore, we use the method of finding the greatest common factor of the sampling periods to obtain the greatest common factor ΔT. A known synchronization time A1 is selected in the distributed generation system. Synchronization time A2 = A1 + ΔT, A3 = A2 + ΔT, A4 = A3 + ΔT, and so on.

[0072] Design corresponding security defense strategies for different network attacks to achieve system security control. When the communication network in the active distribution network is attacked by deception or replay, for example, Figure 3 Middle state x 16 To avoid large signal mutations, the system switches to the synchronization time A2 based on the proximity principle. The switching signal is mainly a fast-changing signal with a short sampling period and high sampling frequency. The impact of the network attack is defined as an uncertain variable Δ, and this randomly changing component is integrated into the "discrete-continuous" hybrid active distribution network system to form an integrated system model:

[0073]

[0074] Y j (k+1)=F j (x j ,k)+B j (u j (k))+D j (ω j (k))+Δ j (t),j∈S2 (5)

[0075] S1∪S2=S:={1,2,…,n} (6)

[0076] where x i (t),u i (t),ω i (t) and x j (t),u j (t),ω j (t) are the system state vector, output vector and external disturbance of the system respectively, and the corresponding matrices A, B, and D are system matrices of appropriate dimensions, Δ i (t) and Δ j (t) represents the components in the continuous and discrete systems respectively, which are dynamically changing and random.

[0077] When the communication network load of the active distribution network is attacked by a dynamic load, the attacker obtains the system frequency by embedding sensors in the power system and uses the attack to make the grid frequency deviate from the standard frequency to destroy the stability of the grid. A frequency deviation range ω is set for this.s When it exceeds this range, the switching system switches to the corresponding power system state equation, solves the non-salient pole configuration optimization problem through the optimization method, and finds the minimum load that needs to be protected, thereby ensuring system stability.

[0078] For power systems with dynamic load change attacks, the load is divided into safety loads and fragile loads Bus load (k represents the kth bus), where is the value in formula (4) is u(t). Based on the collected information, the state equation of the continuous system is established:

[0079]

[0080] where x i (t)=[δ θ ω] T ,δ is the voltage phase angle at the generator bus,ω is the rotor angular frequency deviation at the generator bus,θ is the voltage phase angle at all load buses. Then the coordinated descent method is used to solve the minimum load that needs to be protected, that is, And satisfy 0≤P LP ≤P LV , S is the location of all potential attacker sensors on the bus. The necessary and sufficient condition for system stability is that there exists a positive semidefinite matrix Y such that A T Y+YA<0. Under the condition of system stability, the coordinated descent method is used to solve the minimum load that needs to be protected, that is, And satisfy 0≤P LP ≤P LP , S is the location of all potential attacker sensors on the bus, P LP and the positive semidefinite matrix X s is the variable to be determined. Step S4: Establish a discrete-continuous hybrid system model based on the greatest common factor ΔT and the synchronization time

[0081]

[0082] Y j (k+1)=F j (x j ,k)+B j (u j (k))+D j (ω j (k)),j∈S2 (2)

[0083] S1∪S2=S:={1,2,…,n} (3)

[0084] Furthermore, step S4 specifically includes: according to the greatest common factor ΔT and the synchronization time A discrete-continuous hybrid system model is established. The established system model presents a system state that is a mixture of discrete and continuous systems, and is expressed by the following equation:

[0085]

[0086] Y j (k+1)=F j (x j ,k)+B j (u j (k))+D j (ω j (k)),j∈S2 (2)

[0087] S1∪S2=S:={1,2,…,n} (3)

[0088] in, is the derivative of the system state, F i (x i ,t),i∈S1 represents the fast-changing system state, which appears as a continuous system state (S1 is the time set of the continuous system), F j (x j ,k),j∈S2 represents the system state that changes slowly, which is presented as a discrete system state (S2 is the time set of the discrete system), B i (u i (t)) and B j (u j (k)) represent the control inputs of the continuous system and discrete system respectively, and D i (ω i (t)) and D j (ω j (k)) denote the external perturbations in the continuous and discrete components, respectively.

Claims

1. A method for establishing a discrete-continuous hybrid system model of an active distribution network based on a finite state machine, characterized in that: The following steps are involved: Step S1: Establish a finite state machine model M = (Q, ∑, Δ, q0, F) according to different states and state transition rules in the active distribution network physical system; The step S1 is specifically as follows: the active distribution network physical system includes distributed power sources, and the finite state machine model is M=(Q, ∑, Δ, q0, F), where Q: a finite set of internal states; ∑: trigger event input set; Δ: Q×∑→Q; q0∈Q is the initial state; F⊆Q is a terminal state; Step S2: Obtain the instantaneous voltage, current, frequency and sampling period T of different nodes through sensors s ; Step S2 specifically includes: collecting different parameter information of the physical object through sensors, including state information and event information. The acquired information includes both discrete states and continuous states. Physical objects in the physical system exhibit continuity, while the information system uses discrete binary to abstractly describe the computational objects, which results in different sampling periods for different objects. Step S3: Calculate the sampling period T s The greatest common factor ΔT of , and then the synchronization time within a limited time period is obtained from the greatest common factor ΔT; In the step S3, there are a large number of asynchronous sampling points and a small number of synchronous sampling points in the sampling information of the active distribution network system. s , take the greatest common factor ΔT, and use the initial and final moments of the cycle as the synchronization moments ; Step S4: Based on the greatest common factor ΔT and the synchronization time, a discrete-continuous hybrid system model is established, which is expressed by the following equation: (2) S1∪S2=S:={1,2,...,n} (3) Among them F i (x i， t), i∈S1 represents the fast-changing system state, which appears as a continuous system state, that is, S1 is the time set of the continuous system, F j (x j , k), j∈S2 represents the slowly changing system state, which is presented as a discrete system state, that is, S2 is the time set of the discrete system, B i (u i (t)) and B j (u j (k)) represent the control inputs of the continuous system and discrete system respectively, and D i (ω i (t)) and D j (ω j (k)) denote the external perturbations in the continuous and discrete components, respectively.

2. The method for establishing a discrete-continuous hybrid system model of an active power distribution network based on a finite state machine according to claim 1, characterized in that: Select any synchronization time A1 in the distributed power supply, then the synchronization time A2 = A1 + ΔT, A3 = A2 + ΔT, A4 = A3 + ΔT, and so on.

Images