Generator set networked safety control method and system based on demand driving

Abstract

The invention belongs to the technical field of generator set network security, and particularly discloses a generator set network security control method and system based on demand driving, and the method comprises the following steps: building a state space model represented by the output voltage and current of a generator set; when the network communication is reliable, designing a demand-based scheduling transmission mechanism by using the established state space model, processing the generator set data, and removing the influence of noise and disturbance on the generator set data; when unreliable network communication caused by network attacks exists, an elastic transmission mechanism is designed, data of the generator set are processed, and elastic networked safety control is achieved. According to the demand-driven generator set networked safety control method and system provided by the invention, the influence caused by allowable unreliable network communication can be improved, the cost is controllable, and the use is convenient.

Description

Technical Field

[0001] The present invention relates to the technical field of network security of generator sets, and in particular to a networked security control method and system for generator sets based on demand-driven control, and in particular to a generator set elastic control method based on demand-driven control in the presence of malicious network blocking attacks. Background Art

[0002] In recent years, with the vigorous development of renewable energy, the randomness of renewable energy generation such as wind power and photovoltaic power has made the smooth dispatch of the power system a hot issue. The main power plant generators in the current power system are approaching saturation, and with the increasing demand for carbon peak and carbon neutrality, it is difficult to increase the increment of thermal power units. In the short term, relatively stable clean energy such as hydropower and nuclear power can hardly make up for the current power shortage. Especially in the summer, the air conditioning load surges and the winter due to the decrease in unit power caused by heating, the unit output fluctuations are urgent. The small-power units of local power plants, which are usually easily overlooked, need to increase their output capacity to make up for the shortage of peak electricity consumption. In order to improve the flexible capacity and peak-shaving capacity of the power system, it is necessary to control the voltage and frequency of small generators in local power plants.

[0003] In recent years, dispatching and control schemes based on network communication have been widely used in power systems. However, the use of networks and heterogeneous digital elements makes data transmission vulnerable to various network attacks, such as spoofing attacks, replay attacks, bias injection attacks, zero dynamic attacks, denial of service attacks, etc. Unlike attacks on traditional systems that limit their impact to the network level, malicious attacks on power systems can affect the physical world and cause serious social and livelihood problems. Therefore, in the presence of malicious attacks and unreliable network communications, there is a strong demand for analysis, synthesis, and design methods to ensure the safety and reliability of power systems.

[0004] As a common type of network attack, DoS (Denial of Service) attacks aim to block network communications so that data packets cannot be transmitted. In the research on power systems, the load frequency under DoS attacks is usually controlled, but the relevant research has not proposed corresponding methods from the perspective of transmission mechanism and elastic performance quantification. In the research on elastic control against DoS attacks, the stability analysis of the system is usually given under the assumption of attack frequency and duration. However, traditional control methods are all based on time-driven methods, that is, control is achieved according to a fixed sampling period. Although periodic sampling may be desirable from the perspective of analysis and design, it will lead to unnecessary waste of limited resources. Usually, the dispatching or control signals of the power system do not communicate in real time to send dispatching data, but only communicate when necessary. In this case, the traditional time-driven control algorithm is no longer applicable, and it is necessary to develop a new control algorithm that fully considers the networking factors to solve the unreliable network communication scenario caused by DoS attacks. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present invention provides a demand-driven networked safety control method and system for generator sets, which can improve the tolerance of the impact of unreliable network communications, has controllable costs and is easy to use.

[0006] The present invention adopts the following technical solution.

[0007] In the first aspect, the present invention provides a demand-driven networked safety control method for generator sets, comprising the following steps: establishing a state-space model for characterizing the output voltage and current of the generator sets; when the network communication is reliable, using the established state-space model, designing a demand-based scheduling and transmission mechanism to process the generator set data and remove the impact of noise and disturbance on the generator set data; when the network communication is unreliable due to a network attack, designing a resilient transmission mechanism to process the generator set data and achieve resilient networked safety control.

[0008] Preferably, the establishment of a state-space model for characterizing the output voltage and current of the generator set comprises the following steps: extracting a single generator set connected to the common coupling point of the power grid system, and establishing a dynamic equation of the voltage and current of the single generator set in dq coordinates; and establishing a dynamic state-space model of the generator set from the voltage and current of the d-component and q-component, respectively, according to the established dynamic equation in dq coordinates.

[0009] Preferably, the dynamic equation under dq coordinates is established as follows:

[0010]

[0011] Where GU i represents the generator set connected to the i-th node in the common coupling point, Line ij represents the line connecting generator set i and generator set j, t represents time, V i,dq Represents the output voltage of unit i, V j,dq represents the output voltage of unit j, I ti,dq Represents the output current of unit i, I Li,dq Indicates the load current, I ij,dq Indicates the line current connecting unit i and unit j, L ti Represents the equivalent inductance of unit i, L ij represents the line inductance connecting unit i and unit j, R ij Represents the line resistance connecting unit i and unit j, R ti Represents the equivalent resistance of the series filter, C ti represents the parallel capacitance, k i represents the transformer conversion rate, ω0 represents the dq model rotation speed, V ti,dq Represents the voltage source voltage.

[0012] Preferably, according to the established dynamic equation under dq coordinates, the dynamic state space model of the generator set is established from the voltage and current of the d-component and q-component respectively as follows:

[0013]

[0014] Among them, V i,d Represents the d component of the output voltage of unit i, V i,q represents the q component of the output voltage of unit i, V ti,d represents the d component of the voltage source voltage, V ti,q represents the d component of the voltage source voltage, represents the d component of the line current connecting units i and j, represents the q component of the line current connecting units i and j, I ti,d Represents the d component of the output current of unit i, I ti,q Represents the d component of the output current of unit i.

[0015] Preferably, when the network communication is reliable, the established state space model is used to design a demand-based scheduling and transmission mechanism, the generator set data is processed, and the influence of noise and disturbance on the generator set data is removed, including the following steps: designing an observer integrated with a sensor according to the established spatial state model; monitoring the unit data of the generator set in real time through the sensor, and processing the disturbance and noise of the real-time unit data through the observer to obtain a state observation value; designing a predictor in the control center of the generator set, and the predictor obtains the state prediction value of the generator set according to the unit data of the generator set; constructing a demand function based on the difference between the state observation value and the state prediction value, and designing a demand-based scheduling and transmission mechanism, so that when the demand function is met, the observer transmits the state observation value to the control center of the generator set, and the predictor synchronously receives the state observation value and updates the state prediction value; when the control center receives the state observation value, the state observation value is used as the processed unit data; when the control center does not receive the state observation value, the state prediction value is used as the processed unit data.

[0016] Preferably, when there is unreliable network communication caused by a network attack, an elastic transmission mechanism is designed to process the generator set data to achieve elastic networked security control, including the following steps: when there is unreliable network communication caused by malicious blockage, the malicious blockage is modeled as a DoS attack scenario that meets the duration and attack frequency; an elastic transmission mechanism is designed, and the generator set data is processed by an observer, and the observer sends state observation values ​​to the control center of the generator set based on the elastic transmission mechanism to achieve elastic networked security control.

[0017] Preferably, after the state space model is established, the established state space model is converted into the following model:

[0018]

[0019] y(t)=Cx(t)+ν(t)

[0020] in,

[0021] X ij =ω0L ij ,

[0022] Z ij =|R ij +jX ij |. x(t)=[V i,d ,V i,q ,I ti,d ,I ti,q ] T is the state vector, is the first-order derivative of x(t), u(t)=[V ti,d ,V ti,q ] T is the input vector, y(t) represents V i,d ,V i,q ,I ti,d ,I ti,q Four measured values, ν(t) represents noise, ω(t) = [I Li,d ,I Li,q ] T is the disturbance; V i,d represents the d component of the output voltage of unit i,

[0023] V i,q represents the q component of the output voltage of unit i, V ti,d represents the d component of the voltage source voltage, V ti,q represents the d component of the voltage source voltage, I ti,d Represents the d component of the output current of unit i, I ti,q Represents the d component of the output current of unit i; R ij Represents the line resistance connecting unit i and unit j, R ti Represents the equivalent resistance of the series filter, L ij Indicates the line inductance connecting unit i and unit j, L ti represents the equivalent inductance of unit i, C ti represents the parallel capacitance, ω0 represents the rotation speed of the dq model;

[0024] The measured value of y(t) is obtained, and the four matrices A, B, C, and E are solved using the above formula.

[0025] Preferably, according to the obtained y(t) measurement value and the four matrices A, B, C, and E, the observer is designed as follows:

[0026]

[0027] Among them, x s (t) represents the state observation value, Represents x s The first-order derivative of (t), L represents the observer gain matrix, is obtained by solving the following linear matrix inequality:

[0028]

[0029] Among them, Φ1=A-LC, Y=[IL], I is the unit matrix, λ is any given parameter that makes the linear matrix inequality solvable, and P1 is any given positive definite symmetric matrix that makes the linear matrix inequality solvable.

[0030] Preferably, the observer gain matrix L satisfies the following performance indicators to constrain disturbances and noise:

[0031] ||e s (t)||2<λ||d(t)||2

[0032] in,

[0033] Preferably, the predictor obtains the following calculation formula for the state prediction value of the generator set according to the generator set data:

[0034]

[0035] Among them, x c (t) represents the state observation value x received from the observer s The predicted state value at time (t), Represents the state observation value x that has not been received from the observer s The predicted state value at time (t).

[0036] Preferably, the demand function is constructed based on the difference between the state observation value and the state prediction value as follows:

[0037]

[0038] Among them, σ s ||x s (t)||+σ c ρ is the transmission condition, which is determined by the norm of the observation value and the upper bounds of the disturbance and noise, where ρ = λ||C||κ+κ ν ,κ=||E||κ ω +κ ν , κ ω With κ ν are the upper bounds of disturbance and noise, σ s >0 and σ c >0 is the parameter to be designed.

[0039] Preferably, the parameter σ s The steps to obtain are as follows:

[0040] Calculate the difference between the current state and the predicted state e(t) = x c (t)-x(t), the closed-loop dynamic system is constructed as follows:

[0041]

[0042] Select the Lyapunov function as V(t)=x T (t)Px(t), where P is the Lyapunov equation The only solution of , Q is any given positive definite symmetric matrix;

[0043] For any You can get:

[0044] α1||x(t)||≤V(t)≤α2||x(t)||

[0045]

[0046] Where α1 and α2 are the maximum and minimum eigenvalues ​​of the matrix P, respectively, γ1 is the minimum eigenvalue of the matrix Q, γ2 = ||2PBK||, γ3 = ||2PE||, combined with the constructed demand function and ||e s (t)||=||x(t)-x s (t)||<λ||d(t)|| ∞ , can be calculated within the transmission attempt interval:

[0047] ||x(t)-x c (t)||≤σ s ||x(t)||+σ c ρ+λ||d(t)|| ∞

[0048] Substitute ||d(t)|| ∞ ≤||E||κ ω +κ ν =κ to obtain:

[0049] ||e(t)||=||x(t)-x c (t)||≤σ s ||x(t)||+σ κ κ

[0050] where σ κ =λσ c ||C||+λ+σ c , substituting into:

[0051]

[0052] If γ1-σ is satisfied s If γ2>0, we can get:

[0053]

[0054] This ensures the stability of the control system during normal network communication. γ5=γ1-σ s γ2;

[0055] Based on the above steps, determine the selected parameter σ sThe standard is

[0056] Preferably, the parameter σ c The steps to obtain are as follows:

[0057] For any t∈[t k ,t k+1 ) Calculate the prediction error dynamics as:

[0058]

[0059] It can be calculated that:

[0060]

[0061] in Define the difference between the predicted state and the observed state as e c (t) = x s (t)-x c (t), and its dynamics are calculated as:

[0062]

[0063] in For any t∈[t k ,t k+1 ) by e c (t k )=0, e can be calculated c The upper bound of the (t) norm is:

[0064]

[0065] where ρ = λ||C||κ+κ ν ;

[0066] Reverse the above equation to get tt k ≥ Δ and

[0067]

[0068] By determining the lower bound Δ The corresponding σ is obtained by inversely solving the value of c The value of .

[0069] Preferably, the elastic transmission mechanism is expressed as the following formula:

[0070]

[0071] Among them, Δ * Indicates the DoS interval periodic transmission attempt interval, and satisfies 0<Δ * ≤ Δ , H n Indicates the attack range, is the set of real numbers, is a set of natural numbers, t k Indicates the time of a successful transmission attempt.

[0072] Preferably, the attack frequency constraint is expressed as:

[0073]

[0074] Among them, η and τ D Two parameters representing the attack frequency at the beginning of the time interval [t1, t2), η≥0, τ D ≥ Δ , is the set of real numbers, is a set of natural numbers.

[0075] Preferably, the attack duration constraint is expressed as:

[0076]

[0077] in, and T represent two parameters of duration at the start of the time interval [t1, t2), T>1, is the set of real numbers, is a set of natural numbers.

[0078] Preferably, in order to ensure that the controlled system with bounded disturbance is input-state stable, the following expression is established:

[0079]

[0080]

[0081] According to the above formula, Δ * , τ D and T are constrained.

[0082] In a second aspect, the present invention provides a system utilizing the aforementioned demand-driven networked safety control method for generator sets, the system comprising: an observer, a predictor and a switching controller integrated with a sensor; the sensor is used to monitor the unit data of the generator set in real time; the observer is used to process the unit data, and send a state control value to the control center of the generator set based on a demand-based scheduling transmission mechanism or an elastic transmission mechanism; the predictor is arranged at the control center, and is used to obtain a predicted state observation value based on the measured unit data, and to update the state prediction value when receiving the state control value; the switching controller is used to switch the output state observation value or the state prediction value based on the state feedback of the state control value received by the control center.

[0083] The beneficial effect of the present invention is that, compared with the prior art, the present invention provides a set of networked safety control methods for generator sets driven by demand, mainly for local power plant units that do not participate in provincial grid coordination. The present invention only needs to establish a state space model for the generator set according to the operating mechanism, and does not need to detect and master the specific form and implementation time of malicious network interference. The cost is controllable and easy to use. By designing a demand-based scheduling transmission mechanism, the network communication cost can be minimized, and combined with the designed flexible scheduling mechanism, the scheduling signal transmission can be adaptively adjusted to minimize the deterioration of the control effect caused by network fluctuations.

[0084] Furthermore, the control method of the present invention can achieve a better control effect by relying only on simple state feedback, and can constrain unknown load current interference within a quantifiable range.

[0085] The present invention can be applied to small-power generator sets in local power plants in remote areas and distributed light energy and wind energy generator sets.

[0086] The simulation results of the present invention can fully prove that as long as the parameters are appropriate, the networked controller designed and implemented using the demand-driven switching control algorithm can achieve good control effects. Even in the case of frequent and long-lasting network congestion, the control scheme can still effectively control the generator set, and the whole process is very stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a schematic diagram of the electrical principle of the generator set in the present invention;

[0088] Figure 2 It is a flow chart of the networked safety control method of a generator set based on demand-driven in the present invention;

[0089] Figure 3 It is a schematic diagram of experimental verification of the networked safety control method of a generator set driven by demand in the present invention under normal communication conditions;

[0090] Figure 4 A schematic diagram of a curve and a threshold value based on demand-driven conditions in the present invention;

[0091] Figure 5 It is a schematic diagram of the state response of a generator set of demand-driven control under DoS attack based on the networked safety control method of a generator set driven by the demand in the present invention;

[0092] Figure 6 It is a schematic diagram of the distribution value of the transmission time of the networked safety control method of the generator set driven by demand in the present invention under DoS attack;

[0093] Figure 7 It is a schematic diagram of the evolution of demand-driven transmission conditions of the networked safety control method of generator sets based on demand-driven in the present invention under DoS attack;

[0094] Figure 8 It is a schematic diagram of the change curve of the control input of the networked safety control method of the generator set driven by demand in the present invention under DoS attack. DETAILED DESCRIPTION

[0095] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0096] Embodiment 1 of the present invention provides a demand-driven networked safety control method for a generator set, comprising the following steps:

[0097] Step 1: Establish a state-space model for representing the output voltage and current of the generator set.

[0098] The generator set in the embodiment of the present invention refers to a local power plant unit that does not participate in the provincial grid dispatching, and the power range is usually less than 100MW. Step 1 specifically includes the following steps:

[0099] Step 1.1: Extract the single generator connected to the point of common coupling (PCC) of the power grid system, according to Figure 1 The electrical schematic diagram shown treats the load current as a disturbance and establishes the dynamic equations of the voltage and current of the generator set under the abc coordinates.

[0100] Specifically, a single unit i passes through a non-zero impedance three-phase line (R ij ,L ij) is connected to the rest of the units and includes a DC voltage source V ti , inductance L t , a series resistor R t Characterized filter and a power plant step-up transformer (Y-Δ) with a transformer transformation ratio of k i .

[0101] Park's transformation is used to transform the dynamic equations of the output voltage and current of a single generator set under the abc coordinates into a model under the dq coordinates, as follows:

[0102]

[0103] Where GU i represents the generator set connected to the i-th node in the common coupling point, Line ij represents the line connecting generator set i and generator set j, t represents time, V i,dq Represents the output voltage of unit i, V j,dq represents the output voltage of unit j, I ti,dq Represents the output current of unit i, I Li,dq Indicates the load current, I ij,dq Indicates the line current connecting unit i and unit j, L ti Represents the equivalent inductance of unit i, L ij represents the line inductance connecting unit i and unit j, R ij Represents the line resistance connecting unit i and unit j, R ti Represents the equivalent resistance of the series filter, C ti represents the parallel capacitance, k i represents the transformer conversion rate, ω0 represents the dq model rotation speed, V ti,dq Represents the voltage source voltage.

[0104] Step 1.2: According to equations (1) and (2), the dynamic state space model of the generator set is established from the voltage and current of the d-component and q-component respectively as follows:

[0105]

[0106] Among them, V i,d Represents the d component of the output voltage of unit i, V i,q represents the q component of the output voltage of unit i, V ti,d represents the d component of the voltage source voltage, V ti,q represents the d component of the voltage source voltage, represents the d component of the line current connecting units i and j, represents the q component of the line current connecting units i and j, I ti,d Represents the d component of the output current of unit i, Iti,q Represents the d component of the output current of unit i.

[0107] Step 1.3: Introduce the matrix ABCE to transform equation (3) into the following state space model:

[0108]

[0109] In the formula,

[0110] Where X ij =ω0L ij ,

[0111] Z ij =|R ij +jX ij |. x(t)=[V i,d ,V i,q ,I ti,d ,I ti,q ] T is the state vector, is the first-order derivative of x(t), u(t)=[V ti,d ,V ti,q ] T is the input vector, y(t) is the measured value, representing V i,d ,V i,q ,I ti,d ,I ti,q These four measurements, ν(t) represents the noise, ω(t) = [I Li,d ,I Li,q ] T For disturbance.

[0112] The generator set data is measured by sensors to obtain four measurement values ​​in y(t) and four matrices ABCE.

[0113] Step 2: When the network communication is reliable, the established state space model is used to design a demand-based scheduling transmission mechanism to process the generator set data and remove the impact of noise and disturbance on the generator set data.

[0114] The whole process of step 2 is as follows Figure 2 As shown in the figure, the sensor system (i.e., the dispatch center) uses the observer to obtain the state estimate x from the measurement data. s At the same time, the forecast value backed up by the dispatch center is used to construct the demand-driven mechanism DDM. When the transmission conditions are met, the state estimation value obtained in real time by the observer is sent to the unit control center, and the predictor located in the unit control center updates the state prediction value x c , and use the predicted value to construct the control input u.

[0115] Step 2 specifically includes the following steps:

[0116] Step 2.1: Based on the measured value y(t) and the four ABCE matrices in the state space model, an observer is designed inside the sensor and used as a scheduling center.

[0117] H ∞ The observer is designed as follows:

[0118]

[0119] Among them, x s (t) represents the state observation value, the purpose is to eliminate the influence of disturbance ω(t) and noise ν(t) on the system, Represents x s The first-order derivative of (t), L represents the observer gain matrix, which is obtained by solving the following linear matrix inequality.

[0120]

[0121] Where Φ1 = A-LC, Y = [IL], I is the unit matrix, λ is any given parameter that makes the linear matrix inequality (6) solvable, and P1 is any given positive definite symmetric matrix that makes the inequality solvable. The gain matrix L solved by equation (6) can satisfy H ∞ Performance indicators||e s (t)||2<λ||d(t)||2 constrains disturbance and noise, where,

[0122] Step 2.2: Use sensors to monitor the data of the generator set in real time, and use H ∞ The observer processes the disturbance and noise in the real-time unit data, eliminates the influence of the disturbance and noise on the generator set, and obtains the state observation value;

[0123] Step 2.3: The observer transmits the state observation value back to the control center of the local power plant generator set through a network-based communication method to generate a state control value.

[0124] In step 2.3, network-based communication is used to transmit the state observation values ​​to the local power plant generator control center. In order to reduce the number of network transmissions and reduce the network communication burden, data is transmitted to the control center only when there is a demand. Therefore, a demand-based scheduling transmission mechanism is designed.

[0125] Step 2.4: Run the predictor in the control center of the generator set, and use the predictor to predict the state prediction value of the generator set based on the unit data of the generator set.

[0126] Specifically, in order to provide data for generating control values ​​within the time interval of the data transmission, a predictor based on the generator set model (4) is run in the control center.

[0127] The predictor obtains the state prediction value of the generator set through the prediction function (7).

[0128]

[0129] where x c (t) and The predictor predicts the state value, which receives the state observation value x sent by the dispatch center s (t) is updated.

[0130] Step 2.5: Set up a backup of the predictor at the dispatch center. The dispatch center constructs a demand function based on the difference between the state observation value and the state prediction value, and designs a demand-based dispatch transmission mechanism so that when the demand function is met, the dispatch center sends back the state observation value to the generator set, and the predictor receives the state observation value synchronously and updates the state prediction value.

[0131] Specifically, does the dispatch center send the state observation value x? s (t) The design is driven by the demand-driven mechanism (DDM). The design is to drive the transmission when the difference between the state observation value and the state prediction value is large. The formula is as follows:

[0132]

[0133] The right side of the inequality sign is the transmission condition, which is determined by the norm of the observation value and the upper bounds of the disturbance and noise. Here ρ = λ||C||κ+κ ν ,κ=||E||κ ω +κ ν , κ ω With κ ν are the upper bounds of disturbance and noise, σ s >0 and σ c >0 is the parameter to be designed.

[0134] definition Update the transmission attempt sequence for the predictor and design the transmission attempt time interval to meet

[0135]

[0136] in Δ and The lower and upper bounds of the transmission attempt interval are designed to prevent continuous transmission and prevent long-term false alarm failures.

[0137] In step 2.5, the two design parameters σ in DDM are obtained:s and σ c The method comprises the following steps:

[0138] Step 2.51, parameter σ s The steps for obtaining , specifically include the following steps:

[0139] Step 2.511: Calculate the difference between the current state observation value and the predicted value e(t) = x c (t)-x(t), construct a closed-loop dynamic system:

[0140]

[0141] Where K is the controller gain matrix to be obtained, which can be obtained by the α stability margin pole configuration method, that is, assuming Φ2 = A + BK, then the gain matrix K can be obtained by solving the linear matrix inequality:

[0142]

[0143] We obtain , where P2 is any given positive definite symmetric matrix and satisfies α>0.

[0144] Step 2.512: Choose the Lyapunov function V(t)=x T (t)Px(t), where P is the Lyapunov equation is the unique solution of , where Q is an arbitrarily given positive definite symmetric matrix.

[0145] Step 2.513: For any t≥0, For a set of real numbers, we can get:

[0146] α1||x(t)||≤V(t)≤α2||x(t)|| (12)

[0147]

[0148] where α1 and α2 are the maximum and minimum eigenvalues ​​of the matrix P, respectively, γ1 is the minimum eigenvalue of the matrix Q, γ2 = ||2PBK||, γ3 = ||2PE||. Combining (8) and ||e s (t)||=||x(t)-x s (t)||<λ||d(t)|| ∞ , can be calculated within the transmission attempt interval:

[0149] ||x(t)-x c (t)||≤σ s ||x(t)||+σ c ρ+λ||d(t)|| ∞ (14)

[0150] Substitute ||d(t)|| ∞ ≤||E||κ ω +κ ν =κ to obtain:

[0151] ||e(t)||=||x(t)-x c (t)||≤σ s ||x(t)||+σ κ κ (15)

[0152] where σ κ =λσ c ||C||+λ+σ c .

[0153] Step 2.514: Substitute (16) into (14) to obtain

[0154]

[0155] If γ1-σ is satisfied s γ2>0, then we can get

[0156]

[0157] This ensures the stability of the control system during normal network communication. γ5=γ1-σ s γ2.

[0158] Based on the above steps, the selected parameter σ can be determined s The standard is

[0159] Step 2.52, parameter σ c The steps to obtain .

[0160] In this step, by selecting σ c Make the transmission attempt interval determined by DDM (8) greater than or equal to the selected interval lower bound Δ , the specific calculation steps are as follows:

[0161] Step 2.521: For any t∈[t k ,t k+1 ) Calculate the prediction error dynamics as

[0162]

[0163] It can be calculated

[0164]

[0165] in

[0166] Step 2.522: Define the difference between the predicted value and the observed value as e c (t) = x s (t)-x c (t), and its dynamics are calculated as:

[0167]

[0168] in

[0169] Step 2.523: For any t∈[t k ,t k+1 ) by e c (t k )=0, e can be calculated c The upper bound of the (t) norm is

[0170]

[0171] where ρ = λ||C||κ+κ ν .

[0172] Step 2.524: Inverse (22) to obtain tt k ≥ Δ and

[0173]

[0174] By determining the lower bound Δ The value of can be inversely solved to obtain the corresponding σ c The value of .

[0175] Network-based data transmission is a TCP-type communication method, that is, when the control end receives the observation value data, it will send back an acknowledgment (ACK) signal. When the scheduling end receives the ACK signal, it means that the transmission is successful. The predictor running at the scheduling end also updates the data at the same time. If the scheduling end fails to receive the ACK signal, the current transmission fails. The predictor running at the scheduling end still uses the prediction function to calculate the predicted value.

[0176] Step 2.6, design a switching controller to switch the output state observation value or state prediction value based on the state feedback received by the control center.

[0177] When the control center receives the state observation value, the state observation value is used as the processed unit data; when the control center does not receive the state observation value, the state prediction value is used as the processed unit data.

[0178] The designed switching controller is as follows:

[0179]

[0180] Step 3: When network communication is unreliable due to network attacks, design a flexible transmission mechanism to process the generator set data and implement flexible network security control.

[0181] Step 3.1: When there is a situation where network communication is unreliable due to malicious blocking, the malicious blocking is modeled as a denial of service (DoS) attack scenario that satisfies the duration and attack frequency.

[0182] Specifically, the network channel is of TCP type, that is, when the dispatch center transmits data but fails to receive the confirmation signal sent back by the control center, it is judged that the network environment is unreliable and malicious communication interference occurs.

[0183] Step 3.2: Design a resilient transmission mechanism to improve the system's tolerance to unreliable network environments. Use the observer to process the generator set data. The observer sends state observation values ​​to the generator set's control center based on the resilient transmission mechanism, and realizes networked safety control of the generator set by planning the data transmission sequence.

[0184] Specifically, define the time t of a successful transmission attempt k The first time that DDM is satisfied is

[0185]

[0186] If a transmission attempt occurs within the attack interval H n If the attack interval H is exceeded, the attempt will not succeed. n The set of integers of all transmission attempt times within is is a natural number set, then in the attack interval, the system switches to a periodic update rule, and the transmission attempt period is less than or equal to the lower bound of the demand-driven control update interval to reduce the transmission delay caused by DoS.

[0187] In summary, the design rules for elastic transmission attempts are as follows

[0188]

[0189] where Δ * The DoS interval periodic transmission attempt interval, and satisfies 0<Δ * ≤ Δ .

[0190] In the attack interval, the system switches to a periodic update rule, and the transmission attempt period is less than or equal to the lower bound of the demand-driven control update interval, that is, 0 < Δ * ≤ Δ , to reduce the transmission delay caused by DoS.

[0191] It should be noted that there is an upper limit to the DoS attack intensity that the system can tolerate, that is, after exceeding this limit, the stability of the controlled system can no longer be guaranteed. The following is a detailed description of the method and steps for quantifying the tolerable attack intensity.

[0192] First, two indicators describing DoS attacks are given:

[0193] (1) Attack frequency: In any time interval [t1, t2), 0≤t1<t2, the number of attack stop / start transitions n(t1, t2) satisfies:

[0194]

[0195] where η ≥ 0 and τ D ≥ Δ There are two parameters describing the attack frequency.

[0196] (2) Duration: The duration of any attack in any time interval [t1, t2), 0≤t1<t2. satisfy:

[0197]

[0198] in, and T>1 are two parameters describing the duration.

[0199] Among the four parameters, η is used to describe the number of attacks at the starting position of the time interval [t1, t2), Used to describe the duration at the starting position of the time interval [t1, t2). These two parameters can be any bounded values. It is only necessary to ensure that the above two indicators can effectively describe the actual attack situation.

[0200] Then, the timeline is divided into a set of valid DoS attack sub-intervals and the set of healthy subintervals And meet In order to avoid the overlap of two attack time intervals within the transmission attempt interval, the following definitions and calculations are made:

[0201] definition:

[0202]

[0203] Then the nth effective DoS attack sub-interval is in h0≥0 indicates a DoS attack stop / start transition sequence. May overlap, i.e. hn+1 May belong to definition is the stop / start conversion sequence of the mth effective DoS attack interval, that is,

[0204] ξ0=h0,ξ m+1 = inf{h n >ξ m |h n >h n-1 +λ n-1 +Λ n-1} (30)

[0205] The length of the mth effective DoS attack interval is but

[0206]

[0207] Combining the two indicators describing DoS attacks, we can get

[0208]

[0209] in

[0210] Finally, define is the time sequence of successful transmission attempts. Under the drive of elastic transmission logic, the sequence satisfies z0≤Ψ and in That is to say, the time interval between any two successful transmissions has an upper bound Ψ. Based on this, the upper bound of the accumulated prediction error within the time interval between any two successful transmissions can be calculated as:

[0211]

[0212] in

[0213] Combined with Lyapunov stability analysis, we can conclude that when the indicator parameter τ describing DoS attack D and T satisfy the following expressions, and This ensures that the controlled system with bounded disturbances is input-state stable.

[0214]

[0215] Step 4: Verify the effectiveness of the above solution through simulation.

[0216] Specifically, the simulation experiment contents are as follows: The unit model matrix is ​​selected as

[0217]

[0218] The unknown disturbance is a random number that satisfies uniform distribution in the interval [-0.4, 0.4], and the noise is v(t) = asin(Ω y ),Ω y ∈[0,2π], a is a random number in the interval [0,0.2] that satisfies uniform distribution. The initial condition is x(0)=x p (0) = [0.8, -1, 0, 0.3] T The target is set to [0.08-0.09-0.12-0.63] T .

[0219] Figure 3 This is the experimental verification result under normal communication conditions, which shows the effect of the unit state tracking a given target, namely the voltage and current output by the unit. The four curves in the figure are: the blue line is the d-coordinate voltage, the red line is the q-coordinate voltage, the green line is the d-coordinate current, and the black line is the q-coordinate current.

[0220] Figure 4 The upper part is the evolution of the demand-driven transmission condition, and the lower part is the value of the transmission time; (the black line in the upper part is the demand-driven condition curve, and the red line is the demand-driven threshold.) It can be seen from the figure that when the demand condition exceeds the threshold (such as 9.8s), data transmission is triggered.

[0221] Figure 5 This is the experimental verification result under unreliable communication conditions. The gray shaded area represents the time interval of communication interruption. It can be seen from the figure that when the communication is unreliable, the designed safety control algorithm can still obtain acceptable control performance.

[0222] Figure 6 This is the transmission scheduling verification under unreliable communication. It can be seen from the figure that when in the communication interruption interval, the scheduling information cannot be transmitted and is in a continuous transmission attempt state. When the attack ends, the transmission attempt succeeds and the demand-driven mechanism works normally. When in the communication interruption interval and the demand-driven condition is not met, no transmission attempt is made until the demand-driven condition is met (such as 10.52s).

[0223] Figure 7 is the evolution curve of demand driving conditions and thresholds under unreliable communication, combined with Figure 6 It can be seen that a transmission attempt occurs when the demand driving condition exceeds a threshold.

[0224] Figure 8 is the change curve of the control input.

[0225] The beneficial effect of the present invention is that, compared with the prior art, the present invention provides a set of networked safety control methods for generator sets driven by demand, mainly for local power plant units that do not participate in provincial grid coordination. The present invention only needs to establish a state space model for the generator set according to the operating mechanism, and does not need to detect and master the specific form and implementation time of malicious network interference. The cost is controllable and easy to use. By designing a demand-based scheduling transmission mechanism, the network communication cost can be minimized, and combined with the designed flexible scheduling mechanism, the scheduling signal transmission can be adaptively adjusted to minimize the deterioration of the control effect caused by network fluctuations.

[0226] Furthermore, the control method of the present invention can achieve a better control effect by relying only on simple state feedback, and can constrain unknown load current interference within a quantifiable range.

[0227] The present invention can be applied to small-power generator sets in local power plants in remote areas and distributed light energy and wind energy generator sets.

[0228] The simulation results of the present invention can fully prove that as long as the parameters are appropriate, the networked controller designed and implemented using the demand-driven switching control algorithm can achieve good control effects. Even in the case of frequent and long-lasting network congestion, the control scheme can still effectively control the generator set, and the whole process is very stable.

[0229] 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 it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A networked safety control method for generator sets based on demand-driven, characterized in that: The following steps are involved: Establish a state space model to represent the output voltage and current of the generator set; When network communication is reliable, the established state space model is used to design a demand-based dispatch transmission mechanism to process the generator set data and remove the impact of noise and disturbance on the generator set data; When network communication is unreliable due to network attacks, a flexible transmission mechanism is designed to process the generator set data and achieve flexible network security control.

2. The demand-driven networked safety control method for generator sets according to claim 1 is characterized in that: The method of establishing a state space model for representing the output voltage and current of the generator set comprises the following steps: Extract a single generator set connected to the common coupling point of the power grid system, and establish the dynamic equations of voltage and current of the single generator set in dq coordinates; According to the established dynamic equations under dq coordinates, the dynamic state space model of the generator set is established from the voltage and current of the d-component and q-component respectively.

3. The demand-driven networked safety control method for generator sets according to claim 2 is characterized in that: The dynamic equation established in dq coordinates is as follows: Where GU i represents the generator set connected to the i-th node in the common coupling point, Line ij represents the line connecting generator set i and generator set j, t represents time, V i,dq Represents the output voltage of unit i, V j,dq represents the output voltage of unit j, I ti,dq Represents the output current of unit i, I Li,dq Represents the load current, I ij,dq Indicates the line current connecting unit i and unit j, L ti Represents the equivalent inductance of unit i, L ij represents the line inductance connecting unit i and unit j, R ij Represents the line resistance connecting unit i and unit j, R ti Represents the equivalent resistance of the series filter, C ti represents the parallel capacitance, k i represents the transformer conversion rate, ω0 represents the dq model rotation speed, V ti,dq Represents the voltage source voltage.

4. The demand-driven networked safety control method for generator sets according to claim 3 is characterized in that: According to the established dynamic equations under dq coordinates, the dynamic state space model of the generator set is established from the voltage and current of the d-component and q-component respectively as follows: Among them, V i,d Represents the d component of the output voltage of unit i, V i,q represents the q component of the output voltage of unit i, V ti,d represents the d component of the voltage source voltage, V ti,q represents the d component of the voltage source voltage, represents the d component of the line current connecting units i and j, represents the q component of the line current connecting units i and j, I ti,d The d component of the output current of unit i, I ti,q Represents the d component of the output current of unit i.

5. The demand-driven networked safety control method for generator sets according to claim 1 is characterized in that: When the network communication is reliable, the established state space model is used to design a demand-based scheduling transmission mechanism to process the generator set data and remove the influence of noise and disturbance on the generator set data, including the following steps: Design an observer integrated with the sensor based on the established spatial state model; The sensor monitors the generator set data in real time, and the observer processes the disturbance and noise of the real-time generator set data to obtain the state observation value; Design a predictor in the control center of the generator set, and obtain the state prediction value of the generator set based on the unit data of the generator set; Based on the difference between the state observation value and the state prediction value, a demand function is constructed, and a demand-based scheduling transmission mechanism is designed so that when the demand function is met, the observer transmits the state observation value to the control center of the generator set, and the predictor receives the state observation value synchronously and updates the state prediction value; When the control center receives the state observation value, the state observation value is used as the processed unit data; when the control center does not receive the state observation value, the state prediction value is used as the processed unit data.

6. The demand-driven networked safety control method for generator sets according to claim 5 is characterized in that: When network communication is unreliable due to network attacks, a flexible transmission mechanism is designed to process the generator set data to achieve flexible network security control, including the following steps: When there is a situation where network communication is unreliable due to malicious blocking, the malicious blocking is modeled as a DoS attack scenario that satisfies the duration and attack frequency; An elastic transmission mechanism is designed and the observer is used to process the generator set data. The observer sends the state observation value to the control center of the generator set based on the elastic transmission mechanism to realize elastic networked safety control.

7. The demand-driven networked safety control method for generator sets according to claim 6 is characterized in that: After the state space model is established, the established state space model is converted into the following model: y(t)=Cx(t)+ν(t) in, X ij =ω0L ij , Z ij =|R ij +jX ij |. x(t)=[V i,d ,V i,q ,I ti,d ,I ti,q ] T is the state vector, is the first-order derivative of x(t), u(t)=[V ti,d ,V ti,q ] T is the input vector, y(t) represents V i,d ,V i,q ,I ti,d ,I ti,q Four measured values, ν(t) represents noise, ω(t) = [I Li,d ,I Li,q ] T is the disturbance; V i,d Represents the d component of the output voltage of unit i, V i,q represents the q component of the output voltage of unit i, V ti,d represents the d component of the voltage source voltage, V ti,q represents the d component of the voltage source voltage, I ti,d The d component of the output current of unit i, I ti,q Represents the d component of the output current of unit i; R ij Represents the line resistance connecting unit i and unit j, R ti Represents the equivalent resistance of the series filter, L ij Indicates the line inductance connecting unit i and unit j, L ti represents the equivalent inductance of unit i, C ti represents the parallel capacitance, ω0 represents the rotation speed of the dq model; The measured value of y(t) is obtained, and the four matrices A, B, C, and E are solved using the above formula.

8. The demand-driven networked safety control method for generator sets according to claim 7 is characterized in that: According to the obtained y(t) measurement value and the four matrices A, B, C, and E, the observer is designed as follows: Among them, x s (t) represents the state observation value, Represents x s The first-order derivative of (t), L represents the observer gain matrix, is obtained by solving the following linear matrix inequality: Among them, Φ1=A-LC, Y=[IL], I is the unit matrix, λ is any given parameter that makes the linear matrix inequality solvable, and P1 is any given positive definite symmetric matrix that makes the linear matrix inequality solvable.

9. The demand-driven networked safety control method for generator sets according to claim 8, characterized in that: The observer gain matrix L satisfies the following performance criteria to constrain disturbances and noise: ||e s (t)||2<λ||d(t)||2 in, 10. The demand-driven networked safety control method for generator sets according to claim 8, characterized in that: The predictor obtains the following calculation formula for the state prediction value of the generator set according to the generator set data: Among them, x c (t) represents the state observation value x received from the observer s The predicted state value at time (t), x c (t) represents the state observation value x that has not been received from the observer s The predicted state value at time (t).

11. The demand-driven networked safety control method for generator sets according to claim 9, characterized in that: The demand function is constructed based on the difference between the state observation value and the state prediction value as follows: Among them, σ s ||x s (t)||+σ c ρ is the transmission condition, which is determined by the norm of the observation value and the upper bounds of the disturbance and noise, where ρ = λ||C||κ+κ ν ,κ=||E||κ ω +κ ν , κ ω With κ ν are the upper bounds of disturbance and noise, σ s >0 and σ c >0 is the parameter to be designed.

12. The demand-driven networked safety control method for generator sets according to claim 11, characterized in that: Parameter σ s The steps to obtain are as follows: Calculate the difference between the current state and the predicted state e(t) = x c (t)-x(t), the closed-loop dynamic system is constructed as follows: Select the Lyapunov function as V(t)=x T (t)Px(t), where P is the Lyapunov equation The only solution of , Q is any given positive definite symmetric matrix; For any t≥0, we can get: α1||x(t)||≤V(t)≤α2||x(t)|| Where α1 and α2 are the maximum and minimum eigenvalues ​​of the matrix P, respectively, γ1 is the minimum eigenvalue of the matrix Q, γ2 = ||2PBK||, γ3 = ||2PE||, combined with the constructed demand function and ||e s (t)||=||x(t)-x s (t)||<λ||d(t)|| ∞ , can be calculated within the transmission attempt interval: ||x(t)-x c (t)||≤σ s ||x(t)||+σ c ρ+λ||d(t)|| ∞ Substitute ||d(t)|| ∞ ≤||E||κ ω +κ ν =κ to obtain: ||e(t)||=||x(t)-x c (t)||≤σ s ||x(t)||+σ κ κ where σ κ =λσ c ||C||+λ+σ c , substituting into: If γ1-σ is satisfied s If γ2>0, we can get: This ensures the stability of the control system during normal network communication. γ5=γ1-σ s γ2; Based on the above steps, determine the selected parameter σ s The standard is 13. The demand-driven networked safety control method for generator sets according to claim 12, characterized in that: Parameter σ c The steps to obtain are as follows: For any t∈[t k ,t k+1 ) Calculate the prediction error dynamics as: It can be calculated that: in Define the difference between the predicted state and the observed state as e c (t) = x s (t)-x c (t), and its dynamics are calculated as: in For any t∈[t k ,t k+1 ) by e c (t k )=0, e can be calculated c The upper bound of the (t) norm is: Where p=λ||C||κ+κ ν ; Reverse the above equation to get tt k ≥ Δ and By determining the lower bound Δ The corresponding σ is obtained by inversely solving the value of c The value of .

14. The demand-driven networked safety control method for generator sets according to claim 13, characterized in that: The elastic transmission mechanism is expressed as the following formula: Among them, Δ * Indicates the DoS interval periodic transmission attempt interval, and satisfies 0<Δ * ≤ Δ , Indicates the attack range, is the set of real numbers, is a set of natural numbers, t k Indicates the time of a successful transmission attempt.

15. The demand-driven networked safety control method for generator sets according to claim 14, characterized in that: The attack frequency constraint is expressed as: Among them, η and τ D Two parameters representing the attack frequency at the beginning of the time interval [t1, t2), η≥0, τ D ≥ Δ , is the set of real numbers, is a set of natural numbers.

16. The demand-driven networked safety control method for generator sets according to claim 15, characterized in that: The attack duration constraint is expressed as: in, and T represent two parameters of duration at the start of the time interval [t1, t2), is the set of real numbers, is a set of natural numbers.

17. The demand-driven networked safety control method for generator sets according to claim 16, characterized in that: In order to ensure that the controlled system with bounded disturbance is input-state stable, the following expression is established: According to the above formula, Δ * , τ D and T are constrained.

18. A system using the networked safety control method for generator sets based on demand drive according to any one of claims 1 to 17, characterized in that: The system includes: an observer, a predictor and a switching controller which are integrated with the sensor; The sensor is used to monitor the unit data of the generator set in real time; The observer is used to process the unit data and send the state control value to the control center of the generator set based on the required scheduling transmission mechanism or elastic transmission mechanism; The predictor is arranged in the control center, and is used to obtain the predicted state observation value according to the measured unit data, and to update the state prediction value when receiving the state control value; The switching controller is used to switch the output state observation value or the state prediction value based on the state feedback of the state control value received by the control center.

Images