Battery energy storage secondary frequency modulation control method and system based on dynamic participation factor and adaptive model prediction control

Abstract

The invention provides a battery energy storage secondary frequency modulation control method and system based on dynamic participation factors and adaptive model predictive control, and the method comprises the following steps: S1, taking a frequency response model of a classical double-region interconnected power grid as a research object, and employing a fixed frequency and tie line power deviation control mode TBC to establish a fire storage combined frequency modulation equivalent model; wherein the thermal power generating unit participates in primary and secondary frequency modulation of a power grid, and the battery energy storage only participates in secondary frequency modulation; s2, adopting a double-layer judgment mechanism as a division basis of frequency modulation responsibility between battery energy storage and a thermal power generating unit; s3, taking the frequency modulation responsibility obtained by the dynamic energy storage participation factor as an optimization target, and utilizing an adaptive model to predict and control MPC to issue an instruction signal to enable energy storage to act in advance in frequency modulation control; and S4, formulating a frequency modulation control strategy considering SOC recovery according to the state interval of the regional control deviation signal ACE and the energy storage SOC. According to the method, the conflict between energy storage frequency modulation and SOC recovery is effectively relieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power grid secondary frequency modulation, and particularly relates to a battery energy storage secondary frequency modulation control method and system based on a dynamic participation factor and an adaptive model predictive control. BACKGROUND

[0002] High proportion of renewable energy grid-connected reduces the inertia of the power system, brings more serious intermittency and uncertainty to the power grid, further aggravates the power imbalance between power supply and load, and makes it impossible to guarantee the frequency stability.

[0003] Energy storage can improve the load frequency control performance, reduce the wear and tear caused by the deep action frequency and frequent climbing of traditional generator units, and has become a hot spot in the field of frequency modulation. At the same time, giving full play to the respective advantages of energy storage and traditional units, improving the frequency modulation effect of the power grid and prolonging the service life of energy storage batteries are key problems to promote large-scale application of energy storage. Battery energy storage is a new type of power device with strong frequency regulation capability, which can be bidirectionally charged and discharged, has fast response speed and high regulation accuracy.

[0004] Model predictive control (MPC) can solve the AGC problem with multiple constraints compared with the traditional proportional integral derivative (PID) controller, and is an advanced control method. MPC optimizes the control objective within the set prediction time range by predicting the future dynamic behavior of the system, and has excellent response speed and robustness. Current MPC methods focus more on the frequency regulation capability of energy storage, and the research on SOC recovery is limited, and the problem of reasonably recovering SOC during frequency modulation to avoid the loss of regulation capability of energy storage in the long-term frequency modulation process is not deeply discussed. The MPC objective function can be composed of multiple sub-targets, and the main target of minimizing the system frequency deviation can also consider the performance index of SOC recovery of energy storage, but the completely fixed weight of the objective function is difficult to adapt to the dynamic demand of frequency regulation and SOC recovery of energy storage in different periods, resulting in a conflict between the frequency modulation effect and the long-term availability of energy storage. That is, when the SOC of energy storage needs to be recovered, if the frequency regulation weight is too high, it is easy to cause the SOC to not return to a reasonable level in time, affecting the continuous effectiveness of energy storage; on the contrary, in the case of emergency frequency regulation, if the SOC recovery is overemphasized, it may lead to untimely frequency response and weaken the stability of the power grid. Therefore, it is necessary to study the MPC weight coefficient adjustment method with the objective function of reducing ACE and recovering the SOC of energy storage to balance the demand of energy storage frequency modulation and recovery.

[0005] The existing research ignores the characteristics of continuous change and irregular high-frequency small signal of random load disturbance in the actual power system when performing the responsibility allocation of the fire storage frequency modulation, so that the design of the participation factor lacks flexibility and cannot fully play the respective frequency modulation advantages of the energy storage and the thermal power unit in the complex load fluctuation scenario.

[0006] Based on this, the application provides a power grid secondary frequency modulation method and system based on dynamic participation factor and adaptive model predictive control considering energy storage SOC recovery. SUMMARY

[0007] In view of the deficiencies of the existing research, the application provides a power grid secondary frequency modulation method and system based on dynamic participation factor and adaptive model predictive control considering energy storage SOC recovery. The application realizes dynamic adjustment of the SOC recovery strength in different operating states, effectively alleviates the conflict between energy storage frequency modulation and SOC recovery, and improves the bidirectional frequency modulation capability of the energy storage in the full operating condition range.

[0008] To achieve the above purpose, the technical scheme of the application is as follows:

[0009] In the first aspect, the application provides a battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control, which specifically includes the following steps:

[0010] S1, taking the classical double-area interconnected power grid frequency response model as the research object, and adopting the constant frequency and tie-line power deviation control mode TBC, an equivalent model of fire storage combined frequency modulation is established; wherein the thermal power unit participates in the primary and secondary frequency modulation of the power grid, and the battery energy storage only participates in the secondary frequency modulation;

[0011] S2, a double-layer judgment mechanism is used as the basis for dividing the frequency modulation responsibility between the battery energy storage and the thermal power unit;

[0012] S3, taking the frequency modulation responsibility obtained by the dynamic energy storage participation factor as the optimization target, using the adaptive model predictive control MPC to issue command signals to make the energy storage act in advance in the frequency modulation control, and weakening the impact of the load disturbance on the frequency;

[0013] S4, according to the area control error signal ACE and the state interval of the energy storage SOC, comprehensively considering the power grid frequency modulation demand and the SOC recovery demand in each stage, the dual objectives of system frequency regulation and energy storage SOC recovery are realized by formulating a frequency modulation control strategy considering SOC recovery.

[0014] Preferably, the power grid dispatching center calculates the ACE signal according to the monitored frequency deviation and tie-line power deviation as shown in formula (1):

[0015] ACE = ΔPtie + b1Δf1 (1)

[0016] Where, ΔP tie is the tie-line exchange power deviation, b i is the frequency deviation coefficient, Δf i is the system frequency deviation.

[0017] Preferably, the equivalent model of each frequency-regulating unit in the system is as follows:

[0018] The equivalent transfer function model of the thermal power unit is as follows:

[0019]

[0020] Wherein, T G , T RH , F HP and T CH are the speed governor time constant, reheater time constant, reheater gain and turbine time constant of the thermal power unit respectively.

[0021] The equivalent transfer function model of the battery energy storage is as follows:

[0022]

[0023] Wherein, T B is the response time constant of the battery energy storage output variable.

[0024] Preferably, the ratio of the available capacity of the battery energy storage to the rated capacity represents the state of charge SOC, and the calculation formula is as follows:

[0025]

[0026] Wherein, S OC (t) is the SOC of the battery energy storage at time t; Δt is the sampling time interval; P b (t) is the charging and discharging power of the battery energy storage at time t, and the discharging direction is positive; E N is the rated capacity of the battery energy storage.

[0027] Preferably, the double-layer judgment mechanism is as follows:

[0028] The first layer judgment mechanism: the energy storage participation factor a is divided into two parts with 0.5 as the dividing point, including the larger energy storage participation factor interval 0.5

[0029] When the sensitivity S is less than 0, the system frequency deviation is smaller with the increase of the energy storage participation factor, so the energy storage participation factor is selected in a larger energy storage participation factor interval to facilitate the frequency deviation adjustment; if the sensitivity S is greater than 0, the system frequency deviation is smaller with the decrease of the energy storage participation factor, so the energy storage participation factor is selected in a smaller energy storage participation factor interval to facilitate the frequency recovery;

[0030] The second layer judgment mechanism: according to the actual load fluctuation and the load fluctuation rate at the current time, the load fluctuation in the short term is predicted, and then the predicted result is used to dynamically adjust the energy storage participation factor in the corresponding interval, so as to adaptively determine the time and depth of the energy storage participation frequency regulation.

[0031] Preferably, the dynamic adjustment of the energy storage participation factor in different intervals is realized by means of a fuzzy controller; at the same time, based on the load disturbance ΔP L and the load fluctuation rate γ, the load development trend at the next time is preliminarily judged, as shown in formula (5):

[0032] ΔP L (t+1)=ΔP L (t)+γ(t)Δt (5)

[0033] The fuzzy control rule is designed as follows:

[0034] If the load disturbance is positive; when the absolute value of ΔP L is large and γ>0, a large energy storage participation factor is set in the interval; when the absolute value of ΔP L is small and γ>0, if the absolute value of γ is small, a small energy storage participation factor is set in the interval, and if the absolute value of γ is large, a large energy storage participation factor is set in the interval; when γ<0, a small energy storage participation factor is set in the interval;

[0035] If the system is not affected by disturbance at the current time, but γ>0, a small energy storage participation factor is set in the interval;

[0036] When the load disturbance is negative; when the absolute value of ΔP L is large and γ<0, a large energy storage participation factor is set in the interval; when the absolute value of ΔP L is small and γ<0, if the absolute value of γ is large, a large energy storage participation factor is set in the interval, and if the absolute value of γ is small, a small energy storage participation factor is set in the interval; when γ>0, a small energy storage participation factor is set in the interval;

[0037] If the system is not affected by disturbance at the current time, but γ<0, a small energy storage participation factor is set in the interval.

[0038] Preferably, a double-input single-output fuzzy controller is selected to adjust the energy storage participation factor; the two inputs are the normalized ΔP LWith γ, the fuzzy mapping of high-frequency irregular load state is realized by using Gaussian membership function, so as to facilitate fuzzy reasoning; the output is the energy storage participation factor a, and the membership function thereof is designed based on discrete domain, and a plurality of representative dynamic energy storage participation factor values are selected to cover different load states; ΔP L The quantification factors of ΔP

[0039]

[0040] In the formula, ΔP Lmax and γ max are the maximum values of load disturbance and load change rate respectively; the load disturbance ΔP L and the load change rate γ are normalized, and the domain is [-1, 1], and a takes the discrete domain [0.1, 0.3, 0.5, 0.7, 0.9]; the fuzzy sets NB, NS, ZO, PS and PB are set to describe the value interval of ΔP L , γ and a, wherein NB, NS, ZO, PS and PB respectively represent negative large, negative small, zero, positive small and positive large of the load disturbance and the load fluctuation change rate.

[0041] Preferably, the state space equation of the frequency dynamic response model of the dual-area interconnected power grid with energy storage is shown in formula (7):

[0042]

[0043] In the formula, A, B, G and C respectively represent the state matrix, the control matrix, the input matrix and the output matrix of the system; x, u, d and y respectively represent the state variable, the control variable, the input variable and the output variable of the system; and the specific elements are shown in formulas (8)-(15):

[0044]

[0045] x = [Δf1 ΔP tie ΔP b SOC] T (12)

[0046] u = Δu b (13)

[0047] d = [ΔP L Δf2 ΔP g1 ] T (14)

[0048] y = [ACE a SOC] T (15)

[0049] In the formula, D1 represents the load damping coefficient of region 1, M1 represents the system moment of inertia of region 1, T 12 represents the power synchronization coefficient of the inter-regional tie line, and ΔP b represents the output of the battery energy storage, and Δu b represents the output control signal of the battery energy storage, and Δf2 represents the frequency deviation of region 2, and ΔP g1 represents the output of the thermal power unit in region 1, and ACE a represents the frequency modulation responsibility of the energy storage

[0050] with T s as the sampling period, formula (7) is discretized and converted into the following discrete-time state space model:

[0051]

[0052] In the formula: A, B, and G represent the state matrix, the control matrix, and the input matrix of the system in the discrete time, respectively; and T s is the discrete sampling time interval of the system;

[0053] ACE a = 0 is taken as the core optimization objective of the MPC controller; at time k, the following quadratic performance index function is constructed:

[0054]

[0055] s.t.

[0056] y min ≤(y(k+j|k)≤y max ,j=1,2,...p (18)

[0057] u min ≤(u(k+i|k)≤u max ,i=1,2,...c (19)

[0058] In the formula: Q and R are the output weight diagonal matrix and the control weight diagonal matrix, respectively; p and c are the prediction time domain and the control time domain, respectively; y(k+j|k) is the prediction of the system output at time k+j in the future at time k; u(k+i-1|k) is the prediction of the system control variable at time k+i-1 in the future at time k; y r is the system output reference value, and the reference value is set to 0; y min and y max are the upper and lower limits of the system output variable, that is, the ACE a constraint and the SOC charge and discharge threshold; u min and u maxare the upper and lower limits of the system control variable, i.e. the energy storage output power constraint; by solving the finite time domain optimization problem of the objective function (17), a set of optimal control sequences u(k+i-1|k) is obtained, only the first item in the control sequence is applied to the battery energy storage to achieve immediate optimization adjustment; at the next sampling time, the optimization problem is solved again and the optimization process of applying the first item of the optimal control sequence to the battery energy storage is repeated, and through the rolling optimization method, the optimal performance of the system at different times is ensured.

[0059] Preferably, S4 is specifically as follows:

[0060] According to the threshold size, the ACE is divided into different interval types, including a dead zone, a normal regulation zone, an emergency regulation zone, and a super emergency regulation zone, wherein ACE1 and ACE2 are the demarcation values of adjacent two intervals of the ACE; the SOC is divided into a normal charging and discharging zone, a charging priority zone, a discharging priority zone, and a prohibited charging and discharging zone, wherein S min , S low , S high , S max are the demarcation values of adjacent two intervals of the SOC;

[0061] A weighted item for adjusting the SOC recovery of the energy storage is added to the objective function, and a weight factor of the weighted item is represented by q, and the size of q represents the importance of the SOC recovery;

[0062] According to the ACE and SOC partitioning, the control strategy is as follows:

[0063] (1) When the ACE is located in the dead zone range, at this time, the grid frequency deviation is extremely small, and the frequency regulation demand is basically ignored; without considering the current SOC state, the target is SOC=0.5, the energy storage fully performs SOC recovery and does not participate in frequency regulation, at this time, the weight coefficient q is set to the maximum value 0.5;

[0064] (2) When the ACE is located in the normal regulation zone, the primary task of the energy storage is to respond to the grid frequency regulation demand, supplemented by SOC recovery;

[0065] (3) When the ACE is located in the emergency regulation zone, the recovery of the grid frequency is the highest priority, without considering the state of charge, the energy storage only participates in frequency regulation, and the weight coefficient q is set to 0;

[0066] (4) When the ACE is located in the super emergency regulation zone, it means that the grid frequency has deviated from the normal range, and emergency measures such as generator tripping or load shedding are adopted until the grid frequency is restored to stability.

[0067] In another aspect, the application also provides a battery energy storage secondary frequency modulation control system based on a dynamic participation factor and adaptive model predictive control, comprising a processor, a memory, and a computer program stored in the memory, wherein the processor executes the computer program to specifically execute the steps in any of the above battery energy storage secondary frequency modulation control methods based on a dynamic participation factor and adaptive model predictive control.

[0068] Compared with the prior art, the application has the following beneficial effects:

[0069] A hierarchical judgment mechanism considering the sensitivity, frequency deviation and load fluctuation influencing factors is designed, and a dynamic participation factor strategy based on sensitivity calculation and double fuzzy control to determine the participation degree of thermal storage frequency modulation is proposed. The responsibility allocation of thermal storage frequency modulation is optimized, the rapid response of energy storage and the continuous adjustment ability of thermal power units are exerted, so as to improve the overall frequency modulation performance. A storage frequency modulation and recovery control method based on area control error (ACE) and storage state of charge (SOC) partition control and adaptive model predictive control (MPC) is proposed. The prediction and optimization ability of MPC is used to enhance the foresight and accuracy of the frequency modulation process, realize the optimization of energy storage charging and discharging decision, and flexibly adjust the weight coefficient of the storage SOC recovery term according to the real-time state information, realize the dynamic adjustment of the SOC recovery strength in different operating states, effectively alleviate the conflict between storage frequency modulation and SOC recovery, and improve the bidirectional frequency modulation ability of storage in the whole operating condition range. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 A schematic diagram of thermal storage joint participation in regional power grid frequency modulation;

[0071] Figure 2 A dynamic model diagram of two-area system frequency modulation containing battery energy storage;

[0072] Figure 3 A secondary frequency modulation control strategy architecture diagram;

[0073] Figure 4 A regional power system frequency response model diagram under ACE control mode;

[0074] Figure 5 A sensitivity, energy storage participation factor and system frequency deviation relationship curve;

[0075] Figure 6 An ACE and SOC state interval division diagram;

[0076] Figure 7 A SOC recovery term weight value working condition diagram;

[0077] Figure 8 for the frequency deviation response curve of region 1 under step disturbance;

[0078] Figure 9 for the load continuous disturbance graph;

[0079] Figure 10 for the frequency curve of region 1 under continuous disturbance;

[0080] Figure 11 for the SOC curve of energy storage under continuous disturbance; DETAILED DESCRIPTION

[0081] The technical solutions of the present application will be specifically described below in combination with the accompanying Figures 1-11 , the technical solutions of the present application will be specifically described below in combination with the accompanying

[0082] The present application proposes a battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control, which is specifically as follows:

[0083] Step S1: Taking the classic two-area interconnected power grid frequency response model as the research object, and adopting the constant frequency and tie-line power deviation (Tie-line Bais Control, TBC) control mode, an equivalent model of thermal and battery energy storage combined frequency modulation is established. Among them, the thermal generating unit participates in the primary and secondary frequency modulation of the power grid, and the battery energy storage only participates in the secondary frequency modulation. ACE is the regional control deviation signal, and the power grid dispatching center calculates the ACE signal according to the monitored frequency deviation and tie-line power deviation as follows:

[0084] ACE = ΔP tie +b1Δf1

[0085] In the formula, ΔP tie is the tie-line exchange power deviation, b i is the frequency deviation coefficient, Δf i is the system frequency deviation, and the ACE signal is sent to the frequency modulation power source according to a certain distribution mode after being processed by the controller. The equivalent model of each frequency modulation unit in the system is as follows:

[0086] 1) Thermal generating unit equivalent transfer function model:

[0087]

[0088] In the formula: T G , T RH , F HP and T CH are the time constant of the thermal generating unit governor, the reheater time constant, the reheater gain and the turbine time constant respectively.

[0089] 2) Battery energy storage equivalent transfer function model:

[0090]

[0091] In the formula: T B is the response time constant of the battery energy storage output variable.

[0092] The ratio of the available capacity of the battery energy storage to the rated capacity represents the state of charge SOC, and the calculation formula is as follows:

[0093]

[0094] In the formula: S OC (t) is the size of the battery energy storage SOC at time t; Δt is the sampling time interval; P b (t) is the charging and discharging power of the battery energy storage at time t, and the discharging direction of the energy storage is positive; E N is the rated capacity of the battery energy storage.

[0095] Step S2: Adopt the allocation mode based on the ACE signal. In this mode, the secondary frequency modulation control loop only passes through the PI controller of the thermal power unit, and the energy storage directly responds to part of the ACE signal to exert its rapid characteristics. In the complex frequency domain, the sensitivity principle is applied to analyze the relationship between the action depth of the energy storage participating in frequency modulation, i.e., the participation factor and the system frequency deviation. The sensitivity S is defined as follows: when the independent variable x changes by 1%, the function y(x) will correspondingly change by S%, and S>0, x and y(x) change in the same direction, S<0, x and y(x) change in opposite directions. The system frequency deviation, the thermal power unit and the battery energy storage output variable can be represented as follows:

[0096]

[0097] ΔP b (s) = -abG b (s)Δf(s) (25)

[0098] Substitute formula (24) and formula (25) into formula (23) to obtain:

[0099]

[0100] Further partial derivative of formula (26) gives the sensitivity of the energy storage participation factor a to the frequency deviation:

[0101]

[0102] Since Δf(s) and a are not dimensionally consistent, it may lead to unclear physical meaning, so formula (9) needs to be dimensionless. By expressing the changes of Δf(s) and a in the form of relative change, the dimensional difference is eliminated. The dimensionless result of the sensitivity S is as follows:

[0103]

[0104] In addition, the mutual relationship among the sensitivity, the energy storage participation factor and the system frequency deviation is verified in combination with the step load disturbance. It can be seen that in the initial stage of the disturbance, the sensitivity S is less than 0, and with the increase of the energy storage participation factor a, the S becomes smaller and the system frequency deviation is also smaller. In the final stage of the disturbance, S is greater than 0, and at this time, the smaller the a is, the smaller the sensitivity S is, and the smaller the system frequency deviation is.

[0105] Step S31: A double-layer judgment mechanism is used as the basis for dividing the frequency modulation responsibility between the battery energy storage and the thermal power unit.

[0106] The first layer judgment mechanism: As can be known from the sensitivity principle analysis in step S2, when the sensitivity S is less than 0, the system frequency deviation is smaller with the increase of the energy storage participation factor, so it is more beneficial for the frequency deviation adjustment that the energy storage bears a higher proportion of the frequency modulation responsibility in this stage. If the sensitivity S is greater than 0, the system frequency deviation is smaller with the decrease of the energy storage participation factor, so it is more beneficial for the frequency recovery that the energy storage bears a smaller proportion of the frequency modulation responsibility. Therefore, the energy storage participation factor is divided into two parts by taking 0.5 as the dividing point, including the larger energy storage participation factor interval (0.5 < a < 1) and the smaller energy storage participation factor interval (0 < a < 0.5), and the interval range of the participation factor is determined by the positivity of the sensitivity S. It should be noted that the setting of this dividing point is not directly corresponding to the capacity or power proportion of the energy storage and the thermal power unit, but is a reasonable scheme for balancing the fast response characteristics of the energy storage and the sustained steady adjustment ability of the thermal power unit in different frequency modulation stages. When actually selecting the dividing point of the energy storage participation factor, the overall state of the system (such as the capacity ratio of thermal power and energy storage, power range, load disturbance intensity and other factors) should be adjusted flexibly to adapt to different system conditions and frequency modulation needs.

[0107] The second layer judgment mechanism: The participation factor needs to be selected according to the external conditions such as the system frequency modulation demand and the power frequency modulation capacity, but the fixed size of the participation factor has limitations in the face of continuous irregular load disturbance, and it is difficult to continuously cope with the frequent changes of the frequency modulation needs in the actual scene. Therefore, it is necessary to adjust the energy storage participation factor in advance within the preset reference interval by reasonably predicting the load fluctuation in the next moment. Specifically, the load fluctuation in the short term is predicted according to the actual load fluctuation and the load fluctuation rate at the current moment, and then the participation factor is dynamically adjusted in the corresponding interval to adaptively determine the timing and depth of the energy storage participation in frequency modulation.

[0108] Step S32: Since the second layer judgment mechanism designed has obvious fuzziness and nonlinearity, a fuzzy controller is used to realize the dynamic adjustment of the energy storage participation factor in different intervals. At the same time, based on the current load disturbance ΔP L and the load fluctuation rate γ, the load development trend in the next moment can be preliminarily judged, as shown in formula (29):

[0109] ΔP L (t+1) = ΔP L (t) + γ(t)Δt (29)

[0110] The fuzzy control rule is designed as follows:

[0111] It is assumed that the load disturbance is positive. When ΔP L is large and γ>0, it is judged that the load fluctuation prediction in the next period is large, and the energy storage needs to participate in frequency regulation with large power, so the participation factor is set to be large. When ΔP L is small, γ>0, but γ is small, which indicates that the load increment is small, so it is judged that the load fluctuation prediction in the next period is small, and the energy storage output is reduced accordingly, so the participation factor can be set to be small; if γ is large, it indicates that the load increment is large, and it is judged that the load fluctuation prediction is large, and the energy storage needs to provide large regulation power, so the participation factor is set to be large. When γ<0, it indicates that the load fluctuation is gradually decreasing, that is, the load fluctuation prediction is small, and the energy storage output is reduced, so the participation factor is set to be small. If the system is not affected by the disturbance at the current time, but γ>0, the load fluctuation prediction is relatively small, the energy storage output is small, and the participation factor is set to be small. When the load disturbance is negative, the design idea is similar to that of the positive value, which will not be described here.

[0112] Step S33: The fuzzy controller with two inputs and single output is selected to adjust the energy storage participation factor. The two inputs are ΔP L and γ after normalization, the Gaussian membership function is used to realize the fuzzy mapping of the high-frequency irregular load state more smoothly and continuously, which is convenient for fuzzy reasoning; the output is the energy storage participation factor a, and the membership function is designed based on the discrete domain, and a number of representative dynamic factor values are selected to cover different load states, which simplifies the complexity and improves the response speed of the system. The quantization factor calculation method of ΔP L and γ is shown in formula (30):

[0113]

[0114] In the formula: ΔP Lmax and γ max are the maximum value of the load disturbance and the maximum value of the load change rate respectively.

[0115] The load disturbance ΔP L and the load change rate γ are normalized to the domain [-1, 1], and a takes the discrete domain [0.1, 0.3, 0.5, 0.7, 0.9]. The fuzzy sets NB (negative large), NS (negative small), ZO (zero), PS (positive small), and PB (positive big) are set to describe ΔP LThe value intervals of the input, output and a are shown by graphical method. The fuzzy set intervals of the load disturbance and its change rate and the energy storage participation factor are shown, the logic process of input fuzzification and output defuzzification of the fuzzy controller is corresponded, and the design process of the fuzzy controller from "load state perception" to "frequency modulation responsibility allocation" is embodied. The fuzzy control law intuitively reflects the adjustment strategy of the energy storage participation factor under different load state input combinations through the control surface. The fuzzy control rules are divided into intervals based on the positive and negative of the sensitivity S, and the adjustment logic of the energy storage participation factor under different load scenarios is corresponded, as shown in Tables 1 and 2.

[0116] Table 1 S>0 energy storage participation factor control rule language table

[0117]

[0118] Table 2 S<0 energy storage participation factor control rule language table

[0119]

[0120] Step S41: The frequency modulation responsibility based on the dynamic energy storage participation factor is taken as an optimization target, the energy storage is made to act in advance in frequency modulation control by using the forward-looking instruction signal of the MPC, and the impact of the load disturbance on the frequency is weakened. The state space equation of the frequency dynamic response model of the two-region interconnected power grid containing energy storage is as shown in formula (31):

[0121]

[0122] A, B, G and C represent the state matrix, the control matrix, the input matrix and the output matrix of the system respectively; x, u, d and y represent the state variable, the control variable, the input variable and the output variable of the system respectively. The specific elements are as shown in formulas (32)-(39):

[0123]

[0124] x = [Δf1 ΔP tie ΔP b SOC] T (35)

[0125] u = Δu b (36)

[0126] d = [ΔP L Δf2 ΔP g1 ] T (37)

[0127] y = [ACE a SOC] T (38)

[0128] In the formula, D1 represents the load damping coefficient of region 1, M1 represents the system moment of inertia of region 1, T 12 represents the power synchronization coefficient of the inter-regional tie line, and ΔP b represents the output of the battery energy storage, and Δu b represents the output control signal of the battery energy storage, Δf2 represents the frequency deviation of region 2, and ΔP g1 represents the output of the thermal power unit in region 1, and ACE a represents the size of the frequency regulation responsibility of the energy storage.

[0129] T s is the sampling period, formula (31) is discretized and converted into the following discrete-time state space model:

[0130]

[0131] In the formula: A, B, and G represent the state matrix, control matrix, and input matrix of the system in the discrete time, respectively; and T s is the discrete sampling time interval of the system.

[0132] Step S42: Generally, the AGC of the power system aims to adjust ACE to 0. The frequency regulation responsibility ACE of the energy storage based on the dynamic participation factor a is always strictly proportional to ACE, so ACE a = 0 can be used as the core optimization target of the MPC controller. At time k, the following quadratic performance index function is constructed:

[0133]

[0134] s.t.

[0135] y min ≤(y(k+j|k)≤y max ,j=1,2,...p (41)

[0136] u min ≤(u(k+i|k)≤u max ,i=1,2,...c (42)

[0137] In the formula: Q and R are the output weight diagonal matrix and the control weight diagonal matrix, respectively; p and c are the prediction time domain and the control time domain, respectively; y(k+j|k) is the prediction of the system output at time k+j in the future at time k; u(k+i-1|k) is the prediction of the system control variable at time k+i-1 in the future at time k; y r is the system output reference value, because the control target is ACE a = 0, so the reference value is set to 0. y min , ymax The upper and lower limits of the system output variable, i.e., the system ACE a The constraints and SOC charging and discharging thresholds; u min , u max The upper and lower limits of the system control variable, i.e., the energy storage output power constraint. By solving the finite time domain optimization problem of the objective function (40), a set of optimal control sequences u(k+i-1|k) can be obtained, only the first item in the control sequence is applied to the battery energy storage, and the optimization adjustment is realized immediately. At the next sampling time, the optimization problem is solved again and the above process is repeated, and through the rolling optimization, the optimal performance of the system at different times is ensured.

[0138] Step S51: The present application proposes an adaptive model predictive control method based on ACE and SOC partition to recover SOC. Different control strategies are implemented for different degrees of ACE, which can more finely adjust the frequency deviation and make the frequency regulation measures more effective. Therefore, according to the threshold size, the ACE is divided into different interval types, including the dead zone, the normal regulation zone, the emergency regulation zone, and the super emergency regulation zone. In addition, the battery energy storage is expensive, and long time at too high or too low SOC will accelerate the battery aging and reduce the charging and discharging cycle efficiency. Keeping the SOC in good condition can prolong the battery life and reduce the replacement frequency, which is an important measure to reduce the frequency regulation cost and improve the economy. In this paper, the SOC is divided into the normal charging and discharging zone, the charging priority zone, the discharging priority zone, and the prohibited charging and discharging zone. According to the state interval of ACE and SOC, the frequency regulation demand and SOC recovery demand in each stage are considered comprehensively, and through the development of frequency regulation control strategy considering SOC recovery, the dual objectives of system frequency regulation and energy storage SOC recovery can be realized.

[0139] Step S52: Based on the MPC controller proposed in step S41, a weighted term for adjusting the energy storage SOC recovery is added to the objective function, and the weight factor of this term is represented by q, the size of q represents the importance of SOC recovery. Therefore, the newly designed objective function includes two sub-goals of adjusting ACE a and adjusting the energy storage SOC recovery amount. However, there may be conflicts between different sub-goals, so it is necessary to analyze the importance of frequency regulation and SOC recovery in different stages, and seek the best balance point between the goals by adjusting the weight. Therefore, this paper proposes an MPC weight adaptive method considering energy storage SOC recovery based on ACE and SOC interval information. The control proportion of the energy storage SOC recovery term in the objective function is flexibly adjusted in different interval ranges to improve the system performance.

[0140] The control strategy is developed according to the ACE and SOC partition as follows:

[0141] (1) When ACE is in the dead zone range, the grid frequency deviation is very small at this time, and the frequency regulation demand can be basically ignored. Without considering the current SOC state, the target SOC is 0.5, the energy storage fully restores the SOC and does not participate in frequency regulation, and the weight coefficient q is set to the maximum value 0.5.

[0142] (2) When ACE is in the normal regulation zone, the primary task of the energy storage is to respond to the grid frequency regulation demand, supplemented by moderate SOC recovery.

[0143] (3) When ACE is in the emergency regulation zone, the recovery of the grid frequency is the highest priority, and the state of charge is not considered, the energy storage only participates in frequency regulation, and the weight coefficient q is set to 0.

[0144] (4) When ACE is in the super emergency regulation zone, it means that the grid frequency has deviated from the normal range, and emergency measures such as generator tripping or load shedding are adopted until the grid frequency is restored to stability.

[0145] The application also proposes a battery energy storage secondary frequency modulation control system based on a dynamic participation factor and an adaptive model predictive control, which comprises a processor, a memory and a computer program stored in the memory. When the processor executes the computer program, it specifically executes the steps of any of the above battery energy storage secondary frequency modulation control methods based on a dynamic participation factor and an adaptive model predictive control.

[0146] The following is a specific example of the application.

[0147] The application takes Figure 2 a double-area interconnected grid as shown in the figure as the research object, a battery energy storage system is configured in region 1, and simulation examples under two types of scenes of step and continuous load disturbance are designed in the Matlab / Simulink simulation environment. Four different control methods are used for comparative analysis. Method one is the frequency modulation of thermal power units alone, method two is the step extreme value allocation factor method, method three is the model predictive control strategy, and method four is the dynamic energy storage participation factor model predictive control method proposed in this paper. The rated power of the regional thermal power unit is 300MW, the rated power and capacity of the energy storage are 8MW / 2MWh. 100MW and 50Hz are used as the benchmark value for normalization. The system related parameters are shown in Table 3. The electric vehicle charging load time series must be decomposed by VMD algorithm after pretreatment, and the parameters in VMD algorithm are set as shown in Table 1:

[0148] Table 3 System model parameters Tab.3Model parameters of system

[0149]

[0150] 1s, the simulation time is 80s, and the frequency deviation of the system under different control methods is shown in Figure 8 Table 4.

[0151] Table 4. Performance indexes of frequency regulation under step disturbance

[0152] Tab.4 Performance indexes of frequency regulation under step disturbance

[0153]

[0154] From Figure 8 and Table 4, it can be seen that after the step load disturbance is applied, the system frequency rapidly decreases. The frequency deviation peak and the frequency drop rate of method 4 are the smallest among all methods, only-0.2097 Hz and 0.2526 Hz / s. Method 1 has no energy storage participating in frequency regulation, and the frequency deviation peak is the largest, reaching-0.3602 Hz, and the frequency drop rate is the fastest, 0.4502 Hz / s. When the energy storage participates in secondary frequency regulation, the maximum frequency deviation and frequency drop rate of methods 2, 3 and 4 are significantly improved compared with method 1, indicating that the fire storage combined frequency regulation has better performance than the single frequency regulation of the thermal power unit. Among them, the frequency regulation effect of method 4 is the best, and the maximum frequency deviation and frequency drop rate are reduced by 13.88% and 23.31% respectively compared with method 2; reduced by 8.07% and 32.44% respectively compared with method 3. From the steady-state performance index, method 4 is only second to method 1, and better than the other two methods.

[0155] A continuous load disturbance is applied to region 1 as shown in Figure 9 , and the simulation time is set to 1800s. The frequency change and energy storage SOC simulation results under different control strategies are shown in Figure 10 and Figure 11 , and the evaluation index is shown in Table 5.

[0156] Table 5. Performance indexes of frequency regulation under continuous disturbance

[0157] Tab.5 Performance indexes of frequency regulation under continuous disturbance

[0158]

[0159] From Figure 10From (a) and Table 5, it can be seen that in the continuous load disturbance scenario, the maximum frequency deviation, the root mean square value of the frequency deviation, and the integral of the absolute value of the frequency change of method 4 are the smallest. The maximum frequency deviation is reduced by 32.61%, 21.32%, and 0.48% compared with methods 1, 2, and 3, respectively, and the root mean square value of the frequency deviation is reduced by 48.84%, 40.46%, and 22.31%, respectively. This shows that the introduction of energy storage to participate in secondary frequency modulation and the design of a reasonable control strategy can effectively improve the frequency regulation effect and reduce the frequency fluctuation of the system. In addition, the introduced frequency change amplitude integral index more directly represents the frequency dynamic change situation. From (b) of the frequency change situation, it can be seen that with the increase of the frequency modulation time, the Δf Figure 10 int of the method in this paper is always smaller than that of the other three comparison methods.

[0160] Figure 11 The SOC change curve of the energy storage under different control methods is shown in (c). The SOC fluctuation amplitude of method 4 is the largest, and the reason is that under the same energy storage output power and SOC amplitude constraints, the frequency modulation period is more fully utilized, that is, the energy storage undertakes more secondary frequency modulation tasks, and the action amplitude of the thermal power unit is reduced. The proposed method improves the support capacity of the energy storage in the frequency modulation process, and achieves more significant frequency modulation effect than the other two methods, further reducing the frequency deviation.

[0161] The above is the preferred embodiment of the present application, and any changes made according to the technical solutions of the present application, as long as the resulting functions do not exceed the scope of the technical solutions of the present application, are within the scope of protection of the present application.​

Claims

1. A battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control, characterized in that, Specifically comprising the following steps: S1, taking the classical two-area interconnected power grid frequency response model as the research object, and adopting the constant frequency and tie-line power deviation control mode TBC, an equivalent model of fire storage combined frequency modulation is established; wherein the thermal power unit participates in the primary and secondary frequency modulation of the power grid, and the battery energy storage only participates in the secondary frequency modulation; S2, adopting a double-layer judgment mechanism as the basis for dividing the frequency modulation responsibility between the battery energy storage and the thermal power unit; S3, taking the frequency modulation responsibility obtained by the dynamic energy storage participation factor as the optimization target, using the adaptive model predictive control MPC to issue instruction signals to make the energy storage act in advance in the frequency modulation control, and weakening the impact of load disturbance on the frequency; S4, according to the area control error signal ACE and the state interval of the energy storage SOC, comprehensively considering the frequency modulation demand and the SOC recovery demand of each stage of the power grid, and through the development of a frequency modulation control strategy considering SOC recovery, the dual objectives of system frequency regulation and energy storage SOC recovery are realized.

2. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 1, characterized in that, The grid dispatching center calculates the ACE signal according to the monitored frequency deviation and tie-line power deviation as shown in formula (1): ACE = ΔP tie + b1Δf1 (1) where ΔP tie is the tie-line exchange power deviation, b i is the frequency deviation coefficient, Δf i is the system frequency deviation.

3. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 2, characterized in that, The equivalent model of each frequency modulation unit in the system is as follows: The equivalent transfer function model of the thermal power unit is as follows: wherein: T G , T RH , F HP , and T CH are the time constant of the thermal power unit governor, the reheater time constant, the reheater gain, and the turbine time constant, respectively; The equivalent transfer function model of the battery energy storage is as follows: In the formula: T B is the response time constant of the battery energy storage output variable.

4. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 3, characterized in that, The ratio of the available capacity of the battery energy storage to the rated capacity represents the state of charge SOC, and the calculation formula is as follows: In the formula: S OC (t) is the size of the battery energy storage SOC at time t; Δt is the sampling time interval; P b (t) is the charge and discharge power of the battery energy storage at time t, with the discharge direction of the energy storage being positive; E N is the rated capacity of the battery energy storage.

5. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 4, characterized in that, The double-layer judgment mechanism is specifically as follows: The first layer judgment mechanism: the energy storage participation factor a is divided into two parts with 0.5 as the dividing point, including the larger energy storage participation factor interval 0.5 When the sensitivity S is less than 0, the smaller the system frequency deviation is with the increase of the energy storage participation factor, so the energy storage participation factor is selected in the larger energy storage participation factor interval to facilitate the frequency deviation adjustment; if the sensitivity S is greater than 0, the smaller the system frequency deviation is with the decrease of the energy storage participation factor, so the energy storage participation factor is selected in the smaller energy storage participation factor interval to facilitate the frequency recovery; The second layer judgment mechanism: according to the actual load fluctuation and the load fluctuation rate at the current time, the short-term load fluctuation is predicted, and then the energy storage participation factor is dynamically adjusted in the corresponding interval to adaptively determine the timing and depth of the energy storage participation in frequency modulation.

6. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 5, characterized in that, The dynamic adjustment of the participation factor of the energy storage in different intervals is realized by means of the fuzzy controller; meanwhile, the load development trend at the next moment is preliminarily predicted based on the current moment load disturbance ΔP L and the load fluctuation rate γ, as shown in formula (5): ΔP L (t+1) = ΔP L (t) + γ(t)Δt (5) Design fuzzy control rules: If the load disturbance is positive; when ΔP L is large in absolute value and γ>0, a large energy storage participation factor is set in the interval; when ΔP L is small in absolute value and γ>0, if the absolute value of γ is small, a small energy storage participation factor is set in the interval, if the absolute value of γ is large, a large energy storage participation factor is set in the interval; when γ<0, a small energy storage participation factor is set in the interval; If the system is not affected by disturbance at the current time, but γ>0, set a small energy storage participation factor in the interval; When the load disturbance is negative; when ΔP L is large in absolute value and γ < 0, a large energy storage participation factor is set in the interval; when ΔP L is small in absolute value and γ < 0, if the absolute value of γ is large, a large energy storage participation factor is set in the interval, if the absolute value of γ is small, a small energy storage participation factor is set in the interval; when γ > 0, a small energy storage participation factor is set in the interval; If the system is not affected by disturbance at the current time, but γ<0, set a small energy storage participation factor in the interval.

7. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 6, characterized in that, The fuzzy controller with two inputs and one output is selected to adjust the energy storage participation factor; the two inputs are normalized ΔP L With γ, the fuzzy mapping of high-frequency irregular load state is realized by using Gaussian membership function to facilitate fuzzy reasoning. The output is the energy storage participation factor a, and the membership function is designed based on a discrete argument domain, and several representative dynamic energy storage participation factor values are selected to cover different load states; ΔP L The quantization factor calculation method of a and γ is shown in equation (6): where ΔP Lmax and γ max are the maximum load disturbance and the maximum load rate of change, respectively; the load disturbance ΔP L and the load rate of change γ are normalized to the domain [-1, 1], and a takes the discrete domain [0.1, 0.3, 0.5, 0.7, 0.9]; the fuzzy sets NB, NS, ZO, PS, and PB are set to describe the value range of ΔP L , γ, and a, where NB, NS, ZO, PS, and PB represent negative large, negative small, zero, positive small, and positive large of the load disturbance and the load rate of change, respectively.

8. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 7, characterized in that, The state space equation of the two-area interconnected power grid frequency dynamic response model containing energy storage is as shown in formula (7): In the formula, A, B, G, C represent the state matrix, control matrix, input matrix and output matrix of the system respectively; x, u, d, y are the state variable, control variable, input variable and output variable of the system respectively; the specific elements are as shown in formulas (8)-(15): x = [Δf1 ΔP tie ΔP b SOC] T (12) u = Δu b (13) d = [ΔP L Δf2 ΔP g1 ] T (14) y = [ACE a SOC] T (15) In the formula, D1 represents the load damping coefficient of region 1, M1 represents the system moment of inertia of region 1, T 12 represents the power synchronization coefficient of the inter-regional tie line, ΔP b represents the output size of the battery energy storage, Δu b represents the output control signal of the battery energy storage, Δf2 represents the frequency deviation of region 2, ΔP g1 represents the output size of the thermal power unit of region 1, ACE a represents the size of the frequency modulation responsibility of the energy storage; T s Discretizing equation (7) with a sampling period of T, we obtain the following discrete-time state-space model: In the formula: A, B, G respectively represent state matrix, control matrix, input matrix under system discrete time; T s is the system discrete sampling time interval. The ACE a = 0 as the core optimization objective of the MPC controller; at the k moment, construct the following quadratic performance index function: s.t. y min ≤(y(k+j|k)≤y max ,j=1,2,...p (18) u min ≤(u(k+i|k)≤u max ,i=1,2,...c (19) In the formula, Q and R are output weight diagonal matrix and control weight diagonal matrix respectively, p and c are prediction time domain and control time domain respectively, y(k+j|k) is the prediction of system output at k+j time in the future at k time, u(k+i-1|k) is the prediction of system control variable at k+i-1 time in the future at k time, y r is the system output reference value, and the reference value is set to 0; y min , y max are the upper and lower limits of the system output variable, that is, the system ACE a constraint and the SOC charging and discharging threshold; u min , u max are the upper and lower limits of the system control variable, that is, the energy storage output power constraint; by solving the finite time domain optimization problem of the objective function formula (17), a set of optimal control sequences u(k+i-1|k) is obtained, only the first item in the control sequence is applied to the battery energy storage, and immediate optimization adjustment is realized; At the next sampling time, the optimization problem is solved again and the first term of the optimal control sequence is applied to the optimization process of battery energy storage, ensuring the optimal performance of the system at different times through rolling optimization.

9. The battery energy storage secondary frequency modulation control method based on dynamic participation factor and adaptive model predictive control according to claim 1, characterized in that, The S4 is specifically as follows: ACE is divided into different interval types according to threshold values, including dead zone, normal adjustment zone, emergency adjustment zone, and super-emergency adjustment zone, where ACE1, ACE2, and ACE3 are the boundary values ​​between two adjacent ACE intervals; SOC is divided into normal charge / discharge zone, charging priority zone, discharging priority zone, and prohibited charge / discharge zone, where S... min S low S high S max It is the boundary value between two adjacent intervals of SOC; A weighted term for adjusting the SOC recovery of the energy storage is added to the objective function, and a weight factor of the weighted term is represented by q, and the size of q represents the importance of the SOC recovery; According to the ACE and SOC partition situation, the control strategy is as follows: (1) When the ACE is in the dead zone range, the frequency deviation of the power grid is extremely small, and the frequency regulation demand is basically ignored; without considering the current SOC state, the target is SOC=0.5, the energy storage fully restores the SOC and does not participate in frequency regulation, and the weight coefficient q is set to the maximum value 0.5; (2) When the ACE is in the normal regulation zone, the primary task of the energy storage is to respond to the frequency regulation demand of the power grid, supplemented by SOC recovery; (3) When the ACE is in the emergency regulation zone, the recovery of the power grid frequency is the highest priority, and the influence of the state of charge is not considered, the energy storage only participates in frequency regulation, and the weight coefficient q is set to 0; (4) When the ACE is in the super emergency regulation zone, it means that the frequency of the power grid has deviated from the normal range, and emergency measures such as generator tripping or load shedding are adopted until the frequency of the power grid is restored to stability.

10. A battery energy storage secondary frequency control system based on dynamic participation factors and adaptive model predictive control, characterized in that, The computer program is stored in the memory, and the processor executes the computer program to specifically execute the steps in the battery energy storage secondary frequency modulation control method based on a dynamic participation factor and an adaptive model predictive control according to any one of claims 1-9.

Images