Power system inertia reserve capacity planning method
By establishing a power system inertia reserve capacity planning method, and utilizing the sequential Monte Carlo method and risk preference model, the problem that traditional reserve planning cannot reflect the risk preferences of decision-makers is solved, and differentiated protection of power grid inertia level and frequency stability improvement are achieved.
Patent Information
- Application Number
- CN202411001153.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-12
AI Technical Summary
Existing research focuses on control strategies and optimization of virtual inertia, lacking research on ensuring the inertia level of the power grid from the perspective of reserve planning. Traditional reserve planning methods based on the full accident set cannot reflect the risk preferences of decision-makers, resulting in extreme events being submerged in a large number of frequent events, which cannot meet the system inertia requirements.
The sequential Monte Carlo method was used to establish component state models for conventional units, wind turbines, and photovoltaic units. Random production simulation was used to obtain the system load loss set. A system frequency response model with multiple types of regulation resources including source, storage, and load was established. The minimum demand inertia was evaluated based on RoCoF constraints. A penalty model was constructed to measure RoCoF over-limit losses. A total cost-benefit model of inertia reserve considering risk preference was also constructed to verify the feasibility of the planning scheme.
It can take into account the risk preferences of decision-makers, formulate differentiated backup plans, ensure the inertia level of the power grid, provide more comprehensive theoretical guidance, and improve the frequency stability and economy of the system.
Smart Images

Figure CN118944196A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of power systems, and in particular relates to a method for planning inertia reserve capacity of a power system. Background Art
[0002] As the first line of defense for frequency stability control, inertia can quickly provide power support after the system is disturbed, buying time for the subsequent speed regulator action and the response of other frequency regulation means. Therefore, inertia is crucial to maintaining system frequency stability.
[0003] With the accelerated construction of new power systems, the proportion of wind power, photovoltaic and energy storage inverter interface resources (Inverter Based Resources, IBR) continues to increase, and traditional synchronous units are replaced in large numbers, which makes the system synchronous inertia level continue to decrease, and the power grid presents a significant "low inertia" feature. Too low a grid inertia level will cause the rate of change of frequency (Rate of Change of Frequency, RoCoF) index to deteriorate after the system is disturbed, while too high a RoCoF will cause damage to the internal structure of the synchronous unit, the disconnection of distributed power sources, and even the subsequent frequency regulation means will not have time to respond to the system frequency and will collapse. Therefore, it is urgent to carry out theoretical method research to ensure the level of grid inertia.
[0004] At present, most of the existing research focuses on the control strategy and optimal configuration of virtual inertia, but lacks research on ensuring the inertia level of the power grid from the perspective of backup planning. With the continuous reduction of system synchronous inertia, the traditional backup method will not be able to meet the specific needs of the system for inertia. It is necessary to classify the backup more finely and add inertia backup to ensure the inertia level of the power grid. On the one hand, with the continuous reduction of the synchronous inertia of the power grid and the in-depth exploration of virtual inertia resources, the inertia response will no longer be a natural and free response dominated by the physical laws of traditional components, but will become an important part of participating in the auxiliary services of the power market. Therefore, planning for inertia backup requires the use of risk decision-making methods to weigh the economy and reliability of the scheme. On the other hand, the risk preference of power grid decision makers will affect the planning results. The traditional backup planning method based on the full accident set cannot reflect the risk preference of decision makers, and will also cause extreme events to be submerged in a large number of frequent ordinary events. Therefore, it is urgent to study a power system inertia backup capacity planning method to continuously characterize the changes in the risk focus range of decision makers and formulate differentiated planning schemes.
[0005] The present invention proposes a method for planning the inertia reserve capacity of a power system. The method establishes component state models of conventional units, wind turbines and photovoltaic units based on the sequential Monte Carlo method, and obtains the system load loss set through random production simulation; analyzes the inertia response characteristics, and establishes a system frequency response model containing multiple types of regulation resources such as source-storage-load, so as to achieve rapid tracking of the dynamic trajectory of the frequency after the system is disturbed; evaluates the minimum required inertia of the system based on the RoCoF constraint, and constructs a penalty model to measure the loss caused by the RoCoF limit; and then establishes an inertia reserve total cost-benefit model considering risk preference, and conducts a feasibility test on the planning scheme. This method can take into account the risk preference of decision makers, formulate differentiated reserve planning schemes according to different risk tolerance of power grids, and can provide more comprehensive theoretical guidance for ensuring the inertia level of power grids. Summary of the invention
[0006] The inertia level of power grids with a high proportion of new energy continues to decrease, seriously threatening the safe and stable operation of the power grid. However, existing research focuses on the control strategy and optimal configuration of virtual inertia, and lacks research on ensuring the inertia level of the power grid from the backup planning level; the traditional backup planning method based on the full accident set cannot reflect the risk preference of decision makers, and will also cause extreme events to be submerged in a large number of frequent ordinary events. In response to the above problems, the present invention proposes a method for planning the inertia reserve capacity of a power system. The method establishes component state models of conventional units, wind turbines and photovoltaic units based on the sequential Monte Carlo method, and obtains the system load loss set through random production simulation; analyzes the inertia response characteristics, establishes a system frequency response model containing multiple types of regulation resources such as source-storage-load, and realizes rapid tracking of the dynamic trajectory of the frequency after the system is disturbed; evaluates the minimum required inertia of the system based on the RoCoF constraint, and constructs a penalty model to measure the losses caused by RoCoF exceeding the limit; and then establishes a total cost-benefit model of inertia reserve considering risk preference, and conducts a feasibility test on the planning scheme. This method can take into account the risk preferences of decision makers and formulate differentiated backup planning schemes according to the risk tolerance of different power grids, which can provide more comprehensive theoretical guidance for ensuring the inertia level of the power grid.
[0007] In order to achieve the above object, the technical solution adopted by the present invention is:
[0008] A method for planning the inertia reserve capacity of a power system, such as Figure 1 As shown, the following steps are included:
[0009] S1: Based on the sequential Monte Carlo method, the component state models of conventional units, wind turbines and photovoltaic units are established, and the system load loss set is obtained by random production simulation;
[0010] S2: Analyze the inertia response characteristics and establish a system frequency response model containing multiple types of regulation resources such as source, storage and load;
[0011] S3: Evaluate the minimum required inertia of the system based on the RoCoF constraint, and build a penalty model to measure the loss caused by RoCoF exceeding the limit;
[0012] S4: Then establish the total cost-benefit model of inertia reserve considering risk preference, and conduct feasibility test on the planning scheme.
[0013] Optimization, S11: constructing a wind turbine component state and output model;
[0014] According to the three-state model of the wind turbine, the state transfer matrix can be established:
[0015]
[0016] In the formula, λ w_1 ,λ w_2 μ is the transfer rate of the fan from the running state to the shutdown state and the derated state; w_1 , μ w_2 is the corresponding repair rate;
[0017] Using the Markov method to solve the probability, the duration of each state can be obtained as:
[0018]
[0019] Where: τ 12 , τ 13 is the healthy operation time before the three-state model transitions to derating and shutdown, τ 21 , τ 31 is the derating and downtime time before the healthy operation is restored, and u is a random number uniformly distributed in [0,1];
[0020] There is a nonlinear relationship between wind turbine output and wind speed, and its power output model can be described by a piecewise function:
[0021]
[0022] In the formula, v t is the predicted wind speed; v ci 、v n 、v co are the cut-in wind speed, rated wind speed and cut-out wind speed of the fan respectively; P wn is the rated power of the wind turbine; a, b, c are the unknown coefficients of the wind turbine power characteristic curve, and the solution formula is:
[0023]
[0024] Wind speed can be predicted and obtained using the Weil distribution method, time series method and machine learning method;
[0025] S12: Constructing the photovoltaic unit component status and output model;
[0026] By using the Markov method to solve the probability, the probability and duration of each state of the component can be obtained:
[0027]
[0028] Where λ3 and μ3 are the normal to fault transfer rate and fault repair rate of the two-state model components respectively; P3 and P4 are the probabilities of the components being in operation and shutdown states respectively; τ3 and τ4 are the continuous normal operation time and repair time of the two-state model respectively;
[0029] The power output characteristics of photovoltaic units are described by piecewise functions:
[0030]
[0031] In the formula, G std , P pvn Respectively represent the rated light intensity and the rated power of the photovoltaic unit; R c is a specific light intensity; G bt It is a real-time light intensity sequence, which can be simulated and generated by Homer software using actual measurement data;
[0032] S13: Constructing the component status and output model of conventional units;
[0033] The total power output of conventional units can be obtained based on the capacity value of each conventional unit and the status of its components;
[0034] S14: Establish a system component status model, screen out continuous component failures, output the output of wind power, photovoltaic and conventional units, add up the system unit output, and the part that does not meet the load demand is the disturbance size, which is the power loss collection caused by the abnormal operation status of all components at the sampling point.
[0035] Further preferably, S2 specifically comprises the following steps:
[0036] S21: Analysis and modeling of inertia response and primary frequency regulation characteristics of conventional units;
[0037] The three types of conventional units participate in primary frequency regulation through the speed governor. For a synchronous generator, if the damping is ignored, the rotor motion equation is:
[0038]
[0039] In the formula, f g is the frequency of the generator set; ΔP e is the disturbance power. g is the inertia constant of the generator set, in s;
[0040] From an energy perspective, inertia can be expressed as:
[0041]
[0042] Where J is the moment of inertia; ω is the mechanical angular velocity, S N is the rated capacity of the generator; E k is the kinetic energy stored in the rotor, i.e. the quantified value of inertia in the form of energy, in MW·s;
[0043] S22: Analysis and modeling of virtual inertia response characteristics of converter interface resources;
[0044] The general model of frequency response of power electronic interface resources, whose transfer function can be expressed as:
[0045]
[0046] In the formula, K VI is the virtual inertia coefficient; R VI represents the virtual droop coefficient; T D It is the response delay time;
[0047] From the energy perspective, the virtual inertia can be expressed as:
[0048]
[0049] In the formula, h g is the inertia constant of virtual inertia, unit is s; S B E is the rated capacity of the equipment; k To provide the energy reserve of virtual inertia, that is, the quantified value of inertia in energy form, in MW·s;
[0050] S23: Analysis and modeling of load side resource inertia response and primary frequency regulation characteristics;
[0051] The frequency response transfer function of the asynchronous motor can be expressed as:
[0052]
[0053] In the formula, H am is the inertia time constant of the asynchronous motor. K1, K2 and K3 can be calculated by public derivation:
[0054]
[0055] In the formula, ω r0 is the initial speed of the asynchronous motor, ω0 is the initial value of the system speed, x s and r s is the equivalent leakage reactance and resistance of the stator winding, xr and r r is the equivalent leakage reactance and resistance of the rotor winding, s slip0 is the initial slip ratio, k is the load factor, α is the constant torque coefficient, and ρ is the load mechanical characteristic related index;
[0056] Based on the analysis and modeling of the inertia response and primary frequency regulation characteristics of various regulation resources, a system frequency response model containing multiple types of regulation resources is established, where ΔP d H is the power shortage caused by system disturbance; sys and D sys are the system inertia time constant and damping coefficient respectively; K G , K H , K T , K M and K I They are the capacity ratio coefficients of GTU, HPU, TPU, IM, and IBR respectively.
[0057] Further preferably, in S3, based on the established system frequency response model, the frequency dynamic constraint is analyzed, and the minimum required inertia is evaluated considering the RoCoF constraint;
[0058] In the inertia reserve planning process, the RoCoF constraint priority should be higher than the frequency minimum point constraint;
[0059] When the system inertia meets the RoCoF constraint but does not meet the frequency minimum constraint, the frequency minimum is intercepted by calling the fast frequency response resource means. If the grid inertia level is increased to curb the maximum frequency deviation, not only will the effect be minimal, but it is also far less economical than investing in primary frequency regulation means to intercept the frequency minimum.
[0060] According to formula (7), the system can be obtained under the expected disturbance ΔP d The minimum inertia required to satisfy the RoCoF constraint is:
[0061]
[0062] The system load loss set obtained from the random production simulation is input into equation (18) to obtain the minimum inertia requirement of the system under different RoCoF constraints and compare it with the actual inertia level to characterize the inertia over-limit risk.
[0063] Further optimization, in S3, when planning for inertia reserve, it is necessary to constrain RoCoF and measure the loss caused to the system after RoCoF exceeds the limit, so as to provide guidance for decision makers to plan inertia reserve capacity;
[0064] According to the relevant RoCoF constraints and combined with loss analysis, equation (19) is constructed to simulate the RoCoF over-limit penalty rule:
[0065]
[0066] In the formula, R1, R2, R m-1 and R m are different levels of RoCoF constraints, depending on the distributed grid relay protection setting value and the unit RoCoF tolerance value in the grid; C1, C2, C m-1 and C m It is the loss value corresponding to the RoCoF limit violation, which depends on the distributed generation grid disconnection, synchronous unit damage and low-frequency load reduction cost incremental factors under the corresponding RoCoF constraints.
[0067] Further preferably, S4 specifically comprises the following steps:
[0068] S41: Risk preference cost-benefit analysis;
[0069] The reliability cost L is a random variable and has a probability distribution related to the backup capacity r. Let the probability density distribution function be p L (x, r), through the reliability cost L is greater than or equal to a certain threshold C L The probability of quantifying the risk factor is:
[0070]
[0071] The risk factor γ is the proportion of accidents that the grid dispatchers are most concerned about in the total accident concentration. By changing γ, the decision makers' attention to different risk ranges can be changed, and the changes in their risk preferences can be continuously characterized. When the risk factor γ tends to 0, the decision makers are more inclined to pay attention to extreme accidents that threaten frequency safety. When γ tends to 1, they are more inclined to pay attention to global accidents. Therefore, the conditional reliability cost related to the reserve capacity r under the risk factor γ is:
[0072]
[0073] According to the frequency RoCoF over-limit degree, the corresponding over-limit loss is formulated in formula (1). Assuming that the system has N RoCoF over-limit accidents in the nth year, the total RoCoF over-limit loss in the nth year under the reserve capacity r can be obtained:
[0074]
[0075] The reliability benefit of inertia reserve considering risk preference can be expressed as:
[0076] E(γ,r)=L(γ,r)-L(γ,0) (24)
[0077] S42: Construction of standby total cost-benefit model;
[0078] The cost of adding inertia reserve mainly includes investment cost I(r) and call cost D(r). The call cost is related to the number of effective over-limit times of system RoCoF n and the inertia capacity demand after over-limit H. r And the unit inertia call cost d is related to:
[0079] F(r)=I(r)+D(r) (25)
[0080]
[0081] Where k is the discount rate, σ is the life cycle of the inertia reserve resource, and B represents the total investment cost;
[0082] By combining the reliability benefit of inertia backup and the cost of additional equipment obtained by equations (24) and (25), a total cost-benefit analysis model of backup considering risk preference can be established:
[0083] M(γ,r)=E(γ,r)-F(γ,r) (28)
[0084] Taking the maximization of the final benefit M(γ,r) as the objective function, considering the resource endowment to set the optimal interval of inertia reserve capacity, the random production simulation Y years to obtain the load loss set, by traversing the risk factor γ to change the decision maker's risk preference, the optimal inertia reserve capacity under different risk ranges can be obtained;
[0085] S43: Feasibility check of planning scheme;
[0086] In the inertia reserve planning process, the total cost A(r) includes the cost of adding inertia reserve and the reliability cost of RoCoF exceeding the limit, which also has a probability distribution related to the reserve capacity r. By analogy with formula (22), the total cost under the risk factor γ can be expressed as:
[0087]
[0088] Then the confidence index of the planning scheme can be defined as:
[0089]
[0090] Based on this index, it can be determined whether the plan should be adopted because of the high cost caused by adding inertia reserve r under the risk factor γ. When σ is less than 0, the planning scheme should be rejected; when σ is greater than or equal to 0, the planning scheme can be adopted, and the larger the confidence index, the more worthy the scheme is to be adopted.
[0091] The beneficial effects of the present invention are as follows: the present invention can take into account the risk preferences of decision makers, formulate differentiated backup planning schemes according to different grid risk tolerances, and can provide more comprehensive theoretical guidance for ensuring the inertia level of the grid at the backup planning level. BRIEF DESCRIPTION OF THE DRAWINGS
[0092] Figure 1 Flow chart of the power system inertia reserve capacity planning method considering the risk preference of decision makers;
[0093] Figure 2 Simulate flow chart for random production;
[0094] Figure 3 It is a schematic diagram of the three-state model of a wind turbine;
[0095] Figure 4 It is a schematic diagram of a two-state model;
[0096] Figure 5 It is a schematic diagram of the principle of continuous fault screening;
[0097] Figure 6 It is a schematic diagram of load loss;
[0098] Figure 7 This is the structural diagram of the primary frequency regulation response characteristics of a conventional unit;
[0099] Figure 8 It is a structural diagram of the frequency response model of the power electronic interface resource;
[0100] Fig. 9 It is the structure diagram of the asynchronous motor frequency response model;
[0101] Fig.10 It is the structure diagram of the system frequency response model;
[0102] Fig.11 This is the improved IEEE RTS-79 system structure diagram;
[0103] Fig.12 Optimization diagram for some reserve capacity under higher risk range;
[0104] Fig.13 Optimize the partial reserve capacity diagram for the lower risk range;
[0105] Fig.14 This is a feasibility verification diagram for the planning scheme. DETAILED DESCRIPTION
[0106] The present invention is further described below in conjunction with specific embodiments.
[0107] In order to make the technical solutions and advantages of the present invention clearer, the following is introduced: Fig.11The IEEE RTS-79 system shown in the figure is used as a test system to analyze the effect of inertia reserve investment, and the risk factor is changed to analyze the impact of decision makers' risk preference on the planning scheme; RoCoF differentiated over-limit losses are set to analyze the differences in planning schemes for different power grids and regions; finally, the feasibility of the planning scheme is verified and analyzed.
[0108] The installed capacity and proportion of each unit in the system are shown in Table 1. The virtual droop coefficient R v Take 0.05, and the control delay is 0.1s. Taking the synchronous condenser as an example, the inertia backup resource has an investment cost of 37727.3$ / MW·s. Considering that the synchronous unit provides inertia support free of charge, the call cost is set to 0. The backup life cycle is set to 30a, and the discount rate is 10%. The RoCoF over-limit loss setting is shown in Table 2, L b Take 18, and the power grids in different regions can be flexibly modified according to the actual situation. The present invention uses MATLAB / 2021b to perform relevant calculations. The technical solution of the present invention is clearly and completely described below in conjunction with specific embodiments and drawings.
[0109] Table 1 IEEE RTS-79 system unit parameters
[0110]
[0111] Table 2 RoCoF over-limit loss settings
[0112]
[0113] A method for planning the inertia reserve capacity of a power system, such as Figure 1 As shown, the following steps are included:
[0114] S1: Sequential Monte Carlo random production simulation to obtain the system load loss set;
[0115] To analyze the rationality and feasibility of the inertia reserve capacity planning results, it is necessary to simulate the state of system components. Since the output of wind turbines and photovoltaic units has very obvious uncertainty and timing fluctuations, the Markov chain Monte Carlo method is used for random production simulation. The process is as follows: Figure 2 As shown;
[0116] S11: Construct wind turbine component status and output model;
[0117] In actual operation, due to the influence of external environmental factors such as wind speed fluctuations, when the load is too large and the gearbox oil temperature reaches the critical value and an alarm is triggered, the wind turbine will enter a derated operation state. After the alarm is lifted, it will return to normal operation. Therefore, the wind turbine adopts Figure 3 The three-state model shown;
[0118] According to the three-state model of the wind turbine, the state transfer matrix can be established:
[0119]
[0120] In the formula, λ w_1 ,λ w_2 μ is the transfer rate of the fan from the running state to the shutdown state and the derated state; w_1 , μ w_2 is the corresponding repair rate;
[0121] Using the Markov method to solve the probability, the duration of each state can be obtained as:
[0122]
[0123] Where: τ 12 , τ 13 is the healthy operation time before the three-state model transitions to derating and shutdown; τ 21 , τ 31 is the derating and downtime time before the healthy operation is restored, and u is a random number uniformly distributed in [0,1];
[0124] There is a nonlinear relationship between wind turbine output and wind speed, and its power output model can be described by a piecewise function:
[0125]
[0126] In the formula, v t is the predicted wind speed; v ci 、v n 、v co are the cut-in wind speed, rated wind speed and cut-out wind speed of the fan respectively; P wn is the rated power of the wind turbine; a, b, c are the unknown coefficients of the wind turbine power characteristic curve, and the solution formula is:
[0127]
[0128] Wind speed can be predicted and obtained using the Weil distribution method, time series method and machine learning method;
[0129] S12: Constructing the photovoltaic unit component status and output model;
[0130] According to the operating characteristics of the photovoltaic unit, its shutdown model adopts the following Figure 4 Conventional two-state model shown;
[0131] By using the Markov method to solve the probability, the probability and duration of each state of the component can be obtained:
[0132]
[0133] Where λ3 and μ3 are the normal to fault transfer rate and fault repair rate of the two-state model components respectively; P3 and P4 are the probabilities of the components being in operation and shutdown states respectively; τ3 and τ4 are the continuous normal operation time and repair time of the two-state model respectively;
[0134] The power output of photovoltaic units is subject to external influences of light and has certain volatility and randomness. Before sunrise, the photovoltaic output is 0, and after sunrise, it gradually increases with the increase of photovoltaic illumination until it reaches the output limit, and then it continues to decrease with the decrease of light intensity until the output becomes 0 again at sunset. Therefore, its power output characteristics can also be described by a piecewise function:
[0135]
[0136] In the formula, G std , P pvn Respectively represent the rated light intensity and the rated power of the photovoltaic unit; R c is a specific light intensity; G bt It is a real-time light intensity sequence, which can be simulated and generated by Homer software using actual measurement data;
[0137] S13: Constructing the component status and output model of conventional units;
[0138] Conventional units mainly include thermal power units (TPU), hydropower units (HPU) and gas turbine units (GTU). According to the actual operation process of the power system, their operating conditions can be divided into two states: operation and shutdown. Therefore, conventional units also adopt the following Figure 4 The two-state model shown;
[0139] The probability of each state of the conventional unit and its duration are consistent with the calculation method of the photovoltaic unit. When the Monte Carlo random sampling result is less than P4, the system enters the shutdown state, otherwise it is in the normal operation state. The total power output of the conventional unit can be obtained based on the capacity value of each conventional unit and the state of its components;
[0140] S14: Screening continuous faults to obtain a system load loss set;
[0141] The use of inertia backup focuses on the frequency change within seconds after the system is disturbed, such as the difference in the maximum frequency change rate before and after the backup is put into use. The sampling time scale of random production simulation is usually hourly. When sampling component status, the impact on the system frequency is obviously different if the same component fails for one hour continuously and fails for one hour each in different time periods. For example, Figure 5As shown, the inertia reserve has been put into operation at the time t when the disturbance occurs. The same disturbance will not trigger the means of maintaining the system frequency stability multiple times. The continuous fault state of the unit can be screened to obtain the adaptive fault set, that is, the sampling points of t+1…t+l do not affect the system frequency stability and should be filtered;
[0142] Input the parameters of each component, the probability of failure, and the time of failure repair to establish the system component status model; after screening out the continuous failures of the components, output the output of wind power, photovoltaic and conventional units, such as Figure 6 As shown, the output of the system units is superimposed, and the part that does not meet the load demand is the disturbance size. Specifically, the disturbance size is the power loss collection caused by the abnormal operation status of all components at the sampling point;
[0143] S2: Construct a system frequency response model containing multiple types of regulation resources;
[0144] In order to calculate the difference in system frequency indicators before and after the addition of inertia reserve and evaluate the benefits of adding inertia reserve, it is necessary to establish a system frequency response model to accurately track the dynamic frequency change trajectory after the system is disturbed. First, the frequency response characteristics of conventional units, converter interface resources and load-side resources are analyzed and modeled, and then a system frequency response model containing multiple types of source-storage-load regulation resources is established.
[0145] S21: Analysis and modeling of inertia response and primary frequency regulation characteristics of conventional units;
[0146] The three types of conventional units participate in primary frequency regulation through speed regulators. Their primary frequency regulation response characteristics are as follows: Figure 7 As shown;
[0147] Conventional units have actual rotating rotors, so they can respond spontaneously and without delay when the system is disturbed, converting the rotor kinetic energy into mechanical energy to provide inertia power support. For a synchronous generator, if damping is ignored, its rotor motion equation is:
[0148]
[0149] In the formula, f g is the frequency of the generator set; ΔP e is the disturbance power. g It is the inertia constant of the generator set, in seconds. Its physical meaning is the time that the generator set can last when it uses only the kinetic energy stored in the rotor to provide energy for the load equal to the rated capacity of the generator at rated speed.
[0150] Therefore, from the energy perspective, inertia can also be expressed as:
[0151]
[0152] Where J is the moment of inertia; ω is the mechanical angular velocity, S N is the rated capacity of the generator; E k is the kinetic energy stored in the rotor, i.e. the quantified value of inertia in the form of energy, in MW·s;
[0153] S22: Analysis and modeling of virtual inertia response characteristics of converter interface resources;
[0154] With the development of virtual inertia technology, wind power, photovoltaic and energy storage inverter interface resources (Inverter Based Resources, IBR) can also provide primary frequency regulation and inertia support capabilities for the power system. According to the different grid connection methods and control strategies, IBR is divided into grid-following equipment and grid-building equipment, and the virtual inertia response characteristics and methods of the two are different;
[0155] For network-building equipment, the virtual synchronous machine technology can not only simulate the external characteristics of the synchronous machine, but also actively establish the frequency, so it can be regarded as a synchronous generator set;
[0156] At present, most IBRs are grid-following devices, which passively track the grid frequency signal, provide inertia power support in a form similar to synchronous inertia response, and participate in primary frequency regulation through virtual droop control. In actual operation, due to the delay in frequency measurement, command generation and control implementation, the grid-following IBR participates in frequency regulation with a delay of hundreds of milliseconds. Figure 8 A general frequency response model of power electronic interface resources is given, and its transfer function can be expressed as:
[0157]
[0158] In the formula, K VI is the virtual inertia coefficient; R VI represents the virtual droop coefficient; T D It is the response delay time;
[0159] Since virtual inertia control has a response delay and cannot respond spontaneously at the moment of system disturbance, it cannot reduce the maximum RoCoF like synchronous inertia. However, in actual power grids, since filtering is inevitably involved in frequency measurement, any system that relies on measuring frequency and RoCoF is likely to be unable to detect RoCoF up to the theoretical maximum value. Even if the initial RoCoF exceeds the threshold, the RoCoF relay with a measurement window of hundreds of milliseconds may not be triggered. Therefore, virtual inertia control with a delay of about 100 milliseconds can still play an effective inertia support role.
[0160] Although virtual inertia is realized through the control of power electronic devices, it also requires an actual energy source. For wind turbines, there are three main energy sources involved in virtual inertia response and primary frequency regulation. One is that the wind turbine operates without load reduction and uses the kinetic energy of the wind turbine blade rotor to provide energy; one is that the wind turbine operates at load reduction to provide inertia power support; and another is to use battery energy storage to provide energy for its virtual inertia and frequency regulation. For photovoltaic units, there is no rotational kinetic energy inside, so it is necessary to rely on battery energy storage to provide the energy required for virtual inertia response through load reduction operation. At present, it is generally through the coordinated operation of wind, solar and energy storage to provide more stable and controllable virtual inertia and frequency regulation capabilities. Therefore, from the energy perspective, virtual inertia can also be expressed as:
[0161]
[0162] In the formula, h g is the inertia constant of virtual inertia, unit is s; S B E is the rated capacity of the equipment; k To provide the energy reserve of virtual inertia, that is, the quantified value of inertia in energy form, in MW·s;
[0163] S23: Analysis and modeling of load side resource inertia response and primary frequency regulation characteristics;
[0164] The rotor of the asynchronous motor (Induction Machine, IM) is also electromechanically coupled with the power grid, and can spontaneously respond to system disturbances, release the kinetic energy stored in the rotor, and provide inertia support capabilities. As the proportion of load-side inertia continues to increase, the rotational inertia of the asynchronous motor cannot be ignored. Fig. 9 The structure diagram of the asynchronous motor frequency response model is given, and the frequency response transfer function of the asynchronous motor can be expressed as:
[0165]
[0166] In the formula, H am is the inertia time constant of the asynchronous motor. K1, K2 and K3 can be calculated by public derivation:
[0167]
[0168] In the formula, ω r0 is the initial speed of the asynchronous motor, ω0 is the initial value of the system speed, x s and r s is the equivalent leakage reactance and resistance of the stator winding, x r and r r is the equivalent leakage reactance and resistance of the rotor winding, s slip0 is the initial slip ratio, k is the load factor, α is the constant torque coefficient, and ρ is the load mechanical characteristic related index;
[0169] For Emergency Interruptible Load (EIL), it is connected to the system through an internal feeder branch with a low-frequency relay to receive control. When the frequency drops, a certain amount of load is actively cut off to make up for the system power shortage to maintain frequency stability. The response characteristics of active load shedding simulation inertia mainly depend on the ancillary service agreement signed between the grid operator and the EIL provider and the formulated load shedding control strategy. Generally, the frequency deviation signal can be used to trigger the direct setting of delayed load shedding control;
[0170] Based on the analysis and modeling of the inertia response and primary frequency regulation characteristics of various regulation resources, the following are established: Fig.10 The system frequency response model with multiple types of regulation resources is shown in the figure, where ΔP d H is the power shortage caused by system disturbance; sys and D sys are the system inertia time constant and damping coefficient respectively; K G , K H , K T , K M and K I They are the capacity ratio coefficients of GTU, HPU, TPU, IM, and IBR respectively;
[0171] S3: Minimum required inertia evaluation considering RoCoF constraints;
[0172] Based on the established system frequency response model, the frequency dynamic constraints are analyzed and the minimum required inertia is evaluated considering the RoCoF constraints;
[0173] The existing minimum required inertia assessment is mainly based on the frequency minimum point constraint and the RoCoF constraint to obtain the minimum inertia level required by the system to ensure the stability of the grid frequency under the expected fault. However, different from the theoretical assessment, considering the specific needs and response characteristics of inertia reserve, the minimum required inertia assessment for inertia reserve planning should focus on the RoCoF constraint. In other words, in the inertia reserve planning process, the RoCoF constraint should have a higher priority than the frequency minimum point constraint;
[0174] On a time scale, inertia acts at the initial stage of frequency response at the moment of disturbance and thereafter. Insufficient inertia will directly lead to frequency change rate exceeding the limit. Subsequent measures cannot reverse the fact that the maximum frequency change rate exceeds the limit. For the frequency minimum point constraint, if the inertia and primary frequency regulation capacity are insufficient, it can be contained by low-frequency load reduction measures. In terms of economic indicators, the investment and call cost of inertia resources are generally much higher than that of primary frequency regulation resources due to their extremely high response speed requirements. When the system inertia has met the RoCoF constraint but not the frequency minimum point constraint, the frequency minimum point can be intercepted by calling fast frequency response resources. If the maximum frequency deviation is contained by increasing the grid inertia level, not only will the effect be minimal, but it is also far less economical than investing in primary frequency regulation to intercept the frequency minimum point.
[0175] According to formula (7), the system can be obtained under the expected disturbance ΔP d The minimum inertia required to satisfy the RoCoF constraint is:
[0176]
[0177] The larger the disturbance, the stricter the RoCoF constraint, and the larger the minimum inertia required by the system. The system load loss set obtained by the random production simulation is input into formula (18), and the minimum inertia requirement of the system under different RoCoF constraints is obtained and compared with the actual inertia level, which can characterize the inertia over-limit risk. Different RoCoF constraint levels can reflect the grid decision makers' concern about the risk range and the degree of inertia demand.
[0178] S4: Construct RoCoF over-limit penalty model;
[0179] The system inertia level is directly linked to the RoCoF indicator. As the inertia level continues to decrease, the RoCoF will continue to increase after the grid is disturbed. Once the RoCoF exceeds the limit, it may cause damage to the synchronous units, disconnection of distributed power sources from the grid, and further deterioration of frequency drop. The losses caused cannot be accurately measured.
[0180] When planning for inertia reserve, it is necessary to constrain RoCoF and measure the losses caused to the system when RoCoF exceeds the limit, so as to provide guidance for decision makers to plan inertia reserve capacity. When the losses caused by RoCoF exceeding the limit are small, but the cost of equipping inertia resources is too high, grid decision makers may tend to set less inertia reserve; when the frequency change rate exceeds the limit due to insufficient inertia reserve capacity, causing system losses far exceeding the cost of adding equipment, grid decision makers will tend to set more inertia reserve capacity. Therefore, measuring the losses caused by RoCoF exceeding the limit to the system is crucial for the formulation of decision-making plans;
[0181] After analyzing the consequences of RoCoF exceeding the limit, it can be known that the losses caused are roughly divided into three categories. First, if the RoCoF is too high, it may cause pole slippage in the synchronous unit, causing damage to the internal structure; the synchronous unit needs to be repaired and replaced; secondly, too high RoCoF may trigger the setting value of the distributed power supply relay protection, causing the distributed power supply to be disconnected from the grid, further increasing the power shortage of the system, making the frequency drop worse, and the disconnection of the distributed power supply will cause economic losses. The deterioration of the frequency index will also increase the loss of low-frequency load shedding; in addition, if the RoCoF is too high, it may also cause the speed regulator and other frequency regulation means to collapse before the frequency can be activated. At this time, the loss caused to the system will be difficult to measure. This situation is a large loss-low probability event, which generally occurs in extreme accident scenarios;
[0182] Based on the above analysis, the losses caused by RoCoF over-limit mainly include synchronous unit damage, distributed power supply disconnection losses, low-frequency load reduction costs caused by frequency index deterioration, and system frequency collapse losses in extreme cases. At the equipment level, different units have different RoCoF tolerance capabilities, and different distributed power supply protection setting values are also different. For example, wind turbines can continue to operate stably when RoCoF is as high as 4Hz / s, while synchronous generators are recommended not to exceed 2Hz / s; at the system level, different regions and different power grids have different numbers and capacities of distributed power sources, and different low-frequency load reduction rounds and losses after frequency index deterioration; in addition, if the spatiotemporal distribution characteristics of frequency are taken into account, the frequency dynamics in different regions and power grids are different, and the RoCoF over-limit situations are also different. Therefore, to measure the losses caused by RoCoF over-limit, it is necessary to adapt to the network and the local conditions;
[0183] At present, there are few studies on the RoCoF over-limit losses. According to the relevant RoCoF constraints at home and abroad, and combined with the above loss analysis, we can construct formula (19) to simulate the RoCoF over-limit penalty rule:
[0184]
[0185] In the formula, R1, R2, R m-1 and R m are different levels of RoCoF constraints, depending on the distributed grid relay protection setting value and the unit RoCoF tolerance value in the grid; C1, C2, C m-1 and C m It is the loss value corresponding to the RoCoF limit violation, which depends on the disconnection of distributed generation, the damage of synchronous units and the incremental factor of low-frequency load reduction cost under the corresponding RoCoF constraint;
[0186] The higher the RoCoF, the more serious the loss caused by the limit. Therefore, C1, C2, C m-1 and C mIt shows an increasing law. From the perspective of reserve planning, the severity of RoCoF over-limit loss will affect the formulation of decision-makers' planning schemes. The more serious the RoCoF over-limit loss, the stronger the decision-makers' determination to plan the inertia reserve capacity will be, and the final planned capacity will be more sufficient. Therefore, for different regions and different power grids, the inertia reserve planning results will also be different, which needs to vary from grid to grid and local conditions.
[0187] S4: Obtaining the inertia reserve capacity planning scheme;
[0188] S41: Risk preference cost-benefit analysis;
[0189] In the inertia reserve planning problem, the more extreme the event of concern is, the greater the system load loss, the more serious the RoCoF limit, the greater the loss caused, and the higher the corresponding risk value. Under the inertia reserve capacity r, different accident levels cause different losses. This type of cost is called reliability cost. The reliability cost L is a random variable, and there is a probability distribution related to the reserve capacity r. Let the probability density distribution function be p L (x, r), in order to consider the risk preference of the decision maker, the risk factor γ is defined to reflect the risk range of the decision maker, that is, the severity and rarity of the loss that the decision maker cares about in the global accident. The reliability cost L is greater than or equal to a certain threshold C L The probability of quantifying the risk factor is:
[0190]
[0191] The risk factor γ is the proportion of accidents that the grid dispatchers are most concerned about in the total accident concentration. By changing γ, the decision makers' attention to different risk ranges can be changed, and the changes in their risk preferences can be continuously characterized. When the risk factor γ tends to 0, the decision makers are more inclined to pay attention to extreme accidents that threaten frequency safety. When γ tends to 1, they are more inclined to pay attention to global accidents. Therefore, the conditional reliability cost related to the reserve capacity r under the risk factor γ is:
[0192]
[0193] According to the frequency RoCoF over-limit degree, the corresponding over-limit loss is formulated in formula (1). Assuming that the system has N RoCoF over-limit accidents in the nth year, the total RoCoF over-limit loss in the nth year under the reserve capacity r can be obtained:
[0194]
[0195] In actual calculations, based on the system load loss set of random production simulation, the RoCoF over-limit loss distribution under the global risk range can be calculated, and then the reliability cost can be sorted. Different risk ranges can be selected according to their discrete probability distribution, so as to obtain the reliability cost after adding inertia reserve r under different risk factors γ;
[0196] By adding inertia reserve, RoCoF can be effectively mitigated, thereby reducing RoCoF over-limit losses and improving system reliability. The reduction in RoCoF over-limit losses before and after adding inertia reserve is the reliability benefit. Therefore, the reliability benefit of inertia reserve considering risk preference can be expressed as:
[0197] E(γ,r)=L(γ,r)-L(γ,0) (24)
[0198] S42: Construction of standby total cost-benefit model;
[0199] The cost of adding inertia reserve mainly includes investment cost I(r) and call cost D(r). Transforming retired thermal power units into synchronous phase condensers, building flywheel energy storage and installing power electronic devices to provide virtual inertia can all provide support for inertia reserve. In reserve planning, the equal annual value method is often used to calculate the investment cost of the sampling year, and the call cost is related to the number of effective over-limit times of the system RoCoF n and the inertia capacity demand after over-limit H. r And it is related to the unit inertia call cost d;
[0200] F(r)=I(r)+D(r) (25)
[0201]
[0202] Where k is the discount rate, σ is the life cycle of the inertia reserve resource, and B represents the total investment cost;
[0203] By combining the reliability benefit of inertia backup and the cost of additional equipment obtained by equations (24) and (25), a total cost-benefit analysis model of backup considering risk preference can be established:
[0204] M(γ,r)=E(γ,r)-F(γ,r) (28)
[0205] Taking the maximization of the final benefit M(γ,r) as the objective function, considering the resource endowment to set the optimal interval of inertia reserve capacity, the random production simulation Y years to obtain the load loss set, by traversing the risk factor γ to change the decision maker's risk preference, the optimal inertia reserve capacity under different risk ranges can be obtained;
[0206] S43: Feasibility check of planning scheme;
[0207] The decision-maker's risk is based on his subjective judgment of the severity of the accident. It is necessary to verify the feasibility of the planning results and provide confidence indicators, which helps decision-makers avoid the situation where the backup plan does not meet actual needs due to subjective experience.
[0208] In the inertia reserve planning process, the total cost A(r) includes the cost of adding inertia reserve and the reliability cost of RoCoF exceeding the limit, which also has a probability distribution related to the reserve capacity r. By analogy with formula (22), the total cost under the risk factor γ can be expressed as:
[0209]
[0210] Then the confidence index of the planning scheme can be defined as:
[0211]
[0212] Based on this index, it can be determined whether the plan should be adopted because of the high cost caused by adding inertia reserve r under the risk factor γ. When σ is less than 0, the planning scheme should be rejected; when σ is greater than or equal to 0, the planning scheme can be adopted, and the larger the confidence index, the more worthy the scheme is to be adopted.
[0213] By traversing the risk factors to obtain the optimal inertia reserve capacity under different risk ranges, the impact of decision makers' risk preferences on the backup planning scheme can be effectively characterized. Table 3 gives the optimal inertia reserve capacity and related cost-benefit amounts under different risk factors. Fig.12 and Fig.13 The partial reserve optimization results under high risk factors and low risk factors are given respectively.
[0214] Table 3 Optimal capacity and cost-benefit under some risk factors
[0215]
[0216]
[0217] From Table 3, Fig.12 as well as Fig.13 It is not difficult to find that the risk preference of decision makers has a key impact on the planning results of inertia reserve capacity. As the risk factors decrease, the risk range that decision makers pay attention to shifts from average to extreme, and the optimal inertia reserve capacity also increases from 1125.75MW·s to 30505.79MW·s. This shows that as the accident losses that decision makers are concerned about tend to be extreme, the necessity and determination to add inertia reserves are also increasing. On the other hand, as the risk factors decrease, the reliability benefits and comprehensive benefits brought by adding inertia reserves are significantly improved. The more serious the accident losses, the more considerable the benefits brought by adding inertia reserves, so it is more necessary to carry out inertia reserves.
[0218] Table 4 Optimal reserve capacity under different RoCoF over-limit losses
[0219]
[0220] Table 5 Comprehensive reserve income under different RoCoF over-limit losses
[0221]
[0222]
[0223] Tables 4 and 5 respectively give the optimal inertia reserve capacity and its comprehensive benefits corresponding to each risk factor under different RoCoF over-limit losses. It is not difficult to see that with the increase of RoCoF over-limit losses, the optimal inertia reserve capacity under different risk ranges is increasing, and the comprehensive benefits generated by the additional inertia reserve are also increasing. In other words, the weaker the ability of the power grid to withstand RoCoF over-limit accidents, the higher the demand for additional inertia reserve, and the more considerable the benefits that can be generated by adding inertia reserve. On the contrary, if the power grid itself has sufficient inertia level and each component has a strong tolerance to RoCoF over-limit, the necessity of adding inertia reserve will also be reduced accordingly. Therefore, for different power grids in different regions, it is necessary to formulate differentiated inertia reserve planning schemes based on actual conditions.
[0224] The risk preference of decision makers is influenced by subjective factors. It is necessary to verify the inertia backup plan under different risk factors and give corresponding confidence indicators to judge the feasibility of the plan. The feasibility verification is carried out by taking the planning plan with the optimal backup capacity of 9572.91MW·s when the risk factor is 0.3 as an example. Fig.14 The confidence index of the planning scheme under different risk factors and the verification cost before and after the backup are given. When the risk factor is 0.6, the confidence index of the scheme is 0, and the scheme is at the boundary between feasible and infeasible. When the risk factor is less than 0.6, the backup capacity is acceptable under different risk ranges, and the smaller the risk factor, the higher the confidence index, and the higher the feasibility of the planning scheme; on the contrary, when the risk factor is greater than 0.6, the planning scheme is in an infeasible range. At this time, the total cost caused by the use of additional equipment is too high, and the planning scheme should be rejected.
[0225] From the above analysis, it can be seen that the inertia reserve planning method of the present invention can take into account the risk preferences of decision makers to formulate planning schemes under different risk ranges. The more serious the accident losses that the decision makers are concerned about, the larger the required additional inertia reserve capacity, and the more considerable the benefits that can be generated by the additional equipment. Different power grids have different risk tolerances, and the losses caused by RoCoF exceeding the limit are also different. The more serious the consequences of RoCoF exceeding the limit, the stronger the decision makers' determination to increase inertia reserve, and the higher the capacity and benefits of increasing inertia reserve. For different power grids, it is necessary to formulate differentiated inertia reserve planning schemes according to the actual situation according to the different grids.
[0226] Therefore, the power system inertia reserve capacity planning method proposed in the present invention can take into account the risk preferences of decision makers and formulate differentiated reserve planning schemes for different grid risk tolerances, which can provide more comprehensive theoretical guidance for ensuring the grid inertia level at the reserve planning level.
[0227] The above-described embodiments merely express the implementation methods of the present invention, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.
Claims
1. A method for planning inertia reserve capacity of power system, It is characterized in that The steps include: S1: Based on the sequential Monte Carlo method, the component state models of conventional units, wind turbines and photovoltaic units are established, and the system load loss set is obtained by random production simulation; S2: Analyze the inertia response characteristics and establish a system frequency response model containing multiple types of regulation resources such as source, storage and load; S3: Evaluate the minimum required inertia of the system based on the RoCoF constraint, and build a penalty model to measure the loss caused by RoCoF exceeding the limit; S4: Then establish the total cost-benefit model of inertia reserve considering risk preference, and conduct feasibility test on the planning scheme.
2. A method for planning inertia reserve capacity of a power system according to claim 1, It is characterized in that S1 specifically includes the following steps: S11: Construct wind turbine component status and output model; According to the three-state model of the wind turbine, the state transfer matrix can be established: In the formula, λ w_1 ,λ w_2 μ is the transfer rate of the fan from the running state to the shutdown state and the derated state; w_1 , μ w_2 is the corresponding repair rate; Using the Markov method to solve the probability, the duration of each state can be obtained as: Where: τ 12 , τ 13 is the healthy operation time before the three-state model transitions to derating and shutdown, τ 21 , τ 31 is the derating and downtime time before the healthy operation is restored, and u is a random number uniformly distributed in [0,1]; There is a nonlinear relationship between wind turbine output and wind speed, and its power output model can be described by a piecewise function: In the formula, v t is the predicted wind speed; v ci 、v n 、v co are the cut-in wind speed, rated wind speed and cut-out wind speed of the fan respectively; P wn is the rated power of the wind turbine; a, b, c are the unknown coefficients of the wind turbine power characteristic curve, and the solution formula is: Wind speed can be predicted and obtained using the Weil distribution method, time series method and machine learning method; S12: Constructing the photovoltaic unit component status and output model; By using the Markov method to solve the probability, the probability and duration of each state of the component can be obtained: In the formula, λ 3 , μ 3 are the normal to fault transition rate and fault repair rate of the two-state model components respectively; P 3 , P 4 are the probabilities of the component being in operation and outage respectively; τ 3 , τ 4 They are the continuous normal operation time and repair time of the two-state model respectively; The power output characteristics of photovoltaic units are described by piecewise functions: In the formula, G std , P pvn Respectively represent the rated light intensity and the rated power of the photovoltaic unit; R c is a specific light intensity; G bt It is a real-time light intensity sequence, which can be simulated and generated by Homer software using actual measurement data; S13: Constructing the component status and output model of conventional units; The total power output of conventional units can be obtained based on the capacity value of each conventional unit and the status of its components; S14: Establish a system component status model, screen out continuous component failures, output the output of wind power, photovoltaic and conventional units, add up the system unit output, and the part that does not meet the load demand is the disturbance size, which is the power loss collection caused by the abnormal operation status of all components at the sampling point.
3. A method for planning inertia reserve capacity of a power system according to claim 1, It is characterized in that S2 specifically includes the following steps: S21: Analysis and modeling of inertia response and primary frequency regulation characteristics of conventional units; The three types of conventional units participate in primary frequency regulation through the speed governor. For a synchronous generator, if the damping is ignored, the rotor motion equation is: In the formula, f g is the frequency of the generator set; ΔP e is the disturbance power. g is the inertia constant of the generator set, in s; From an energy perspective, inertia can be expressed as: Where J is the moment of inertia; ω is the mechanical angular velocity, S N is the rated capacity of the generator; E k is the kinetic energy stored in the rotor, i.e. the quantified value of inertia in the form of energy, in MW·s; S22: Analysis and modeling of virtual inertia response characteristics of converter interface resources; The general model of frequency response of power electronic interface resources, whose transfer function can be expressed as: In the formula, K VI is the virtual inertia coefficient; R VI represents the virtual droop coefficient; T D It is the response delay time; From the energy perspective, the virtual inertia can be expressed as: In the formula, h g is the inertia constant of virtual inertia, unit is s; S B E is the rated capacity of the equipment; k To provide the energy reserve of virtual inertia, that is, the quantified value of inertia in energy form, in MW·s; S23: Analysis and modeling of load side resource inertia response and primary frequency regulation characteristics; The frequency response transfer function of the asynchronous motor can be expressed as: In the formula, H am is the inertia time constant of the asynchronous motor, K 1 , K 2 and K 3 It can be calculated by public deduction: K ms =kω r0 p(1-a)(1-s) slip0 ) ρ-1 (16) In the formula, ω r0 is the initial speed of the asynchronous motor, ω 0 is the initial value of system speed, x s and r s is the equivalent leakage reactance and resistance of the stator winding, x r and r r is the equivalent leakage reactance and resistance of the rotor winding, s slip0 is the initial slip ratio, k is the load factor, α is the constant torque coefficient, and ρ is the load mechanical characteristic related index; Based on the analysis and modeling of the inertia response and primary frequency regulation characteristics of various regulation resources, a system frequency response model containing multiple types of regulation resources is established, where ΔP d H is the power shortage caused by system disturbance; sys and D sys are the system inertia time constant and damping coefficient respectively; K G , K H , K T , K M and K I They are the capacity ratio coefficients of GTU, HPU, TPU, IM, and IBR respectively.
4. A method for planning inertia reserve capacity of a power system according to claim 1, Features: In S3, based on the constructed system frequency response model, the frequency dynamic constraints are analyzed and the minimum required inertia is evaluated considering the RoCoF constraint; In the inertia reserve planning process, the RoCoF constraint priority should be higher than the frequency minimum point constraint; When the system inertia meets the RoCoF constraint but does not meet the frequency minimum constraint, the frequency minimum is intercepted by calling the fast frequency response resource means. If the grid inertia level is increased to curb the maximum frequency deviation, not only will the effect be minimal, but it is also far less economical than investing in primary frequency regulation means to intercept the frequency minimum. According to formula (7), the system can be obtained under the expected disturbance ΔP d The minimum inertia required to satisfy the RoCoF constraint is: The system load loss set obtained from the random production simulation is input into equation (18) to obtain the minimum inertia requirement of the system under different RoCoF constraints and compare it with the actual inertia level to characterize the inertia over-limit risk.
5. A method for planning inertia reserve capacity of a power system according to claim 1, Features: In S3, when planning for inertia reserve, it is necessary to constrain RoCoF and measure the loss caused to the system when RoCoF exceeds the limit, so as to provide guidance for decision makers to plan inertia reserve capacity; According to the relevant RoCoF constraints and combined with loss analysis, equation (19) is constructed to simulate the RoCoF over-limit penalty rule: In the formula, R 1 , R 2 , R m-1 and R m C is the RoCoF constraint of different levels, which depends on the setting value of the distributed grid relay protection and the RoCoF tolerance value of the unit in the grid; 1 , C 2 , C m-1 and C m It is the loss value corresponding to the RoCoF limit violation, which depends on the distributed generation grid disconnection, synchronous unit damage and low-frequency load reduction cost incremental factors under the corresponding RoCoF constraints.
6. A method for planning inertia reserve capacity of a power system according to claim 1, Features: S4 specifically includes the following steps: S41: Risk preference cost-benefit analysis; The reliability cost L is a random variable and has a probability distribution related to the backup capacity r. Let the probability density distribution function be p L (x, r), through the reliability cost L is greater than or equal to a certain threshold C L The probability of quantifying the risk factor is: The risk factor γ is the proportion of accidents that the grid dispatchers are most concerned about in the total accident concentration. By changing γ, the decision makers' attention to different risk ranges can be changed, and the changes in their risk preferences can be continuously characterized. When the risk factor γ tends to 0, the decision makers are more inclined to pay attention to extreme accidents that threaten frequency safety. When γ tends to 1, they are more inclined to pay attention to global accidents. Therefore, the conditional reliability cost related to the reserve capacity r under the risk factor γ is: According to the frequency RoCoF over-limit degree, the corresponding over-limit loss is formulated in formula (1). Assuming that the system has N RoCoF over-limit accidents in the nth year, the total RoCoF over-limit loss in the nth year under the reserve capacity r can be obtained: The reliability benefit of inertia reserve considering risk preference can be expressed as: E(γ,r)=L(γ,r)-L(γ,0) (24) S42: Construction of standby total cost-benefit model; The cost of adding inertia reserve mainly includes investment cost I(r) and call cost D(r). The call cost is related to the number of effective over-limit times of system RoCoF n and the inertia capacity demand after over-limit H. r And the unit inertia call cost d is related to: F(r)=I(r)+D(r) (25) Where k is the discount rate, σ is the life cycle of the inertia reserve resource, and B represents the total investment cost; By combining the reliability benefit of inertia backup and the cost of additional equipment obtained by equations (24) and (25), a total cost-benefit analysis model of backup considering risk preference can be established: M(γ,r)=E(γ,r)-F(γ,r) (28) Taking the maximization of the final benefit M(γ,r) as the objective function, considering the resource endowment to set the optimal interval of inertia reserve capacity, the random production simulation Y years to obtain the load loss set, by traversing the risk factor γ to change the decision maker's risk preference, the optimal inertia reserve capacity under different risk ranges can be obtained; S43: Feasibility check of planning scheme; In the inertia reserve planning process, the total cost A(r) includes the cost of adding inertia reserve and the reliability cost of RoCoF exceeding the limit, which also has a probability distribution related to the reserve capacity r. By analogy with formula (22), the total cost under the risk factor γ can be expressed as: Then the confidence index of the planning scheme can be defined as: Based on this index, it can be determined whether the plan should be adopted because of the high cost caused by adding inertia reserve r under the risk factor γ. When σ is less than 0, the planning scheme should be rejected; when σ is greater than or equal to 0, the planning scheme can be adopted, and the larger the confidence index, the more worthy the scheme is to be adopted.
Citation Information
Cited By
Power system inertia distribution optimization method and device based on node inertia and medium
CN119627982A
Receiving end power grid emergency frequency control method based on optimal frequency track
CN119965901A
Photovoltaic stochastic programming method and system based on bearing limit
CN122553139A
A photovoltaic random planning method and system based on bearing limit
CN122553139B
