A provincial power grid frequency control method based on a double-layer cooperative mechanism

Abstract

The application discloses a kind of provincial power grid frequency control methods based on double-layer cooperative mechanism, comprising:1, the system frequency deviation and regional control deviation of real-time calculation provincial power grid;2, construct double-layer cooperative mechanism based on working mode switching-frequency difference allocation;3, establish optimization control model based on frequency modulation index-superposition model predictive control, and solve using Gurobi solver, obtain frequency control scheme.The application can effectively reduce the fluctuation amplitude of system frequency and system power generation cost, and can improve the frequency modulation performance of system, to ensure the safety and reliability of power grid during secondary frequency modulation.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of research on collaborative control optimization methods for large-scale distributed resources and frequency modulation units, and more particularly to a provincial power grid frequency control method based on a double-layer collaborative mechanism. BACKGROUND

[0002] With the over-consumption of fossil energy leading to increasingly serious environmental pollution problems, China is actively promoting the transformation of the national energy structure and the greenization of electricity. However, with the continuous increase in the scale of new energy installations and the proportion of power generation, the frequency regulation capability of traditional generating units is gradually weakening, and it is urgent to explore and apply various frequency modulation resources to enhance the frequency regulation capability of the power system.

[0003] It is of great significance to build a frequency modulation model for the collaborative control of multiple types of resources such as thermal power units, energy storage, wind farms, photovoltaic power stations, and load aggregators, to achieve a stable and safe power system. In the resource calling process, heterogeneous frequency modulation resources differ in capacity and response speed. In order to maximize the role of heterogeneous resources with excellent frequency modulation performance in frequency regulation, the current common methods include using an improved reinforcement learning algorithm to update the participation factor of different types of units in real time, classifying AGC units according to response speed, or evaluating the reliability of AGC units based on frequency modulation accuracy probability indicators. Although the above methods screen frequency modulation resources according to relevant indicators, the screening results are not dynamically updated, and the impact of the screening process on frequency regulation is not effectively fed back. Therefore, it is necessary to reasonably select resources with excellent frequency modulation performance to perform frequency modulation tasks to respond to the dynamic changes in system operation state, and feedback the regulation results to the performance indicators. This strategy helps to alleviate the frequency modulation pressure of the system under emergency conditions such as external power in large power areas, while limiting the opportunity for resources with poor frequency modulation performance to participate in frequency modulation under normal conditions, thereby improving the adaptability and control performance of the system under various conditions.

[0004] The rapid development of new energy technology has promoted the clean transformation of electricity production. However, in a power system with a high proportion of new energy penetration, the randomness and uncertainty of new energy output pose higher requirements for the frequency stability control of the system. To cope with the uncertainty of new energy output, the current common methods include robust optimization, optimal configuration of energy storage systems, and model predictive control. The advantage of model predictive control is that it can continuously adjust the control strategy according to the real-time state of the system and new energy prediction information to cope with uncertainty. However, model predictive control has the problem that the operating points not executed in the window do not constitute a guiding role, resulting in a low degree of utilization of new energy prediction information. SUMMARY

[0005] To address the problems existing in the prior art, this invention proposes a provincial power grid frequency control method based on a two-layer collaborative mechanism, which aims to effectively reduce the fluctuation amplitude of the system frequency, thereby reducing the system's power generation cost and improving the system's frequency regulation performance, thus ensuring the safety of the power grid during secondary frequency regulation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The present invention discloses a provincial power grid frequency control method based on a two-layer collaborative mechanism, characterized by its application in the secondary frequency regulation stage of the provincial power grid. The provincial power grid includes wind farms, photovoltaic power plants, energy storage, load aggregators, virtual power plants, and thermal power units, forming a provincial frequency collaborative control system with the control center. The provincial power grid frequency control method includes the following steps:

[0008] Step 1: Real-time calculation of the provincial power grid's performance. System frequency deviation at time and regional control deviation :

[0009] Step 2: Based on the real-time calculation results of Step 1, construct a two-layer collaborative mechanism based on working mode switching and frequency difference allocation;

[0010] Step 3: Based on the two-layer collaborative mechanism in Step 2, construct a control strategy and model based on frequency modulation index-superposition model predictive control:

[0011] Step 3.1: Construct a working mode switching strategy based on frequency modulation index;

[0012] Step 3.2: For the resources in frequency adjustment mode in Step 3.1, construct a frequency difference allocation strategy based on superposition model predictive control;

[0013] Step 3.3: Based on the predicted states of multiple time periods obtained by the frequency difference allocation strategy, construct a frequency control model and call the Gurobi solver to solve the frequency control model to obtain a frequency control scheme, including: frequency regulation objects, the optimal regulation power of thermal power units, energy storage, and virtual power plants participating in frequency regulation, the optimal load reduction of load aggregators participating in frequency regulation, and the optimal amount of wind and solar curtailment participating in frequency regulation.

[0014] The provincial power grid frequency control method based on a two-layer cooperative mechanism described in this invention is characterized in that step 1 includes:

[0015] Step 1.1: Calculate using equation (1) System power deviation generated by provincial power grid at any time :

[0016] (1)

[0017] In formula (1): and respectively the system load demand and the system tie-line power of the provincial power grid at the moment; , , and respectively the output power of the i-th thermal power unit, the j-th photovoltaic power station, the k-th wind farm and the l-th virtual power plant at the moment; respectively and respectively the output of the i-th energy storage at the moment and the load demand of the j-th load aggregator; ; and respectively the maximum number of thermal power units, photovoltaic power stations, wind farms, virtual power plants, energy storages and load aggregators, and ; the total number of resources;

[0018] Step 1.2, calculating the system frequency deviation of the provincial power grid at the moment using formula (2):

[0019] (2)

[0020] In formula (2): and respectively the inertia time constant and the load damping coefficient, is the time variable in the integral; is the power deviation amount under the time variable ;

[0021] Step 1.3, when the frequency deviation exceeds the safe range, calculating the regional control deviation of the provincial power grid at the moment using formula (3):

[0022] (3)

[0023] In formula (3), is the frequency offset coefficient.

[0024] Further, the step 2 comprises:

[0025] Step 2.1, defining ​​​​​​Instant resource working mode vector ; wherein, the working mode of the instant th resource, including: tracking plan mode and adjusting frequency mode; when =0, it means that the instant th resource is in tracking plan mode; when =1, it means that the instant th resource is in adjusting frequency mode;

[0026] Step 2.2, establish the relationship between the working mode switching and the frequency difference allocation period by using formula (4)-(5):

[0027] (4)

[0028] (5)

[0029] In formula (4)-(5), is the time of detecting frequency deviation, is the time interval of working mode switching, and represent the th switching period and the th switching period of working mode, and , is the total number of switching periods, is the time interval of frequency difference allocation period, is the total number of frequency difference allocation periods, and let the th switching period of working mode under the th frequency difference allocation period be ;

[0030] Step 2.3, establish the decision process of working mode switching and frequency difference allocation by using formula (6)-(7):

[0031] (6)

[0032] (7)

[0033] In formula (6)-(7), is the environmental information of the th frequency difference allocation period , is the system load demand amount of the ultra-short-term prediction under the th frequency difference allocation period ; is the​ the frequency difference allocation period the system tie-line power of the lower short-term prediction; the first the frequency difference allocation period the power generation of the wind farm of the lower short-term prediction; the first the frequency difference allocation period the power generation of the photovoltaic power station of the lower short-term prediction; the first the resource working mode vector within the switching period the first the frequency difference allocation period the actual output of all resources, wherein, the first the frequency difference allocation period the actual output of the first resource; the area control deviation of the first the frequency difference allocation period the action performed, wherein, the first the output change amount of the first resource;

[0034] Step 2.4, defining the first the frequency difference allocation period the frequency regulation resource set in the first , wherein, the first the frequency difference allocation period the working mode of the first resource; defining the first the frequency difference allocation period the resource set for tracking the plan is , and , and updating the first resource response result by using formula (8)-formula (10):

[0035] (8)

[0036] (9)

[0037] (10)

[0038] In formula (8)-formula (10), , , , respectively represent the first​ Frequency difference allocation period Thermal power units, energy storage, virtual power plants, and load aggregators participating in frequency regulation , They represent the first Frequency difference allocation period Wind farms and photovoltaic power stations participating in frequency regulation and For the first Frequency difference allocation period and the Frequency difference allocation period Next The actual power of each resource For the first Frequency difference allocation period Next The power generation capacity of each resource For the first Frequency difference allocation period Next The planned operating point power of each resource.

[0039] Furthermore, step 3.1 includes:

[0040] Step 3.1.1, calculate the first... The first switching period Frequency regulation index of each resource And based on the frequency modulation index, for the first In each switching period, I resources are sorted in ascending order, and the sequence number of each sorted resource is recorded as . ,in, Indicates the sorted order of the first... The serial number of each resource;

[0041] Step 3.1.2: Calculate the gain margin in both cases using equation (11):

[0042] (11)

[0043] In equation (11), for Next The actual output value of each resource , The first The maximum and minimum output of each resource; Indicates the first Regional control deviation during each switching period;

[0044] Step 3.1.3: Obtain the resource sequence number that satisfies the boundary point shown in equation (12). Thus, the first the working mode corresponding to the first resource to the first resource is set to "1", and the working mode corresponding to the first resource to the first resource is set to "0";

[0045] (12).

[0046] Further, the step 3.2 comprises:

[0047] Step 3.2.1, the control center sets the prediction window length under the control of the control center to , and uses formula (13)-(14) to construct a superposition model to predict the prediction model of the control:

[0048] (13)

[0049] In formula (13), represents the input state of the first time period in the first frequency difference allocation period , and , wherein, , , , , , , , are the thermal power, energy storage power, load aggregated merchant power, wind power output, photovoltaic power output, virtual power plant power and system power deviation of the first time period in the first period; ; N is the total number of time periods in the prediction window; is the input state of the first time period in the first period; is the disturbance input of the first time period in the first period, and , wherein, , , are the ultra-short-term prediction power of the tie line, the ultra-short-term prediction power of the load, and the power change amount of the tracking plan resource in the first time period in the first period; is the power deviation of the first time period in the first period; is the adjustment instruction of the first time period in the first period, and ,in, , , , , , , They are respectively Next Power regulation commands for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation adjustment commands for each time period. for Next Control variables for each time period, and ,in, , , , , , , They are respectively Next The power control variables for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation control variables for each time period. , and These are the coefficient matrices for the state variables, control variables, and disturbances, respectively, where T denotes transpose;

[0050] Step 3.2.2: Obtain the adjustment command using equation (14):

[0051] (14)

[0052] In equation (14), and These are the two corresponding weight matrices. For the first The first switching period within the first Frequency difference allocation period Issue the instruction for the first time slot; for The control variables for the first time period.

[0053] Furthermore, the frequency control model in step 3.3 includes:

[0054] Step 3.3.1: Construct an objective function J using equations (15)-(18) to minimize the cost per kilowatt-hour, wind and solar curtailment penalty, and frequency deviation penalty.

[0055] (15)

[0056] (16)

[0057] (17)

[0058] (18)

[0059] In formula (15) to formula (18), is the penalty of the frequency difference in the first time period under the prediction window is the penalty of the frequency difference in the first time period under the prediction window is the output change cost of the thermal power unit and the flexible resource participating in frequency modulation in the first time period under the prediction window is the penalty of wind and light curtailment when adjusting the frequency in the first time period under the prediction window is is the frequency deviation of the system in the first time period under the prediction window is the frequency deviation of the system in the first time period under the prediction window is the frequency difference penalty coefficient corresponding to the first time period under the prediction window is the frequency difference penalty coefficient corresponding to the first time period under the prediction window represents the output control variable of all resources in the first time period under the prediction window represents the output control variable of all resources in the first time period under the prediction window and are the output of the super short-term predicted wind power and photovoltaic in the first time period under the prediction window and are the actual output of the wind power and photovoltaic in the first time period under the prediction window and are the actual output of the wind power and photovoltaic in the first time period under the prediction window and are the control variable of the wind power and photovoltaic output in the first time period under the prediction window and represent the two-norm under different weight matrices and are the weight matrices determined by the generation cost and the wind and light curtailment penalty Step 3.3.2, decompose the objective function J, and construct the objective function under the prediction window by formula (19):

[0060] (19)

[0061] (19)

[0062] Step 3.3.3, establish the output constraints of the thermal power unit, wind farm, photovoltaic power station, virtual power plant, energy storage, and load aggregator by formula (20) to formula (27):

[0063] (20)​​​

[0064] (twenty one)

[0065] In equations (20)-(21), , They are respectively Next The first time period Output and control variables of each thermal power unit , The first Maximum power output of each thermal power unit when climbing uphill or downhill. , The first The maximum and minimum output of each thermal power unit;

[0066] (twenty two)

[0067] In equation (22), They are respectively Next The first time period The first photovoltaic power station, the first The power of a wind farm They are respectively Next The first time period The first photovoltaic power station, the first Control variables for a wind farm; , They are respectively Short-term forecast The first time period The first photovoltaic power station, the first The maximum power generation capacity of each wind farm;

[0068] (twenty three)

[0069] In equation (23), and They are respectively Next The first time period The output and control variables of a virtual power plant For the first Within the first switching period Unregulated power from a virtual power plant , The first The pre-clearing plan during the switching period is the first The adjustable range of a virtual power plant;

[0070] (twenty four)

[0071] (25)

[0072] (26)

[0073] In equations (24)-(26), and They are respectively Next The first time period The output and control variables of an energy storage system. express Next The first time period The state of charge of an energy storage device. , The first The upper and lower limits of the output power of each energy storage device. , For the first The upper and lower limits of the state of charge of a single energy storage battery. , , The first The efficiency coefficient, charge / discharge coefficient, and capacity of an energy storage device;

[0074] (27)

[0075] In equation (27), and They are respectively Next The first time period The load and load that can be reduced by the load aggregator. and The first Within the first switching period The load aggregator has both no-shrinkage and shrinkable loads.

[0076] The present invention provides an electronic device, comprising a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the provincial power grid frequency control method, and the processor is configured to execute the program stored in the memory.

[0077] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the steps of the provincial power grid frequency control method.

[0078] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0079] 1、The application constructs a double-layer collaborative mechanism based on working mode switching-frequency difference allocation, realizes adaptive conversion of two working modes of frequency difference allocation and tracking plan, and compared with the fixed resources in a single working mode in the prior art, the proposed mechanism can dynamically screen frequency modulation resources according to the frequency modulation index, realize collaborative frequency modulation of heterogeneous resources, and make the control resources more flexible.

[0080] 2、The working mode switching strategy based on the frequency modulation index in the application screens resources with excellent performance to participate in frequency modulation from three aspects of response time, response accuracy and adjustment rate, compared with the prior art which only screens from a single aspect, the evaluation standard is more comprehensive, and the frequency modulation performance of the system is improved.

[0081] 3、The improved superimposed model predictive control method in the application superimposes the adjustment instruction of the last period, so that the result is more forward-looking, compared with the prior art which is only affected by the current optimization variable, the leading control ability of the power grid is enhanced, while the frequency fluctuation of the system is suppressed, and the smoothness of resource output is improved. BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 The application is a provincial frequency collaborative control system.

[0083] Figure 2 The application is a working mode switching-frequency difference allocation collaborative control flowchart.

[0084] Figure 3 The application is a working mode switching-frequency difference allocation double-layer sequential decision process diagram.

[0085] Figure 4 The application is the output of each resource under strategy 2 and strategy 3. DETAILED DESCRIPTION

[0086] In the embodiment, a double-layer collaborative mechanism for wind farms, photovoltaic power stations, energy storage, load aggregators, virtual power plants and thermal power units of a provincial power grid is constructed, adaptive conversion of two working modes of frequency difference allocation and tracking plan is realized, collaborative frequency modulation of heterogeneous resources is realized, and an optimization method based on a frequency modulation index-superimposed model predictive control is proposed. The method switches the working mode through a dynamically updated frequency modulation index, realizes screening of frequency modulation resources, and reasonably allocates the frequency difference of the system based on the improved superimposed model predictive control, and can better cope with the uncertainty of new energy output. Specifically, a provincial power grid frequency control method based on a double-layer collaborative mechanism comprises:

[0087] Step 1: Real-time calculation of the system frequency deviation and the regional control deviation of the provincial power grid.

[0088] For example Figure 1As shown, due to the uncertainty of new energy output, system load demand and system tie-line power, there is an imbalance between the power generation and power consumption in the system, and the imbalance is calculated by formula (1) System power deviation generated by the provincial grid at time t

[0089] (1)

[0090] In formula (1): and are the system load demand and the system tie-line power of the provincial grid at time t, respectively; and are the output powers of the i-th thermal power unit, the j-th photovoltaic power station, the k-th wind farm and the l-th virtual power plant at time t, respectively; are the output of the m-th energy storage and the load demand of the n-th load aggregator at time t, respectively; are the maximum numbers of the thermal power unit, the photovoltaic power station, the wind farm, the virtual power plant, the energy storage and the load aggregator, respectively, and ; is the total number of resources.

[0091] The system frequency deviation generated by the provincial grid at time t is calculated by formula (2)

[0092] (2)

[0093] In formula (2): and are the inertia time constant and the load damping coefficient, is the time variable in the integral; is the power deviation at time variable .

[0094] When the frequency deviation exceeds the safe range , the regional control deviation generated by the provincial grid at time t is calculated by formula (3)

[0095] ​​​​​​​​​​​​​​​​​​ (3)

[0096] In equation (3), This is the frequency offset coefficient.

[0097] Step 2: Construct a two-layer collaborative mechanism based on working mode switching and frequency difference allocation;

[0098] Step 2.1, Resource Working Mode:

[0099] Resource operating modes are divided into two categories: tracking plans and adjusting frequency. (Definition) The time-resource working mode vector is denoted as ,in for Time of the first The working mode of each resource. When When it is 0, it means that in Anytime, Any Resource In the tracking plan mode, the resource maintains its planned daily operation without being affected by frequency adjustment commands; when When it is 1, it means Anytime, Any Resource It is in frequency adjustment mode, meaning the resource responds to frequency difference allocation commands.

[0100] Step 2.2, according to Figure 2 The working mode switching-frequency difference allocation coordinated control process utilizes the relationship between working mode switching and frequency difference allocation cycle in equations (4)-(5):

[0101] (4)

[0102] (5)

[0103] In equations (4)-(5), The time for detecting frequency deviation, This is the time interval for switching working modes. and The first one representing the working mode Each switching period and Each switching period, and , This represents the total number of time slots that can be switched. The time interval for allocating the frequency difference period. The total number of cycles allocated to the frequency difference, let the working mode's... The first switching period The frequency difference allocation period is denoted as .

[0104] Step 2.3, as follows Figure 3As shown, the decision-making process for working mode switching and frequency difference allocation is established using equations (6)-(7):

[0105] (6)

[0106] (7)

[0107] In equations (6)-(7), For the first Frequency difference allocation period Environmental information, For the first Frequency difference allocation period The system load demand forecast for the very short term; For the first Frequency difference allocation period System tie-line power for ultra-short-term forecasts; For the first Frequency difference allocation period Wind farm power generation forecast in the ultra-short term; For the first Frequency difference allocation period The power generation of photovoltaic power plants is predicted in the ultra-short term; For the first Resource working mode vector within each switching period; For the first Frequency difference allocation period The actual output of all resources, among which, For the first Frequency difference allocation period Next The actual output of each resource; for The area control deviation below; For the first Frequency difference allocation period The action to be performed for Next The change in the output of each resource.

[0108] Step 2.4, Update resource response results:

[0109] Definition of the first Frequency difference allocation period Down-modulation resource set is ,in, For the first Frequency difference allocation period Next The working mode of each resource; the response control center of thermal power units, energy storage, virtual power plants, load aggregators, wind farms, and photovoltaic power plants participating in frequency regulation. The adjustment command is sent and the response result is used as... Initial power:

[0110] (8)

[0111] (9)

[0112] In equations (8)-(9), , , , They represent the first Frequency difference allocation period Thermal power units, energy storage, virtual power plants, and load aggregators participating in frequency regulation , They represent the first Frequency difference allocation period Wind farms and photovoltaic power stations participating in frequency regulation For the first The frequency difference allocation period under the first The adjustment power of each resource and For the first Frequency difference allocation period and the Frequency difference allocation period Next The actual power of each resource For the first Frequency difference allocation period Next The power generation capacity of each resource.

[0113] Definition of the first Frequency difference allocation period The resource set for the next tracking plan is ,and The power of the tracking plan resources changes only with the day-ahead plan's operating point:

[0114] (10)

[0115] In equation (10), For the first Frequency difference allocation period Next The planned operating point power of each resource.

[0116] Step 3: Control strategy and model based on frequency modulation index-superimposed model predictive control;

[0117] Step 3.1, the working mode switching strategy based on the frequency modulation index:

[0118] Step 3.1.1, the adjustment speed sub-index of the first resource is calculated by using formula (11) - formula (14) the response time sub-index and the adjustment accuracy sub-index of the first resource in the first switching period :

[0119] (11)

[0120] (12)

[0121] (13)

[0122] (14)

[0123] In formula (11) - formula (14), is the average adjustment rate of the first resource from receiving the instruction to achieving the instruction in the first switching period is the time used for receiving the instruction and achieving the instruction, and satisfies ; is the average response time of the first resource in the first switching period is the time difference between receiving the instruction and the actual output exceeding the adjustment dead zone and no longer returning in the first switching period, and the adjustment dead zone is defined as 1% of the rated power is the average adjustment accuracy of the first resource in the first switching period, that is, the difference between the actual output and the required output of the instruction

[0124] The frequency modulation index of the first resource in the first switching period is calculated , and according to the frequency modulation index, the first resource in the first switching period is sorted in ascending order, and the serial number of each sorted resource is recorded as , wherein represents the serial number of the first resource after sorting ​​​​​​​​​​​​​​​

[0125] Step 3.1.2: Calculate the gain margin in both cases using equation (15):

[0126] (15)

[0127] In equation (15), for Next The actual output value of each resource , The first The maximum and minimum output of each resource; Indicates the first Regional control deviation during each switching period.

[0128] Step 3.1.3: Obtain the resource sequence number that satisfies the boundary point shown in equation (16). Thus, the first The resource to the first The working mode corresponding to each resource is set to "1", and the first resource... The resource to the first The working mode corresponding to each resource is set to "0";

[0129] (16)

[0130] Step 3.2, Frequency Difference Allocation Strategy Based on Superposition Model Predictive Control:

[0131] Step 3.2.1, Prediction Model:

[0132] Control Center The prediction window length for the time period is The state is composed of multiple time-period prediction information and the current time-period environmental information in the window. . Indicates in Short-term forecasts for the future System load during the time period, of which for Short-term forecast System load during a given time period. Similarly, this will not be elaborated upon here.

[0133] Window The change in output over a period of time constitutes the motion. . Indicates the first One resource in Making decisions shapes the future. The change in output over a period of time, of which for Next optimization The first time period Power change command for each resource.

[0134] Construct a prediction model using equations (17) and (18):

[0135] (17)

[0136] In equation (17), Indicates the first Frequency difference allocation period Next The input status for each time period, and ,in, , , , , , , They are respectively Next Thermal power unit power, energy storage power, load aggregator power, wind power output, photovoltaic power output, virtual power plant power, and system power deviation for each time period; N represents the total number of time periods within the prediction window; for Next Input status for each time period; for Next The disturbance input for each time period, and ,in, , , They are respectively Next The ultra-short-term forecast power of the tie line, the ultra-short-term forecast power of the load, and the power change of the tracking plan resources for each time period. for Next Power deviation over a given period; for Next Adjustment instructions for each time period, and ,in, , , , , , , They are respectively Next Power regulation commands for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation adjustment commands for each time period. for Next Control variables for each time period, and ,in, , , , , , , They are respectively Next The power control variables for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation control variables for each time period. , and These are the coefficient matrices for the state variables, control variables, and disturbances, respectively, with T representing the transpose.

[0137] Step 3.2.2: Using equation (18), the previous window instruction is superimposed on the control variable as an adjustment instruction:

[0138] (18)

[0139] In equation (18), and These are the corresponding weight matrices. For the previous window The instruction for the first time period is issued.

[0140] Step 3.3: Construct the frequency control model and call the solver to solve it:

[0141] Step 3.3.1: Construct an objective function J using equations (19)-(22) to minimize the cost per kilowatt-hour, wind and solar curtailment penalty, and frequency deviation penalty.

[0142] (19)

[0143] (20)

[0144] (twenty one)

[0145] (twenty two)

[0146] In equations (19)-(22), the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window the first time period under the prediction window

[0147] Step 3.3.2, decompose the objective function J, thereby constructing the objective function under the decision-making period using formula (23):

[0148] (23)

[0149] Step 3.3.3, related resource output constraints:

[0150] (1) thermal power unit constraints:

[0151] (24)

[0152] (25)

[0153] In formula (24)-(25), , are the output and control variable of the i-th thermal power unit in the first time period under the prediction window, respectively, , , , ,​​ are the maximum up and down ramping power of the i-th thermal power unit, respectively, , are the maximum and minimum power of the i-th thermal power unit, respectively.

[0154] (2) Wind farm and photovoltaic power station constraints:

[0155] (26)

[0156] In formula (26), , are the power of the j-th photovoltaic power station and the k-th wind farm in the i-th time period, respectively, are the control variables of the j-th photovoltaic power station and the k-th wind farm in the i-th time period, respectively, , are the maximum power generation of the j-th photovoltaic power station and the k-th wind farm in the i-th time period, respectively. (3) Virtual power plant constraints: (27)

[0157] In formula (27), and

[0158] are the power and control variable of the i-th virtual power plant in the j-th time period, is the unregulated power of the i-th virtual power plant in the j-th switching time period, ,

[0159] are the upper and lower adjustable ranges of the i-th virtual power plant in the j-th switching time period. (4) Energy storage constraints: (28) (5) Other constraints:

[0160]

[0161] (28)

[0162] ​​​​​​​​​​​​​​​​​​​​(29)

[0163] (30)

[0164] In formula (28) to formula (30), and respectively, the output and control variable of the first energy storage in the next time period, the state of charge of the first energy storage in the next time period, the upper and lower limits of the output power of the first energy storage, the upper and lower limits of the state of charge of the first energy storage, the efficiency coefficient, the charge and discharge coefficient and the capacity of the first energy storage. respectively,

[0165] (5) Load aggregator constraint:

[0166] (31)

[0167] In formula (31), and respectively, the load and the reducible load of the first load aggregator in the next time period, and respectively, the non-reducible load and the reducible load of the first load aggregator in the next switching time period. Step 3.2.4. The constructed objective function and constraints constitute a frequency control model, and a Gurobi solver is called to solve the frequency control model to obtain a frequency control scheme, including: frequency modulation objects, optimal adjustment power of thermal power units, energy storage and virtual power plants participating in frequency modulation, optimal reduced load of load aggregators participating in frequency modulation, and optimal abandoned wind and light quantity participating in frequency modulation.

[0168] In the embodiment, an electronic device includes a memory for storing a program supporting a processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0169] In the embodiment, an electronic device includes a memory for storing a program supporting a processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0170] ​​​​​​​​​​​​​In this embodiment, a computer readable storage medium stores a computer program, and the computer program is run by a processor to execute the steps of the above method.

[0171] For this example, the following three strategies are compared and analyzed in the scenario of 12:00-12:30 in summer.

[0172] (1) Method 1: The working mode is not switched, and the model predictive control method is used for frequency difference allocation.

[0173] (2) Method 2: The working mode is automatically switched, and the model predictive control method is used for frequency difference allocation.

[0174] (3) Method 3 (the method of the present application): The working mode is automatically switched, and the improved superimposed model predictive control method is used for frequency difference allocation.

[0175] Table 1 Number of frequency modulation resources in each period

[0176]

[0177] As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources.

[0178] As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources. Figure 4 As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources. As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources. As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources. As shown in Table 1, method 1 fixes 14 control objects to respond to the frequency modulation instruction according to the ultra-short-term prediction data at 12:00. Although this method is stable, it lacks flexibility, ignores the collaborative potential between resources and the changes in power grid characteristics at different times, and can lead to the situation that some resources are idle while other resources are over-regulated during the peak period of photovoltaic output. The automatic switching of the working mode in methods 2 and 3 makes the control resources more flexible, realizing the collaborative frequency modulation of heterogeneous resources.

[0179] Although the above embodiments have been described, those skilled in the art can make other changes and modifications to the embodiments once they know the basic creative concept, so the above description is only an embodiment of the present application, and does not limit the patent protection scope of the present application. Any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or directly or indirectly applied to other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A provincial power grid frequency control method based on a two-layer collaborative mechanism. Its characteristic is that it is applied to the secondary frequency regulation stage of a provincial power grid, which includes wind farms, photovoltaic power stations, energy storage, load aggregators, virtual power plants, and thermal power units, and forms a provincial frequency collaborative control system with the control center; the provincial power grid frequency control method includes the following steps: Step 1: Real-time calculation of the provincial power grid's performance. System frequency deviation at time and regional control deviation : Step 2: Based on the real-time calculation results of Step 1, construct a two-layer collaborative mechanism based on working mode switching and frequency difference allocation; Step 3: Based on the two-layer collaborative mechanism in Step 2, construct a control strategy and model based on frequency modulation index-superposition model predictive control: Step 3.1: Construct a working mode switching strategy based on frequency modulation index; Step 3.2: For the resources in frequency adjustment mode in Step 3.1, construct a frequency difference allocation strategy based on superposition model predictive control; Step 3.3: Based on the predicted states of multiple time periods obtained by the frequency difference allocation strategy, construct a frequency control model and call the Gurobi solver to solve the frequency control model to obtain a frequency control scheme, including: frequency regulation objects, the optimal regulation power of thermal power units, energy storage, and virtual power plants participating in frequency regulation, the optimal load reduction of load aggregators participating in frequency regulation, and the optimal amount of wind and solar curtailment participating in frequency regulation.

2. The provincial power grid frequency control method based on a two-layer collaborative mechanism as described in claim 1, characterized in that, Step 1 includes: Step 1.1: Calculate using equation (1) System power deviation generated by provincial power grid at any time : (1) In formula (1): and They are respectively Real-time system load demand and system tie-line power of the provincial power grid; , , and They are respectively Time of the first One thermal power unit, the first The first photovoltaic power station, the first The first wind farm and the first The output power of a virtual power plant; and They are respectively Time of the first The output of the first energy storage and the first The load power demand of a load aggregator; and These are the maximum number of thermal power units, photovoltaic power plants, wind farms, virtual power plants, energy storage, and load aggregators, respectively. ; The total number of resources; Step 1.2: Calculate using equation (2) System frequency deviation generated by provincial power grid at any time : (2) In formula (2): and These are the inertial time constant and the load damping coefficient, respectively. The time variable in the integral; For time variables Power deviation; Step 1.3, when the frequency deviation When the safety range is exceeded, use equation (3) to calculate. Regional control deviations generated by provincial power grids at all times : (3) In equation (3), This is the frequency offset coefficient.

3. The provincial power grid frequency control method based on a two-layer collaborative mechanism as described in claim 2, characterized in that, Step 2 includes: Step 2.1, Definition Time-based resource working mode vector ;in, for Time of the first The operating modes of each resource include: tracking plan mode and frequency adjustment mode; when =0 indicates Time of the first The resource is in tracking plan mode; when When =1, it means Time of the first The resource is in frequency adjustment mode; Step 2.2: Use equations (4)-(5) to establish the relationship between working mode switching and frequency difference allocation cycle: (4) (5) In equations (4)-(5), The time for detecting frequency deviation, This is the time interval for switching working modes. and The first one representing the working mode Each switching period and Each switching period, and , This represents the total number of time slots that can be switched. The time interval for allocating the frequency difference period. The total number of cycles allocated to the frequency difference, let the working mode's... The first switching period The frequency difference allocation period is denoted as ; Step 2.3: Establish the decision-making process for working mode switching and frequency difference allocation using equations (6)-(7): (6) (7) In equations (6)-(7), For the first Frequency difference allocation period Environmental information, For the first Frequency difference allocation period The system load demand forecast for the very short term; For the first Frequency difference allocation period System tie-line power for ultra-short-term forecasts; For the first Frequency difference allocation period Wind farm power generation forecast in the ultra-short term; For the first Frequency difference allocation period The power generation of photovoltaic power plants is predicted in the ultra-short term; For the first Resource working mode vector within each switching period; For the first Frequency difference allocation period The actual output of all resources, among which, For the first Frequency difference allocation period Next The actual output of each resource; for The area control deviation below; For the first Frequency difference allocation period The action to be performed for Next The change in the output of each resource; Step 2.4, Define the first Frequency difference allocation period Down-modulation resource set is ,in, For the first Frequency difference allocation period Next The working mode of the resource; define the first Frequency difference allocation period The resource set for the next tracking plan is ,and And update the first equation using equations (8)-(10). Resource response results: (8) (9) (10) In equations (8)-(10), , , , They represent the first Frequency difference allocation period Thermal power units, energy storage, virtual power plants, and load aggregators participating in frequency regulation , They represent the first Frequency difference allocation period Wind farms and photovoltaic power stations participating in frequency regulation and For the first Frequency difference allocation period and the Frequency difference allocation period Next The actual power of each resource For the first Frequency difference allocation period Next The power generation capacity of each resource For the first Frequency difference allocation period Next The planned operating point power of each resource.

4. The provincial power grid frequency control method based on a two-layer collaborative mechanism as described in claim 3, characterized in that, Step 3.1 includes: Step 3.1.1, calculate the first... The first switching period Frequency regulation index of each resource And based on the frequency modulation index, for the first In each switching period, I resources are sorted in ascending order, and the sequence number of each sorted resource is recorded as . ,in, Indicates the sorted order of the first... The serial number of each resource; Step 3.1.2: Calculate the gain margin in both cases using equation (11): (11) In equation (11), for Next The actual output value of each resource , The first The maximum and minimum output of each resource; Indicates the first Regional control deviation during each switching period; Step 3.1.3: Obtain the resource sequence number that satisfies the boundary point shown in equation (12). Thus, the first The resource to the first The working mode corresponding to each resource is set to "1", and the first resource... The resource to the first The working mode corresponding to each resource is set to "0"; (12)。 5. The provincial power grid frequency control method based on a two-layer collaborative mechanism as described in claim 4, characterized in that, Step 3.2 includes: Step 3.2.1: Instruct the control center to... The prediction window length is The prediction model for superimposed model predictive control is constructed using equations (13) and (14): (13) In equation (13), Indicates the first Frequency difference allocation period Next The input status for each time period, and ,in, , , , , , , They are respectively Next Thermal power unit power, energy storage power, load aggregator power, wind power output, photovoltaic power output, virtual power plant power, and system power deviation for each time period; N represents the total number of time periods within the prediction window; for Next Input status for each time period; for Next The disturbance input for each time period, and ,in, , , They are respectively Next The ultra-short-term forecast power of the tie line, the ultra-short-term forecast power of the load, and the power change of the tracking plan resources for each time period. for Next Power deviation over a given period; for Next Adjustment instructions for each time period, and ,in, , , , , , , They are respectively Next Power regulation commands for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation adjustment commands for each time period. for Next Control variables for each time period, and ,in, , , , , , , They are respectively Next The power control variables for thermal power units, energy storage, load aggregators, wind farms, photovoltaic power plants, virtual power plants, and system power deviation control variables for each time period. , and These are the coefficient matrices for the state variables, control variables, and disturbances, respectively, where T denotes transpose; Step 3.2.2: Obtain the adjustment command using equation (14): (14) In equation (14), and These are the two corresponding weight matrices. For the first The first switching period within the first Frequency difference allocation period Issue the instruction for the first time slot; for The control variables for the first time period.

6. The provincial power grid frequency control method based on a two-layer collaborative mechanism as described in claim 5, characterized in that, The frequency control model in step 3.3 includes: Step 3.3.1: Construct an objective function J using equations (15)-(18) to minimize the cost per kilowatt-hour, wind and solar curtailment penalty, and frequency deviation penalty. (15) (16) (17) (18) In equations (15)-(18), For the prediction window Frequency difference penalty for each time period To predict the output change costs of thermal power units and flexible resources participating in frequency regulation during the first period of the forecast window, The penalty for curtailing wind and solar power when adjusting the frequency during the first period under the prediction window; for Next System frequency deviation over a given time period For the first Frequency difference penalty coefficient corresponding to each time period; express Output control variables for all resources in the next time period; and They are respectively The output of wind power and photovoltaic power in the first period is predicted in the ultra-short term; and Wind power and solar power respectively Actual output during the first period; and They are respectively The control variables for wind power and solar power output in the first period are... and Represents the L2 norm under different weight matrices. and These are weight matrices determined by power generation costs and penalties for wind and solar curtailment, respectively. Step 3.3.2, based on the decision-making period The objective function J is decomposed, and then constructed using equation (19). The objective function is as follows : (19) Step 3.3.3: Use equations (20)-(27) to establish output constraints for thermal power units, wind farms, photovoltaic power plants, virtual power plants, energy storage, and load aggregators: (20) (21) In equations (20)-(21), , They are respectively Next The first time period Output and control variables of each thermal power unit , The first Maximum power output of each thermal power unit when climbing uphill or downhill. , The first The maximum and minimum output of each thermal power unit; (22) In equation (22), They are respectively Next The first time period The first photovoltaic power station, the first The power of a wind farm They are respectively Next The first time period The first photovoltaic power station, the first Control variables for a wind farm; , They are respectively Short-term forecast The first time period The first photovoltaic power station, the first The maximum power generation capacity of each wind farm; (23) In equation (23), and They are respectively Next The first time period The output and control variables of a virtual power plant For the first Within the first switching period Unregulated power from a virtual power plant , The first The pre-clearing plan during the switching period is the first The adjustable range of a virtual power plant; (24) (25) (26) In equations (24)-(26), and They are respectively Next The first time period The output and control variables of an energy storage system. express Next The first time period The state of charge of an energy storage device. , The first The upper and lower limits of the output power of each energy storage device. , For the first The upper and lower limits of the state of charge of a single energy storage battery. , , The first The efficiency coefficient, charge / discharge coefficient, and capacity of an energy storage device; (27) In equation (27), and They are respectively Next The first time period The load and load that can be reduced by the load aggregator. and The first Within the first switching period The load aggregator's unreduced load and the load that can be reduced.

7. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing any of the provincial power grid frequency control methods of claims 1-6, and the processor is configured to execute the program stored in the memory.

8. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is run by the processor, it executes the steps of any of the provincial power grid frequency control methods described in claims 1-6.

Images