A multi-source collaborative optimization scheduling method of CPS standard

Abstract

The present application relates to the technical field of power system control, and more particularly to a CPS standard multi-source collaborative optimization scheduling method, which is executed in a preset optimization period and scheduling period, and takes the actual grid-connected power of new energy and the climbing rate and power increase and decrease instruction variables of AGC units as decision variables, and is used for the collaborative optimization scheduling of new energy and AGC units under the condition of meeting the CPS standard. The present application has the advantages that the actual grid-connected power of new energy is changed from a fixed input variable to a decision variable, so that the new energy is changed from an uncontrollable "disturbance source" to a virtual frequency regulation resource which can be actively utilized, when the system has power surplus or the AGC regulation capacity is insufficient, the active reduction of the grid-connected output of new energy is equivalent to virtual downward regulation, and additional regulation freedom is provided for ACE control, and in the second scheme, the active power abandonment rate is 1.49%, the P2 examination point is improved from C level to A level, and the CPS2 qualified rate is improved from 75% to 100%, which proves that the abandonment of electricity as an active control means is effective.

Description

Technical Field

[0001] This invention relates to the field of power system control technology, and in particular to a multi-source collaborative optimization scheduling method based on the CPS standard. Background Technology

[0002] my country's new energy industry has achieved leapfrog development. By 2025, the installed capacity of new energy in Liaoning Province's power grid had exceeded 35 million kilowatts, with the combined capacity penetration rate of wind and solar power exceeding 41%, and the proportion of clean energy power generation capacity and electricity generation both exceeding 50%. However, the output of new energy is significantly affected by natural conditions. Wind power power fluctuations can reach 10% to 15% of the rated capacity per minute, and photovoltaic power output is also highly volatile due to factors such as cloud cover. This directly leads to a sharp rise and fall in the net load of the power grid, posing a severe challenge to the active power balance of the AGC system.

[0003] The AGC system is a core technology for frequency and tie-line power control in power systems. Its function is to compensate for net load fluctuations by adjusting the output of AGC units in real time. In 2024, the Northeast China Power Grid comprehensively revised its assessment method for tie-line power control management, making significant improvements in three aspects: the allocation of the B coefficient, the introduction of the inherent ACE standard deviation of the control area, the allocation of CPS assessment indicators, and the quantification of the grid dispatch CFC mode, forming a four-level discrete assessment and evaluation system of A / B / C / D.

[0004] The current new energy-thermal power dispatching method has the following core problems: (1) The output of new energy sources is regarded as an unadjustable fixed value. AGC units passively track the fluctuations of new energy sources, and the adjustment potential of the grid-connected power of new energy sources is not utilized. In the traditional scheduling model, the output of new energy sources is only used as a deterministic input in the power balance calculation. When the system has a power surplus or the AGC adjustment capacity is insufficient, there is no mechanism to use power curtailment as a control measure.

[0005] (2) Traditional AGC optimization scheduling models do not fully consider the differences in regulation characteristics between hydropower and thermal power units. Hydropower units have the characteristics of extremely fast ramping speed, extremely short minimum continuous ramping time, and low regulation cost, but existing models use homogenization for AGC units and do not utilize hydropower as a priority regulation resource.

[0006] (3) Most existing studies are based on traditional CPS assessment standards, and do not adopt the factor-based CPS assessment standards revised by the Northeast Power Grid in 2024, nor do they reflect the frequency regulation contribution factor of the grid dispatch CFC mode. This results in a mismatch between the dispatch scheme and the actual assessment system.

[0007] (4) Existing AGC optimization scheduling methods mostly adopt a single open-loop optimization mode and lack a real-time feedback closed-loop correction mechanism. Open-loop optimization relies entirely on ultra-short-term forecast data. When the forecast error of new energy sources is large (especially the minute-level wind power fluctuation), the CPS assessment index may deviate significantly from the expected value, resulting in failure to pass the assessment. Summary of the Invention

[0008] The purpose of this invention is to provide a multi-source collaborative optimization scheduling method for the CPS standard, which solves the problem of CPS performance failure caused by large fluctuations in renewable energy output and passive tracking by AGC units under high-proportion renewable energy grid connection. By using renewable energy grid-connected power as a decision variable, an optimization model with CPS1 / CPS2 constraints is constructed. A two-stage scheduling framework of intraday open-loop optimization and real-time closed-loop feedback is adopted, and the rapid regulation capability of hydropower is utilized in a differentiated manner to achieve collaborative scheduling of renewable energy and AGC units, thereby improving the CPS performance compliance rate and grid frequency quality.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A multi-source collaborative optimization scheduling method based on the CPS standard is implemented on a rolling basis with a preset optimization cycle and scheduling period. It uses the actual grid-connected power of new energy sources and the ramp rate and power increase / decrease indicators of AGC units as decision variables. Under the condition of meeting the CPS standard, it is used for the collaborative optimization scheduling of new energy sources and AGC units. The specific steps are as follows: S1. Construct an AGC unit output model that includes decision variables for grid-connected renewable energy power. S2. Construct a collaborative scheduling optimization model with the goal of minimizing the combined costs of AGC auxiliary adjustment and power curtailment penalty. S3. Establish a constraint system that includes system power balance, frequency deviation, CPS1 and CPS2 indicators; S4. Use an adaptive evolutionary programming algorithm to solve the intraday open-loop optimization problem; S5. Introduce a closed-loop feedback controller for real-time closed-loop feedback correction: within a predetermined time, adjust the correction amount according to the measured ACE, distribute the correction amount to each AGC unit according to the adjustable capacity ratio of hydropower priority, and prioritize reducing the grid-connected output of new energy for new energy-thermal power coordinated regulation when there is a power surplus. S6. Set up a comparison scheme to verify the collaborative scheduling scheme, which is used to evaluate the control performance of the collaborative scheduling method in improving CPS performance indicators in scenarios with high penetration of new energy.

[0010] In S1, the AGC unit output model containing the decision variables of renewable energy grid-connected power is constructed as follows: S11. Object selection: AGC unit group consisting of hydropower units and thermal power units was selected; S12, Decision Variables: (1) Power increase / decrease indicator variables and climbing rate For decision variables of thermal power and hydropower units; in: ; (2) Power output from actual grid connection of new energy sources For new energy decision-making variables; in: ; The available power output for combined wind and solar power is expressed in MW. S13, Unit Characteristic Parameters: The hydropower unit's climbing speed is 0~100MW / min, and the minimum continuous climbing time is 1min; The climbing rate of thermal power units is 0~60MW / min or 0~10MW / min, and the minimum continuous climbing time is 1~4min; Each AGC unit participates in primary and secondary frequency regulation.

[0011] In S12, the decision variables for renewable energy transform the curtailment behavior from a passive loss into an active control measure. When the system experiences power surplus or insufficient AGC regulation capacity, the actual grid-connected output of renewable energy is actively reduced. This is equivalent to virtual downward adjustment, which forms a coordinated control with the increase or decrease of output of hydropower units.

[0012] In S2, the objective function of the cooperative scheduling optimization model is: ; in: The adjustment cost for AGC units is calculated as the product of the deviation of the planned output of each unit and the cost coefficient, in yuan. The penalty fee for abandoning renewable energy is calculated as the product of the abandoned amount and the penalty coefficient, and the unit is yuan. and As weight; This is the normalized baseline value for AGC unit adjustment costs, in yuan. This is the normalized baseline value for the fee for power curtailment, in yuan; This represents the objective function value.

[0013] In S2; AGC auxiliary adjustment fee, the expression is: ; in: Here is the cost coefficient for the i-th AGC unit: 0.25 yuan / kWh for hydropower units, 0.25 yuan / kWh for fast thermal power units, and 0.50 yuan / kWh for slow thermal power units. The planned power output of the i-th AGC unit during time period t, in MW; The actual output of the i-th AGC unit during time period t-1 is expressed in MW. Let be the power increase / decrease indicator variable for the i-th AGC unit during time period t; The duration of the scheduling period is expressed in hours (h). The penalty fee for abandoned electricity is expressed as: ; in: The amount of renewable energy wasted during time period t is expressed in MWh. This is the power curtailment penalty factor, expressed in yuan / kWh; The combined wind and solar power output for time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW.

[0014] In S3, a constraint system is established that includes system power balance, frequency deviation, CPS1, and CPS2 indices, including: S31. The power balance equation states that the difference between the total power generation output of the system and the total load and network losses is equal to zero. The expression is: ; in: The output of the i-th AGC unit during time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW. The total system load for time period t is expressed in MW. The system network loss during time period t is expressed in MW. S32. The frequency deviation is solved directly using the quasi-steady-state closed-loop formula for frequency, and the expression is: ; in: Frequency deviation, in Hz; This refers to the power imbalance, measured in MW. The system unit adjustable power is expressed in MW / Hz. The unit regulating power of the interconnected power grid is expressed in MW / Hz. The total system regulation power is expressed in MW / Hz. constraint This represents the upper limit of the allowable frequency deviation, in Hz. The constraint system for the S33, CPS1, and CPS2 indicators includes: The first constraint of the control performance standard, CPS1, is within the assessment period. That is, the CPS1 value is not less than 100%; CPS assessment uses a four-level evaluation system: A / B / C / D. It was grade A at the time; Furthermore, when the control performance standard item 2, CPS2, is qualified, it is grade B; Furthermore, a grade C is assigned when CPS2 fails to meet the standard. It was classified as Grade D at the time; in: The value is CPS1. CPS2 passing means that the absolute value of the 15-minute average of the regional control deviation ACE does not exceed the CPS2 assessment limit during the assessment period. CPS2 failure refers to the absolute value of the 15-minute average of the area control deviation (ACE) exceeding the CPS2 assessment limit during the assessment period.

[0015] In S4, the coding method using the adaptive evolutionary programming algorithm is as follows: the coding unit is composed of the power increase / decrease indicator variable of each AGC unit, the ramp rate, and the grid-connected output of new energy sources. The coding dimension is: (2 × number of AGC units + 1) × 15 time periods; An adaptive Gaussian mutation and a random competition parameter q are used to search for the optimal scheduling scheme. The constraint handling adopts a combination of penalty function method and direct correction method, where q is the number of competing individuals in the random competition selection.

[0016] In S4, the encoding method for the adaptive evolutionary programming algorithm is expressed as follows: ; ; in: Let $\frac{i}{i}$ be the power increase / decrease indicator variable for the $i$-th AGC unit during time period $t$, with values ​​of ${1}$, $0$, and $-1$. Let be the ramp rate of the i-th AGC unit during time period t, in MW / min; The actual grid-connected power output of new energy sources during time period t is expressed in MW. This represents the total number of AGC units, in units. This represents the total number of scheduling periods; Let be the adjustable capacity upper limit direction indicator variable for the i-th AGC unit during time period t; Let be the adjustable capacity lower limit direction indicator variable of the i-th AGC unit during time period t; The maximum adjustable capacity of the i-th AGC unit during time period t is expressed in MW. The combined wind and solar power output for time period t is expressed in MW. Encode decision variables into vectors; This constitutes a 105-dimensional search space; Population size 60, maximum number of generations 80, random competition parameter q=15, penalty coefficient ; Adaptive variable asynchronous length, the expression is: ; in: Let be the variable length of the j-th dimension decision variable for the i-th individual; Let j be the current value of the j-th dimension decision variable for the i-th individual; Let be the upper limit of the range of values ​​for the j-th dimension decision variable; This represents the lower bound of the range of values ​​for the j-th dimension decision variable; The fitness function is: ; ; in: The fitness function value; Let i be the objective function value corresponding to the i-th individual; This is the penalty term for the k-th constraint. This represents the total number of constraints. Let $\frac{k}{k}$ be the absolute value of the limit exceeded by the k-th constraint. Let be the penalty coefficient for the k-th constraint. This represents the upper limit value of the k-th constraint. This is the lower limit value of the k-th constraint. This represents the actual value of the k-th constraint under the current solution; The chosen strategy introduces a concentration regulation mechanism.

[0017] In S5, the closed-loop feedback controller is a PI controller, which has anti-integral saturation measures such as ACE dead zone and integral limiting. The correction amount is distributed among AGC units according to the product of priority coefficient and adjustable margin. The priority coefficient of hydropower units is greater than that of thermal power units. When there is a power surplus, the ratio of the adjustment space of new energy to the PI correction amount determines the proportion of new energy to be borne, and the remaining adjustment amount is distributed to each AGC unit according to the weighted margin.

[0018] In S6, a comparison scheme is set up to verify the cooperative scheduling scheme, including: Option 1 involves CPS regulation via AGC units only, with full grid connection of renewable energy. Option 2: By coordinating AGC units with new energy sources for CPS regulation, the grid-connected output of new energy sources can be actively reduced; The two schemes were compared in simulation under the same initial conditions. By comparing the mean of CPS1, the pass rate of CPS2, the AGC regulation cost, the curtailment rate and the frequency deviation, the control performance of the cooperative scheduling method under the CPS test was evaluated.

[0019] Compared with the prior art, the beneficial effects of the present invention are: 1. By changing the actual grid-connected power of new energy sources from a fixed input variable to a decision variable, new energy sources are transformed from uncontrollable "disturbance sources" into "virtual frequency regulation resources" that can be actively utilized. When the system has a power surplus or insufficient AGC regulation capacity, actively reducing the grid-connected power output of new energy sources is equivalent to "virtual downward regulation," providing additional regulation freedom for ACE control. Example 1 shows that Scheme 2, with an active curtailment rate of 1.49%, improves the P2 assessment point from level C (CPS2 failure) to level A, and the CPS2 pass rate from 75% to 100%, proving the effectiveness of curtailment as an active control measure. 2. A constraint system containing a frequency quasi-steady-state closed-loop formula and CPS1 / CPS2 indicators was constructed. Through two-stage scheduling of intraday open-loop optimization and real-time closed-loop feedback, the CPS performance indicators were significantly improved. In Example 1, the 60-minute average KCPS1 of Scheme 2 increased from 164.2% in Scheme 1 to 190.2%. In Example 2, the average CPS1 increased from 119.7% to 173.8%, and the CPS2 pass rate increased from 75% to 100%. This verifies that the present invention can effectively improve frequency control performance under different system scales and different new energy penetration rates. 3. Differentiated modeling of hydropower units and thermal power units is carried out, and a priority coefficient for hydropower units is introduced in the PI correction allocation to make full use of the rapid ramping capability of hydropower (0~100MW / min, minimum continuous ramping time 1min). In Example 1, the G1 hydropower unit of Scheme 2 undertakes the main rapid correction task, reduces the pressure of reverse regulation of slow thermal power units, and improves the allocation structure of regulation resources. 4. It overcomes the shortcomings of relying on ultra-short-term forecast data for single open-loop optimization. On the basis of 15-minute rolling EP optimization, it adds PI closed-loop feedback correction every minute. Intraday open-loop optimization provides the optimal output benchmark, and real-time closed-loop feedback compensates for the prediction deviation caused by minute-level fluctuations of new energy sources. This makes the scheduling scheme both forward-looking and robust. In Example 1, the ACE curve of Scheme 2 converges within the CPS2 limit at the assessment points P2 to P4, which verifies the effectiveness of PI closed-loop feedback correction. 5. Replacing the traditional B coefficient allocation with factor allocation, introducing factor quantification of the frequency modulation contribution of the network dispatching CFC mode, and adopting a four-level discrete evaluation system of A / B / C / D to replace the continuous CPS value, so that the dispatching model matches the actual assessment rules and improves the engineering practical value of the dispatching scheme. Attached Figure Description

[0020] Figure 1 This is the topology diagram of the IEEE 14-node system.

[0021] Figure 2 This is the system load forecast curve.

[0022] Figure 3 This is a comparison chart of the predicted and actual available power output of wind and solar power.

[0023] Figure 4 This is a comparison chart of the KCPS1 rolling values ​​for the two schemes.

[0024] Figure 5 This is a comparison chart of the area control deviation (ACE) of the two schemes.

[0025] Figure 6 This is a comparison chart of the frequency changes of the two schemes.

[0026] Figure 7 This is a comparison chart of the tie-line power of the two schemes.

[0027] Figure 8 This is a comparison chart of the real-time PI correction values ​​for the two AGC unit schemes.

[0028] Figure 9 This is a cost comparison chart of the two options.

[0029] Figure 10 This is a graph showing the changes in the amount of abandoned electricity and the cumulative cost of abandoned electricity under Scheme 2. Detailed Implementation

[0030] The present invention will now be described in detail with reference to the accompanying drawings, but it should be noted that the implementation of the present invention is not limited to the following embodiments.

[0031] The following embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments. Unless otherwise specified, the methods used in the following embodiments are conventional methods.

[0032] Example 1:

[0033] A multi-source collaborative optimization scheduling method based on the CPS standard is implemented on a rolling basis with an optimization cycle of 15 minutes and a scheduling period of 1 minute. The method uses the actual grid-connected power of new energy sources and the ramp rate and power increase / decrease indicators of AGC units as decision variables. Under the condition of meeting the CPS standard, it is used for the collaborative optimization scheduling of new energy sources and AGC units. The specific steps are as follows: S1. Construct an AGC unit output model that includes decision variables for grid-connected renewable energy power. See Figure 1 Using the modified IEEE 14-node system to simulate a provincial control area, an AGC unit output model containing new energy grid-connected power decision variables was constructed.

[0034] S11. Object selection: An AGC (Automatic Generation Control) unit group consisting of hydropower units and thermal power units was selected. (See...) Figure 1 In the system, G1 is a hydropower AGC unit connected to node 1, with a rated capacity of 350MW; G2 is a fast thermal power AGC unit connected to node 2, with a rated capacity of 200MW; G3 is a non-AGC thermal power unit connected to node 6, with a rated capacity of 200MW; and G4 is a slow thermal power AGC unit connected to node 8, with a rated capacity of 240MW. Wind farms are connected to node 3, and photovoltaic power stations are connected to node 6; both are 200MW-class renewable energy sources and are connected to the external power grid through node 5. The total renewable energy penetration rate is approximately 30%.

[0035] S12, Decision Variables: (1) Power increase / decrease indicator variables and climbing rate For decision variables of thermal power and hydropower units; in: ; A value of 1 indicates an increase in power, -1 indicates a decrease in power, and 0 indicates that the power remains unchanged.

[0036] For example, during a certain optimization cycle at the P2 assessment point (9:15-9:30), when the system experiences a power deficit, the G1 hydropower unit... =1, =60MW / min, indicating that the unit increases its output at a rate of 60MW / min; when the system has a power surplus, =-1 indicates a reduction in output.

[0037] (2) Power output from actual grid connection of new energy sources For new energy decision-making variables; in: The available output for combined wind and solar power is expressed in MW; specifically, it is the sum of available wind power output and available solar power output. ; Available wind power output; Power can be generated from photovoltaic power; For example, at time t=9:20, by Figure 3 It is known that the available output of wind power is approximately 120MW, and the available output of photovoltaic power is approximately 110MW. =230MW, actual grid-connected power output of new energy sources The selection can be made by the optimization model within the range of 0 to 230MW.

[0038] The decision variables for renewable energy transform the behavior of curtailment from a passive loss to an active control measure. When the system has a power surplus or insufficient AGC regulation capacity, the actual grid-connected output of renewable energy is actively reduced. This is equivalent to virtual downward adjustment, which forms a coordinated control with the increase or decrease of output of hydropower units.

[0039] S13, Unit Characteristic Parameters: Hydropower Unit G1: Climbing rate is 0-100MW / min, minimum continuous climbing time is 1min, initial output is 120MW, cost factor is 0.25 yuan / kWh, and unit regulating power is 50MW / Hz; Fast-moving thermal power unit G2: climbing rate 0-60MW / min, minimum continuous climbing time 1min, initial output 120MW, cost factor 0.25 yuan / kWh, unit regulating power 50MW / Hz; Slow-speed thermal power unit G4: climbing rate 0-10MW / min, minimum continuous climbing time 4min, initial output 168MW, cost factor 0.50 yuan / kWh, unit regulating power 26.7MW / Hz; Each AGC unit participates in both primary and secondary frequency regulation.

[0040] S2. Construct a collaborative scheduling optimization model with the goal of minimizing the combined costs of AGC auxiliary adjustment and power curtailment penalty. The simulation period was from 9:00 AM to 10:00 AM, a total of 60 minutes, with an optimization cycle of 15 minutes and a scheduling period of 1 minute within each cycle. The system load curve is shown below. Figure 2 The load fluctuates within the range of approximately 756–783 MW. The available power output curves for wind and solar power are as follows: Figure 3 As shown, wind power output is roughly in the range of 70–176 MW, and photovoltaic power output is roughly in the range of 94–145 MW.

[0041] The collaborative scheduling optimization model has the following objective function: ; in: The adjustment cost for AGC units is calculated as the product of the deviation of the planned output of each unit and the cost coefficient, in yuan. The penalty fee for abandoning renewable energy is calculated as the product of the abandoned amount and the penalty coefficient, and the unit is yuan. and The weights are set to 0.6 and 0.4 respectively. The normalized baseline value for AGC unit adjustment costs, in yuan, is obtained through independent optimization of AGC costs. The normalized benchmark value for the power curtailment penalty fee is set at 1100 yuan; This represents the objective function value.

[0042] AGC auxiliary adjustment fee, the expression is: ; Taking the G1 hydropower unit at t=9:01 as an example: =122MW, =120MW, =0.25 yuan / kWh, =1 / 60h, then the adjustment cost of G1 at 9:01 is 0.25×122-120×1 / 60=0.0083 yuan.

[0043] in: The cost coefficient for the i-th AGC unit is 0.25 yuan / kWh for hydropower unit G1, 0.25 yuan / kWh for fast thermal power unit G2, and 0.50 yuan / kWh for slow thermal power unit G4. The planned power output of the i-th AGC unit during time period t, in MW; The actual output of the i-th AGC unit during time period t-1 is expressed in MW. Let $\frac{i}{t}$ be the power increase / decrease indicator variable for the $i$-th AGC unit during time period $t$, with a value of ${1}$, $0$, and $-1$. The duration of the scheduling period is set to 1 / 60h (i.e., 1 minute). The penalty fee for abandoned electricity is expressed as: ; Taking t=9:20 as an example, =230MW, if optimization decision =210MW, then the amount of power wasted is 20MW. =1 / 60h, the penalty for abandoning electricity is 0.30×20×1 / 60=0.10 yuan.

[0044] in: The amount of renewable energy wasted during time period t is expressed in MWh. This is the curtailment penalty factor, expressed in yuan / kWh, and is set to 0.30 yuan / kWh. The combined wind and solar power output for time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW.

[0045] S3. Establish a constraint system including system power balance, frequency deviation, CPS1 and CPS2 indicators, including: S31. The power balance equation states that the difference between the total power generation output of the system and the total load and network losses is equal to zero. The expression is: ; Taking t=9:15 as an example: =135MW, =130MW, =170MW, =220MW, =765MW, =10MW; Then 135+130+170+220-765-110=-110MW≠0, this scheme does not meet the power balance constraint, and the output of each unit needs to be adjusted through optimization algorithm to make the equation true.

[0046] in: The output of the i-th AGC unit during time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW. The total system load for time period t is in MW. See [link / reference]. Figure 2 ; The system network loss during time period t is expressed in MW. S32. The frequency deviation is solved directly using the quasi-steady-state closed-loop formula for frequency, and the expression is: ; Taking t=9:15 as an example, if the power imbalance... =-15MW (meaning the system has a power deficit of 15MW). =2006.25MW / Hz, =2006.25MW / Hz; =-(-15) / (2006.25+2006.25)=15 / 4012.5=0.00374Hz, the frequency increases by 0.00374Hz.

[0047] in: This is the frequency deviation, measured in Hz. It is the power imbalance quantity, measured in MW, and is determined by the difference between the system's power generation output and the load and network losses. The system unit regulating power is taken as 2006.25 MW / Hz; The unit regulating power of the interconnected power grid is taken as 2006.25 MW / Hz; + The total unit regulating power of the system is taken as 4012.5 MW / Hz; constraint , The upper limit of the allowable frequency deviation is set to 0.2Hz.

[0048] The constraint system for the S33, CPS1, and CPS2 indicators includes: The constraint for CPS1 (the first indicator of the control performance standard) is that the CPS1 value shall not be lower than 100% during the assessment period; CPS assessment uses a four-level evaluation system: A / B / C / D. It was grade A at the time; Furthermore, when the control performance standard item 2, CPS2, is qualified, it is grade B; Furthermore, a grade C is assigned when CPS2 fails to meet the standard. It was classified as Grade D at the time; in: The value is CPS1. CPS2 qualification means that the absolute value of the 15-minute average of the regional control deviation ACE does not exceed the CPS2 assessment limit during the assessment period. In this embodiment, the CPS2 assessment limit is 7.8MW. CPS2 failure refers to the absolute value of the 15-minute average of the area control deviation (ACE) exceeding the CPS2 assessment limit during the assessment period.

[0049] CPS1 and CPS2 are calculated according to the new revised standard of Northeast Power Grid in 2024, introducing factors that reflect the responsibility share of the control area in the whole network disturbance and factors that quantify the frequency regulation contribution of the grid dispatch CFC mode.

[0050] S4. Use an adaptive evolutionary programming algorithm to solve the intraday open-loop optimization problem; An optimization cycle is defined as 15 minutes, with each cycle comprising 15 scheduling periods (each minute being one period), for a total of 4 optimization cycles. The individual code is composed of power increase / decrease indicators for each AGC unit, ramp rate, and grid-connected power output from renewable energy sources.

[0051] See Figure 4The intraday open-loop optimization problem is solved by using an adaptive evolutionary programming algorithm to obtain the output plans of each AGC unit and the grid-connected output plans of new energy sources for each time period.

[0052] The coding method using the adaptive evolutionary programming algorithm is as follows: Each AGC unit's power increase / decrease indicator variable, ramp rate, and grid-connected renewable energy output collectively constitute the coded individual, with the coding dimensions being: (2 × number of AGC units + 1) × 15 time periods; If there are 3 AGC units (G1 hydropower, G2 fast thermal power, G4 slow thermal power), then the coding dimension is: (2×3+1)×15=105 dimensions.

[0053] An adaptive Gaussian mutation and a random competition parameter q are used to search for the optimal scheduling scheme. The constraint handling adopts a combination of penalty function method and direct correction method, where q is the number of competing individuals in the random competition selection.

[0054] The encoding method for the adaptive evolutionary programming algorithm is expressed as follows: ; ; ; ; Taking the first AGC unit (G1 hydropower unit) during the time period t=1 as an example: , MW / min indicates the power increase / decrease indication and ramp rate of the unit during the first dispatch period.

[0055] in: Let $\frac{i}{t}$ be the power increase / decrease indicator variable for the $i$-th AGC unit during time period $t$, with a value of ${1}$, $0$, and $-1$. Let G be the ramp rate of the i-th AGC unit during time period t, in MW / min. The range of values ​​is determined according to the unit type: G1 for hydropower units is 0 to 100 MW / min, G2 for fast thermal power units is 0 to 60 MW / min, and G4 for slow thermal power units is 0 to 10 MW / min. The actual grid-connected power output of new energy sources during time period t is expressed in MW, and the value ranges from 0 to the combined available power output of wind and solar power. N represents the total number of AGC units, which is 3. T represents the total number of scheduling periods, with each optimization cycle consisting of 15 periods. This is used to encode the decision variables into vectors; This constitutes a 105-dimensional search space; Population size 60, maximum number of generations 80, random competition parameter q=15, penalty coefficient M=10 6 ; Adaptive variable asynchronous length, the expression is: ; in: , ; n is the dimension of the decision variables, which is set to 105; Taking the second-dimensional decision variable of the first individual as an example: if the current =0.1, random number =0.5, =-0.3, then:

[0056] Let be the variable length of the j-th dimension decision variable for the i-th individual; This is the global scaling factor for variation. These are random numbers distributed according to a standard normal distribution. The j-th dimension of the i-th individual is a random number from a standard normal distribution. This is the scaling factor for individual variation; The fitness function is: ; ; For example, the objective function value =1.15, power balance constraint exceeded limit =5MW (allowable deviation is 0), CPS1 constraint exceeds limit =2% (CPS1=98%, lower than 100%) = =10 6 Then the penalty term = 10 6 ×5 2 +10 6 ×0.02 2 ≈25×10 6 , =1 / (1.15+25×10 6 ) ≈ 4 × 10 -8 .

[0057] The fitness function value; Let i be the objective function value corresponding to the i-th individual; Let $\frac{k}{k}$ be the absolute value of the limit exceeded by the k-th constraint. is the penalty coefficient for the k-th constraint.

[0058] This represents the upper limit value of the k-th constraint. This is the lower limit value of the k-th constraint. This represents the actual value of the k-th constraint under the current solution; The selection strategy introduces a concentration regulation mechanism to prevent premature convergence.

[0059] S5. Introduce a closed-loop feedback controller for real-time closed-loop feedback correction: The scheduling period is 1 minute. At each minute, the adjustment correction amount is calculated based on the measured ACE. The correction amount is distributed to each AGC unit according to the adjustable capacity ratio of hydropower priority. When there is a power surplus, the grid-connected output of new energy is reduced first to achieve coordinated regulation of new energy and thermal power.

[0060] See Figure 8 The minute-by-minute variation curves of the real-time PI correction for each AGC unit under the two schemes are presented. The real-time PI correction is mainly allocated to the G1 hydropower unit and the G2 fast thermal power unit, while the correction for the G4 slow thermal power unit is relatively small due to its lower ramp rate and higher regulation cost.

[0061] The closed-loop feedback controller is a PI controller, and the parameters of the PI controller are proportional coefficients. =0.5, the integral coefficient is taken as =0.05, and includes anti-integral saturation measures such as ACE dead zone and integral limit; the correction amount is allocated among AGC units according to the product of priority coefficient and adjustable margin, and the allocation formula is: ; Taking the power surplus at t=9:16 as an example: the total correction value of the PI controller output. =5MW (needs downward adjustment), priority factor for G1 hydropower units =1.2, adjustable margin (downward). =15MW, G4 thermal power unit =0.8, adjustable margin =8MW. Therefore: G1 allocation =1.2×20 / (1.2×20+1.0×15+0.8×8)×5=24 / (24+15+6.4)×5=2.64=1.2×20 / (1.2×20+1.0×15+0.8×8)×5=24 / (24+15+6.4)×5=2.64MW; G2 allocation =15 / (24+15+6.4)×5=1.65=15 / (24+15+6.4)×5=1.65MW; G4 Allocation =6.4 / (24+15+6.4)×5=0.71=6.4 / (24+15+6.4)×5=0.71MW.

[0062] in: The PI correction amount allocated to the i-th AGC unit at time t is expressed in MW. The priority coefficient of the i-th unit (the priority coefficient of hydropower units is greater than that of thermal power units) is used to reflect the regulation priority of different units. The priority coefficient of hydropower units is set larger so that they can undertake more regulation tasks in the PI correction allocation, while the priority coefficient of thermal power units is relatively smaller. The adjustable capacity margin of the i-th unit at time t is expressed in MW. It refers to the remaining capacity space that the unit can adjust upward or downward. Specifically, the upper or lower margin is taken according to the direction of power surplus or deficit. Let j be the priority coefficient of the j-th AGC unit; The adjustable capacity margin of the j-th AGC unit at time t, in MW; This represents the total correction output of the PI controller at time t, in MW. j is the index of the AGC unit number, j=1,2,......,N; i is the index of the AGC unit number, i=1,2,......,N; N represents the total number of AGC units, in units.

[0063] This is a weighted sum of the products of priority coefficients and adjustable margins for all AGC units, used to normalize the allocation ratio.

[0064] When there is a power surplus, the ratio of the downward adjustment space of new energy to the PI correction amount determines the proportion of new energy to be borne, and the remaining adjustment amount is allocated to each AGC unit according to the weighted margin.

[0065] Among them, the priority coefficient of hydropower units is greater than that of thermal power units; when there is a power surplus, the ratio of the downward adjustment space of new energy to the PI correction amount determines the proportion of new energy to be borne, and the remaining adjustment amount is allocated to each AGC unit according to the weighted margin.

[0066] Through simulation, Figure 5 The minute-by-minute ACE variation curves for the two schemes are given, and the dashed line marking the CPS2 assessment limit (7.8MW) is shown in the figure.

[0067] S6. Set up a comparison scheme to verify the collaborative scheduling scheme; Two comparison schemes with the same initial state are set up to verify the collaborative scheduling scheme and to evaluate the control performance of the collaborative scheduling method in improving CPS performance indicators in scenarios with high penetration of new energy.

[0068] Set up a comparison scheme to verify the collaborative scheduling scheme, including: Option 1 involves CPS regulation via AGC units only, with all renewable energy sources connected to the grid (no wind or solar curtailment). Option 2: By coordinating AGC units with new energy sources for CPS regulation, it is permissible to actively reduce the grid-connected output of new energy sources (in cases of power curtailment). The two schemes were compared in simulation under the same initial conditions: At the first assessment point (P1, 9:00-9:15), both schemes had their new energy sources fully connected to the grid to ensure that the initial state was consistent. The simulation period lasted 60 minutes (9:00-10:00), including four 15-minute assessment points (P1, P2, P3, P4).

[0069] The control performance of the collaborative dispatching method under CPS assessment is evaluated by comparing the mean of CPS1, the pass rate of CPS2, the AGC regulation cost, the curtailment rate, and the frequency deviation.

[0070] A modified IEEE 14-node system was used to simulate a provincial control area. In the system, G1 is a hydropower AGC unit connected to node 1, with a rated capacity of 350MW, initial output of 120MW, ramp rate of 0–100MW / min, cost factor of 0.25 yuan / kWh, and unit regulating power of 50MW / Hz; G2 is a fast-speed thermal power AGC unit connected to node 2, with a rated capacity of 200MW, initial output of 120MW, ramp rate of 0–60MW / min, cost factor of 0.25 yuan / kWh, and unit regulating power of 50MW / Hz; G3 is a non-AGC thermal power unit connected to node 6, with a rated capacity of 200MW; and G4 is a slow-speed thermal power AGC unit connected to node 8, with a rated capacity of 240MW, initial output of 168MW, ramp rate of 0–10MW / min, cost factor of 0.50 yuan / kWh, and unit regulating power of 26.7MW / Hz. Wind farm access node 3 and photovoltaic power station access node 6 are both 200MW-class renewable energy sources, with a total renewable energy penetration rate of approximately 30%. In the interconnection parameters with the external grid, the frequency deviation coefficient B = 200.625MW / 0.1Hz, and the planned power of the tie line... =200MW.

[0071] Under these system conditions, two comparison schemes are set up: Option 1 is an AGC-only regulation scheme, in which new energy sources are always connected to the grid at full available output, without any active wind or solar curtailment. Option 2 is a coordinated adjustment scheme between AGC and new energy sources, which allows for proactive adjustment of the grid-connected power output of new energy sources through EP optimization and PI correction.

[0072] Both schemes adopted full grid connection at the first assessment point (P1, 9:00-9:15) to ensure consistent initial scenarios. The simulation period was 60 minutes from 9:00 to 10:00, including four 15-minute assessment points. The effectiveness of the cooperative scheduling method was verified by comparing the average CPS1, CPS2 pass rate, AGC regulation cost, curtailment rate, and maximum frequency deviation.

[0073] For detailed simulation results, see Figures 4 to 10 : (a) Verification of control variables Figure 4 As shown, for both schemes, the KCPS1 of the first assessment point P1 is 75.6% (Level D), the average ACE of 15 minutes is 5.6MW, and the CPS2 is qualified. The initial conditions are exactly the same, and the control variables are valid.

[0074] (II) Comparison of CPS1 Indicators The four assessment points in Scheme 1 are as follows: P1 is Grade D (KCPS1=75.6%), P2 is Grade C (KCPS1=128.2%, CPS2 is unqualified), P3 is Grade A (KCPS1=262.3%), and P4 is Grade B (KCPS1=190.7%). The average KCPS1 score over 60 minutes is 164.2%.

[0075] The four assessment points in Scheme 2 are as follows: P1 is Grade D (KCPS1=75.6%), P2 is Grade A (KCPS1=239.7%), P3 is Grade A (KCPS1=228.0%), and P4 is Grade A (KCPS1=217.6%). The average KCPS1 score over 60 minutes is 190.2%.

[0076] (III) Comparison of ACE and CPS2 Figure 5 As shown, in Scheme 1, the ACE fluctuation at assessment point P2 is relatively large, with the absolute value of the 15-minute average ACE reaching 13.9MW, exceeding the limit of 7.8MW, resulting in a failure of CPS2. In Scheme 2, the ACE curve is more convergent overall, with the 15-minute average ACE from P2 to P4 all far below the limit, achieving a 100% pass rate for CPS2.

[0077] (iv) Frequency deviation comparison Figure 6 As shown, the system frequency of both schemes fluctuates around the rated value of 50Hz. The maximum absolute value of the frequency deviation of Scheme 1 is 0.0921Hz, and the maximum absolute value of the frequency deviation of Scheme 2 is 0.1092Hz, both of which are controlled within the allowable range of ±0.2Hz.

[0078] (v) Comparison of tie line power Figure 7 As shown, the planned power of the tie line is adjusted in steps every 15 minutes around 200MW. The actual power of the tie line in Scheme 1 fluctuates significantly around the planned value at the P2 assessment point, while the power of the tie line in Scheme 2 more closely follows the planned value curve, with a significantly smaller deviation range. The power deviation of the tie line in both schemes does not exceed the ±20MW tie line safety limit.

[0079] (vi) Comparison of PI adjustment allocation Figure 8 As shown, in Scheme 2, the G1 hydropower unit undertakes more rapid correction tasks, demonstrating the role of hydropower priority coefficient and large-capacity rapid ramp-up capability; at the same time, the new energy side participates in downward adjustment when there is a power surplus, reducing the pressure of relying solely on the reverse adjustment of slow-speed thermal power units.

[0080] (vii) Comparison of Costs and Curtailment Rate Figure 9As shown, the AGC adjustment cost and total cost of Option 1 (without active power curtailment) are approximately RMB 5,178; the AGC adjustment cost of Option 2 is approximately RMB 5,038, which is slightly lower than Option 1, but at the same time, it incurs a power curtailment penalty cost of approximately RMB 1,091, and the total cost is approximately RMB 6,129, which is higher than Option 1.

[0081] Figure 10 As shown, Scheme 2 involves active power curtailment during the P2-P4 period, with the curtailment mainly concentrated in the P2 phase and a few subsequent periods of power surplus. The total curtailed power in Scheme 2 is approximately 3638 kWh, with a cumulative curtailment rate of 1.49%. Calculated using a curtailment penalty coefficient of 0.30 yuan / kWh, the cumulative curtailment penalty cost is approximately 1091 yuan.

[0082] The simulation results demonstrate that, under conditions of high renewable energy penetration, treating renewable energy grid-connected power as a dispatchable resource, and employing a two-stage dispatch framework combining intraday EP optimization with real-time PI correction, can significantly improve CPS performance indicators at the cost of approximately 1.49% active curtailment. Scheme Two upgrades the P2 assessment point from level C (CPS2 failure) to level A, increases the CPS2 pass rate from 75% to 100%, and raises the 60-minute KCPS1 average from 164.2% to 190.2%. Curtailment is no longer a passive loss but rather an actively available virtual frequency regulation capacity, expanding the equivalent adjustable resource boundary of the provincial power grid.

[0083] Example 2:

[0084] In this embodiment, the multi-source collaborative optimization scheduling method of the CPS standard is the same as that in Embodiment 1. Based on this, the modified IEEE 39-node system (New England system) simulates a provincial control area to verify the applicability and effectiveness of the method of the present invention under different system scales and different new energy penetration rates.

[0085] S1. Construct an AGC unit output model that includes decision variables for grid-connected renewable energy power. A modified IEEE 39-node system was used to simulate a provincial control area. In this system, five hydropower units (G1-G5) were selected to form a hydropower AGC (Automatic Generation Control) unit group, and four thermal power units (G6-G9) were selected to form a thermal power AGC unit group. Wind farms were connected to nodes 16 and 21, and photovoltaic power plants were connected to nodes 24 and 27. The total installed capacity of new energy sources was 1800MW, and the penetration rate of new energy sources was approximately 35%.

[0086] Power increase / decrease indicator variable and climbing rate For thermal power and hydropower units, the actual grid-connected power output of new energy sources is used as the decision variable. These are the decision variables for new energy sources. Specifically, the ramp rate for hydropower units is 0–120 MW / min, with a minimum sustained ramp time of 1 minute; the ramp rate for thermal power units is 0–50 MW / min or 0–8 MW / min, with a minimum sustained ramp time of 1–5 minutes. All AGC units participate in both primary and secondary frequency regulation.

[0087] S2. Construct a collaborative scheduling optimization model with the goal of minimizing the combined costs of AGC auxiliary adjustment and power curtailment penalty. The simulation period was from 2:00 PM to 3:00 PM, a total of 60 minutes, with an optimization cycle of 15 minutes and a scheduling period of 1 minute within each cycle. The total system load fluctuated within the range of 4000–4300 MW. The available wind and solar power outputs were generated based on measured data, with wind power output roughly in the range of 350–520 MW and solar power output roughly in the range of 580–720 MW due to ample afternoon sunlight.

[0088] The objective function is the same as in Example 1: ; in: The adjustment cost for the AGC unit is in yuan. The penalty fee for abandoning renewable energy is in yuan. =0.6; =0.4; =25,000 yuan, which is the normalized benchmark value for AGC adjustment costs; =5000 yuan, which is the normalized benchmark value for the power abandonment penalty fee.

[0089] The expression for AGC auxiliary adjustment cost is: ; The cost coefficient for the i-th AGC unit is 0.20 yuan / kWh for hydropower units, 0.28 yuan / kWh for fast thermal power units, and 0.55 yuan / kWh for slow thermal power units. The planned power output of the i-th AGC unit during time period t, in MW; The actual output of the i-th AGC unit during time period t-1 is expressed in MW. The duration of the scheduling period is set to 1 / 60h (i.e., 1 minute). Taking the G1 hydropower unit at 14:01 as an example: =420MW, =415MW, =0.20 yuan / kWh, then the adjustment cost is: 0.20 × |420−415 | × 1 / 60 = 0.0167 yuan.

[0090] The expression for the power curtailment penalty fee is: ; The curtailment penalty factor is set at 0.30 yuan / kWh; The combined wind and solar power output for time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW.

[0091] S3. Establish a constraint system that includes system power balance, frequency deviation, CPS1 and CPS2 indicators; System control area frequency deviation coefficient =4300MW / Hz, frequency deviation coefficient of interconnected power grid =2100MW / Hz, total system unit regulating power =6400MW / Hz. Planned power of the tie line. =400MW, maximum allowable frequency deviation =0.2Hz.

[0092] The power balance equation is: ; in: This represents the total system load. This is due to network loss in the system.

[0093] The frequency deviation can be solved directly using the quasi-steady-state closed-loop formula: ; It is the power imbalance quantity, measured in MW, and is determined by the difference between the system's power generation output and the load and network losses. The system unit adjustable power is expressed in MW / Hz. The unit regulating power of the interconnected power grid is expressed in MW / Hz.

[0094] The CPS1 constraint is that the CPS1 value shall not be lower than 100% during the assessment period. The CPS assessment adopts a four-level evaluation system of A / B / C / D, the same as in Example 1.

[0095] S4. Solve the intraday open-loop optimization problem using an adaptive evolutionary programming algorithm. Each AGC unit's power increase / decrease indicator variable, ramp rate, and grid-connected renewable energy output constitute a single coded unit. With 9 AGC units, the coding dimension is (2×9+1)×15=285 dimensions.

[0096] Population size is set to 100, maximum number of generations to 120, random competition parameter q=20, and penalty coefficient M=10. 6 .

[0097] The adaptive variable-length asynchronous expression is: ; in: ; ; n is the dimension of the decision variables; Let be the variable length of the j-th dimension decision variable for the i-th individual; This is the global scaling factor for variation. These are random numbers distributed according to a standard normal distribution. The j-th dimension of the i-th individual is a random number from a standard normal distribution. This is the scaling factor for individual variation; The fitness function is: ; The fitness function value; The objective function value; This is the limit exceeded value for the k-th constraint. This is the penalty coefficient.

[0098] S5. Introduce a closed-loop feedback controller for real-time closed-loop feedback correction. The scheduling period is 1 minute, and the adjustment correction is calculated based on the measured ACE at each minute. The PI controller parameters are: proportional coefficient. =0.6, integral coefficient =0.08, with anti-integral saturation measures including ACE dead zone (±2MW) and integral limit (±30MW).

[0099] The correction amount is allocated among AGC units according to the product of the priority coefficient and the adjustable margin, and the allocation formula is as follows:

[0100] Priority coefficient for hydropower units =1.2, priority coefficient for thermal power units =0.8. When there is a power surplus, the ratio of the downward adjustment space of new energy to the PI correction amount determines the proportion of new energy to be borne, and the remaining adjustment amount is allocated to each AGC unit according to the weighted margin.

[0101] S6. Set up a comparison scheme to verify the collaborative scheduling scheme. This embodiment uses a modified IEEE 39-node system to simulate a provincial control area and sets up two comparative schemes for simulation verification.

[0102] In the system, G1-G5 are hydropower units, and G6-G9 are thermal power units. Wind farms are connected to nodes 16 (rated capacity 800MW) and 21 (rated capacity 400MW), and photovoltaic power plants are connected to nodes 24 (rated capacity 300MW) and 27 (rated capacity 300MW). The total renewable energy penetration rate is approximately 35%. In the interconnection parameters with the external grid, the frequency deviation coefficient B = 215MW / 0.1Hz, and the planned power of the tie line... =400MW.

[0103] Under this system condition, two comparative schemes were set up. Scheme 1 is an AGC-only regulation scheme, where renewable energy is always fully connected to the grid; Scheme 2 is a coordinated regulation scheme of AGC and renewable energy, allowing active adjustment of renewable energy grid-connected output. Both schemes adopted full grid connection at the first assessment point (P1, 14:00-14:15) to ensure consistent initial scenario. The simulation period was 60 minutes from 14:00 to 15:00, including four 15-minute assessment points. By comparing the CPS1 mean, CPS2 pass rate, AGC regulation cost, curtailment rate, and maximum frequency deviation, the control performance of the coordinated dispatch method under the CPS assessment was evaluated.

[0104] The simulation results are as follows: (I) Comparison of CPS1 Indicators The four assessment points in Scheme 1 are as follows: P1 is grade C (KCPS1=95.6%), P2 is grade D (KCPS1=82.3%, CPS2 is unqualified), P3 is grade C (KCPS1=135.7%), and P4 is grade B (KCPS1=165.2%). The average KCPS1 score over 60 minutes is 119.7%.

[0105] The four assessment points in Scheme 2 are as follows: P1 is grade C (KCPS1=95.6%), P2 is grade B (KCPS1=175.4%), P3 is grade A (KCPS1=215.3%), and P4 is grade A (KCPS1=208.7%). The average KCPS1 score over 60 minutes is 173.8%.

[0106] The average CPS1 of Option 2 is about 45% higher than that of Option 1.

[0107] (II) Comparison of CPS2 pass rates In Scheme 1, the absolute value of the ACE 15-minute average at the P2 assessment point reached 14.2MW, exceeding the limit of 9.6MW, and the CPS2 pass rate was 75%. In Scheme 2, the ACE curve was more convergent overall, and the ACE 15-minute average at P2 to P4 was far below the limit, with the CPS2 pass rate reaching 100%.

[0108] (III) Frequency Deviation Comparison The system frequencies of both schemes fluctuate around the rated value of 50Hz. The maximum absolute value of the frequency deviation for Scheme 1 is 0.105Hz, and the maximum absolute value of the frequency deviation for Scheme 2 is 0.088Hz, both controlled within the allowable range of ±0.2Hz.

[0109] (iv) Comparison of tie line power The actual power of the tie line in Scheme 1 fluctuated significantly around the planned value at the P2 assessment point, while the tie line power in Scheme 2 more closely followed the planned value curve, with a significantly smaller deviation range. The tie line power deviations of both schemes did not exceed the ±30MW tie line safety limit.

[0110] (v) Comparison of Costs and Curtailment Rate Option 1 has an AGC adjustment cost of 25,360 yuan. Option 2 has an AGC adjustment cost of 23,820 yuan, which is about 6.1% lower than Option 1, but incurs a curtailment penalty of about 4,820 yuan, bringing the total cost to 28,640 yuan, which is higher than Option 1. Option 2 has a total curtailed electricity volume of 16,070 kWh, with a cumulative curtailment rate of approximately 1.82%.

[0111] The simulation results above verify the effectiveness and versatility of the present invention under different system scales (39-node system) and different renewable energy penetration rates (35%): at the cost of approximately 1.82% active curtailment and a certain increase in total cost, the average CPS1 rate is increased from 119.7% to 173.8%, the CPS2 pass rate is increased from 75% to 100%, and the P2 assessment point is improved from level D (CPS2 failure) to level B, achieving a significant improvement in CPS assessment indicators.

Claims

1. A multi-source collaborative optimization scheduling method based on the CPS standard, characterized in that, The system is executed on a rolling basis with a preset optimization cycle and scheduling period. The actual grid-connected power of new energy sources and the ramp rate and power increase / decrease indicators of AGC units are used as decision variables. Under the condition of meeting the CPS standard, it is used for the coordinated optimization scheduling of new energy sources and AGC units. The specific steps are as follows: S1. Construct an AGC unit output model that includes decision variables for grid-connected renewable energy power. S2. Construct a collaborative scheduling optimization model with the goal of minimizing the combined costs of AGC auxiliary adjustment and power curtailment penalty. S3. Establish a constraint system that includes system power balance, frequency deviation, CPS1 and CPS2 indicators; S4. Use an adaptive evolutionary programming algorithm to solve the intraday open-loop optimization problem; S5. Introduce a closed-loop feedback controller for real-time closed-loop feedback correction: within a predetermined time, adjust the correction amount according to the measured ACE, distribute the correction amount to each AGC unit according to the adjustable capacity ratio of hydropower priority, and prioritize reducing the grid-connected output of new energy for new energy-thermal power coordinated regulation when there is a power surplus. S6. Set up a comparison scheme to verify the collaborative scheduling scheme, which is used to evaluate the control performance of the collaborative scheduling method in improving CPS performance indicators in scenarios with high penetration of new energy.

2. The multi-source collaborative optimization scheduling method of the CPS standard according to claim 1, characterized in that, In S1, the AGC unit output model that includes new energy grid-connected power decision variables is as follows: S11. Object selection: AGC unit group consisting of hydropower units and thermal power units was selected; S12, Decision Variables: (1) Power increase / decrease indicator variables and climbing rate For decision variables of thermal power and hydropower units; in: ； (2) Power output from actual grid connection of new energy sources For new energy decision-making variables; in: ； The available power output for combined wind and solar power is expressed in MW. S13, Unit Characteristic Parameters: The hydropower unit's climbing speed is 0~100MW / min, and the minimum continuous climbing time is 1min; The climbing rate of thermal power units is 0~60MW / min or 0~10MW / min, and the minimum continuous climbing time is 1~4min; Each AGC unit participates in primary and secondary frequency regulation.

3. The multi-source collaborative optimization scheduling method of the CPS standard according to claim 2, characterized in that, In S12, the new energy decision variables transform the curtailment behavior from a passive loss to an active control measure. When the system has a power surplus or insufficient AGC regulation capacity, the actual grid-connected output of new energy sources is actively reduced. This is equivalent to virtual downward adjustment, which forms a coordinated control with the increase or decrease of output of hydropower units.

4. The multi-source collaborative optimization scheduling method of the CPS standard according to claim 1, characterized in that, In S2, the objective function of the cooperative scheduling optimization model is: ； in: The adjustment cost for AGC units is calculated as the product of the deviation of the planned output of each unit and the cost coefficient, in yuan. The penalty fee for abandoning renewable energy is calculated as the product of the abandoned amount and the penalty coefficient, and the unit is yuan. and As weight; This is the normalized baseline value for AGC unit adjustment costs, in yuan. This is the normalized baseline value for the fee for power curtailment, in yuan; This represents the objective function value.

5. A multi-source collaborative optimization scheduling method for CPS standard according to claim 1, characterized in that, In S2; The AGC auxiliary adjustment cost is expressed as: ； in: Here is the cost coefficient for the i-th AGC unit: 0.25 yuan / kWh for hydropower units, 0.25 yuan / kWh for fast thermal power units, and 0.50 yuan / kWh for slow thermal power units. The planned power output of the i-th AGC unit during time period t, in MW; The actual output of the i-th AGC unit during time period t-1 is expressed in MW. Let be the power increase / decrease indicator variable for the i-th AGC unit during time period t; The duration of the scheduling period is expressed in hours (h). The aforementioned power abandonment penalty fee is expressed as follows: ； in: The amount of renewable energy wasted during time period t is expressed in MWh. This is the power curtailment penalty factor, expressed in yuan / kWh; The combined wind and solar power output for time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW.

6. The multi-source collaborative optimization scheduling method of the CPS standard according to claim 1, characterized in that, In S3, the establishment of a constraint system including system power balance, frequency deviation, CPS1, and CPS2 indices includes: S31. The power balance equation states that the difference between the total power generation output of the system and the total load and network losses is equal to zero. The expression is: ； in: The output of the i-th AGC unit during time period t is expressed in MW. The actual grid-connected power output of new energy sources during time period t is expressed in MW. The total system load for time period t is expressed in MW. The system network loss during time period t is expressed in MW. S32. The frequency deviation is solved directly using the quasi-steady-state closed-loop formula for frequency, and the expression is: ； in: Frequency deviation, in Hz; This refers to the power imbalance, measured in MW. The system unit adjustable power is expressed in MW / Hz. The unit regulating power of the interconnected power grid is expressed in MW / Hz. The total system regulation power is expressed in MW / Hz. constraint This represents the upper limit of the allowable frequency deviation, in Hz. The constraint system for the S33, CPS1, and CPS2 indicators includes: The first constraint of the control performance standard, CPS1, is within the assessment period. That is, the CPS1 value is not less than 100%; CPS assessment uses a four-level evaluation system: A / B / C / D. It was grade A at the time; Furthermore, when the control performance standard item 2, CPS2, is qualified, it is grade B; Furthermore, a grade C is assigned when CPS2 fails to meet the standard. It was classified as Grade D at the time; in: The value is CPS1. CPS2 passing means that the absolute value of the 15-minute average of the regional control deviation ACE does not exceed the CPS2 assessment limit during the assessment period. CPS2 failure refers to the absolute value of the 15-minute average of the area control deviation (ACE) exceeding the CPS2 assessment limit during the assessment period.

7. A multi-source collaborative optimization scheduling method for CPS standard according to claim 1, characterized in that, In S4, the coding method using the adaptive evolutionary programming algorithm is as follows: the power increase / decrease indicator variable of each AGC unit, the ramp rate, and the grid-connected output of new energy sources are used to jointly constitute the coding unit, and the coding dimension is: (2 × number of AGC units + 1) × 15 time periods; An adaptive Gaussian mutation and a random competition parameter q are used to search for the optimal scheduling scheme. The constraint handling adopts a combination of penalty function method and direct correction method, where q is the number of competing individuals in the random competition selection.

8. A multi-source collaborative optimization scheduling method for CPS standard according to claim 7, characterized in that, In S4, the encoding method of the adaptive evolutionary programming algorithm is expressed as follows: ； ； in: Let $\frac{i}{i}$ be the power increase / decrease indicator variable for the $i$-th AGC unit during time period $t$, with values ​​of ${1}$, $0$, and $-1$. Let be the ramp rate of the i-th AGC unit during time period t, in MW / min; The actual grid-connected power output of new energy sources during time period t is expressed in MW. This represents the total number of AGC units, in units. This represents the total number of scheduling periods; Let be the adjustable capacity upper limit direction indicator variable for the i-th AGC unit during time period t; Let be the adjustable capacity lower limit direction indicator variable of the i-th AGC unit during time period t; The maximum adjustable capacity of the i-th AGC unit during time period t is expressed in MW. The combined wind and solar power output for time period t is expressed in MW. Encode decision variables into vectors; This constitutes a 105-dimensional search space; Population size 60, maximum number of generations 80, random competition parameter q=15, penalty coefficient ; Adaptive variable asynchronous length, the expression is: ； in: Let be the variable length of the j-th dimension decision variable for the i-th individual; Let j be the current value of the j-th dimension decision variable for the i-th individual; Let be the upper limit of the range of values ​​for the j-th dimension decision variable; This represents the lower bound of the range of values ​​for the j-th dimension decision variable; The fitness function is: ； ； in: The fitness function value; Let i be the objective function value corresponding to the i-th individual; This is the penalty term for the k-th constraint. This represents the total number of constraints. Let $\frac{k}{k}$ be the absolute value of the limit exceeded by the k-th constraint. Let be the penalty coefficient for the k-th constraint. This represents the upper limit value of the k-th constraint. This is the lower limit value of the k-th constraint. This represents the actual value of the k-th constraint under the current solution; The chosen strategy introduces a concentration regulation mechanism.

9. A multi-source collaborative optimization scheduling method for CPS standard according to claim 1, characterized in that, In S5, the closed-loop feedback controller is a PI controller, which has anti-integral saturation measures such as ACE dead zone and integral limiting. The correction amount is distributed among AGC units according to the product of priority coefficient and adjustable margin. The priority coefficient of hydropower units is greater than that of thermal power units. When there is a power surplus, the ratio of the adjustment space of new energy to the PI correction amount determines the proportion of new energy to be borne, and the remaining adjustment amount is distributed to each AGC unit according to the weighted margin.

10. A multi-source collaborative optimization scheduling method for CPS standard according to claim 1, characterized in that, In S6, the setting of a comparison scheme to verify the cooperative scheduling scheme includes: Option 1 involves CPS regulation via AGC units only, with full grid connection of renewable energy. Option 2: By coordinating AGC units with new energy sources for CPS regulation, the grid-connected output of new energy sources can be actively reduced; The two schemes were compared in simulation under the same initial conditions. By comparing the mean of CPS1, the pass rate of CPS2, the AGC regulation cost, the curtailment rate and the frequency deviation, the control performance of the cooperative scheduling method under the CPS test was evaluated.

Images