Flexible load-oriented day-ahead intra-day inter-group secondary cooperative regulation and control method

Abstract

The invention provides a flexible load-oriented day-ahead intra-day inter-group secondary cooperative regulation and control method, and belongs to the technical field of power flexible load scheduling control. The problem that active participation of user side flexible resources in power grid regulation and control is uncertain is solved; comprising the following steps: S10, obtaining related characteristic parameters and historical electric quantity data of user side flexible resources; s20, establishing a flexible resource evaluation index, dividing flexible resources into different flexible load clusters through an improved K-means clustering method, and forming a layered regulation and control framework of flexible loads after aggregation; s30, in combination with a day-ahead load prediction method, calculating the maximum adjustable capacity of user side flexible resources, and in combination with a parallel Markov chain model, preliminarily making a flexible load day-ahead scheduling plan; s40, tracking the load power change of the user side in the day in real time, and correcting the adjustment capacity in time based on a flood filling algorithm to realize secondary cooperative regulation and control of the flexible load cluster; the method is applied to a power system.

Description

Technical Field

[0001] The present invention provides a day-ahead intra-group secondary coordinated control method for flexible loads, belonging to the technical field of power flexible load dispatching and control. Background Art

[0002] With the continuous expansion of flexible resources such as temperature control equipment, electric vehicles and energy storage equipment on the user side, the traditional distribution network has gradually transformed into an active distribution network that can actively participate in grid regulation. At the same time, due to the decentralized access of flexible loads, the flexible loads of the distribution network have the characteristics of a large number of users, limited individual capacity, and low regulation reliability. It is not only difficult to fully tap the maximum regulation potential of flexible loads on the user side, but also brings great challenges to the power dispatching and control on the load side, which has become an urgent problem to be solved in order to further improve the safe operation and stability of the power grid. Summary of the invention

[0003] In order to solve the uncertainty problem of user-side flexible resources actively participating in power grid regulation, the present invention proposes a day-ahead intra-group secondary collaborative regulation method for flexible loads.

[0004] The technical solution adopted by the present invention is: a method for secondary coordinated control of flexible loads within a day before the day, comprising the following steps: S10: Obtain relevant characteristic parameters and historical power data of user-side flexible resources; S20: Establish flexible resource evaluation indicators, and divide flexible resources into different flexible load clusters through improved K-means clustering method, and form a hierarchical control architecture of flexible loads after aggregation; S30: Combined with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resources is calculated, and combined with the parallel Markov chain model, a preliminary day-ahead dispatch plan for the flexible load is formulated; S40: Real-time tracking of daily user-side load power changes, and timely correction of the regulation capacity based on the flood filling algorithm to achieve secondary coordinated regulation of the flexible load cluster.

[0005] Furthermore, in step S10, the user-side flexible resources include temperature control equipment, electric vehicles, and distributed energy storage.

[0006] Furthermore, the flexible resource evaluation index includes a load characteristic index, a regulation characteristic index and a response characteristic index.

[0007] Furthermore, the calculation formula of the load characteristic index is as follows: f 1= P av / P max ; Where:f 1 is the load characteristic index, P av is the daily average load, P max is the maximum daily load; The calculation formula of the regulation characteristic index is as follows: ; Where: f 2 is the regulation characteristic index, t res is the load response time, Δ t is the time scale, S res It is the regulatory capacity that can participate in the response after the indicator is issued; The calculation formula of the response characteristic index is as follows: ; Where: f 3 is the response characteristic index, t lim is the upper limit of load response time, v res is the daily fluctuation rate of the flexible load.

[0008] Furthermore, the improved K-means clustering method is improved by using the entropy weight method, and the feature weight values ​​corresponding to the flexible resource evaluation index are calculated by the entropy weight method.

[0009] Furthermore, the steps of dividing the flexible resources into different flexible load clusters by improving the K-means clustering method are as follows: 1) Select cluster centers based on hierarchical clustering results C I , and A The cluster centers are A Input, copy B Distribute to B Computers; 2) Daily load sample set P A×B All samples in P j =[ p j1 , p j2 , p j3 ,…, p jB ] is distributed to each computer node, and then calculated to the I Cluster Centers C I The weighted Euclidean distance of ; 3)A Inputs are dispatched to c Distributed computing cores are used to calculate the minimum weighted Euclidean distance for each sample. ， Divide it into cluster centers C I The cluster in which it is located; 4) For each cluster, update the cluster center C I ; 5) Repeat steps 3) and 4) until the new cluster center becomes stable or the loss function converges.

[0010] Furthermore, the hierarchical control architecture of the flexible load formed after aggregation in step S20 includes an aggregator layer, a load proxy layer and a user device layer. The user device layer uploads load parameters and operation status information to the load proxy through a smart socket or a smart switch, and performs dynamic timing control of the controllable load in the form of direct call according to the control indicators allocated and issued by the load proxy to achieve the demand side response target; The load proxy layer monitors the load changes within the flexible load cluster in real-time time scale, continuously updates the load status and response potential parameters, and performs secondary distribution of target power among the flexible load clusters according to the control target and the updated response potential, thereby controlling the user equipment. The aggregator layer calculates the maximum response potential range of various flexible load clusters and feeds it back to the dispatch center, which further obtains the adjustable target power. After evaluation, it makes an initial allocation and hands it over to the corresponding load agents for management. At the same time, it relies on a two-way communication mechanism to strengthen coordination and control.

[0011] Furthermore, in step S30, in combination with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resource is calculated by performing aggregate modeling on multiple flexible loads, wherein the aggregate modeling expression of multiple flexible loads is as follows: ; Where: for t Always control the internal temperature of the equipment. for t+ 1. External ambient temperature at time 1, R is the equivalent thermal resistance, C is the equivalent heat capacity, Δ t is the time scale, λ The electric heating and cooling power of the temperature control equipment The conversion efficiency of for t The state of charge of the electric vehicle at all times, for t The rated charging power of the electric vehicle at the moment, EEV is the power battery capacity of electric vehicles, η For the charging efficiency of electric vehicles, for t The state of charge of the energy storage device at all times, for t The rated charging and discharging power of the energy storage device at all times, E ES is the battery capacity of the energy storage device, α is the charging and discharging efficiency index of the energy storage device, and its expression is as follows: ; Where: η c and η d are the charging and discharging efficiency of the energy storage device, respectively; The constraints for the aggregate modeling of multivariate flexible loads are as follows: ; Where: δ is the temperature offset acceptable to the user, T set is the set temperature value; , They are t The upper and lower limits of the state of charge of the electric vehicle at all times; , They are t The upper and lower limits of the state of charge of the energy storage device at all times; The maximum adjustable capacity of the user-side flexible resource is calculated as follows: ; Where: and are the calculated values ​​of the maximum adjustable increase and decrease capacity of the load aggregator, respectively; N TC , N EV and N ES are the corresponding load agent numbers of temperature control equipment, electric vehicles, and energy storage equipment; λ TC , λ EV and λ ES are the capacity increase state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. u TC , u EV and u ESare the capacity reduction state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. For the i Temperature control equipment in t The electric heating and cooling power at all times, For the i Electric vehicles in t Rated charging power at the moment, For the i Energy storage devices in t Rated charge and discharge power at the moment.

[0012] Furthermore, in step S30, the steps of preliminarily formulating a flexible load day-ahead dispatching plan in combination with the parallel Markov chain model are as follows: 1) Assumptions Q For flexible load operation state, Q 1=1 indicates running or charging status, Q 2=0 means closed state, Q 3=-1 indicates discharge state; 2) If Q 1. Q 2. Q 3 Three states are randomly converted, and the state conversion probability matrix corresponding to different flexible loads is obtained according to the Markov chain model; 3) According to the maximum likelihood estimation, the state transition probability is replaced by the state transition frequency of different flexible loads at a certain time scale to obtain the different state transition probabilities of the flexible load; 4) Based on the day-ahead forecast results of flexible loads and the calculation method of the maximum adjustable capacity, the upper and lower limits of the load aggregator’s adjustable capacity are further calculated to formulate a day-ahead dispatch plan for flexible loads.

[0013] Furthermore, in step S40, the metric range is prioritized from two aspects of spatial distance and time interval based on the flood filling algorithm, and the peak and valley capacity is allocated first, and then diffused to nearby connected areas until all scheduling capacities are allocated.

[0014] The beneficial effects of the present invention compared with the prior art are as follows: The flexible resource evaluation index provided by the present invention quantifies the characteristics of flexible load resources from three aspects: load characteristics, regulation characteristics and response characteristics, and is divided into different flexible load clusters by improving the K-means clustering method. It can provide strong support for cluster coordination control of aggregators, load agents and user equipment in combination with the hierarchical control architecture of flexible loads, and realize the participation of flexible load resources in power grid peak regulation applications after aggregation control. The method for calculating the maximum adjustable capacity of user-side flexible resources provided by the present invention realizes the unification of the calculation methods of the response capacity of different types of flexible loads by aggregate modeling of multiple flexible loads such as temperature control equipment, electric vehicles, and energy storage equipment, and formulates a flexible load day-ahead scheduling plan based on a Markov chain model, which is conducive to improving the effective utilization of flexible resources on the load side and enhancing the ability of the distribution network to cope with load shocks. The secondary coordinated control method of the flexible load cluster based on the flood filling algorithm provided by the present invention can track the intraday load power changes in real time, correct the adjustment capacity in time, give priority to the allocation of peak-valley capacity differences, and then diffuse them to nearby connected capacity areas until all scheduling capacities are allocated, thereby minimizing the adjustable capacity difference, which is beneficial to improving the safety and flexibility of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further described below in conjunction with the accompanying drawings: Figure 1 is a flow chart of the method of the present invention; Figure 2 It is a flexible load resource cluster division result diagram obtained based on the improved K-means clustering method given in the implementation of the present invention; Figure 3 A diagram of a hierarchical control architecture of a flexible load in an embodiment of the present invention; Figure 4 It is a day-ahead load forecast diagram of three types of flexible loads, namely, temperature control equipment, electric vehicle, and energy storage equipment, in an embodiment of the present invention. In the diagram, TC represents temperature control equipment, EV represents electric vehicle, and ES represents energy storage equipment; Figure 5 This is a diagram showing the initial allocation of the regulating capacity and the timing control result of the flexible load facing the day-ahead scheduling plan in an embodiment of the present invention; Figure 6 It is a diagram showing the secondary distribution of the flexible load regulation capacity and the intra-day timing control result in an embodiment of the present invention; Figure 7 This is a diagram of capacity adjustment deviation for each time period of a single day taking into account the influence of external uncertain factors in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] like Figure 1-7 As shown, the present invention proposes a day-ahead and intra-day secondary collaborative control method for flexible loads in view of the uncertainty characteristics of flexible resources on the user side. It is a cluster control method for flexible loads, which can not only divide different types of flexible loads into corresponding flexible load clusters, but also formulate a day-ahead scheduling plan for flexible loads in combination with load forecasting methods, and track intra-day power changes and make timely adjustments, and perform secondary collaborative control of flexible load clusters to achieve smoothing of power fluctuations during the entire control period and reduce the peak-shaving pressure brought by the demand side to the upper-level power grid.

[0017] The specific implementation steps of the method of the present invention include: S10: Obtain relevant characteristic parameters and historical power data of flexible resources such as user-side temperature control equipment, electric vehicles, and distributed energy storage; S20: Establish flexible resource evaluation indicators, and divide flexible resources into different flexible load clusters through improved K-means clustering method, and form a hierarchical control architecture of flexible loads after aggregation; S30: Combined with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resources is calculated, and combined with the parallel Markov chain model, a preliminary day-ahead dispatch plan for the flexible load is formulated; S40: Real-time tracking of daily user-side load power changes, and timely correction of the regulation capacity based on the flood filling algorithm to achieve secondary coordinated regulation of the flexible load cluster.

[0018] The flexible resource evaluation index includes load characteristic index, regulation characteristic index and response characteristic index. The calculation formula of load characteristic index is as follows: f 1= P av / P max ; Where: f 1 is the load characteristic index, P av is the daily average load, P max The maximum daily load.

[0019] The calculation formula of the regulation characteristic index is as follows: ; Where: f 2 is the regulation characteristic index, which is used to evaluate regulation-related characteristics; t res is the load response time, i.e. the time interval from when the command starts to change to when the load starts to respond; Δ t It is a time scale, or called a unit reference duration or a minimum adjustment cycle. From the perspective of load regulation characteristics, it specifically refers to a minimum complete cycle duration of the flexible load participating in the regulation response, such as every 15 minutes, 30 minutes or 1 hour, etc. It is generally determined according to the actual situation, and 1 hour is selected in this application example. S res It is the adjustment capacity that can participate in the response after the indicator is issued.

[0020] The calculation formula of the response characteristic index is as follows: ; Where:f 3 is the response characteristic index, t lim is the upper limit of load response time, v res is the daily fluctuation rate of the flexible load.

[0021] The improved K-means clustering method proposed in the present invention improves the distributed K-means algorithm by using the entropy weight method, and calculates the feature weight value corresponding to the flexible resource evaluation index by the entropy weight method. The calculation method is as follows: ; ; ; Where: m is the number of indicator features, n is the number of load samples; p ij Daily load sample set P A×B The i Row, No. j Column data; e ij is the sample information probability, E j is the entropy calculation result, w j is the weight value of the feature index.

[0022] The clustering steps of improving the distributed K-means algorithm using the entropy weight method are as follows: 1) Select cluster centers based on hierarchical clustering results C I , and A The cluster centers are A Input, copy B Distribute to B computers.

[0023] 2) P A×B All samples in P j =[ p j1 , p j2 , p j3 ,…, p jB ] is distributed to each computer node, and then calculated to the I Cluster Centers C I The weighted Euclidean distance is calculated as follows: ; Where: c il For the I The cluster center l Column coordinates, p il The daily load sample set i Line l Column data, w l Load sample set P A×B No. l The column data is calculated using the entropy weight method to obtain the characteristic weight value of the flexible resource evaluation index.

[0024] 3) A Inputs are dispatched to c Distributed computer cores and perform a separate calculation once to find the minimum weighted Euclidean distance corresponding to each sample ， Divide it into cluster centers C I The cluster in which the

[0025] 4) For each cluster, update the cluster center C I ,remember C I The number of samples in the cluster is , then the calculation formula is: ; 5) Repeat steps 3) and 4) until the new cluster center tends to be stable or the loss function tends to converge. The formula of the loss function is as follows: ; The physical meaning of using square in the loss function is the representation of the sum of Euclidean distances. There are two main considerations: First, it highlights the impact of the difference in Euclidean distances between different samples and cluster centers. Because the larger the difference after squaring, the greater the contribution to the loss function. During the training process, the model is guided to optimize in a more accurate and rapid direction, and comprehensively reflects the clustering division of the model on multiple samples. Second, from a mathematical point of view, the square function is smooth and differentiable, which makes it easy to calculate the gradient when using optimization algorithms such as gradient descent to optimize the loss function, ensuring feasibility and stability, and helping to more efficiently find the model parameters that minimize the loss function.

[0026] The flexible resources are divided into different flexible load clusters by the improved K-means clustering method mentioned above. The clustering results are as follows: Figure 2 shown.

[0027] like Figure 3As shown in the figure, the hierarchical control architecture of flexible loads includes an aggregator layer, a load agent layer, and a user device layer. The user device layer uploads information such as load parameters and operating status to the load agent through devices such as smart sockets or smart switches. According to the control indicators allocated and issued by the load agent, dynamic timing control of controllable loads is performed in the form of direct calls to achieve the demand-side response target.

[0028] The load proxy layer monitors the load changes within the flexible load cluster in real time, continuously updates parameters such as load status and response potential, and redistributes the target power among the flexible load clusters according to the control target and the updated response potential, thereby controlling the user equipment.

[0029] The aggregator layer calculates the maximum response potential range of various flexible load clusters and feeds it back to the dispatch center, which further obtains the adjustable target power. After evaluation, it makes an initial allocation and hands it over to the corresponding load agents for management. At the same time, it relies on a two-way communication mechanism to strengthen coordination and control.

[0030] In step S30, in combination with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resource is calculated by performing aggregate modeling on multiple flexible loads, wherein the aggregate modeling expression of multiple flexible loads is as follows: ; Where: for t Always control the internal temperature of the equipment. for t+ 1. External ambient temperature at time 1, R is the equivalent thermal resistance, C is the equivalent heat capacity, Δ t is the time scale, λ The electric heating and cooling power of the temperature control equipment The conversion efficiency of for t The state of charge of the electric vehicle at any time. for t The rated charging power of the electric vehicle at the moment, E EV is the power battery capacity of electric vehicles, η For the charging efficiency of electric vehicles, for t The state of charge of the energy storage device at all times, for t The rated charging and discharging power of the energy storage device at all times, E ES is the battery capacity of the energy storage device, α is the charging and discharging efficiency index of the energy storage device, and its expression is as follows: ; Where: η c and η d are the charging and discharging efficiency of the energy storage device, respectively.

[0031] The constraints for the aggregate modeling of multivariate flexible loads are as follows: ; Where: δ is the temperature offset acceptable to the user, T set is the set temperature value; considering the optimal power maintenance range of electric vehicle lithium-ion batteries, The upper and lower limits of are set to 90% and 20% respectively; considering the safety of energy storage equipment and the maximum power limit of charging and discharging, The upper and lower limits are set to 80% and 30% respectively.

[0032] The maximum adjustable capacity of the user-side flexible resource is calculated as follows: ; Where: and are the calculated values ​​of the maximum adjustable increase and decrease capacity of the load aggregator, respectively; N TC , N EV and N ES are the corresponding load agent numbers of temperature control equipment, electric vehicles, and energy storage equipment; λ TC , λ EV and λ ES are the capacity increase state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. u TC , u EV and u ES are the capacity reduction state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. For the i Temperature control equipment in t The electric heating (cooling) power at the moment, For the i Electric vehicles in t Rated charging power at the moment, For the i Energy storage devices in t Rated charge and discharge power at the moment.

[0033] Taking three load agents as objects, the day-ahead load forecasting is performed for three types of flexible loads, namely temperature control equipment, electric vehicles, and energy storage equipment. The total load of the day is formed by stacking and summing. The time granularity is set to 1h. The load forecasting results are as follows: Figure 4 shown.

[0034] The steps to initially formulate a flexible load day-ahead dispatch plan in combination with the parallel Markov chain model are as follows: 1) Assumptions Q For flexible load operation state, Q 1=1 indicates running or charging status, Q 2=0 means closed state, Q 3=-1 indicates the discharge state.

[0035] 2) If Q 1. Q 2. Q 3 The three states are randomly converted, then the state conversion probability matrix corresponding to the temperature control device, electric vehicle and energy storage device is A It can be expressed as: ; Where: A 1,1 To continue to run or charge the conversion probability, A 1,2 is the transition probability from the running or charging state to the off state, A 1,3 is the transition probability from the running or charging state to the discharging state; A 2,1 is the transition probability from the off state to the running or charging state, A 2,2 To keep the transition probability of the closed state, A 2,3 is the transition probability from the off state to the discharge state; A 3,1 is the transition probability from the discharge state to the operation or charging state, A 3,2 is the transition probability from the discharge state to the off state, A 3,3 The transition probability of continuing to maintain the discharge state.

[0036] 3) According to the maximum likelihood estimation, the state transition frequency of different flexible loads at a certain time scale is used to replace the probability, and the transition probability of different states of the flexible load is obtained, that is: ; Where: T N,Nis the state transition probability convergence matrix under maximum likelihood estimation conditions.

[0037] 4) Based on the day-ahead forecast results of flexible loads and the calculation method of the maximum adjustable capacity, the upper and lower limits of the load aggregator’s adjustable capacity are further calculated to formulate a day-ahead dispatch plan for flexible loads.

[0038] According to the above steps, the initial allocation and timing control results of the regulation capacity of different flexible loads facing the day-ahead dispatch plan can be obtained, such as Figure 5 shown.

[0039] The main purpose of timely correcting the regulation capacity based on the flood filling algorithm in step S40 is to achieve secondary coordinated regulation of the flexible load cluster. The flood filling algorithm prioritizes the measurement range from the two aspects of spatial distance and time interval, allocates peak and valley capacity first, and then spreads it to nearby connected areas until all scheduling capacities are allocated.

[0040] The secondary coordinated control method of the flexible load cluster needs to divide the adjustable capacity into the source point list and the node list according to the priority range of the adjustable capacity. The steps of the secondary coordinated control of the flexible load cluster are as follows: 1) Select one source point from the source point list one by one, and calculate the closest source point based on feature similarity. O of x If there is a node above the threshold, the node is added to the source point list; if x If all nodes are below the threshold, the source node O Divide into a node list, and then select the next source point from the source point list in turn until the source point list is traversed and the measurement is completed; 2) Select a node from the node list one by one X , calculate the distance from the node X Recent x nodes, and perform measurements in turn. If all nodes are measured successfully, the node is added to the source point list, otherwise the node is added to the node list. Finally, all nodes will be added to the source point list and the node list respectively. 3) Check the source point list and node list separately until both the source point list and the node list are cleared, that is, no new nodes can be calculated in the specified area.

[0041] With a time interval of 15 minutes, the dispatch capacity is secondary allocated for 4 periods within each hour, a total of 96 periods. The results of secondary allocation of flexible load regulation capacity and intraday timing control are as follows: Figure 6 shown.

[0042] Due to the actual load changes and the influence of external uncertainties, the capacity deviation in each period of a day is as follows: Figure 7 As shown in the figure, the analysis shows that the deviation capacity of the aggregator participating in load regulation is concentrated in the peak and valley periods, but the overall capacity is small, approximately in the range of -15 to 20kW, and the deviation control amount is within 5%, indicating that the method proposed in the present invention has better flexible load coordinated control and dynamic response characteristics, and has achieved the expected effect.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for secondary coordinated control of flexible loads within a day before the day, characterized by: The following steps are involved: S10: Obtain relevant characteristic parameters and historical power data of user-side flexible resources; S20: Establish flexible resource evaluation indicators, and divide flexible resources into different flexible load clusters through improved K-means clustering method, and form a hierarchical control architecture of flexible loads after aggregation; S30: Combined with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resources is calculated, and combined with the parallel Markov chain model, a preliminary day-ahead dispatch plan for the flexible load is formulated; S40: Real-time tracking of daily user-side load power changes, and timely correction of the regulation capacity based on the flood filling algorithm to achieve secondary coordinated regulation of the flexible load cluster.

2. The method for secondary coordinated control of flexible loads within a day before the day according to claim 1 is characterized by: In step S10, the user-side flexible resources include temperature control equipment, electric vehicles, and distributed energy storage.

3. The method for secondary coordinated control of flexible loads within a day before the day according to claim 1, characterized in that: The flexible resource evaluation index includes a load characteristic index, a regulation characteristic index and a response characteristic index.

4. The method for secondary coordinated control of flexible loads within a day before the day according to claim 3 is characterized by: The calculation formula of the load characteristic index is as follows: f 1= P av / P max ； Where: f 1 is the load characteristic index, P av is the daily average load, P max is the maximum daily load; The calculation formula of the regulation characteristic index is as follows: ； Where: f 2 is the regulation characteristic index, t res is the load response time, Δ t is the time scale, S res It is the regulatory capacity that can participate in the response after the indicator is issued; The calculation formula of the response characteristic index is as follows: ； Where: f 3 is the response characteristic index, t lim is the upper limit of load response time, v res is the daily fluctuation rate of the flexible load.

5. The method for secondary coordinated control of flexible loads within a day before the day according to claim 1, characterized in that: The improved K-means clustering method is improved by using the entropy weight method, and the feature weight values ​​corresponding to the flexible resource evaluation indicators are calculated by the entropy weight method.

6. The method for secondary coordinated control between groups for flexible loads in the day before and within the day according to claim 5, characterized in that: The steps of dividing flexible resources into different flexible load clusters by improving the K-means clustering method are as follows: 1) Select cluster centers based on hierarchical clustering results C I , and A The cluster centers are A Input, copy B Distribute to B Computers; 2) Daily load sample set P A×B All samples in P j =[ p j1 , p j2 , p j3 ,…, p jB ] is distributed to each computer node, and then calculated to the I Cluster Centers C I The weighted Euclidean distance of ; 3) A Inputs are dispatched to c Distributed computing cores are used to calculate the minimum weighted Euclidean distance for each sample. ， Divide it into cluster centers C I The cluster in which it is located; 4) For each cluster, update the cluster center C I ; 5) Repeat steps 3) and 4) until the new cluster center becomes stable or the loss function converges.

7. The method for secondary coordinated control of flexible loads within a day before the day according to claim 6 is characterized by: The hierarchical control architecture of the flexible load formed after aggregation in step S20 includes an aggregator layer, a load agent layer and a user equipment layer. The user equipment layer uploads load parameters and operation status information to the load agent through a smart socket or a smart switch, and performs dynamic timing control of the controllable load in the form of direct call according to the control indicators allocated and issued by the load agent to achieve the demand side response target; The load proxy layer monitors the load changes within the flexible load cluster in real-time time scale, continuously updates the load status and response potential parameters, and performs secondary distribution of target power among the flexible load clusters according to the control target and the updated response potential, thereby controlling the user equipment. The aggregator layer calculates the maximum response potential range of various flexible load clusters and feeds it back to the dispatch center, which further obtains the adjustable target power. After evaluation, it makes an initial allocation and hands it over to the corresponding load agents for management. At the same time, it relies on a two-way communication mechanism to strengthen coordination and control.

8. The method for secondary coordinated control of flexible loads within a day before the day according to claim 1, characterized in that: In step S30, in combination with the day-ahead load forecasting method, the maximum adjustable capacity of the user-side flexible resource is calculated by performing aggregate modeling on multiple flexible loads, wherein the aggregate modeling expression of multiple flexible loads is as follows: ； Where: for t Always control the internal temperature of the equipment. for t+ 1. External ambient temperature at time 1, R is the equivalent thermal resistance, C is the equivalent heat capacity, Δ t is the time scale, λ The electric heating and cooling power of the temperature control equipment The conversion efficiency of for t The state of charge of the electric vehicle at all times, for t The rated charging power of the electric vehicle at the moment, E EV is the power battery capacity of electric vehicles, η For the charging efficiency of electric vehicles, for t The state of charge of the energy storage device at all times, for t The rated charging and discharging power of the energy storage device at all times, E ES is the battery capacity of the energy storage device, α is the charging and discharging efficiency index of the energy storage device, and its expression is as follows: ； Where: η c and η d are the charging and discharging efficiency of the energy storage device, respectively; The constraints for the aggregate modeling of multivariate flexible loads are as follows: ； Where: δ is the temperature offset acceptable to the user, T set is the set temperature value; , They are t The upper and lower limits of the state of charge of the electric vehicle at all times; , They are t The upper and lower limits of the state of charge of the energy storage device at all times; The maximum adjustable capacity of the user-side flexible resource is calculated as follows: ； Where: and are the calculated values ​​of the maximum adjustable increase and decrease capacity of the load aggregator, respectively; N TC , N EV and N ES are the corresponding load agent numbers of temperature control equipment, electric vehicles, and energy storage equipment; λ TC , λ EV and λ ES are the capacity increase state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. u TC , u EV and u ES are the capacity reduction state transition probabilities of temperature control equipment, electric vehicles, and energy storage equipment, respectively. For the i Temperature control equipment in t The electric heating and cooling power at all times, For the i Electric vehicles in t Rated charging power at the moment, For the i Energy storage devices in t Rated charge and discharge power at the moment.

9. The method for secondary coordinated control of flexible loads within a day before the day according to claim 8, characterized in that: The steps of preliminarily formulating the flexible load day-ahead dispatching plan in combination with the parallel Markov chain model in step S30 are as follows: 1) Assumptions Q For flexible load operation state, Q 1=1 indicates running or charging status, Q 2=0 means closed state, Q 3=-1 indicates discharge state; 2) If Q 1. Q 2. Q 3 Three states are randomly converted, and the state conversion probability matrix corresponding to different flexible loads is obtained according to the Markov chain model; 3) According to the maximum likelihood estimation, the state transition probability is replaced by the state transition frequency of different flexible loads at a certain time scale to obtain the different state transition probabilities of the flexible load; 4) Based on the day-ahead forecast results of flexible loads and the calculation method of the maximum adjustable capacity, the upper and lower limits of the load aggregator’s adjustable capacity are further calculated to formulate a day-ahead dispatch plan for flexible loads.

10. The method for secondary coordinated control of flexible loads within a day before the day according to claim 1, characterized in that: In step S40, the measurement range is prioritized from the two aspects of spatial distance and time interval based on the flood filling algorithm, and the peak and valley capacity is allocated first, and then diffused to nearby connected areas until all scheduling capacities are allocated.

Images