Hierarchical and partitioned autonomous distributed resource cooperative control method and system for Internet of Things terminal

Abstract

The invention discloses a layered and partitioned autonomous distributed resource cooperative control method and system for an Internet of Things terminal. The method comprises the following steps: establishing an equivalent model for a transformer area layer based on an internal autonomous principle; performing distributed resource cooperative control of the transformer area according to the equivalent model, and calculating source load storage basic data and active and reactive regulation and control capacity in the transformer area; calculating a comprehensive modularity zoning index, and performing regional division on the feeder line layer according to the comprehensive modularity zoning index; establishing a scheduling model of the feeder layer based on the divided area of the feeder layer and the source load storage basic data and the active and reactive regulation capacity in the transformer area; performing distributed resource cooperative control of the feeder layer according to the feeder layer scheduling model; and carrying out interval distributed coordination optimization on the scheduling model of the feeder layer by adopting an optimization ADMM algorithm, and solving to obtain an optimized scheduling result of scheduling operation in different time scales in a layered and partitioned manner. According to the method, multi-type resource complementation characteristics are considered, and coordinated regulation and control of the power distribution network system can be completed under multiple time scales.

Description

Technical Field

[0001] The present invention belongs to the technical field of resource collaborative control, and more specifically, relates to a distributed resource collaborative control method for hierarchical and partitioned autonomy of Internet of Things terminals. Background Art

[0002] With the massive addition of distributed resources, the traditional centralized operation and management methods of distribution networks are no longer applicable, so hierarchical and zoned control and refined energy management of distribution networks become possible. The large-scale penetration of distributed power sources can, on the one hand, be used as a flexible resource to coordinate and deal with the operation needs of the distribution network and improve the resilience of the distribution network; on the other hand, due to the dispersed spatial distribution, many types, large scale and difficulty in management of distributed resources, excessive penetration will bring a series of operation safety problems to the distribution network. Therefore, it is urgent to study a set of coordinated control strategies for distributed resources in distribution networks to solve the current problem of many types and dispersed dispatching resources and to realize the diversified regulation needs of distribution networks such as economy and safety.

[0003] The patent CN117239725A discloses a distributed flexible resource aggregation and hierarchical collaborative scheduling method and device, the method comprising: obtaining the operating data of all distributed flexible resources contained in the target area; dynamically clustering all distributed flexible resources according to the operating data to obtain the grouping result; determining the multi-dimensional aggregation information of the target area according to the grouping result; building a hierarchical collaborative control model based on the multi-dimensional aggregation information; the hierarchical collaborative control model includes a first layer model, a second layer model and a third layer model; iteratively optimizing the first objective function, the second objective function and the third objective function through the resource scheduling optimization algorithm, and outputting the optimized scheduling scheme of the distributed flexible resources. However, in order to simultaneously consider the internal autonomous division of the substation and the optimal operation of the distribution network as a whole, and to consider the needs of medium and long-term energy scheduling management, it is difficult to achieve the coordination of different control objectives of the distribution network control under the hierarchical and partitioned collaborative framework, as well as the coordination of medium and long-term scheduling and short-term scheduling. In addition, it is not possible to consider how to achieve the use of flexible resources in a partitioned and layered manner to solve low-probability, high-severity scenario problems under the condition of global optimization + local autonomy after a large number of distributed energy sources are added. Summary of the invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a distributed resource collaborative control method for hierarchical and partitioned autonomy of IoT terminals.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of the present invention proposes a distributed resource collaborative control method for hierarchical and partitioned autonomy of IoT terminals, which is characterized by comprising:

[0007] An equivalent model is established for the substation layer based on the principle of internal autonomy; the equivalent model includes feeder-substation layer power exchange conditions, substation target functions and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model as the goal, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power;

[0008] The feeder layer is randomly divided into regions, and the modularity index value, active power balance index value and reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are integrated into the comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value;

[0009] A dispatching model for the feeder layer is established based on the areas divided by the feeder layer, and the dispatching model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized dispatching model, and the distributed resources of the feeder layer area are adjusted while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer;

[0010] The optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area, and the feeder layer is adjusted according to the optimized distributed resources in the feeder layer area.

[0011] Preferably, the feeder-substation layer power exchange condition is that the active and reactive power variables exchanged at the feeder area side at all feeder layers and substation layer coupling nodes in the substation are respectively equal to the active and reactive power variables exchanged at the corresponding substation layer.

[0012] Preferably, the station area objective function is:

[0013] minF=C DG +C ESS +C pgrid

[0014]

[0015] In the formula: C DG is the penalty cost for abandoning light, C ESS is the operation and maintenance cost of distributed energy storage, C pgrid is the penalty cost for power interaction between the substation and feeder layer, is the penalty cost coefficient for abandoned light, is the abandoned power of the lth photovoltaic in the area at time t, N DG is the number of distributed photovoltaics within the area; and are the charging and discharging power of the sth distributed energy storage at time t, is the operation and maintenance cost of energy storage per unit power, N ESS is the amount of distributed energy storage within the substation; is the penalty cost coefficient of power interaction between the substation and the feeder side of the upper grid at time t, P t UP is the actual value of the interaction power between the station layer and the feeder layer at time t; T is the time length;

[0016] The station area constraints include power balance constraints, distributed energy storage constraints, and distributed photovoltaic power generation constraints.

[0017] Preferably, the feeder layer is randomly divided into regions, a modularity index value, an active power balance index value and a reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are merged into a comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value, specifically:

[0018] The time scene with the highest photovoltaic penetration rate is selected to divide the feeder layer into regions, and the feeder layer is randomly divided into regions. After the division, the modularity index value, active balance index value and reactive balance index value under the time scene are calculated respectively, and the three index values ​​are linearly added with the set weights to obtain the comprehensive modularity partition index. The maximum comprehensive modularity partition index is taken as the objective function, and the pre-trained genetic algorithm is used to optimize the regional division.

[0019] Preferably, the calculation of the modularity index value is specifically as follows:

[0020] The weight w of the edge between the connecting nodes i and j in the feeder layer is calculated by the electrical distance i j:

[0021]

[0022] Where: ΔV i Indicates the change in the amplitude of node voltage i; ΔQ j is the reactive power variation of node j; They represent the sensitivity of the reactive power change of node j to the voltage amplitude of node i and node j respectively; d ij It reflects the ratio change of the degree of influence of node i voltage when the reactive power of node j changes; N represents the number of nodes in the distribution network; D ij represents the electrical distance between nodes i and j; d iβ ,d jβ They respectively represent the ratio of the impact of node i and node j voltages when the reactive power of node β changes;

[0023] The calculation formula of modularity index value is as follows:

[0024]

[0025] Where: χ is the modularity index value; Ω is the feeder layer node set; w i j is the weight of the edge connecting nodes i and j, w i is the sum of the weights of the edges connected to node i, w j is the sum of the weights of the edges connected to node j, and σ is the sum of the weights of the edges between all nodes; when i and j are in the same partition, δ(i,j) is 1, otherwise it is 0.

[0026] Preferably, the active and reactive power balance index values ​​are specifically:

[0027] The reactive power balance degree is calculated by randomly dividing the feeder layer area:

[0028]

[0029] Where: N clu Indicates the number of feeder layer areas, is the net reactive power value of the νth feeder area at time t, is the maximum reactive power provided by the νth feeder area at time t, is the reactive power load demand value of the νth feeder area at time t;

[0030] Active power balance index value, calculation formula is:

[0031]

[0032]

[0033] Where: is the net active power value of the νth feeder area at time t; is the maximum active power provided by the νth feeder area at time t, and is the active power demand value of the νth feeder area at time t.

[0034] Preferably, the establishment of a scheduling model for the feeder layer is specifically as follows:

[0035] The dispatch model of the feeder layer includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer;

[0036] The objective function of the νth feeder area is:

[0037]

[0038] Where: is the cost of abandoning renewable energy in the νth feeder area; is the energy storage operation and maintenance cost of the νth feeder area; is the main grid electricity purchase cost for the νth feeder area ν; is the network loss cost of the νth feeder area;

[0039] The feeder layer constraints include Distflow constraints, power balance constraints, energy storage power station charge state constraints, node voltage constraints, photovoltaic power station power generation constraints, and main grid-feeder interaction power upper and lower limit constraints; and all feeder layer areas must satisfy the feeder-station layer power exchange.

[0040] Preferably, the optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources of the feeder layer area, specifically:

[0041] The initial values ​​of global variables are set according to the measured data of the distribution network, and the initial values ​​of Lagrange multipliers and penalty parameters in all feeder layer regions are set to 0.

[0042] The global variables are active power, reactive power, voltage, current, and phase difference between voltage and current at the input and output ends of the feeder layer; the boundary data are information exchanged between different partitions or nodes, including voltage, current, active power, reactive power, control instructions, and status information; the penalty parameter is the reactive compensation value optimized in each iteration of the feeder layer area;

[0043] Establish the Lagrangian augmented function corresponding to the objective function in the dispatching model of the feeder layer. The Lagrangian augmented function corresponding to any two adjacent feeder layer regions a and b is as follows:

[0044]

[0045]

[0046] Where: k is the penalty parameter for this iteration, x a 、x b are decision variables within feeder layer region a and feeder layer region b respectively; the decision variables are active and reactive power change values ​​of the corresponding feeder layer region after optimization calculation;

[0047] F a 、F b is the objective function of feeder layer area a and feeder layer area b; These are the power interaction results from upstream node i to node j obtained by this iterative optimization calculation within feeder layer area a and feeder layer area b; are the fixed reference values ​​for the next iteration respectively; are the Lagrange multipliers of feeder layer region a and feeder layer region b corresponding to the kth iteration respectively.

[0048] Update the decision variables within each feeder layer area; optimize and solve the Lagrangian augmented function within each area in parallel, and use the solution as the fixed value for the next iteration after the update

[0049] Repeatedly update the decision variables within each feeder layer region until the judgment algorithm converges, and output the final Lagrangian augmented function values ​​corresponding to feeder layer region a and feeder layer region b. As the distributed resources of the optimized feeder layer area a and feeder layer area b.

[0050] Preferably, the optimized ADMM algorithm adaptively updates the penalty parameter based on the spectral step size obtained according to the gradient descent method;

[0051] Update the penalty parameter based on the spectral step size as follows:

[0052] Among them, ψ k+1 is the updated penalty parameter, ψ k is the penalty parameter for this iteration, and Respectively and the correlation of the changes in the auxiliary variables of the Lagrange multiplier, a k and b k F a 、F b The spectral step size is obtained by the mixed step size rule; ε cor is the set threshold, F a 、F b The amount of change.

[0053] The second aspect of the present invention proposes a distributed resource collaborative control system with hierarchical partition autonomy of IoT terminals using the method described in the first aspect of the present invention, characterized in that it includes:

[0054] Equivalent model building module: used to establish an equivalent model for the substation layer based on the internal autonomy principle; the equivalent model includes feeder-substation layer power exchange conditions, substation target function and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power;

[0055] Area division module: used to randomly divide the feeder layer into regions, calculate the modularity index value, active and reactive power balance index value according to the divided feeder layer regions, and merge the three index values ​​into a comprehensive modularity partition index value, and divide the feeder layer into regions according to the comprehensive modularity partition index value;

[0056] Scheduling model building module: used to establish a scheduling model for the feeder layer based on the areas divided by the feeder layer. The scheduling model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized scheduling model, and to adjust the distributed resources of the feeder layer area while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer;

[0057] Optimization module: It is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area by using the optimized alternating direction multiplier ADMM algorithm, and adjust the feeder layer according to the distributed resources in the optimized feeder layer area.

[0058] The beneficial effect of the present invention is that, compared with the prior art, in view of the scene characteristics of high dynamic, high dimensional and high intelligent entities within large-scale complex new distribution networks, as well as the difficulties in efficient consumption and coordinated regulation of hierarchical and partitioned resources and the low efficiency of centralized solution, the present invention uses comprehensive indicators to divide the feeder layer into regions, which is conducive to in-depth analysis of various resources of the distribution network and in various scenarios. By constructing an equivalent model and a scheduling model for the coordinated optimization of hierarchical and partitioned distribution networks at multiple time scales, the complementary characteristics of multiple types of resources are taken into account, the coordinated regulation of the distribution network system is completed at multiple time scales, and the optimized ADMM algorithm is used for interval distributed coordinated optimization, which smoothes the random power fluctuations of distributed power sources and realizes efficient consumption of distributed power sources and economical operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0061] like Figure 1 As shown, embodiment 1 of the present invention proposes a distributed resource collaborative control method for hierarchical partition autonomy of IoT terminals, which is characterized by comprising:

[0062] An equivalent model is established for the substation layer based on the principle of internal autonomy; the equivalent model includes feeder-substation layer power exchange conditions, substation target functions and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model as the goal, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power;

[0063] The feeder layer is randomly divided into regions, and the modularity index value, active power balance index value and reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are integrated into the comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value;

[0064] A dispatching model for the feeder layer is established based on the areas divided by the feeder layer, and the dispatching model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized dispatching model, and the distributed resources of the feeder layer area are adjusted while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer;

[0065] The optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area, and the feeder layer is adjusted according to the optimized distributed resources in the feeder layer area.

[0066] Preferably, the feeder-substation layer power exchange condition is that the active and reactive power variables exchanged at the feeder area side at all feeder layers and substation layer coupling nodes in the substation are respectively equal to the active and reactive power variables exchanged at the corresponding substation layer.

[0067] The formula for the power exchange condition at the feeder-station level is:

[0068]

[0069] in, are the active power and reactive power exchanged on the feeder area side at the coupling nodes of the feeder layer and the substation layer, respectively. are the active power and reactive power exchanged at the corresponding substation layer respectively; M is the total number of substations;

[0070] Preferably, the station area objective function is:

[0071] minF=C DG +C ESS +C pgrid

[0072]

[0073] In the formula: C DGis the penalty cost for abandoning light, C ESS is the operation and maintenance cost of distributed energy storage, C pgrid is the penalty cost for power interaction between the substation and feeder layer, is the penalty cost coefficient for abandoned light, is the abandoned power of the lth photovoltaic in the area at time t, N DG is the number of distributed photovoltaics within the area; and are the charging and discharging power of the sth distributed energy storage at time t, is the operation and maintenance cost of energy storage per unit power, N ESS is the amount of distributed energy storage within the substation; is the penalty cost coefficient of power interaction between the substation and the feeder side of the upper grid at time t, P t UP is the actual value of the interaction power between the station layer and the feeder layer at time t; T is the time length;

[0074] The station area constraints include power balance constraints, distributed energy storage constraints, and distributed photovoltaic power generation constraints.

[0075] Specifically, the station constraint formulas are:

[0076] 1) Power balance constraints

[0077]

[0078] Where: is the actual active power output of distributed photovoltaics in the area at time t; It is the total load demand within the substation at time t.

[0079] 2) Distributed energy storage constraints

[0080]

[0081] Where: is the energy state of energy storage s at time t; are the initial energy state and the end-of-cycle energy state of the energy storage s, respectively; Δt represents the duration of energy storage charging and discharging; are the charge and discharge efficiency respectively; are the minimum and maximum values ​​of the state of charge, respectively; They are the charge and discharge states of energy storage s at time t. In a time period, the energy storage can be in either the charge or discharge state, represented by 0 and 1; are the maximum values ​​of charging power and discharging power of energy storage s respectively.

[0082] 3) Distributed photovoltaic output constraints

[0083]

[0084] Where: It is the maximum value of the distributed photovoltaic output in the area at time t.

[0085] Preferably, the feeder layer is randomly divided into regions, a modularity index value, an active power balance index value and a reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are merged into a comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value, specifically:

[0086] The time scene with the highest photovoltaic penetration rate is selected to divide the feeder layer into regions, and the feeder layer is randomly divided into regions. After the division, the modularity index value, active balance index value and reactive balance index value under the time scene are calculated respectively, and the three index values ​​are linearly added with the set weights to obtain the comprehensive modularity partition index. The maximum comprehensive modularity partition index is taken as the objective function, and the pre-trained genetic algorithm is used to optimize the regional division.

[0087] Preferably, the calculation of the modularity index value is specifically as follows:

[0088] Specifically, this embodiment selects the time scene with the highest photovoltaic penetration rate as the time scene at 13:00,

[0089] At this point, the photovoltaic penetration rate is 120%.

[0090] Specifically, the calculation formula of the comprehensive modularity partition index ξ is:

[0091]

[0092] Among them, χ is the modularity index; It is the active power balance index; is the reactive power balance; α is the weight of the modularity index; β is the weight of the active power balance index, and δ is the weight of the reactive power balance; specifically, α, β, and δ are 0.6, 0.2, and 0.2, respectively.

[0093] Specifically, this embodiment uses a genetic algorithm to perform region division, specifically, taking the region as a population, initializing the population, performing crossover and mutation operations according to the genetic algorithm, optimizing the selection of the population, until the comprehensive partition index reaches the maximum, or reaches the set maximum number of iterations, and then stopping the iteration. At this time, the population is the divided region;

[0094] The maximum number of iterations is set to 500, the population size is 40, the crossover probability is 0.3-0.6, and the mutation probability is (0.1, 0.5). To ensure convergence, the two best individuals after each iteration are set not to perform crossover and mutation operations. After about 172 iterations, the comprehensive partition index of this embodiment reaches a maximum value of 0.869.

[0095] Preferably, the calculation of the modularity index value is specifically as follows:

[0096] The weight w of the edge between the connecting nodes i and j in the feeder layer is calculated by the electrical distance i j:

[0097]

[0098] Where: ΔV i Indicates the change in the amplitude of node voltage i; ΔQ j is the reactive power variation of node j; They represent the sensitivity of the reactive power change of node j to the voltage amplitude of node i and node j respectively; d ij It reflects the ratio change of the degree of influence of node i voltage when the reactive power of node j changes; N represents the number of nodes in the distribution network; D ij represents the electrical distance between nodes i and j; d iβ ,d jβ They respectively represent the ratio of the impact of node i and node j voltages when the reactive power of node β changes;

[0099] The calculation formula of modularity index value is as follows:

[0100]

[0101] Where: χ is the modularity index value; Ω is the feeder layer node set; w ij is the weight of the edge connecting nodes i and j, w i is the sum of the weights of the edges connected to node i, w j is the sum of the weights of the edges connected to node j, and σ is the sum of the weights of the edges between all nodes; when i and j are in the same partition, δ(i,j) is 1, otherwise it is 0.

[0102] Preferably, the active and reactive power balance index values ​​are specifically:

[0103] The reactive power balance degree is calculated by randomly dividing the feeder layer area:

[0104]

[0105] Where: N clu Indicates the number of feeder layer areas, is the net reactive power value of the νth feeder area at time t, is the maximum reactive power provided by the νth feeder area at time t, is the reactive power load demand value of the νth feeder area at time t;

[0106] Active power balance index value, calculation formula is:

[0107]

[0108] Where: is the net active power value of the νth feeder area at time t; is the maximum active power provided by the νth feeder area at time t, and is the active power demand value of the νth feeder area at time t.

[0109] Preferably, the establishment of a scheduling model for the feeder layer is specifically as follows:

[0110] The dispatch model of the feeder layer includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer;

[0111] The objective function of the νth feeder area is:

[0112]

[0113] Where: is the cost of abandoning renewable energy in the νth feeder area; is the energy storage operation and maintenance cost of the νth feeder area; is the main grid electricity purchase cost for the νth feeder area ν; is the network loss cost of the νth feeder area;

[0114] It should be noted that the calculation formulas for the above costs are:

[0115] 1) Penalty costs for abandoning new energy

[0116]

[0117] Where: is the number of PV plants within the νth feeder area; π DG The unit power generation loss cost of the photovoltaic power station; is the predicted value of active output of the photovoltaic power station at node j at time t; is the actual active output value of the photovoltaic power station at node j at time t; is the abandoned power value of the PV power station at node j at time t.

[0118] 2) Operating costs of energy storage power stations

[0119]

[0120] Where: is the number of energy storage power stations within the νth feeder area; π ESS is the operating cost of the unit energy storage power station; is the charging power and discharging power of the energy storage station at node j at time t.

[0121] 3) Main grid electricity purchase cost

[0122]

[0123] Where: is the active power purchased by the νth feeder area from the upper main grid at time t; upper is the real-time electricity price of the power grid at time t.

[0124] 4) Network loss cost

[0125]

[0126] Where: N ν represents the number of internal nodes in the νth feeder area; is the total network loss of the νth feeder area at time t; π ploss is the unit network loss cost; R ij is the resistance value of the line between nodes i and j; I ij is the current passing between nodes i and j, P ij , Q ij are the active and reactive power flowing from upstream node i to node j respectively; V i is the voltage at node i.

[0127] The feeder layer constraints include Distflow flow constraints, power balance constraints, energy storage power station charge state constraints, node voltage constraints, photovoltaic power station power generation constraints, and main grid-feeder interaction power upper and lower limit constraints; and the sum of the feeder-station layer power exchanges in all feeder layer areas is the feeder-station layer power exchange within the station.

[0128] It should be noted that the formulas for the constraints of each feeder layer are as follows:

[0129] (2) Feeder layer constraints

[0130] 1) Distflow Constraints

[0131]

[0132] Where: i∈ζ(j), i∈ψ(j) respectively represent the set of first (last) nodes of the branch to which node i belongs with node j as the last (first) node; Xij is the reactance value of the line between nodes i and j; P ij,t , Q ij,t is the active and reactive power flowing from upstream node i to node j at time t; P jk,t , Q jk,t is the active power and reactive power transferred between nodes j and k at time t; P j,t , Q j,t They represent the net injected active power and reactive power of node j at time t respectively; I ij,t is the current of branch ij at time t; U i,t , U j,t are the upstream and downstream node voltages at time t respectively.

[0133] Since the following equation is a nonlinear constraint, it needs to be converted into an inequality constraint according to the second-order cone relaxation principle:

[0134]

[0135] 2) Power balance constraints

[0136]

[0137] Where: are the active power and reactive power of the load at node j at time t respectively;

[0138] The power purchased by node j from the main grid at time t; is the reactive power output of the photovoltaic at node j at time t.

[0139] 3) Constraints on the state of charge of energy storage power stations

[0140]

[0141] Where: E j,t is the energy state of the energy storage power station at node j at time t; E is the charging and discharging efficiency of the energy storage station at node j; j,0 、E j,T are the initial energy state and the end-of-cycle energy state of the energy storage power station at node j, respectively; are the maximum charging power and discharging power of the energy storage power station at node j respectively; E j,m i n、E j,max are the minimum and maximum state of charge at node j, respectively; They are the charging and discharging states of the energy storage power station at node j at time t. In a time period, the energy storage power station is only in the charging or discharging state, which are represented by 0 and 1 respectively.

[0142] 4) Node voltage constraints

[0143]

[0144] U pre =U N

[0145] Where: and are the minimum and maximum limits of the square of the voltage at node i, U pre is the voltage amplitude at the substation outlet, U N is the rated voltage amplitude of the system.

[0146] 5) Output constraints of photovoltaic power stations

[0147]

[0148] Where: It is the maximum value of active and reactive output of the PV power station at node j.

[0149] 6) Main grid-feeder interaction power upper and lower limit constraints

[0150]

[0151] Where: is the interaction power between the νth feeder area and the main grid at time t; are the minimum and maximum values ​​of active power exchange between the νth feeder area and the main grid, respectively; are the minimum and maximum values ​​of reactive power exchange between the νth feeder area and the main grid, respectively.

[0152] Preferably, the optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources of the feeder layer area, specifically:

[0153] The initial values ​​of global variables are set according to the measured data of the distribution network, and the initial values ​​of Lagrange multipliers and penalty parameters in all feeder layer regions are set to 0.

[0154] The global variables are active power, reactive power, voltage, current, and phase difference between voltage and current at the input and output ends of the feeder layer; the boundary data are information exchanged between different partitions or nodes, including voltage, current, active power, reactive power, control instructions, and status information; the penalty parameter is the reactive compensation value optimized in each iteration of the feeder layer area;

[0155] Establish the Lagrangian augmented function corresponding to the objective function in the dispatching model of the feeder layer. The Lagrangian augmented function corresponding to any two adjacent feeder layer regions a and b is as follows:

[0156]

[0157] Where: k is the penalty parameter for this iteration, x a 、x b are decision variables within feeder layer region a and feeder layer region b respectively; the decision variables are active and reactive power change values ​​of the corresponding feeder layer region after optimization calculation;

[0158] F a 、F b is the objective function of feeder layer area a and feeder layer area b; These are the power interaction results from upstream node i to node j obtained by this iterative optimization calculation within feeder layer area a and feeder layer area b; are the fixed reference values ​​for the next iteration respectively; are the Lagrange multipliers of feeder layer region a and feeder layer region b corresponding to the kth iteration respectively.

[0159] Update the decision variables within each feeder layer area; optimize and solve the Lagrangian augmented function within each area in parallel, and use the solution as the fixed value for the next iteration after the update

[0160] Repeatedly update the decision variables within each feeder layer region until the judgment algorithm converges, and output the final Lagrangian augmented function values ​​corresponding to feeder layer region a and feeder layer region b. As the distributed resources of the optimized feeder layer area a and feeder layer area b.

[0161] Preferably, the optimized ADMM algorithm adaptively updates the penalty parameter based on the spectral step size obtained according to the gradient descent method;

[0162] It should be noted that the convergence criterion for judging the convergence of the algorithm is the original residual r k+1 and the dual residual s k+1 Whether it tends to converge to the accuracy; the convergence criterion formula is:

[0163]

[0164] Where δ is the convergence accuracy.

[0165] Preferably, the penalty parameter is updated based on the spectrum step size, specifically:

[0166] Update the auxiliary variable y of the Lagrange multiplier k , the formula is:

[0167]

[0168] Update the Lagrange multiplier, the formula is:

[0169]

[0170] In the formula, are the Lagrange multipliers of the feeder layer region a and the feeder layer region b corresponding to the next iteration respectively.

[0171] Calculate the steepest descent and minimum gradient in the feeder layer region a as and The steepest descent and minimum gradient in the feeder layer region b are and The calculation is as follows:

[0172]

[0173] F a 、F b The spectral step length is a k and b k , calculated by the mixed step rule, as shown below:

[0174]

[0175] Update the penalty parameter based on the spectral step size:

[0176]

[0177] Among them, ψ k+1 is the updated penalty parameter, ψ k is the penalty parameter for this iteration, and Respectively and the auxiliary variable y of its Lagrange multiplier k The change in Δy k The correlation of k and b k F a 、F b The spectral step size is obtained by the mixed step size rule; ε cor is the set threshold, F a 、F b The amount of change.

[0178] Embodiment 2 of the present invention proposes a distributed resource collaborative control system with hierarchical partition autonomy of IoT terminals using the method described in Embodiment 1 of the present invention, including:

[0179] Equivalent model building module: used to establish an equivalent model for the substation layer based on the internal autonomy principle; the equivalent model includes feeder-substation layer power exchange conditions, substation target function and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power;

[0180] Area division module: used to randomly divide the feeder layer into regions, calculate the modularity index value, active and reactive power balance index value according to the divided feeder layer regions, and merge the three index values ​​into a comprehensive modularity partition index value, and divide the feeder layer into regions according to the comprehensive modularity partition index value;

[0181] Scheduling model building module: used to establish a scheduling model for the feeder layer based on the areas divided by the feeder layer. The scheduling model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized scheduling model, and to adjust the distributed resources of the feeder layer area while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer;

[0182] Optimization module: It is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area by using the optimized alternating direction multiplier ADMM algorithm, and adjust the feeder layer according to the distributed resources in the optimized feeder layer area.

[0183] 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 above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed resource collaborative control method for hierarchical and partitioned autonomy of IoT terminals, characterized in that: include: An equivalent model is established for the substation layer based on the principle of internal autonomy; the equivalent model includes feeder-substation layer power exchange conditions, substation target functions and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model as the goal, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power; The feeder layer is randomly divided into regions, and the modularity index value, active power balance index value and reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are integrated into the comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value; A dispatching model for the feeder layer is established based on the areas divided by the feeder layer, and the dispatching model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized dispatching model, and the distributed resources of the feeder layer area are adjusted while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer; The optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area, and the feeder layer is adjusted according to the optimized distributed resources in the feeder layer area.

2. According to claim 1, a distributed resource collaborative control method for hierarchical and partitioned autonomy of IoT terminals is characterized by: The feeder-station layer power exchange condition is that the active and reactive power variables exchanged at the feeder area side at all feeder layers and station layer coupling nodes in the station area are respectively equal to the active and reactive power variables exchanged at the corresponding station layer.

3. The method for distributed resource collaborative control of hierarchical and partitioned autonomous IoT terminals according to claim 2, characterized in that: The objective function of the station area is: minF=C DG +C ESS +C pgrid In the formula: C DG is the penalty cost for abandoning light, C ESS is the operation and maintenance cost of distributed energy storage, C pgrid is the penalty cost for power interaction between the substation and feeder layer, is the penalty cost coefficient for abandoned light, is the abandoned power of the lth photovoltaic in the area at time t, N DG is the number of distributed photovoltaics within the substation; and are the charging and discharging power of the sth distributed energy storage at time t, is the operation and maintenance cost of energy storage per unit power, N ESS is the amount of distributed energy storage within the substation; is the penalty cost coefficient for power interaction between the substation and the feeder side of the upper grid at time t, is the actual value of the interaction power between the station layer and the feeder layer at time t; T is the time length; The station area constraints include power balance constraints, distributed energy storage constraints, and distributed photovoltaic power generation constraints.

4. The method for distributed resource collaborative control of hierarchical and partitioned autonomy of IoT terminals according to claim 1, characterized in that: The feeder layer is randomly divided into regions, and the modularity index value, active power balance index value and reactive power balance index value are calculated according to the divided feeder layer regions, and the three index values ​​are integrated into a comprehensive modularity partition index value, and the feeder layer is divided into regions according to the comprehensive modularity partition index value, specifically: The time scene with the highest photovoltaic penetration rate is selected to divide the feeder layer into regions, and the feeder layer is randomly divided into regions. After the division, the modularity index value, active balance index value and reactive balance index value under the time scene are calculated respectively, and the three index values ​​are linearly added with the set weights to obtain the comprehensive modularity partition index. The maximum comprehensive modularity partition index is taken as the objective function, and the pre-trained genetic algorithm is used to optimize the regional division.

5. The method for distributed resource collaborative control of hierarchical and partitioned autonomy of IoT terminals according to claim 4, characterized in that: The calculation of the modularity index value is specifically as follows: The weight w of the edge between the connecting nodes i and j in the feeder layer is calculated by the electrical distance ij : Where: ΔV i represents the change in the amplitude of node voltage i; ΔQ j is the reactive power variation of node j; They represent the sensitivity of the reactive power change of node j to the voltage amplitude of node i and node j respectively; d ij It reflects the ratio change of the degree of influence of node i voltage when the reactive power of node j changes; N represents the number of nodes in the distribution network; D ij represents the electrical distance between nodes i and j; d iβ d jβ They respectively represent the ratio of the impact of node i and node j voltages when the reactive power of node β changes; The calculation formula of modularity index value is as follows: Where: χ is the modularity index value; Ω is the set of feeder layer nodes; w ij is the weight of the edge connecting nodes i and j, w i is the sum of the weights of the edges connected to node i, w j is the sum of the weights of the edges connected to node j, and σ is the sum of the weights of the edges between all nodes; when i and j are in the same partition, δ(i,j) is 1, otherwise it is 0.

6. The method for distributed resource collaborative control of hierarchical and partitioned autonomy of IoT terminals according to claim 4, characterized in that: The active and reactive power balance index values ​​are specifically: The reactive power balance degree is calculated by randomly dividing the feeder layer area: Where: N clu Indicates the number of feeder layer areas, is the net reactive power value of the νth feeder area at time t, is the maximum reactive power provided by the νth feeder area at time t, is the reactive power load demand value of the νth feeder area at time t; Active power balance index value, calculation formula is: Where: is the net active power value of the νth feeder area at time t; is the maximum active power provided by the νth feeder area at time t, and is the active power demand value of the νth feeder area at time t.

7. The method for distributed resource collaborative control of hierarchical and partitioned autonomous IoT terminals according to claim 1, characterized in that: The establishment of the scheduling model of the feeder layer is specifically as follows: The dispatch model of the feeder layer includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; The objective function of the νth feeder area is: Where: is the cost of abandoning renewable energy in the νth feeder area; is the energy storage operation and maintenance cost of the νth feeder area; is the main grid electricity purchase cost for the νth feeder area ν; is the network loss cost of the νth feeder area; The feeder layer constraints include Distflow constraints, power balance constraints, energy storage power station charge state constraints, node voltage constraints, photovoltaic power station power generation constraints, and main grid-feeder interaction power upper and lower limit constraints; and all feeder layer areas must satisfy the feeder-station layer power exchange.

8. The method for distributed resource collaborative control of hierarchical and partitioned autonomy of IoT terminals according to claim 1, characterized in that: The optimized alternating direction multiplier ADMM algorithm is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources of the feeder layer area, specifically: The initial values ​​of global variables are set according to the measured data of the distribution network, and the initial values ​​of Lagrange multipliers and penalty parameters in all feeder layer regions are set to 0. The global variables are active power, reactive power, voltage, current, and phase difference between voltage and current at the input and output ends of the feeder layer; the boundary data are information exchanged between different partitions or nodes, including voltage, current, active power, reactive power, control instructions, and status information; the penalty parameter is the reactive compensation value optimized in each iteration of the feeder layer area; Establish the Lagrangian augmented function corresponding to the objective function in the dispatching model of the feeder layer. The Lagrangian augmented function corresponding to any two adjacent feeder layer regions a and b is as follows: Where: k is the penalty parameter for this iteration, x a 、x b are the decision variables within feeder layer region a and feeder layer region b respectively; The decision variable is the active and reactive power change value of the corresponding feeder layer area after optimization calculation; F a 、F b is the objective function of feeder layer area a and feeder layer area b; These are the power interaction results from upstream node i to node j obtained by this iterative optimization calculation within feeder layer area a and feeder layer area b; are the fixed reference values ​​for the next iteration respectively; are the Lagrange multipliers of feeder layer region a and feeder layer region b corresponding to the kth iteration respectively. Update the decision variables within each feeder layer area; optimize and solve the Lagrangian augmented function within each area in parallel, and use the solution as the fixed value for the next iteration after the update Repeatedly update the decision variables within each feeder layer region until the judgment algorithm converges, and output the final Lagrangian augmented function values ​​corresponding to feeder layer region a and feeder layer region b. As the distributed resources of the optimized feeder layer area a and feeder layer area b.

9. The method for distributed resource collaborative control of hierarchical and partitioned autonomy of IoT terminals according to claim 1, characterized in that: The optimized ADMM algorithm adaptively updates the penalty parameter based on the spectral step size obtained according to the gradient descent method; Update the penalty parameter based on the spectral step size as follows: Among them, ψ k+1 is the updated penalty parameter, ψ k is the penalty parameter for this iteration, and Respectively and the correlation of the changes in the auxiliary variables of the Lagrange multiplier, a k and b k F a 、F b The spectral step size is obtained by the mixed step size rule; ε cor is the set threshold, F a 、F b The amount of change.

10. A distributed resource collaborative control system with hierarchical partition autonomy of IoT terminals using the method according to any one of claims 1 to 9, characterized in that: include: Equivalent model building module: used to establish an equivalent model for the substation layer based on the internal autonomy principle; the equivalent model includes feeder-substation layer power exchange conditions, substation target function and substation constraint conditions, with the objective of minimizing the substation target function in the equivalent model, and adjusting the distributed resources of the substation under the conditions of satisfying the feeder-substation layer power exchange conditions and substation constraint conditions; the distributed resources include active power and reactive power; Area division module: used to randomly divide the feeder layer into regions, calculate the modularity index value, active and reactive power balance index value according to the divided feeder layer regions, and merge the three index values ​​into a comprehensive modularity partition index value, and divide the feeder layer into regions according to the comprehensive modularity partition index value; Scheduling model building module: used to establish a scheduling model for the feeder layer based on the areas divided by the feeder layer. The scheduling model includes the objective function of each area of ​​the feeder layer and the constraints of the feeder layer; the objective is to minimize the objective function of each area of ​​the feeder layer in the optimized scheduling model, and to adjust the distributed resources of the feeder layer area while satisfying the power exchange conditions of the feeder-station layer and the constraints of the feeder layer; Optimization module: It is used to perform interval distributed coordination optimization on the scheduling model of the feeder layer and the distributed resources in the feeder layer area by using the optimized alternating direction multiplier ADMM algorithm, and adjust the feeder layer according to the distributed resources in the optimized feeder layer area.

Images