A real-time control method for air-conditioning load in a distribution network, an electronic device and a storage medium

Abstract

The present invention belongs to the technical field related to distribution network optimization and scheduling, and discloses a real-time control method for air-conditioning loads in a distribution network, an electronic device, and a storage medium. The method comprises: step S1: constructing a real-time aggregation control model for air-conditioning loads, including an objective function and constraints, wherein the objective function is to minimize the operating cost of the distribution network, and the constraints include air-conditioning load model constraints, transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints; step S2: solving the real-time aggregation control model for air-conditioning loads to obtain a real-time control decision for air-conditioning loads in the distribution network. The real-time control method for air-conditioning loads in a distribution network proposed by the present invention achieves good results in terms of accuracy, safety, and economy for optimizing the scheduling of HVAC systems with multiple chillers.

Description

Technical Field

[0001] The present invention belongs to the technical field related to distribution network optimization and scheduling, and more specifically, relates to a real-time control method for air-conditioning load in a distribution network, an electronic device, and a storage medium. Background Art

[0002] During the peak summer season, grid loads often remain high for extended periods. This can cause overheating of power transmission and distribution equipment, such as transformers and cables, in some areas, affecting their insulation performance. To ensure the safe operation of distribution network equipment, the power dispatching center will implement load limits to remove some feeder loads. However, such emergency load control measures can severely impact normal economic development and the quality of life of consumers. Furthermore, the uncertainties brought about by the high penetration of renewable energy and the diversification of power loads also increase the difficulty of ensuring safe and economical dispatch of distribution networks.

[0003] As a critical component of the load side, air conditioning loads account for a significant portion of summer peak loads. Therefore, efficient management of air conditioning loads helps ensure a balanced supply and demand during peak periods. Compared to split-type air conditioners, centralized multi-chiller HVAC systems offer continuous power adjustment and large unit capacity, making them widely used in commercial and office buildings. Current research on multi-chiller HVAC systems often fails to consider their energy efficiency and heterogeneity, resulting in suboptimal scheduling.

[0004] Therefore, it is urgent to propose a scheduling scheme for centralized multi-chiller HVAC, which can not only ensure the safety of the distribution network, but also achieve optimized scheduling more accurately and alleviate the contradiction between supply and demand balance during peak power consumption periods in the distribution network. Summary of the Invention

[0005] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a real-time control method for air-conditioning load in a distribution network, an electronic device and a storage medium, the purpose of which is to perform safe and economical scheduling optimization for centralized multi-chiller HVAC, thereby solving the technical problem of balancing supply and demand during peak electricity consumption periods in the distribution network.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for real-time control of air-conditioning load in a distribution network is provided, which comprises:

[0007] Step S1: Construct a real-time aggregation control model for air conditioning loads, including an objective function and constraints. The objective function is to minimize the operating cost of the distribution network. The constraints include air conditioning load model constraints, transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints:

[0008] The air conditioning load model constraint is to substitute the state equation of the virtual energy storage state of charge of a discretized single air conditioning load into the calculation formula for the total working power consumed by the air conditioning load of the corresponding node, and obtain the state equation of the discretized aggregated air conditioning virtual energy storage state of charge after aggregating the power of all air conditioning loads at the node based on a unified control signal. The unified control signal includes an assumption that the virtual energy storage state of charge and the rate of change of the virtual energy storage state of charge of each room served by all air conditioning loads at the corresponding node are the same.

[0009] The transformer thermal circuit model is constrained to be a discretized transformer temperature state equation;

[0010] The cable thermal circuit model is constrained to be a discretized cable temperature state equation;

[0011] Step S2: Solve the real-time aggregation control model of the air-conditioning load to obtain a real-time control decision of the air-conditioning load of the distribution network.

[0012] In one embodiment, obtaining a state equation of a virtual energy storage state of charge of a discretized single air conditioning load includes:

[0013] Substitute the single air-conditioning load virtual energy storage charge state expression into the simplified first-order thermodynamic equivalent model to obtain the state equation of the single air-conditioning load into the virtual energy storage charge state;

[0014] The state equation of the virtual energy storage state of charge of a single air-conditioning load is solved by setting the time step, and the state equation of the virtual energy storage state of charge of the discretized single air-conditioning load is obtained.

[0015] In one embodiment, the air conditioning load model constraint is:

[0016]

[0017] Where, E k,t is the virtual energy storage charge state of the air conditioner at node k during period t, is the total working power consumed by the air conditioning load of node k during period t, k , Β k are the constant coefficients of the state equation of the air-conditioning virtual energy storage state of charge after node k aggregation; R k,t is the time-varying coefficient of the state equation of the air-conditioning virtual energy storage state of charge after node k aggregation, and are the slope and intercept of the optimal linear energy consumption curve of air conditioner n at node k, respectively, α' k,n,r and β' k,n,r are the constant coefficients of the discretized state equation of room r on node k, which is managed by air conditioner n. k,n,r,tis the time-varying coefficient of the discretized state equation of room r at node k, which is managed by air conditioner n. C is the air conditioning load set on node k, N R The collection of rooms responsible for a single air conditioning load.

[0018] In one embodiment, the operating cost of the distribution network in period t includes: the cost of the distribution network purchasing electricity from the main grid in period t System network loss cost System curtailment costs of wind and solar power Comfort adjustment cost for air-conditioning users and system load shedding costs

[0019]

[0020] Where N node 、N branch and N HVAC They are the distribution network node set, branch set and node set containing air conditioning load, The electricity price purchased by the distribution network system from the main grid during period t; is the ratio of node e shedding conventional load during period t; is the normal load power of node e during period t; is the air conditioning load power of node k during period t; P 0 and P k are the no-load loss and short-circuit loss of the transformer respectively; ψ t is the transformer load factor during period t; I ef,t is the current flowing through branch ef during period t, r ef is the resistance value of branch ef; w pv and w wt These are the penalty fees for unit curtailment of solar power and wind power respectively; and are the maximum photovoltaic power generation and the maximum wind power generation during period t respectively; P t pv and P t wt are the actual photovoltaic power generation and wind power generation power in period t; E k,t w is the virtual energy storage charge state of the air conditioning load at node k during period t; pmv is the unit indoor temperature adjustment cost; w cut It is the unit load shedding power cost.

[0021] In one embodiment, solving the real-time aggregation control model for air conditioning load includes:

[0022] The real-time aggregation control model of the air-conditioning load is reconstructed into a Markov decision process, which includes five parts: state variables, decision variables and random variables, transfer function and objective function. The state variables include parameters reflecting the current state information of the distribution network, the decision variables include decision parameters to be solved, the random variables include random factors reflecting the operation process of the distribution network, and the transfer function reflects the transfer relationship between different states of the distribution network.

[0023] A piecewise linear function is used to approximate the value function of the virtual energy storage state of charge of the air-conditioning load of each node in the state variable, and an approximate state value function of the virtual energy storage state of charge of the air-conditioning load of each node is obtained and substituted into the Bellman equation in the Markov decision process;

[0024] Solve the Bellman equation to obtain the global optimal decision of the distribution network.

[0025] In one embodiment: the virtual energy storage charge state E of the air conditioning load at node k in period t k,t The approximate state value function is

[0026]

[0027] Where, is the slope of the mth segment of the virtual energy storage state of charge of the air conditioning load at node k during period t, M is the number of segments of the virtual energy storage state of charge of the air conditioning load at each node, and E m,k,t is the length of the mth segment of the virtual energy storage charge state of the air-conditioning load at node k in period t mapped to the horizontal axis.

[0028] In one embodiment, the piecewise linear function has a slope, and a process of determining the slope includes:

[0029] Step S01: Acquire multiple sets of training scenarios, each set of training scenarios corresponding to a different prediction period, each set of training scenarios including a wind power maximum output curve generated based on the predicted value of wind power output and the corresponding prediction error distribution, a photovoltaic output curve generated based on the predicted value of photovoltaic output and the corresponding prediction error distribution, and a conventional load curve generated based on the predicted value of conventional load and the corresponding prediction error distribution;

[0030] Step S02: Select one of the unselected training scenarios as the current scenario, solve the global optimal decision of the distribution network under the current training scenario, and obtain the slope sampling value of the piecewise linear function during the solution process. The slope sampling value and the slope used for solving the global optimal decision of the distribution network under the current scenario are weighted averaged to obtain the updated slope and substitute it into the slope used for solving the global optimal decision of the distribution network under the next set of training scenarios;

[0031] Step S03: Repeat step S02 until the global optimal decision of the distribution network under the last set of training scenarios is solved, and the slope obtained by solving the last set of training scenarios is used as the slope used by the piecewise linear function.

[0032] In one embodiment, in step S02, the mth segment slope sampling value of the piecewise linear function in the nth iteration t period is The slope of the mth segment of the virtual energy storage state of charge of the air-conditioning load at node k in the nth training period t is

[0033]

[0034]

[0035] Where, ΔE k is the virtual energy storage charge state E of the air conditioning load at node k during period t k,t The differential of The virtual energy storage charge state E is the air conditioning load k,t and the differential ΔE k The approximate state value function of the sum of , ρ is the slope update step size.

[0036] According to another aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0037] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0038] In general, the above technical solutions conceived by the present invention, compared with the prior art, provide a method for real-time control of air-conditioning load in a distribution network, which has the following beneficial effects:

[0039] The present invention provides a real-time control method for air-conditioning loads in a distribution network. The method constructs a real-time aggregation control model for air-conditioning loads. The objective function is to minimize cost, thereby ensuring the economic efficiency of the scheduling scheme. The constraints include, on the one hand, air-conditioning load model constraints, and on the other hand, transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints. The air-conditioning load model constraints take into account the energy efficiency ratio and heterogeneity of HVAC, as well as the fact that HVAC loads have energy storage characteristics similar to conventional energy storage devices. Thus, the air-conditioning is established as a virtual energy storage model, which is more conducive to solving the model and thus can make scheduling more accurate. The transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints ensure the safety of optimized scheduling. Therefore, the real-time control method for air-conditioning loads in a distribution network proposed by the present invention has good results in terms of accuracy, safety, and economy of optimized scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a schematic diagram of the topological structure of a distribution network according to an embodiment of the present application;

[0041] Figure 2 This is a flowchart of the steps of the method for real-time control of air-conditioning load in the distribution network in one embodiment of the present application;

[0042] Figure 3 is a flow chart of the model solving steps in one embodiment of the present application;

[0043] Figure 4 This is a training scenario for wind power generation in one embodiment of the present application;

[0044] Figure 5 This is a photovoltaic power generation training scenario in one embodiment of the present application;

[0045] Figure 6 This is a conventional load training scenario in one embodiment of the present application;

[0046] Figure 7 This is a graph showing the trend of outdoor temperature changes in one embodiment of the present application;

[0047] Figure 8 is a temperature curve of the power transmission and transformation equipment obtained by simulation in an embodiment of the present application;

[0048] Figure 9 It is the benchmark power consumption, actual power consumption and virtual energy storage charge state after HVAC load aggregation obtained by simulation in one embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0050] Example 1

[0051] like Figure 1 The figure shows a schematic diagram of the distribution network topology structure of an embodiment of the present application. The distribution network has a large number of nodes. Some nodes are connected to wind power plants (WT) or photovoltaic power plants (PV), and some nodes are connected to HVAC loads of multiple chillers. The HVAC loads have energy storage characteristics similar to conventional energy storage devices. A node may be connected to multiple HVAC loads, and a single HVAC load may be responsible for temperature control of multiple rooms. Each node is also connected to conventional loads, which are loads other than HVAC loads of multiple chillers. The high penetration rate of new energy, the diversification of power loads, and the fluctuations in power consumption and electricity prices during different periods of time will bring uncertainty to regulation, thereby increasing the difficulty of safe and economical scheduling of the distribution network.

[0052] Based on this, the present invention proposes a real-time control method for air-conditioning load in a distribution network, which can achieve more economical and reliable scheduling while ensuring the security of the power grid.

[0053] like Figure 2 The figure shows a flow chart of the steps of the real-time control method of the air-conditioning load of the distribution network in one embodiment of the present application, which mainly includes the following steps:

[0054] Step S1: Construct a real-time aggregation control model for air conditioning loads, including an objective function and constraints. The objective function is to minimize the operating cost of the distribution network. The constraints include air conditioning load model constraints, transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints.

[0055] The air conditioning load model constraint is to substitute the state equation of the virtual energy storage charge state of a discretized single air conditioning load into the total working power equation consumed by the air conditioning load of the corresponding node, and obtain the state equation of the discretized aggregated air conditioning virtual energy storage charge state after aggregating the nodes based on a unified control signal. The unified control signal includes the assumption that the virtual energy storage charge state and the rate of change of the virtual energy storage charge state of each room served by all air conditioning loads at the corresponding node are the same.

[0056] The transformer thermal circuit model is constrained to be a discretized transformer temperature state equation;

[0057] The cable thermal circuit model is constrained to be a discretized cable temperature state equation;

[0058] Step S2: Solve the real-time aggregation control model of air-conditioning load to obtain the real-time control decision of air-conditioning load in the distribution network.

[0059] It is understandable that before modeling, it is necessary to collect the technical parameters of the HVAC load and the two types of power transmission and transformation equipment in the distribution network: transformers and cables.

[0060] HVAC load parameters include the capacity of each building chiller at each node, the number of rooms served by a single HVAC unit, the room equivalent thermal resistance, room equivalent heat capacity, room equivalent thermal resistance, room temperature range, and room temperature variation range.

[0061] The technical parameters of the transformer include rated capacity, rated voltage, no-load loss, short-circuit loss, hot spot temperature upper limit, and initial top oil temperature.

[0062] The technical parameters of the cable include cable conductor resistance, conductor temperature upper limit, initial conductor temperature

[0063] The operating costs of the distribution network can be flexibly selected based on actual conditions. In one embodiment, the operating costs include the cost of the distribution network purchasing electricity from the main grid, system network loss costs, system wind and solar curtailment costs, air conditioning user comfort adjustment costs, and system load shedding costs.

[0064] Specifically, the objective function can be:

[0065]

[0066] Where, T is the distribution network optimization period; C is the total operating cost of the distribution network within the distribution network optimization period T; C t is the operating cost of the distribution network during period t; and are the electricity purchase cost of the distribution network from the main grid, the system network loss cost, the system wind and solar curtailment cost, the HVAC user comfort adjustment cost, and the system load shedding cost during period t; N node 、N branch and N HVAC They are the distribution network node set, branch set and node set containing HVAC loads; The electricity price purchased by the distribution network system from the main grid during period t; is the load shedding ratio of node e in period t; is the normal load power of node e during period t; is the HVAC load power of node k during period t; P 0 and P k are the no-load loss and short-circuit loss of the transformer respectively; ψ t is the transformer load factor during period t; I ef,tis the current flowing through branch ef during period t, r ef is the resistance value of branch ef; w pv and w wt These are the penalty fees for unit curtailment of solar power and wind power respectively; and are the maximum photovoltaic power generation and the maximum wind power generation during period t respectively; P t pv and P t wt are the actual photovoltaic power generation and wind power generation power in period t; E k,t is the virtual energy storage charge state of the HVAC load at node k during period t; w pmv is the unit indoor temperature adjustment cost; w cut The unit load shedding power cost. Among the above parameters, the load shedding ratio Air conditioning load power Actual photovoltaic power generation P t pv and actual wind power generation power P t wt is the decision variable to be decided.

[0067] The following uses air conditioning and refrigeration as an example to introduce the process of building air conditioning load model constraints.

[0068] Step S11 : establishing an optimal linear energy consumption curve of HVAC considering the load distribution strategy.

[0069] The general form of the HVAC optimal linear energy consumption curve is:

[0070]

[0071] Where, and P t c are the total cooling capacity and total power of HVAC during period t, k c and l c are the slope and intercept of the fitted curve, respectively.

[0072] In one embodiment, the process of determining the slope and intercept of the optimal linear energy consumption curve is as follows:

[0073] Step S111: Establishing a HVAC line energy consumption model considering the multi-chilled water unit load distribution strategy:

[0074]

[0075] Where q i,t and p i,t are the cooling capacity and cooling power of the i-th HVAC unit in period t, respectively, i,nare the coefficients of the fitted polynomial; is the rated cooling capacity of the i-th chiller; N I is the chiller set in the HVAC system; θ i,t is the start and stop status of the i-th chiller in period t.

[0076] Step S112: Solving the above equation can obtain the minimum cooling power consumed by a single HVAC at all cooling capacity sampling points. The least squares method is used for linear fitting to obtain the optimal linear energy consumption curve of the HVAC within the rated capacity.

[0077] Step S12: Establishing a state equation of the virtual energy storage charge state of a single HVAC load.

[0078] It includes the following sub-steps:

[0079] Step S121: Use a simplified first-order thermodynamic equivalent model to describe the temperature change inside the room cooled by the HVAC:

[0080]

[0081] Among them, T t o and T t i are the outdoor and indoor temperatures during period t; C room and R room are the equivalent heat capacity and thermal resistance of the room respectively; The cooling capacity of a single room.

[0082] Step S122: defining the virtual energy storage charge state of a single HVAC load as:

[0083]

[0084] in, T is the virtual energy storage charge state of a single HVAC load during period t; max and T min are the upper and lower limits of the indoor temperature respectively; ΔT is the room temperature variation range; T set Set the temperature for the room.

[0085] Step S123: Substitute the definition of the virtual energy storage state of charge of a single HVAC load into the simplified first-order thermodynamic equivalent model to derive the state equation of the virtual energy storage state of charge of a single HVAC load:

[0086]

[0087] Among them, α and β are the constant coefficients of the state equation; γ t The time-varying coefficients of the equation of state.

[0088] Step S13: discretizing the state equation of the virtual energy storage charge state of a single HVAC load.

[0089] By setting the time step, the state equation of the virtual energy storage state of charge of a single HVAC load is solved to obtain the state equation of the virtual energy storage state of charge of the discretized HVAC load:

[0090]

[0091] in, is the virtual energy storage charge state of a single HVAC load in period t-1; α' and β' are the constant coefficients of the discretized state equation; γ t ' is the time-varying coefficient of the discretized state equation; Δt is the time step.

[0092] Step S14: Calculate the total cooling power of all rooms in the same node after aggregation.

[0093] The cooling power consumed by a single HVAC unit is:

[0094]

[0095] Where N R The set of rooms that are responsible for a single HVAC load, is the cooling power consumed by room r during period t.

[0096] Assume N C is the HVAC load set of multiple chillers at node k, and the total cooling power after the HVAC load aggregation of node k can be expressed as:

[0097]

[0098] Where n is the HVAC load index on node k; and are the cooling power and cooling capacity consumed by room r cooled by HVAC n at node k in period t, respectively; and are the slope and intercept of the optimal linear energy consumption curve of HVAC n at node k, respectively.

[0099] Step S15: Substitute the state equation of the virtual energy storage charge state of the discretized single HVAC load into the total cooling power and aggregate it based on the unified control signal to obtain the state equation of the discretized aggregated air conditioning virtual energy storage charge state of the corresponding node, which is the air conditioning load model constraint.

[0100] The unified control signal is: all rooms on node k that are cooled by HVAC will receive the same signal control, that is, the and Should be controlled to the same value:

[0101]

[0102] in, and is the unified control signal of node k, is the virtual energy storage charge state of room r cooled by HVAC n at node k during period t.

[0103] Combining equations (7), (9), and (10), we can obtain the discretized form of the state equation of the aggregated virtual energy storage state of charge, which is the HVAC load model constraint:

[0104]

[0105] Among them, E k,t is the virtual energy storage charge state of the HVAC node k during period t; E k,t-1 is the virtual energy storage charge state of the HVAC node k in period t-1; is the total cooling power consumed by the HVAC load at node k during period t; α' k,n,r and β' k,n,r is the constant coefficient of the discretized state equation of room r cooled by HVAC n at node k; γ' k,n,r,t is the time-varying coefficient of the discretized state equation of room r cooled by HVAC n at node k; k and Β k is the constant coefficient of the state equation of the HVAC virtual energy storage state of charge after node k aggregation; R k,t is the time-varying coefficient of the state equation of the HVAC virtual energy storage state of charge after node k aggregation.

[0106] When establishing the discretized equivalent thermal circuit model of the transformer and the cable in step S1, it is necessary to construct the transformer temperature state equation and the cable temperature state equation according to the physical structure of the transformer and the cable, and discretize them.

[0107] Specifically, based on the physical structure of the transformer, the state equation of the transformer top oil temperature during period t can be listed as follows:

[0108]

[0109]

[0110] in, and are the transformer top oil temperature, bottom oil temperature and ambient temperature during period t; R o The equivalent thermal resistance of the transformer winding heat transfer to the underlying insulating oil; Cth is the equivalent heat capacity of transformer insulating oil, winding and oil tank; R bt The equivalent thermal resistance for heat transfer between the top and bottom layers of insulating oil is R ath is the equivalent thermal resistance on the air side; P 0 and P k are the no-load loss and short-circuit loss of the transformer respectively; ψ t is the transformer load factor during period t; x is the transformer oil index; is the heat generated by the transformer during operation during period t.

[0111] Transformer hot spot temperature Can be expressed as the average winding temperature Top oil temperature and average oil temperature The relevant function formula is calculated as follows:

[0112]

[0113] Where H is the hot spot temperature coefficient of the transformer. and ambient temperature θ t a Can calculate the average transformer winding temperature and bottom oil temperature

[0114]

[0115] Discretizing the state equation of the transformer top oil temperature during period t, we can obtain:

[0116]

[0117] According to the physical structure of the cable, the state equation of the cable conductor temperature during period t can be listed:

[0118]

[0119]

[0120] Among them, C line is the equivalent heat capacity of the cable; R line is the equivalent thermal resistance of the cable; is the cable conductor temperature during period t; I is the heat generated by the cable during system operation during period t; t is the current flowing through the cable during period t; r is the resistance of the cable conductor. Discretizing the state equation for the temperature of the cable conductor during period t yields:

[0121]

[0122] Transformer hot spot temperature upper limit constraint:

[0123]

[0124] in, is the upper limit of the transformer hot spot temperature.

[0125] Cable conductor temperature upper limit constraint:

[0126]

[0127] in, The upper limit of cable conductor temperature.

[0128] In one embodiment, the above-mentioned real-time aggregation and control model for air conditioning loads may also be configured with other related constraints, such as HVAC load constraints, new energy output constraints, power flow constraints, voltage constraints, and node power balance constraints.

[0129] More specifically, the HVAC load constraints:

[0130]

[0131] Among them, E k,t is the virtual energy storage charge state of the HVAC node k during period t.

[0132] Load shedding constraints:

[0133]

[0134] New energy output constraints:

[0135]

[0136] For the branch ef connecting nodes e and f, the power flow constraint is:

[0137]

[0138] Where, e,f∈N node , ef∈N branch ;x ef and r ef are the reactance and resistance of branch ef respectively; v e,t and v f,t are the squares of the voltage amplitudes at nodes e and f during period t; i ef,t is the square of the current amplitude of branch ef during period t; P ef,t and Q ef,t are the active and reactive powers flowing through the head end of branch ef during period t; P in,f,t and Q in,f,tare the active and reactive powers injected into node f during period t, respectively; j:f→j represents the child node of node f.

[0139] Voltage Constraints:

[0140]

[0141] Among them, V e,t is the voltage amplitude of node e during period t; V max and V min are the upper and lower limits of the distribution network voltage amplitude respectively.

[0142] Power balance constraints:

[0143]

[0144] in, and are the active and reactive powers injected into node f by the upper power grid during period t, respectively; is the reactive load power of node f during period t.

[0145] After establishing the real-time aggregation control model of air-conditioning load, the model can be solved by using existing methods such as model predictive control method and short-sighted method to obtain the optimal decision.

[0146] Example 2

[0147] This embodiment proposes a specific model solving process, introduces the Markov decision process, and utilizes the powerful optimization ability of the Markov decision process in a random environment to quickly find the optimal decision with higher decision accuracy than other conventional methods.

[0148] like Figure 3 Shown is a flow chart of the model solving steps in one embodiment of the present application, which includes the following steps.

[0149] Step S21: Reconstruct the real-time aggregation control model of air-conditioning load into a Markov decision process. The Markov decision process includes five parts: state variables, decision variables and random variables, transfer function and objective function: state variables include parameters reflecting the current distribution network state information, decision variables include decision parameters to be solved, random variables include random factors reflecting the operation process of the distribution network, and transfer function reflects the transfer relationship between different states of the distribution network.

[0150] Specifically, the state variable S t It is used to reflect the current state of the distribution network system. This paper expresses it as the maximum wind and solar output value during period t. and Conventional active load power of each node Virtual energy storage charge state E of HVAC load at each node k,t and the time-varying coefficient R of the HVAC load state equation k,t ,Right now:

[0151]

[0152] Decision variable x t Reflects the decision made by the system based on the status information during period t, including the actual wind and solar power output P during period t t wt and P t pv , HVAC cooling power of each node and the ratio of conventional load shedding at each node Right now:

[0153]

[0154] Random variables are used to describe random factors in the operation of the distribution network, including wind and solar power output forecast errors. and Forecast error of conventional load Right now:

[0155]

[0156] State transition function S t+1 =f(S t ,x t ,W t+1 ) is used to reflect the transfer relationship between different states of the distribution network system, namely:

[0157]

[0158] Among them, z is the index; S t+1 (z), W t+1 (z), S t (z), x t (z) are S t+1 、 W t+1 、S t 、x t the zth element of ; is the day-ahead forecast value of wind and solar power output and conventional load. k and Β k is the constant coefficient of the HVAC load state equation.

[0159] The objective function of air conditioning load control in the distribution network at time t is:

[0160]

[0161] Among them, C t (S t ,x t ,W t ) is the distribution network objective function; and They are respectively the cost of the distribution network purchasing electricity from the main grid during period t, the system network loss cost, the system load shedding cost, the HVAC user comfort adjustment cost and the system wind and solar power curtailment cost.

[0162] Step S22: Use a piecewise linear function to approximate the value function of the virtual energy storage charge state of the air-conditioning load of each node in the state variable, obtain the approximate state value function of the virtual energy storage charge state of the air-conditioning load of each node and substitute it into the Bellman equation in the Markov decision process.

[0163] Firstly, a piecewise linear function is used to calculate the virtual energy storage state of charge E of each node HVAC load in the Markov decision process state variable. k,t Perform state-value function approximation:

[0164]

[0165] in, is the set of state variables after decision making; is the state value function after the decision in period t; is the approximate state value function during period t; is the approximate state value function related to the virtual energy storage charge state of the HVAC load; M is the number of segments of the virtual energy storage charge state of the HVAC load at each node; is the slope of the mth segment of the virtual energy storage state of charge at node k in period t; E m,k,t is the length of the mth segment of the virtual energy storage charge state of node k in period t mapped to the horizontal axis.

[0166] Then, substitute the approximate state value function into the Bellman equation in the Markov decision process:

[0167]

[0168] Among them, V t (S t )and are the state value functions before and after decision making respectively; is the set of state variables after decision making.

[0169] Step S23: Solve the Bellman equation to obtain the global optimal decision of the distribution network in the corresponding time period.

[0170] In one embodiment, all the above optimization problems are solved using the Gurobi solver on the Matlab platform.

[0171] In one embodiment, the slope of the piecewise linear function may be determined by training. The specific training steps are as follows.

[0172] Step S01: Obtain multiple groups of training scenarios, each group of training scenarios corresponds to a different prediction period, and each group of training scenarios includes a wind power output curve generated based on the predicted value of wind power output and the corresponding prediction error distribution, a photovoltaic output curve generated based on the predicted value of photovoltaic output and the corresponding prediction error distribution, and a conventional load curve generated based on the predicted value of conventional load and the corresponding prediction error distribution.

[0173] Assume that the forecast errors of wind power output and conventional load obey normal distribution, and their specific distributions are N(0,0.05 2 ), the Monte Carlo sampling method is used to generate 200 sets of offline training scenarios for training the slope of the piecewise linear function. The trained piecewise linear function slope will be used for subsequent online optimization.

[0174] In the above training scenarios, the wind power generation training scenario is as follows: Figure 4 As shown in Figure 2, the training scenario for photovoltaic power generation is as follows: Figure 5 As shown, the training scenario of conventional load is as follows Figure 6 shown.

[0175] During online optimization, the same method is used to generate test scenarios to simulate the changes in random factors within a day.

[0176] Step S02: Select one of the unselected training scenarios as the current scenario, solve the global optimal decision of the distribution network under the current training scenario, and obtain the slope sampling value of the piecewise linear function during the solution process. The slope sampling value and the slope used to solve the global optimal decision of the distribution network under the current scenario are weighted averaged to obtain the updated slope and iterate into the slope used to solve the global optimal decision of the distribution network under the next set of training scenarios.

[0177] In one embodiment, the mth segment slope sampling value of the piecewise linear function in the nth iteration t period is The slope of the mth segment of the virtual energy storage state of charge of the air-conditioning load at node k in the nth training period t is

[0178]

[0179] Where, ΔE k is the virtual energy storage charge state E of the air conditioning load at node k during period t k,t The differential of The virtual energy storage charge state E is the air conditioning loadk,t and the differential ΔE k The approximate state value function of the sum of , ρ is the slope update step size.

[0180] Step S03: Repeat step S02 until the global optimal decision of the distribution network under the last set of training scenarios is solved, and the slope obtained by solving the last set of training scenarios is used as the slope of the piecewise linear function.

[0181] Due to the uncertainty on both the source and load sides in the optimal scheduling of distribution networks, it is necessary to search for the best answer in this random environment. By utilizing the powerful optimization ability of the Markov decision process in a random environment, the optimal decision can be found quickly and the decision accuracy is higher than other conventional methods.

[0182] The improved IEEE-33 node system is simulated and verified. The specific topology is as follows Figure 1 shown.

[0183] The first step is to collect the technical parameters of HVAC loads and two types of power transmission and transformation equipment in the distribution network: transformers and cables.

[0184] In this embodiment, the capacity of the building chiller for each HVAC load at each node is shown in Table 1:

[0185] Table 1

[0186]

[0187] In this embodiment, the room parameters are shown in Table 2:

[0188] Table 2

[0189] parameter Numerical Number of rooms in a single building 50 Room equivalent thermal resistance / (℃ / kW) 1.5-2.5 Room equivalent heat capacity / (kWh / ℃) 1.5-2.5 Room temperature setting / (℃) 22-24 Room temperature variation range / (℃) 0-3

[0190] In this embodiment, the technical parameters of the two power transmission and transformation equipment, transformers and cables, are shown in Table 3:

[0191] Table 3

[0192]

[0193]

[0194] In this embodiment, the electricity prices for each period are shown in Table 4:

[0195] Table 4

[0196] name Time Electricity price / yuan Valley Section 22:00-07:00 the next day 0.27 Flat section 07:00-10:00、14:00-17:00 0.70 peak 12:00-14:00、17:00-20:00 1.15 peak 10:00-12:00、20:00-22:00 1.30

[0197] In this embodiment, the outdoor temperature change trend is as follows: Figure 7 shown.

[0198] In the second step, a real-time aggregation control model for air-conditioning load is constructed and solved according to the description of the above embodiment to obtain a real-time control decision for air-conditioning load in the distribution network.

[0199] The optimal decision is used to simulate the temperature curve of the power transmission and transformation equipment. Figure 8 As shown in Figure 2, after taking into account the temperature constraints of the power transmission and transformation equipment, the distribution network regulates the HVAC load power on the load side to reduce the risk of overheating of the power transmission and transformation equipment. Taking node 25 as an example, the baseline power consumption, actual power consumption and virtual energy storage charge state after the HVAC load aggregation of this node are shown in Figure 2. Figure 9 As shown. Figure 9 It can be seen that, on the one hand, the HVAC load of this node is cooled in advance during the two periods of 07:00-08:00 and 17:00-18:00, reducing power consumption during the high electricity price period and thus reducing operating costs; on the other hand, power is reduced during the two periods of 12:00-15:00 and 18:00-22:00 to reduce load peaks, prevent the temperature of the transmission and transformation equipment from exceeding the limit, and ensure the safe and stable operation of the distribution network.

[0200] Example 3

[0201] The present invention also relates to an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0202] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The processor may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The memory may be used to store computer programs and / or modules, and the processor may perform various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory.

[0203] Example 4

[0204] The present invention also relates to a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when the computer program is executed by a processor.

[0205] Specifically, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0206] The technical features of the above embodiments can be combined in any combination. To keep the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. It should be noted that the phrases "in one embodiment", "for example", "and another example" in this application are intended to illustrate this application and are not intended to limit this application.

[0207] The above embodiments merely illustrate several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art could make numerous variations and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application.

Claims

1. A real-time control method for air conditioning load in a distribution network, characterized in that: include: Step S1: Construct a real-time aggregation control model for air conditioning loads, including an objective function and constraints. The objective function is to minimize the operating cost of the distribution network. The constraints include air conditioning load model constraints, transformer thermal circuit model constraints, cable thermal circuit model constraints, transformer hot spot temperature upper limit constraints, and cable conductor temperature upper limit constraints: The air conditioning load model constraint is to substitute the state equation of the virtual energy storage state of charge of a discretized single air conditioning load into the total working power calculation formula consumed by the air conditioning load of the corresponding node, and aggregate the power of all air conditioning loads at the node based on a unified control signal to obtain the state equation of the discretized aggregated air conditioning virtual energy storage state of charge. The unified control signal includes the assumption that the virtual energy storage state of charge and the rate of change of the virtual energy storage state of charge of each room responsible for all air conditioning loads at the corresponding node are the same. The air conditioning load model constraint is specifically: Where, E k,t is the virtual energy storage charge state of the air conditioner at node k during period t, is the total working power consumed by the air conditioning load of node k during period t, A k 、B k are the constant coefficients of the state equation of the air-conditioning virtual energy storage state of charge after node k aggregation; R k,t is the time-varying coefficient of the state equation of the air-conditioning virtual energy storage state of charge after node k aggregation, and are the slope and intercept of the optimal linear energy consumption curve of air conditioner n at node k, respectively, α' k,n,r and β' k,n,r are the constant coefficients of the discretized state equation of room r on node k, which is managed by air conditioner n. k,n,r,t is the time-varying coefficient of the discretized state equation of room r at node k, which is managed by air conditioner n. C is the air conditioning load set on node k, N R The set of rooms responsible for a single air conditioning load, is the cooling power consumed by room r cooled by HVAC n at node k during period t; The transformer thermal circuit model is constrained to be a discretized transformer temperature state equation; The cable thermal circuit model is constrained to be a discretized cable temperature state equation; Step S2: Solve the real-time aggregation control model of the air-conditioning load to obtain a real-time control decision for the air-conditioning load of the distribution network.

2. The method for real-time control of air-conditioning load in a distribution network according to claim 1, characterized in that: Obtain the state equation of the virtual energy storage charge state of a discretized single air conditioning load, including: Substitute the single air-conditioning load virtual energy storage charge state expression into the simplified first-order thermodynamic equivalent model to obtain the state equation of the single air-conditioning load into the virtual energy storage charge state; The state equation of the virtual energy storage state of charge of a single air-conditioning load is solved by setting the time step, and the state equation of the virtual energy storage state of charge of the discretized single air-conditioning load is obtained.

3. The method for real-time control of air conditioning load in a distribution network according to claim 1, characterized in that: The operating cost of the distribution network in period t includes: the cost of the distribution network purchasing electricity from the main grid in period t System network loss cost System curtailment costs of wind and solar power Comfort adjustment cost for air-conditioning users and system load shedding costs Where N node 、N branch and N HVAC They are the distribution network node set, branch set and node set containing air conditioning load, The electricity price purchased by the distribution network system from the main grid during period t; is the ratio of node e shedding conventional load during period t; is the normal load power of node e during period t; is the air conditioning load power of node k during period t; P 0 and P k are the no-load loss and short-circuit loss of the transformer respectively; ψ t is the transformer load factor during period t; I ef,t is the current flowing through branch ef during period t, r ef is the resistance value of branch ef; w pv and w wt These are the penalty fees for unit curtailment of solar power and wind power respectively; and are the maximum photovoltaic power generation and the maximum wind power generation during period t respectively; P t pv and P t wt are the actual photovoltaic power generation and wind power generation power in period t; E k,t is the virtual energy storage charge state of the air conditioning load at node k during period t; w pmv is the unit indoor temperature adjustment cost; w cut It is the unit load shedding power cost.

4. The method for real-time control of air conditioning load in a distribution network according to claim 1, wherein: Solving the real-time aggregation control model of the air conditioning load includes: The real-time aggregation control model of the air-conditioning load is reconstructed into a Markov decision process, which includes five parts: state variables, decision variables and random variables, transfer function and objective function. The state variables include parameters reflecting the current state information of the distribution network, the decision variables include decision parameters to be solved, the random variables include random factors reflecting the operation process of the distribution network, and the transfer function reflects the transfer relationship between different states of the distribution network. A piecewise linear function is used to approximate the value function of the virtual energy storage state of charge of the air-conditioning load of each node in the state variable, and an approximate state value function of the virtual energy storage state of charge of the air-conditioning load of each node is obtained and substituted into the Bellman equation in the Markov decision process; Solve the Bellman equation to obtain the global optimal decision of the distribution network.

5. The method for real-time control of air-conditioning load in a distribution network according to claim 4, characterized in that: The virtual energy storage charge state E of the air conditioning load at node k during period t k,t The approximate state value function is Where, is the slope of the mth segment of the virtual energy storage state of charge of the air conditioning load at node k during period t, M is the number of segments of the virtual energy storage state of charge of the air conditioning load at each node, and E m,k,t is the length of the mth segment of the virtual energy storage charge state of the air-conditioning load at node k in period t mapped to the horizontal axis.

6. The method for real-time control of air-conditioning load in a distribution network according to claim 4, characterized in that: The piecewise linear function has a slope, and a process of determining the slope includes: Step S01: Acquire multiple sets of training scenarios, each set of training scenarios corresponding to a different prediction period, each set of training scenarios including a wind power maximum output curve generated based on the predicted value of wind power output and the corresponding prediction error distribution, a photovoltaic maximum output curve generated based on the predicted value of photovoltaic output and the corresponding prediction error distribution, and a conventional load curve generated based on the predicted value of conventional load and the corresponding prediction error distribution; Step S02: Select one of the unselected training scenarios as the current scenario, solve the global optimal decision of the distribution network under the current training scenario, and obtain the slope sampling value of the piecewise linear function during the solution process. The slope sampling value and the slope used for solving the global optimal decision of the distribution network under the current scenario are weighted averaged to obtain the updated slope and substitute it into the slope used for solving the global optimal decision of the distribution network under the next set of training scenarios; Step S03: Repeat step S02 until the global optimal decision of the distribution network under the last set of training scenarios is solved, and the slope obtained by solving the last set of training scenarios is used as the slope used by the piecewise linear function.

7. The method for real-time control of air-conditioning load in a distribution network according to claim 6, characterized in that: In step S02, the mth segment slope sampling value of the piecewise linear function in the nth iteration t period is The slope of the mth segment of the virtual energy storage state of charge of the air-conditioning load at node k in the nth training period t is Where, ΔE k is the virtual energy storage charge state E of the air conditioning load at node k during period t k,t The differential of The virtual energy storage charge state E is the air conditioning load k,t and the differential ΔE k The approximate state value function of the sum of , ρ is the slope update step size.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images