Distributed coordinated control method of residential loads considering dynamic hot spot temperature of transformer

Abstract

The present invention relates to a distributed coordinated control method for resident loads considering the dynamic hot spot temperature of transformers, comprising the following steps: 1) constructing a rolling optimization model for resident load electricity consumption according to a transformer aging model and a resident load model respectively; 2) modularizing the centralized problem of the rolling optimization model for resident load electricity consumption, introducing a Lagrange multiplier vector, and using a distributed algorithm to solve the centralized problem, and finally obtaining an optimal coordinated control scheme. Compared with the prior art, the present invention has the advantages of improving the economic efficiency of distribution network operation, ensuring the safe operation of transformers, intelligent dynamic congestion management, no need to collect load privacy data, adapting to the coordinated control of large-scale loads, and reflecting the differences in loads.

Description

Technical Field

[0001] The invention relates to the field of coordinated control of residential loads, and in particular to a distributed coordinated control method of residential loads taking into account the dynamic hot spot temperature of a transformer. Background Art

[0002] With the steady growth of air-conditioning load and the explosive growth of electric vehicle (EV) load, the seasonal peak load and periodic peak load problems of the power grid have become more prominent, and the congestion of the distribution network has become more serious. Coordinated control of flexible loads on the user side is an economical and effective means to solve the congestion problem. However, most of the existing congestion management only takes the transformer capacity as a fixed and limited target without considering its aging factor. In fact, the life loss of the transformer is mainly related to its winding hot-spot temperature (HST), and HST has a complex nonlinear relationship with the ambient temperature and the transformer's historical load. On the one hand, when the transformer HST is low, its overload capacity can be reasonably utilized; on the other hand, when the HST is too high, even if the transformer is not overloaded, it will accelerate aging.

[0003] With the continuous increase in EV penetration, researchers have begun to pay attention to the impact of EV centralized charging on transformer life in recent years. The existing literature "The impact of electric vehicle access on the service life of distribution transformers, Sun Jianlong, Li Yan, Hu Guowei, et al." proposed an intelligent charging strategy based on particle swarm algorithm to minimize the life loss of transformers while meeting the charging needs of users; the literature "Affonso CDM, Kezunovic M. Technical and Economic Impact of PV-BESS Charging Station on Transformer Life: A Case Study" reduces the transformer overload problem in commercial building parking lots with photovoltaic access by coordinating EV and battery energy storage; the literature "Gray MK, Morsi W G. On the role of prosumers owning rooftop solar photovoltaic in reducing the impact on transformer's aging due to plug-in electric vehicles charging" studies the effect of configuring distributed photovoltaics in residential buildings on improving transformer aging caused by EV charging. However, the above studies only analyze the aging of transformers a posteriori, and do not directly incorporate it into the optimization model, so it is difficult to achieve optimal charging management.

[0004] In order to achieve the overall optimization of transformer life loss and EV users, the paper "Olsen DJ, Sarker MR, Ortega-Vazquez M A. Optimal Penetration of Home Energy Management Systems in Distribution Networks Considering Transformer Aging" proposes a home energy management system that can extend the life of the transformer and reduce the operating cost of the distribution network; the paper "Sarker MR, Olsen DJ, Ortega-Vazquez M A. Co-Optimization of Distribution Transformer Aging and Energy Arbitrage Using Electric Vehicles" proposes a centralized smart charging strategy to reduce transformer aging; the paper "Hilshey AD, Hines PDH, Rezaei P, et al. Estimating the Impact of Electric Vehicle Smart Charging on Distribution Transformer Aging" combines the transformer thermal aging model with EV vehicle behavior and proposes a smart charging strategy to alleviate transformer aging caused by centralized EV charging. However, the above methods have the following common problems:

[0005] (1) The control strategy only considers EVs and does not coordinate with other flexible loads, especially air conditioning loads. In fact, there is a peak overlap between air conditioning loads and charging loads during certain periods in summer, so it is necessary to give full play to the flexibility of air conditioning loads;

[0006] (2) Using a centralized control structure. Although the centralized structure is convenient for using the complete information of the entire system to obtain the best solution, for a large number of user-side resources, this method has prominent user privacy issues and scalability issues. Summary of the invention

[0007] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a distributed coordinated control method for residential loads taking into account the dynamic hot spot temperature of the transformer. This method does not need to collect load privacy data, and can give full play to the distributed computing capabilities of each load to solve the scale problem of user-side resources.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers comprises the following steps:

[0010] 1) Construct a rolling optimization model for residential load electricity consumption based on the transformer aging model and the residential load model respectively;

[0011] 2) The centralized problem of the rolling optimization model of residential load electricity consumption is modularized and the Lagrange multiplier vector is introduced. A distributed algorithm is used to solve the centralized problem and finally the optimal coordinated control scheme is obtained.

[0012] In step 1), the objective function of the rolling optimization model for residential load electricity consumption is:

[0013]

[0014] Among them, P T,t is the transformer power, μ t For electricity price, is the aging cost of the transformer, ω(s i,t ) is the loss penalty term related to load satisfaction, M is the length of the rolling optimization time domain, N is the number of resident loads, and the subscript i is the resident load number. is the dynamic hot spot temperature HST of the winding, t is the scheduling period, and t0 is the starting time of the scheduling period.

[0015] The constraints of the residential load electricity rolling optimization model include:

[0016] Power balance constraints:

[0017]

[0018] Transformer operation constraints:

[0019]

[0020] Residential load constraints:

[0021]

[0022] Among them, P t base is the predicted value of base load output, P t RES is the predicted value of renewable energy output, η is the transformer efficiency, is the maximum short-time overload power of the transformer, is the upper limit of the winding dynamic hot spot temperature HST, is the feasible region of power for residential loads, P i,t The power of residential load.

[0023] In the transformer aging model described above, the transformer aging cost The expression is:

[0024]

[0025]

[0026]

[0027] Among them, C T is the transformer cost, L exp is the expected life of the transformer, V t is the relative aging rate of the transformer, is the top oil temperature, is the temperature rise of the hot spot relative to the top oil.

[0028] The top oil temperature and the temperature rise of the hot spot relative to the top oil There is an internal temperature accumulation process, and the dynamic equations are:

[0029]

[0030]

[0031] in, is the transformer load factor, is the transformer rated power, is the top oil temperature at time t-1, R is the ratio of load loss to no-load loss at rated current, x is the exponential power of total loss relative to the top oil temperature rise of the tank, Δθ or is the steady-state temperature rise of the hot spot relative to the environment under rated load, is the ambient temperature, ΔT is the control period, τ0 is the oil time constant, is the temperature rise of the hot spot relative to the top oil at time t-1, y is the exponential power of the current relative to the winding temperature rise, Δθ hr is the temperature rise of the hot spot relative to the top oil at rated load, To consider the basic hot spot temperature rise before the effect of the changing oil flow through the hot spot, To consider the hot spot temperature rise caused by the oil flow rate change through the hot spot, and to meet τ w is the winding time constant, k 11 , k 21 , k 22 is the thermal model constant.

[0032] In the residential load model, residential loads include air conditioning loads and EVs.

[0033] For air conditioning load, the corresponding power is expressed as:

[0034] P i,t =a i s i,t+1 +b i s i,t +c i,t

[0035]

[0036]

[0037] Among them, a i <0, b i , c i,t >0 is the parameter, s i,t 、s i,t+1 are the normalized load satisfaction at time t and t+1, and θ i are the upper and lower limits of the room temperature set by the user, θ i,t ,θ i,t+1 is the room temperature of the air conditioning load at time t and t+1, θ t,a is the ambient temperature at time t, R i,a , C i,a are the equivalent thermal resistance and thermal capacity respectively, ΔT is the control period, Q i,t is the cooling capacity of the air conditioner at time t;

[0038] Its loss penalty term ω(s) related to load satisfaction at time t i,t ) is:

[0039]

[0040]

[0041]

[0042] Among them, w is the weight coefficient, are the upper and lower limits of the air conditioning power at time t, is the average main grid electricity price in M ​​time periods starting from time t, is the maximum power of the air conditioner, which can be taken as its rated power;

[0043] The power feasible region of the air conditioning load is expressed as:

[0044]

[0045] For EV, its power feasible region is expressed as:

[0046]

[0047] Among them, P i,N is the EV rated power, t i,a is the time to start charging, t i,d For departure time, is the target power, The power level at the start of charging.

[0048] The step 2) specifically comprises the following steps:

[0049] 21) Introduce the Lagrange multiplier vector λ to construct the Lagrange function, then:

[0050]

[0051] Among them, λ t is the Lagrange multiplier at time t;

[0052] 22) Get the dual function f(λ) of the centralized problem, then:

[0053]

[0054]

[0055]

[0056]

[0057]

[0058] Among them, f T (λ) is the optimization sub-problem of the transformer, f i (λ) is the optimization sub-problem of resident load i;

[0059] 23) Given the Lagrange multipliers, f T (λ) and f i (λ) is solved to obtain the optimal solution of the subproblem at the current step and Based on this, the subgradient method is used to update the Lagrange multiplier, and the solution is obtained by rolling iteration, and finally the optimal coordinated control solution is obtained.

[0060] In the step 23), the expression for updating the Lagrange multiplier is:

[0061]

[0062] Among them, λ t + is the updated Lagrange multiplier.

[0063] Compared with the prior art, the present invention has the following advantages:

[0064] 1. The maximum load is determined dynamically based on the hot spot temperature of the transformer winding, rather than using a fixed transformer capacity constraint. This can not only reasonably utilize the transformer's overload capacity and improve the economy of the entire distribution network operation, but also ensure the safety of transformer operation, thus achieving more intelligent dynamic congestion management.

[0065] Second, the use of a distributed control structure can give full play to the distributed computing capabilities of each load and adapt to the coordinated control of large-scale loads.

[0066] 3. There is no need to collect load privacy data and no need to directly control the load.

[0067] Fourth, it can adapt to multiple types of flexible loads such as EV and air conditioners. Each type of flexible load only needs to define the power feasible domain and satisfaction loss penalty item according to the electricity consumption preference, and allows each load to adopt a different optimization time domain, which fully reflects the differences in loads. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 Schematic diagram of the distributed rolling optimization process.

[0069] Figure 2 Output and ambient temperature curves for baseload and renewable energy.

[0070] Figure 3 Figure 2 shows the power curves of transformer, EV and air conditioning loads when dynamic congestion management is used.

[0071] Figure 4 This is the simulation result when static congestion management is used.

[0072] Figure 5 These are the simulation results when each user operates in a purely selfish and uncoordinated mode.

[0073] Figure 6 This is the simulation result when users use electricity freely.

[0074] Figure 7 Figure 7a shows the transformer HST and relative aging conditions, where Figure 7b shows the transformer relative aging rate. DETAILED DESCRIPTION

[0075] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0076] Example

[0077] The present invention provides a method for distributed coordinated control of residential loads taking into account the dynamic hot spot temperature of transformers, which specifically includes the following contents:

[0078] 1. Rolling Optimization Model

[0079] 1.1 Optimization Problem Model

[0080] The present invention establishes the following residential load electricity consumption rolling optimization model:

[0081]

[0082]

[0083]

[0084]

[0085] Where: i=1…N is the flexible load number, t is the scheduling period, M is the length of the rolling optimization time domain, and t0 is the starting time of the scheduling period.

[0086] In the objective function (1a), the first term is the total electricity purchase cost of the distribution network, where μ t is the electricity price, P T,t is the transformer power; Item 2 Reflects the aging cost of the transformer, where: is the winding HST; the third term ω(s i,t ) represents the satisfaction loss penalty term of the load, where s i is the satisfaction level of load i.

[0087] Constraint (1b) ensures power balance, where P i,t is the power of load i, and are the predicted values ​​of base load and renewable energy resource (RES) output respectively, and η is the transformer efficiency.

[0088] Constraint (1c) is the transformer operation constraint, where: is the maximum short-time overload power of the transformer, is the upper limit of the winding HST.

[0089] Constraint (1d) is the load constraint, where: is the feasible domain of load power, which is related to its electricity preference.

[0090] 1.2 Distributed solution method

[0091] Establishing and solving the above centralized optimization problem requires all subsystems to report their own key information, which may cause privacy leakage. In order to use distributed algorithms to solve the above centralized problem, the problem needs to be modularized. The solution process of the distributed rolling optimization problem is as follows: Figure 1 As shown, the Lagrangian multiplier vector λ is introduced, and the equality constraint of equation (1) is relaxed to obtain the following Lagrangian function:

[0092]

[0093] Using the dual decomposition principle, the dual function of problem (1) is:

[0094]

[0095] In the above formula, f T (λ) is the optimization sub-problem of the transformer:

[0096]

[0097] And f i (λ) is the optimization sub-problem of load i:

[0098]

[0099] The multiplier λ is updated according to the subgradient method:

[0100]

[0101] Where: They are the optimal solutions to subproblems (4) and (5) respectively when the multiplier λ is given.

[0102] (1) About the multiplier λ t Meaning: For the transformer sub-optimization problem (4), it can be seen that when its inequality condition is inactive (i.e., both the transformer power and HST are below the limit), the optimal power satisfies:

[0103]

[0104] At this time, the multiplier λ t It has a clear meaning, which is the marginal aging cost of the transformer added to the current electricity price. When equation (6) converges, the present invention uses λ t As the clearing price, for the load optimization subproblem (5), each load is optimized by the multiplier λ t The aging costs of the transformer are taken into account, so that the transformer life loss can be synergistically reduced.

[0105] (2) Regarding the universality of the load optimization model, Figure 1It can be seen that the flexible load and the distribution network dispatcher responsible for coordinated control only need to exchange multiplier signals and power signals, which improves the versatility of the method while protecting privacy. For different flexible loads, it can be seen from formula (5) that it is only necessary to give the satisfaction s according to the preference. i The definition of load power is determined

[0106] 2. Transformer aging model

[0107] For the aging cost model in the transformer optimization sub-problem (4), the transformer aging cost can be expressed as:

[0108]

[0109] Where: L exp is the expected life of the transformer, C T is the transformer cost, V t is the relative aging rate of the transformer, which can be calculated by the following formula:

[0110]

[0111] In the above formula, is the HST of the transformer at time t, which can be decomposed into the following two items for calculation:

[0112]

[0113] Where: the first term is the top oil temperature, and the second term is the temperature rise of the hot spot relative to the top oil. Both of the above two terms have an internal temperature accumulation process, and their dynamic equations are as follows:

[0114]

[0115]

[0116] Where: is the transformer load factor at time t, where is the rated power of the transformer; is the ambient temperature, ΔT is the control period, which is 15 minutes in the present invention. is the temperature rise of the hot spot relative to the top oil at time t-1 The former represents the basic hotspot temperature rise before considering the influence of the changing oil flow passing through the hotspot, and the latter represents the hotspot temperature rise caused by the oil flow change rate passing through the hotspot, and satisfies The meanings of the remaining parameters are shown in Table 1.

[0117] Table 1 Thermal model parameters of oil-immersed distribution transformer

[0118]

[0119]

[0120] It is easy to prove that the transformer aging cost C is related to the transformer power P T,t Therefore, the transformer optimization subproblem (4) is a convex optimization problem, and it is easy to obtain the global optimal solution.

[0121] 3. Residential load optimization model

[0122] 3.1 Air conditioning load optimization model

[0123] Without loss of generality, the present invention studies the coordinated control of refrigeration and air conditioning. The thermodynamic process of residential air conditioning can be modeled using an equivalent thermal parameters (ETP) model:

[0124]

[0125] Where: θ i,t is the room temperature of air conditioner i at time t, θ t,a is the ambient temperature at time t, R i,a and C i,a are the equivalent thermal resistance and thermal capacitance, Q i,t is the cooling capacity of air conditioner i at time t.

[0126] Define the normalized satisfaction s of air conditioner i at time t i for:

[0127]

[0128] Where: and θ i are the upper and lower limits of the room temperature set by the user. It is easy to know that when the room temperature is within the range allowed by the user, s i,t ∈[0,1],s i,t The closer it is to 0.5, the closer the indoor temperature is to the ideal temperature, the higher the user satisfaction is, and the more flexible the air conditioner is.

[0129] The main residential air conditioners in my country are variable frequency air conditioners, which can achieve continuous adjustment of cooling capacity and electric power by adjusting the compressor frequency. There is a linear relationship between electric power and cooling capacity as follows:

[0130]

[0131] Where: p1, p2, q1, q2 are all model coefficients.

[0132] Substituting equations (13) and (14) into equation (15), we can obtain the following air conditioning dynamic equation:

[0133] P i,t =a i s i,t+1 +b i s i,t +c i,t (16)

[0134] Where: a i <0,b i ,c i,t >0 can be derived from equations (13) and (14).

[0135] From formula (16), we can see that in order to make the room temperature reach the lower limit of the allowable satisfaction (s i,t+1 =0) or upper limit (s i,t+1 =1), the maximum and minimum power of air conditioner i satisfy:

[0136]

[0137]

[0138] Where: For the maximum power of the air conditioner, you can take its rated power.

[0139] The satisfaction loss of air conditioner i at time t is defined as:

[0140]

[0141] Where: w is the weight coefficient, It is the average main grid electricity price in M ​​time periods starting from time t, obtained based on historical electricity price data.

[0142] The power feasible range of air conditioning load is:

[0143]

[0144] At this point, substituting equations (19) and (20) into equation (5) constitutes the optimization subproblem of air conditioning load. It is easy to see that this problem is a convex optimization problem.

[0145] 3.2EV Optimization Model

[0146] Assume that the rated power of EV is P i,N The charging time is t i,a , the departure time is t i,d , the target power is Etarg i. The present invention assumes that as long as the target power requirement is met, the satisfaction loss is no longer considered. Thus, in formula (5), EV only needs to consider the following charging power feasible domain:

[0147]

[0148] It is worth noting that the distributed optimization method of the present invention fully considers the differences in EV charging behavior, and each EV does not need to use a unified optimization time domain. i,t =0.

[0149] 4. Simulation analysis

[0150] 4.1 Example Design

[0151] A simulation system was established based on the basic situation of a residential load source grid-load interaction project. The transformer capacity is 630kVA, the power factor is considered to be 0.9, the transformer model parameters are shown in Table 1, and the HST upper limit is 120℃. 100 households were considered in the simulation, and each household was equipped with 2 air conditioners. EV is considered in the long term, with an average of 1 EV per household, using slow charging, a rated charging power of 3.7kW, and a capacity randomly distributed between 20-25kWh. The power curves of base load and renewable energy are shown in Figure 1. Figure 2 The time-of-use electricity price for residents is shown in Table 2. The simulation season is summer, the simulation duration is 2 weeks, and the ambient temperature curve is also shown in Table 2. Figure 2 .

[0152] Table 2 Time-of-use electricity price (yuan / kWh)

[0153] Time 23:00~7:00(Valley) 8:30~11:30,16:00~21:00(peak) Other time periods (average) price 0.43 1.17 0.8

[0154] Four representative schemes were set up for simulation, among which Scheme 1 is the scheme of the present invention, and the others are comparative schemes:

[0155] Solution 1: Dynamic blocking management, with the transformer HST not exceeding 110°C as the condition, rationally utilize the transformer short-term overload capacity (1.2 times overload), and dynamically determine the transformer maximum load;

[0156] Scheme 2: Static congestion management, ignoring HST, limiting the maximum load of the transformer to the rated capacity. This scheme is a special case of Scheme 1, that is, the HST constraint is removed from equation (1c);

[0157] Option 3: Pure self-interest model, assuming that each household is equipped with an energy management system that can achieve self-optimization based on electricity prices, but is not coordinated with each other;

[0158] Solution 4: Free operation, assuming that each household is not equipped with an energy management system and the loads are all running freely.

[0159] 4.2 Analysis of blocking management effect

[0160] Taking the rated active power of the transformer as the base value, Figures 3 to 6 The transformer load curves under each scheme on days 2-4 are plotted.

[0161] A. Dynamic congestion management.

[0162] Figure 3 The power curves of transformer, EV and air conditioning load when dynamic congestion management is adopted. It can be seen from the figure:

[0163] (1) EVs are coordinated to charge during the off-peak hours (23:00-7:00), which prevents concentrated charging and reduces charging costs;

[0164] (2) Taking advantage of the thermal inertia of buildings, air conditioners will increase electricity consumption to cool down the room in advance before the high electricity price period arrives, thereby reducing electricity costs;

[0165] (3) In this scheme, during the time periods of 6:00-7:00 and 8:00-8:30, since the transformer HST is not high and the historical load is relatively low, the transformer will use its short-term overload capacity to operate beyond the rated power;

[0166] (4) The transformer explicitly adds its marginal aging cost to the electricity bill, thereby further guiding the flexible load to use electricity in an orderly manner and avoiding congestion. The additional electricity price for each period can be intuitively seen in the figure. When the transformer HST is low, the marginal aging cost is very small and can be ignored.

[0167] B. Static blocking management.

[0168] Figure 4 The simulation results when static congestion management is adopted are shown in the figure. It can be seen from the figure that: (1) Compared with Scheme 1, this scheme can prevent the transformer power from exceeding the limit throughout the day, achieving the purpose of static congestion management; (2) Compared with Scheme 1, since the transformer aging cost is not considered, the EV load is more concentrated and the peak value is higher. During the period of 0:00-6:00, the transformer is continuously running at full load, which will cause its HST to continue to increase; (3) When the transformer power exceeds the limit, the system automatically generates a very high congestion price based on the original electricity price, thereby suppressing the flexible load power.

[0169] C. Purely self-interest model.

[0170] Figure 5 The simulation results are shown below when each user operates in a purely self-interested and uncoordinated mode. As shown in the figure: during the 6:00-7:00 period, the air conditioning load uses the valley electricity price to cool down the room in advance; during the 22:00-23:00 period, the air conditioning load increases the electricity consumption level due to the lower electricity price. The above two air conditioning load peaks are superimposed on the charging load, resulting in a serious overload of up to 1.4 times the transformer. Figure 7It can be seen that this will cause the transformer HST to exceed 120°C, causing irreversible damage to the transformer insulation strength. The above shows that when some users configure a home energy management system, the effect of time-of-use electricity prices can be brought into play. However, when the proportion of home energy management systems is very high, new load peaks may be generated during low electricity price periods.

[0171] Figure 6 The simulation results are shown in the case of free electricity consumption by users. As can be seen from the figure: EVs are connected to the power grid in the evening, which, on the one hand, overlaps with the original load peak, thus causing serious overload problems for the transformer; on the other hand, the corresponding time periods are mostly peak hours and normal hours, and the electricity bill expenditure is relatively high.

[0172] It should be noted that in Schemes 3 and 4, regardless of whether there is congestion or not, the electricity price is the original time-of-use electricity price, without adding transformer aging or congestion costs.

[0173] 4.3 Analysis of transformer HST and relative aging rate

[0174] The following analysis shows the HST and relative aging rate of the transformer under the four schemes. Figure 7 As shown in Figure (7a), the 110℃ hot spot temperature control line is the normal operation line of the transformer. Within this range, the relative aging rate of the transformer is less than 1; when it exceeds this line, the transformer can still be overloaded for a short time, but it will accelerate aging. 120℃ is the safe operation limit temperature of the transformer. Analysis Figure 7 It can be seen that:

[0175] (1) The free power consumption mode will cause the transformer HST to exceed 120°C for a long time, and the corresponding transformer aging speed will be more than 20 times. This is not allowed in actual operation;

[0176] (2) Under the pure self-interest model, guided by the time-of-use electricity price, the HST limit-exceeding situation is alleviated, but it will still exceed 120℃ for a short time;

[0177] (3) After adopting static blocking management, the load power consumption is more orderly. However, due to the failure to consider the transformer HST, the transformer continued to operate at full power from 0:00 to 6:00, and the internal hot spots exceeded 110°C due to the continuous accumulation of heat.

[0178] (4) Compared with the previous schemes, the dynamic blocking scheme can control HST below 110°C. This is due to its ability to appropriately relax power constraints when HST is low, and limit power when HST is high. Its significance is not only to reduce the aging speed of the transformer, but more importantly, this scheme can maintain HST at a low level, thereby ensuring the safe operation of the transformer and reducing the probability of unplanned power outages during peak loads.

[0179] 4.4 Solution Summary

[0180] Table 3 compares various schemes, among which the percentage of cost savings compared with Scheme 4 is given in brackets in the column of electricity cost. It can be seen that Schemes 1-3 are all effective in reducing electricity costs; Scheme 1 is the best in reducing transformer aging rate. More importantly, only Scheme 1 can effectively implement congestion management and control HST, thus having advantages in ensuring transformer equipment safety and improving power supply reliability.

[0181] Table 3 Comparison results of the schemes

[0182]

[0183] In summary, the life loss and operational safety of the transformer are closely related to its hot spot temperature. Compared with the static congestion management method with rated capacity as a constraint, the dynamic congestion management method with hot spot temperature as the control target adopted by the present invention can take into account factors such as the ambient temperature and historical load of the transformer operation, thus being more reasonable.

[0184] By coordinating the control of various flexible loads such as EV and air conditioners, the present invention can exert greater user-side flexibility and avoid the peak-peak superposition phenomenon in certain time periods compared to considering only a single load type.

[0185] The distributed coordinated control method of residential loads proposed in the present invention can achieve the above objectives well. First, the transformer can participate in dynamic bidding based on its aging model, thereby implementing dynamic congestion management; second, each load does not need to expose privacy data and direct control authority to the outside world. At the same time, the load is allowed to have different optimization time domains, reflecting the differences in loads, which are crucial for demand-side electricity management; finally, the method does not need to establish a large-scale centralized optimization model and has good scalability.

Claims

1. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers, characterized in that: The following steps are involved: 1) Construct a rolling optimization model for residential load electricity consumption based on the transformer aging model and the residential load model respectively; 2) The centralized problem of the rolling optimization model of residential load electricity consumption is modularized, the Lagrange multiplier vector is introduced, and the distributed algorithm is used to solve the centralized problem, and finally the optimal coordinated control scheme is obtained; In step 1), the objective function of the rolling optimization model for residential load electricity consumption is: Among them, P T,t is the transformer power, μ t For electricity price, is the aging cost of the transformer, ω(s i,t ) is the loss penalty term related to load satisfaction, M is the length of the rolling optimization time domain, N is the number of resident loads, and the subscript i is the resident load number. is the dynamic hot spot temperature HST of the winding, t is the scheduling period, and t0 is the starting time of the scheduling period; The constraints of the residential load electricity rolling optimization model include: Power balance constraints: Transformer operation constraints: Residential load constraints: Among them, P t base is the predicted value of base load output, P t RES is the predicted value of renewable energy output, η is the transformer efficiency, is the maximum short-time overload power of the transformer, is the upper limit of the winding dynamic hot spot temperature HST, is the feasible region of power for residential loads, P i,t The power of residential load.

2. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 1, characterized in that: In the transformer aging model described above, the transformer aging cost The expression is: Among them, C T is the transformer cost, L exp is the expected life of the transformer, V t is the relative aging rate of the transformer, is the top oil temperature, is the temperature rise of the hot spot relative to the top oil.

3. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 2, characterized in that: The top oil temperature and the temperature rise of the hot spot relative to the top oil There is an internal temperature accumulation process, and the dynamic equations are: in, is the transformer load factor, is the transformer rated power, is the top oil temperature at time t-1, R is the ratio of load loss to no-load loss at rated current, x is the exponential power of total loss relative to the top oil temperature rise of the tank, △θ or is the steady-state temperature rise of the hot spot relative to the environment under rated load, is the ambient temperature, △T is the control period, t0 is the oil time constant, is the temperature rise of the hot spot relative to the top oil at time t-1, y is the exponential power of the current relative to the winding temperature rise, △θ hr is the temperature rise of the hot spot relative to the top oil at rated load, To consider the basic hot spot temperature rise before the effect of the changing oil flow through the hot spot, To consider the hot spot temperature rise caused by the oil flow rate change through the hot spot, and to meet t w is the winding time constant, k 11 , k 21 , k 22 is the thermal model constant.

4. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 1, characterized in that: In the residential load model, residential loads include air conditioning loads and EVs.

5. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 4, characterized in that: For air conditioning load, the corresponding power is expressed as: P i,t =a i s i,t+1 +b i s i,t +c i,t Among them, a i <0, b i , c i,t >0 is the parameter, s i,t 、s i,t+1 are the normalized load satisfaction at time t and t+1, and θ i are the upper and lower limits of the room temperature set by the user, θ i,t ,θ i,t+1 is the room temperature of the air conditioning load at time t and t+1, θ t,a is the ambient temperature at time t, R i,a , C i,a are equivalent thermal resistance and thermal capacitance respectively, △T is the control period, Q i,t is the cooling capacity of the air conditioner at time t; Its loss penalty term ω(s) related to load satisfaction at time t i,t ) is: Among them, w is the weight coefficient, are the upper and lower limits of the air conditioning power at time t, is the average main grid electricity price in M ​​time periods starting from time t, is the maximum power of the air conditioner; The power feasible region of the air conditioning load is expressed as:

6. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 4, characterized in that: For EV, its power feasible region is expressed as: Among them, P i,N is the EV rated power, t i,a is the time to start charging, t i,d For departure time, is the target power, The power level at the start of charging.

7. A method for distributed coordinated control of residential loads considering the dynamic hot spot temperature of transformers according to claim 1, characterized in that: The step 2) specifically comprises the following steps: 21) Introduce the Lagrange multiplier vector λ to construct the Lagrange function, then: Among them, λ t is the Lagrange multiplier at time t; 22) Get the dual function f(λ) of the centralized problem, then: Among them, f T (λ) is the optimization sub-problem of the transformer, f i (λ) is the optimization sub-problem of resident load i; 23) Given the Lagrange multipliers, f T (λ) and f i (λ) is solved to obtain the optimal solution of the subproblem at the current step and Based on this, the subgradient method is used to update the Lagrange multiplier, and the solution is obtained by rolling iteration, and finally the optimal coordinated control scheme is obtained.

8. A method for distributed coordinated control of residential loads taking into account the dynamic hot spot temperature of transformers according to claim 7, characterized in that: In the step 23), the expression for updating the Lagrange multiplier is: in, is the updated Lagrange multiplier.

Images