A two-layer distributed control method for constant temperature loads in smart grids

Abstract

The present invention discloses a two-layer distributed control method for constant temperature loads of smart grids. First, a two-layer distributed control structure is constructed, which includes two functional modules: an optimization layer and a control layer. The optimization layer realizes the economic optimization scheduling of TCL energy consumption by introducing a bill splitting function and based on the aggregation game theory. At the same time, by introducing the "comfort state" indicator, a fair distribution mechanism is designed to balance the comfort experience of different users. The control layer receives the signal output by the optimization layer and adopts a state-related variable exponential coefficient control algorithm. While ensuring the convergence speed, it effectively avoids the singularity problem caused by the sign function and realizes the fixed-time precise tracking of the TCL operating state. The present invention realizes the coordinated optimization of user comfort and system economy, and has the advantages of clear structure, high control accuracy, strong privacy protection and suitability for large-scale applications. It has broad promotion value and application prospects.

Description

Technical Field

[0001] The present invention belongs to the field of smart grids, and in particular relates to a double-layer distributed control method for constant temperature loads of smart grids. Background Art

[0002] With the increasing global emphasis on sustainable development and the dual carbon goals, smart grids have garnered widespread attention as a key technology for achieving efficient energy management and optimizing electricity consumption. Within smart grids, thermostatically controlled loads (TCLs), such as air conditioners and refrigerators, have become a key research focus due to their widespread use and significant energy consumption. Effective management of TCLs not only contributes to the stable operation of the grid but also significantly improves energy efficiency and reduces unnecessary electricity waste.

[0003] However, traditional TCLs management methods often ignore the key factor of user comfort. The sole pursuit of economic efficiency may lead to a decline in user comfort experience, which in turn affects user acceptance and satisfaction with smart grid technology. Therefore, while optimizing the economic performance of TCLs, how to ensure user comfort has become an urgent problem to be solved. Most existing research focuses on the optimization of a single objective, such as focusing only on economy or only on comfort. Although some studies have attempted to balance the two through centralized control methods, these methods usually face problems such as high computational complexity, poor real-time response capabilities, and insufficient privacy protection. In addition, in practical applications, due to factors such as temperature fluctuations, changes in user behavior, equipment aging, and power supply fluctuations, the system needs to have higher robustness and adaptability to cope with various uncertainties.

[0004] Therefore, it is urgent to design a control structure that can ensure user comfort while taking into account the load economic dispatch goals, and has good distributed characteristics, system robustness and scalability to adapt to the dispatch needs of large-scale intelligent loads in the future. Summary of the Invention

[0005] To solve the above problems, the present invention discloses a two-layer distributed control method for constant temperature loads in smart grids. It constructs a two-layer distributed control structure consisting of an optimization layer and a control layer. By adopting a scientific algorithm, economic optimization and comfort control can be deployed in different control layers respectively without interfering with each other, thereby effectively improving the operating efficiency and stability of the system.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] A two-layer distributed control method for constant temperature loads in smart grids includes the following steps:

[0008] S1. Construct a two-layer distributed control structure, with the upper layer being the optimization layer and the lower layer being the control layer. The optimization layer is responsible for optimizing the economy and comfort of the objective function to achieve optimal economic benefits and ensure fair sharing of comfort status among TCLs; the control layer is responsible for accurately adjusting the actual energy consumption and comfort status of each TCL based on the reference signal output by the optimization layer. The TCLs described in the present invention represent all constant temperature control loads in the system, and TCL represents a specific constant temperature control load;

[0009] S2. Build an economic optimization module for the optimization layer. This module is based on game theory and determines the reference power value of each TCL according to the cost function of each TCL. First, set the cost function. The cost function of each TCL is set to the product of the partition function and the total cost function. The partition function is a function that allocates energy costs to each TCL, so it should be a continuous and strictly increasing function, and the sum of the partition functions of all users is 1; the convexity of the partition function is used to ensure the convexity of a single cost function to establish a concave N-person game problem. Through this mechanism, it can ensure that each TCL achieves the global optimum while pursuing its own best decision;

[0010] S3. Use the aggregate game method to solve the concave N-person game problem, so that each TCL can achieve the global optimality while ensuring that its own decision is optimal and obtain the optimal energy consumption of each TCL. ;

[0011] S4. Build a comfort optimization module for the optimization layer, define the comfort state of TCL as a ratio of 0 to 1, and ultimately achieve fair sharing of user comfort and obtain a reference comfort state. ;

[0012] S5. The output of the optimization layer is passed to the control layer, that is, the reference values ​​obtained in steps S3 and S4 are passed to the control layer. The control layer realizes the coordinated optimization of user comfort and system economy through the variable exponential coefficient timing stabilization algorithm.

[0013] Furthermore, in S1, communication is established within the optimization layer using an undirected connected graph consisting of nodes and edges. Each node represents a TCL, and each edge represents a direct connection or relationship between two nodes. Edges are non-directional. For any pair of nodes in the graph, there is at least one path connecting them, indicating that information can be transmitted throughout the network, with no isolated components.

[0014] Furthermore, in S2, a concave N-person game problem is determined using a pricing mechanism determined by a partitioning function, where the partitioning function pricing mechanism is used to make the cost function of TCL a strictly convex function to ensure the unity of individual optimality and global optimality. represents the energy consumption of the kth TCL, represents the energy consumption of all TCLs except the kth TCL, is the cost function of the k-th TCL, is the total cost function, and the set K / k represents the set of TCLs in K except the kth TCL, so represents the total energy consumption of all TCLs except the kth TCL, represents the sum of the mth power of the energy consumption of all individual TCLs except the kth TCL. m is a controllable parameter in the pricing strategy. The cost function determined by the split function is as follows:

[0015] .

[0016] Among them, the split function:

[0017]

[0018] is a continuous, non-negative and strictly increasing function, and the sum of all TCL bill splitting functions is 1; that is, , ; Assume the total cost function is a continuous, positive, increasing convex function, it is easy to get Strictly convex.

[0019] Furthermore, in S3, the aggregate game method is used to solve the concave N-player game problem. The advantage is that participants do not need to share their energy consumption with others, but only need to share their own estimation of total energy consumption, thus protecting the privacy of users. is the adjacent set of participant k,

[0020] ,

[0021] , here is a small positive parameter, is a fixed positive parameter, and is an intermediate variable, and the Nash equilibrium seeking strategy adopted is as follows:

[0022] .

[0023] Furthermore, the comfort optimization module of the optimization layer in S4 involves a flexible TCL model, which can achieve continuous power adjustment from 0 to rated power through variable frequency air conditioning or inverter air conditioning. The internal temperature dynamics equation of the i-th flexible TCL can be described as a first-order ordinary differential equation:

[0024]

[0025] in are the internal temperature, heat capacity and thermal resistance of the i-th TCL, is the ambient temperature, which is considered constant in a short period of time. is the thermal coefficient, which is positive for cooling and negative for heating. The power consumed by the flexible TCL can be continuously controlled from 0 to rated power.

[0026] Assumptions The lower temperature limit of the comfortable temperature is The upper temperature limit for comfortable temperature. Is the user's maximum tolerance for temperature changes, is the user's most comfortable temperature set point. To quantify the user's sensitivity to temperature, an index called comfort state is introduced, which is defined as follows:

[0027]

[0028] The comfort level is the ratio of the temperature deviation from the set temperature to the maximum tolerance range. As the temperature changes from the lower temperature boundary to the upper temperature boundary, the corresponding comfort level changes from 0 to 1. Notably, when the temperature is equal to the set temperature, the comfort level is 0.5. This definition means that users who are less sensitive to temperature can tolerate a wider range of temperature values. The goal of this invention is to achieve fair sharing of comfort levels across all TCLs. Specifically,

[0029] .

[0030] set up is a positive parameter, To solve this problem, the following distributed algorithm is used:

[0031]

[0032] in, are the Laplace matrix and incidence matrix of the communication graph, respectively. .error .error It eventually converges to 0, achieving fair sharing of the comfort state.

[0033] Furthermore, in S5, the output of the optimization layer is passed to the control layer. In the optimization layer (steps S3 and S4), the aggregation game method is used for economic optimization scheduling to achieve the best economic performance and ensure that the comfort state of all TCLs is shared fairly. Specifically, in step S3, each TCL minimizes its own cost function based on its own actions and the linear aggregation actions of the entire system, thereby achieving the global optimal solution while also achieving the individual optimal solution. In step S4, in order to ensure the user's comfortable experience, the goal is set to ensure that the comfort state of all TCLs is shared fairly to avoid excessive sacrifice of the experience of some users. Then, according to the reference signal calculated by the optimization layer (power reference value) and (Comfort state reference value), this information will be passed to the control layer. The main goal of the control layer is to ensure that the actual energy consumption and comfort state of each TCL can accurately track these reference values, that is, This not only helps to achieve the economic benefits of the entire system, but also ensures that the user's comfort is not affected.

[0034] Furthermore, in S6, a state-dependent variable exponential coefficient algorithm is used in the control layer to avoid the system's singularity problem and enhance the system's robustness and adaptability. This algorithm can cope with disturbances caused by factors such as temperature fluctuations, user behavior, equipment aging, and power fluctuations, ensuring that the system can still operate stably in complex and changing environments. Assume that the state of TCL is , whose coefficients are set to ,in and , there is disturbance , For a given and Assumptions , make

[0035] ,

[0036] The sign function is used to describe the dynamic changes in energy consumption. Since the discontinuity of the sign function at zero will cause singularity problems, a state-dependent smoothing mechanism is introduced to solve the singularity problem. The system is globally stable, and the settling time (here e represents a natural constant, and other places represent errors)

[0037] .

[0038] The algorithm provides better performance, enabling TCLs to quickly respond to changes in the reference signal within a fixed time, thereby improving the response speed and stability of the entire system.

[0039] The beneficial effects of the present invention are:

[0040] This paper proposes a two-tier distributed control structure for the coordinated optimization of economic efficiency and comfort of TCLs in smart grids. This layered architecture offers structural advantages for multi-objective control tasks. Economic optimization and comfort control can be deployed in separate control layers without interfering with each other, effectively improving system efficiency and stability.

[0041] In the optimization layer, an aggregated game approach based on game theory is employed. Each participating node only needs to exchange local estimates of total energy consumption, without sharing its actual energy consumption information. This protects user privacy while significantly reducing communication overhead and system coupling. This distributed optimization strategy achieves the dual-objective optimization of system economy and user comfort while ensuring user information security, effectively enhancing the coordination and scalability of the control system.

[0042] In the control layer, a fixed-time control algorithm based on a state-dependent variable exponential coefficient is introduced. This algorithm exhibits excellent robustness and nonlinear adaptability, capable of handling complex disturbance environments such as temperature fluctuations, changes in user behavior, and device aging. By designing state-dependent control gains and introducing a smooth transition mechanism, this algorithm effectively avoids the discontinuity and singularity issues caused by traditional sign functions at zero, ensuring the continuity of the system response process and the numerical stability of the control law. Specifically, when the system error approaches zero, the control behavior achieves smooth convergence, avoiding the oscillation or overshoot common in traditional control.

[0043] In summary, the present invention provides an efficient, stable and easy-to-implement solution for multi-participant load control systems through the organic combination of structural optimization, algorithm innovation and information interaction mechanism, and has good engineering adaptability and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a step block diagram of the present invention.

[0045] Figure 2 This is a structural diagram of the double-layer distributed control framework described in the present invention.

[0046] Figure 3 This is a TCL communication diagram in an embodiment of the present invention.

[0047] Figure 4 This is a curve diagram of energy consumption and comfort state changes in the optimized layer in the example of the present invention.

[0048] Figure 5 This is a curve diagram of the change of energy consumption error and comfort state error defined in the control layer in the example of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0050] The specific implementation steps of the present invention are as follows Figure 1 As shown, the double-layer distributed control framework structure is as follows Figure 2 As shown in the TCL communication diagram Figure 3 shown.

[0051] The present invention provides a two-layer distributed control method for constant temperature loads in a smart grid, comprising the following steps:

[0052] S1. Construct a two-layer distributed control structure, with the upper layer being the optimization layer and the lower layer being the control layer. Figure 2 As shown in , each TCL in the optimization layer only needs to share its own estimate of total energy consumption, without sharing its actual energy consumption, and pass the output reference value to the control layer. The control layer is responsible for accurately adjusting the actual energy consumption and comfort state of each TCL based on the reference signal output by the optimization layer. Figure 3 The adjacency matrix of the undirected communication graph shown is:

[0053]

[0054] S2. Build an economic optimization module in the optimization layer. This module determines the reference power value of each TCL based on the cost function of each TCL based on game theory. First, a concave N-person game problem is determined using a pricing mechanism determined by a partitioning function. The partitioning function pricing mechanism is used to make the cost function of the TCL a strictly convex function to ensure the unity of individual optimality and global optimality. Assume represents the energy consumption of the kth TCL, represents the energy consumption of all TCLs except the kth TCL, is the cost function of the k-th TCL, is the total cost function, and the set K / k represents the set of TCLs in K except the kth TCL, so represents the total energy consumption of all TCLs except the kth TCL, represents the sum of the mth power of the energy consumption of all individual TCLs except the kth TCL. m is a controllable parameter in the pricing strategy. The cost function determined by the split function is as follows:

[0055]

[0056] Among them, the split function:

[0057]

[0058] In this example, , just need to meet

[0059]

[0060] This ensures the strict convexity of the partitioning function.

[0061] S3. Use the aggregate game method to solve the concave N-player game problem, design a Nash equilibrium search strategy based on the dynamic average consistency principle, and obtain the optimal energy consumption of each TCL Assumptions is the adjacent set of participant k,

[0062] ,

[0063] , here is a small positive parameter, is a fixed positive parameter, and is an intermediate variable, and the Nash equilibrium seeking strategy adopted is as follows:

[0064]

[0065] The algorithm designs a Nash equilibrium seeking strategy based on the principle of dynamic average consistency. The power change curve is as follows: Figure 4 shown.

[0066] S4. Build a comfort optimization module for the optimization layer, define the comfort state of TCL as a ratio of 0 to 1, and ultimately achieve fair sharing of user comfort and obtain a reference comfort state. , here The lower temperature limit of the comfortable temperature is The upper temperature limit for comfortable temperature. Is the user's maximum tolerance for temperature changes , is the user's most comfortable temperature set point. To quantify the user's sensitivity to temperature, the comfort state is introduced and defined as follows:

[0067]

[0068] The comfort level is the ratio of the temperature deviation from the set temperature to the maximum tolerance range. As the temperature changes from the lower temperature boundary to the upper temperature boundary, the corresponding comfort level changes from 0 to 1. Notably, when the temperature is equal to the set temperature, the comfort level is 0.5. This definition means that users who are less sensitive to temperature can tolerate a wider range of temperature values. The goal of this invention is to achieve fair sharing of comfort levels across all TCLs. Specifically,

[0069] .

[0070] set up is a positive parameter, To solve this problem, the following distributed algorithm is used:

[0071]

[0072] in, are the Laplace matrix and incidence matrix of the communication graph, respectively. .error , the comfort state change curves of each TCL are as follows Figure 4 shown.

[0073] S5. The output of the optimization layer is passed to the control layer, that is, the reference values ​​obtained in steps S3 and S4 are passed to the control layer. The control goal of the control layer is to make the energy consumption and comfort state of each TCL reach the reference values ​​at the same time. The main goal of the control layer is to ensure that the actual energy consumption and comfort state of each TCL can accurately track these reference values, that is, This not only helps to achieve the economic benefits of the entire system, but also ensures that the user's comfort is not affected.

[0074] S6. Use a state-dependent fixed time-varying exponential coefficient algorithm in the control layer. Assume that the state of TCL is , whose coefficients are set to ,in and , there is disturbance , For a given and Assumptions , make

[0075]

[0076] The system is globally stable at the same time, and the settling time (where e represents a natural constant and other places represent errors);

[0077] .

[0078] In this case, ,error The stabilization time can be calculated. , its changing curve is as follows Figure 5 shown.

[0079] It should be noted that the above content merely illustrates the technical idea of ​​the present invention and cannot be used to limit the scope of protection of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications all fall within the scope of protection of the claims of the present invention.

Claims

1. A two-layer distributed control method for constant temperature loads in smart grids, characterized by: The steps include: S1. Construct a two-layer distributed control structure, with the upper layer being the optimization layer and the lower layer being the control layer. The optimization layer is used to optimize the reference energy consumption and comfort state of the TCLs, and the control layer is used to track the reference value output by the optimization layer and achieve precise control. S2. Build an economic optimization module for the optimization layer. This module is based on game theory and determines the reference power value of each TCL according to the cost function of each TCL. First, set the cost function. The cost function of each TCL is set to the product of the partition function and the total cost function. The partition function is a function that allocates energy costs to each TCL. Therefore, it should be a continuous and strictly increasing function, and the sum of the partition functions of all users is 1. Use the convexity of the partition function to ensure the convexity of the individual cost function to establish a concave N-player game problem. S3. Use the aggregate game method to solve the concave N-person game problem, so that each TCL can achieve the global optimality while ensuring that its own decision is optimal and obtain the optimal energy consumption of each TCL. ; S4. Build a comfort optimization module for the optimization layer, define the comfort state of TCL as a ratio of 0 to 1, and ultimately achieve fair sharing of user comfort and obtain a reference comfort state. ; S5. The output of the optimization layer is passed to the control layer, that is, the reference values ​​obtained in steps S3 and S4 are passed to the control layer. The goal of the control layer is to make the energy consumption and comfort state of each TCL reach the reference values ​​at the same time; S6. Use a state-dependent variable exponential coefficient algorithm in the control layer: it can cope with disturbances caused by various factors and ensure that the system can still operate stably in complex and changing environments.

2. A two-tier distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: In step S2, a concave N-player game problem is determined using a pricing mechanism determined by a partitioning function, wherein the partitioning function pricing mechanism is used to make the cost function of TCL a strictly convex function to ensure the unity of individual optimality and global optimality; assuming represents the energy consumption of the kth TCL, represents the energy consumption of all TCLs except the kth TCL, is the cost function of the k-th TCL, is the total cost function, and the set K / k represents the set of TCLs in K except the kth TCL, so represents the total energy consumption of all TCLs except the kth TCL, represents the sum of the m-th power of the energy consumption of all individual TCLs except the k-th TCL. m is a controllable parameter in the pricing strategy. The cost function determined by the partitioning function is as follows: ； Among them, the split function: ； The split function is a continuous, non-negative and strictly increasing function, and the sum of all TCL bill split functions is 1; that is, , ; Assume the total cost function is a continuous, positive, increasing convex function, it is easy to get Strictly convex.

3. The dual-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: In step S3, the aggregate game method is used to solve the concave N-player game problem. Assume is the adjacent set of participant k, ， , here is a small positive parameter, is a fixed positive parameter, and is an intermediate variable, and the Nash equilibrium seeking strategy adopted is as follows: 。 4. The dual-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: In step S4, the comfort optimization module of the optimization layer is built. The lower temperature limit of the comfortable temperature is The upper temperature limit for the allowed comfortable temperature; It is the user's maximum tolerance for temperature changes; The temperature set point that is most comfortable for the user. To quantify the user's sensitivity to temperature, an index called comfort state is introduced and is defined as follows: ； The comfort state is the ratio of the temperature deviation from the set temperature to the maximum tolerance range. When the temperature changes from the lower temperature boundary to the upper temperature boundary, the corresponding comfort state changes from 0 to 1. Specifically: ； set up is a positive parameter, , are the thermal resistance and thermal coefficient of the kth TCL. To solve this problem, the following distributed algorithm is used: ； in, are the Laplace matrix and incidence matrix of the communication graph, respectively. ;error ;error It eventually converges to 0, achieving fair sharing of the comfort state.

5. The double-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: In step S5, the reference signal calculated by the optimization layer is and , passed to the control layer; the main goal of the control layer is to ensure that the actual energy consumption and comfort status of each TCL can accurately track these reference values, that is, This not only helps to achieve the economic benefits of the entire system, but also ensures that the user's comfort is not affected.

6. The double-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: In step S6, a state-dependent variable exponential coefficient algorithm is used in the control layer. Assuming that the state of TCL is , whose coefficients are set to ,in and , there is disturbance , For a given and Assumptions , make ， The sign function is used to describe the dynamic change of energy consumption. Since the discontinuity of the sign function at zero point will bring about singularity problem, a state-dependent smoothing mechanism is introduced to solve the singularity problem. The system is globally stable and the stabilization time is 。 7. The double-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: The inter-node communication structure involved in the optimization layer and the control layer is constructed using an undirected connected graph, which consists of several nodes and edges. Each edge represents a direct connection or relationship between two nodes, and the edge is non-directional. For any pair of nodes in the graph, there is at least one path connecting them, which means that information is transmitted throughout the network and there are no isolated parts.

8. The double-layer distributed control method for constant temperature loads in smart grids according to claim 1, characterized in that: The method is applicable to a flexible TCL model, wherein the TCL realizes continuous power regulation from 0 to rated power through a variable frequency air conditioner or an inverter air conditioner; the internal temperature dynamics equation of the i-th flexible TCL is described as a first-order ordinary differential equation: ； in are the internal temperature, heat capacity and thermal resistance of the i-th TCL, is the ambient temperature, which is considered constant in a short period of time. is the thermal coefficient, cooling is positive and heating is negative, The power consumed by the flexible TCL can be continuously controlled from 0 to rated power.

Images