Privacy-preserving method for constructing building thermal dynamic aggregation model

Abstract

Disclosed in the present invention is a privacy-preserving method for constructing a building thermal dynamic aggregation model, comprising: firstly, establishing a thermal dynamic model of a single building area, and establishing a building thermal dynamic aggregation model on the basis of an aggregation equation; then on the basis of a measurement equation, carrying out parameter estimation by means of a least squares method, and introducing a regularization term to solve the sparsity problem of aggregation coefficients, to obtain a compact parameter estimation model of the building thermal dynamic aggregation model; and finally, establishing a privacy-preserving parameter estimation method of the building thermal dynamic aggregation model. According to the present method, aggregation modeling is performed on massive buildings by building load aggregators to participate in operation and control of energy systems; moreover, privacy information of building users can be protected from leakage, thereby promoting the mining of building thermal inertia and improving the flexibility of operation and regulation of power systems.

Description

A privacy-preserving method for constructing a building thermal dynamic aggregation model Technical Field

[0001] The present invention belongs to the technical field of privacy protection and demand-side response in power systems, and mainly relates to a method for constructing a privacy-protected building thermal dynamic aggregation model. Background Art

[0002] With the increasing proportion of intermittent renewable energy, the power system is undergoing significant transformation. The power system requires more flexible resources to support its secure and economical operation. The inherent thermal inertia of buildings provides considerable flexibility for heating and cooling, making it a potential demand response resource. The concept of leveraging building thermal inertia for power system operation and control has attracted widespread attention from both academic and engineering communities.

[0003] However, putting building flexibility resources into practical use remains challenging. Firstly, due to the sheer number of buildings, direct information exchange between the energy system and the vast number of building users creates a significant computational and communication burden. Secondly, current approaches to mining building flexibility lack the means to protect user privacy, posing a risk of privacy leakage.

[0004] Summary of the Invention

[0005] This invention addresses the problems existing in the prior art and provides a privacy-preserving method for constructing a building thermal dynamic aggregation model. First, a thermal dynamic model of a single building area is established. Based on the aggregation equation, the building thermal dynamic aggregation model is constructed. Then, based on the measurement equation, the least squares method is used for parameter estimation. Regularization terms are introduced to address the sparsity of the aggregation coefficient, resulting in a compact parameter estimation model for the building thermal dynamic aggregation model. Finally, a privacy-preserving method for estimating the parameters of the building thermal dynamic aggregation model is established. This method uses a building load aggregator to aggregate and model a large number of buildings to participate in the operation and control of the energy system. This method also protects the privacy of building users, thereby promoting the mining of building thermal inertia and enhancing the flexibility of power system operation and regulation.

[0006] In order to achieve the above-mentioned objectives, the technical solution adopted by the present invention is: a privacy-preserving method for constructing a thermal dynamic aggregation model of a building, establishing a thermal dynamic aggregation model of a building cluster and its corresponding parameter estimation model, and according to the block coordinate descent theory, transforming the original non-convex parameter estimation problem into two convex optimization problems for iterative solution, and realizing the protection of user privacy information based on a transformation-based encryption method and a secure aggregation protocol.

[0007] As an improvement of the present invention, a privacy-preserving method for constructing a building thermal dynamic aggregation model includes the following steps:

[0008] S1, establish a building thermal dynamic aggregation model; first establish a thermal dynamic model of a single building area, and then establish a building thermal dynamic aggregation model based on the aggregation equation:

[0009] in, is the aggregate indoor temperature of the building cluster, represents the thermal power of sub-region i at time tm, Indicates the outdoor temperature at time tm, represents the solar radiation power at time tm, Describe the role of user activities, is the aggregation coefficient of the i-th building area, and the set K={1,2,…,K} represents each building area. Aggregate model parameters for building thermal dynamics;

[0010] S2. Establish a parameter estimation model for the building thermal dynamic aggregation model: Based on the measurement equation, the least squares method is used for parameter estimation, and a regularization term is introduced to address the sparsity problem of the aggregation coefficient. The parameter estimation model of the building thermal dynamic aggregation model in a compact form is obtained:

[0011] in, I M is an M-dimensional 1 vector, is the Kronecker product;

[0012] S3, establish a privacy-preserving building thermal dynamic aggregation model parameter estimation method: decompose the parameter estimation model of the building thermal dynamic aggregation model established in step S2 into two quadratic programming sub-problems, whose mathematical expressions are as follows:

[0013] For sub-problem 1, the building load aggregator completes the privacy-preserving calculation based on the secure aggregation protocol and solves sub-problem 1 to obtain α(ξ), β(ξ), γ(ξ), θ(ξ), τ occ (ξ); For sub-problem 2, the building load aggregator introduces a random transformation matrix and completes the privacy-preserving calculation based on the secure aggregation protocol. Solving sub-problem 2, we get ξ(α), β(α), γ(α), θ(α), τ occ (α); The building load aggregator sets the initial value of the iteration and sets the difference δ between the objective functions of subproblems 1 and 2 to satisfy δ<10 -6As the iteration termination condition, the privacy protection calculations of subproblems 1 and 2 are repeated in sequence until the iteration converges.

[0014] As an improvement of the present invention, the thermal dynamic model of a single building area in step S1 is specifically:

[0015] in, Indicates the indoor temperature. are the parameters of the thermal dynamic model of a single building area, and the set M = {0, 1, …, M} represents the model order;

[0016] The polymerization equation is specifically:

[0017] in, represents the indoor temperature of building area i at time t.

[0018] As an improvement of the present invention, the measurement equation in step S2 is specifically:

[0019] Among them, ε t represents independent and identically distributed Gaussian noise.

[0020] As another improvement of the present invention, in step S3, the privacy-preserving calculation method for sub-problem 1 is as follows:

[0021] S31: Building load aggregators will i Passed to the i-th building area;

[0022] S32: The i-th building area is based on calculate in Representation matrix Column i of

[0023] S33: Generate a random vector for the i-th building area And share it with all other building areas, collect the random vectors shared by other building areas, and perform the following calculations:

[0024] S34: Upload the i-th building area to building load aggregators;

[0025] S35: Building load aggregator calculation c0ξ and

[0026] S36: Building load aggregator solves subproblem 1 to obtain α(ξ), β(ξ), γ(ξ), θ(ξ), τ occ (ξ).

[0027] As another improvement of the present invention, in step S3, a random transformation matrix is ​​introduced, and sub-problem 2 is further transformed into:

[0028] remember for The information required by the building load aggregator to solve subproblem 2 is WW T ,1 T W T , which can be expressed in the form of and respectively:

[0029] remember Based on the secure aggregation protocol, the privacy-preserving calculation method for sub-problem 2 proceeds as follows:

[0030] S31': Generate a random matrix W for the i-th building area [i] ,u i,j ,p i,j ,q i,j And share it with all other building areas, collect the random matrices shared by other building areas, and perform the following calculations:

[0031] S32': Upload the i-th building area to building load aggregators;

[0032] S33': Building load aggregator calculation WW T ,1 T W T :

[0033] S34': Building load aggregator solves subproblem 2

[0034] Compared with existing technologies, this invention offers significant advantages: It proposes a privacy-preserving method for constructing a building thermal dynamic aggregation model. This method can help building load aggregators estimate model parameters for building clusters while protecting user privacy information (indoor temperature, cooling / heating power, etc.). This method investigates privacy-preserving non-convex parameter estimation methods, utilizes block coordinate descent to address non-convex problems, and employs a transformation-based encryption method and a secure aggregation protocol to achieve privacy protection. This method is groundbreaking and demonstrates excellent computational accuracy and privacy protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] FIG1 is a flowchart of the steps of a method for constructing a privacy-protected building thermal dynamic aggregation model according to the present invention;

[0036] FIG2 is a schematic diagram of the structure of a building load aggregator-building cluster system applicable to Example 2 of the present invention;

[0037] FIG3 is a comparison diagram of the predicted value, the true value, and the actual indoor temperature of each area of ​​the building cluster aggregation state in Example 2 of the present invention;

[0038] FIG4 is a comparison diagram of indoor temperature-related information of a certain area of ​​a building cluster before and after encryption in Example 2 of the present invention. DETAILED DESCRIPTION

[0039] 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.

[0040] Example 1

[0041] A privacy-preserving building thermal dynamic aggregation model construction method is applied to a building load aggregator-building cluster system, specifically including the steps shown in Figure 1:

[0042] S1. Establish a thermal dynamic aggregation model for buildings;

[0043] S11. Establish a regional thermal dynamic model for a single building:

[0044] in, Indicates the indoor temperature. represents the thermal power, Indicates the outdoor temperature, represents the solar radiation power, Describe the role of user activities, are the parameters of the thermal dynamic model of a single building area, and the set M = {0, 1, …, M} represents the model order;

[0045] S12. Establish the aggregation equation:

[0046] in, is the aggregate indoor temperature of the building cluster, is the aggregation coefficient of the i-th building area; it is necessary to ensure that the sum of the aggregation coefficients of the building cluster is 1 and the aggregation coefficients are all positive numbers. The set K = {1, 2, …, K} represents each building area;

[0047] S13. Establishing a building thermal dynamic aggregation model:

[0048] Combining the single building area thermal dynamic model established in step S11 with the aggregation equation established in step S12, the building thermal dynamic aggregation model is obtained as follows:

[0049] in, are the building thermal dynamic aggregation model parameters.

[0050] S2. Establish a parameter estimation model for the building thermal dynamic aggregation model;

[0051] S21. Establish the measurement equation:

[0052] Among them, ε t represents independent and identically distributed Gaussian noise;

[0053] S22. Establish a parameter estimation model for the building thermal dynamic aggregation model:

[0054] Based on the measurement equation established in step S21, the least squares method is used to estimate the parameters, and a regularization term is introduced to solve the sparsity problem of the aggregation coefficient.

[0055] The parameter estimation model of the building thermal dynamic aggregation model in a compact form is obtained as follows:

[0056] in,

[0057] I M is an M-dimensional 1 vector, is the Kronecker product.

[0058] S3. Establish a privacy-preserving parameter estimation method for the building thermal dynamic aggregation model;

[0059] S31, iterative algorithm based on block coordinate descent

[0060] Fixing ξ and α respectively, the parameter estimation model of the building thermal dynamic aggregation model established in step S22 is decomposed into two quadratic programming sub-problems, and their mathematical expressions are as follows:

[0061] S32. Sub-problem 1: Privacy-preserving calculation method

[0062] Based on the Secure Aggregation Protocol (SAP), the privacy-preserving calculation method for sub-problem 1 is designed as follows (steps 1-4 must be performed in all building areas):

[0063] (1) Building load aggregators will i Passed to the i-th building area;

[0064] (2) The i-th building area is based on calculate (in Representation matrix ith column of );

[0065] (3) The i-th building area generates a random vector And share it with all other building areas, collect the random vectors shared by other building areas, and perform the following calculations:

[0066] (4) Upload the i-th building area To building load aggregators

[0067] (5) Calculation of building load aggregator c0ξ and

[0068] (6) The building load aggregator solves subproblem 1 and obtains α(ξ), β(ξ), γ(ξ), θ(ξ), τ occ (ξ).

[0069] S33. Sub-problem 2: Privacy-preserving calculation method

[0070] In subproblem 2, α is known, so we define Subproblem 2 is equivalently transformed into:

[0071] Introducing random transformation matrix

[0072] Sub-problem 2 is further transformed into

[0073] remember for The information required by the building load aggregator to solve subproblem 2 is WW T ,1 T W T , which can be expressed in the form of and respectively:

[0074] remember Then, based on the Secure Aggregation Protocol (SAP), the privacy-preserving calculation method for sub-problem 2 is designed as follows:

[0075] (1) The random matrix W is generated in the i-th building area [i] ,u i,j ,p i,j ,q i,j And share it with all other building areas, collect the random matrices shared by other building areas, and perform the following calculations:

[0076] (2) Upload the i-th building area To building load aggregators

[0077] (3) Calculation of building load aggregator WW T ,1 T W T :

[0078] (4) Building load aggregator solves subproblem 2.

[0079] In this embodiment, the overall process of the parameter estimation method of the privacy-preserving building thermal dynamic aggregation model is as follows: the building load aggregator sets the initial value of the iteration, sets the difference δ between the objective functions of subproblem 1 and subproblem 2 to satisfy δ<10 -6 As the iteration termination condition, the privacy protection calculation method of sub-problem 1 in S32 and sub-problem 2 in S33 is repeated in sequence until the iteration converges.

[0080] Example 2

[0081] The building load aggregator-building cluster system in this embodiment consists of a building load aggregator and seven buildings (containing a total of 64 building areas). The aggregator performs aggregation modeling on the seven buildings, as shown in Figure 2. A total of 1440 simulation data items are used, of which 1080 items are training data and the remaining 360 items are testing data. The simulation interval is set to 30 minutes, the user activity cycle is 24 hours, the random matrix follows a normal distribution with a mean of 0.1 and a standard deviation of 0.1, the thermal dynamic aggregation model order M is set to 2, and the penalty coefficient λ is set to 100.

[0082] According to the steps of the present invention, the parameter estimation model of the privacy-protected thermal dynamic aggregation model is solved. Figure 3 is a comparison diagram of the predicted value, true value, and actual indoor temperature of each area of ​​the building cluster aggregation state in Example 2; Figure 4 is a comparison diagram of the indoor temperature-related information of a certain area of ​​the building cluster in Example 2 before and after encryption. As can be seen from Figure 3, the predicted value of the aggregation state is very close to the actual value, and the model prediction is accurate; as can be seen from Figure 4, the difference between the sensitive information before and after encryption is large and there is no pattern to follow, and the encryption effect is good. Therefore, the parameter estimation model of the privacy-protected thermal dynamic aggregation model of the present invention has high accuracy, good prediction effect, and can effectively protect user privacy information from leakage.

[0083] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0084] 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 method for constructing a privacy - protected building thermal dynamic aggregation model, characterized in that: A thermal dynamic aggregation model of a building cluster and its corresponding parameter estimation model are established. According to the block - coordinate descent theory, the original non - convex parameter estimation problem is transformed into two convex optimization problems for iterative solution, and the protection of user privacy information is realized according to the transformation - based encryption method and the secure aggregation protocol.

2. The method for constructing a privacy - protected building thermal dynamic aggregation model according to claim 1, characterized in that , including the following steps: S1. Establish a building thermal dynamic aggregation model; First, establish the thermal dynamic model of a single building area, and based on the aggregation equation, establish the building thermal dynamic aggregation model: Among them, is the aggregated indoor temperature of a building cluster, Indicates the thermal power of sub-region i at time t - m Indicates the outdoor temperature from time t to time m. Indicates the solar radiation power from time t to time m. Describe the role of user activities, is the aggregation coefficient for the i-th building area, and the set K = {1, 2, …, K} represents each building area, is the parameter of the building thermal dynamic aggregation model; S2. Establish a parameter estimation model for the building thermal dynamic aggregation model: Based on the measurement equation, use the least squares method for parameter estimation, and introduce a regularization term to solve the sparsity problem of the aggregation coefficient, obtaining a parameter estimation model for the building thermal dynamic aggregation model in a compact form: Among them, I M is an M-dimensional 1 vector, is the Kronecker product; S3. Establish a parameter estimation method for the privacy-protected building thermal dynamic aggregation model: Decompose the parameter estimation model of the building thermal dynamic aggregation model established in step S2 into two quadratic programming sub-problems, and their mathematical expressions are as follows respectively: For sub-problem 1, the building load aggregator completes privacy-preserving calculations based on the secure aggregation protocol and solves sub-problem 1 to obtain α(ξ), β(ξ), γ(ξ), θ(ξ), τ occ (ξ); for sub-problem 2, the building load aggregator introduces a random transformation matrix, completes privacy-preserving calculations based on the secure aggregation protocol, and solves sub-problem 2 to obtain ξ(α), β(α), γ(α), θ(α), τ occ (α); the building load aggregator sets the initial iteration value and sets the difference δ before the objective functions of sub-problem 1 and sub-problem 2 to satisfy δ < 10 -6 as the iteration termination condition, and sequentially repeats the privacy-preserving calculations of sub-problem 1 and sub-problem 2 until the iteration converges.

3. The method for constructing a privacy - protected building thermal dynamic aggregation model according to claim 2, characterized in that: The specific thermal dynamic model of a single building area in the step S1 is as follows: Among them, Indicates the indoor temperature, is the parameter of the thermal dynamic model of a single building area, and the set M = {0, 1, …, M} represents the model order; The specific aggregation equation is as follows: Among them, represents the indoor temperature of building area i at time t.

4. The method for constructing a privacy - protected building thermal dynamic aggregation model according to claim 2, characterized in that: The measurement equation in the step S2 is specifically as follows: where ε t represents independent and identically distributed Gaussian noise.

5. The method for constructing a privacy - protected building thermal dynamic aggregation model according to claim 2, characterized in that: In the step S3, the privacy - protected calculation method process of sub - problem 1 is as follows: S31: The building load aggregator transmits ξ i to the i-th building area; S32: The i-th building area is based on Calculation Among them Indicates a matrix the i - th column of; S33: The i-th building area generates a random vector And share it with all other building areas, and collect the random vectors shared by other building areas, and perform the following calculations: S34: Upload by the i-th building area to the building load aggregator; S35: The building load aggregator calculates c 0 ξ and S36: The building load aggregator solves sub-problem 1 to obtain α(ξ), β(ξ), γ(ξ), θ(ξ), τ occ (ξ).

6. The method for constructing a privacy - protected building thermal dynamic aggregation model according to claim 2, characterized in that: In the step S3, a random transformation matrix is introduced, and sub-problem 2 is further transformed into: Remember For For the i-th column, the information required for the building load aggregator to solve sub-problem 2 is WW T ,1 T W T , can be respectively expressed in the forms of and: Record Based on the secure aggregation protocol, the privacy - protected calculation method process of sub - problem 2 is as follows: S31’: The i-th building area generates a random matrix W [i] , u i,j , p i,j , q i,j And share it with all other building areas, and collect the random matrices shared by other building areas, and perform the following calculations: S32’: Upload from the i-th building area to the building load aggregator; S33’: Building load aggregator calculation WW T ,1 T W T : S34’: The building load aggregator solves sub - problem 2.

Images