Air conditioner load parameter trusted aggregation method based on cloud-edge encryption

Abstract

The application relates to the technical field of air conditioner load, and particularly discloses an air conditioner load parameter credible aggregation method based on cloud-edge encryption, which is based on a cloud-edge collaborative architecture, uses a public cloud server as a cloud platform, and uses an edge computing device as a user-side intelligent terminal; a data transmission mode based on a file transmission protocol is used for cloud-edge communication; the cloud-edge collaborative architecture comprises three cloud servers; the cloud server 3 is responsible for distributing an encryption key to users and distributing a decryption key to the cloud server 1; the cloud server 2 is responsible for clustering user parameters; and the cloud server 1 is responsible for decrypting the clustered parameters and solving a demand response load reduction optimization problem to determine an optimal strategy. In this way, it is ensured that each cloud server can only master partial information of all users, the safety of data flow is enhanced, the clustered user parameters are encrypted on the cloud, the user data privacy is protected, and the solving scale of the demand response load reduction optimization problem is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air conditioning load, and more particularly, to an air conditioning load parameter trusted aggregation method based on cloud-edge encryption. BACKGROUND

[0002] Air conditioning load is an important adjustable load, which can be used to realize load shedding of power system, i.e. demand response. There is a clear master-slave relationship between demand response implementer (usually load aggregation subject) and air conditioning load. The load aggregation subject as the "master" is responsible for formulating the demand response plan; the air conditioning user as the "slave" adjusts its own power consumption behavior according to the plan. In this scenario, a bi-level optimization problem of demand response load shedding is formed, in which the profit maximization of the load aggregation subject and the minimum comprehensive cost of the user's air conditioning are balanced. For the solution of the optimal demand response strategy in the bi-level optimization model, in the existing method, the load shedding implementer obtains the air conditioning user's own parameters, converts the user's optimization problem into Karush-Kuhn-Tucker equivalent conditions, and integrates them into the constraint conditions of the load shedding implementer's benefit maximization problem to form a Mathematical Programs with Equilibrium Constraints (MPEC) problem, and performs linearization to obtain a Mixed-Integer Linear Program (MILP), and finally the optimal load shedding strategy is solved.

[0003] However, this method has some obvious limitations. Specifically, the air conditioning power consumption preferences and other private data of users participating in demand response will be directly exposed in the database of the load shedding implementer, so the personal data security of the users cannot be fully guaranteed. Moreover, the seasonal characteristics of air conditioning load are obvious, and the dedicated server only runs in a few periods of the year, while it is idle in other periods, which will cause waste of computing resources. At the same time, the participation of a large number of users brings the problem of variable explosion to the MPEC or MILP of demand response load shedding optimization problem, which brings difficulties to the solution of the optimal strategy.

[0004] Therefore, an air conditioning load parameter trusted aggregation method based on cloud-edge encryption is expected. SUMMARY

[0005] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a cloud-edge encryption-based air conditioner load parameter trusted aggregation method, which is based on a cloud-edge collaborative architecture, uses a public cloud server as a cloud platform, an edge computing device as a user-side intelligent terminal, and adopts a file transfer protocol-based data transmission mode for cloud-edge communication. The cloud-edge collaborative architecture includes three cloud servers. Cloud server 3 is responsible for distributing encryption keys to users and distributing decryption keys to cloud server 1. Cloud server 2 is responsible for clustering user parameters. Cloud server 1 is responsible for decrypting the clustered parameters and solving a demand response load reduction optimization problem to determine an optimal strategy. In this way, each cloud server can only master partial information of all users, enhancing the security of data flow, and by encrypting the clustered user parameters on the cloud, the privacy of user data is protected while the size of the demand response load reduction optimization problem is reduced.

[0006] According to an aspect of the present application, a cloud-edge encryption-based air conditioner load parameter trusted aggregation method is provided, which includes:

[0007] The third cloud server generates an encryption key and a decoding key, and stores the encryption key and the decoding key in a first FTP folder;

[0008] The edge side obtains the encryption key from the first FTP folder, encrypts the air conditioner load parameter using the encryption key to obtain an encrypted air conditioner load parameter, and transmits the encrypted air conditioner load parameter to a second FTP folder, wherein the air conditioner load parameter includes an air conditioner temperature preference coefficient and a most comfortable temperature setting;

[0009] The second cloud server extracts the encrypted air conditioner load parameter from the second FTP folder and performs clustering analysis on the encrypted air conditioner load parameter to obtain a clustering result, and then writes the clustering result into a third FTP folder;

[0010] The first cloud server extracts the decoding key from the first FTP folder and the clustering result from the third FTP folder, and calculates an optimal demand response strategy based on the clustering result.

[0011] In the above cloud-edge encryption-based air conditioner load parameter trusted aggregation method, the first cloud server extracts the decoding key from the first FTP folder and the clustering result from the third FTP folder, and calculates an optimal demand response strategy based on the clustering result, including using an MPEC or MILP method to process the clustering result to obtain an optimal demand response strategy, the optimal demand response strategy being a recommended air conditioner temperature setting.

[0012] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, the first cloud server extracts the decoding key from the first FTP folder and extracts the clustering result from the third FTP folder, and calculates the optimal demand response strategy based on the clustering result, including: decrypting the clustering result using the decoding key to obtain a decrypted clustering result; embedding and encoding each clustering center in the decrypted clustering result using a class embedding matrix to obtain a set of air conditioner parameter clustering center embedding and encoding vectors; performing feature dynamic aggregation on the set of air conditioner parameter clustering center embedding and encoding vectors to obtain an air conditioner parameter global significant integration representation vector; and generating the optimal demand response strategy based on the air conditioner parameter global significant integration representation vector.

[0013] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, the feature dynamic aggregation on the set of air conditioner parameter clustering center embedding and encoding vectors to obtain an air conditioner parameter global significant integration representation vector includes: determining an air conditioner parameter global clustering initial center vector based on a feature distribution field of the set of air conditioner parameter clustering center embedding and encoding vectors; and performing significant modulation dynamic aggregation on the set of air conditioner parameter clustering center embedding and encoding vectors to obtain the air conditioner parameter global significant integration representation vector based on the spatial span of each air conditioner parameter clustering center embedding and encoding vector in the set of air conditioner parameter clustering center embedding and encoding vectors relative to the air conditioner parameter global clustering initial center vector.

[0014] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, the determination of the air conditioner parameter global clustering initial center vector based on the feature distribution field of the set of air conditioner parameter clustering center embedding and encoding vectors includes: calculating a static energy factor of each air conditioner parameter clustering center embedding and encoding vector in the set of air conditioner parameter clustering center embedding and encoding vectors to obtain a set of air conditioner parameter static energy factors; and selecting an air conditioner parameter clustering center embedding and encoding vector corresponding to the maximum value in the set of air conditioner parameter static energy factors as the air conditioner parameter global clustering initial center vector.

[0015] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, the calculation of the static energy factor of each air conditioner parameter clustering center embedding and encoding vector in the set of air conditioner parameter clustering center embedding and encoding vectors to obtain a set of air conditioner parameter static energy factors includes: calculating the kurtosis of the air conditioner parameter clustering center embedding and encoding vector, and inputting the kurtosis into a sigmoid activation function to obtain the air conditioner parameter static energy factor.

[0016] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, based on the spatial span of each air conditioner parameter cluster center embedding coding vector in the set of air conditioner parameter cluster center embedding coding vectors relative to the air conditioner parameter global cluster initial center vector, the set of air conditioner parameter cluster center embedding coding vectors is dynamically aggregated based on the significance modulation to obtain the air conditioner parameter global significant integrated representation vector, including: based on the spatial span between each air conditioner parameter cluster center embedding coding vector in the set of air conditioner parameter cluster center embedding coding vectors and the air conditioner parameter global cluster initial center vector, and the set of air conditioner parameter static energy factors, the dynamic aggregation energy factor of each air conditioner parameter cluster center embedding coding vector in the set of air conditioner parameter cluster center embedding coding vectors is calculated to obtain the set of air conditioner parameter dynamic aggregation energy factors; the set of air conditioner parameter dynamic aggregation energy factors is input into the gate mask unit to obtain the set of air conditioner parameter dynamic aggregation weight factors; based on the set of air conditioner parameter dynamic aggregation weight factors, the weighted sum of the set of air conditioner parameter cluster center embedding coding vectors is calculated to obtain the air conditioner parameter global significant integrated representation vector.

[0017] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, the dynamic aggregation energy factor of each air conditioner parameter cluster center embedding coding vector in the set of air conditioner parameter cluster center embedding coding vectors is calculated to obtain the set of air conditioner parameter dynamic aggregation energy factors, including: taking the square value of the number of feature vectors separated between the air conditioner parameter cluster center embedding coding vector and the air conditioner parameter global cluster initial center vector as a spatial span coefficient, calculating the weighted ratio between the product of the static energy factor of the air conditioner parameter cluster center embedding coding vector and the static energy factor of the air conditioner parameter global cluster initial center vector and the spatial span coefficient to obtain the air conditioner parameter dynamic aggregation energy factor.

[0018] In the cloud-edge encryption-based air conditioner load parameter credible aggregation method, based on the air conditioner parameter global significant integrated representation vector, the optimal demand response strategy is generated, including: inputting the air conditioner parameter global significant integrated representation vector into the decoder-based optimal demand response strategy generator to obtain the optimal demand response strategy.

[0019] Compared with the prior art, the cloud-edge encryption-based air conditioner load parameter credible aggregation method provided by the application uses a public cloud server as a cloud platform and an edge computing device as a user-side intelligent terminal, adopts a file transfer protocol-based data transmission mode for cloud-edge communication, and comprises a cloud-edge collaborative architecture, wherein the cloud-edge collaborative architecture comprises three cloud servers, cloud server 3 is responsible for distributing an encryption key to users and distributing a decryption key to cloud server 1, cloud server 2 is responsible for clustering user parameters, and cloud server 1 is responsible for decrypting the clustered parameters and solving a demand response load reduction optimization problem to determine an optimal strategy. In this way, it is ensured that each cloud server can only master partial information of all users, the security of data flow is enhanced, the clustering of user parameters is encrypted on the cloud, the privacy of user data is protected, and the scale of solving the demand response load reduction optimization problem is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application, when taken in conjunction with the accompanying drawings. The drawings provided in the present application are used to provide further understanding of embodiments of the present application, and constitute a part of the specification, and are used to explain the present application together with embodiments of the present application, and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0021] Figure 1 A flowchart of cloud-edge encryption clustering of user personal data in the cloud-edge encryption-based air conditioner load parameter credible aggregation method according to the embodiments of the present application.

[0022] Figure 2 A flowchart of a parameter Euclidean distance clustering method of a user in the cloud-edge encryption-based air conditioner load parameter credible aggregation method according to the embodiments of the present application.

[0023] Figure 3 A flowchart of substep S4 of the cloud-edge encryption-based air conditioner load parameter credible aggregation method according to the embodiments of the present application.

[0024] Figure 4 A data flow schematic diagram of substep S4 of the cloud-edge encryption-based air conditioner load parameter credible aggregation method according to the embodiments of the present application.

[0025] Figure 5 A flowchart of substep S43 of the cloud-edge encryption-based air conditioner load parameter credible aggregation method according to the embodiments of the present application. DETAILED DESCRIPTION

[0026] As shown in the present application and claims, unless the context clearly indicates otherwise, the words "a," "an," "the," and / or "this" are not limited in scope to the singular, but rather include the plural. Generally, the terms "comprises" and "comprising" are not intended to be construed as limiting, but rather, are intended to be construed as including steps and elements that are specifically identified, but also including other steps or elements not specifically identified.

[0027] While the present application makes various references to certain modules in the system according to embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative, and different aspects of the system and method can use different modules.

[0028] Flowcharts are used in the present application to illustrate operations performed by the system according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in sequence. Rather, various steps can be processed in reverse order or simultaneously, as desired. Other operations can also be added to or removed from these processes.

[0029] In the following, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It is obvious that the described embodiments are only a part of the embodiments of the present application, and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.

[0030] To solve the above technical problems, the present application proposes an optimized cloud-edge encryption-based air conditioner load parameter trusted aggregation method, which is based on a cloud-edge collaborative architecture, uses a public cloud server as a cloud platform, an edge computing device as a user-side intelligent terminal, and adopts a file transfer protocol-based data transmission method for cloud-edge communication. The cloud-edge collaborative architecture includes three cloud servers. Cloud server 3 is responsible for distributing encryption keys to users and distributing decryption keys to cloud server 1. Cloud server 2 is responsible for clustering user parameters. Cloud server 1 is responsible for decrypting the clustered parameters and solving the demand response load reduction optimization problem to determine the optimal strategy. In this way, each cloud server can only master partial information of all users, enhancing the security of data flow, and by encrypting the clustered user parameters on the cloud, the user data privacy is protected while the size of the demand response load reduction optimization problem is reduced.

[0031] Figure 1 A flowchart of cloud-edge encryption clustering of user personal data in the cloud-edge encryption-based air conditioner load parameter trusted aggregation method according to embodiments of the present application. As shown in Figure 1As shown, the cloud-edge encryption-based air conditioning load parameter credible aggregation method comprises the following steps: S1, a third cloud server generates an encryption key and a decoding key, and stores the encryption key and the decoding key in a first FTP folder; S2, an edge side obtains the encryption key from the first FTP folder, encrypts air conditioning load parameters using the encryption key to obtain encrypted air conditioning load parameters, and transmits the encrypted air conditioning load parameters to a second FTP folder, wherein the air conditioning load parameters comprise an air conditioning temperature preference coefficient and a most comfortable temperature setting; S3, a second cloud server extracts the encrypted air conditioning load parameters from the second FTP folder and performs cluster analysis on the encrypted air conditioning load parameters to obtain a clustering result, and then writes the clustering result into a third FTP folder; and S4, a first cloud server extracts the decoding key from the first FTP folder and extracts the clustering result from the third FTP folder, and calculates an optimal demand response strategy based on the clustering result.

[0032] In the cloud-edge encryption-based air conditioning load parameter credible aggregation method, in the step S1, the third cloud server generates an encryption key and a decoding key, and stores the encryption key and the decoding key in a first FTP folder. The cloud server 3 generates a key, and writes the encryption key X and the decryption key X' into the FTP1 folder. In order to maximize the characteristics of the original data of the user, linear encryption and decryption are selected, that is, the encrypted data is multiplied by a constant, and the decrypted data is divided by the same constant.

[0033] In the cloud-edge encryption-based air conditioning load parameter credible aggregation method, in the step S2, the edge side obtains the encryption key from the first FTP folder, encrypts air conditioning load parameters using the encryption key to obtain encrypted air conditioning load parameters, and transmits the encrypted air conditioning load parameters to a second FTP folder, wherein the air conditioning load parameters comprise an air conditioning temperature preference coefficient and a most comfortable temperature setting. It can be understood that the key of the user in the demand response is to minimize the overall cost, and the core is the temperature discomfort cost, which is a quadratic form, indicating that the farther the air conditioning setting temperature deviates from the ideal temperature, the higher the discomfort cost of the user:

[0034] C discomfort,m = ω m Δi(T set,m -T ideal,m ) 2

[0035] wherein ω m is a temperature preference coefficient of the user m, T set,m is an air conditioning temperature setting of the user m, and T ideal,mis the most comfortable temperature value of the user m.

[0036] Based on this, the present application focuses on the air conditioning temperature preference coefficient ω m and the most comfortable temperature setting T ideal,m Perform clustering. The user reads the encryption key X from the FTP1 folder, encrypts the temperature preference coefficient and the most comfortable temperature setting of the user, and writes the parameters such as the most comfortable temperature setting into the FTP2 folder. The encryption formula is:

[0037] ω′ m = X (ω m ) = τω m

[0038] T′ ideal,m = X (T ideal,m ) = τT ideal,m

[0039] Where ω′ m represents the temperature preference coefficient encrypted by the key, T′ ideal,m represents the most comfortable temperature setting encrypted by the key, τ represents a constant for user parameter encryption, and the total number of users is M.

[0040] In the above cloud-edge encryption-based air conditioning load parameter trusted aggregation method, the step S3, the second cloud server extracts the encrypted air conditioning load parameter from the second FTP folder and performs clustering analysis on the encrypted air conditioning load parameter to obtain a clustering result, and then writes the clustering result into a third FTP folder. After the cloud server 2 reads the encrypted air conditioning parameters from the FTP2 folder, it performs clustering using the K-means algorithm based on the Euclidean distance (Euclidean Distance), avoiding the influence of data encryption operation on the clustering result. Then, write the user parameters after clustering to FTP3 folder. The detailed steps of data clustering are shown in Figure 2 :

[0041] Step 3.1: Initialize K cluster centers.

[0042] The cloud server 2 selects K samples as the initial cluster centers o = o1,...,o k ,...,o K :

[0043] o k = (T′ ideal,k , ω′ k )

[0044] Step 3.2: Divide the data into K classes.

[0045] Cloud server 2 computes the Euclidean distance of each sample i in the dataset to the k cluster centers and divides it to the class Ξ corresponding to the cluster center with the smallest distance k In:

[0046]

[0047] Step 3.3: Recompute the cluster centers of the K classes of data.

[0048] Cloud server 2 recomputes the cluster centers of all data in each class (the centroid of all samples belonging to the class):

[0049]

[0050] Where x k is the number of sample data contained in the set Ξ k .

[0051] Step 3.4: Determine whether the iteration is over.

[0052] Cloud server 2 repeats the operations of 3.2 and 3.3 until the appropriate centroid is selected: so that within each cluster, the distance d cri of the sample to the centroid is as small as possible, ensuring that the samples within each cluster have high similarity:

[0053]

[0054] The K groups of data after clustering are as follows:

[0055] {(T′ ideal,1 ,ω′1),...,(T′ ideal,k ,ω′ k ),...,(T′ ideal,K ,ω′ K )}

[0056] In the above cloud-edge encryption-based air conditioning load parameter credible aggregation method, the step S4, the first cloud server extracts the decoding key from the first FTP folder and extracts the clustering result from the third FTP folder, and calculates the best demand response strategy based on the clustering result. Cloud server 1 reads the decryption key in FTP1 folder and reads the user parameters after clustering in FTP3 folder, and then performs data decryption operation:

[0057]

[0058] At this point, the user's air conditioner temperature setting preference coefficient and the most comfortable temperature setting will be reduced from M to K, and the variables of the MPEC or MILP problem will also be reduced to K / M times of the original. The cloud server 1 solves the demand response load reduction optimization problem (MPEC or MILP problem) containing K groups of variables according to the clustered user parameters to obtain the optimal demand response strategy.

[0059] In particular, the present application takes into account the construction of MPEC and MILP problem models needs to consider the actual situation and constraints of the problem. However, in practical applications, there are often various complexities and uncertainties, which may cause the constructed MILP model to be unable to fully adapt to the actual situation. For example, when some parameters or constraints in the problem change, the model structure may need to be redefined or the parameters adjusted and solved, which further limits the flexibility and adaptability of the model. In another embodiment of the present application, an optimized optimal demand response strategy solving method is proposed, which embeds and encodes each cluster center in the decrypted clustering result by introducing a deep learning algorithm to learn the intrinsic features of the air conditioner parameter cluster center, and then based on the feature distribution field of each air conditioner parameter cluster center, the feature saliency modulation and dynamic aggregation are performed to further identify and strengthen the key user load features, and based on this, the intelligent decoding prediction of the best demand response strategy is realized. In this way, the adaptability and flexibility of the model can be effectively improved, and the strategy can be dynamically adjusted according to real-time data to cope with environmental changes and user behavior uncertainties, thereby providing more accurate and personalized load reduction schemes for users and grid operators, achieving efficient use of energy and minimization of cost.

[0060] Figure 3 Flowchart for sub-step S4 of the cloud-edge encryption-based air conditioner load parameter trusted aggregation method according to the embodiment of the present application. Figure 4 Data flow diagram for sub-step S4 of the cloud-edge encryption-based air conditioner load parameter trusted aggregation method according to the embodiment of the present application. As shown in Figure 3 and Figure 4 As shown in the steps S4, it includes the steps of: S41, decrypting the clustering result using the decoding key to obtain a decrypted clustering result; S42, embedding and encoding each cluster center in the decrypted clustering result using a class embedding matrix to obtain a set of air conditioner parameter cluster center embedding code vectors; S43, performing feature dynamic aggregation on the set of air conditioner parameter cluster center embedding code vectors to obtain an air conditioner parameter global significant integrated representation vector; S44, generating the best demand response strategy based on the air conditioner parameter global significant integrated representation vector.

[0061] Specifically, the step S41, using the decoding key to decrypt the clustering result to obtain the decrypted clustering result. Here, the specific operation of using the decoding key to decrypt the clustering result has been described in detail above, and therefore, the repeated description thereof will be omitted.

[0062] Specifically, the step S42, using the class embedding matrix to embed encode each cluster center in the decrypted clustering result to obtain a set of air conditioner parameter cluster center embedding encoding vectors. It should be understood that each cluster center in the decrypted clustering result respectively represents the air conditioner use preference and comfort requirement of different user groups. In order to effectively capture the relevance and difference between the air conditioner load parameter characteristics of different user groups, so as to better formulate the demand response strategy, the present application uses the class embedding matrix to embed encode each cluster center in the decrypted clustering result, so as to map each cluster center data to a unified low-dimensional continuous vector space. Through the embedding encoding technology, the inherent characteristics of the original clustering information are extracted, and the potential association structure between the air conditioner load parameter characteristics of different user groups is revealed based on the relative position relationship of each cluster center in the feature space, thereby providing more useful data support for subsequent demand response strategy formulation.

[0063] Specifically, the step S43, performing feature dynamic aggregation on the set of air conditioner parameter cluster center embedding encoding vectors to obtain an air conditioner parameter global significant integrated representation vector. It should be understood that in order to further analyze the overall characteristics of the user air conditioner load parameters from a global perspective, it is necessary to aggregate the features of the set of air conditioner parameter cluster center embedding encoding vectors, so as to integrate the air conditioner use preference and comfort requirement information of different user groups, and realize the global understanding of the user air conditioner load demand. In particular, in the feature aggregation process, in order to effectively improve the discrimination and representativeness of the aggregated features, the present application proposes a dynamic aggregation method based on feature distribution field, which dynamically adjusts the weight of feature aggregation by analyzing the relative position relationship and feature energy distribution between each cluster center, thereby strengthening the identification and expression of key features and suppressing the interference of non-key features, improving the efficiency and accuracy of feature aggregation. Among them, Figure 5 The flowchart of the sub-step S43 of the cloud-edge encryption based air conditioner load parameter trusted aggregation method according to the embodiment of the present application. As shown in FIG. 6, the step S43 includes steps S431-S433. Figure 5As shown, the step S43 comprises steps of: S431, determining an air conditioner parameter global clustering initial center vector based on a feature distribution field of the set of air conditioner parameter clustering center embedded encoding vectors; S432, performing significant modulation dynamic aggregation on the set of air conditioner parameter clustering center embedded encoding vectors to obtain the air conditioner parameter global significant integrated representation vector based on a spatial span of each air conditioner parameter clustering center embedded encoding vector in the set of air conditioner parameter clustering center embedded encoding vectors relative to the air conditioner parameter global clustering initial center vector.

[0064] More specifically, the step S431 further comprises: calculating a static energy factor of each air conditioner parameter clustering center embedded encoding vector in the set of air conditioner parameter clustering center embedded encoding vectors to obtain a set of air conditioner parameter static energy factors; and selecting an air conditioner parameter clustering center embedded encoding vector corresponding to a maximum value in the set of air conditioner parameter static energy factors as the air conditioner parameter global clustering initial center vector.

[0065] In one specific example of the present application, calculating a static energy factor of each air conditioner parameter clustering center embedded encoding vector in the set of air conditioner parameter clustering center embedded encoding vectors to obtain a set of air conditioner parameter static energy factors comprises: calculating a kurtosis of the air conditioner parameter clustering center embedded encoding vector and inputting the kurtosis into a sigmoid activation function to obtain the air conditioner parameter static energy factor. That is, the static energy factor of each air conditioner parameter clustering center embedded encoding vector is calculated based on the kurtosis thereof to evaluate the stability and importance thereof in the feature space, and the air conditioner parameter clustering center embedded encoding vector corresponding to the maximum static energy factor is selected as the air conditioner parameter global clustering initial center vector to determine a cluster core of the feature distribution of each air conditioner parameter clustering center, thereby providing a stable reference point for subsequent feature aggregation.

[0066] More specifically, the step S432 further comprises: calculating a dynamic aggregation energy factor of each air conditioner parameter clustering center embedded encoding vector in the set of air conditioner parameter clustering center embedded encoding vectors to obtain a set of air conditioner parameter dynamic aggregation energy factors based on a spatial span between each air conditioner parameter clustering center embedded encoding vector in the set of air conditioner parameter clustering center embedded encoding vectors and the air conditioner parameter global clustering initial center vector, and the set of air conditioner parameter static energy factors; inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit to obtain a set of air conditioner parameter dynamic aggregation weight factors; and calculating a weighted sum of the set of air conditioner parameter clustering center embedded encoding vectors based on the set of air conditioner parameter dynamic aggregation weight factors to obtain the air conditioner parameter global significant integrated representation vector.

[0067] In a specific example of the present application, the dynamic aggregation energy factor of each air conditioner parameter cluster center embedding encoding vector in the set of air conditioner parameter cluster center embedding encoding vectors is calculated to obtain a set of air conditioner parameter dynamic aggregation energy factors, including: taking the square value of the number of feature vectors between the air conditioner parameter cluster center embedding encoding vector and the air conditioner parameter global cluster initial center vector as a space span coefficient, calculating the weighted ratio between the product of the static energy factor of the air conditioner parameter cluster center embedding encoding vector and the static energy factor of the air conditioner parameter global cluster initial center vector and the space span coefficient to obtain the air conditioner parameter dynamic aggregation energy factor.

[0068] In a specific example of the present application, the set of air conditioner parameter dynamic aggregation energy factors is input into a gating mask unit, including: inputting the set of air conditioner parameter dynamic aggregation energy factors into a sigmoid function for normalization processing to obtain a set of normalized air conditioner parameter dynamic aggregation energy factors; based on a preset mask threshold, the set of normalized air conditioner parameter dynamic aggregation energy factors is inactivated to obtain the set of air conditioner parameter dynamic aggregation weight factors.

[0069] That is, by considering the feature space proximity of each air conditioner parameter cluster center embedding encoding vector relative to the cluster core of the air conditioner parameter cluster center feature distribution and the importance difference of the intrinsic attribute, the contribution of the air conditioner parameter cluster center embedding encoding vector in the global feature of the user air conditioner load parameter is more accurately measured. Then, based on the gating mask mechanism, the dynamic aggregation energy factor obtained is nonlinearly transformed and screened to generate the corresponding weight, so as to weight and aggregate the set of air conditioner parameter cluster center embedding encoding vectors, thereby enhancing the expression of important features and suppressing the influence of irrelevant or noise features, to obtain an air conditioner parameter global significant integrated representation vector. In this way, not only the overall trend and pattern of the user group in air conditioner use preference and comfort demand are captured, but also the user preference and comfort demand information that has a greater impact on demand response strategy making can be given priority consideration, thereby providing a solid data foundation for formulating an efficient demand response strategy.

[0070] Correspondingly, the step S43 includes: processing the set of air conditioner parameter cluster center embedding encoding vectors by a feature dynamic aggregation formula to obtain the air conditioner parameter global significant integrated representation vector, wherein the feature dynamic aggregation formula is:

[0071] X1={x1,x2,...,x i ,...,x n}

[0072]

[0073] xc = x m

[0074]

[0075]

[0076] w si = mask(w i )

[0077]

[0078] wherein X1 represents a set of the air conditioner parameter cluster center embedding encoding vectors, x1, x2, x i , x m and x n represent the first, the second, the i-th, the m-th and the n-th air conditioner parameter cluster center embedding encoding vector in the set of the air conditioner parameter cluster center embedding encoding vectors respectively, n is the number of the air conditioner parameter cluster center embedding encoding vectors, x ij represents the feature value at the j-th position in the i-th air conditioner parameter cluster center embedding encoding vector, μ i and σ i 4 represent the feature mean and the square of the feature variance of the i-th air conditioner parameter cluster center embedding encoding vector respectively, E{·} represents the expectation value of a set, sigmoid represents a sigmoid activation function, represents the static energy factor of the i-th air conditioner parameter cluster center embedding encoding vector, argmax represents the index corresponding to the maximum value, m represents the index of the maximum static energy factor in the set of the air conditioner parameter static energy factors, x c represents the air conditioner parameter global cluster initial center vector, represents the static energy factor of the air conditioner parameter global cluster initial center vector, a and b are different weight parameters, Count(x i → x m ) represents the number of feature vectors between the i-th air conditioner parameter cluster center embedding encoding vector and the air conditioner parameter global cluster initial center vector, e xi represents the dynamic aggregated energy factor of the i-th air conditioner parameter cluster center embedding encoding vector, w i represents the i-th normalized air conditioner parameter dynamic aggregated energy factor, mask(·) represents a mask processing, θ is a preset mask threshold, w si represents the i-th air conditioner parameter dynamic aggregated weight factor, V represents the air conditioner parameter global significant integration representation vector.

[0079] Specifically, the step S44 generates the optimal demand response strategy based on the air conditioner parameter global significant integration representation vector. In a specific example of the present application, the step S44 includes: inputting the air conditioner parameter global significant integration representation vector into a decoder-based optimal demand response strategy generator to obtain the optimal demand response strategy. In the technical solution of the present application, the decoder is based on a multi-layer neural network architecture, which can effectively learn the user air conditioner load information contained in the air conditioner parameter global significant integration representation vector and establish a mapping relationship between the user air conditioner load information and the demand response strategy through training optimization, thereby realizing intelligent prediction of the optimal demand response strategy.

[0080] In a preferred example of the present application, inputting the air conditioner parameter global significant integration representation vector into the decoder-based optimal demand response strategy generator to obtain the optimal demand response strategy includes: determining an air conditioner parameter global significant integration correlation response matrix and an air conditioner parameter global significant integration distance response matrix based on the correlation value and the distance value of the feature value of the air conditioner parameter global significant integration representation vector; performing matrix multiplication on the air conditioner parameter global significant integration representation vector and the air conditioner parameter global significant integration correlation response matrix to obtain an air conditioner parameter global significant integration correlation response vector; performing matrix multiplication on the air conditioner parameter global significant integration distance response matrix and the transpose vector of the air conditioner parameter global significant integration representation vector to obtain an air conditioner parameter global significant integration distance response vector; performing point multiplication on the air conditioner parameter global significant integration correlation response matrix and the air conditioner parameter global significant integration distance response matrix, and then performing matrix multiplication on the result and the transpose vector of the air conditioner parameter global significant integration representation vector to obtain an air conditioner parameter global significant integration logical bias vector; performing point addition on the air conditioner parameter global significant integration correlation response vector, the air conditioner parameter global significant integration distance response vector and the air conditioner parameter global significant integration logical bias vector to obtain an optimized air conditioner parameter global significant integration representation vector; and inputting the optimized air conditioner parameter global significant integration representation vector into the decoder-based optimal demand response strategy generator to obtain the optimal demand response strategy.

[0081] Correspondingly, the optimization process of the air conditioner parameter global significant integration representation vector is represented by the following optimization formula:

[0082] M1(i,j)=v i ×v j

[0083]

[0084] v i ,v j ∈V∈R 1×L

[0085] V1∈R 1×L

[0086] V2∈R L×1

[0087]

[0088]

[0089] wherein, M1 is a global significant integration correlation response matrix of air conditioning parameters, M2 is a global significant integration distance response matrix of air conditioning parameters, M1(i,j) is an eigenvalue at a position (i,j) in the global significant integration correlation response matrix of air conditioning parameters, M2(i,j) is an eigenvalue at a position (i,j) in the global significant integration distance response matrix of air conditioning parameters, v i and v j are the i th and j th eigenvalues in the global significant integration representation vector of air conditioning parameters respectively, V1 is a point addition vector between the global significant integration correlation response vector of air conditioning parameters and the global significant integration distance response vector of air conditioning parameters, V2 is a global significant integration logical bias vector of air conditioning parameters, R is a real number set, L is the length of the global significant integration representation vector of air conditioning parameters, and represents point multiplication by position, represents matrix multiplication, represents point addition by position, V is a global significant integration representation vector of air conditioning parameters, and V' is an optimized global significant integration representation vector of air conditioning parameters.

[0090] That is, in the technical solution of the present application, when each air conditioning parameter cluster center embedding code vector in the set of air conditioning parameter cluster center embedding code vectors respectively represents the class embedding code features of each cluster center in the decrypted clustering result, and when the feature distribution field domain is dynamically aggregated based on the mimetic field, the differences in the class embedding code features of each cluster center will cause a large area of simple repetition in the dynamic aggregation of the feature distribution field domain, affect the logical dependency of the field domain dynamic aggregation, and reduce the accuracy of the decoding result based on the clustering distribution.

[0091] Therefore, by taking the self-association matrix and the self-distance matrix of the air conditioning parameter global significant integration representation vector as the statistical-based no-reference distribution response framework of the air conditioning parameter global significant integration representation vector, feature distribution simple repetition is avoided based on retrieval-enhanced reverse response construction for the air conditioning parameter global significant integration representation vector, and surface combined mapping of the air conditioning parameter global significant integration representation vector is avoided by ensuring the retrieval-response context-based intrinsic mapping logic of the air conditioning parameter global significant integration representation vector, so as to realize logical dependent mapping of the air conditioning parameter global significant integration representation vector to a decoding target domain while maintaining intuition-based response correlation, thereby improving the accuracy of the optimal demand response strategy obtained by the air conditioning parameter global significant integration representation vector input based on the decoder-based optimal demand response strategy generator.

[0092] In summary, the cloud-edge encryption-based air conditioning load parameter trusted aggregation method based on the embodiments of the present application is illustrated, which uses a public cloud server as a cloud platform and an edge computing device as a user-side intelligent terminal based on a cloud-edge collaborative architecture, and uses a file transfer protocol-based data transmission method for cloud-edge communication. The cloud-edge collaborative architecture includes three cloud servers, cloud server 3 is responsible for distributing encryption keys to users and distributing decryption keys to cloud server 1, cloud server 2 is responsible for clustering user parameters, and cloud server 1 is responsible for decrypting the clustered parameters and solving the demand response load reduction optimization problem to determine the optimal strategy. In this way, it is ensured that each cloud server can only master partial information of all users, enhancing the security of data flow, and by encrypting the clustered user parameters on the cloud, the privacy of user data is protected while reducing the solution size of the load reduction optimization problem in demand response.

[0093] The basic principles of the application are described above in combination with specific embodiments, but it should be noted that the advantages, advantages, effects, etc. mentioned in the application are only examples and are not limiting, and these advantages, advantages, effects, etc. cannot be considered as the application must have. In addition, the specific details of the above embodiments are only for the purpose of example and for the purpose of understanding, and are not limited to the application which must use the above specific details to realize.

[0094] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in a certain embodiment can be referred to the relevant description of other embodiments. In the several embodiments provided by the present application, it should be understood that the disclosed method can be implemented in other ways. For example, the above-described system embodiments are merely illustrative. For example, the unit division is only a logical function division, and there can be another division manner in actual implementation. The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e. they can be located in one place or distributed on a plurality of network units. In actual implementation, some or all of the units can be selected according to the actual needs to achieve the purposes of the embodiments.

[0095] It is apparent that the present application is not limited to the foregoing exemplary embodiments, but can be carried out in other concrete forms without departing from the spirit or essential characteristics of the present application. Accordingly, the embodiments are to be considered in all respects as illustrative and not restrictive, the scope of the present application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims should not be construed as limiting the scope of the claims.

[0096] Further, it is clear that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. Multiple units referred to in system claims can also be implemented by one unit by means of software or hardware.

[0097] Finally, it should be noted that the above description is given for illustrative and descriptive purposes only. Furthermore, the above embodiments are merely used to illustrate the technical solutions of the present application but not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application.

Claims

1. A cloud-edge encryption based method for trustworthy aggregation of air conditioning load parameters, the method comprising: The method comprises the following steps: The third cloud server generates an encryption key and a decoding key, and stores the encryption key and the decoding key in a first FTP folder; The edge side acquires the encryption key from the first FTP folder, encrypts an air conditioner load parameter using the encryption key to obtain an encrypted air conditioner load parameter, and transmits the encrypted air conditioner load parameter to a second FTP folder, wherein the air conditioner load parameter comprises an air conditioner temperature preference coefficient and a most comfortable temperature setting; The second cloud server extracts the encrypted air conditioner load parameter from the second FTP folder and performs cluster analysis on the encrypted air conditioner load parameter to obtain a cluster result, and then writes the cluster result into a third FTP folder; The first cloud server extracts the decoding key from the first FTP folder and extracts the cluster result from the third FTP folder, and calculates an optimal demand response strategy based on the cluster result.

2. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 1, characterized in that, The first cloud server extracts the decoding key from the first FTP folder and extracts the cluster result from the third FTP folder, and calculates an optimal demand response strategy based on the cluster result, comprising: processing the cluster result using an MPEC or MILP method to obtain an optimal demand response strategy, wherein the optimal demand response strategy is a recommended air conditioner temperature setting. 3.The cloud-edge based encryption enabled air conditioning load parameter trustful aggregation method according to claim 1, characterized in that, The first cloud server extracts the decoding key from the first FTP folder and extracts the cluster result from the third FTP folder, and calculates an optimal demand response strategy based on the cluster result, comprising: decrypting the cluster result using the decoding key to obtain a decrypted cluster result; embedding and encoding each cluster center in the decrypted cluster result using a class embedding matrix to obtain a set of air conditioner parameter cluster center embedding and encoding vectors; performing feature dynamic aggregation on the set of air conditioner parameter cluster center embedding and encoding vectors to obtain an air conditioner parameter global significant integration representation vector; generating the optimal demand response strategy based on the air conditioner parameter global significant integration representation vector.

4. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 3, characterized in that, The feature dynamic aggregation on the set of air conditioner parameter cluster center embedding and encoding vectors to obtain an air conditioner parameter global significant integration representation vector comprises: determining an air conditioner parameter global cluster initial center vector based on a feature distribution field of the set of air conditioner parameter cluster center embedding and encoding vectors; performing significant modulation dynamic aggregation on the set of air conditioner parameter cluster center embedding and encoding vectors based on a spatial span of each air conditioner parameter cluster center embedding and encoding vector in the set of air conditioner parameter cluster center embedding and encoding vectors relative to the air conditioner parameter global cluster initial center vector to obtain the air conditioner parameter global significant integration representation vector.

5. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 4, characterized in that, The determination of the air conditioner parameter global cluster initial center vector based on the feature distribution field of the set of air conditioner parameter cluster center embedding and encoding vectors comprises: calculating a static energy factor of each air conditioner parameter cluster center embedding and encoding vector in the set of air conditioner parameter cluster center embedding and encoding vectors to obtain a set of air conditioner parameter static energy factors; The air conditioner parameter global significant integrated representation vector is obtained by performing significant modulation dynamic aggregation on the set of air conditioner parameter cluster center embedded encoding vectors based on the spatial span of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors relative to the air conditioner parameter global cluster initial center vector.

6. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 5, characterized in that, The set of air conditioner parameter static energy factors is obtained by calculating the static energy factor of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors, including: The kurtosis of the air conditioner parameter cluster center embedded encoding vector is calculated, and the kurtosis is input into a sigmoi d activation function to obtain the air conditioner parameter static energy factor.

7. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 6, characterized in that, The set of air conditioner parameter cluster center embedded encoding vectors is subjected to significant modulation dynamic aggregation based on the spatial span of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors relative to the air conditioner parameter global cluster initial center vector, to obtain the air conditioner parameter global significant integrated representation vector, including: The set of air conditioner parameter dynamic aggregation energy factors is obtained by calculating the dynamic aggregation energy factor of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors based on the spatial span between each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors and the air conditioner parameter global cluster initial center vector, and the set of air conditioner parameter static energy factors. The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit.

8. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 7, characterized in that, The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The set of air conditioner parameter dynamic aggregation energy factors is obtained by calculating the dynamic aggregation energy factor of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors, including:

9. The cloud-edge encryption based air conditioning load parameter trustful aggregation method according to claim 8, characterized in that, The set of air conditioner parameter dynamic aggregation energy factors is obtained by calculating the dynamic aggregation energy factor of each air conditioner parameter cluster center embedded encoding vector in the set of air conditioner parameter cluster center embedded encoding vectors, including: The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The set of air conditioner parameter dynamic aggregation weight factors is obtained by inputting the set of air conditioner parameter dynamic aggregation energy factors into a gating mask unit. The optimal demand response strategy is generated based on the air conditioner parameter global significant integrated representation vector, including: The optimal demand response strategy is generated based on the air conditioner parameter global significant integrated representation vector, including: The optimal demand response strategy is generated based on the air conditioner parameter global significant integrated representation vector, including:

Images