Automatic construction method and system of energy consumption model based on AI algorithm

Abstract

The present invention provides an automatic construction method and system for energy consumption models based on AI algorithms, the system includes an edge computing node and a central server; the edge computing node includes a local data acquisition unit and a first encryption channel component; the central server includes a remote data receiving unit and a remote model training unit; an encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through an encrypted communication link; the energy consumption model is trained in combination with the AI ​​algorithm to obtain an energy consumption prediction model; the energy consumption prediction model is automatically built to the edge computing node through an encrypted communication link. In the present invention, the security and privacy of the data are guaranteed; the training and construction of the energy consumption model are completed by the central server, which overcomes the defect that the local device cannot complete high-intensity model training.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for automatically building an energy consumption model based on an AI algorithm. Background Art

[0002] In industrial, commercial and residential fields, we can build adaptive energy consumption models (energy consumption prediction models) to predict energy consumption in various scenarios through real-time monitoring and analysis of energy consumption data, which will help optimize energy distribution, reduce waste and improve energy efficiency.

[0003] However, the current energy consumption model faces the following problems when building:

[0004] In energy management systems, since energy consumption data involves a large amount of sensitive information (such as user electricity usage habits, enterprise production data, etc.), the security and privacy of the above data are at risk of being leaked during transmission;

[0005] The computing power of local energy consumption detection equipment in various scenarios is limited, making it difficult to complete high-intensity energy consumption model training and construction. Summary of the invention

[0006] The main purpose of the present invention is to provide a method and system for automatically building an energy consumption model based on an AI algorithm, aiming to overcome the risk of easy leakage of energy consumption data and the defect that the performance of local equipment is limited and cannot complete high-intensity model training.

[0007] To achieve the above-mentioned object, the present invention provides an automatic construction system of energy consumption model based on AI algorithm, comprising an edge computing node and a central server; wherein the edge computing node comprises a local data acquisition unit and a first encryption channel component; the central server comprises a remote data receiving unit and a remote model training unit;

[0008] An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data;

[0009] The first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through the encrypted communication link;

[0010] The remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; and automatically builds the energy consumption prediction model to the edge computing node through the encrypted communication link.

[0011] Furthermore, the edge computing node also includes a retraining module;

[0012] The edge computing node receives the energy consumption prediction model through the retraining module, the local data acquisition unit collects energy consumption data in the actual application scenario of the edge computing node, and optimizes and trains the energy consumption prediction model based on the energy consumption data of the actual application scenario to obtain an optimized energy consumption prediction model.

[0013] Furthermore, the step of automatically building the energy consumption prediction model to the edge computing node through the encrypted communication link includes:

[0014] Obtain model parameters of the energy consumption prediction model, package the model parameters, and automatically transmit them to the edge computing node through the encrypted communication link; so that the edge computing node updates the model parameters of its pre-stored initial energy consumption prediction model based on the model parameters, thereby constructing the energy consumption prediction model.

[0015] Further, the central server also includes a second encryption channel component;

[0016] The second encryption channel component is used to encrypt the model parameters and then securely transmit them to each edge computing node through the encrypted communication link.

[0017] Furthermore, an encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit, including:

[0018] The local data acquisition unit and the remote data receiving unit respectively generate a public key and a private key that conform to national encryption standards;

[0019] The local data acquisition unit exchanges public keys with the remote data receiving unit, and uses a key exchange protocol of national encryption standards to negotiate a session key based on the two public keys;

[0020] Run the transport layer security protocol based on the national encryption standard, use the negotiated session key to configure the encryption parameters in the secure socket layer handshake process, and complete the configuration of the encrypted communication link; the session key is used as data encryption and data decryption in the encrypted communication link.

[0021] Furthermore, the remote model training unit is an AI chip based on the Ampere architecture;

[0022] The first encryption channel component is an encryption card.

[0023] Further, the key exchange protocol includes the SM2 elliptic curve cryptographic algorithm protocol;

[0024] The transport layer security protocol includes TLS or DTLS protocol based on national encryption standards.

[0025] Furthermore, the edge computing node and the central server pre-negotiate a building request code; when the edge computing node needs to build an energy consumption model, the edge computing node sends the building request code to the central server for verification; after the verification is passed, the building of the energy consumption model begins.

[0026] Furthermore, the edge computing node and the central server pre-negotiate to establish a request code, including:

[0027] Obtaining the node code and identification address of the edge computing node, and obtaining the service character matrix pre-stored in the central server;

[0028] Updating the service character matrix based on the identification address to obtain an updated character matrix;

[0029] The construction request code is generated based on the node code and the update character matrix.

[0030] The present invention also provides a method for automatically building an energy consumption model based on an AI algorithm, which is applied to any of the above-mentioned systems for automatically building an energy consumption model based on an AI algorithm; the method at least comprises:

[0031] An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data;

[0032] The first encryption channel component encrypts the energy consumption training data and then securely transmits it to the remote model training unit through the encrypted communication link;

[0033] The remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; and automatically builds the energy consumption prediction model to the edge computing node through the encrypted communication link.

[0034] The present invention provides an automatic construction method and system for energy consumption models based on AI algorithms, the system comprising an edge computing node and a central server; wherein the edge computing node comprises a local data acquisition unit and a first encryption channel component; the central server comprises a remote data receiving unit and a remote model training unit; an encrypted communication link based on national secret standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data; the first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through the encrypted communication link; the remote model training unit receives the energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; the energy consumption prediction model is automatically constructed to the edge computing node through the encrypted communication link. In the present invention, an encrypted communication link based on national secret standards is established between the edge computing node and the central server, and the energy consumption training data is encrypted and then securely transmitted through the encrypted communication link, thereby ensuring the security and privacy of the energy consumption data; at the same time, the training and construction of the energy consumption model is completed through the central server, overcoming the defect that the local device performance is limited and high-intensity model training cannot be completed. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a structural block diagram of a system for automatically building an energy consumption model based on an AI algorithm in one embodiment of the present invention;

[0036] Figure 2 It is a structural block diagram of a system for automatically building an energy consumption model based on an AI algorithm in another embodiment of the present invention;

[0037] Figure 3 It is a structural block diagram of an energy consumption model automatic construction system based on AI algorithm in another embodiment of the present invention;

[0038] Figure 4 It is a structural block diagram of an energy consumption model automatic construction system based on AI algorithm in another embodiment of the present invention;

[0039] Figure 5 It is a structural block diagram of an energy consumption model automatic construction system based on AI algorithm in another embodiment of the present invention;

[0040] Figure 6 It is a schematic diagram of the steps of a method for automatically building an energy consumption model based on an AI algorithm in one embodiment of the present invention.

[0041] The implementation, functional features and advantages of the present invention will be further described in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0043] Reference Figure 1 In one embodiment of the present invention, an automatic construction system of an energy consumption model based on an AI algorithm is provided, comprising an edge computing node and a central server; wherein the edge computing node comprises a local data acquisition unit and a first encryption channel component; the central server comprises a remote data receiving unit and a remote model training unit;

[0044] An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data;

[0045] The first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through the encrypted communication link;

[0046] The remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; and automatically builds the energy consumption prediction model to the edge computing node through the encrypted communication link.

[0047] In this embodiment, an automatic energy consumption model building system based on AI algorithm is proposed, which mainly includes two parts: edge computing nodes and central servers. Among them, edge computing nodes refer to data sources (such as sensors, devices) and end users that are closer to the device. In other words, the above-mentioned edge computing nodes are local devices set in various actual application scenarios (such as local energy consumption management devices, safety detection equipment, etc.); the above-mentioned central server is a remote cloud server.

[0048] The above edge computing nodes include:

[0049] The local data acquisition unit is used to collect energy consumption related data based on sensors on energy devices and energy devices in the scene. The above data includes but is not limited to meter readings, temperature, humidity and other environmental parameters.

[0050] The first encryption channel component is used to encrypt the collected energy consumption data to ensure that the data will not be accessed by unauthorized persons during transmission. The first encryption channel component is usually an encryption card.

[0051] The above-mentioned central server includes:

[0052] The remote data receiving unit is used to receive the encrypted energy consumption training data from the edge computing node, which is usually a communication module.

[0053] The remote model training unit uses the received data to train the energy consumption model, and uses an AI algorithm (such as a machine learning algorithm) to optimize the model performance to obtain the above-mentioned energy consumption prediction model, and then automatically builds the energy consumption prediction model to the edge computing node through the encrypted communication link.

[0054] In this embodiment, an encrypted communication link is first established based on the national encryption standard to ensure the security of data transmission. The national encryption standard refers to a series of cryptographic algorithms and technical standards formulated by the State Cryptography Administration, such as SM2, SM3, and SM4. The first encryption channel component on the edge computing node uses a specific encryption algorithm (such as a symmetric encryption algorithm) to encrypt the energy consumption training data, and then sends it to the central server through an encrypted communication link. After the remote data receiving unit of the central server receives the encrypted data, it uses the corresponding decryption algorithm to restore the original data.

[0055] Specifically, in a specific embodiment, the local data acquisition unit of the edge computing node regularly collects energy consumption data from field devices. The first encryption channel component encrypts the data using the encryption algorithm in the national encryption standard. The encrypted data is sent to the central server through an encrypted communication link. After the central server receives the data, it uses the corresponding decryption algorithm to restore the original data. The remote model training unit uses the decrypted data to train the energy consumption prediction model. At the same time, the model performance is evaluated and adjusted if necessary. The trained model is then re-encrypted and returned to the edge computing node through an encrypted communication link. Each edge computing node uses the deployed model to perform real-time energy consumption prediction.

[0056] Specifically, the above solution has the following advantages:

[0057] Security: Use national encryption standards to ensure secure data transmission.

[0058] Real-time: Edge computing nodes can perform fast data processing and prediction locally.

[0059] Efficiency: The central server is responsible for training complex models, reducing the computational burden on edge nodes.

[0060] Flexibility: Models can be updated or optimized as needed without changing hardware configuration.

[0061] To summarize, in this embodiment, an encrypted communication link based on national encryption standards is established between the edge computing node and the central server, and the energy consumption training data is encrypted and then securely transmitted through the encrypted communication link, thereby ensuring the security and privacy of the energy consumption data; at the same time, the training and construction of the energy consumption model are completed through the central server, thereby overcoming the defect that the local device performance is limited and cannot complete high-intensity model training.

[0062] Reference Figure 2 ,In one embodiment, the edge computing node further includes a retraining module;

[0063] The edge computing node receives the energy consumption prediction model through the retraining module, the local data acquisition unit collects energy consumption data in the actual application scenario of the edge computing node, and optimizes and trains the energy consumption prediction model based on the energy consumption data of the actual application scenario to obtain an optimized energy consumption prediction model.

[0064] In this embodiment, the above-mentioned edge computing node also includes a retraining module; since the energy consumption data in each actual application scenario is different, for each different application scenario, the edge computing node in the actual application scenario can collect the energy consumption data in the actual application scenario where the edge computing node is located through a local data acquisition unit, and optimize the energy consumption prediction model based on the energy consumption data of the actual application scenario through the retraining module to obtain an optimized energy consumption prediction model, so that the energy consumption prediction model can adapt to the current application scenario and improve the prediction accuracy of subsequent energy consumption.

[0065] Reference Figure 3 In one embodiment, the step of automatically building the energy consumption prediction model to the edge computing node through the encrypted communication link includes:

[0066] Obtain model parameters of the energy consumption prediction model, package the model parameters, and automatically transmit them to the edge computing node through the encrypted communication link; so that the edge computing node updates the model parameters of its pre-stored initial energy consumption prediction model based on the model parameters, thereby constructing the energy consumption prediction model.

[0067] In this embodiment, first, it is necessary to extract the model parameters from the trained energy consumption prediction model. The above model parameters are obtained after training and are used to define the key values ​​of the model structure and behavior. In order to facilitate transmission, the model parameters will be packaged into a data packet. Usually, a serialization process is involved, that is, converting the model parameters into a data format that can be transmitted.

[0068] Since it involves the transmission of sensitive information, an encrypted communication link is used to protect the security of the data. This can be achieved through a variety of encryption protocols, such as TLS (Transport Layer Security), to ensure that the data will not be accessed or tampered with by unauthorized persons during transmission. Through the above encrypted channel, the model parameters are automatically sent to the designated edge computing node. This process is usually automated and does not require human intervention. After receiving the model parameters, the edge computing node uses the above model parameters to update its local initial energy consumption prediction model. This means that the model on the node will become more accurate and can better predict the actual energy consumption.

[0069] After the parameters are updated, the edge computing node has the latest energy consumption prediction model, which can be used to monitor and predict the energy consumption of edge devices in real time, thereby optimizing energy management and utilization efficiency.

[0070] In this embodiment, through the above technical solution, the energy consumption prediction model is remotely, securely, and automatically deployed and updated, which improves the flexibility and security of the system and reduces maintenance costs. This is very important for edge computing applications in scenarios such as the Internet of Things (IoT) and smart grids.

[0071] Reference Figure 4 , in one embodiment, the central server further comprises a second encryption channel component;

[0072] The second encryption channel component is used to encrypt the model parameters and then securely transmit them to each edge computing node through the encrypted communication link.

[0073] In this embodiment, a component dedicated to encryption is integrated in the central server, which is a second encryption channel component, which can be an encryption card. The main task of the above component is to encrypt the model parameters before transmission.

[0074] After the model parameters are packaged, the second encryption channel component will encrypt the above model parameters. The encryption process uses a pre-defined encryption algorithm and key to ensure that even if the data is intercepted during transmission, it cannot be easily cracked. The encrypted model parameters are securely transmitted to each edge computing node through a pre-set encrypted communication link. The encrypted communication link here uses protocols such as SSL / TLS that comply with national encryption standards to ensure the security and integrity of data during transmission.

[0075] The second encryption channel component is part of the central server and is mainly responsible for encrypting model parameters. It can use symmetric encryption or asymmetric encryption methods, depending on the design requirements and security requirements of the system. For example, the AES (Advanced Encryption Standard) symmetric encryption algorithm or the RSA (Rivest-Shamir-Adleman) asymmetric encryption algorithm can be used.

[0076] In this embodiment, the above technical solution ensures the safe transmission of model parameters and also realizes the automation of model update. The above solution is particularly suitable for scenarios where model parameters need to be updated frequently, such as real-time monitoring and prediction of energy consumption.

[0077] In one embodiment, an encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit, including:

[0078] The local data acquisition unit and the remote data receiving unit respectively generate a public key and a private key that conform to national encryption standards;

[0079] The local data acquisition unit exchanges public keys with the remote data receiving unit, and uses a key exchange protocol of national encryption standards to negotiate a session key based on the two public keys;

[0080] Run the transport layer security protocol based on the national encryption standard, use the negotiated session key to configure the encryption parameters in the secure socket layer handshake process, and complete the configuration of the encrypted communication link; the session key is used as data encryption and data decryption in the encrypted communication link.

[0081] In this embodiment, the above technical solution proposes a process for establishing an encrypted communication link based on national encryption standards, which aims to protect the confidentiality and integrity of data during transmission and ensure that only legitimate recipients can decrypt and read the data.

[0082] First, the local data acquisition unit and the remote data receiving unit each independently generate a pair of public and private keys that meet national encryption standards. Usually, this step uses an asymmetric encryption algorithm, such as the SM2 elliptic curve cryptography algorithm. The public key can be made public, while the private key needs to be kept confidential.

[0083] The local data acquisition unit and the remote data receiving unit exchange public keys with each other in a secure manner. The above secure manner means that even if the public key is intercepted by a third party, it will not cause security risks, because the public key itself is used to encrypt rather than decrypt data. Then, the two parties use the key exchange protocol in the national secret standard (such as the key exchange protocol based on SM2) to negotiate a shared session key based on the two public keys. The session key will be used in the subsequent data encryption and decryption process to improve efficiency and reduce the complexity of key management.

[0084] Both parties run the Transport Layer Security (TLS) protocol based on the national secret standard, such as the national secret TLS version. This protocol defines a set of rules to ensure the security of data during transmission. During the TLS handshake process, both parties use the negotiated session key to configure encryption parameters, including but not limited to encryption algorithms, message authentication code (MAC) algorithms, etc. After the handshake is completed, the encrypted communication link is configured.

[0085] When the encrypted communication link is configured, all data transmission will be encrypted and decrypted using the session key. This means that the sender uses the session key to encrypt the data, and the receiver uses the same session key to decrypt the data. The encrypted communication link ensures the confidentiality, integrity, and non-repudiation of the data during transmission.

[0086] In summary, the above technical solution generates public and private keys by using the asymmetric encryption algorithm in the national secret standard, and negotiates the session key using the key exchange protocol in the national secret standard, and then configures the transport layer security protocol based on the national secret standard, and finally establishes a secure encrypted communication link. The above link can effectively protect data from eavesdropping and tampering, ensuring the confidentiality and integrity of the data.

[0087] Reference Figure 5 In one embodiment, the remote model training unit is an AI chip based on the Ampere architecture;

[0088] The first encryption channel component is an encryption card. The second encryption channel component is an encryption card. The retraining module is a local AI chip, and its performance may not be as good as that of the AI ​​chip on the central server. This is because its training task is lighter and does not require strong computing power support.

[0089] In this embodiment, the above-mentioned remote model training unit is a GPU using the NVIDIA Ampere architecture, which is used for training and reasoning of machine learning models and is widely used in the field of deep learning due to its efficient parallel processing capabilities.

[0090] The above encryption card is a hardware device specially used to perform encryption operations. Encryption cards usually contain a special encryption processor that can accelerate the encryption and decryption process while providing hardware-level security. Encryption cards can be used to implement functions such as data encryption, decryption, and key management.

[0091] In one embodiment, the key exchange protocol includes the SM2 elliptic curve cryptographic algorithm protocol;

[0092] The transport layer security protocol includes TLS or DTLS protocol based on national encryption standards.

[0093] In this embodiment, the key exchange protocol includes the SM2 elliptic curve cryptographic algorithm protocol, which is a public key encryption algorithm based on elliptic curves developed by the State Cryptography Administration. The SM2 algorithm is applicable to the key exchange protocol, allowing both parties to negotiate a shared session key for subsequent data encryption and decryption. The advantage of elliptic curve cryptography is that it can provide higher security even with a smaller key length.

[0094] TLS or DTLS protocol based on national secret standards: TLS (Transport Layer Security) is a security protocol used to encrypt Internet communications, while DTLS (Datagram Transport Layer Security) is a variant of TLS used to transmit UDP-based applications. TLS or DTLS protocol based on national secret standards refers to a version that uses the encryption algorithm and key exchange protocol in the national secret standards to enhance the security of data transmission.

[0095] In one embodiment, the edge computing node and the central server pre-negotiate a building request code; when the edge computing node needs to build an energy consumption model, the edge computing node sends the building request code to the central server for verification; after the verification is passed, the building of the energy consumption model begins.

[0096] In this embodiment, a method for secure communication between an edge computing node and a central server is proposed for identity authentication and authorization before building an energy consumption model on the edge side. Specifically, the edge computing node and the central server need to negotiate a request code in advance. When the edge computing node needs to build an energy consumption model, it sends the request code to the central server for verification. If the verification is successful, the central server will allow the edge computing node to start building the energy consumption model.

[0097] Specifically, a request code needs to be negotiated in advance between the edge computing node and the central server. The request code can be a string of numbers, letters, or a mixed string, or encrypted information. The request code is used to identify the identity of the edge computing node and the validity of the request. When the edge computing node needs to build an energy consumption model, it sends the pre-negotiated request code to the central server. After receiving the request code, the central server will verify whether the request code is consistent with the pre-negotiated one. If the verification is successful, it means that the edge computing node has the right to request the construction of the energy consumption model.

[0098] In summary, this technical solution establishes a secure and effective communication mechanism between the edge computing node and the central server by pre-negotiating the request code to ensure the correct construction of the energy consumption model.

[0099] In one embodiment, the edge computing node and the central server pre-negotiate a request code, including:

[0100] Obtaining the node code and identification address of the edge computing node, and obtaining the service character matrix pre-stored in the central server;

[0101] Updating the service character matrix based on the identification address to obtain an updated character matrix;

[0102] The construction request code is generated based on the node code and the update character matrix.

[0103] In this embodiment, this scheme proposes a mechanism for generating a request code, which is used to generate a request code for building an energy consumption model between an edge computing node and a central server. The generation of the request code is based on the specific information and service character matrix of the edge computing node and is completed through a series of steps. This ensures that only authorized edge computing nodes can initiate a building request, thereby improving the security and credibility of the system.

[0104] First, obtain the node code and identification address of the edge computing node; each edge computing node has a unique code to distinguish different nodes. The identification address refers to the unique identifier of the edge computing node in the network, such as IP address, MAC address or other forms of network identifiers.

[0105] Get the service character matrix pre-stored in the central server. The service character matrix is ​​a two-dimensional array pre-stored in the central server, which contains a series of characters. The above matrix can be used to generate the basic data for the construction request code.

[0106] Then, the service character matrix is ​​updated based on the identification address, and the service character matrix in the central server is updated according to the identification address of the edge computing node. The specific methods of updating include replacing certain characters, adding new characters, or rearranging the characters in the matrix according to a certain rule.

[0107] Finally, the setup request code is generated based on the node code and the updated character matrix, and the final setup request code is generated by combining the node code of the edge computing node and the updated character matrix. It can use a variety of algorithms, such as hash functions, encryption algorithms or other encoding technologies to ensure the uniqueness and security of the request code.

[0108] In this embodiment, by using the node code and identification address to update the service character matrix, and generating a build request code based on the above information, the uniqueness of the request code can be ensured, and unauthorized edge computing nodes can be prevented from initiating build requests. The above technical solution is suitable for scenarios where energy consumption models need to be built on edge computing nodes, especially in environments that require high security. For example, in the fields of industrial automation, smart city management, Internet of Things (IoT) applications, etc., the above technical solution can effectively protect the security of the system and the integrity of the data.

[0109] In summary, this technical solution ensures the secure generation of the build request code through specific node information and service character matrix, thereby ensuring the secure communication between the edge computing node and the central server and the effective construction of the energy consumption model.

[0110] In one embodiment, the updating of the service character matrix based on the identification address to obtain an updated character matrix includes:

[0111] The total number of columns of the service character matrix is ​​obtained as n;

[0112] Extract all characters in the identification address and combine them in sequence to obtain an identification character combination;

[0113] Create a blank matrix with n rows and m columns; add the characters in the identification character combination to the blank matrix one by one in order to obtain an identification matrix; wherein n*m≥the number of characters in the identification address>n*(m-1);

[0114] The service character matrix is ​​multiplied by the identification matrix, and the obtained matrix is ​​used as the update character matrix.

[0115] In this embodiment, the service character matrix is ​​a predefined matrix including a series of characters. It can be pre-stored in the central server. It is understandable that there is usually only one character at each position in the above service character matrix, which can be a number or an English character. The service character matrix is ​​a matrix with multiple rows and columns, and the total number of columns is n.

[0116] The identification address is a unique identifier of the edge computing node in the network, such as an IP address, a MAC address, or other forms of network identifiers. The identification character combination is to extract all characters from the identification address and sequentially combine them into a new string or character sequence.

[0117] In order to deform and update the above service character matrix to change its adaptability, a blank matrix with n rows and m columns can be created, where n is the total number of columns in the service character matrix. Then, the characters in the identification character combination are added to the blank matrix one by one in sequence until all characters are filled in the matrix. It should be noted that in this process, it is necessary to satisfy n* m ≥ the number of characters in the identification address>n * (m-1), that is, the number of characters in the identification address must be between n * (m-1) and n*m. This means that the number of characters in the identification address can just fill the entire blank matrix or at least fill n*(m-1)+1 positions, that is, the last row of the matrix cannot be completely empty, and there is at least one character.

[0118] Finally, the service character matrix is ​​multiplied with the identification matrix, and the resultant matrix is ​​used as the update character matrix. Here, it should be noted that in order for the two matrices to be multiplied, the number of columns of the first matrix must be the same as the number of rows of the second matrix. In this embodiment, the number of columns n of the service character matrix is ​​the same as the number of rows n of the identification matrix.

[0119] For example: Assume that the service character matrix is ​​a 3×3 matrix and the identification address is an IP address "192.168.1.1".

[0120] Get the total number of columns of the service character matrix n = 3.

[0121] Extract all characters in the identification address to obtain the identification character combination: 19216811.

[0122] Create a 3×3 blank matrix and fill it with the identification character combination to get the identification matrix.

[0123] In this example, m must be equal to 3 because the number of characters in the identification address is 8, satisfying 3×3 ≥ 8>3×(3-1).

[0124] Finally, the service character matrix is ​​multiplied by the identification matrix to obtain the updated character matrix.

[0125] In summary, the above technical solution provides a method for generating an update character matrix based on the edge computing node identification address, which provides a basis for the subsequent generation of the request code. The above method ensures the uniqueness and security of the request code, which is essential for the secure communication between the edge computing node and the central server.

[0126] In one embodiment, the node code is a digital code; based on the node code and the update character matrix, generating the construction request code includes:

[0127] Get a preset number;

[0128] The number corresponding to the node code and the preset number are used as the major axis and minor axis of an ellipse to construct an ellipse;

[0129] Adding the ellipse and the updated character matrix to the same layer, with the centers of the ellipse and the updated character matrix overlapping with the center of the layer;

[0130] Scaling / stretching the ellipse so that two ends of the major axis of the ellipse are tangent to two sides of the updated character matrix;

[0131] The character formed by the intersection of the ellipse and the updated character matrix is ​​obtained as a target character; and the target characters are combined in sequence to obtain the construction request code.

[0132] In this embodiment, this scheme describes the specific process of how to generate a build request code through a node code and an update character matrix.

[0133] First, a preset number is obtained. The preset number is a predefined value and can be any integer or floating point number. For example, the preset number can be 8.

[0134] The node code is a numeric code. The node code should be a unique number to distinguish different edge computing nodes. For example, the node code is 12.

[0135] Use the node code and the preset number as the major axis and minor axis length of the ellipse to construct an ellipse. It is understandable that when determining the minor axis and the major axis, the relationship between the number corresponding to the node code and the preset number should be determined first, and the larger number is used as the major axis length; if the numbers are equal, a circle (equivalent to a special ellipse) is generated. This is mathematical common sense and will not be elaborated here.

[0136] In order to overlay the ellipse and the updated character matrix, you need to create a graphic layer and place the ellipse and the updated character matrix on the layer. At the same time, make sure that the center of the ellipse and the updated character matrix coincides with the center of the layer. Then adjust the size of the ellipse so that the two ends of its long axis are tangent to the two sides of the updated character matrix.

[0137] At this time, find the character at the intersection of the ellipse and the updated character matrix from the layer as the target character, and combine the target characters in sequence to form a build request code.

[0138] For example, assuming that the node code is 5 and the preset number is 3, the character matrix is ​​updated to be a 5×5 matrix, which includes some characters from A to Z and some special symbols (such as #, &).

[0139] Node code = 5, as the length of the major axis of the ellipse; preset number = 3, as the length of the minor axis of the ellipse. At this time, you can construct an ellipse with a major axis of 5 and a minor axis of 3.

[0140] Place the ellipse and the updated character matrix on the same layer, making sure their centers coincide with the center of the layer. Adjust the size of the ellipse so that its long axis is tangent to both sides of the updated character matrix.

[0141] The target characters at the intersection of the ellipse and the update character matrix can be A, D, E, M, N, etc. At this time, the target characters are combined in sequence, such as ADENM, which is the construction request code.

[0142] In summary, the above technical solution provides a method for generating a request code based on a node code and an updated character matrix, which ensures the uniqueness and security of the request code, and is essential for secure communication between edge computing nodes and central servers.

[0143] Reference Figure 6 In one embodiment of the present invention, a method for automatically building an energy consumption model based on an AI algorithm is provided, which is applied to the system for automatically building an energy consumption model based on an AI algorithm described in any of the above embodiments; the method at least includes:

[0144] Step S1, establishing an encrypted communication link based on national encryption standards between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data;

[0145] Step S2, the first encryption channel component encrypts the energy consumption training data and then securely transmits the data to the remote model training unit through the encrypted communication link;

[0146] Step S3, the remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; and automatically builds the energy consumption prediction model to the edge computing node through the encrypted communication link.

[0147] In this embodiment, for the specific implementation of the above method embodiment, please refer to the above system embodiment, which will not be repeated here.

[0148] In summary, the method and system for automatically building an energy consumption model based on an AI algorithm provided in an embodiment of the present invention include an edge computing node and a central server; wherein the edge computing node includes a local data acquisition unit and a first encryption channel component; the central server includes a remote data receiving unit and a remote model training unit; an encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data; the first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through the encrypted communication link; the remote model training unit receives the energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; the energy consumption prediction model is automatically built to the edge computing node through the encrypted communication link. In the present invention, an encrypted communication link based on national encryption standards is established between the edge computing node and the central server, and the energy consumption training data is encrypted and then securely transmitted through the encrypted communication link, thereby ensuring the security and privacy of the energy consumption data; at the same time, the training and construction of the energy consumption model are completed through the central server, thereby overcoming the defect that the local equipment performance is limited and cannot complete high-intensity model training.

[0149] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0150] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0151] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An automatic energy consumption model building system based on AI algorithm, characterized in that: It includes an edge computing node and a central server; wherein the edge computing node includes a local data acquisition unit and a first encryption channel component; the central server includes a remote data receiving unit and a remote model training unit; An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data; The first encryption channel component is used to encrypt the energy consumption training data and then securely transmit it to the remote model training unit through the encrypted communication link; The remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; the energy consumption prediction model is automatically built to the edge computing node through the encrypted communication link; The edge computing node and the central server pre-negotiate a request code for building; when the edge computing node needs to build an energy consumption model, the edge computing node sends the request code for building to the central server for verification; after the verification is passed, the building of the energy consumption model begins; The edge computing node and the central server pre-negotiate to establish a request code, including: Obtaining the node code and identification address of the edge computing node, and obtaining the service character matrix pre-stored in the central server; Updating the service character matrix based on the identification address to obtain an updated character matrix; Based on the node code and the update character matrix, generate the build request code; The node code is a digital code; based on the node code and the update character matrix, the construction request code is generated, including: Get a preset number; The number corresponding to the node code and the preset number are used as the major axis and minor axis of an ellipse to construct an ellipse; Adding the ellipse and the updated character matrix to the same layer, with the centers of the ellipse and the updated character matrix overlapping with the center of the layer; Scaling / stretching the ellipse so that two ends of the major axis of the ellipse are tangent to two sides of the updated character matrix; The character formed by the intersection of the ellipse and the updated character matrix is ​​obtained as a target character; and the target characters are combined in sequence to obtain the construction request code.

2. The automatic construction system of energy consumption model based on AI algorithm according to claim 1 is characterized in that: The edge computing node also includes a retraining module; The edge computing node receives the energy consumption prediction model through the retraining module, the local data acquisition unit collects energy consumption data in the actual application scenario of the edge computing node, and optimizes and trains the energy consumption prediction model based on the energy consumption data of the actual application scenario to obtain an optimized energy consumption prediction model.

3. The automatic construction system of energy consumption model based on AI algorithm according to claim 1 is characterized in that: The step of automatically building the energy consumption prediction model to the edge computing node through the encrypted communication link includes: Obtain model parameters of the energy consumption prediction model, package the model parameters, and automatically transmit them to the edge computing node through the encrypted communication link; so that the edge computing node updates the model parameters of its pre-stored initial energy consumption prediction model based on the model parameters, thereby constructing the energy consumption prediction model.

4. The automatic construction system of energy consumption model based on AI algorithm according to claim 3 is characterized in that: The central server also includes a second encryption channel component; The second encryption channel component is used to encrypt the model parameters and then securely transmit them to each edge computing node through the encrypted communication link.

5. The automatic construction system of energy consumption model based on AI algorithm according to claim 1 is characterized in that: An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit, including: The local data acquisition unit and the remote data receiving unit respectively generate a public key and a private key that conform to national encryption standards; The local data acquisition unit exchanges public keys with the remote data receiving unit, and uses a key exchange protocol of national encryption standards to negotiate a session key based on the two public keys; Run the transport layer security protocol based on the national encryption standard, use the negotiated session key to configure the encryption parameters in the secure socket layer handshake process, and complete the configuration of the encrypted communication link; the session key is used as data encryption and data decryption in the encrypted communication link.

6. The automatic construction system of energy consumption model based on AI algorithm according to claim 1 is characterized in that: The remote model training unit is an AI chip based on the Ampere architecture; The first encryption channel component is an encryption card.

7. The automatic construction system of energy consumption model based on AI algorithm according to claim 5 is characterized in that: The key exchange protocol includes the SM2 elliptic curve cryptographic algorithm protocol; The transport layer security protocol includes TLS or DTLS protocol based on national encryption standards.

8. A method for automatically building an energy consumption model based on an AI algorithm, characterized in that: Applicable to the automatic construction system of energy consumption model based on AI algorithm according to any one of claims 1 to 7; the method at least comprises: An encrypted communication link based on national encryption standards is established between the local data acquisition unit and the remote data receiving unit; the local data acquisition unit is used to collect energy consumption training data; The first encryption channel component encrypts the energy consumption training data and then securely transmits it to the remote model training unit through the encrypted communication link; The remote model training unit receives energy consumption training data through the encrypted communication link, and trains the energy consumption model in combination with the AI ​​algorithm to obtain an energy consumption prediction model; the energy consumption prediction model is automatically built to the edge computing node through the encrypted communication link; The edge computing node and the central server pre-negotiate a request code for building; when the edge computing node needs to build an energy consumption model, the edge computing node sends the request code for building to the central server for verification; after the verification is passed, the building of the energy consumption model begins; The edge computing node and the central server pre-negotiate to establish a request code, including: Obtaining the node code and identification address of the edge computing node, and obtaining the service character matrix pre-stored in the central server; Updating the service character matrix based on the identification address to obtain an updated character matrix; Based on the node code and the update character matrix, generate the build request code; The node code is a digital code; based on the node code and the update character matrix, the construction request code is generated, including: Get a preset number; The number corresponding to the node code and the preset number are used as the major axis and minor axis of an ellipse to construct an ellipse; Adding the ellipse and the updated character matrix to the same layer, with the centers of the ellipse and the updated character matrix overlapping with the center of the layer; Scaling / stretching the ellipse so that two ends of the major axis of the ellipse are tangent to two sides of the updated character matrix; The character formed by the intersection of the ellipse and the updated character matrix is ​​obtained as a target character; and the target characters are combined in sequence to obtain the construction request code.

Images