Photovoltaic system load prediction method and system based on trusted computing

Abstract

The invention discloses a photovoltaic system load prediction method and system based on trusted computing, and relates to the technical field of photovoltaic power generation, and the method comprises the steps: carrying out the trusted authentication of all users participating in federated learning modeling based on a trusted computing module, and carrying out the information interaction when a participant is verified to be not tampered and trusted; the participant constructs a local load prediction model based on the interaction data, pre-processes the data, and initializes the local load prediction model to perform federal calculation cooperative training; and on the basis of a federated learning mode, aggregating the local prediction models of the plurality of participating users to form a global load prediction model, and predicting the load of the photovoltaic system on the basis of the global load prediction model. According to the method, the credibility of users participating in modeling and the safety of the whole federal learning process environment are guaranteed, and consistent lossless aggregation of the model performance obtained by multi-party calculation and the centralized training model performance is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a photovoltaic system load prediction method and system based on trusted computing. Background Art

[0002] Photovoltaic power generation system is a power generation system that uses the photovoltaic effect of semiconductor materials to convert solar radiation energy into electrical energy. The power generation process does not pollute the environment or damage the ecology. In recent years, photovoltaics have developed rapidly in my country. In 2022, my country's new photovoltaic installed capacity reached 87GW. The promotion of photovoltaic power generation systems has responded well to the country's relevant policies on reducing carbon emissions and reduced our dependence on conventional power generation methods. However, the power supply of photovoltaic power generation is limited, so photovoltaic power generation still has great limitations. If the power consumption is not reasonably allocated, it will cause further waste of photovoltaic power generation resources. Therefore, the photovoltaic power generation system should change the power supply strategy to the "personalization" direction of regional users, and use power user data to analyze user power consumption behavior so as to better dispatch power. However, since the demand for power load is constantly changing and difficult to predict, this greatly increases the difficulty of controlling the power grid, and the power system urgently needs a good solution. With its excellent security and reliability, federated learning is very suitable for the needs of regional user behavior analysis and is a very effective predictive model modeling method. However, there are huge information security risks hidden behind these calculations, which need to be prevented from the aspects of secure access to photovoltaic power stations, information transmission, and information protection.

[0003] Therefore, how to accurately model and predict power load, and how to build a related information security assurance system based on trusted computing technology, is an important research direction for photovoltaic power generation systems. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a photovoltaic system load prediction method and system based on trusted computing to solve the problems in existing photovoltaic systems, such as limited power generation, waste of power resources caused by failure to supply power on demand, and inability to ensure the security and privacy of user power usage data.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a photovoltaic system load forecasting method based on trusted computing, including: based on a trusted computing module, performing trusted authentication on all users participating in federated learning modeling, and performing information interaction when the participants are verified to be untampered and trustworthy;

[0009] The participant builds a local load forecasting model based on the interactive data, pre-processes the data, and initializes the local load forecasting model for federated computing collaborative training;

[0010] Based on a federated learning approach, local prediction models of multiple participating users are aggregated to form a global load prediction model, and the load of the photovoltaic system is predicted based on the global load prediction model.

[0011] As a preferred solution of the photovoltaic system load forecasting method based on trusted computing described in the present invention, wherein: trusted authentication of all users participating in federated learning modeling includes: establishing a secure channel including demand information and integrity report exchange;

[0012] The participant completes the integrity measurement of its own system to generate an integrity report based on the integrity requirement of the trusted computing module, and digitally signs the integrity report using a private key;

[0013] The trusted computing module verifies the validity of the signature through the public key of the participant, confirming that the report has not been tampered with and originated from the participant; the trusted computing module also compares the measurement value in its locally stored platform configuration register with the measurement value of the participant's integrity report to check whether there is any unauthorized modification behavior in the participant's system;

[0014] When all verifications are passed, the trusted computing module determines that the participant is trustworthy and allows it to conduct federated learning.

[0015] As a preferred solution of the photovoltaic system load prediction method based on trusted computing described in the present invention, wherein: the participating party constructs a local load prediction model based on the interactive data, including: the local load prediction model input is the demand information of the participating users, and each participating user is processed by their own load prediction model to output their own load prediction value.

[0016] As a preferred solution of the photovoltaic system load prediction method based on trusted computing described in the present invention, wherein: preprocessing the data and initializing the local load prediction model for federated computing collaborative training includes: loading the preprocessed data into the local load prediction model, using the data of each participating user to train the local load prediction model until the local load prediction model converges and the training is completed.

[0017] As a preferred solution of the photovoltaic system load prediction method based on trusted computing described in the present invention, the local prediction models of multiple participating users are aggregated to form a global load prediction model based on a federated learning method, including: packaging the parameters of the local load prediction model after the training is completed, and uploading them to the central server to perform a federated learning local prediction model aggregation process.

[0018] As a preferred solution of the photovoltaic system load prediction method based on trusted computing described in the present invention, it also includes: the prediction values ​​of all participating users are summarized and aggregated, and the predictions of different users are integrated to form a global load prediction model.

[0019] As a preferred solution of the photovoltaic system load prediction method based on trusted computing described in the present invention, the multi-party aggregate computing is expressed as:

[0020]

[0021] in, is the aggregated result of the model parameters of all participants, I is the unit matrix, and reversible, is the matrix Ψ i The inverse matrix of j is the covariance matrix of the feature matrix of the jth sub-dataset, W i is the model parameter corresponding to the i-th sub-dataset.

[0022] In a second aspect, the present invention provides a photovoltaic system load forecasting system based on trusted computing, including: a trusted verification module for performing trusted authentication on all users participating in federated learning modeling based on the trusted computing module, and performing information interaction when the participants are verified to be untampered and trustworthy;

[0023] A local load forecasting model building module, used by the participant to build a local load forecasting model based on the interactive data, pre-process the data, and initialize the local load forecasting model for federated computing collaborative training;

[0024] The global load prediction model building module is used to aggregate the local prediction models of multiple participating users to form a global load prediction model based on a federated learning method, and predict the load of the photovoltaic system based on the global load prediction model.

[0025] In a third aspect, the present invention provides a computing device, comprising:

[0026] Memory and processor;

[0027] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the photovoltaic system load prediction method based on trusted computing are implemented.

[0028] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the photovoltaic system load prediction method based on trusted computing.

[0029] Compared with the prior art, the present invention has the following beneficial effects: the present invention uses trusted computing technology to build a trusted execution environment, ensure the credibility of users, and the security of the entire federated learning process. It uses trusted computing combined with federated learning to propose a lossless user electricity behavior analysis model, which not only avoids the risk of data leakage, but also obtains a prediction model with excellent performance. By using the model collaborator, the LSTM prediction model of user data is obtained, and the time series pattern attention mechanism is introduced to increase the weight of important factors and reduce the weight of unimportant elements, so that the LSTM model pays more attention to the data features related to the target prediction. At the same time, in the aggregation stage of the federated learning model, a linear regression-based algorithm is used to achieve a lossless aggregation operation with the performance of the multi-party model consistent with the centralized training model, which can help photovoltaic companies obtain a load prediction model with better performance without being limited by the number of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0031] Figure 1 A schematic diagram of a method and system flow of a photovoltaic system load prediction method based on trusted computing according to an embodiment of the present invention;

[0032] Figure 2 A schematic diagram of photovoltaic system load prediction according to a photovoltaic system load prediction method and system based on trusted computing according to an embodiment of the present invention;

[0033] Figure 3 A schematic diagram of a trusted authentication step of a photovoltaic system load prediction method based on trusted computing and a system according to an embodiment of the present invention;

[0034] Figure 4Schematic diagram of a lossless user behavior analysis model based on federated learning for a photovoltaic system load prediction method and system based on trusted computing according to an embodiment of the present invention

[0035] Figure 5 A schematic diagram of the unit structure of an LSTM long short-term memory model of a photovoltaic system load forecasting method based on trusted computing and a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0037] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0038] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0039] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0040] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0041] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0042] Example 1

[0043] Reference Figure 1 to Figure 4 , is an embodiment of the present invention, which provides a photovoltaic system load prediction method based on trusted computing, such as Figure 1 As shown, including:

[0044] S100: Based on the trusted computing module, all users participating in the federated learning modeling are authenticated. When the participants are verified to be untampered and trustworthy, information exchange is performed;

[0045] S200: The participants build a local load forecasting model based on the interactive data, pre-process the data, and initialize the local load forecasting model for federated computing collaborative training;

[0046] S300: Based on the federated learning method, the local prediction models of multiple participating users are aggregated to form a global load prediction model, and the load of the photovoltaic system is predicted based on the global load prediction model.

[0047] It should be noted that the existing photovoltaic system has certain limitations, and the power generation capacity is limited. Failure to supply power on demand will cause a waste of power resources. Therefore, it is necessary to predict the power load of regional users, formulate better power supply strategies, and better dispatch power. At the same time, when performing load forecasting, the security of user power data is difficult to guarantee, and an information security assurance system needs to be built. In addition, the existing regional user power behavior analysis model cannot meet the needs of privacy and intelligence, so the holders of user data are unwilling to share data, and regional user data has problems such as insufficient data volume and one-sided information contained in the data, making it difficult to jointly analyze or model.

[0048] The embodiments of this application aim at two problems: how to build a trusted execution environment under the photovoltaic system and how to perform load forecasting. Through trusted computing technology, a trusted execution environment is built to ensure the credibility of users and the security of the entire federated learning process. Trusted computing combined with federated learning is used to propose a lossless user electricity behavior analysis model, which not only avoids the risk of data leakage, but also obtains a prediction model with excellent performance. By using model collaborators, an LSTM prediction model of user data is obtained, and the TPA mechanism is introduced to increase the weight of important factors and reduce the weight of unimportant elements, so that the LSTM model pays more attention to data features related to target prediction. At the same time, in the aggregation stage of the federated learning model, a linear regression-based algorithm is used to achieve a lossless aggregation operation with the performance of the multi-party model consistent with the centralized training model, which can help photovoltaic companies obtain a photovoltaic system load forecasting model with better performance without being limited by the number of users.

[0049] Reference Figure 2 It should be noted that before all users participating in federated learning modeling are authenticated and managed securely, the trusted computing module needs to be started. In the embodiment of the present application, the trusted computing module includes a trusted hardware platform and trusted basic software; the trusted hardware platform also includes a trusted platform control module TPCM and a trusted cryptographic module TCP; TPCM provides a trusted root for the trusted computing module, and TPCM relies on the cryptographic services and system measurement functions of the hardware trusted cryptographic module TCM to support the operation of the trusted basic software.

[0050] Specifically, TPCM is started first to measure the trustworthiness of the system kernel. Only after the measurement verification is the control transferred to the operating system, and then the authority is handed over to the basic input and output system BIOS after measurement, and the BIOS performs integrity measurement. Finally, the complete measurement value is placed in the state register of TCM according to the startup sequence, thus forming a linear static trust chain to ensure the trusted operation of the software. The value written to this register must follow the rules:

[0051] PCR inew =Hash(PCR iold value)

[0052] It should be noted that the trusted basic software can control access requests through a variety of mechanisms, such as support mechanisms, control mechanisms, judgment mechanisms, measurement mechanisms, etc.

[0053] In the embodiment of the present application, all users participating in the federated learning modeling are authenticated, including: establishing a secure channel including the exchange of demand information and integrity reports;

[0054] Based on the integrity requirements of the trusted computing module, the participants complete their own system integrity measurement to generate an integrity report, and use the private key to digitally sign the integrity report;

[0055] The trusted computing module verifies the validity of the signature through the public key of the participant, confirming that the report has not been tampered with and originated from the participant; the trusted computing module also compares the measurement values ​​in its locally stored platform configuration register with the measurement values ​​of the participant's integrity report to check whether there is any unauthorized modification of the participant's system;

[0056] When all verifications are passed, the trusted computing module determines that the participant is trustworthy and allows it to conduct federated learning.

[0057] Specifically, Figure 3 As shown, the steps of establishing a secure channel including the exchange of requirement information and integrity report include:

[0058] A1: The participants submit their own trusted integrity reports to TPCM.

[0059] A2: TPCM generates a verification report based on the trusted integrity report received.

[0060] A3: The policy implementation component requests the security policy from the policy judgment component.

[0061] A4: TPCM requests the security policy integrity report from the policy judgment component.

[0062] A5: The policy judgment component returns the integrity report of the security policy to TPCM.

[0063] A6: The policy judgment component returns the security policy to the policy implementation component.

[0064] A7: The policy implementation component performs corresponding operations or adjustments based on the received security policy.

[0065] A8: The policy implementation component records the execution process and results in the database.

[0066] It should be noted that after being authenticated by the trusted computing module, the environment is deemed safe and trustworthy, and in the federated learning process, all information transmission behaviors must be carried out under the management of the trusted computing module. Data transmission is allowed only after the policy implementation component has authenticated the identity of the data transmitter, and all information must be transmitted in a specific channel allocated by the trusted computing module, so as to achieve secure management of the federated learning process by the trusted computing module.

[0067] In an embodiment of the present application, the participants construct a local load prediction model based on the interactive data, including: the local load prediction model input is the demand information of the participating users, each participating user is processed by their own load prediction model, and the output is their own load prediction value.

[0068] In an embodiment of the present application, data is preprocessed and a local load prediction model is initialized for federated computing collaborative training, including: loading the preprocessed data into the local load prediction model, training the local load prediction model using the data of each participating user until the local load prediction model converges and the training is completed.

[0069] In a preferred embodiment, the photovoltaic system load forecasting model is constructed using federated learning technology. First, each user builds his own forecasting model, and then the subject generates the final global model through a series of model aggregation calculations. The framework is as follows: Figure 4 As shown, the specific process is as follows:

[0070] B1: User 1, User 2, etc. each provide their data as input.

[0071] B2: Each user’s data is processed by their own LSTM (Long Short-Term Memory) model. LSTM is a special type of recurrent neural network used to process sequence data, such as time series data.

[0072] B3: The processed data enters the time series weighting module. This module may be responsible for assigning different weights to data at different time points in order to better capture the time series characteristics.

[0073] B4: The weighted data enters the Softmax layer. The Softmax function is usually used for multi-classification problems, which converts a set of arbitrary real numbers into a probability distribution.

[0074] B5: The output of the Softmax layer is the final prediction value.

[0075] B6: The forecast values ​​of all users are summarized and model aggregation is performed. This step will integrate the forecasts from different users to form a global load forecasting model.

[0076] B7: The final result is a global load forecasting model that makes more accurate forecasts based on the contributions of all users.

[0077] In a preferred embodiment, the load forecasting model of the present application selects the LSTM long short-term memory model. The unit structure of LSTM is as follows Figure 5 As shown, where x t Represents the input data at the current moment; c t-1 、c tRespectively represent the internal state of the previous moment and the current moment; h t-1 、h t Respectively represent the state of the hidden layer at the previous moment and the current moment; ☉ represents the multiplication of matrix elements, represents element-wise multiplication, represents the addition of matrix elements; σ and tanh represent sigmoid and tanh activation functions respectively; f t 、i t and t Respectively represent the calculation variables of the forget gate, input gate and output gate at the current moment, Indicates the unit status of the input at this moment:

[0078] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0079] f t =σ(W i *[h t-1 ,x t ]+b i )

[0080]

[0081] o t =σ(W o *[h t-1 ,x t ]+b o )

[0082] h t =o t *tanh(c t )

[0083] Among them, W f , W i , W o are the weight matrices of the forget gate, input gate, and output gate respectively; b f 、b i 、b o They are the output bias items of the forget gate, input gate, and output gate respectively; * represents the Hadamard product of the matrix.

[0084] In a preferred embodiment, a temporal pattern attention mechanism is also introduced into the LSTM network to enhance the model's attention to business-related data, thereby reducing prediction errors and enhancing robustness. The specific principles are as follows:

[0085] In the LSTM iteration process, the hidden state of the forward unit is ht , the unit status information is s t , the time attention weight vector l corresponding to each historical moment at the current moment t for:

[0086] l t = relu(W d h t +b d )

[0087] Among them, l t =[l 1,t ,l 2,t ,…,l k,t ], k is the time window length of the input sequence; W d is the trainable weight matrix; b d is the bias vector for calculating the temporal attention weights.

[0088] Use the Softmax function for normalization to get the temporal attention weight β t :

[0089] β t =[β 1,t ,β 2,t ,…,β i,t ,…β k,t ]

[0090] Among them, β i,t is the attention weight value at the i-th moment, and the comprehensive temporal information state c is further obtained t :

[0091]

[0092] In an embodiment of the present application, based on a federated learning method, aggregating the local prediction models of multiple participating users to form a global load prediction model includes: packaging the parameters of the local load prediction model after training, uploading them to a central server, and performing a federated learning local prediction model aggregation process.

[0093] In the embodiment of the present application, it also includes: the predicted values ​​of all participating users are summarized and aggregated, and the predictions of different users are integrated to form a global load prediction model.

[0094] In the embodiment of the present application, multi-party aggregate computing is expressed as:

[0095]

[0096] in, is the aggregated result of the model parameters of all participants, I is the unit matrix, and reversible, is the matrix Ψ i The inverse matrix of j is the covariance matrix of the feature matrix of the jth sub-dataset, W i is the model parameter corresponding to the i-th sub-dataset.

[0097] In another optional embodiment, the present application may also select a recurrent neural network RNN ​​as a load forecasting model; the implementation steps may include:

[0098] C1: Collect historical load data and other relevant features, such as weather data, holiday information, etc.; divide the data into training set and test set.

[0099] C2: Define an RNN model. You can choose a single-layer or multi-layer RNN. Use a suitable activation function, such as the tanh function or the ReLU function.

[0100] C3: Use the training set data to train the RNN model; adjust the model parameters through the back propagation algorithm to minimize the prediction error.

[0101] C4: Use the test set data to evaluate the performance of the model and apply the trained model for load forecasting.

[0102] In another optional embodiment, the present application may also select a convolutional neural network (CNN) as a load forecasting model; the implementation steps may include:

[0103] D1: Collect historical load data and other relevant features; divide the data into training set and test set.

[0104] D2: Define a CNN model, including several convolutional layers, pooling layers, and fully connected layers; use appropriate activation functions, such as the ReLU function.

[0105] D3: Use the training set data to train the CNN model; adjust the model parameters through the back propagation algorithm to minimize the prediction error.

[0106] D4: Use the test set data to evaluate the performance of the model; apply the trained model to load forecasting.

[0107] It should be noted that the present application prefers LSTM network as a load forecasting model because it is suitable for modeling and forecasting time series data. Through the mechanism of forget gate, input gate and control gate, long-term dependencies can be captured efficiently. The gating mechanism enables LSTM to selectively forget or save previous information and apply it to the current state. LSTM has excellent performance in processing time series data and is suitable for analyzing and predicting data with strong correlation before and after, such as photovoltaic system load.

[0108] Example 2

[0109] In an exemplary embodiment, a photovoltaic system load prediction system based on trusted computing is also provided in this embodiment, including:

[0110] The trusted verification module is used to authenticate all users participating in the federated learning modeling based on the trusted computing module. When the participants are verified to be untampered and trustworthy, information exchange is carried out;

[0111] The local load forecasting model building module is used by the participants to build a local load forecasting model based on the interactive data, pre-process the data, and initialize the local load forecasting model for federated computing collaborative training;

[0112] The global load forecasting model building module is used to aggregate the local forecasting models of multiple participating users to form a global load forecasting model based on federated learning, and predict the load of the photovoltaic system based on the global load forecasting model.

[0113] In an optional embodiment, the photovoltaic system load prediction system based on trusted computing may perform the following steps:

[0114] Trusted authentication of all users participating in federated learning modeling includes: establishing a secure channel for the exchange of requirement information and integrity reports;

[0115] Based on the integrity requirements of the trusted computing module, the participants complete their own system integrity measurement to generate an integrity report, and use the private key to digitally sign the integrity report;

[0116] The trusted computing module verifies the validity of the signature through the public key of the participant, confirming that the report has not been tampered with and originated from the participant; the trusted computing module also compares the measurement values ​​in its locally stored platform configuration register with the measurement values ​​of the participant's integrity report to check whether there is any unauthorized modification of the participant's system;

[0117] When all verifications are passed, the trusted computing module determines that the participant is trustworthy and allows it to conduct federated learning.

[0118] The input of the local load forecasting model is the demand information of participating users. Each participating user processes it through their own load forecasting model and outputs their own load forecast value.

[0119] The preprocessed data is loaded into the local load forecasting model, and the local load forecasting model is trained using the data of each participating user until the local load forecasting model converges and the training is completed.

[0120] The parameters of the local load prediction model after training are encapsulated and uploaded to the central server for the federated learning local prediction model aggregation process.

[0121] The forecast values ​​of all participating users are summarized and aggregated, integrating the forecasts of different users to form a global load forecasting model.

[0122] Multi-party aggregate calculation is expressed as:

[0123]

[0124] in, is the aggregated result of the model parameters of all participants, I is the unit matrix, and reversible, is the matrix Ψ i The inverse matrix of j is the covariance matrix of the feature matrix of the jth sub-dataset, W i is the model parameter corresponding to the i-th sub-dataset.

[0125] This embodiment further provides a computing device, which is applicable to the case of a photovoltaic system load prediction method based on trusted computing, including:

[0126] A memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the photovoltaic system load prediction method based on trusted computing as proposed in the above embodiment.

[0127] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for photovoltaic system load prediction based on trusted computing as proposed in the above embodiment is implemented.

[0128] The storage medium proposed in this embodiment and the method for implementing photovoltaic system load prediction based on trusted computing proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0129] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

[0130] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention 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 invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A photovoltaic system load forecasting method based on trusted computing, characterized in that: include: Based on the trusted computing module, all users participating in the federated learning modeling are authenticated. When the participants are verified to be untampered and trustworthy, information exchange is carried out. The participant builds a local load forecasting model based on the interactive data, pre-processes the data, and initializes the local load forecasting model for federated computing collaborative training; Based on a federated learning approach, local prediction models of multiple participating users are aggregated to form a global load prediction model, and the load of the photovoltaic system is predicted based on the global load prediction model.

2. The photovoltaic system load prediction method based on trusted computing according to claim 1, characterized in that: Trusted authentication of all users participating in federated learning modeling includes: establishing a secure channel for the exchange of requirement information and integrity reports; The participant completes the integrity measurement of its own system based on the integrity requirement of the trusted computing module to generate an integrity report, and digitally signs the integrity report using a private key; The trusted computing module verifies the validity of the signature through the public key of the participant, confirming that the report has not been tampered with and originated from the participant; the trusted computing module also compares the measurement value in its locally stored platform configuration register with the measurement value of the participant's integrity report to check whether there is any unauthorized modification behavior in the participant's system; When all verifications are passed, the trusted computing module determines that the participant is trustworthy and allows it to conduct federated learning.

3. The photovoltaic system load prediction method based on trusted computing according to claim 2, characterized in that: The participant constructs a local load forecasting model based on the interactive data, including: the local load forecasting model input is the demand information of the participating users, each participating user is processed by their own load forecasting model, and outputs their own load forecast value.

4. The photovoltaic system load prediction method based on trusted computing according to claim 2 or 3, characterized in that: Preprocessing the data and initializing the local load forecasting model for federated computing collaborative training includes: loading the preprocessed data into the local load forecasting model, training the local load forecasting model using the data of each participating user, until the local load forecasting model converges and the training is completed.

5. The photovoltaic system load prediction method based on trusted computing according to claim 4, characterized in that: Based on the federated learning method, the local prediction models of multiple participating users are aggregated to form a global load prediction model, including: packaging the parameters of the local load prediction model after the training is completed, and uploading them to the central server to perform the federated learning local prediction model aggregation process.

6. The photovoltaic system load prediction method based on trusted computing according to claim 5, characterized in that: Also includes: The forecast values ​​of all participating users are summarized and aggregated, integrating the forecasts of different users to form a global load forecasting model.

7. The photovoltaic system load prediction method based on trusted computing according to claim 6, characterized in that: include: Multi-party aggregate calculation is expressed as: in, is the aggregated result of the model parameters of all participants, I is the unit matrix, and reversible, is the matrix Ψ i The inverse matrix of j is the covariance matrix of the feature matrix of the jth sub-dataset, W i is the model parameter corresponding to the i-th sub-dataset.

8. A photovoltaic system load forecasting system based on trusted computing, characterized in that: include: The trusted verification module is used to authenticate all users participating in the federated learning modeling based on the trusted computing module. When the participants are verified to be untampered and trustworthy, information exchange is carried out; A local load forecasting model building module, used by the participant to build a local load forecasting model based on the interactive data, pre-process the data, and initialize the local load forecasting model for federated computing collaborative training; The global load prediction model building module is used to aggregate the local prediction models of multiple participating users to form a global load prediction model based on a federated learning method, and predict the load of the photovoltaic system based on the global load prediction model.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the photovoltaic system load prediction method based on trusted computing as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the photovoltaic system load prediction method based on trusted computing as described in any one of claims 1 to 7.

Images