KR102570617B1 - System, method and program for transacting power forecasting ai model based on blockchain - Google Patents

Abstract

It relates to a blockchain-based power consumption prediction artificial intelligence model trading system, method, and program for guaranteeing the property rights of the power consumption prediction artificial intelligence model. A blockchain network unit including a plurality of blockchain nodes that trains the generated power consumption prediction artificial intelligence model, and an NFT market that registers NFT (Non-Fungible Token) metadata corresponding to the learned power consumption prediction artificial intelligence model. A place unit and a power consumption prediction artificial intelligence model purchase unit that purchases a power consumption prediction artificial intelligence model registered in the NFT marketplace unit, wherein the NFT marketplace unit includes a power consumption prediction artificial intelligence model of the NFT metadata. When the purchase price is generated, the purchase price is distributed according to the preset profit distribution information, and the distributed purchase price is used to generate the power consumption data provider and the power consumption prediction artificial intelligence model that provides the power consumption data. It is characterized in that each payment is made to one block chain node.

Description

Blockchain-based power consumption prediction artificial intelligence model trading system, method and program {SYSTEM, METHOD AND PROGRAM FOR TRANSACTING POWER FORECASTING AI MODEL BASED ON BLOCKCHAIN}

The present disclosure relates to a blockchain-based power consumption prediction artificial intelligence model trading system, and more specifically, to a blockchain-based power consumption prediction artificial intelligence model trading system, method, and program for guaranteeing property rights of the power consumption prediction artificial intelligence model. .

In general, artificial intelligence (AI), including machine learning, performs prediction tasks such as regression, classification, and clustering based on what a computer learns on its own based on data. say that As machine learning artificial intelligence shows excellent performance, various industries are developing and applying excellent prediction models based on artificial intelligence.

Recently, the electric power industry is also applying AI to various fields based on the accumulated big data, and AI-based application algorithms are showing excellent performance. Statistical techniques such as ARIMA and LSTM-based deep learning are used in actual work to predict peak power demand to maintain stability in power supply and demand. It not only predicts SMP and maximum power supply by using AI, but also creates and applies a health index that predicts the lifespan of equipment based on facility sensor data.

This artificial intelligence model is a microgrid power supply and demand predictor, a power consumption predictor that predicts power consumption considering the peak power consumption value using LSTM (Long Short Term Memory) and RMSE (Root Mean Square Error), It has been applied to various technologies such as DC power relay switch system for surplus power P2P direct transaction service.

However, these existing technologies have limitations in the accuracy and reliability of power consumption prediction due to the lack of power consumption data necessary for learning the artificial intelligence model for power consumption prediction, and secure a lot of power consumption data and deep learning servers. In order to do this, a lot of cost can be incurred, and in particular, it was difficult to install a high-performance deep learning server that shortens the learning time.

Therefore, in the future, the power consumption data provider and the deep learning server provider are compensated for the resources used to generate the power consumption prediction artificial intelligence model to induce participation, thereby solving the problem of lack of resources for learning the artificial intelligence model, Ultimately, it is required to develop a power consumption prediction artificial intelligence model trading system that can guarantee the property rights of the resources used for learning the power consumption prediction artificial intelligence model.

Republic of Korea Patent Publication No. 10-2020-0109196 (2020. 09. 22)

One object of the present disclosure to solve the above problems is to register a power consumption prediction artificial intelligence model learned based on power consumption data in the NFT marketplace, and when the registered power consumption prediction artificial intelligence model is purchased, in advance. The purchase price is distributed in accordance with the set revenue distribution information and paid to the power consumption data provider and the deep learning server provider, respectively, to compensate the power consumption data provider and the deep learning server provider for the resources used in generating the power consumption prediction artificial intelligence model. Blockchain-based power consumption prediction artificial intelligence that can induce active participation by paying a To provide intelligent model trading systems, methods and programs.

The problems to be solved by the present disclosure are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

A blockchain-based power consumption prediction artificial intelligence model trading system according to an embodiment of the present disclosure for solving the above problems is a power consumption data providing unit providing power consumption data, pre-generated based on the power consumption data A blockchain network unit including a plurality of blockchain nodes that trains a power consumption prediction artificial intelligence model, and an NFT marketplace unit that registers NFT (Non-Fungible Token) metadata corresponding to the learned power consumption prediction artificial intelligence model. and a power consumption prediction artificial intelligence model purchase unit for purchasing a power consumption prediction artificial intelligence model registered in the NFT marketplace unit, wherein the NFT market place unit purchases the power consumption prediction artificial intelligence model of the NFT metadata. When the purchase price is generated, the purchase price is distributed according to the preset profit distribution information, and the power consumption data providing unit that provides the power consumption data and the block that generates the power consumption prediction artificial intelligence model It is characterized by payment to each chain node.

In an embodiment, the power consumption data providing unit transmits a broadcast to a plurality of blockchain nodes included in the blockchain network unit, and the fastest among response signals received from the blockchain nodes. It is characterized in that the power consumption data is provided to an occupied node that is a blockchain node that has transmitted a response signal.

In an embodiment, the power consumption data providing unit, when providing the power consumption data, transmits a power consumption prediction artificial intelligence model generation request signal and power consumption data to an occupied node that is a blockchain node that has transmitted the fastest response signal. characterized in that they are provided at the same time.

The reason for this is to generate a power consumption prediction artificial intelligence model in advance while the power consumption data is being transmitted, since the amount of power consumption data is large and it takes a long time to transmit.

For example, if it takes about 20 ms to send a model creation request and about 100 ms to send data, the delay rate of service delivery is reduced by creating a model in advance during data transmission, rather than creating a model after data transmission. can make it

In an embodiment, the blockchain network unit includes a plurality of blockchain nodes, and among the plurality of blockchain nodes, a blockchain node that transmits a response signal to the broadcast of the power consumption data providing unit most quickly is preoccupied. It is characterized in that it is selected as a node.

In an embodiment, the preemption node generates the power consumption prediction AI model when receiving the power consumption prediction artificial intelligence model generation request signal from the power consumption data provider, and the power consumption data provider generates the power consumption prediction artificial intelligence model. When consumption data is received, the generated power consumption prediction artificial intelligence model is trained based on the power consumption data, and when the generated power consumption prediction artificial intelligence model is completed learning, the NFT corresponding to the power consumption prediction artificial intelligence model It is characterized by generating meta data.

In an embodiment, the preoccupied node may start learning the power consumption prediction artificial intelligence model immediately upon receiving the power consumption data when training the generated power consumption prediction artificial intelligence model.

In an embodiment, the preemption node, when learning the generated power consumption prediction artificial intelligence model, proceeds with transfer learning of the pre-generated power consumption prediction artificial intelligence model based on the received power consumption data, and epoch If (Epoch) exceeds a first set number of times, learning is performed by releasing layer freezing of the power consumption prediction artificial intelligence model, and if the epoch is more than a second set number of times, learning of the power consumption prediction artificial intelligence model characterized by completion.

In an embodiment, if the epoch is less than the second set number of times, the preemption node checks whether the loss value of the verification dataset among the power consumption data is improved during a specific number of epochs, and improves the loss value of the verification dataset If there is no, it is characterized by early stopping the learning of the power consumption prediction artificial intelligence model.

In an embodiment, the preemption node, when generating NFT metadata corresponding to the power consumption prediction artificial intelligence model, predicts the power consumption based on information of the generated power consumption prediction artificial intelligence model and data information for learning. It is characterized by generating NFT meta data corresponding to the artificial intelligence model.

In an embodiment, the preemption node, when generating NFT metadata corresponding to the power consumption prediction artificial intelligence model, when NFT metadata corresponding to the power consumption prediction artificial intelligence model is generated, the generated NFT metadata It is characterized by transmitting to the NFT marketplace unit through an NFT metadata transmission protocol.

In an embodiment, when the NFT marketplace unit distributes the purchase price, the power consumption prediction artificial intelligence model purchase unit purchases the power consumption prediction artificial intelligence model and pays the purchase price in cryptocurrency, the cryptocurrency Revenue distribution information including the distribution ratio between the power consumption data provider of the power consumption data providing unit that provided the power consumption data for the purchase price and the deep learning server provider of the blockchain node that generated the power consumption prediction artificial intelligence model and distributing the distributed cryptocurrency to the cryptocurrency wallet address of the power consumption data provider and the cryptocurrency wallet address of the deep learning server provider, respectively.

A power consumption prediction artificial intelligence model trading method according to an embodiment of the present disclosure is a power consumption prediction artificial intelligence model trading system including a power consumption data providing unit, a blockchain network unit having a plurality of blockchain nodes, and an NFT marketplace unit. As a power consumption prediction artificial intelligence model trading method, when the blockchain network unit receives a broadcast from the power consumption data providing unit, the plurality of blockchain nodes based on the response signal of the blockchain node to the broadcast Selecting a preempted node from among, generating a power consumption prediction artificial intelligence model by the preempted node, and when the preoccupied node receives power consumption data from the power consumption data providing unit, the power consumption data is determined based on the power consumption data. Learning a prediction artificial intelligence model, generating non-fungible token (NFT) metadata corresponding to the learned power consumption prediction artificial intelligence model, and transmitting the NFT marketplace unit to the NFT marketplace unit; When the power consumption prediction artificial intelligence model of the data is purchased, the purchase price is distributed according to the preset revenue distribution information, and the power consumption data provider and the power consumption prediction artificial intelligence model are generated by the power consumption data providing unit that provided the power consumption data. It is characterized by including the step of paying each to the cryptocurrency wallet address of the deep learning server provider of one blockchain node.

A computer program that provides a power consumption prediction artificial intelligence model trading method of a power consumption prediction artificial intelligence model trading system according to another embodiment of the present disclosure for solving the above problems is combined with a computer that is hardware and any of the above methods It is stored on a medium to perform a method.

In addition to this, another method for implementing the present disclosure, another system, and a computer readable recording medium recording a computer program for executing the method may be further provided.

As described above, according to the present disclosure, the power consumption prediction artificial intelligence model learned based on power consumption data is registered in the NFT marketplace, and when the registered power consumption prediction artificial intelligence model is purchased, the purchase is made in accordance with the preset revenue distribution information. The payment is distributed and paid to the power consumption data provider and the deep learning server provider, respectively, to induce active participation by paying compensation for the resources used to generate the power consumption prediction artificial intelligence model to the power consumption data provider and the deep learning server provider. , it can solve the problem of lack of resources for learning artificial intelligence models, and guarantee the property rights of resources used for learning power consumption prediction artificial intelligence models.

In addition, the present disclosure, when learning a power consumption prediction artificial intelligence model, completes preparation for learning before transmitting big data (power consumption data) through a request for generating a power consumption prediction artificial intelligence model, shortens the learning time, and transfers Unnecessary consumption of computing power can be prevented through learning, automatic learning rate adjustment, and early termination.

In addition, the present disclosure can automatically calculate the model NFT price through a multiple linear regression algorithm in proportion to the accuracy of the generated power consumption prediction artificial intelligence model, the number of learning data, and the learning time, and to provide NFT marketplaces from buyers. Cryptocurrency paid through this is distributed according to the revenue distribution information of the power consumption data provider and the deep learning server provider, and you can receive compensation for the resources used for learning the power consumption prediction artificial intelligence model.

The effects of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 is a schematic diagram for explaining a blockchain-based power consumption prediction artificial intelligence model trading system according to an embodiment of the present disclosure.
2 is a diagram for explaining a blockchain node of a blockchain-based power consumption prediction artificial intelligence model trading system according to the present disclosure.
3 is a diagram for explaining NFT metadata corresponding to the artificial intelligence model for predicting power consumption according to the present disclosure.
4 and 5 are flowcharts for explaining a blockchain-based power consumption prediction artificial intelligence model transaction method according to the present disclosure.
6 is a diagram for explaining a transaction process of a blockchain-based power consumption prediction artificial intelligence model according to the present disclosure.
7 is a schematic diagram illustrating a network function of an artificial intelligence model for predicting power consumption according to an embodiment of the present disclosure.

Advantages and features of the present disclosure, and methods of achieving them, will become clear with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present disclosure is not limited to the embodiments disclosed below and may be implemented in various different forms, but only the present embodiments make the disclosure of the present disclosure complete, and those skilled in the art to which the present disclosure belongs It is provided to fully inform the scope of the present disclosure, which is only defined by the scope of the claims.

Terminology used herein is for describing the embodiments and is not intended to limit the present disclosure. In this specification, singular forms also include plural forms unless specifically stated otherwise in a phrase. As used herein, "comprises" and/or "comprising" does not exclude the presence or addition of one or more other elements other than the recited elements. Like reference numerals throughout the specification refer to like elements, and “and/or” includes each and every combination of one or more of the recited elements. Although "first", "second", etc. are used to describe various components, these components are not limited by these terms, of course. These terms are only used to distinguish one component from another. Accordingly, it goes without saying that the first element mentioned below may also be the second element within the technical spirit of the present disclosure.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings commonly understood by those skilled in the art to which this disclosure belongs. In addition, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless explicitly specifically defined.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

Prior to the description, the meaning of the terms used in this specification will be briefly described. However, it should be noted that the explanation of terms is intended to help the understanding of the present specification, and is not used in the sense of limiting the technical spirit of the present disclosure unless explicitly described as limiting the present disclosure.

In this specification, neural networks, artificial neural networks, and network functions may often be used interchangeably.

Also, throughout this specification, a neural network, a neural network, and a network function may be used with the same meaning. A neural network may consist of a set of interconnected computational units, which may be generally referred to as “nodes”. These “nodes” may also be referred to as “neurons”. A neural network includes at least two or more nodes. Nodes (or neurons) constituting neural networks may be interconnected by one or more “links”.

1 is a schematic diagram for explaining a blockchain-based power consumption prediction artificial intelligence model trading system according to an embodiment of the present disclosure.

As shown in FIG. 1, the blockchain-based power consumption prediction artificial intelligence model trading system of the present disclosure includes a power consumption data providing unit 30 and a blockchain network unit 10 including a plurality of blockchain nodes 110. ), an NFT marketplace unit 40, and a power consumption prediction artificial intelligence model purchase unit 50.

Here, the power consumption data providing unit 30 is a terminal of a power consumption data provider providing power consumption data, such as PC (Personal Computer), network TV (Network TV), HBBTV (Hybrid Broadcast Broadband TV), smart TV ( Standing devices such as Smart TV) and IPTV (Internet Protocol TV), and mobile devices such as smart phones, tablet PCs, notebooks, and personal digital assistants (PDAs) mobile device or handheld device) may be included.

In addition, the network that communicates and connects the power consumption data provider 30 and the external server includes both wired and wireless networks, and performs pairing or/and data transmission/reception between the power consumption data provider 30 and the external server. It refers to a communication network that supports various communication standards or protocols for

These wired/wireless networks include all communication networks currently or to be supported in the future according to standards, and can support all one or more communication protocols therefor.

Such wired/wireless networks include, for example, Universal Serial Bus (USB), Composite Video Banking Sync (CVBS), Component (Component), S-Video (Analog), Digital Visual Interface (DVI), High Definition Multimedia Interface (HDMI), Networks for wired connections such as RGB and D-SUB, communication standards and protocols for them, Bluetooth, RFID (Radio Frequency Identification), infrared communication (IrDA: Infrared Data Association), UWB (Ultra Wideband), ZigBee (ZigBee), Digital Living Network Alliance (DLNA), Wireless LAN (WLAN) (Wi-Fi), Wireless broadband (Wibro), World Interoperability for Microwave Access (Wimax), High Speed Downlink Packet Access (HSDPA), LTE/LTE -A (Long Term Evolution/LTE-Advanced), it may be formed by a network for wireless connection such as Wi-Fi direct and a communication standard or protocol for it.

As such, the power consumption data providing unit 30 that provides power consumption data transmits a broadcast to a plurality of blockchain nodes 110 included in the blockchain network unit 10, and blocks the blockchain nodes. Power consumption data may be provided to the preoccupied node 111, which is the blockchain node 110 that has transmitted the fastest response signal among the response signals received from (110).

Here, when the power consumption data providing unit 30 provides the power consumption data, the power consumption prediction artificial intelligence model generation request signal and the power consumption data are sent to the preoccupied node 111, which is the blockchain node that transmitted the fastest response signal. can be provided simultaneously.

The reason for this is to generate a power consumption prediction artificial intelligence model in advance while the power consumption data is being transmitted, since the amount of power consumption data is large and it takes a long time to transmit.

For example, if it takes about 20 ms to send a model creation request and about 100 ms to send data, the delay rate of service delivery is reduced by creating a model in advance during data transmission, rather than creating a model after data transmission. can make it

Next, the block-chain network unit 10 may include a block-chain node 110 that trains a power consumption prediction artificial intelligence model generated in advance based on power consumption data.

Here, the blockchain network unit 10 may include a plurality of blockchain nodes 110, and among the plurality of blockchain nodes 110, a response signal to the broadcast of the power consumption data providing unit 30 The blockchain node 110 that transmits the fastest can be selected as the preemption node 111.

Here, the preemption node 111 generates a power consumption prediction artificial intelligence model when receiving a power consumption prediction artificial intelligence model generation request signal from the power consumption data provider 30, and generates power from the power consumption data provider 30. When the consumption data is received, the power consumption prediction artificial intelligence model generated based on the power consumption data is trained, and when the generated power consumption prediction artificial intelligence model is trained, the NFT metadata corresponding to the power consumption prediction artificial intelligence model (20 ) can be created.

In addition, when generating the power consumption prediction artificial intelligence model, the preemption node 111 may generate the power consumption prediction artificial intelligence model at the time of receiving the power consumption prediction artificial intelligence model generation request signal. but not limited thereto.

Here, the preemption node 111 may recognize the time when the power consumption prediction artificial intelligence model generation request signal is received as the generation start point of the power consumption prediction artificial intelligence model.

In addition, the generated power consumption prediction artificial intelligence model is a product for performing learning based on power consumption data in the preoccupied node 111, and can serve as an artificial intelligence program itself that helps predict power consumption.

Next, when the preemption node 111 learns the generated power consumption prediction artificial intelligence model, it may start learning the power consumption prediction artificial intelligence model immediately upon receiving power consumption data.

Here, the preemption node 111 may recognize the time at which power consumption data is received as the learning start time of the power consumption prediction artificial intelligence model.

In addition, the preemption node 111, when learning the generated power consumption prediction artificial intelligence model, proceeds with transfer learning of the pre-generated power consumption prediction artificial intelligence model based on the received power consumption data, Epoch If the first set number of times is exceeded, the layer of the power consumption prediction artificial intelligence model is unfrozen and learning is performed, and if the epoch is the second set number or more, the power consumption prediction artificial intelligence model learning may be completed.

For example, the epoch of the first set number of times may be about the 50th epoch, and the epoch of the second set number of times may be about the 100th epoch.

In addition, if the epoch is less than the second set number of times, the preemption node 111 checks whether the loss value of the verification dataset among the power consumption data is improved during a specific number of epochs, and if there is no improvement in the loss value of the verification dataset, the power consumption Learning of predictive AI models can be early stopped.

For example, the specific number of epochs may be about 5 epochs, which is only an example, but is not limited thereto.

In addition, the preemption node 111 may calculate and predict the accuracy of the trained power consumption prediction AI model when training the generated power consumption prediction AI model.

In addition, the preemption node 111 may calculate the power consumption prediction artificial intelligence model price when generating the NFT metadata 20 corresponding to the power consumption prediction artificial intelligence model.

Here, the preemption node 111 calculates the price of the power consumption prediction artificial intelligence model, based on the multiple linear regression algorithm proportional to the accuracy of the power consumption prediction artificial intelligence model, the number of learning data, and the learning time. The intelligent model price can be calculated.

In addition, when the preemption node 111 generates the NFT metadata 20 corresponding to the power consumption prediction artificial intelligence model, the power consumption prediction artificial intelligence model is based on the information of the generated power consumption prediction artificial intelligence model and the learning data information. NFT meta data 20 corresponding to the intelligence model can be generated.

Here, the information of the generated power consumption prediction artificial intelligence model may include accuracy information of the power consumption prediction artificial intelligence model, weight file, NFT price information, deep learning server provider cryptocurrency wallet address of the preoccupied node, and revenue distribution information. The learning data information may include a power consumption data collection period, power consumption data collection location information, the number of power consumption data, and a power consumption data provider cryptocurrency wallet address of a power consumption data provider.

In addition, when the NFT meta data 20 corresponding to the power consumption prediction artificial intelligence model is generated, the preemption node 111 generates the NFT meta data 20 corresponding to the power consumption prediction artificial intelligence model. Metadata may be transmitted to the NFT marketplace unit 40 through an NFT metadata transmission protocol.

Next, the NFT marketplace unit 40 may register NFT (Non-Fungible Token) metadata 20 corresponding to the learned power consumption prediction artificial intelligence model.

Here, the NFT marketplace unit 40 distributes the purchase price in accordance with the preset profit distribution information when the power consumption prediction artificial intelligence model of the NFT meta data is purchased and the purchase price is generated, and the distributed purchase price is used as power consumption. The power consumption data providing unit 30 that provided the data and the preoccupied node 111, which is a blockchain node that generated the power consumption prediction artificial intelligence model, can be paid respectively.

For example, when the NFT marketplace unit 40 distributes the purchase price, the power consumption prediction artificial intelligence model purchasing unit 50 purchases the power consumption prediction artificial intelligence model and pays the purchase price in cryptocurrency. Distribution of the purchase price for the power consumption data between the power consumption data provider of the power consumption data providing unit 30 and the deep learning server provider of the preoccupied node 111, which is a blockchain node that created the power consumption prediction artificial intelligence model. It is distributed according to the profit distribution information including the ratio, and the cryptocurrency of the distributed purchase price can be paid to the cryptocurrency wallet address of the power consumption data provider and the cryptocurrency wallet address of the deep learning server provider.

Next, the power consumption prediction artificial intelligence model purchase unit 50 may purchase the power consumption prediction artificial intelligence model registered in the NFT marketplace unit 40 .

Here, the power consumption prediction artificial intelligence model purchase unit 50 is a terminal of a power consumption prediction artificial intelligence model buyer who purchases a power consumption prediction artificial intelligence model registered in the NFT marketplace unit 40, a PC (Personal Computer), Network TV (Network TV), HBBTV (Hybrid Broadcast Broadband TV), Smart TV (Smart TV), IPTV (Internet Protocol TV), such as standing devices, smart phones (Smart Phones), tablet PCs (Tablet PCs) ), a notebook, a personal digital assistant (PDA), and the like may all be included.

In addition, the network that communicates and connects the power consumption prediction artificial intelligence model purchase unit 50, the NFT marketplace unit 40 and the external server includes both wired and wireless networks, and the power consumption prediction artificial intelligence model purchase unit 50 It collectively refers to a communication network that supports various communication standards or protocols for pairing or / and data transmission and reception between the NFT marketplace unit 40 and external servers.

These wired/wireless networks include all communication networks currently or to be supported in the future according to standards, and can support all one or more communication protocols therefor.

Such wired/wireless networks include, for example, Universal Serial Bus (USB), Composite Video Banking Sync (CVBS), Component (Component), S-Video (Analog), Digital Visual Interface (DVI), High Definition Multimedia Interface (HDMI), Networks for wired connections such as RGB and D-SUB, communication standards and protocols for them, Bluetooth, RFID (Radio Frequency Identification), infrared communication (IrDA: Infrared Data Association), UWB (Ultra Wideband), ZigBee (ZigBee), Digital Living Network Alliance (DLNA), Wireless LAN (WLAN) (Wi-Fi), Wireless broadband (Wibro), World Interoperability for Microwave Access (Wimax), High Speed Downlink Packet Access (HSDPA), LTE/LTE -A (Long Term Evolution/LTE-Advanced), it may be formed by a network for wireless connection such as Wi-Fi direct and a communication standard or protocol for it.

As such, according to the present disclosure, the power consumption prediction artificial intelligence model learned based on power consumption data is registered in the NFT marketplace, and when the registered power consumption prediction artificial intelligence model is purchased, the purchase is made in accordance with preset revenue distribution information. The payment is distributed and paid to the power consumption data provider and the deep learning server provider, respectively, to induce active participation by paying compensation for the resources used to generate the power consumption prediction artificial intelligence model to the power consumption data provider and the deep learning server provider. , it can solve the problem of lack of resources for learning artificial intelligence models, and guarantee the property rights of resources used for learning power consumption prediction artificial intelligence models.

In addition, the present disclosure, when learning a power consumption prediction artificial intelligence model, completes preparation for learning before transmitting big data (power consumption data) through a request for generating a power consumption prediction artificial intelligence model, shortens the learning time, and transfers Unnecessary consumption of computing power can be prevented through learning, automatic learning rate adjustment, and early termination.

In addition, the present disclosure can automatically calculate the model NFT price through a multiple linear regression algorithm in proportion to the accuracy of the generated power consumption prediction artificial intelligence model, the number of learning data, and the learning time, and to provide NFT marketplaces from buyers. Cryptocurrency paid through this is distributed according to the revenue distribution information of the power consumption data provider and the deep learning server provider, and you can receive compensation for the resources used for learning the power consumption prediction artificial intelligence model.

2 is a diagram for explaining a blockchain node of a blockchain-based power consumption prediction artificial intelligence model trading system according to the present disclosure.

As shown in FIG. 2, the blockchain network unit 10 of the present disclosure may include a plurality of blockchain nodes 110, and among the plurality of blockchain nodes 110, a power consumption data providing unit ( The blockchain node 110 that transmits the response signal to the broadcast of 30) the fastest may be selected as the preoccupied node 111.

Here, the blockchain node 110 includes a power consumption data receiving unit 120 receiving power consumption data, a power consumption prediction artificial intelligence model learning unit 131 and a power consumption prediction artificial intelligence model prediction unit 132. It may include a power consumption prediction artificial intelligence model generation unit 130, a power consumption prediction artificial intelligence model NFT metadata generation unit 140, and a power consumption prediction artificial intelligence model NFT metadata transmission unit 150.

For example, the power consumption data receiving unit 120 may receive power consumption data information for learning a power consumption prediction artificial intelligence model and a cryptocurrency wallet address of a power consumption data provider.

Subsequently, the power consumption prediction artificial intelligence model generation unit 130 generates a power consumption prediction artificial intelligence model, trains the generated power consumption prediction artificial intelligence model based on the power consumption data, and generates a power consumption prediction artificial intelligence model. accuracy can be calculated.

Here, the power consumption prediction artificial intelligence model learning unit 131 may perform learning of a power consumption prediction artificial intelligence model generated in advance based on the received power consumption data.

The power consumption prediction artificial intelligence model learning unit 131 performs transfer learning of a pre-generated power consumption prediction artificial intelligence model based on the received power consumption data, and if the epoch exceeds a first set number of times, power consumption Learning is performed by releasing layer freezing of the prediction AI model, and learning of the power consumption prediction AI model may be completed when the number of epochs is greater than or equal to the second set number.

At this time, if the epoch is less than the second set number of times, the power consumption prediction artificial intelligence model learning unit 131 checks whether the loss value of the verification dataset among the power consumption data is improved during a specific number of epochs, and the loss of the verification dataset If there is no improvement in the value, the learning of the power consumption prediction artificial intelligence model can be terminated early (early stopping).

In addition, the power consumption prediction artificial intelligence model prediction unit 132 may calculate and predict the accuracy of the power consumption prediction artificial intelligence model that has been learned.

Subsequently, the power consumption prediction artificial intelligence model NFT metadata generation unit 140 generates NFT metadata corresponding to the power consumption prediction artificial intelligence model based on the information of the generated power consumption prediction artificial intelligence model and the training data information. can

Here, the power consumption prediction artificial intelligence model NFT metadata generation unit 140 may calculate the power consumption prediction artificial intelligence model price when generating NFT metadata corresponding to the power consumption prediction artificial intelligence model.

For example, the power consumption prediction artificial intelligence model NFT metadata generator 140 is a power consumption prediction artificial intelligence model based on a multiple linear regression algorithm proportional to the accuracy of the power consumption prediction artificial intelligence model, the number of learning data, and the learning time. price can be calculated.

Here, the multiple linear regression algorithm may generate a multiple linear regression model (power consumption prediction artificial intelligence model NFT price calculation model) that is relearned based on the generated transaction information.

The model is shared by all nodes, and each node can set the NFT price in proportion to the accuracy of the power consumption prediction artificial intelligence model it creates, the number of learning data, and the learning time.

The initial weight value of the model for pricing is set from the blockchain node that created the AI model, and can be relearned whenever a transaction occurs.

Through this, buyers, data providers, and block chain node providers can set an acceptable AI model NFT price to encourage participation and purchase.

For example, assume that when node 1 is an occupied node, the generated power consumption data prediction artificial intelligence model is an LSTM-based model, and when node 2 is an occupied node, the generated power consumption data prediction artificial intelligence model is a GRU-based model. If so, regardless of the type of each model, price is set based only on the accuracy of the generated model, the number of training data, and the learning time (these data are also included in the artificial intelligence model NFT) Multiple linear regression model (power consumption prediction artificial The intelligence model (NFT price calculation model) can be learned and shared by all nodes.

When transaction data occurs in the NFT marketplace, the data of the artificial intelligence model traded (for example, accuracy of 95%, the number of training data is 12,000,150, and the learning time is 5h24m13s) and the sold price (for example, 0.21 ETH) A multiple linear regression model (power consumption prediction artificial intelligence model NFT price calculation model) can be learned using as input and output.

Then, the power consumption prediction artificial intelligence model NFT metadata generation unit 140 generates NFT metadata corresponding to the power consumption prediction artificial intelligence model based on the information of the generated power consumption prediction artificial intelligence model and the learning data information. can

Next, when NFT metadata corresponding to the power consumption prediction artificial intelligence model is generated, the power consumption prediction artificial intelligence model NFT metadata transmission unit 150 transmits the generated NFT metadata to the NFT marketplace through the NFT metadata transmission protocol. can transmit

3 is a diagram for explaining NFT metadata corresponding to the artificial intelligence model for predicting power consumption according to the present disclosure.

As shown in FIG. 3 , the present disclosure may generate NFT metadata 20 corresponding to the power consumption prediction artificial intelligence model based on the information of the power consumption prediction artificial intelligence model and the training data information.

Here, the NFT metadata 20 includes accuracy information 210 of the power consumption prediction artificial intelligence model, weight file 220 of the power consumption prediction artificial intelligence model, NFT price information 230 of the power consumption prediction artificial intelligence model, Training data information of power consumption prediction AI model (240), power consumption data provider cryptocurrency wallet address (250), deep learning server provider cryptocurrency wallet address (260), and revenue of power consumption data provider and deep learning server provider Distribution information 270 may be included.

In addition, the learning data information 240 of the power consumption prediction artificial intelligence model may include a power consumption data collection period 241, power consumption data collection location information 242, and the number of power consumption data 243.

4 and 5 are flowcharts for explaining a blockchain-based power consumption prediction artificial intelligence model transaction method according to the present disclosure.

As shown in FIGS. 4 and 5, the present disclosure may receive a broadcast from a power consumption data provider in the blockchain network 10 (S10).

Subsequently, the present disclosure may select a preempted node from among a plurality of blockchain nodes based on the response signal of the blockchain node to the broadcast (S20).

For example, according to the present disclosure, among a plurality of blockchain nodes, a blockchain node that transmits a response signal to the broadcast of the power consumption data providing unit most quickly may be selected as a preoccupied node.

And, according to the present disclosure, a power consumption prediction artificial intelligence model may be generated through the preemption node 111 (S30).

Next, according to the present disclosure, power consumption data may be received from the power consumption data providing unit through the preoccupied node (S40).

Next, according to the present disclosure, transfer learning of a pre-generated power consumption prediction artificial intelligence model based on power consumption data received through an occupied node may be started.

And, according to the present disclosure, it is possible to check whether an epoch exceeds a first set number of times through the preemption node (S50).

For example, the epoch of the first set number of times may be about the 50th epoch, which is only an example, but is not limited thereto.

Next, in the present disclosure, if the number of epochs exceeds the first set number through the preemption node, the layer (layer) of the power consumption prediction AI model may be frozen and learning may be performed (S60).

Subsequently, the present disclosure may check whether the number of epochs is greater than or equal to the second set number (S70).

For example, the epoch of the second set number of times may be about the 100th epoch, which is only an example, but is not limited thereto.

In addition, in the present disclosure, if the epoch is equal to or greater than the second set number of times, the learning of the power consumption prediction artificial intelligence model is completed, and NFT (Non-Fungible Token) metadata corresponding to the learned power consumption prediction artificial intelligence model can be generated. Yes (S100).

In addition, according to the present disclosure, if the number of epochs is less than the second set number, it may be checked whether the loss value of the verification dataset among the power consumption data is improved during the specific number of epochs (S80).

For example, the specific number of epochs may be about 5 epochs, which is only an example, but is not limited thereto.

Subsequently, in the present disclosure, if there is no improvement in the loss value of the verification dataset, the learning of the power consumption prediction artificial intelligence model is early stopped (S90), and the NFT corresponding to the learned power consumption prediction artificial intelligence model (Non- Fungible Token) metadata can be created (S100).

Here, the present disclosure can calculate the power consumption prediction artificial intelligence model price based on a multiple linear regression algorithm proportional to the accuracy of the power consumption prediction artificial intelligence model, the number of learning data, and the learning time through the preoccupied node.

Next, the present disclosure may output the generated power consumption prediction artificial intelligence model NFT metadata 20 (S110).

Here, the NFT metadata 20 includes accuracy information of the power consumption prediction AI model, weight file of the power consumption prediction AI model, NFT price information of the power consumption prediction AI model, and training data of the power consumption prediction AI model. information, power consumption data provider cryptocurrency wallet address, deep learning server provider cryptocurrency wallet address, and revenue distribution information between the power consumption data provider and the deep learning server provider.

And, in the present disclosure, the power consumption prediction artificial intelligence model NFT metadata may be transmitted and registered to the NFT marketplace unit (S120).

Next, in the present disclosure, it is possible to check whether the power consumption prediction artificial intelligence model registered in the NFT market place unit is purchased through the purchaser of the power consumption prediction artificial intelligence model (S130).

Subsequently, the present disclosure discloses the power consumption data provider and power consumption of the power consumption data providing unit that distributes the purchase price in accordance with the preset revenue distribution information when the power consumption prediction artificial intelligence model of the NFT meta data is purchased and provides the power consumption data. Each payment can be made to the cryptocurrency wallet address of the deep learning server provider of the blockchain node that created the predictive artificial intelligence model (S140).

6 is a diagram for explaining a transaction process of a blockchain-based power consumption prediction artificial intelligence model according to the present disclosure.

As shown in FIG. 6, the present disclosure provides a terminal of a power consumption data provider 30, a blockchain network 10 including a plurality of blockchain nodes 110, a terminal of an NFT marketplace 40, and It may include a terminal of the power consumption prediction artificial intelligence model purchaser 50 .

First, the terminal of the power consumption data provider 30 may transmit a broadcast to a plurality of blockchain nodes 110 included in the blockchain network 10.

In addition, the plurality of blockchain nodes 110 included in the blockchain network 10 may transmit a response signal to the broadcast to the terminal of the power consumption data provider 30.

Subsequently, the terminal of the power consumption data provider 30 may transmit a power consumption prediction artificial intelligence model generation request signal to the preoccupied node 111, which is a blockchain node that has transmitted the fastest response signal.

Here, the blockchain network 10 selects the blockchain node 110 that transmits the response signal to the broadcast of the terminal of the power consumption data provider 30 the fastest among the plurality of blockchain nodes 110 as the preoccupied node ( 111) can be selected.

Next, the blockchain node selected as the preemption node 111 may generate a power consumption prediction artificial intelligence model when receiving a power consumption prediction artificial intelligence model generation request signal from the terminal of the power consumption data provider 30 .

Here, the preemption node 111 may recognize the time when the power consumption prediction artificial intelligence model generation request signal is received as the generation start point of the power consumption prediction artificial intelligence model.

In addition, the generated power consumption prediction artificial intelligence model is a product for performing learning based on power consumption data in the preoccupied node 111, and can serve as an artificial intelligence program itself that helps predict power consumption.

Also, when the preemption node 111 receives the power consumption data from the terminal of the power consumption data provider 30, it can learn the power consumption prediction artificial intelligence model generated based on the power consumption data.

Here, the preemption node 111 may start learning the power consumption prediction artificial intelligence model immediately upon receiving the power consumption data.

For example, the preemption node 111 may recognize the time at which power consumption data is received as the learning start time of the power consumption prediction artificial intelligence model.

In addition, the preemption node 111, when learning the generated power consumption prediction artificial intelligence model, proceeds with transfer learning of the pre-generated power consumption prediction artificial intelligence model based on the received power consumption data, Epoch If the first set number of times is exceeded, the layer of the power consumption prediction artificial intelligence model is unfrozen and learning is performed, and if the epoch is the second set number or more, the power consumption prediction artificial intelligence model learning may be completed.

In addition, if the epoch is less than the second set number of times, the preemption node 111 checks whether the loss value of the verification dataset among the power consumption data is improved during a specific number of epochs, and if there is no improvement in the loss value of the verification dataset, the power consumption Learning of predictive AI models can be early stopped.

Subsequently, the preemption node 111 may generate NFT metadata corresponding to the power consumption prediction artificial intelligence model when the generated power consumption prediction artificial intelligence model is completed learning.

Here, the preemption node 111 may generate NFT metadata corresponding to the power consumption prediction artificial intelligence model based on the information of the generated power consumption prediction artificial intelligence model and the training data information.

Next, when NFT metadata corresponding to the power consumption prediction AI model is generated, the preemption node 111 may transmit the generated NFT metadata to a terminal of the NFT marketplace 40 through an NFT metadata transmission protocol.

And, the terminal of the NFT marketplace 40 may register NFT (Non-Fungible Token) metadata 20 corresponding to the learned power consumption prediction artificial intelligence model.

Subsequently, the terminal of the power consumption prediction artificial intelligence model purchaser 50 may request purchase of the power consumption prediction artificial intelligence model registered in the NFT marketplace 40 and pay cryptocurrency as the purchase price.

Next, when the power consumption prediction artificial intelligence model of the NFT metadata is purchased and the purchase price is generated, the terminal of the NFT marketplace 40 distributes the purchase price in accordance with the preset profit distribution information, and distributes the distributed purchase price to power It is possible to pay each to the terminal of the power consumption data provider 30 that provided the consumption data and the server of the preoccupied node 111, which is a blockchain node that generated the artificial intelligence model for predicting power consumption.

For example, when the terminal of the NFT marketplace 40 distributes the purchase price, when the terminal of the power consumption prediction artificial intelligence model purchaser 50 purchases the power consumption prediction artificial intelligence model and pays the purchase price in cryptocurrency Includes the distribution ratio between the power consumption data provider 30, which provided the power consumption data, and the deep learning server provider of the preoccupied node 111, which is a blockchain node that created the power consumption prediction artificial intelligence model, for the purchase price for cryptocurrency It is distributed according to the revenue distribution information, and the cryptocurrency of the distributed purchase price can be paid to the cryptocurrency wallet address of the power consumption data provider and the cryptocurrency wallet address of the deep learning server provider.

As such, according to the present disclosure, the power consumption prediction artificial intelligence model learned based on power consumption data is registered in the NFT marketplace, and when the registered power consumption prediction artificial intelligence model is purchased, the purchase is made in accordance with preset revenue distribution information. The payment is distributed and paid to the power consumption data provider and the deep learning server provider, respectively, to induce active participation by paying compensation for the resources used to generate the power consumption prediction artificial intelligence model to the power consumption data provider and the deep learning server provider. , it can solve the problem of lack of resources for learning artificial intelligence models, and guarantee the property rights of resources used for learning power consumption prediction artificial intelligence models.

In addition, the present disclosure, when learning a power consumption prediction artificial intelligence model, completes preparation for learning before transmitting big data (power consumption data) through a request for generating a power consumption prediction artificial intelligence model, shortens the learning time, and transfers Unnecessary consumption of computing power can be prevented through learning, automatic learning rate adjustment, and early termination.

In addition, the present disclosure can automatically calculate the model NFT price through a multiple linear regression algorithm in proportion to the accuracy of the generated power consumption prediction artificial intelligence model, the number of learning data, and the learning time, and to provide NFT marketplaces from buyers. Cryptocurrency paid through this is distributed according to the revenue distribution information of the power consumption data provider and the deep learning server provider, and you can receive compensation for the resources used for learning the power consumption prediction artificial intelligence model.

Here, the multiple linear regression algorithm may generate a multiple linear regression model (power consumption prediction artificial intelligence model NFT price calculation model) that is relearned based on the generated transaction information.

The model is shared by all nodes, and each node can set the NFT price in proportion to the accuracy of the power consumption prediction artificial intelligence model it creates, the number of learning data, and the learning time.

The initial weight value of the model for pricing is set from the blockchain node that created the AI model, and can be relearned whenever a transaction occurs.

Through this, buyers, data providers, and block chain node providers can set an acceptable AI model NFT price to encourage participation and purchase.

For example, assume that when node 1 is an occupied node, the generated power consumption data prediction artificial intelligence model is an LSTM-based model, and when node 2 is an occupied node, the generated power consumption data prediction artificial intelligence model is a GRU-based model. If so, regardless of the type of each model, price is set based only on the accuracy of the generated model, the number of training data, and the learning time (these data are also included in the artificial intelligence model NFT) Multiple linear regression model (power consumption prediction artificial The intelligence model (NFT price calculation model) can be learned and shared by all nodes.

When transaction data occurs in the NFT marketplace, the data of the artificial intelligence model traded (for example, accuracy of 95%, the number of training data is 12,000,150, and the learning time is 5h24m13s) and the sold price (for example, 0.21 ETH) A multiple linear regression model (power consumption prediction artificial intelligence model NFT price calculation model) can be learned using as input and output.

In addition, the neural network model, which is the above-described power consumption prediction artificial intelligence model, may be a deep neural network. Throughout this specification, neural network, network function, and neural network may be used interchangeably. A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can reveal latent structures in data. In other words, it can identify the latent structure of a photo, text, video, sound, or music (e.g., what objects are in the photo, what the content and emotion of the text are, what the content and emotion of the audio are, etc.). . Deep neural networks include a convolutional neural network (CNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), and a deep belief network (DBN). , Q network, U network, Siamese network, and the like.

A convolutional neural network is a type of deep neural network and includes a neural network including a convolutional layer. A convolutional neural network is a type of multilayer perceptron designed to use minimal preprocessing. A CNN may be composed of one or several convolutional layers and artificial neural network layers combined therewith. CNNs can additionally utilize weights and pooling layers. Thanks to this structure, CNNs can fully utilize the two-dimensional structure of the input data. Convolutional neural networks can be used to recognize objects in images. A convolutional neural network can represent and process image data as a matrix having dimensions. For example, in the case of image data encoded in RGB (red-green-blue), each R, G, and B color may be represented as a two-dimensional (eg, two-dimensional image) matrix. That is, the color value of each pixel of the image data can be a component of a matrix, and the size of the matrix can be the same as the size of the image. Thus, image data can be represented by three two-dimensional matrices (a three-dimensional data array).

In a convolutional neural network, a convolutional process (input/output of a convolutional layer) can be performed by moving the convolutional filter and multiplying the convolutional filter with matrix components at each position of the image. A convolutional filter may be composed of an n*n matrix. A convolutional filter can consist of a fixed shape filter, which is usually smaller than the total number of pixels in an image. That is, when an m*m image is input to a convolutional layer (for example, a convolutional layer whose size of convolutional filter is n*n), a matrix representing n*n pixels including each pixel of the image It can be component multiplied with this convolutional filter (ie, multiplication of each component of the matrix). A component matching the convolutional filter can be extracted from the image by multiplication with the convolutional filter. For example, a 3*3 convolutional filter to extract top and bottom linear components from an image would be constructed as [[0,1,0], [0,1,0], [0,1,0]] can When a 3*3 convolutional filter for extracting up and down linear components from an image is applied to an input image, up and down linear components matching the convolutional filter may be extracted from the image and output. The convolutional layer may apply a convolutional filter to each matrix for each channel representing an image (ie, R, G, B colors in the case of an R, G, B coded image). The convolutional layer may extract features matching the convolutional filter from the input image by applying the convolutional filter to the input image. The filter value of the convolutional filter (that is, the value of each element of the matrix) may be updated by backpropagation during the learning process of the convolutional neural network.

A subsampling layer is connected to the output of the convolutional layer to simplify the output of the convolutional layer, thereby reducing memory usage and computational complexity. For example, if the output of the convolutional layer is input to a pooling layer with a 2*2 max pooling filter, the image is compressed by outputting the maximum value included in each patch for each 2*2 patch in each pixel of the image. can The aforementioned pooling may be a method of outputting a minimum value of a patch or an average value of a patch, and any pooling method may be included in the present disclosure.

A convolutional neural network may include one or more convolutional layers and subsampling layers. A convolutional neural network may extract features from an image by repeatedly performing a convolutional process and a subsampling process (eg, max pooling described above). Through an iterative convolutional process and subsampling process, a neural network can extract global features of an image.

An output of the convolutional layer or the subsampling layer may be input to a fully connected layer. A fully connected layer is a layer in which all neurons in one layer are connected to all neurons in neighboring layers. A fully connected layer may refer to a structure in which all nodes of each layer are connected to all nodes of other layers in a neural network.

7 is a schematic diagram illustrating a network function of an artificial intelligence model for predicting power consumption according to an embodiment of the present disclosure.

Throughout this specification, the terms computational model, neural network, network function, and neural network may be used interchangeably. A neural network may consist of a set of interconnected computational units, which may generally be referred to as nodes. These nodes may also be referred to as neurons. A neural network includes one or more nodes. Nodes (or neurons) constituting neural networks may be interconnected by one or more links.

In a neural network, one or more nodes connected through a link may form a relative relationship of an input node and an output node. The concept of an input node and an output node is relative, and any node in an output node relationship with one node may have an input node relationship with another node, and vice versa. As described above, an input node to output node relationship may be created around a link. More than one output node can be connected to one input node through a link, and vice versa.

In a relationship between an input node and an output node connected through one link, the value of data of the output node may be determined based on data input to the input node. Here, a link interconnecting an input node and an output node may have a weight. The weight may be variable, and may be changed by a user or an algorithm in order to perform a function desired by the neural network. For example, when one or more input nodes are interconnected by respective links to one output node, the output node is set to a link corresponding to values input to input nodes connected to the output node and respective input nodes. An output node value may be determined based on the weight.

As described above, in the neural network, one or more nodes are interconnected through one or more links to form an input node and output node relationship in the neural network. Characteristics of the neural network may be determined according to the number of nodes and links in the neural network, an association between the nodes and links, and a weight value assigned to each link. For example, when there are two neural networks having the same number of nodes and links and different weight values of the links, the two neural networks may be recognized as different from each other.

A neural network may be composed of a set of one or more nodes. A subset of nodes constituting a neural network may constitute a layer. Some of the nodes constituting the neural network may form one layer based on distances from the first input node. For example, a set of nodes having a distance of n from the first input node may constitute n layers. The distance from the first input node may be defined by the minimum number of links that must be passed through to reach the corresponding node from the first input node. However, the definition of such a layer is arbitrary for explanation, and the order of a layer in a neural network may be defined in a method different from the above. For example, a layer of nodes may be defined by a distance from a final output node.

An initial input node may refer to one or more nodes to which data is directly input without going through a link in relation to other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes among nodes in the neural network. Also, the hidden node may refer to nodes constituting the neural network other than the first input node and the last output node.

In the neural network according to an embodiment of the present disclosure, the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as the number of nodes progresses from the input layer to the hidden layer. can In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes of the input layer may be less than the number of nodes of the output layer and the number of nodes decreases as the number of nodes increases from the input layer to the hidden layer. there is. In addition, the neural network according to another embodiment of the present disclosure is a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as the number of nodes increases from the input layer to the hidden layer. can A neural network according to another embodiment of the present disclosure may be a neural network in the form of a combination of the aforementioned neural networks.

A deep neural network (DNN) may refer to a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can reveal latent structures in data. In other words, it can identify the latent structure of a photo, text, video, sound, or music (e.g., what objects are in the photo, what the content and emotion of the text are, what the content and emotion of the audio are, etc.). . Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted boltzmann machines (RBMs). machine), a deep belief network (DBN), a Q network, a U network, a Siamese network, a Generative Adversarial Network (GAN), and the like. The description of the deep neural network described above is only an example and is not limited thereto.

In one embodiment of the present disclosure, the network function may include an autoencoder. An autoencoder may be a type of artificial neural network for outputting output data similar to input data. An auto-encoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between input and output layers. The number of nodes of each layer may be reduced from the number of nodes of the input layer to an intermediate layer called the bottleneck layer (encoding), and then expanded symmetrically with the reduction from the bottleneck layer to the output layer (symmetrical to the input layer). Autoencoders can perform non-linear dimensionality reduction. The number of input layers and output layers may correspond to dimensions after preprocessing of input data. In the auto-encoder structure, the number of hidden layer nodes included in the encoder may decrease as the distance from the input layer increases. If the number of nodes in the bottleneck layer (the layer with the fewest nodes located between the encoder and decoder) is too small, a sufficient amount of information may not be conveyed, so more than a certain number (e.g., more than half of the input layer, etc.) ) may be maintained.

The neural network may be trained using at least one of supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. Learning of the neural network may be a process of applying knowledge for the neural network to perform a specific operation to the neural network.

A neural network can be trained in a way that minimizes output errors. In the learning of the neural network, the learning data is repeatedly input into the neural network, the output of the neural network for the training data and the error of the target are calculated, and the error of the neural network is transferred from the output layer of the neural network to the input layer in the direction of reducing the error. It is a process of updating the weight of each node of the neural network by backpropagating in the same direction. In the case of teacher learning, the learning data in which the correct answer is labeled is used for each learning data (ie, the labeled learning data), and in the case of comparative teacher learning, the correct answer may not be labeled in each learning data. That is, for example, learning data in the case of teacher learning regarding data classification may be data in which each learning data is labeled with a category. Labeled training data is input to a neural network, and an error may be calculated by comparing an output (category) of the neural network and a label of the training data. As another example, in the case of comparative history learning for data classification, an error may be calculated by comparing input learning data with a neural network output. The calculated error is back-propagated in a reverse direction (ie, from the output layer to the input layer) in the neural network, and the connection weight of each node of each layer of the neural network may be updated according to the back-propagation. The amount of change in the connection weight of each updated node may be determined according to a learning rate. The neural network's computation of input data and backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of iterations of the learning cycle of the neural network. For example, a high learning rate may be used in the early stage of neural network training to increase efficiency by allowing the neural network to quickly obtain a certain level of performance, and a low learning rate may be used in the late stage to increase accuracy.

In neural network learning, generally, training data can be a subset of real data (ie, data to be processed using the trained neural network), and therefore, errors for training data are reduced, but errors for real data are reduced. There may be incremental learning cycles. Overfitting is a phenomenon in which errors for actual data increase due to excessive learning on training data. For example, a phenomenon in which a neural network that has learned a cat by showing a yellow cat does not recognize that it is a cat when it sees a cat other than yellow may be a type of overfitting. Overfitting can act as a cause of increasing the error of machine learning algorithms. Various optimization methods can be used to prevent such overfitting. To prevent overfitting, methods such as increasing the training data, regularization, inactivating some nodes in the network during learning, and using a batch normalization layer should be applied. can

The method according to an embodiment of the present disclosure described above may be implemented as a program (or application) to be executed in combination with a server, which is hardware, and stored in a medium.

The aforementioned program is C, C++, JAVA, machine language, etc. It may include a code coded in a computer language of. These codes may include functional codes related to functions defining necessary functions for executing the methods, and include control codes related to execution procedures necessary for the processor of the computer to execute the functions according to a predetermined procedure. can do. In addition, these codes may further include memory reference related codes for which location (address address) of the computer's internal or external memory should be referenced for additional information or media required for the computer's processor to execute the functions. there is. In addition, when the processor of the computer needs to communicate with any other remote computer or server in order to execute the functions, the code uses the computer's communication module to determine how to communicate with any other remote computer or server. It may further include communication-related codes for whether to communicate, what kind of information or media to transmit/receive during communication, and the like.

The storage medium is not a medium that stores data for a short moment, such as a register, cache, or memory, but a medium that stores data semi-permanently and is readable by a device. Specifically, examples of the storage medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc., but are not limited thereto. That is, the program may be stored in various recording media on various servers accessible by the computer or various recording media on the user's computer. In addition, the medium may be distributed to computer systems connected through a network, and computer readable codes may be stored in a distributed manner.

Steps of a method or algorithm described in connection with an embodiment of the present disclosure may be implemented directly in hardware, implemented in a software module executed by hardware, or a combination thereof. A software module may include random access memory (RAM), read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, hard disk, removable disk, CD-ROM, or It may reside in any form of computer readable recording medium well known in the art to which this disclosure pertains.

Although the embodiments of the present disclosure have been described with reference to the accompanying drawings, those skilled in the art to which the present disclosure belongs will understand that the present disclosure can be implemented in other specific forms without changing the technical spirit or essential features. You will be able to. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

Claims (20)

As a blockchain-based power consumption prediction artificial intelligence model trading system,
a power consumption data providing unit providing power consumption data;
A block-chain network unit including a plurality of block-chain nodes for learning a pre-generated power consumption prediction artificial intelligence model based on the power consumption data;
an NFT marketplace unit registering non-fungible token (NFT) metadata corresponding to the learned power consumption prediction artificial intelligence model; and,
A power consumption prediction artificial intelligence model purchase unit for purchasing a power consumption prediction artificial intelligence model registered in the NFT marketplace unit,
The NFT marketplace unit,
When the power consumption prediction artificial intelligence model of the NFT meta data is purchased and the purchase price is generated, the purchase price is distributed according to the preset revenue distribution information, and the distributed purchase price is used as the power consumption data provided with the power consumption data. Payment is made to the provider and the blockchain node that generated the power consumption prediction artificial intelligence model, respectively;
The power consumption data providing unit,
Sending a broadcast to a plurality of blockchain nodes included in the blockchain network unit,
Provides the power consumption data to an occupied node, which is a blockchain node that has transmitted the fastest response signal among response signals received from the blockchain nodes;
When providing the power consumption data, a power consumption prediction artificial intelligence model generation request signal and the power consumption data are simultaneously provided to an occupied node, which is a blockchain node that transmitted the fastest response signal. Intelligent model trading system. delete delete According to claim 1,
The blockchain network unit,
Including multiple blockchain nodes,
A power consumption prediction artificial intelligence model trading system, characterized in that selecting a blockchain node that transmits a response signal to the broadcast of the power consumption data provider fastest among the plurality of blockchain nodes as an occupied node. According to claim 4,
The preemption node,
generating the power consumption prediction artificial intelligence model when receiving a request signal for generating the power consumption prediction artificial intelligence model from the power consumption data providing unit;
Upon receiving the power consumption data from the power consumption data provider, learning the generated power consumption prediction artificial intelligence model based on the power consumption data;
The power consumption prediction artificial intelligence model trading system, characterized in that for generating NFT meta data corresponding to the power consumption prediction artificial intelligence model when the generated power consumption prediction artificial intelligence model is completed. According to claim 5,
The preemption node,
When generating the power consumption prediction artificial intelligence model, the power consumption prediction artificial intelligence model trading system, characterized in that for generating the power consumption prediction artificial intelligence model at the time of receiving the power consumption prediction artificial intelligence model generation request signal. According to claim 5,
The generated power consumption prediction artificial intelligence model,
As an output for performing learning based on the power consumption data in the preoccupied node, the power consumption prediction artificial intelligence model trading system, characterized in that it plays a role of the artificial intelligence program itself that helps predict power consumption. According to claim 5,
The preemption node,
When learning the generated power consumption prediction artificial intelligence model, the power consumption prediction artificial intelligence model trading system, characterized in that to perform the learning start of the power consumption prediction artificial intelligence model immediately upon receiving the power consumption data. According to claim 5,
The preemption node,
When learning the generated power consumption prediction artificial intelligence model, transfer learning of the pre-generated power consumption prediction artificial intelligence model is performed based on the received power consumption data, and if the epoch exceeds the first set number of times The power consumption prediction artificial intelligence model, characterized in that the learning is performed by releasing the layer (layer) freezing of the power consumption prediction artificial intelligence model, and the learning of the power consumption prediction artificial intelligence model is completed when the epoch is equal to or greater than a second set number of times. Intelligent model trading system. According to claim 9,
The preemption node,
If the epoch is less than the second set number of times, it is checked whether the loss value of the verification dataset among the power consumption data is improved during a specific number of epochs, and if there is no improvement in the loss value of the verification dataset, the power consumption prediction artificial intelligence model A power consumption prediction artificial intelligence model trading system characterized by early stopping learning. According to claim 5,
The preemption node,
When the generated power consumption prediction artificial intelligence model is trained, the power consumption prediction artificial intelligence model trading system, characterized in that for calculating and predicting the accuracy of the learned power consumption prediction artificial intelligence model. According to claim 5,
The preemption node,
When generating NFT metadata corresponding to the power consumption prediction artificial intelligence model, the power consumption prediction artificial intelligence model price is calculated, characterized in that the power consumption prediction artificial intelligence model trading system. According to claim 12,
The preemption node,
When calculating the price of the power consumption prediction artificial intelligence model, the power consumption prediction artificial intelligence model price is calculated based on a multiple linear regression algorithm proportional to the accuracy of the power consumption prediction artificial intelligence model, the number of learning data, and the learning time Power consumption prediction artificial intelligence model trading system, characterized in that. According to claim 5,
The preemption node,
When generating NFT metadata corresponding to the power consumption prediction artificial intelligence model, NFT metadata corresponding to the power consumption prediction artificial intelligence model is generated based on the information of the generated power consumption prediction artificial intelligence model and learning data information. An artificial intelligence model trading system for predicting power consumption, characterized in that for generating. According to claim 5,
The preemption node,
When generating NFT metadata corresponding to the power consumption prediction artificial intelligence model, when NFT metadata corresponding to the power consumption prediction artificial intelligence model is generated, the generated NFT metadata is transmitted through the NFT metadata transmission protocol. Power consumption prediction artificial intelligence model trading system, characterized in that transmitted to the marketplace unit. According to claim 5,
The preemption node,
a power consumption data receiving unit receiving the power consumption data;
Generating the power consumption prediction artificial intelligence model, training the generated power consumption prediction artificial intelligence model based on the power consumption data, and generating a power consumption prediction artificial intelligence model that calculates the accuracy of the power consumption prediction artificial intelligence model wealth;
an NFT metadata generating unit for generating NFT metadata corresponding to the power consumption prediction artificial intelligence model based on the information of the generated power consumption prediction artificial intelligence model and learning data information; and,
Power consumption prediction artificial intelligence model trading system, characterized in that it comprises an NFT metadata transmission unit for transmitting the generated NFT metadata to the NFT marketplace unit through an NFT metadata transmission protocol. According to claim 16,
The power consumption prediction artificial intelligence model generation unit,
a power consumption prediction artificial intelligence model learning unit that performs learning of a power consumption prediction artificial intelligence model generated in advance based on the received power consumption data; and,
Power consumption prediction artificial intelligence model trading system comprising a power consumption prediction artificial intelligence model prediction unit for calculating the accuracy of the generated power consumption prediction artificial intelligence model According to claim 1,
The NFT marketplace unit,
When distributing the purchase price, if the power consumption prediction artificial intelligence model purchasing unit purchases the power consumption prediction artificial intelligence model and pays the purchase price in cryptocurrency, the purchase price for the cryptocurrency provides the power consumption data Distributed according to the revenue distribution information including the distribution ratio between the power consumption data provider of one power consumption data provider and the deep learning server provider of the blockchain node that generated the power consumption prediction artificial intelligence model, and the distribution of the purchased price The power consumption prediction artificial intelligence model trading system, characterized in that the cryptocurrency is paid to the cryptocurrency wallet address of the power consumption data provider and the cryptocurrency wallet address of the deep learning server provider, respectively. In the power consumption prediction artificial intelligence model trading method of a power consumption prediction artificial intelligence model trading system including a power consumption data providing unit, a blockchain network unit having a plurality of blockchain nodes, and an NFT marketplace unit,
Selecting, by the blockchain network unit, a preoccupied node from among the plurality of blockchain nodes based on a response signal of the blockchain node to the broadcast when a broadcast is received from the power consumption data providing unit;
generating, by the preemption node, an artificial intelligence model for predicting power consumption;
When the preemption node receives power consumption data from the power consumption data provider, it learns the power consumption prediction artificial intelligence model based on the power consumption data, and NFTs corresponding to the learned power consumption prediction artificial intelligence model ( Non-Fungible Token) generating metadata and transmitting them to the NFT marketplace unit; and
When the NFT marketplace unit purchases the power consumption prediction artificial intelligence model of the NFT meta data, the power consumption data provider of the power consumption data providing unit distributes the purchase price corresponding to the preset revenue distribution information and provides the power consumption data Paying each to the cryptocurrency wallet address of the deep learning server provider of the blockchain node that generated the power consumption prediction artificial intelligence model,
The power consumption data providing unit,
Transmitting the broadcast to a plurality of blockchain nodes included in the blockchain network unit;
Provides the power consumption data to the preoccupied node, which is a blockchain node that has transmitted the fastest response signal among response signals received from the blockchain nodes;
When providing the power consumption data, a power consumption prediction artificial intelligence model generation request signal and the power consumption data are simultaneously provided to the preoccupied node, which is a blockchain node that transmitted the fastest response signal. Artificial intelligence model trading method. Combined with a computer that is hardware, a power consumption prediction artificial intelligence model transaction method of a power consumption prediction artificial intelligence model transaction system stored in a medium to perform the power consumption prediction artificial intelligence model transaction method of claim 19. A computer program.

Images