A digital data conversion method and system based on artificial intelligence

Abstract

The present invention belongs to the field of digital data conversion technology and discloses a digital data conversion method and system based on artificial intelligence. The method comprises the following steps: building a digital data display platform, deploying a blockchain network, using an artificial intelligence algorithm to construct a digital data conversion engine, and connecting the digital data conversion engine to the digital data display platform; based on the digital data display platform, collecting real-time multimodal data, using the digital data conversion engine to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features; based on the real-time multimodal fusion features, using the digital data conversion engine to perform digital data conversion to obtain real-time digital data, and using the digital data display platform to visualize the real-time digital data. The present invention solves the problems of the prior art, such as single data source, low data processing efficiency, insufficient data security, and poor visualization effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital data conversion, and in particular relates to a digital data conversion method and system based on artificial intelligence. Background Art

[0002] Digital data refers to all types of data generated, collected, and stored in the process of achieving digital immortality. Digital data can record an individual's thoughts, experiences, emotions, and memories in detail, thereby preserving their unique identity in the digital world. For individuals, this data is central to achieving digital immortality, crucial to their existence and continuity in the digital world. Digital data holds immense economic value and can drive the development of related industries, such as digital storage, cloud computing, and virtual reality. It can also spawn entirely new business models and services, such as digital legacy management and virtual companionship. In short, digital data is not only of significant importance at the individual level but also has profound implications across multiple dimensions, including technological, social, cultural, economic, legal, and scientific. However, the accumulation and application of digital data also presents a host of challenges and controversies.

[0003] Existing digital data conversion technology has the following defects:

[0004] 1) Single data source: Existing technologies often focus on feature extraction from single-modal data, such as images or audio, while ignoring the complementarity and correlation between multimodal data. For complex multimodal data, existing feature extraction methods may not fully capture deep and subtle features, resulting in information loss and an inability to fully reproduce an individual's digital identity.

[0005] 2) Inefficient data processing: Existing technologies often consume a lot of time and computing resources when processing large-scale multimodal data, resulting in low data processing efficiency and difficulty in meeting real-time requirements.

[0006] 3) Insufficient data security: Since digital data involves a large amount of private information, existing technologies lack effective security mechanisms for the storage and management of digital data, making it vulnerable to security threats such as data tampering and leakage;

[0007] 4) Poor visualization effects: Existing technologies often lack intuitiveness and interactivity in digital data visualization, and are unable to clearly display digital data, affecting user experience. Summary of the Invention

[0008] In order to solve the problems of single data source, low data processing efficiency, insufficient data security and poor visualization effect in the existing technology, the purpose of the present invention is to provide a digital data conversion method and system based on artificial intelligence.

[0009] The technical solution adopted in the present invention is:

[0010] A digital data conversion method based on artificial intelligence comprises the following steps:

[0011] Build a digital data display platform, deploy a blockchain network, use artificial intelligence algorithms, build a digital data conversion engine, and connect the digital data conversion engine to the digital data display platform;

[0012] Based on the digital data display platform, real-time multimodal data is collected, and the digital data conversion engine is used to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features;

[0013] According to the real-time multimodal fusion characteristics, a digital data conversion engine is used to convert digital data to obtain real-time digital data, and a digital data display platform is used to visualize the real-time digital data.

[0014] Furthermore, the digital data display platform is provided with a multimodal data uploading module, a conversion requirement collection module, and a digital data visualization module;

[0015] The digital data conversion engine is provided with a feature fusion extraction model, a digital portrait generation model, a conversion strategy generation model and a digital data conversion model.

[0016] Furthermore, building a digital data display platform, deploying a blockchain network, using artificial intelligence algorithms, constructing a digital data conversion engine, and connecting the digital data conversion engine to the digital data display platform includes the following steps:

[0017] On the software program platform, a digital data display architecture is constructed, and a multimodal data upload module, a conversion requirement collection module, and a digital data visualization module are constructed to obtain a digital data display platform;

[0018] Connect several data servers as data nodes in a distributed manner, set up the IPFS system and smart contracts, obtain a blockchain network, and connect the blockchain network to the digital data display platform;

[0019] Based on the multimodal data uploading module of the digital data display platform, a number of historical multimodal data are collected, and the number of historical multimodal data are preprocessed to obtain a number of preprocessed historical multimodal data;

[0020] Based on several pre-processed historical multimodal data, a feature fusion extraction model is constructed using multimodal feature fusion and deep learning algorithms, and several historical multimodal fusion features are generated;

[0021] Based on several historical multimodal fusion features, key feature screening and deep learning algorithms are used to build a digital portrait generation model, and generate several historical key multimodal features and corresponding historical digital portraits;

[0022] Based on several historical digital portraits and several preset conversion requirement information, an enhanced reinforcement learning algorithm is used to build a conversion strategy generation model, and several historical conversion strategies and corresponding historical conversion strategy generation experiences are generated;

[0023] Based on several key historical multimodal features, corresponding historical digital portraits, and historical conversion strategies, a generative artificial intelligence algorithm is used to build a digital data conversion model;

[0024] The feature fusion extraction model, digital portrait generation model, conversion strategy generation model and digital data conversion model are integrated to obtain a digital data conversion engine, which is then connected to the digital data display platform.

[0025] Furthermore, the feature fusion extraction model is constructed based on the FPN-DBN-LSTM-Attention algorithm, and the feature fusion extraction model includes an image feature extraction module constructed based on the FPN algorithm, an audio feature extraction module constructed based on the DBN algorithm, a text feature extraction module constructed based on the LSTM algorithm, and a feature fusion module constructed based on the Attention mechanism. The feature fusion module is respectively connected to the image feature extraction module, the audio feature extraction module, and the text feature extraction module.

[0026] Furthermore, the digital portrait generation model is constructed based on the RF-Elman algorithm, and the digital portrait generation model includes a key feature screening module constructed based on the RF algorithm and a digital portrait generation module constructed based on the Elman algorithm, which are connected in sequence.

[0027] Furthermore, the conversion strategy generation model is constructed based on the MPO-MOGRPO algorithm, and the conversion strategy generation model includes a meta-strategy optimization module constructed based on the MPO algorithm and a conversion strategy generation module constructed based on the MOGRPO algorithm. The conversion strategy generation module includes an objective function set, a policy network, an experience replay pool and an intelligent agent. The intelligent agent is connected to the objective function set and the policy network respectively, and the meta-strategy optimization module is connected to the policy network.

[0028] Furthermore, the digital data conversion model is constructed based on the MLP-AIGC algorithm, and the digital data conversion model includes a feature processing module constructed based on the MLP algorithm and a digital data conversion module group constructed based on the AIGC algorithm, which are connected in sequence. The digital data conversion module group includes a text generation module constructed based on the Transformer algorithm, an image generation module constructed based on GAN, an audio generation module constructed based on the WaveNet algorithm, and a video generation module constructed based on VGAN. The feature processing module is respectively connected to the text generation module, the image generation module, the audio generation module, and the video generation module.

[0029] Furthermore, based on the digital data display platform, real-time multimodal data is collected, and a digital data conversion engine is used to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features, including the following steps:

[0030] Based on the digital data display platform, real-time multimodal data and real-time conversion requirement information are collected, and the real-time multimodal data is preprocessed to obtain preprocessed real-time multimodal data;

[0031] Use the feature fusion extraction model of the digital data conversion engine to extract real-time multimodal features from pre-processed real-time multimodal data;

[0032] The real-time multimodal features are fused to obtain the real-time multimodal fusion features of the preprocessed real-time multimodal data.

[0033] Furthermore, based on the real-time multimodal fusion feature, a digital data conversion engine is used to convert digital data to obtain real-time digital data, and a digital data display platform is used to visualize the real-time digital data, including the following steps:

[0034] Using the digital portrait generation model of the digital data conversion engine, digital portrait generation is performed based on real-time multimodal fusion features to obtain real-time key multimodal features and corresponding real-time digital portraits;

[0035] According to the real-time digital portrait and real-time conversion requirement information, a conversion strategy generation model of the digital data conversion engine is used to generate a conversion strategy to obtain a real-time conversion strategy;

[0036] According to the real-time key multimodal features, real-time digital portraits and real-time conversion strategies, the digital data conversion model of the digital data conversion engine is used to convert digital data to obtain real-time digital data;

[0037] Use the blockchain network to distribute and store real-time digital data, and use the digital data display platform to visualize real-time digital data.

[0038] A digital data conversion system based on artificial intelligence is used to implement a digital data conversion method. The system includes a system initialization unit, a feature fusion extraction unit and a digital data conversion unit connected in sequence.

[0039] The beneficial effects of the present invention are:

[0040] The present invention provides an artificial intelligence-based digital data conversion method and system. By collecting real-time multimodal data (including images, audio, video, text, etc.) and using a digital data conversion engine to perform feature fusion extraction, the method can fully capture deep and subtle features, thereby comprehensively and accurately reproducing the digital identity of an individual and avoiding the problem of information loss caused by single-modal data. The digital data conversion engine is constructed using artificial intelligence algorithms, which significantly improves the efficiency of data processing and can process large-scale multimodal data in a short time, meeting real-time requirements and being suitable for scenarios such as real-time interaction and dynamic display. By deploying a blockchain network, the distributed storage of real-time digital data is realized, the security of digital data is enhanced, and effective prevention is achieved. It prevents security threats such as data tampering and leakage, and protects users' privacy information; uses a digital data display platform to visualize real-time digital data, provides an intuitive and interactive display method, enables users to clearly understand and operate digital data, and significantly improves user experience; dynamically generates real-time conversion strategies based on real-time multimodal fusion characteristics and real-time conversion requirement information, and uses digital data conversion models to perform flexible digital data conversion to adapt to diverse application scenarios and needs; it is not only suitable for scenarios such as virtual image reconstruction and digital memorial construction, but can also be widely used in cultural heritage, education, entertainment and other fields. It meets various interactive needs through realistic digital reproduction and promotes the development and application of digital immortality technology.

[0041] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of the digital data conversion method based on artificial intelligence in the present invention.

[0043] Figure 2 It is a structural block diagram of the digital data conversion system based on artificial intelligence in the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments.

[0045] Example 1:

[0046] like Figure 1 As shown, this embodiment provides a digital data conversion method based on artificial intelligence, comprising the following steps:

[0047] S1: Build a digital data display platform, deploy a blockchain network, use artificial intelligence algorithms, build a digital data conversion engine, and connect the digital data conversion engine to the digital data display platform;

[0048] The digital data display platform is equipped with a multimodal data upload module, a conversion requirement collection module, and a digital data visualization module;

[0049] The multimodal data upload module is used to collect real-time multimodal data from users after the user terminal is connected to the digital data display platform. Real-time multimodal data includes real-time image data, real-time video data, real-time text data, and real-time audio data.

[0050] The conversion requirement collection module is used to collect the user's real-time conversion requirement information; the real-time conversion requirement information includes the user's requirements for the scene, focus, characters, and style of the real-time digital data;

[0051] Digital data visualization module, used to visualize real-time digital data, improving the intuitiveness and convenience of users viewing and checking real-time digital data;

[0052] The digital data conversion engine is provided with a feature fusion extraction model, a digital portrait generation model, a conversion strategy generation model and a digital data conversion model;

[0053] A feature fusion extraction model is used to extract real-time multimodal features of the preprocessed real-time multimodal data and perform feature fusion on the real-time multimodal features to obtain real-time multimodal fusion features of the preprocessed real-time multimodal data;

[0054] A digital portrait generation model is used to generate digital portraits based on real-time multimodal fusion features, obtaining real-time key multimodal features and corresponding real-time digital portraits;

[0055] A conversion strategy generation model is used to generate a conversion strategy based on the real-time digital portrait and the real-time conversion requirement information using the conversion strategy generation model of the digital data conversion engine to obtain a real-time conversion strategy;

[0056] A digital data conversion model is used to convert digital data using the digital data conversion model of the digital data conversion engine according to real-time key multimodal features, real-time digital portraits, and real-time conversion strategies to obtain real-time digital data;

[0057] Building a digital data display platform, deploying a blockchain network, using artificial intelligence algorithms, building a digital data conversion engine, and connecting the digital data conversion engine to the digital data display platform includes the following steps:

[0058] S1-1: On the software platform, a digital data display architecture is constructed, and a multimodal data upload module, a conversion requirement collection module, and a digital data visualization module are constructed to obtain a digital data display platform;

[0059] S1-2: Distribute and connect several data servers as data nodes, set up the InterPlanetary File System (IPFS) system and smart contracts, obtain a blockchain network, and connect the blockchain network to the digital data display platform;

[0060] Blockchain network, which is used for distributed storage of real-time digital data, improving the reliability and security of real-time digital data storage;

[0061] S1-3: Based on the multimodal data uploading module of the digital data display platform, a number of historical multimodal data are collected and preprocessed to obtain a number of preprocessed historical multimodal data;

[0062] S1-4: Based on some pre-processed historical multimodal data, use multimodal feature fusion and deep learning algorithms to build a feature fusion extraction model and generate some historical multimodal fusion features;

[0063] The feature fusion extraction model is built based on the Feature Pyramid Networks (FPN)-Deep Belief Nets (DBN)-Long Short-Term Memory (LSTM)-Attention algorithm. It includes an image feature extraction module based on the FPN algorithm, an audio feature extraction module based on the DBN algorithm, a text feature extraction module based on the LSTM algorithm, and a feature fusion module based on the Attention mechanism. The feature fusion module is connected to the image feature extraction module, the audio feature extraction module, and the text feature extraction module, respectively.

[0064] The image feature extraction module combines the top-down and bottom-up paths of the FPN algorithm to extract image features at different scales, capturing multi-level and multi-scale information in the image, such as edges, textures, and shapes. The audio feature extraction module uses the DBN algorithm to perform deep feature learning on audio data, extracting implicit features in the audio, such as frequency, rhythm, pitch, and timbre, to achieve data dimensionality reduction and feature selection, reduce redundant information, and improve the efficiency of subsequent processing. The text feature extraction module uses in-depth modeling of text sequences, and LSTM can better understand the semantic information of the text and improve the semantic expression ability of text features. The feature fusion module uses the Attention mechanism to perform weighted fusion of features from images, audio, and text, highlighting important information and suppressing irrelevant information.

[0065] S1-5: Based on several historical multimodal fusion features, use key feature screening and deep learning algorithms to build a digital portrait generation model, and generate several historical key multimodal features and corresponding historical digital portraits;

[0066] The digital portrait generation model is constructed based on the Random Forest (RF)-Elman algorithm, and the digital portrait generation model includes a key feature screening module constructed based on the RF algorithm and a digital portrait generation module constructed based on the Elman algorithm, which are connected in sequence;

[0067] The key feature screening module uses the random forest (RF) algorithm to evaluate and screen the importance of fused multimodal features, identifying the most critical and representative feature subset for digital portrait generation from a large number of features. The digital portrait generation module uses the Elman algorithm to model the filtered key features and generate digital portraits, capturing the temporal dependencies between features and generating coherent and consistent digital portraits, thus achieving efficient and accurate conversion of multimodal data into digital portraits.

[0068] S1-6: Based on several historical digital portraits and several preset conversion requirement information, use the enhanced reinforcement learning algorithm to build a conversion strategy generation model, and generate several historical conversion strategies and corresponding historical conversion strategy generation experience;

[0069] The conversion strategy generation model is built based on the Meta-Policy Optimization (MPO)-Multi-Objective Group Relative Policy Optimization (MOGRPO) algorithm. It includes a meta-policy optimization module based on the MPO algorithm and a conversion strategy generation module based on the MOGRPO algorithm. The conversion strategy generation module includes an objective function set, a policy network, an experience replay pool, and an intelligent agent. The intelligent agent is connected to the objective function set and the policy network, respectively, and the meta-policy optimization module is connected to the policy network.

[0070] The meta-policy optimization module is used to convert the network parameters of the policy network in the policy generation module so that these parameters can quickly adapt to new and unseen digital portraits, thereby improving the generalization ability of the model. Even in the case of unseen digital portraits, the policy network can be updated based on previous learning experience, thereby improving the adaptability of the conversion policy generation model. The objective function set of the conversion policy generation module can handle multiple conflicting optimization objectives, such as minimizing the digital data conversion time, minimizing the digital data conversion calculation amount, maximizing the digital data conversion efficiency, etc., and generate conversion strategies that balance these objectives. The intelligent agent learns historical conversion strategies through the experience replay pool and continuously optimizes its own policy generation capabilities. The intelligent agent controls the policy network based on the learned experience to generate more effective conversion strategies. The design of the experience replay pool and the intelligent agent enables the model to continuously learn and optimize, thereby improving the quality of policy generation. The conversion policy generation module adopts a group exploration method, which can avoid falling into local optimal solutions to a certain extent. The policy network outputs the distribution probability of actions under a given state. The conversion policy generation module directly updates the policy network through gradients, eliminating the Critic model in traditional reinforcement learning, making the algorithm structure more concise.

[0071] Based on several historical digital portraits and several preset conversion requirements, an enhanced reinforcement learning algorithm is used to build a conversion strategy generation model, and several historical conversion strategies and corresponding historical conversion strategy generation experiences are generated, including the following steps:

[0072] S1-6-1: Use the MPO-MOGRPO algorithm to build an initial conversion strategy generation model; the initial conversion strategy generation model includes an initial meta-strategy optimization module and an initial conversion strategy generation module;

[0073] S1-6-2: The optimization goal of digital data conversion is used as the scenario for meta-strategy optimization. Based on the analysis results of several historical data, the initial meta-strategy optimization module is trained to obtain the final meta-strategy optimization module.

[0074] S1-6-3: Use the final meta-strategy optimization module to initialize the policy network of the initial conversion strategy generation module under different scenarios to obtain the initialized policy network;

[0075] S1-6-4: According to the different scenarios of the meta-strategy optimization module, that is, the optimization goal of digital data conversion, set the objective function set for the initial conversion strategy generation module with the initialized policy network, set the experience replay pool, and set the action space and state space for the agent of the initial conversion strategy generation module;

[0076] S1-6-5: Use the transition strategy generation problem as a simulation environment, and obtain an optimized transition strategy generation module based on the initialized policy network and the agent with action space and state space.

[0077] S1-6-6: Traverse all objective functions in the objective function set, optimize and train the optimized conversion strategy generation module based on a number of historical digital portraits and a number of preset conversion requirement information, obtain the final conversion strategy generation module, and generate a number of historical conversion strategy generation experiences;

[0078] S1-6-7: Integrate the final meta-strategy optimization module and the final conversion strategy generation module to obtain the final conversion strategy generation model, and store several historical conversion strategy generation experiences in the experience replay pool;

[0079] S1-7: Based on several key historical multimodal features, corresponding historical digital portraits, and historical conversion strategies, use generative artificial intelligence algorithms to build a digital data conversion model;

[0080] The digital data conversion model is constructed based on the Multi-Layer Perceptron (MLP)-Artificial Intelligence Generated Content (AIGC) algorithm, and includes a feature processing module constructed based on the MLP algorithm and a digital data conversion module group constructed based on the AIGC algorithm, which are connected in sequence. The digital data conversion module group includes a text generation module constructed based on the Transformer algorithm, an image generation module constructed based on Generative Adversarial Networks (GAN), an audio generation module constructed based on the Wavelet Transformation Network WaveNet algorithm, and a video generation module constructed based on the Video Generative Adversarial Networks (VGAN). The feature processing module is connected to the text generation module, image generation module, audio generation module, and video generation module, respectively.

[0081] AIGC algorithm refers to an algorithm that uses artificial intelligence technology to generate content. This type of algorithm generally includes natural language processing, deep learning, generative adversarial networks, variational autoencoders and other technologies, and is used to create text, images, audio, video and other forms of generative digital data; the feature processing module processes the historical key multimodal features and the features of the corresponding historical digital portraits to obtain processed historical key multimodal features and processed historical digital portrait features; according to the historical conversion strategy, the corresponding text generation module, image generation module, audio generation module and / or video generation module in the digital data conversion module group are scheduled to perform digital data conversion based on the processed historical key multimodal features and the processed historical digital portrait features to obtain the corresponding historical digital data;

[0082] S1-8: Integrate the feature fusion extraction model, the digital portrait generation model, the conversion strategy generation model, and the digital data conversion model to obtain a digital data conversion engine, and connect it to the digital data display platform;

[0083] S2: Based on the digital data display platform, real-time multimodal data is collected and the digital data conversion engine is used to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features. The steps include:

[0084] S2-1: Based on the digital data display platform, collect real-time multimodal data and real-time conversion requirement information, and preprocess the real-time multimodal data to obtain preprocessed real-time multimodal data;

[0085] Perform frame interception processing on video data to obtain corresponding frames of image data, perform denoising and enhancement processing on the frame image data and the original image data to obtain pre-processed real-time image data; perform format conversion and noise reduction processing on the audio to obtain pre-processed real-time audio data; perform magnitude normalization and digitization processing on the text data to obtain pre-processed real-time text data; improve data quality and provide data support for subsequent multimodal feature fusion extraction;

[0086] The preprocessed real-time multimodal data includes preprocessed real-time image data, preprocessed real-time text data, and preprocessed real-time audio data;

[0087] S2-2: Using the feature fusion extraction model of the digital data conversion engine to extract real-time multimodal features from the preprocessed real-time multimodal data, including the following steps:

[0088] S2-2-1: Input the pre-processed real-time multimodal data into the feature fusion extraction model of the digital data conversion engine;

[0089] S2-2-2: using an image feature extraction module of a feature fusion extraction model to extract real-time image features of the preprocessed real-time image data from the preprocessed real-time multimodal data;

[0090] S2-2-3: Using the audio feature extraction module of the feature fusion extraction model, extracting real-time audio features of the preprocessed real-time audio data in the preprocessed real-time multimodal data;

[0091] S2-2-3: Using the text feature extraction module of the feature fusion extraction model, extracting real-time text features of the preprocessed real-time text data in the preprocessed real-time multimodal data;

[0092] S2-2-3: Integrate real-time image features, real-time audio features, and real-time text features to obtain real-time multimodal features;

[0093] S2-3: Based on the preset attention weight value, the feature fusion module of the feature fusion extraction model is used to fuse the real-time multimodal features to obtain the real-time multimodal fusion features of the preprocessed real-time multimodal data;

[0094] S3: Based on the real-time multimodal fusion features, a digital data conversion engine is used to convert digital data to obtain real-time digital data, and the real-time digital data is visualized using a digital data display platform, including the following steps:

[0095] S3-1: Using the digital portrait generation model of the digital data conversion engine, digital portrait generation is performed based on real-time multimodal fusion features to obtain real-time key multimodal features and corresponding real-time digital portraits, including the following steps:

[0096] S3-1-1: Input the real-time multimodal fusion features into the digital portrait generation model of the digital data conversion engine;

[0097] S3-1-2: Use the key feature screening module of the digital portrait generation model to screen the real-time key multimodal features of the real-time multimodal fusion features;

[0098] Real-time key multimodal features include action preference features, behavior features, body features, appearance features, and scene features in real-time image features; intonation features, language features, speech speed features, and timbre features in real-time audio features; and writing habit features and text preference features in real-time text features.

[0099] S3-1-3: Based on the real-time key multimodal features, the digital portrait generation module of the digital portrait generation model is used to generate a digital portrait to obtain a real-time digital portrait;

[0100] Real-time digital portraits include action preference tags, behavior tags, body tags, appearance tags, scene tags, etc. for characters; intonation tags, language tags, speech speed tags, timbre tags, etc. for audio; and writing habit tags and text preference tags for text. Real-time digital portraits are used to describe and characterize the characteristics of digital data.

[0101] S3-2: Based on the real-time digital profile and the real-time conversion requirement information, a conversion strategy generation model of the digital data conversion engine is used to generate a conversion strategy to obtain a real-time conversion strategy, including the following steps:

[0102] S3-2-1: Based on the real-time digital portrait, the meta-strategy optimization module of the conversion strategy model is used to update the policy network of the conversion strategy generation module to obtain an updated policy network;

[0103] S3-2-2: Randomly extract several historical transition strategy generation experiences from the experience replay pool, generate several possible transition actions based on these historical transition strategy generation experiences, update the action space of the agent in the transition strategy generation module, and obtain an updated action space;

[0104] S3-2-3: Analyze the real-time digital portrait and real-time conversion requirement information to obtain several real-time states, and update the state space of the intelligent agent of the conversion strategy generation module according to the real-time states to obtain an updated state space;

[0105] S3-2-4: Select a real-time objective function from the objective function set of the transition strategy generation module, and based on the real-time objective function, use the agent to control the updated policy network to generate a probability distribution of all possible transition actions in the updated action space corresponding to each real-time state in the updated state space;

[0106] S3-2-5: The possible transition action with the highest probability distribution in the updated action space is used as the execution transition action of the corresponding real-time state;

[0107] S3-2-6: Integrate the execution transition actions of all real-time states in the updated state space to obtain the real-time transition strategy;

[0108] The real-time conversion strategy includes real-time module call decisions of the text generation module, image generation module, audio generation module and / or video generation module in the digital data conversion module group, real-time digital data conversion decisions of each module, real-time digital data conversion content decisions, real-time digital data combination decisions, etc.;

[0109] S3-3: Based on the real-time key multimodal features, real-time digital portraits, and real-time conversion strategies, the digital data conversion model of the digital data conversion engine is used to perform digital data conversion to obtain real-time digital data, including the following steps:

[0110] S3-3-1: Input the real-time key multimodal features, real-time digital profiles, and real-time conversion strategies into the digital data conversion engine’s digital data conversion model;

[0111] S3-3-2: Calling the text generation module, image generation module, audio generation module and / or video generation module in the digital data conversion module group according to the real-time module call decision of the real-time conversion strategy;

[0112] S3-3-3: Input the real-time key multimodal features and the real-time digital portrait into the text generation module, the image generation module, the audio generation module and / or the video generation module according to the real-time digital data conversion decision of the real-time conversion strategy;

[0113] S3-3-3: Based on the real-time digital data conversion content decision of the real-time conversion strategy, use the text generation module, the image generation module, the audio generation module, and / or the video generation module to generate corresponding portions of real-time digital data according to the real-time key multimodal features and the real-time digital portrait;

[0114] The partial real-time digital data includes real-time generated text data generated by a text generation module, real-time generated image data by an image generation module, real-time generated audio data by an audio generation module, and / or real-time generated video data by a video generation module;

[0115] S3-3-4: combining several parts of real-time digital data according to the real-time digital data combination decision of the real-time conversion strategy to obtain real-time digital data;

[0116] In the virtual image reconstruction application, the user's virtual image is obtained by converting multimodal data such as historical photos, audio and video, and text records into digital data; in the digital memorial hall construction application, the digital memorial hall of historical figures is constructed by converting multimodal data such as historical photos, audio and video, and text records into digital data; in the cultural heritage application, the original appearance of cultural heritage is reproduced by converting relevant multimodal data of cultural heritage, such as cultural relics photos, historical audio and documentary records into digital data;

[0117] S3-4: Use the blockchain network to distribute and store real-time digital data, and use the digital data display platform to visualize the real-time digital data, including the following steps:

[0118] S3-4-1: Store the real-time digital data into the IPFS system of the blockchain network and return the corresponding real-time digital data hash value;

[0119] S3-4-2: Call the smart contract of the blockchain network to generate real-time transaction data based on the real-time digital data hash value, timestamp, etc.

[0120] S3-4-3: Use several data nodes connected in a distributed manner on the blockchain network to reach consensus on real-time transaction data.

[0121] Example 2:

[0122] like Figure 2 As shown, this embodiment provides an artificial intelligence-based digital data conversion system for implementing a digital data conversion method. The system includes a system initialization unit, a feature fusion extraction unit, and a digital data conversion unit connected in sequence;

[0123] A system initialization unit, which is used to build a digital data display platform, deploy a blockchain network, use artificial intelligence algorithms, build a digital data conversion engine, and connect the digital data conversion engine to the digital data display platform;

[0124] A feature fusion extraction unit is used to collect real-time multimodal data based on the digital data display platform, and use the digital data conversion engine to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features;

[0125] The digital data conversion unit is used to perform digital data conversion using a digital data conversion engine based on real-time multimodal fusion features to obtain real-time digital data, and to visualize the real-time digital data using a digital data display platform.

[0126] The present invention provides an artificial intelligence-based digital data conversion method and system. By collecting real-time multimodal data (including images, audio, video, text, etc.) and using a digital data conversion engine to perform feature fusion extraction, the method can fully capture deep and subtle features, thereby comprehensively and accurately reproducing the digital identity of an individual and avoiding the problem of information loss caused by single-modal data. The digital data conversion engine is constructed using artificial intelligence algorithms, which significantly improves the efficiency of data processing and can process large-scale multimodal data in a short time, meeting real-time requirements and being suitable for scenarios such as real-time interaction and dynamic display. By deploying a blockchain network, the distributed storage of real-time digital data is realized, the security of digital data is enhanced, and effective prevention is achieved. It prevents security threats such as data tampering and leakage, and protects users' privacy information; uses a digital data display platform to visualize real-time digital data, provides an intuitive and interactive display method, enables users to clearly understand and operate digital data, and significantly improves user experience; dynamically generates real-time conversion strategies based on real-time multimodal fusion characteristics and real-time conversion requirement information, and uses digital data conversion models to perform flexible digital data conversion to adapt to diverse application scenarios and needs; it is not only suitable for scenarios such as virtual image reconstruction and digital memorial construction, but can also be widely used in cultural heritage, education, entertainment and other fields. It meets various interactive needs through realistic digital reproduction and promotes the development and application of digital immortality technology.

[0127] The present invention is not limited to the above optional embodiments. Anyone can derive various other forms of products based on the teachings of the present invention. The above specific embodiments should not be construed as limiting the scope of protection of the present invention. The scope of protection of the present invention shall be based on the scope defined in the claims, and the description can be used to interpret the claims.

Claims

1. A digital data conversion method based on artificial intelligence, characterized in that: The steps include: On the software program platform, a digital data display architecture is constructed, and a multimodal data upload module, a conversion requirement collection module, and a digital data visualization module are constructed to obtain a digital data display platform; Connect several data servers as data nodes in a distributed manner, set up the IPFS system and smart contracts, obtain a blockchain network, and connect the blockchain network to the digital data display platform; Based on the multimodal data uploading module of the digital data display platform, a number of historical multimodal data are collected, and the number of historical multimodal data are preprocessed to obtain a number of preprocessed historical multimodal data; Based on several pre-processed historical multimodal data, a feature fusion extraction model is constructed using multimodal feature fusion and deep learning algorithms, and several historical multimodal fusion features are generated; The feature fusion extraction model includes an image feature extraction module based on the FPN algorithm, an audio feature extraction module based on the DBN algorithm, a text feature extraction module based on the LSTM algorithm, and a feature fusion module based on the Attention mechanism. Based on several historical multimodal fusion features, key feature screening and deep learning algorithms are used to build a digital portrait generation model, and generate several historical key multimodal features and corresponding historical digital portraits; The digital portrait generation model includes a key feature screening module based on the RF algorithm and a digital portrait generation module based on the Elman algorithm, which are connected in sequence; Based on several historical digital portraits and several preset conversion requirement information, an enhanced reinforcement learning algorithm is used to build a conversion strategy generation model, and several historical conversion strategies and corresponding historical conversion strategy generation experiences are generated; The conversion strategy generation model includes a meta-strategy optimization module based on the MPO algorithm and a conversion strategy generation module based on the MOGRPO algorithm; Based on several key historical multimodal features, corresponding historical digital portraits, and historical conversion strategies, a generative artificial intelligence algorithm is used to build a digital data conversion model; The digital data conversion model includes a feature processing module based on the MLP algorithm and a digital data conversion module group based on the AIGC algorithm, which are connected in sequence. The digital data conversion module group includes a text generation module based on the Transformer algorithm, an image generation module based on the GAN, an audio generation module based on the WaveNet algorithm, and a video generation module based on the VGAN. Integrate the feature fusion extraction model, digital portrait generation model, conversion strategy generation model and digital data conversion model to obtain a digital data conversion engine, and connect it to the digital data display platform; Based on the digital data display platform, real-time multimodal data is collected, and the digital data conversion engine is used to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features; According to the real-time multimodal fusion characteristics, a digital data conversion engine is used to convert digital data to obtain real-time digital data, and a digital data display platform is used to visualize the real-time digital data.

2. The artificial intelligence-based digital data conversion method according to claim 1, characterized in that: The digital data display platform is provided with a multimodal data uploading module, a conversion requirement collection module and a digital data visualization module; The digital data conversion engine is provided with a feature fusion extraction model, a digital portrait generation model, a conversion strategy generation model and a digital data conversion model.

3. The method for digital data conversion based on artificial intelligence according to claim 2, characterized in that: Based on the digital data display platform, real-time multimodal data is collected, and the digital data conversion engine is used to perform feature fusion extraction on the real-time multimodal data to obtain real-time multimodal fusion features, including the following steps: Based on the digital data display platform, real-time multimodal data and real-time conversion requirement information are collected, and the real-time multimodal data is preprocessed to obtain preprocessed real-time multimodal data; Use the feature fusion extraction model of the digital data conversion engine to extract real-time multimodal features from pre-processed real-time multimodal data; The real-time multimodal features are fused to obtain the real-time multimodal fusion features of the preprocessed real-time multimodal data.

4. The artificial intelligence-based digital data conversion method according to claim 3, characterized in that: Based on the real-time multimodal fusion characteristics, a digital data conversion engine is used to convert digital data to obtain real-time digital data, and a digital data display platform is used to visualize the real-time digital data, including the following steps: Using the digital portrait generation model of the digital data conversion engine, digital portrait generation is performed based on real-time multimodal fusion features to obtain real-time key multimodal features and corresponding real-time digital portraits; According to the real-time digital portrait and real-time conversion requirement information, a conversion strategy generation model of the digital data conversion engine is used to generate a conversion strategy to obtain a real-time conversion strategy; According to the real-time key multimodal features, real-time digital portraits and real-time conversion strategies, the digital data conversion model of the digital data conversion engine is used to convert digital data to obtain real-time digital data; Use the blockchain network to distribute and store real-time digital data, and use the digital data display platform to visualize real-time digital data.

5. An artificial intelligence-based digital data conversion system for implementing the digital data conversion method according to any one of claims 1 to 4, characterized in that: The system comprises a system initialization unit, a feature fusion extraction unit and a digital data conversion unit which are connected in sequence.

Images