AI broadcast interaction platform and method

Abstract

The invention belongs to the technical field of nuclear power artificial intelligence, and particularly relates to an AI broadcast interaction platform and method. Comprising a broadcast control system, a broadcast task decomposition module and a semantic understanding module, the broadcast control system supports docking of broadcast devices of different brands through an adapter, and the broadcast task decomposition module decomposes a task and distributes the task to a plurality of broadcast devices for execution; the semantic understanding module generates a broadcast task from input content, and the broadcast data analysis module analyzes abnormal data and execution frequency of the broadcast task and proposes a workflow optimization suggestion. The platform has the beneficial effects that the execution efficiency and reliability of the broadcast task are improved, multiple access modes are supported, and meanwhile, an intelligent voice recognition function is provided to improve the automation level of the broadcast task.

Description

Technical Field

[0001] This invention belongs to the field of nuclear power artificial intelligence technology, specifically relating to an AI broadcast interaction platform and method. Background Technology

[0002] Most existing nuclear power plants already have wired public address systems. These systems rely on human operators at each plant location to answer phone calls requesting broadcasts, who then broadcast the message to the designated area. This method has the following problems: 1. Inefficiency: Broadcasts cannot be made while a human is answering the phone; conversely, phone calls cannot be answered while a broadcast is being made. 2. Time constraints: It cannot work 24 hours a day without interruption, making it difficult to meet the broadcasting needs for overtime work during major repairs; 3. Location Restriction: Broadcasts can only be initiated at the broadcasting station; broadcasts cannot be initiated from other locations. 4. No record of broadcasts: There is no record of how many broadcasts were sent or to whom they were broadcast. In special circumstances, one can only rely on human memory to recall the information. During the busy overhaul period, this manual broadcasting method was unable to meet the heavy broadcasting demand, affecting overall work efficiency.

[0003] A single day of normal operation of a nuclear power unit generates millions or even tens of millions of yuan in economic benefits. Improving the efficiency of nuclear power plant overhauls can not only reduce overhaul costs but also enable the unit to return to normal operation sooner. Summary of the Invention

[0004] The purpose of this invention is to provide an AI broadcast interaction platform and method that can connect to broadcast systems of different brands and support task decomposition, adapter hot-swappable, semantic understanding, and data analysis and optimization.

[0005] The technical solution of the present invention is as follows: An AI broadcast interaction platform, comprising a broadcast control system, a broadcast task decomposition module, and a semantic understanding module. The broadcast control system supports connection to broadcast devices of different brands through an adapter. The broadcast task decomposition module decomposes tasks and distributes them to multiple broadcast devices for execution. The semantic understanding module generates broadcast tasks from input content. The broadcast data analysis module analyzes abnormal data and execution frequency of broadcast tasks and proposes workflow optimization suggestions.

[0006] The broadcast control system includes a broadcast controller and multiple broadcast adapters. The broadcast controller is connected to multiple broadcast adapters, and each broadcast adapter is connected to a broadcast device. The broadcast controller provides multiple broadcast interfaces, which can access broadcast tasks from telephone systems, SMS systems, web pages, APPs, and mini-programs.

[0007] The broadcast control system includes an adapter manager and a task scheduling module. The adapter manager automatically detects the currently connected broadcast devices and assigns them corresponding broadcast adapters. The task scheduling module is responsible for the decomposition and distribution of broadcast tasks to ensure that different broadcast devices can execute tasks synchronously.

[0008] The semantic understanding module generates broadcast tasks directly through voice translation and natural language processing when a user makes a phone call and engages in intelligent voice dialogue to input the broadcast task.

[0009] When a user sends a text message, the semantic understanding module extracts key information about the broadcast task and generates the broadcast task.

[0010] The semantic understanding module generates broadcast tasks when users directly input the content and execution conditions of the broadcast task through web pages, apps, or mini-programs.

[0011] The broadcast data analysis module analyzes the speech content using NLP algorithms, extracts key information such as broadcast time, location, and audience, and generates an executable broadcast task.

[0012] In the aforementioned broadcast task decomposition module, the broadcast task is decomposed into multiple sub-tasks, which are executed independently by different brand broadcast systems according to their own conditions.

[0013] An AI broadcast interaction method includes the following steps: Collect conversation data: Collect user conversation data, including voice and text; Data annotation: The collected data is annotated, including the intent, entities, and relationships in the dialogue; Training the model: The model is trained using labeled data. The trained model includes a speech recognition model, a semantic understanding model, and a dialogue generation model. Test model: Test the accuracy and efficiency of the model; Deploying the model: Deploying the trained model into the system to enable intelligent dialogue functionality; Broadcast data analysis: Records the execution status of broadcast tasks in real time and automatically generates broadcast statistical reports.

[0014] The beneficial effects of this invention are: the platform improves the execution efficiency and reliability of broadcast tasks, supports multiple access methods, and provides intelligent voice recognition function to improve the automation level of broadcast tasks. Attached Figure Description

[0015] Figure 1 This invention provides an architecture diagram of an AI broadcast interaction platform. Figure 2 This is a schematic diagram of a broadcast control system; Figure 3 A schematic diagram of the broadcast task decomposition module; Figure 4 Flowchart for intelligent semantic understanding of text messages; Figure 5 Flowchart for voice translation of broadcast content; Figure 6 This is a schematic diagram of the broadcast adapter mode; Figure 7 A flowchart for intelligent voice interaction on a telephone. Detailed Implementation

[0016] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1 As shown, an AI broadcast interaction platform includes a broadcast control system, a broadcast task decomposition module, a semantic understanding module, and a broadcast data analysis module. The broadcast control system supports connection to broadcast equipment from different brands via an adapter, which is hot-swappable. The broadcast task decomposition module can decompose tasks and distribute them to multiple broadcast devices for execution. The semantic understanding module generates broadcast tasks using natural language processing technology. It can analyze abnormal data and execution frequency of broadcast tasks and propose workflow optimization suggestions. like Figure 2 As shown, the broadcast control system includes a broadcast controller and multiple broadcast adapters. The broadcast controller connects to each of the broadcast adapters, and each broadcast adapter connects to a broadcast device. The broadcast adapters support integration with broadcast devices from different domestic and international brands, enabling unified management and operation of multi-brand broadcast systems. The broadcast adapters are hot-swappable without affecting broadcast tasks, ensuring task continuity. The broadcast controller also provides multiple broadcast interfaces, allowing access to broadcast tasks from various channels such as telephone systems, SMS systems, web pages, apps, and mini-programs. A task queue is used to assign broadcast tasks to the adapters of each broadcast system. The broadcast control system also includes an adapter manager and a task scheduling module. The adapter manager automatically detects currently connected broadcast devices and assigns them corresponding broadcast adapters. The task scheduling module is responsible for the decomposition and distribution of broadcast tasks, ensuring that different broadcast devices can execute tasks synchronously. The hot-swapping of adapters relies on the system's automatic detection and configuration capabilities; when an adapter fails or is replaced, the system automatically switches without affecting the current broadcast task.

[0018] TCP communication messages contain only the following basic commands to ensure that these commands can be implemented across different broadcast systems: Broadcast task queries are sent as heartbeat messages at regular intervals. A broadcast task includes the task ID, broadcast content, and broadcast partition information. Broadcast the task execution result, including the task ID and the execution result; Broadcast system information queries are sent from the controller to the adapter; The results of the broadcast system information query are returned to the adapter by the controller.

[0019] The semantic understanding module generates broadcast tasks from input content. When a user makes a phone call, they input the broadcast task via intelligent voice dialogue; the task is then directly generated through voice translation and natural language processing. Similarly, when a user sends a text message, the semantic understanding module extracts key information to generate the broadcast task. Users can also directly input the content and execution conditions of the broadcast task through web pages, apps, or mini-programs, and the semantic understanding module generates the task. The semantic understanding module also features intelligent voice recognition, supporting the matching of personnel and departments through natural language processing to improve task assignment accuracy. Furthermore, it includes voice translation functionality, directly extracting names and department information from speech without matching through the address book, significantly improving the accuracy of broadcast-based task finding.

[0020] The broadcast data analysis module can analyze abnormal data and execution frequency of broadcast tasks, and propose workflow optimization suggestions. Through this module, users can input broadcast requests via voice input through telephone, APP, etc. The system uses NLP algorithms to analyze the voice content, extracting key information such as broadcast time, location, and target audience, and generating executable broadcast tasks. Users do not need to manually input or match, greatly improving the convenience of task assignment. The system can record and summarize broadcast records. By analyzing the broadcast frequency of broadcast time, work group, personnel, and position, the platform can provide optimization suggestions based on data statistics, helping users improve their workflows. Users can directly issue broadcast tasks via voice. Users can freely describe broadcast requirements; the system extracts key information through the broadcast data analysis module and automatically generates broadcast tasks, eliminating the need for tedious manual input.

[0021] The platform supports broadcast partition combination tasks and supports the execution of partition tasks across broadcast systems.

[0022] In the broadcast task decomposition module, broadcast tasks can be broken down into multiple sub-tasks, which are executed independently by different brand broadcast systems based on their own conditions. The execution of each broadcast task does not interfere with each other, thus improving task reliability. During the broadcast task decomposition process, the platform automatically assigns tasks to appropriate broadcast partitions and supports cross-system task combinations.

[0023] An AI broadcast interaction method includes the following steps: Collect conversation data: Collect user conversation data, including voice and text; Data annotation: The collected data is annotated, including the intent of the dialogue, entities, relationships, etc. Training the model: The model is trained using labeled data. The trained model includes a speech recognition model, a semantic understanding model, and a dialogue generation model. Test model: Test the accuracy and efficiency of the model; Deploy the model: Deploy the trained model into the system to enable intelligent dialogue functionality.

[0024] Broadcast data analysis: The system records the execution status of broadcast tasks in real time and automatically generates broadcast statistical reports. By analyzing broadcast execution data, such as the frequency of use by time, personnel, and zones, the system can identify abnormal broadcasts and provide optimization suggestions to help improve broadcast efficiency and accuracy.

[0025] The specific implementation process is as follows: Data logging: Records the execution status of broadcast tasks, including broadcast time, broadcast content, broadcast zone, broadcast personnel, and other information; Data statistics: Calculate the execution frequency of broadcast tasks and generate broadcast statistics reports; Data analysis: Analyze the execution data of broadcast tasks to identify abnormal broadcasts; Anomaly detection: Identify abnormal broadcasts using anomaly detection algorithms; Abnormal Broadcast Analysis: Manually analyze the causes of abnormal broadcasts; Optimization suggestions: Based on the analysis results of abnormal broadcasts, suggestions for workflow optimization are proposed.

Claims

1. An AI broadcasting interactive platform, characterized in that: The system includes a broadcast control system, a broadcast task decomposition module, and a semantic understanding module. The broadcast control system supports connection to broadcast devices of different brands through an adapter. The broadcast task decomposition module decomposes tasks and distributes them to multiple broadcast devices for execution. The semantic understanding module generates broadcast tasks from input content. The broadcast data analysis module analyzes abnormal data and execution frequency of broadcast tasks and proposes workflow optimization suggestions.

2. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: The broadcast control system includes a broadcast controller and multiple broadcast adapters. The broadcast controller is connected to multiple broadcast adapters, and each broadcast adapter is connected to a broadcast device. The broadcast controller provides multiple broadcast interfaces, which can access broadcast tasks from telephone systems, SMS systems, web pages, APPs, and mini-programs.

3. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: The broadcast control system includes an adapter manager and a task scheduling module. The adapter manager automatically detects the currently connected broadcast devices and assigns them corresponding broadcast adapters. The task scheduling module is responsible for the decomposition and distribution of broadcast tasks to ensure that different broadcast devices can execute tasks synchronously.

4. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: The semantic understanding module generates broadcast tasks directly through voice translation and natural language processing when a user makes a phone call and engages in intelligent voice dialogue to input the broadcast task.

5. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: When a user sends a text message, the semantic understanding module extracts key information about the broadcast task and generates the broadcast task.

6. The AI ​​broadcast interaction platform as described in claim 1, characterized in that: The semantic understanding module generates broadcast tasks when users directly input the content and execution conditions of the broadcast task through web pages, apps, or mini-programs.

7. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: The broadcast data analysis module analyzes the speech content using NLP algorithms, extracts key information such as broadcast time, location, and audience, and generates an executable broadcast task.

8. The AI ​​broadcasting interactive platform as described in claim 1, characterized in that: In the aforementioned broadcast task decomposition module, the broadcast task is decomposed into multiple sub-tasks, which are executed independently by different brand broadcast systems according to their own conditions.

9. An AI broadcast interaction method, characterized in that, Includes the following steps: Collect conversation data: Collect user conversation data, including voice and text; Data annotation: The collected data is annotated, including the intent, entities, and relationships in the dialogue; Training the model: The model is trained using labeled data. The trained model includes a speech recognition model, a semantic understanding model, and a dialogue generation model. Test model: Test the accuracy and efficiency of the model; Deploying the model: Deploying the trained model into the system to enable intelligent dialogue functionality; Broadcast data analysis: Records the execution status of broadcast tasks in real time and automatically generates broadcast statistical reports.

Images