Intelligent arrangement method and device for urban governance application scene

Abstract

The invention provides an intelligent arrangement method and device for urban governance application scenarios, and the method comprises the steps: carrying out the unified and automatic management of an AI visual algorithm model, edge computing resources and access equipment through employing an AI resource pool, and enabling the access equipment to at least comprise video monitoring equipment; selecting corresponding video monitoring equipment from the AI resource pool to realize video stream input on the visual arrangement platform according to an application scene of urban governance, selecting a required AI visual algorithm model, and selecting edge computing resources required for operation according to the AI visual algorithm model; arranging and optimizing an application scene according to the selected video monitoring equipment, the AI vision algorithm model and the edge computing resources, and generating a scene task required by the application scene; and executing the scene task for event prediction and response. The system has the advantages that compared with a traditional city management system, the system has higher flexibility, automation and intelligence, and the overall efficiency, precision and flexibility of city management can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to an intelligent orchestration method and device for urban governance application scenarios.

Background Art

[0002] The goal of urban governance is to achieve sustainable urban development, improve the quality of life of residents, optimize resource utilization, and achieve environmental protection, etc. Urban management systems are the products of the urban governance process. By collecting, analyzing, and utilizing various data of the city, they can timely discover and handle various problems existing in the urban governance process, thereby enhancing the intelligent level of urban governance. However, the existing urban management systems have the following defects in actual use:

[0003] 1. In terms of resource management: The existing urban management systems usually adopt static resource allocation and cannot be dynamically adjusted according to actual needs, which easily causes resource waste or performance bottlenecks; at the same time, the scheduling and optimization of resources mostly rely on manual operations, with high operation and maintenance costs and also reducing the response speed of the system;

[0004] 2. In terms of algorithm models: The update of algorithm models depends on manual intervention and cannot respond in a timely manner to environmental changes or the input of new data, resulting in affecting the prediction accuracy of the models; the adopted algorithm models are single, lacking flexibility and adaptability when dealing with complex and changeable urban governance scenarios;

[0005] 3. In terms of scenario orchestration: Usually, professional technical personnel are required to write code to configure application scenarios, resulting in a long development cycle, high deployment difficulty, and once the application scenario configuration is determined, it is difficult to perform real-time adjustment and optimization;

[0006] 4. In terms of event response: Most events are passive responses, lacking prediction and early intervention mechanisms, unable to timely discover potential problems, and with low response efficiency;

[0007] 5. In terms of system collaboration and decision-making: Different urban management systems usually operate independently, unable to effectively share and collaborate, and decision-making depends on manual analysis and historical experience, with low decision-making efficiency.

[0008] In view of the above existing problems, the inventor of this case conducted in-depth research on this problem, and thus this case was born.

Summary of the Invention

[0009] The technical problem to be solved by the present invention is to provide an intelligent orchestration method and device for urban governance application scenarios, to solve the defects existing in the existing urban management systems in terms of resource management, algorithm models, scenario orchestration, event response, system collaboration and decision-making, to enhance the efficiency, accuracy, and flexibility of urban governance, and to provide strong technical support for the construction of smart cities.

[0010] The present invention is implemented as follows:

[0011] In a first aspect, an intelligent orchestration method for an urban governance application scenario, the orchestration method comprising the following steps:

[0012] Using an AI resource pool to uniformly and automatically manage AI vision algorithm models, edge computing resources, and access devices, where the access devices at least include video surveillance devices;

[0013] On a visualization orchestration platform, according to the application scenario of urban governance, select the corresponding video surveillance devices from the AI resource pool to achieve video stream input, select the required AI vision algorithm models, and select the edge computing resources required for operation according to the AI vision algorithm models; Orchestrate and optimize the application scenario based on the selected video surveillance devices, AI vision algorithm models, and edge computing resources, and generate the scenario tasks required for the application scenario;

[0014] Execute the scenario tasks for event prediction and response.

[0015] Further, the using of the AI resource pool to uniformly and automatically manage the AI vision algorithm models specifically includes:

[0016] Model version control: Manage the versions of the trained AI vision algorithm models and record the relevant information of the AI vision algorithm models;

[0017] Model evaluation and monitoring: Automatically evaluate the performance of the AI vision algorithm models using evaluation metrics and perform real-time monitoring on the AI vision algorithm models running online;

[0018] Model optimization and update: Automatically optimize the model parameters, adjust the model structure, or retrain the model according to the actual application effects and data changes of the AI vision algorithm models;

[0019] Model deployment and inference: Deploy the trained AI vision algorithm models to edge devices or cloud servers and provide a unified API interface to call the required AI vision algorithm models through the API interface to achieve model inference;

[0020] Multi-model collaboration: Support the parallel operation of multiple AI vision algorithm models and automatically select the appropriate AI vision algorithm models to execute tasks according to the actual application scenarios and real-time data.

[0021] Further, the using of the AI resource pool to uniformly and automatically manage the edge computing resources specifically includes:

[0022] Using an AI algorithm to real-time monitor the resource consumption of the edge computing resources and dynamically adjust the allocation of the edge computing resources according to the resource consumption;

[0023] Introduce an intelligent scheduling mechanism to optimize resource allocation through the intelligent scheduling mechanism;

[0024] Utilize a machine learning model to monitor hardware and software failures in real time, automatically detect and feedback potential problems, and automatically reallocate resources or enable a backup system through a self-healing mechanism after a failure is detected.

[0025] Furthermore, the unified automatic management of access devices using the AI resource pool includes:

[0026] Connect various video surveillance devices and various sensors to the AI resource pool, and use the AI resource pool to monitor and manage the connected video surveillance devices and sensors; analyze and preprocess the raw data collected by various video surveillance devices and various sensors, and store the analyzed and processed data using a distributed storage system.

[0027] Furthermore, the specific arrangement and optimization of the application scenario according to the selected video surveillance device, AI vision algorithm model, and edge computing resources include:

[0028] In the graphical interface provided by the visual arrangement platform, according to the selected video surveillance device, AI vision algorithm model, and edge computing resources, create an application scenario by dragging and configuring or create an application scenario using the scenario library provided by the visual arrangement platform, and configure scenario parameters according to actual application requirements;

[0029] Define various rules or policies using a rule engine, and perform automated management and control of the application scenario through the defined rules or policies; dynamically adjust the running application scenario according to the connected real-time data;

[0030] Display the running status of the application scenario in the graphical interface provided by the visual arrangement platform.

[0031] Furthermore, the event prediction and response specifically include:

[0032] Spatio-temporal data analysis: Identify and predict potential scenario events by combining historical data, real-time data, or spatial information;

[0033] Alarm management: Set different alarm levels according to the severity and priority of the scenario event;

[0034] Intelligent linkage response: When an alarm is triggered, automatically link relevant resources for response according to the alarm information;

[0035] Emergency plan management: Pre-develop various emergency plans and execute the corresponding emergency plan according to the trigger of the scenario event;

[0036] Scenario event subscription: Push scenario events to third-party business systems through message middleware or API interfaces.

[0037] Furthermore, the orchestration method further includes system collaboration and data sharing, specifically including:

[0038] Data interface standardization: Adopt a unified data interface standard to achieve data exchange between different systems;

[0039] Data sharing: Establish a data sharing platform and use the data sharing platform to centrally manage relevant data for urban governance;

[0040] System integration: Integrate with other systems to achieve message intercommunication and collaborative work with other systems.

[0041] Furthermore, the orchestration method further includes providing decision support, specifically including:

[0042] Data visualization: Visualize various data analysis results in the form of graphs and tables;

[0043] Decision support model: Build a decision support model and use the decision support model to automatically generate decision recommendation plans based on data analysis results;

[0044] Situation analysis: Conduct comprehensive analysis and evaluation on the operation situations of various application scenarios in urban governance, and provide decision-making references based on the analysis and evaluation results.

[0045] Furthermore, the orchestration method further includes execution and feedback, specifically including:

[0046] Task dispatch: Convert decisions into specific tasks and dispatch the specific tasks to executors;

[0047] Process monitoring: Monitor the execution process of specific tasks, and track the progress and completion status of specific tasks;

[0048] Feedback collection and analysis: Collect feedback information after the execution of specific tasks, and analyze the execution effects and existing problems based on the feedback information;

[0049] Continuous optimization: Continuously optimize the system based on feedback information.

[0050] Second, an intelligent orchestration device for urban governance application scenarios, the device includes an AI management module, a scenario orchestration module, and a task execution module;

[0051] The AI management module is used to uniformly and automatically manage AI vision algorithm models, edge computing resources, and access devices using an AI resource pool, where the access devices at least include video surveillance devices;

[0052] The scenario orchestration module is used to select corresponding video surveillance devices from the AI resource pool on the visual orchestration platform according to the application scenarios of urban governance to achieve video stream input, select the required AI vision algorithm models, and select the edge computing resources required for operation according to the AI vision algorithm models; orchestrate and optimize the application scenarios based on the selected video surveillance devices, AI vision algorithm models, and edge computing resources, and generate the scenario tasks required for the application scenarios.

[0053] The task execution module is used to execute scenario tasks for event prediction and response.

[0054] By improving aspects such as resource management, algorithm models, scenario orchestration, event response, system coordination and decision-making, the present invention has higher flexibility, automation and intelligence compared with traditional urban management systems, can effectively improve the overall efficiency, accuracy and flexibility of urban governance, and thus provide strong technical support for the construction of smart cities. Specifically, it includes the following aspects:

[0055] 1. The AI resource pool is adopted to uniformly and automatically manage the AI vision algorithm models. On the one hand, it can conveniently optimize and update the AI vision algorithm models automatically according to the actual application effects and real-time data changes, ensure that the AI vision algorithm models are always in the best state, thereby improving the adaptability and intelligence of the system, and at the same time ensuring the prediction accuracy of the models; on the other hand, it supports the parallel operation of multiple AI vision algorithm models, can automatically select the most suitable AI vision algorithm model according to different application scenarios and requirements, can improve the usage flexibility and adaptability, and at the same time improve the processing efficiency and accuracy.

[0056] 2. The AI resource pool is adopted to uniformly and automatically manage the edge computing resources. On the one hand, the AI-based resource pool management can well realize the dynamic adjustment of computing resources, so that in the actual use process, the edge computing resources can be intelligently adjusted according to the resource consumption situation, thereby avoiding resource waste or performance bottlenecks and improving the resource utilization rate; on the other hand, the AI can be used to automatically detect the system performance and potential faults, and perform automatic repair through the self-healing mechanism, which can greatly reduce manual intervention, improve the system stability and response speed, and also reduce the operation and maintenance costs.

[0057] 3. By connecting various video surveillance devices and various sensors to the AI resource pool, in the actual use process, it can well realize the data collection of multiple data sources, that is, realize multi-modal data collection; at the same time, based on the AI-based resource pool management, it has the ability to quickly and real-time process data, can well support automatically extracting key information from unstructured data through AI algorithms, and realize intelligent analysis.

[0058] 4. By using a low-code or no-code platform to provide a graphical interface, when in specific use, application scenarios can be quickly created by dragging and configuring or by using the provided scenario library, enabling non-technical personnel to easily define and adjust application scenarios, which helps to accelerate project deployment. At the same time, the application scenarios can be automatically managed and controlled through defined rules or policies, and can also be dynamically adjusted according to the docked real-time data during the running of the application scenarios, that is, real-time adjustment and optimization can be achieved, which helps to greatly improve the usage flexibility and response speed.

[0059] 5. By based on the AI vision algorithm model and combined with historical data, real-time data or spatial information, it can well achieve active identification and prediction of potential scenario events, and issue early warnings or automatically trigger the execution of relevant emergency plans in advance, which can effectively improve the response efficiency. At the same time, the scenario events can be pushed to a third-party business system through a message middleware or API interface to achieve intelligent subscription of scenario events, which helps to improve the timeliness and accuracy of event processing.

[0060] 6. By adopting a unified data interface standard, establishing a data sharing platform or system integration, seamless docking with other systems can be well achieved, so as to share data and work collaboratively with other systems, forming an intelligent urban governance platform, which can greatly improve the governance efficiency and information interoperability.

[0061] 7. By based on data analysis and AI prediction capabilities, it can well provide intelligent decision-making support for urban governors and improve the governance effect.

Description of the Drawings

[0062] The following further describes the present invention with reference to the accompanying drawings in combination with embodiments.

[0063] Figure 1 It is a flowchart of an intelligent orchestration method for an urban governance application scenario of the present invention;

[0064] Figure 2 A structural schematic diagram of an intelligent orchestration device for an urban governance application scenario.

Detailed Embodiments

[0065] In order to better understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below in combination with the accompanying drawings of the specification and specific embodiments.

[0066] It should be noted here that the orientation or positional relationship indicated by terms such as "center", "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing these embodiments and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. In addition, terms such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features.

[0067] Please refer to Figure 1 As shown, an intelligent orchestration method for an urban governance application scenario of the present invention, the orchestration method includes the following steps:

[0068] The AI resource pool is used to uniformly and automatically manage the AI vision algorithm model, edge computing resources, and access devices. Among them, the access devices at least include video surveillance devices;

[0069] On the visual orchestration platform, according to the application scenario of urban governance, select the corresponding video surveillance device from the AI resource pool to realize video stream input, select the required AI vision algorithm model, and select the edge computing resources required for operation according to the AI vision algorithm model; perform the orchestration and optimization of the application scenario according to the selected video surveillance device, AI vision algorithm model, and edge computing resources, and generate the scenario tasks required for the application scenario;

[0070] Execute the scenario tasks for event prediction and response.

[0071] When the present invention is specifically implemented, the AI resource pool provides five aspects of content, including device management, edge computing resource management, AI algorithm supermarket, video transcoding, and one standard with N actual data. Among them, device management refers to the management of various video surveillance devices, sensors, positioning devices, etc. based on AI technology, including video device data docked from video platforms such as Hikvision and Dahua; edge computing resource management refers to the management of various resources such as edge nodes, CPUs, edge computing boxes, GPU servers, etc. based on AI technology; the AI algorithm supermarket is used to uniformly manage the AI vision algorithm models; video transcoding is used to support transcoding of video streams, such as H265 to H264, etc.; one standard with N actual data is used to provide analysis data for other business aids of the created application scenario, and one standard with N actual data may include population data, enterprise data, geographical data, housing data, etc.

[0072] In some embodiments of the present invention, the unified automatic management of the AI vision algorithm model using the AI resource pool specifically includes:

[0073] Model version control: Manage the versions of the trained AI vision algorithm models, record the relevant information of the AI vision algorithm models, so as to facilitate the traceability and backtracking of the AI vision algorithm models. Among them, the relevant information of the AI vision algorithm models may include training parameters, data sets, performance metrics, etc.;

[0074] Model evaluation and monitoring: Automatically evaluate the performance of the AI vision algorithm models using evaluation metrics, and perform real-time monitoring on the AI vision algorithm models running online, so as to detect the decline or abnormal conditions of the model performance in a timely manner. Among them, the evaluation metrics used may include accuracy, recall rate, F1 value, etc.;

[0075] Model optimization and update: Automatically optimize the model parameters, adjust the model structure or retrain the model according to the actual application effects and data changes of the AI vision algorithm models, so as to improve the accuracy and robustness of the models; In the specific implementation of the present invention, the AI resource pool can automatically optimize and update the AI vision algorithm models based on AI algorithms and deep learning technologies according to the real-time collected relevant data (such as traffic flow, environmental monitoring data, video surveillance data, etc.), and adopt the methods of online learning and incremental learning to ensure that the AI vision algorithm models can also quickly adapt under new data and environmental conditions, reducing the need for manual intervention;

[0076] Model deployment and inference: Deploy the trained AI vision algorithm models to edge devices or cloud servers to support efficient model inference, and provide a unified API interface to call the required AI vision algorithm models through the API interface to achieve model inference;

[0077] Multi-model collaboration: Support the parallel operation of multiple AI vision algorithm models, and automatically select the appropriate AI vision algorithm model to execute tasks according to the actual application scenarios and real-time data; For example, in the application scenario of traffic management, the system can select a traffic flow monitoring model based on image recognition or a traffic prediction model based on time series analysis, and can specifically perform intelligent model switching according to data characteristics.

[0078] In the specific implementation of the present invention, before using the AI resource pool to perform unified automatic management on the AI vision algorithm models, it also includes creating AI vision algorithm models. The specific creation process of the AI vision algorithm models is as follows:

[0079] Data processing: Process the collected relevant data, including data cleaning, formatting and other related operations;

[0080] Model development: Develop or select an AI vision algorithm model suitable for the scenario task;

[0081] Model training: Use the processed data to train the developed or selected AI vision algorithm model to improve the accuracy of the AI vision algorithm model;

[0082] Model evaluation: Use performance metrics to evaluate the trained AI vision algorithm model to ensure the effectiveness of the AI vision algorithm model;

[0083] Model deployment: Deploy the trained AI vision algorithm model to the actual application environment to process data and make predictions in real time through the AI vision algorithm model.

[0084] In the present invention, by using the AI resource pool to uniformly and automatically manage the AI vision algorithm model, on the one hand, it can conveniently optimize and update the AI vision algorithm model automatically according to the actual application effect and real-time data changes, ensure that the AI vision algorithm model is always in the best state, thereby improving the adaptability and intelligence of the system, and at the same time ensuring the prediction accuracy of the model; on the other hand, it supports the parallel operation of multiple AI vision algorithm models, can automatically select the most suitable AI vision algorithm model according to different application scenarios and requirements, can improve the flexibility and adaptability of use, and at the same time improve the processing efficiency and accuracy.

[0085] In some embodiments of the present invention, the uniform and automatic management of edge computing resources by using the AI resource pool specifically includes:

[0086] Use the AI algorithm to monitor the resource consumption of edge computing resources (such as computing power, storage capacity, bandwidth, etc.) in real time, and dynamically adjust the allocation of edge computing resources according to the resource consumption; for example, during peak hours or in case of emergencies, the allocation of edge computing resources can be dynamically adjusted according to the resource consumption to avoid system overload;

[0087] Introduce an intelligent scheduling mechanism to optimize the resource allocation through the intelligent scheduling mechanism. For example, in the specific implementation of the present invention, the resource allocation can be optimized based on factors such as the priority of tasks, resource requirements, and current load;

[0088] Use the machine learning model to monitor hardware and software failures in real time, automatically detect and feedback potential problems to reduce manual intervention, and automatically reallocate resources or enable the standby system through the self-healing mechanism after a failure is found to ensure the continuous and stable operation of the system.

[0089] In the present invention, the edge computing resources are uniformly and automatically managed by using an AI resource pool. On the one hand, the resource pool management based on AI can well realize the dynamic adjustment of computing resources, so that during the actual use, the edge computing resources can be intelligently adjusted according to the resource consumption situation, thus avoiding resource waste or performance bottlenecks and improving the resource utilization rate. On the other hand, the AI can be used to automatically detect the system performance and potential faults, and automatically repair them through a self-healing mechanism, which can greatly reduce manual intervention, improve the system stability and response speed, and also reduce the operation and maintenance costs.

[0090] In some embodiments of the present invention, the unified and automatic management of access devices by using the AI resource pool includes:

[0091] Connect various video surveillance devices (including high-definition cameras, PTZ cameras, bayonet cameras, etc.) and various sensors (such as environmental monitoring sensors, manhole cover sensors, water level sensors, etc.) to the AI resource pool, and use the AI resource pool to monitor and manage the connected video surveillance devices and sensors. When the present invention is specifically implemented, for video surveillance devices, it supports multiple protocols such as GB / T28181, RTSP, and ONVIF for access, and for sensors, it can use protocols such as MQTT and CoAP to realize data transmission.

[0092] Analyze and preprocess the raw data collected by various video surveillance devices and various sensors. For example, the collected raw data can be cleaned, filtered, format-converted, and standardized to remove noise, duplicate data, invalid information, etc. to ensure the data quality; and a distributed storage system (such as HDFS, object storage, etc.) is used to store the analyzed and processed data. By using a distributed storage system, it can well realize the storage of a large amount of video, image, and structured data, and support efficient data retrieval and access.

[0093] In the present invention, by connecting various video surveillance devices and various sensors to the AI resource pool, during the actual use, it can well realize the data collection of multiple data sources, that is, realize multi-modal data collection; at the same time, based on the resource pool management of AI, it has the ability to quickly and real-time process data, and can well support the automatic extraction of key information from unstructured data (such as video, image) through AI algorithms (such as image recognition, object detection, scene analysis, etc.) and realize intelligent analysis.

[0094] In some embodiments of the present invention, the arrangement and optimization of the application scenario according to the selected video surveillance device, AI vision algorithm model, and edge computing resources specifically include:

[0095] In the graphical interface provided by the visualization orchestration platform, according to the selected video surveillance devices, AI vision algorithm models, and edge computing resources, create application scenarios by dragging and configuring, or create application scenarios using the scenario library provided by the visualization orchestration platform, and configure scenario parameters according to actual application requirements, such as monitoring areas, alarm thresholds, analysis algorithms, etc. In the specific implementation of the present invention, the visualization orchestration platform specifically uses a low-code or no-code platform and provides a graphical interface, so that users can quickly create application scenarios by dragging and configuring without writing complex code; of course, some templates for application scenarios can also be preset and placed in the scenario library of the visualization orchestration platform, such as common application scenarios in urban governance (such as traffic management, security monitoring, environmental monitoring, etc.), so that users can quickly combine and customize the required application scenarios using the templates.

[0096] Define various rules or policies using a rule engine, such as event trigger conditions, alarm levels, response processes, etc., and automate the management and control of application scenarios through the defined rules or policies; for example, scenario tasks can be generated through the defined rules or policies, and the scenario tasks can be various required tasks such as planned running time periods, frequencies, etc.

[0097] Dynamically adjust the running application scenarios according to the docked real-time data. For example, during special events (such as large-scale events, traffic accidents), the signal lights, positions of surveillance cameras can be adjusted in real time through the platform, or an emergency response plan can be activated; at the same time, in the specific implementation of the present invention, optimization plans can also be automatically recommended according to historical data and real-time monitoring situations to improve the flexibility and response speed of urban governance.

[0098] Display the running status of the application scenario in the graphical interface provided by the visualization orchestration platform. For example, the running status, event occurrence situations, resource usage situations, etc. of the application scenario can be displayed in the graphical interface to facilitate users' monitoring and management.

[0099] The present invention uses a low-code or no-code platform to provide a graphical interface, so that in specific use, application scenarios can be quickly created by dragging and configuring, or quickly created using the provided scenario library, enabling non-technical personnel to easily define and adjust application scenarios, which helps to accelerate project deployment; at the same time, the application scenarios can also be automatically managed and controlled through the defined rules or policies, and the running application scenarios can be dynamically adjusted according to the docked real-time data, that is, real-time adjustment and optimization can be achieved, which helps to greatly improve the usage flexibility and response speed.

[0100] In some embodiments of the present invention, the event prediction and response specifically include:

[0101] Spatio-temporal data analysis: Identify and predict potential scenario events by combining historical data, real-time data, or spatial information, such as garbage overflow events, crowd gathering events, road waterlogging events, vehicle violation events, etc.; In the specific implementation of the present invention, analysis methods such as time series analysis and spatial analysis can be used and combined with historical data, real-time data, and spatial information to predict potential events, such as traffic congestion, safety hazards, etc.;

[0102] Alarm management: Set different alarm levels according to the severity and priority of scenario events, and can send alarm information to relevant personnel in a timely manner through methods such as text messages, emails, and APP push;

[0103] Intelligent linkage response: When an alarm is triggered, automatically link relevant resources to respond according to the alarm information, such as dispatching police, adjusting traffic lights, activating emergency plans, etc.;

[0104] Emergency plan management: Pre-develop various emergency plans and execute the corresponding emergency plans according to the triggered scenario events to improve the emergency response efficiency;

[0105] Scenario event subscription: Push scenario events to third-party business systems through message middleware or API interfaces. In the specific implementation of the present invention, it can support the subscription of various scenario events such as urban management applications, public security monitoring applications, living environment applications, and social public management applications; For example, the urban management department can choose to subscribe to the real-time traffic flow information in a specific area, and the environmental protection department can subscribe to pollutant exceeding standard events, etc.; At the same time, when subscribing, it also supports customizing the priority, processing process, etc. of the event to ensure the accuracy and efficiency of event response.

[0106] The present invention can well realize the active identification and prediction of potential scenario events by combining the AI vision algorithm model with historical data, real-time data, or spatial information, and send out alarms in advance or automatically trigger the execution of relevant emergency plans, which can effectively improve the response efficiency; At the same time, scenario events can be pushed to third-party business systems through message middleware or API interfaces to realize intelligent subscription of scenario events, which helps to improve the timeliness and accuracy of event processing.

[0107] In some embodiments of the present invention, the orchestration method further includes system collaboration and data sharing, specifically including:

[0108] Data interface standardization: Adopt a unified data interface standard to realize data exchange between different systems;

[0109] Data sharing: Establish a data sharing platform and use the data sharing platform to centrally manage the relevant data of urban governance to break data islands;

[0110] System integration: Integrate with other systems to achieve message intercommunication and collaborative work with other systems; for example, it can be integrated with the urban comprehensive management platform, transportation system, emergency command system, etc. to achieve information intercommunication and collaborative work.

[0111] By adopting a unified data interface standard, establishing a data sharing platform or system integration, the present invention can well achieve seamless docking with other systems, thereby enabling data sharing and collaborative work with other systems, forming an intelligent urban governance platform, and greatly improving the governance efficiency and information intercommunication.

[0112] In some embodiments of the present invention, the orchestration method further includes providing decision support, specifically including:

[0113] Data visualization: Visualize various data analysis results in the form of graphs and tables. For example, various data analysis results can be displayed in the form of maps, reports, etc. to facilitate managers to intuitively understand the operation status of the city;

[0114] Decision support model: Construct a decision support model, and use the decision support model to automatically generate decision recommendation schemes based on data analysis results. For example, it can analyze the relationship between traffic flow and accident occurrence and recommend the optimal traffic control strategy; in the specific implementation of the present invention, the construction of the decision support model also includes steps such as data processing, model development, model training, and model evaluation, and specifically can refer to the creation process of the AI vision algorithm model;

[0115] Situation analysis: Comprehensively analyze and evaluate the operation situation of various application scenarios in urban governance, and provide decision-making references based on the analysis and evaluation results to facilitate managers to make decisions.

[0116] Based on data analysis and AI prediction capabilities, the present invention can well provide intelligent decision support for urban governors and improve the governance effect.

[0117] Furthermore, the orchestration method further includes execution and feedback, specifically including:

[0118] Task assignment: Convert the decision into a specific task and assign the specific task to the executor;

[0119] Process monitoring: Monitor the execution process of the specific task and track the progress and completion status of the specific task;

[0120] Feedback collection and analysis: Collect the feedback information after the execution of the specific task, and analyze the execution effect and existing problems according to the feedback information;

[0121] Continuous optimization: Continuously optimize the system based on feedback information. For example, the system functions, algorithm models, governance strategies, etc. can be continuously optimized to achieve continuous improvement and enhancement of the system.

[0122] In summary, through improvements in aspects such as resource management, algorithm models, scenario orchestration, event response, system collaboration, and decision-making, the present invention has higher flexibility, automation, and intelligence compared with traditional urban management systems, and can effectively improve the overall efficiency, accuracy, and flexibility of urban governance, thereby providing strong technical support for the construction of smart cities.

[0123] Embodiment 2

[0124] Please refer to Figure 2 As shown, an intelligent orchestration device for an urban governance application scenario of the present invention includes an AI management module, a scenario orchestration module, and a task execution module;

[0125] The AI management module is used to uniformly and automatically manage the AI vision algorithm model, edge computing resources, and access devices by using an AI resource pool. Among them, the access devices at least include video surveillance devices;

[0126] The scenario orchestration module is used to select corresponding video surveillance devices from the AI resource pool on the visual orchestration platform to realize video stream input according to the application scenarios of urban governance, select the required AI vision algorithm model, and select the edge computing resources required for operation according to the AI vision algorithm model; perform the orchestration and optimization of the application scenario according to the selected video surveillance devices, AI vision algorithm model, and edge computing resources, and generate the scenario tasks required for the application scenario;

[0127] The task execution module is used to execute the scenario tasks for event prediction and response.

[0128] In the specific implementation of the present invention, the AI resource pool provides five aspects of content, including device management, edge computing resource management, AI algorithm supermarket, video transcoding, and one standard multi-entity data. Among them, device management refers to the management of various video surveillance devices, sensors, positioning devices, etc. based on AI technology, including video device data docked from video platforms such as Hikvision and Dahua; edge computing resource management refers to the management of various resources such as edge nodes, CPUs, edge computing boxes, and GPU servers based on AI technology; the AI algorithm supermarket is used to uniformly manage the AI vision algorithm models; video transcoding is used to support the transcoding of video streams, such as H265 to H264, etc.; one standard multi-entity data is used to provide analysis data for other business assistance of the created application scenarios, and the one standard multi-entity data can include population data, enterprise data, geographical data, housing data, etc.

[0129] In some embodiments of the present invention, in the AI management module, the unified automatic management of the AI vision algorithm model using the AI resource pool specifically includes:

[0130] Model version control: Manage the versions of the trained AI vision algorithm models, record the relevant information of the AI vision algorithm models, so as to facilitate the traceability and backtracking of the AI vision algorithm models. Among them, the relevant information of the AI vision algorithm models may include training parameters, data sets, performance metrics, etc.;

[0131] Model evaluation and monitoring: Automatically evaluate the performance of the AI vision algorithm models using evaluation metrics, and perform real-time monitoring on the AI vision algorithm models running online, so as to detect model performance degradation or abnormal situations in a timely manner. Among them, the evaluation metrics used may include accuracy, recall rate, F1 value, etc.;

[0132] Model optimization and update: Automatically optimize model parameters, adjust model structures or retrain models according to the actual application effects and data changes of the AI vision algorithm models, in order to improve the accuracy and robustness of the models; In the specific implementation of the present invention, the AI resource pool can automatically optimize and update the AI vision algorithm models based on AI algorithms and deep learning technologies according to the relevant data collected in real time (such as traffic flow, environmental monitoring data, video surveillance data, etc.), and adopt the methods of online learning and incremental learning to ensure that the AI vision algorithm models can quickly adapt under new data and environmental conditions, reducing the need for manual intervention;

[0133] Model deployment and inference: Deploy the trained AI vision algorithm models to edge devices or cloud servers to support efficient model inference, and provide a unified API interface to call the required AI vision algorithm models through the API interface to achieve model inference;

[0134] Multi-model collaboration: Support the parallel operation of multiple AI vision algorithm models, and automatically select the appropriate AI vision algorithm model to execute tasks according to the actual application scenarios and real-time data; For example, in the application scenario of traffic management, the system can select a traffic flow monitoring model based on image recognition or a traffic prediction model based on time series analysis, and specifically can perform intelligent model switching according to data characteristics.

[0135] In the specific implementation of the present invention, before the unified automatic management of the AI vision algorithm model using the AI resource pool, it also includes creating an AI vision algorithm model. The specific creation process of the AI vision algorithm model is as follows:

[0136] Data processing: Process the collected relevant data, including data cleaning, formatting and other related operations;

[0137] Model Development: Develop or select an AI vision algorithm model suitable for the scenario task;

[0138] Model Training: Use the processed data to train the developed or selected AI vision algorithm model to improve the accuracy of the AI vision algorithm model;

[0139] Model Evaluation: Evaluate the trained AI vision algorithm model using performance metrics to ensure the effectiveness of the AI vision algorithm model;

[0140] Model Deployment: Deploy the trained AI vision algorithm model to the actual application environment to process data and make predictions in real time through the AI vision algorithm model.

[0141] In the present invention, by using the AI resource pool to uniformly and automatically manage the AI vision algorithm model, on the one hand, it can conveniently optimize and update the AI vision algorithm model automatically according to the actual application effect and real-time data changes, ensuring that the AI vision algorithm model is always in the best state, thereby improving the adaptability and intelligence of the system, and at the same time ensuring the prediction accuracy of the model; on the other hand, it supports the parallel operation of multiple AI vision algorithm models, and can automatically select the most suitable AI vision algorithm model according to different application scenarios and requirements, which can improve the flexibility and adaptability of use, and at the same time improve the processing efficiency and accuracy.

[0142] In some embodiments of the present invention, in the AI management module, the specific method of using the AI resource pool to uniformly and automatically manage the edge computing resources includes:

[0143] Use the AI algorithm to monitor the resource consumption of the edge computing resources in real time (such as computing power, storage capacity, bandwidth, etc.), and dynamically adjust the allocation of the edge computing resources according to the resource consumption; for example, during peak hours or in emergency situations, the allocation of the edge computing resources can be dynamically adjusted according to the resource consumption to avoid system overload;

[0144] Introduce an intelligent scheduling mechanism to optimize the resource allocation through the intelligent scheduling mechanism. For example, in the specific implementation of the present invention, the resource allocation can be optimized based on factors such as the priority of the task, resource requirements, and current load;

[0145] Use the machine learning model to monitor the hardware and software failures in real time, automatically detect and feedback potential problems to reduce manual intervention, and automatically reallocate resources or enable the standby system through the self-healing mechanism after a failure is found to ensure the continuous and stable operation of the system.

[0146] In the present invention, the edge computing resources are uniformly and automatically managed by using an AI resource pool. On the one hand, the resource pool management based on AI can well realize the dynamic adjustment of computing resources, so that in the actual use process, the edge computing resources can be intelligently adjusted according to the resource consumption situation, thus avoiding resource waste or performance bottlenecks and improving the resource utilization rate. On the other hand, the AI can be used to automatically detect the system performance and potential faults, and automatically repair them through the self-healing mechanism, which can greatly reduce manual intervention, improve the system stability and response speed, and also reduce the operation and maintenance costs.

[0147] In some embodiments of the present invention, in the AI management module, the unified and automatic management of the access devices by using the AI resource pool includes:

[0148] Connect various video surveillance devices (including high-definition cameras, PTZ cameras, bayonet cameras, etc.) and various sensors (such as environmental monitoring sensors, manhole cover sensors, water level sensors, etc.) to the AI resource pool, and use the AI resource pool to monitor and manage the connected video surveillance devices and sensors. When the present invention is specifically implemented, for video surveillance devices, it supports multiple protocols such as GB / T28181, RTSP, and ONVIF for access, and for sensors, it can use protocols such as MQTT and CoAP to achieve data transmission;

[0149] Analyze and preprocess the original data collected by various video surveillance devices and various sensors. For example, the collected original data can be cleaned, filtered, format-converted, and standardized to remove noise, duplicate data, invalid information, etc. to ensure the data quality; and a distributed storage system (such as HDFS, object storage, etc.) is used to store the analyzed and processed data. By using a distributed storage system, it can well realize the storage of a large amount of video, image, and structured data, and support efficient data retrieval and access.

[0150] In the present invention, by connecting various video surveillance devices and various sensors to the AI resource pool, in the actual use process, it can well realize the data collection of multiple data sources, that is, realize multi-modal data collection; at the same time, based on the resource pool management of AI, it has the ability to quickly and real-time process data, and can well support the automatic extraction of key information from unstructured data (such as video, image) through AI algorithms (such as image recognition, object detection, scene analysis, etc.) and realize intelligent analysis.

[0151] In some embodiments of the present invention, in the scenario orchestration module, the orchestration and optimization of the application scenario according to the selected video surveillance device, AI vision algorithm model, and edge computing resources specifically include:

[0152] In the graphical interface provided by the visualization orchestration platform, according to the selected video surveillance devices, AI vision algorithm models, and edge computing resources, create application scenarios by dragging and configuring or create application scenarios using the scenario library provided by the visualization orchestration platform, and configure scenario parameters according to actual application requirements, such as monitoring areas, alarm thresholds, analysis algorithms, etc. In the specific implementation of the present invention, the visualization orchestration platform specifically adopts a low-code or no-code platform and provides a graphical interface, so that users can quickly create application scenarios by dragging and configuring without writing complex code; of course, some templates of application scenarios can also be preset and placed in the scenario library of the visualization orchestration platform, such as common application scenarios in urban governance (such as traffic management, security monitoring, environmental monitoring, etc.), so that users can quickly combine and customize the required application scenarios using the templates.

[0153] Define various rules or policies using a rule engine, such as event trigger conditions, alarm levels, response processes, etc., and automate the management and control of application scenarios through the defined rules or policies; for example, scenario tasks can be generated through the defined rules or policies, and the scenario tasks can be various required tasks such as planned running time periods, frequencies, etc.

[0154] Dynamically adjust the running application scenarios according to the docked real-time data. For example, during special events (such as large-scale events, traffic accidents), the signal lights, positions of surveillance cameras can be adjusted in real time through the platform or an emergency response plan can be activated; at the same time, in the specific implementation of the present invention, optimization plans can also be automatically recommended according to historical data and real-time monitoring situations to improve the flexibility and response speed of urban governance.

[0155] Display the running status of the application scenario in the graphical interface provided by the visualization orchestration platform. For example, the running status, event occurrence situations, resource usage situations, etc. of the application scenario can be displayed in the graphical interface to facilitate users' monitoring and management.

[0156] The present invention uses a low-code or no-code platform to provide a graphical interface, so that in specific use, application scenarios can be quickly created by dragging and configuring or quickly created using the provided scenario library, enabling non-technical personnel to easily define and adjust application scenarios, which helps to accelerate project deployment; at the same time, the application scenarios can also be automatically managed and controlled through the defined rules or policies, and the running application scenarios can be dynamically adjusted according to the docked real-time data, that is, real-time adjustment and optimization can be achieved, which helps to greatly improve the usage flexibility and response speed.

[0157] In some embodiments of the present invention, in the task execution module, the event prediction and response specifically include:

[0158] Spatiotemporal data analysis: Identify and predict potential scenario events by combining historical data, real-time data, or spatial information, such as garbage overflow events, crowd gathering events, road waterlogging events, vehicle violation events, etc.; In the specific implementation of the present invention, analysis methods such as time series analysis and spatial analysis can be used and combined with historical data, real-time data, and spatial information to predict potential events, such as traffic congestion, safety hazards, etc.;

[0159] Alarm management: Set different alarm levels according to the severity and priority of scenario events, and the alarm information can be sent to relevant personnel in a timely manner through methods such as text messages, emails, APP push, etc.;

[0160] Intelligent linkage response: When an alarm is triggered, automatically link relevant resources to respond according to the alarm information, such as dispatching police force, adjusting traffic lights, activating emergency plans, etc.;

[0161] Emergency plan management: Pre-develop various emergency plans and execute the corresponding emergency plans according to the triggered scenario events to improve the emergency response efficiency;

[0162] Scenario event subscription: Push scenario events to third-party business systems through message middleware or API interfaces. In the specific implementation of the present invention, it can support the subscription of various scenario events such as urban management applications, public security monitoring applications, living environment applications, and social public management applications; For example, urban management departments can choose to subscribe to real-time traffic flow information in specific areas, and environmental protection departments can subscribe to events of excessive pollutants, etc.; At the same time, when subscribing, it also supports customizing the priority, processing process, etc. of events to ensure the accuracy and efficiency of event response.

[0163] Through the present invention, based on the AI vision algorithm model and combined with historical data, real-time data, or spatial information, it can well achieve the active identification and prediction of potential scenario events, and issue alarms in advance or automatically trigger the execution of relevant emergency plans, which can effectively improve the response efficiency; At the same time, scenario events can be pushed to third-party business systems through message middleware or API interfaces to realize intelligent subscription of scenario events, which helps to improve the timeliness and accuracy of event processing.

[0164] In some embodiments of the present invention, the device further includes a collaborative sharing module, and the collaborative sharing module is used to achieve system collaboration and data sharing, specifically including:

[0165] Standardization of data interfaces: Adopt a unified data interface standard to achieve data exchange between different systems;

[0166] Data sharing: Establish a data sharing platform and use the data sharing platform to centrally manage relevant data for urban governance to break data islands;

[0167] System integration: Integrate with other systems to achieve message intercommunication and collaborative work with other systems; for example, it can be integrated with the urban comprehensive management platform, transportation system, emergency command system, etc. to achieve information intercommunication and collaborative work.

[0168] By adopting a unified data interface standard, establishing a data sharing platform or system integration, the present invention can well achieve seamless docking with other systems, thereby enabling data sharing and collaborative work with other systems, forming an intelligent urban governance platform, which can greatly improve governance efficiency and information intercommunication.

[0169] In some embodiments of the present invention, the device further includes a decision-making module, and the decision-making module is used to provide decision-making support, specifically including:

[0170] Data visualization: Visualize various data analysis results in the form of graphs and tables. For example, various data analysis results can be displayed in the form of maps, reports, etc. to facilitate managers to intuitively understand the operation status of the city;

[0171] Decision-making support model: Construct a decision-making support model, and use the decision-making support model to automatically generate decision-making suggestion schemes based on data analysis results. For example, it can analyze the relationship between traffic flow and accident occurrence and recommend the optimal traffic control strategy; in the specific implementation of the present invention, the construction of the decision-making support model also includes steps such as data processing, model development, model training, and model evaluation, and specifically can refer to the creation process of the AI vision algorithm model;

[0172] Situation analysis: Comprehensively analyze and evaluate the operation situation of various application scenarios in urban governance, and provide decision-making references based on the analysis and evaluation results to facilitate managers to make decisions.

[0173] Based on data analysis and AI prediction capabilities, the present invention can well provide intelligent decision-making support for urban governors and improve governance effects.

[0174] Furthermore, the device further includes an execution feedback module, and the execution feedback module is used to achieve execution and feedback, specifically including:

[0175] Task assignment: Convert decisions into specific tasks and assign the specific tasks to executors;

[0176] Process monitoring: Monitor the execution process of specific tasks and track the progress and completion status of specific tasks;

[0177] Feedback collection and analysis: Collect the feedback information after the execution of specific tasks, and analyze the execution effect and existing problems according to the feedback information;

[0178] Continuous optimization: Continuously optimize the system based on feedback information. For example, continuously optimize system functions, algorithm models, governance strategies, etc. to achieve continuous improvement and enhancement of the system.

[0179] In summary, by improving aspects such as resource management, algorithm models, scenario orchestration, event response, system collaboration, and decision-making, the present invention has higher flexibility, automation, and intelligence compared to traditional urban management systems, can effectively improve the overall efficiency, accuracy, and flexibility of urban governance, and thus provides strong technical support for the construction of smart cities.

[0180] Although the specific embodiments of the present invention have been described above, those skilled in the art of this technology should understand that the specific embodiments we described are illustrative rather than used to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be covered by the scope protected by the claims of the present invention.

Claims

1. An intelligent arrangement method for urban governance application scenarios, characterized in that: The arrangement method comprises the following steps: An AI resource pool is used to automatically manage AI visual algorithm models, edge computing resources, and access devices in a unified manner, where the access devices include at least video surveillance devices. On the visual orchestration platform, according to the application scenarios of urban governance, select the corresponding video surveillance equipment from the AI ​​resource pool to implement video stream input, select the required AI visual algorithm model, and select the edge computing resources required for operation based on the AI ​​visual algorithm model; orchestrate and optimize the application scenarios based on the selected video surveillance equipment, AI visual algorithm model and edge computing resources, and generate the scenario tasks required for the application scenarios; Perform scenario tasks to predict and respond to events.

2. The intelligent arrangement method of urban governance application scenarios according to claim 1, characterized in that: The use of the AI ​​resource pool to uniformly and automatically manage the AI ​​vision algorithm model specifically includes: Model version control: perform version management on the trained AI vision algorithm model and record relevant information of the AI ​​vision algorithm model; Model evaluation and monitoring: Use evaluation indicators to automatically evaluate the performance of AI vision algorithm models and monitor the AI ​​vision algorithm models running online in real time; Model optimization and update: Automatically optimize model parameters, adjust model structure, or retrain the model based on the actual application effect and data changes of the AI ​​vision algorithm model; Model deployment and reasoning: Deploy the trained AI vision algorithm model to edge devices or cloud servers, and provide a unified API interface to call the required AI vision algorithm model through the API interface to realize model reasoning; Multi-model collaboration: supports multiple AI vision algorithm models running in parallel, and automatically selects the appropriate AI vision algorithm model to perform tasks based on actual application scenarios and real-time data.

3. The intelligent arrangement method of urban governance application scenarios according to claim 1 is characterized in that: The use of the AI ​​resource pool to uniformly and automatically manage edge computing resources specifically includes: Use AI algorithms to monitor the resource consumption of edge computing resources in real time and dynamically adjust the allocation of edge computing resources based on resource consumption; Introduce intelligent scheduling mechanism to optimize resource allocation; Use machine learning models to monitor hardware and software failures in real time, automatically detect and feedback potential problems, and automatically reallocate resources or enable backup systems through self-healing mechanisms when failures are detected.

4. The intelligent arrangement method of urban governance application scenarios according to claim 1, characterized in that: The use of the AI ​​resource pool to uniformly and automatically manage access devices includes: Connect various video surveillance devices and sensors to the AI ​​resource pool, and use the AI ​​resource pool to monitor and manage the connected video surveillance devices and sensors; analyze and pre-process the raw data collected by various video surveillance devices and sensors, and use a distributed storage system to store the analyzed and processed data.

5. The intelligent arrangement method of urban governance application scenarios according to claim 1 is characterized in that: The arrangement and optimization of application scenarios based on the selected video surveillance equipment, AI visual algorithm model and edge computing resources specifically include: In the graphical interface provided by the visual orchestration platform, create application scenarios by dragging and dropping and configuring according to the selected video surveillance equipment, AI vision algorithm model and edge computing resources, or create application scenarios using the scenario library provided by the visual orchestration platform, and configure scenario parameters according to actual application requirements; Use the rule engine to define various rules or strategies, and use the defined rules or strategies to automatically manage and control application scenarios; dynamically adjust the running application scenarios based on the real-time data of the connection; The operation status of the application scenario is displayed in the graphical interface provided by the visual orchestration platform.

6. The intelligent arrangement method of urban governance application scenarios according to claim 1, characterized in that: The event prediction and response specifically include: Spatiotemporal data analysis: combining historical data, real-time data or spatial information to identify and predict potential scenario events; Alarm management: set different alarm levels according to the severity and priority of scene events; Intelligent linkage response: When an alarm is triggered, relevant resources are automatically linked to respond according to the alarm information; Plan management: Prepare various emergency plans in advance and execute corresponding emergency plans according to the triggering of scene events; Scenario event subscription: Push scenario events to third-party business systems through message middleware or API interface.

7. The intelligent arrangement method of urban governance application scenarios according to claim 1, characterized in that: The arrangement method also includes system collaboration and data sharing, specifically including: Data interface standardization: adopt unified data interface standards to realize data exchange between different systems; Data sharing: Establish a data sharing platform and use it to centrally manage relevant data on urban governance; System integration: Integrate with other systems to achieve message exchange and collaborative work with other systems.

8. The intelligent arrangement method of urban governance application scenarios according to claim 1, characterized in that: The arrangement method also includes providing decision support, specifically including: Data visualization: Visualize various data analysis results in the form of graphs and tables; Decision support model: Build a decision support model and use it to automatically generate decision suggestions based on data analysis results; Situation analysis: Comprehensively analyze and evaluate the operating status of various application scenarios in urban governance, and provide decision-making references based on the analysis and evaluation results.

9. The intelligent arrangement method of urban governance application scenarios according to claim 8, characterized in that: The arrangement method also includes execution and feedback, specifically including: Task distribution: convert decisions into specific tasks and distribute specific tasks to executors; Process monitoring: monitor the execution process of specific tasks and track the progress and completion of specific tasks; Feedback collection and analysis: Collect feedback information after the execution of specific tasks, and analyze the execution effect and existing problems based on the feedback information; Continuous optimization: Continuously optimize the system based on feedback information.

10. An intelligent arrangement device for urban governance application scenarios, characterized in that: The device includes an AI management module, a scene arrangement module and a task execution module; The AI ​​management module is used to use the AI ​​resource pool to uniformly and automatically manage the AI ​​visual algorithm model, edge computing resources and access devices, wherein the access devices at least include video surveillance devices; The scenario orchestration module is used to select the corresponding video surveillance device from the AI ​​resource pool to implement video stream input, select the required AI visual algorithm model, and select the edge computing resources required for operation according to the AI ​​visual algorithm model on the visual orchestration platform according to the application scenario of urban governance; Orchestrate and optimize application scenarios based on the selected video surveillance equipment, AI vision algorithm models, and edge computing resources, and generate scenario tasks required for the application scenarios; The task execution module is used to execute scenario tasks to predict and respond to events.

Images