Artificial Intelligence-Based Workflow Automatic Generation and Computing Power Allocation Method and Device

Abstract

An embodiment of the present application provides a method and device for automatically generating a workflow and allocating computing power based on artificial intelligence. The method includes: determining corresponding task requirements and model types corresponding to the task requirements according to a workflow generation instruction and a preset natural language analysis model, and determining a corresponding task structure according to the task requirements and the dependency relationships between the model types; determining a corresponding algorithm model and a computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in a preset model library, constructing a task workflow according to the model matching relationship in the algorithm model and the task structure, and scheduling corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds a preset computing power threshold, updating the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow. The present application can effectively improve the efficiency and accuracy of workflow generation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and specifically to a method and device for automatically generating a workflow and allocating computing power based on artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence technology, all walks of life are actively exploring how to apply artificial intelligence algorithms to actual scenarios. However, current artificial intelligence algorithms are often relatively independent and lack effective series and association mechanisms. For example, in a comprehensive intelligent analysis system, the results generated by each algorithm are often relatively independent and difficult to effectively combine and associate. This results in the inability to obtain complete and meaningful result information in scenarios where multiple algorithms act together.

[0003] In traditional methods, to solve this problem, a manual arrangement method is usually adopted, that is, the user needs to manually arrange the processes of each algorithm to obtain the comprehensive result. However, this manual arrangement method has problems of low efficiency and easy occurrence of errors. Especially in the case of complex association and dependency relationships between algorithms, manual arrangement often cannot meet the needs of users.

[0004] In addition, due to the large differences in the needs of users for combined results, existing technologies often cannot effectively respond. Each project party may have different ways of combining results, and the manual customization process is cumbersome and time-consuming, and prone to errors. This results in users often being unable to obtain comprehensive results that meet their needs in actual applications.

[0005] In summary, there are some challenges in current technologies that limit the effectiveness of artificial intelligence algorithms in actual applications. Especially in the series arrangement and result combination of multiple algorithms, existing technologies often cannot effectively respond to the needs of users, and new methods need to be sought to solve these problems. Summary of the Invention

[0006] Aiming at the problems in the prior art, this application provides a method and device for automatically generating a workflow and allocating computing power based on artificial intelligence, which can effectively improve the efficiency and accuracy of workflow generation.

[0007] To solve at least one of the above problems, this application provides the following technical solutions:

[0008] In the first aspect, this application provides a method for automatically generating a workflow and allocating computing power based on artificial intelligence, including:

[0009] Receive the workflow generation instruction sent by the user, determine the corresponding task requirements and the model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine the corresponding task structure according to the task requirements and the dependency relationships between the model types;

[0010] Determine the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each of the model types in the preset model library, construct a task workflow according to the model matching relationship between the algorithm model and the task structure, and schedule the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy;

[0011] Monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow. When the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0012] Further, the receiving the workflow generation instruction sent by the user, and determining the corresponding task requirements and the model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model includes:

[0013] Receive the workflow generation instruction sent by the user, parse the instruction of the workflow generation instruction to obtain the corresponding structured data;

[0014] Input the structured data into a preset pre-trained language model to obtain the task requirements output by the pre-trained language model, and determine the model types corresponding to the task requirements according to the preset association rules.

[0015] Further, the determining the corresponding task structure according to the task requirements and the dependency relationships between the model types includes:

[0016] Determine the corresponding task execution order according to the dependency relationships between the model types;

[0017] Determine the corresponding task structure according to the task requirements and the task execution order.

[0018] Further, the determining the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each of the model types in the preset model library includes:

[0019] Determine the corresponding computing power requirements according to the model complexity of each of the model types in the preset model library;

[0020] Determine the resource utilization rate of each model type according to the computing power requirements of each model type and the computing resources of the hardware devices in the current system, and determine the computing power allocation strategy according to the maximized resource utilization rate.

[0021] Further, constructing a task workflow according to the model matching relationship in the algorithm model and the task structure, and scheduling the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy, includes:

[0022] Convert the model matching relationship in the task structure into a directed graph of the task workflow, where the nodes in the directed graph of the task workflow represent algorithm models, and the edges in the directed graph of the task workflow represent the dependency relationships between algorithm models;

[0023] Construct a task workflow according to the algorithm model and the directed graph of the task workflow, and schedule the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy determined by the maximized resource utilization rate.

[0024] Further, when the real-time computing power occupancy data exceeds the preset computing power threshold, updating the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, includes:

[0025] When the real-time computing power occupancy data exceeds the preset computing power threshold, update the task priority of the algorithm models to be run according to the computing power requirements of the algorithm models to be run in the task workflow;

[0026] Update the computing power allocation strategy according to the current system computing power and the algorithm models to be run with updated task priorities.

[0027] Further, updating the computing power allocation strategy according to the current system computing power and the algorithm models to be run with updated task priorities, includes:

[0028] Determine the adjustment value when the task priority of the algorithm model to be run is updated;

[0029] Adjust the system computing power scheduled for the algorithm model to be run by the current system computing power according to the adjustment value.

[0030] In a second aspect, the present application provides a device for automatically generating a workflow and allocating computing power based on artificial intelligence, including:

[0031] A task structure determination module, configured to receive a workflow generation instruction sent by a user, determine corresponding task requirements and a model type corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine a corresponding task structure according to the task requirements and the dependency relationship between the model types;

[0032] A system computing power scheduling module, configured to determine a corresponding algorithm model and a computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in a preset model library, construct a task workflow according to the algorithm model and the model matching relationship in the task structure, and schedule corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy;

[0033] A computing power allocation update module, configured to monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow, and when the real-time computing power occupancy data exceeds a preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0034] In a third aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, the steps of the method for automatically generating a workflow and allocating computing power based on artificial intelligence are implemented.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for automatically generating a workflow and allocating computing power based on artificial intelligence are implemented.

[0036] In a fifth aspect, the present application provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the method for automatically generating a workflow and allocating computing power based on artificial intelligence are implemented.

[0037] As can be seen from the above technical solution, the present application provides a method and device for automatically generating a workflow and allocating computing power based on artificial intelligence. By using a workflow generation instruction and a preset natural language analysis model, the corresponding task requirements and the model types corresponding to the task requirements are determined, and the corresponding task structure is determined according to the task requirements and the dependency relationships between the model types; according to the current system computing power and the computing power requirements of each model type in the preset model library, the corresponding algorithm model and computing power allocation strategy are determined, a task workflow is constructed according to the model matching relationship between the algorithm model and the task structure, and the corresponding system computing power is scheduled for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds the preset computing power threshold, the computing power allocation strategy is updated according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0039] Figure 1 FIG. 1 is one of the schematic flowcharts of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0040] Figure 2 FIG. 2 is another schematic flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0041] Figure 3 FIG. 3 is yet another schematic flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0042] Figure 4 FIG. 4 is still another schematic flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0043] Figure 5 FIG. 5 is yet still another schematic flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0044] Figure 6 FIG. 6 is another schematic flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in an embodiment of the present application;

[0045] Figure 7Seventh flowchart of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in the embodiments of the present application;

[0046] Figure 8 Structural diagram of the device for automatically generating a workflow and allocating computing power based on artificial intelligence in the embodiments of the present application;

[0047] Figure 9 Structural diagram of the electronic device in the embodiments of the present application. Detailed implementation manners

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the scope of protection of the present application.

[0049] In the technical solutions of the present application, the acquisition, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations.

[0050] Considering that in the prior art, a manual arrangement method is usually adopted, that is, the user needs to manually arrange the processes of each algorithm to obtain the comprehensive result. However, this manual arrangement method has the problems of low efficiency and easy occurrence of errors. Especially in the case where the correlation and dependency relationship between algorithms are complex, manual arrangement often cannot meet the needs of users. The present application provides a method and device for automatically generating a workflow and allocating computing power based on artificial intelligence, determines the corresponding task requirements and the model types corresponding to the task requirements through workflow generation instructions and a preset natural language analysis model, and determines the corresponding task structure according to the task requirements and the dependency relationship between each model type; determines the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, constructs a task workflow according to the model matching relationship in the algorithm model and the task structure, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds the preset computing power threshold, updates the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0051] To effectively improve the efficiency and accuracy of workflow generation, the present application provides an embodiment of a method for automatically generating a workflow and allocating computing power based on artificial intelligence. Refer to Figure 1 , the method for automatically generating a workflow and allocating computing power based on artificial intelligence specifically includes the following content:

[0052] Step S101: Receive a workflow generation instruction sent by a user, determine corresponding task requirements and model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine a corresponding task structure according to the task requirements and the dependency relationships between the model types;

[0053] Optionally, in this embodiment, in this technical solution, step S101 first involves receiving a workflow generation instruction sent by a user. This means that the system receives an input from the user, which contains a description and requirements of the task that the user hopes to execute. For example, the user may send an instruction asking the system to generate a natural language processing workflow for analyzing a series of text data and extracting key information therefrom. This instruction may include a description of the task, the required input data, the expected output result, and so on.

[0054] Next, according to the received workflow generation instruction and the preset natural language analysis model, the system will perform parsing and analysis. The preset natural language analysis model can be a well-trained machine learning model for understanding and interpreting the natural language input provided by the user. The system will use this model to understand the user's instruction, identify the task requirements, and determine the required model types. For example, if the user describes a text classification task, the system will identify that the task requirement is to classify the text, and the model type may be a text classifier model.

[0055] According to the task requirements and the dependency relationships between the model types, the system further determines a corresponding task structure. This means that the system will consider the relevance and execution order between different tasks, as well as the dependency relationships between different models. For example, in a text analysis workflow, it may be necessary to perform text preprocessing first, then apply a text classification model for classification, and finally perform post-processing on the classification results. Therefore, the task structure will include the order and association of these steps.

[0056] In this way, through step S101, the system can determine the task requirements and model types according to the user's workflow generation instruction, and establish a corresponding task structure, laying a foundation for the subsequent workflow generation and execution.

[0057] Step S102: Determine a corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each of the model types in the preset model library, construct a task workflow according to the algorithm model and the model matching relationship in the task structure, and schedule corresponding system computing power for each algorithm model in the task workflow;

[0058] Optionally, in this embodiment, in step S102, the system determines the corresponding algorithm model and computing power allocation strategy according to the computing power situation of the current system and the computing power requirements of each model type in the preset model library. The purpose of this step is to ensure that the system can effectively execute the workflow and complete various tasks using limited computing power resources.

[0059] First, the system evaluates the computing power of the current system, including aspects such as processor performance, memory capacity, and the number of GPUs / CPUs. This will provide a benchmark for the system to determine the maximum computing power that the system can provide.

[0060] Secondly, the system evaluates each model type according to the computing power requirements of each model type in the preset model library. Different models may have different computational complexities and resource consumptions. For example, a deep learning model may require more GPU resources, while a simple statistical model may only require CPU resources.

[0061] Next, the system determines the appropriate algorithm model and computing power allocation strategy according to the computing power situation and the model information in the model library. This means that the system will select a model suitable for the current computing power situation and formulate a reasonable computing power allocation plan to ensure the best utilization of the system's resources.

[0062] When constructing the task workflow, the system organizes each model into a complete workflow according to the model matching relationship in the task structure, in accordance with the execution order and dependencies of the tasks. This can ensure that the tasks are executed in the expected order and manner.

[0063] Finally, according to the computing power allocation strategy, the system allocates the corresponding system computing power to each algorithm model in the task workflow. This may involve considerations such as parallel processing of different models, resource optimization, and task scheduling, in order to maximize the operating efficiency and performance of the system.

[0064] Through step S102, the system can determine the appropriate algorithm model and computing power allocation strategy according to the current computing power situation and task requirements, construct an effective task workflow, and provide appropriate system computing power for each model, thereby achieving the efficient execution of tasks.

[0065] Step S103: Monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow. When the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0066] Optionally, in this embodiment, in step S103, the system monitors the real-time computing power occupancy of each algorithm model during the operation of the task workflow. The purpose of this step is to ensure that the system can promptly perceive changes in computing power occupancy and make corresponding adjustments to ensure that the task can be effectively executed within the preset computing power range.

[0067] First, the system continuously collects and monitors the real-time computing power occupancy data of each algorithm model in the task workflow. This data may involve the usage of resources such as CPU, GPU, and memory, as well as the computing load and running status of each model.

[0068] Second, the system compares the collected real-time computing power occupancy data with the preset computing power threshold. When the real-time computing power occupancy of a certain algorithm model exceeds the preset threshold, the system triggers an update of the computing power allocation strategy.

[0069] Next, the system updates the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow. This may include re-evaluating the priorities of each model and adjusting the system computing power allocation plan to ensure that the system can meet the execution requirements of the task.

[0070] Finally, according to the updated computing power allocation strategy, the system dynamically adjusts the system computing power of the algorithm models to be run. This may involve real-time monitoring and adjustment of the running status of the models to ensure the reasonable utilization of system resources and the smooth execution of the task.

[0071] Through step S103, the system can timely monitor and adjust the computing power allocation strategy during the operation of the task workflow to cope with real-time changes in computing power occupancy, ensuring the effective utilization of system resources and the efficient execution of tasks.

[0072] As can be seen from the above description, the method for automatically generating a workflow and allocating computing power based on artificial intelligence provided by the embodiment of the present application can determine the corresponding task requirements and the model types corresponding to the task requirements through workflow generation instructions and a preset natural language analysis model, and determine the corresponding task structure according to the dependencies between the task requirements and each model type; determine the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, construct a task workflow according to the model matching relationship in the algorithm model and the task structure, and schedule the corresponding system computing power for each algorithm model in the task workflow; when the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0073] In an embodiment of the artificial intelligence-based workflow automatic generation and computing power allocation method of the present application, refer to Figure 2 , and it may specifically include the following content:

[0074] Step S201: Receive the workflow generation instruction sent by the user, parse the workflow generation instruction, and obtain the corresponding structured data;

[0075] Step S202: Input the structured data into a preset pre-trained language model, obtain the task requirements output by the pre-trained language model, and determine the model type corresponding to the task requirements according to the preset association rules.

[0076] Optionally, in this embodiment, in step S201, the system will first receive the workflow generation instruction sent by the user and parse these instructions. The purpose of this step is to convert the natural language instructions input by the user into structured data that can be understood and processed by the computer, and prepare for the subsequent extraction of task requirements and determination of the model type.

[0077] The process of parsing the workflow generation instruction involves natural language processing and text parsing techniques. The system will parse the instructions input by the user through techniques such as word segmentation, part-of-speech tagging, and syntactic analysis, and convert them into a structured data form recognizable by the computer, such as a tree structure, a relationship graph, etc.

[0078] In step S202, the system inputs the parsed structured data into a preset pre-trained language model for processing. The pre-trained language model is usually a deep learning model trained based on a large-scale text corpus, which can understand and infer natural language, so as to extract the semantic information and task requirements therein.

[0079] Through the pre-trained language model, the system can convert the structured data into task requirements. The output of this step is the semantic understanding of the user input instructions and the expression of task requirements, providing a basis for the subsequent determination of the model type.

[0080] At the same time, the system will also match the task requirements with the predefined model types according to the preset association rules. These association rules may be based on domain knowledge, experience summary, or pre-set logical rules, and are used to guide the system to determine the appropriate model type according to the task requirements.

[0081] For example, if the workflow generation instruction sent by the user is "perform object detection and image segmentation on an image", after parsing in step S201, the system will obtain a structured data representation of "image -> object detection -> image segmentation". Then, through step S202, the system inputs the structured data into a pre-trained language model to obtain an expression of the task requirements, such as "detect objects from the image and segment the image". Finally, the system determines the corresponding model types according to the preset association rules, which may be an object detection model and an image segmentation model.

[0082] In an embodiment of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in the present application, refer to Figure 3 , it may specifically include the following content:

[0083] Step S301: Determine the corresponding task execution order according to the dependency relationships between the model types;

[0084] Step S302: Determine the corresponding task structure according to the task requirements and the task execution order.

[0085] Optionally, in this embodiment, in step S301, the system will determine the task execution order according to the dependency relationships between the various model types. This step is to ensure that each task in the workflow can be executed in the correct order to meet the task requirements and ensure the smooth operation of the entire process.

[0086] The dependency relationships between model types generally include two aspects: data dependency and logical dependency. Data dependency refers to the situation where the input data of some models is provided by the output data of other models, while logical dependency refers to the situation where some models can only be executed after other models have been completed.

[0087] The system will build a task execution graph or a task execution sequence based on these dependency relationships to ensure that each task can be correctly executed on the premise of meeting its input data requirements and execution conditions. This can avoid execution order problems caused by data dependency and logical dependency, and ensure the correctness and effectiveness of the entire task process.

[0088] In step S302, the system will determine the structure of the entire task according to the task requirements and the determined task execution order. This step combines the task requirements and the task execution order to form a complete task structure graph or a task flow graph, clearly showing the position and execution order of each task in the entire process.

[0089] The task structure usually includes information such as the input, output, execution order of tasks, and the association relationships between various tasks. Through the task structure, the role and execution order of each task in the entire workflow can be clearly understood, which helps in task management and monitoring, and improves the execution efficiency and stability of the workflow.

[0090] For example, if image feature extraction needs to be performed first in the workflow, and then object detection and image segmentation are carried out, the system will determine the task execution order according to the dependency relationships between model types. It may first execute the image feature extraction task, and then sequentially execute the object detection and image segmentation tasks. Then, according to the task requirements and task execution order, the entire task structure is determined to form a clear task process, ensuring that each task is executed in the correct order and conditions.

[0091] In an embodiment of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in the present application, referring to Figure 4 , it may specifically include the following content:

[0092] Step S401: Determine the corresponding computing power requirements according to the model complexity of each model type in the preset model library;

[0093] Step S402: Determine the resource utilization rate of each model type according to the computing power requirements of each model type and the computing resources of the current system's hardware devices, and determine the computing power allocation strategy according to the maximized resource utilization rate.

[0094] Optionally, in this embodiment, in step S401, the system will determine the corresponding computing power requirements according to the model complexity of each model type in the preset model library. Model complexity can usually be measured by indicators such as the number of model parameters, the amount of computation, or the complexity of the network structure. More complex models usually require more computing resources for training and inference, so their computing power requirements will also increase accordingly.

[0095] Through the model complexity information in the preset model library, the system can classify different types of models and determine the corresponding computing power requirements according to their complexity. This can provide a basis and foundation for the subsequent computing power allocation strategy, ensuring that the system can reasonably utilize computing resources for task processing.

[0096] In step S402, the system will determine the resource utilization rate of each model type according to the computing power requirements of each model type and the computing resources of the current system's hardware devices. The resource utilization rate can be obtained by calculating the ratio of the computing resources of the current system to the computing power requirements of each model type. By calculating the resource utilization rate, the resource utilization situation of the current system for different types of models can be evaluated, and the model types with higher resource utilization rates in the system can be found.

[0097] After determining the resource utilization rate, the system will formulate a computing power allocation strategy based on the principle of maximizing resource utilization rate. This means that the system will make the best use of the computing resources of the current system to maximize the utilization rate. This can ensure that the system can efficiently process tasks and make full use of the performance of hardware devices.

[0098] For example, if the computing power requirement of a certain model type in the system is low and the computing resources of the current system are relatively sufficient, then the system can allocate more computing resources to this model type to improve its resource utilization rate. On the contrary, if the computing power requirement of a certain model type is high and the computing resources of the system are limited, then the system may limit the resource allocation of this model type to avoid resource waste and overloading of the system load.

[0099] In an embodiment of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in this application, refer to Figure 5 , it may specifically include the following content:

[0100] Step S501: Convert the model matching relationship in the task structure into a task workflow directed graph, where the nodes in the task workflow directed graph represent algorithm models, and the edges in the task workflow directed graph represent the dependency relationships between algorithm models;

[0101] Step S502: Construct a task workflow based on the algorithm models and the task workflow directed graph, and schedule the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy determined by maximizing the resource utilization rate.

[0102] Optionally, in this embodiment, in step S501, the system converts the model matching relationship in the task structure into a task workflow directed graph. The task workflow directed graph is a graph structure used to describe the task execution order and dependency relationships, where the nodes represent algorithm models and the edges represent the dependency relationships between algorithm models. By converting the task structure into a task workflow directed graph, the system can clearly show the execution order and dependency relationships between each algorithm model, which is helpful for subsequent task scheduling and execution.

[0103] In step S502, the system constructs a task workflow based on the algorithm models and the task workflow directed graph. The task workflow is a task execution sequence composed of each algorithm model, which is constructed according to the order of the nodes and the connection relationships of the edges in the task workflow directed graph. The system schedules the corresponding system computing power for each algorithm model in the task workflow according to the previously determined computing power allocation strategy. This can ensure that the system can efficiently execute the task workflow and meet the task requirements of users on the premise of maximizing resource utilization rate.

[0104] For example, assume that a certain task structure includes three models: image processing, feature extraction, and object detection. The output of the image processing model serves as the input of the feature extraction model, and the output of the feature extraction model serves as the input of the object detection model. Such a model matching relationship can be converted into the relationship between nodes and edges in a task workflow directed graph. Then, the system constructs a task workflow based on the task workflow directed graph and schedules the corresponding system computing power for each model according to the computing power allocation strategy that maximizes resource utilization. In this way, the task workflow can be efficiently executed, and the task processes of image processing, feature extraction, and object detection can be completed.

[0105] In an embodiment of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in the present application, referring to Figure 6 , it may specifically include the following content:

[0106] Step S601: When the real-time computing power occupancy data exceeds a preset computing power threshold, update the task priority of the algorithm model to be run in the task workflow according to the computing power requirements of the algorithm model to be run;

[0107] Step S602: Update the computing power allocation strategy according to the current system computing power and the algorithm model to be run with the updated task priority.

[0108] Optionally, in this embodiment, in step S601, the system monitors the real-time computing power occupancy data of each algorithm model during the operation of the task workflow. If the real-time computing power occupancy of a certain algorithm model exceeds the preset computing power threshold, the system will update its task priority according to the computing power requirements of the model. Specifically, if the computing power requirements of a certain algorithm model are high and the real-time computing power occupancy exceeds the threshold, then the task priority of this model will be increased, enabling it to obtain system resources faster and complete the task as soon as possible. By increasing the task priority of the algorithm model to be run, the system can effectively schedule resources and improve the resource utilization rate and task execution efficiency of the system.

[0109] In step S602, the system updates the algorithm model to be run according to the current system computing power and the updated task priority of the algorithm model, and accordingly updates the computing power allocation strategy. Specifically, the system will readjust the resource allocation of each algorithm model according to the updated task priority, and preferentially allocate system computing power to the model with a high priority to ensure that the system resources are fully utilized and the task can be completed on time. This can enable the system to maintain the stability and efficiency of the system while dynamically adjusting the computing power allocation, and adapt to the changes in resource requirements under different task loads.

[0110] In an embodiment of the method for automatically generating a workflow and allocating computing power based on artificial intelligence in the present application, referring to Figure 7 , it may specifically include the following content:

[0111] Step S701: Determine the adjustment value when the task priority of the to-be-run algorithm model is updated;

[0112] Step S702: Adjust the system computing power scheduled for the to-be-run algorithm model by the current system computing power according to the adjustment value.

[0113] Optionally, in this embodiment, in step S701, the system determines the adjustment value when the task priority of the to-be-run algorithm model is updated. This adjustment value is the result of comprehensive consideration of factors such as the computing power occupancy of the current system, the computing power requirements of the to-be-run algorithm model, and the task scheduling strategy of the system. Specifically, if the computing power requirements of a certain algorithm model are high and the real-time computing power occupancy exceeds the preset computing power threshold, the system will determine the task priority adjustment value of the model according to the preset adjustment rules and priority update strategy. For example, the size of the adjustment value can be determined according to factors such as the degree of exceeding the threshold of the real-time computing power occupancy and the importance of the to-be-run algorithm model, so as to more accurately reflect the change of the current task priority.

[0114] In step S702, the system adjusts the system computing power scheduled for the to-be-run algorithm model by the current system according to the adjustment value determined in step S701. Specifically, the system adjusts the system resource allocation ratio of the to-be-run algorithm model according to the size of the adjustment value. If the adjustment value is positive, the system resource allocation ratio of the to-be-run algorithm model is increased to improve its task priority and speed up the task completion speed; if the adjustment value is negative, its system resource allocation ratio is reduced to lower its task priority and release system resources, so that other tasks can obtain more system resources. In this way, the system can flexibly adjust the computing power allocation strategy according to the real-time task requirements and system resource conditions to ensure the stability and efficiency of the system.

[0115] In a specific example of this application, when considering an intelligent security system that includes functions such as face recognition, safety helmet wearing detection, personnel crossing detection, and work clothing wearing detection, we need a complete workflow to manage and execute these tasks. The following is a detailed description of how each step is applied to such a system:

[0116] In step S101, the system first receives a workflow generation instruction from the user, for example: "Monitor at a construction site." The system determines the task requirements and the corresponding model types according to this instruction and the preset natural language analysis model. In this example, the task requirements include construction site monitoring, and the model types may include face recognition, safety helmet wearing detection, etc.

[0117] In step S102, the system determines the algorithm model and the computing power allocation strategy based on the current system computing power and the computing power requirements of each model type in the preset model library. For example, a face recognition model may require more computing power resources, while a safety helmet wearing detection model may require fewer resources. The system constructs a task workflow according to these requirements and the computing power allocation strategy, and allocates the corresponding system computing power to each model.

[0118] In steps S201 and S202, the system parses the workflow generation instruction of the user to obtain structured data, and inputs it into the preset pre-trained language model to obtain the task requirements and the corresponding model types. For example, the system may obtain the task requirements: "construction site monitoring" and the corresponding model types: face recognition, safety helmet wearing detection, etc.

[0119] In steps S301 and S302, the system determines the task execution order according to the dependency relationship between each model type, and determines the task structure according to the task requirements. For example, the system may determine that face recognition needs to be executed before safety helmet wearing detection. Then, the system constructs a task workflow according to this task structure.

[0120] In steps S401 and S402, the system determines the corresponding computing power requirements according to the model complexity of each model type in the preset model library, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy of maximizing resource utilization. For example, according to the complexity of the face recognition model and the complexity of the safety helmet wearing detection model, the system determines their respective computing power requirements and allocates the corresponding computing power according to the hardware device resources of the system.

[0121] In steps S501 and S502, the system converts the model matching relationship in the task structure into a directed graph of the task workflow, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy of maximizing resource utilization. For example, the system determines that the face recognition model needs to be executed before the safety helmet wearing detection model according to the directed graph of the task workflow, and allocates appropriate system resources to each model according to the computing power allocation strategy.

[0122] In steps S601 and S602, when the system monitors the computing power occupancy data in real time, if it exceeds the preset computing power threshold, it will update the task priority of the algorithm model to be run, and readjust the system computing power according to the updated priority. For example, if the computing power requirement of the face recognition model suddenly increases, the system will correspondingly increase its priority and reallocate the system resources.

[0123] In steps S701 and S702, the system determines the adjustment values of the task priorities of the algorithm models to be run, and accordingly adjusts the system computing power scheduled for the algorithm models to be run based on these adjustment values. For example, if the system decides to increase the task priority of the face recognition model, the system will accordingly adjust its task priority and reallocate system resources.

[0124] Through these steps, the system can effectively manage and execute various intelligent analysis tasks in the intelligent security system, ensuring the stable and efficient operation of the system.

[0125] To effectively improve the efficiency and accuracy of workflow generation, this application provides an embodiment of an artificial intelligence-based workflow automatic generation and computing power allocation device for implementing all or part of the artificial intelligence-based workflow automatic generation and computing power allocation method. Refer to Figure 8 The artificial intelligence-based workflow automatic generation and computing power allocation device specifically includes the following:

[0126] A task structure determination module 10, configured to receive a workflow generation instruction sent by a user, determine corresponding task requirements and model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine a corresponding task structure according to the task requirements and the dependency relationship between the model types;

[0127] A system computing power scheduling module 20, configured to determine corresponding algorithm models and computing power allocation strategies according to the current system computing power and the computing power requirements of each model type in a preset model library, construct a task workflow according to the model matching relationship between the algorithm models and the task structure, and schedule corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy;

[0128] A computing power allocation update module 30, configured to monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow, and when the real-time computing power occupancy data exceeds a preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0129] As can be seen from the above description, the workflow automatic generation and computing power allocation device based on artificial intelligence provided by the embodiments of the present application can determine the corresponding task requirements and the model types corresponding to the task requirements through workflow generation instructions and a preset natural language analysis model, and determine the corresponding task structure according to the task requirements and the dependency relationships between the model types; determine the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, construct a task workflow according to the model matching relationship in the algorithm model and the task structure, and schedule the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0130] At the hardware level, in order to effectively improve the efficiency and accuracy of workflow generation, the present application provides an embodiment of an electronic device for implementing all or part of the content in the above-mentioned artificial intelligence-based workflow automatic generation and computing power allocation method. The electronic device specifically includes the following:

[0131] A processor, a memory, a communication interface, and a bus; wherein, the processor, the memory, and the communication interface complete communication with each other through the bus; the communication interface is used to implement information transmission between the artificial intelligence-based workflow automatic generation and computing power allocation device and related devices such as the core business system, the user terminal, and the relevant database. The logic controller can be a desktop computer, a tablet computer, a mobile terminal, etc., and this embodiment is not limited thereto. In this embodiment, the logic controller can be implemented with reference to the embodiments of the artificial intelligence-based workflow automatic generation and computing power allocation method and the embodiments of the artificial intelligence-based workflow automatic generation and computing power allocation device, and the content is incorporated herein, and the repeated parts will not be elaborated.

[0132] It can be understood that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0133] In practical applications, part of the method for automatically generating a workflow and allocating computing power based on artificial intelligence can be executed on the side of the electronic device as described above, or all operations can be completed in the client device. Specifically, it can be selected according to the processing capacity of the client device and the limitations of the user usage scenario, etc. This application does not make any limitations in this regard. If all operations are completed in the client device, the client device may further include a processor.

[0134] The above-mentioned client device may have a communication module (i.e., a communication unit), which can communicate with a remote server to achieve data transmission with the server. The server may include a server on the side of the task scheduling center. In other implementation scenarios, it may also include a server of an intermediate platform, such as a server of a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or may include a server cluster composed of multiple servers, or a server structure of a distributed device.

[0135] Figure 9 It is a schematic block diagram of the system composition of the electronic device 9600 according to an embodiment of the present application. As Figure 9 shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It should be noted that this Figure 9 is exemplary; other types of structures can also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0136] In one embodiment, the function of the method for automatically generating a workflow and allocating computing power based on artificial intelligence can be integrated into the central processing unit 9100. Among them, the central processing unit 9100 can be configured to perform the following controls:

[0137] Step S101: Receive a workflow generation instruction sent by a user, determine corresponding task requirements and the model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine a corresponding task structure according to the dependencies between the task requirements and each of the model types;

[0138] Step S102: Determine a corresponding algorithm model and a computing power allocation strategy according to the current system computing power and the computing power requirements of each of the model types in the preset model library, construct a task workflow according to the model matching relationship in the algorithm model and the task structure, and schedule corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy;

[0139] Step S103: Monitor the real-time computing power occupancy data of each algorithm model during the running of the task workflow. When the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation policy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation policy.

[0140] As can be seen from the above description, the electronic device provided in the embodiment of the present application determines the corresponding task requirements and the model types corresponding to the task requirements through the workflow generation instruction and the preset natural language analysis model, and determines the corresponding task structure according to the dependencies between the task requirements and various model types; determines the corresponding algorithm models and computing power allocation policies according to the current system computing power and the computing power requirements of various model types in the preset model library, constructs a task workflow according to the model matching relationship between the algorithm models and the task structure, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation policy; when the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation policy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0141] In another embodiment, the workflow automatic generation and computing power allocation device based on artificial intelligence can be separately configured from the central processing unit 9100. For example, the workflow automatic generation and computing power allocation device based on artificial intelligence can be configured as a chip connected to the central processing unit 9100, and the functions of the workflow automatic generation and computing power allocation method based on artificial intelligence are realized through the control of the central processing unit.

[0142] As Figure 9 shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It should be noted that the electronic device 9600 does not necessarily have to include Figure 9 all the components shown in Figure 9 ; in addition, the electronic device 9600 may further include

[0143] components not shown in Figure 9 ; reference can be made to the prior art.

[0144] Among them, the memory 9140 can be, for example, one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices. It can store the above information related to failures, and in addition, it can also store programs for executing relevant information. And the central processing unit 9100 can execute the program stored in the memory 9140 to achieve information storage or processing, etc.

[0145] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to supply power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display can be, for example, an LCD display, but is not limited thereto.

[0146] The memory 9140 can be a solid-state memory. For example, a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It can also be a memory that stores information even when powered off, can be selectively erased, and has more data. An example of this memory is sometimes referred to as an EPROM, etc. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 can include an application / function storage unit 9142, which is used to store application programs and function programs or the processes for operating the electronic device 9600 through the central processing unit 9100.

[0147] The memory 9140 can also include a data storage unit 9143, which is used to store data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 can include various drivers of the electronic device for communication functions and / or for executing other functions of the electronic device (such as a messaging application, an address book application, etc.).

[0148] The communication module 9110 is a transmitter / receiver 9110 that transmits and receives signals via the antenna 9111. The communication module (transmitter / receiver) 9110 is coupled to the central processing unit 9100 to provide input signals and receive output signals, which can be the same as in the case of a conventional mobile communication terminal.

[0149] Based on different communication technologies, in the same electronic device, multiple communication modules 9110 can be provided, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module, etc. The communication module (transmitter / receiver) 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby implementing normal telecommunication functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is also coupled to a central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0150] Embodiments of the present application also provide a computer-readable storage medium capable of implementing all steps of the artificial intelligence-based workflow automatic generation and computing power allocation method with the execution subject being a server or a client in the above embodiments. A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, it implements all steps of the artificial intelligence-based workflow automatic generation and computing power allocation method with the execution subject being a server or a client in the above embodiments. For example, when the processor executes the computer program, the following steps are implemented:

[0151] Step S101: Receive a workflow generation instruction sent by a user, determine corresponding task requirements and model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine a corresponding task structure according to the task requirements and the dependency relationship between the model types.

[0152] Step S102: Determine a corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in a preset model library, construct a task workflow according to the model matching relationship in the algorithm model and the task structure, and schedule corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy.

[0153] Step S103: Monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow. When the real-time computing power occupancy data exceeds a preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0154] As can be seen from the above description, the computer-readable storage medium provided by the embodiments of the present application determines the corresponding task requirements and the model types corresponding to the task requirements through workflow generation instructions and a preset natural language analysis model, and determines the corresponding task structure according to the task requirements and the dependency relationships between the model types; determines the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, constructs a task workflow according to the matching relationship between the algorithm model and the models in the task structure, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds the preset computing power threshold, updates the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0155] An embodiment of the present application further provides a computer program product that can implement all the steps in the above-mentioned artificial intelligence-based workflow automatic generation and computing power allocation method with the execution subject being a server or a client. When the computer program / instructions are executed by a processor, the steps of the above-mentioned artificial intelligence-based workflow automatic generation and computing power allocation method are implemented. For example, the computer program / instructions implement the following steps:

[0156] Step S101: Receive a workflow generation instruction sent by a user, determine the corresponding task requirements and the model types corresponding to the task requirements according to the workflow generation instruction and a preset natural language analysis model, and determine the corresponding task structure according to the task requirements and the dependency relationships between the model types;

[0157] Step S102: Determine the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, construct a task workflow according to the matching relationship between the algorithm model and the models in the task structure, and schedule the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy;

[0158] Step S103: Monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow. When the real-time computing power occupancy data exceeds the preset computing power threshold, update the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, and dynamically adjust the system computing power of the algorithm models to be run according to the updated computing power allocation strategy.

[0159] As can be seen from the above description, the computer program product provided by the embodiments of the present application determines the corresponding task requirements and the model types corresponding to the task requirements through workflow generation instructions and a preset natural language analysis model, and determines the corresponding task structure according to the task requirements and the dependency relationships between the model types; determines the corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in the preset model library, constructs a task workflow according to the model matching relationship between the algorithm model and the task structure, and schedules the corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; when the real-time computing power occupancy data exceeds the preset computing power threshold, updates the computing power allocation strategy according to the current system computing power and the computing power requirements of the algorithm models to be run in the task workflow, thereby effectively improving the efficiency and accuracy of workflow generation.

[0160] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, apparatus, or computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0162] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the steps of the function specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 in one block or a plurality of blocks.

[0164] In the present invention, specific embodiments are used to elaborate the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention. At the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for automatically generating a workflow and allocating computing power based on artificial intelligence, characterized in that, The method includes: Receiving a workflow generation instruction sent by a user, parsing the workflow generation instruction to obtain corresponding structured data; Inputting the structured data into a preset pre-trained language model to obtain task requirements output by the pre-trained language model, determining a corresponding model type according to a preset association rule, determining a corresponding task execution order according to the dependency relationship between the model types, and determining a corresponding task structure according to the task requirements and the task execution order; Determining a corresponding algorithm model and a computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in a preset model library, including determining corresponding computing power requirements according to the model complexity of each model type in the preset model library; determining the resource utilization rate of each model type according to the computing power requirements of each model type and the computing resources of the hardware devices of the current system, and determining a computing power allocation strategy according to the maximized resource utilization rate, constructing a task workflow according to the matching relationship between the algorithm model and the models in the task structure, and scheduling corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; Monitoring the real-time computing power occupancy data of each algorithm model during the operation of the task workflow, and when the real-time computing power occupancy data of the algorithm model exceeds a preset computing power threshold, updating the task priority of the algorithm model to be run according to the computing power requirements of the algorithm model to be run in the task workflow; Updating the computing power allocation strategy according to the current system computing power and the algorithm model to be run with the updated task priority, and dynamically adjusting the system computing power of the algorithm model to be run according to the updated computing power allocation strategy.

2. The method for automatically generating a workflow and allocating computing power based on artificial intelligence according to claim 1, wherein The constructing a task workflow according to the matching relationship between the algorithm model and the models in the task structure, and scheduling corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy includes: Converting the model matching relationship in the task structure into a directed graph of the task workflow, where the nodes in the directed graph of the task workflow represent algorithm models, and the edges in the directed graph of the task workflow represent the dependency relationship between algorithm models; Constructing a task workflow according to the algorithm model and the directed graph of the task workflow, and scheduling corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy determined by the maximized resource utilization rate.

3. The method for automatically generating a workflow and allocating computing power based on artificial intelligence according to claim 1, wherein The updating the computing power allocation strategy according to the current system computing power and the algorithm model to be run with the updated task priority includes: Determining an adjustment value when the task priority of the algorithm model to be run is updated; Correspondingly adjusting the system computing power scheduled for the algorithm model to be run according to the adjustment value.

4. An artificial intelligence-based workflow automatic generation and computing power allocation device, characterized in that, The device includes: A task structure determination module, configured to receive a workflow generation instruction sent by a user, parse the instruction of the workflow generation instruction to obtain corresponding structured data, input the structured data into a preset pre-trained language model, obtain task requirements output by the pre-trained language model, determine a corresponding model type according to a preset association rule, determine a corresponding task execution order according to the dependency relationship between the model types, and determine a corresponding task structure according to the task requirements and the task execution order; A system computing power scheduling module, configured to determine a corresponding algorithm model and computing power allocation strategy according to the current system computing power and the computing power requirements of each model type in a preset model library, including determining corresponding computing power requirements according to the model complexity of each model type in the preset model library; determining the resource utilization rate of each model type according to the computing power requirements of each model type and the computing resources of the hardware devices of the current system, and determining a computing power allocation strategy according to the maximized resource utilization rate, constructing a task workflow according to the matching relationship between the algorithm model and the models in the task structure, and scheduling corresponding system computing power for each algorithm model in the task workflow according to the computing power allocation strategy; A computing power allocation update module, configured to monitor the real-time computing power occupancy data of each algorithm model during the operation of the task workflow, when the real-time computing power occupancy data of the algorithm model exceeds a preset computing power threshold, update the task priority of the algorithm model to be run according to the computing power requirements of the algorithm model to be run in the task workflow, update the computing power allocation strategy according to the current system computing power and the algorithm model to be run with the updated task priority, and dynamically adjust the system computing power of the algorithm model to be run according to the updated computing power allocation strategy.

5. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the steps of the method for automatically generating a workflow and allocating computing power based on artificial intelligence according to any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the steps of the method for automatically generating a workflow and allocating computing power based on artificial intelligence according to any one of claims 1 to 3 are implemented.

Images