Multi-task resource intelligent distribution system based on deep learning

Abstract

The invention discloses a multi-task resource intelligent allocation system based on deep learning, and belongs to the technical field of multi-task resource intelligent allocation. Comprising a data acquisition preprocessing module, a knowledge graph construction module, a context sensing module, a self-adaptive task identification module, a resource state sensing module, an intelligent dynamic resource allocation module and a feedback optimization adjustment module, an intelligent dynamic resource allocation module obtains an identified task list and characteristics thereof from a self-adaptive task identification module and obtains latest resource state information from a resource state sensing module, an optimal resource allocation scheme is formulated, in the actual allocation process, an ant colony algorithm is adopted to find an approximate optimal solution, and the optimal resource allocation efficiency is improved. According to the resource allocation method, the resource allocation scientificity and rationality are ensured, the demand of the current task is considered, the dynamic optimal allocation of the resources is realized, and in this way, the resource waste can be minimized and the resource use efficiency can be improved while the task is ensured to be completed on time.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-task resource intelligent allocation, and specifically refers to a multi-task resource intelligent allocation system based on deep learning. Background Art

[0002] With the development of cloud computing, big data, and artificial intelligence, how to efficiently manage and allocate limited computing resources has become a key issue. Traditional resource allocation methods usually rely on preset rules or static models, which are difficult to adapt to rapidly changing workload requirements, resulting in resource waste or inefficient task processing. In addition, traditional methods often lack effective global optimization strategies when facing concurrent execution of multiple tasks. However, the existing multi-task resource intelligent allocation still has certain defects. The existing multi-task resource intelligent allocation relies on static task and resource databases, lacks the ability to dynamically understand and update the relationship between tasks and resources in real time, has limitations in task identification and classification, and mainly relies on predefined rules or simple machine learning models. It fails to make full use of contextual information and real-time environmental conditions. When faced with complex and changing task environments, the recognition accuracy is low, the flexibility is insufficient, and it is easy to misclassify or miss key task features. Traditional resource allocation methods are usually based on static rules or simple heuristic algorithms, which fail to fully consider the actual needs of tasks and the real-time status of resources. These methods can often only provide suboptimal solutions and are difficult to cope with possible changes in the future, resulting in resource waste and task delays. To this end, a multi-task resource intelligent allocation system based on deep learning is proposed. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-task resource intelligent allocation system based on deep learning to solve the problems raised in the above background technology.

[0004] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a multi-task resource intelligent allocation system based on deep learning, comprising a data acquisition preprocessing module, a knowledge graph construction module, a context perception module, an adaptive task identification module, a resource status perception module, an intelligent dynamic resource allocation module and a feedback optimization adjustment module; The data collection and preprocessing module is used to collect information about task requirements and resource status from multiple sources and preprocess the collected data; The knowledge graph construction module is used to construct a knowledge graph of task and resource relationships based on the preprocessed data; The context perception module is used to perceive the real-time context status according to the real-time environmental conditions and the knowledge graph information; The adaptive task identification module is used to analyze and classify tasks based on the knowledge graph combined with situational state information; The resource status perception module is used to monitor the status of all available resources in real time according to the data acquisition preprocessing module; The intelligent dynamic resource allocation module is used to formulate optimal resource allocation according to the identified tasks and the status of available resources; The feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and make dynamic adjustments.

[0005] The data acquisition preprocessing module is wirelessly connected to the knowledge graph construction module and the resource status perception module, the knowledge graph construction module is wirelessly connected to the context perception module and the adaptive task identification module, the context perception module is wirelessly connected to the adaptive task identification module, the adaptive task identification module and the resource status perception module are wirelessly connected to the intelligent dynamic resource allocation module, and the intelligent dynamic resource allocation module is wirelessly connected to the feedback optimization and adjustment module.

[0006] Among them, the data collection and preprocessing module collects information about task requirements and resource status from multiple sources and preprocesses the collected data; according to the task type and application scenario, a fixed time interval is set to regularly collect data, and the collected data is denoised and filled with missing values ​​for data cleaning, and different data after data cleaning are converted into a unified format, and numerical data are standardized.

[0007] The knowledge graph construction module constructs a knowledge graph of the relationship between tasks and resources based on the preprocessed data; acquires data from the data collection preprocessing module in real time, and sets the task category , classify the collected data x, and the implementation formula is: In the formula, Indicates that the input x belongs to the task category The probability of represents the adjustment factor, Represents the semantic similarity function, which measures the input x and the task category The semantic similarity of represents the i-th task, x represents the input data, and j represents the index of all possible task categories; Relationship extraction is performed based on the classified task categories, and the implementation formula is: , In the formula, Represents a given task and And input x, the probability of the existence of relation R, and represents the weight parameter, Represents the evaluation task and The probability of co-occurrence in the same context, It represents the measure of the information strength of the input x about the relation R. represents probability; Set the system perception coefficient , the synergy between measurement task identification classification and relation extraction is implemented as:

[0008] In the formula, represents the cooperative perception coefficient, represents the normalization constant, For all possible task pairs To sum, Input x belongs to the task category The probability of Represents a given task and And input x, the probability of the existence of relationship R; create a knowledge graph based on task categories and relationship information, and update the knowledge graph in real time based on real-time data.

[0009] The knowledge graph construction module uses the preprocessed data to construct a knowledge graph of task and resource relationships. By acquiring and classifying the data from the data acquisition preprocessing module in real time, it realizes task category identification and further performs relationship extraction to create a dynamically updated knowledge graph. It not only captures explicit entity relationships, but also reveals implicit association patterns and provides deep semantic information support. By introducing the collaborative perception coefficient, it can measure the synergy between task identification classification and relationship extraction, thereby optimizing the structure and content of the knowledge graph and helping to improve the accuracy of task parsing and the rationality of resource allocation.

[0010] Among them, the situational awareness module perceives the real-time situational status based on the real-time environmental conditions and knowledge graph information; obtains the latest data in real time through integrated sensors, API interfaces and other sources, and extracts situational information and rules from the knowledge graph, integrates the real-time data with the extracted situational information to form a comprehensive situational data set, analyzes the integrated situational data information, identifies the key status of the current situation based on the situational analysis results, evaluates the identified status, and automatically adjusts the situational category and judgment rules based on the latest data changes.

[0011] The adaptive task identification module analyzes and classifies tasks according to the knowledge graph combined with the situational status information; receives real-time situational status information from the situational awareness module, and obtains the latest knowledge graph data from the knowledge graph construction module. Let the task description be , the knowledge graph is G, and the task description Combined with the knowledge graph G information, the implicit representation of the task is generated, and the implementation formula is:

[0012] In the formula, The implicit representation vector representing the task, represents the task description, G represents the knowledge graph, Represents the task representation function based on the knowledge graph; Let the situation state be c, and calculate the task category The similarity with situation c is calculated, and the weight of each task category is dynamically adjusted. The implementation formula is: , In the formula, represents the task weight of situational awareness, represents the adjustment factor, Represents the context similarity function, evaluating the task category The matching degree with the current situation c, represents the i-th type of task, and c represents the current situation state.

[0013] Among them, the adaptive task identification module, assuming the parameter matrix is , according to the implicit representation vector of the task amount Projection to each task category The parameter matrix Get , combined with the task weight of context awareness, and added to the task representation projection result to form a comprehensive score , converting the combined scores of all categories into a probability distribution , the implementation formula is:

[0014] In the formula, Represents a given task description and c, belonging to the task category The probability of represents the adjustment factor, Represents a task representation projection, Representation and Task Categories The relevant parameter matrix, represents the task weight of situation awareness; The impact of each task classification is reflected through context information, and the implementation formula is: , In the formula, represents the impact of contextual information on task classification, represents the normalization constant, Represents all possible task categories Perform the summation.

[0015] The adaptive task identification module parses and classifies tasks by combining knowledge graphs and situational status information, significantly improving the accuracy and flexibility of task identification. It first generates an implicit representation vector of the task, then calculates the similarity between the task category and the current situation, and dynamically adjusts the weight of each task category. It not only considers the characteristics of the task itself, but also combines the influence of the current situation. It can more accurately identify and classify tasks in complex environments, and project the task representation onto each task category through the parameter matrix to form a comprehensive score, which is converted into a probability distribution, realizing the probabilistic output of task classification, which helps to improve the efficiency and quality of overall task management.

[0016] Among them, the resource status perception module monitors the status of all available resources in real time according to the data acquisition and preprocessing module; receives real-time resource status data from the data acquisition and preprocessing module, and performs instant evaluation of the key indicators of each resource according to pre-set rules to determine whether its current status is normal, and monitors the resource status in real time.

[0017] Among them, the intelligent dynamic resource allocation module is used to formulate the optimal resource allocation according to the identified tasks and the status of available resources; obtain the identified task list and its characteristics from the adaptive task identification module, and obtain the latest resource status information from the resource status perception module, train the prediction model based on historical data, predict the resource demand change trend in the future period of time, and find the approximate optimal solution through the ant colony algorithm based on the identified tasks and the status of available resources to formulate the optimal resource allocation.

[0018] Among them, the feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and dynamically adjust; status information of each task; collect real-time status information of each task in real time, and continuously obtain the latest resource usage provided by the resource status perception module, and conduct instant evaluation of the current resource allocation effect according to pre-set rules to determine whether the expected goal has been achieved. If the expected goal has not been achieved, the corresponding adjustment measures are automatically triggered. According to the evaluation results and adjustment strategies, the optimal adjustment suggestions are generated, and the resource configuration parameters are dynamically adjusted according to the optimal adjustment suggestions.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses the preprocessed data to construct a knowledge graph of task and resource relationships through a knowledge graph construction module. By acquiring and classifying the data from the data acquisition preprocessing module in real time, a dynamically updated knowledge graph is created, which not only captures explicit entity relationships, but also reveals implicit association patterns. By introducing a collaborative perception coefficient, the synergy between task identification classification and relationship extraction can be measured, thereby optimizing the structure and content of the knowledge graph. 2. The present invention uses an adaptive task identification module to parse and classify tasks by combining knowledge graphs and situational state information, which significantly improves the accuracy and flexibility of task identification. First, the implicit representation vector of the task is generated, and then the similarity between the task category and the current situation is calculated. The weight of each task category is dynamically adjusted. Not only the characteristics of the task itself are considered, but also the influence of the current situation is combined, so that tasks can be more accurately identified and classified in complex environments. The task representation is projected onto each task category through a parameter matrix to form a comprehensive score, which is converted into a probability distribution, realizing the probabilistic output of task classification, which helps to improve the efficiency and quality of overall task management; 3. The present invention uses the intelligent dynamic resource allocation module to obtain the identified task list and its characteristics from the adaptive task identification module, and obtains the latest resource status information from the resource status perception module, and formulates the optimal resource allocation plan. In the actual allocation process, the ant colony algorithm is used to find the approximate optimal solution, which ensures the scientificity and rationality of resource allocation. It not only considers the needs of the current task, but also takes into account possible changes in the future, and realizes the dynamic optimization configuration of resources. In this way, it can minimize resource waste and improve resource utilization efficiency while ensuring that tasks are completed on time; 4. The present invention receives real-time resource status data from the data acquisition and preprocessing module through the resource status perception module, and instantly evaluates the key indicators of each resource according to pre-set rules. It can not only determine whether the current status of the resource is normal, but also monitor the changing trend of the resource status in real time, discover potential problems in time, and respond in the first time to avoid task delays caused by resource failure or shortage, maximize resource utilization, reduce waste, and improve overall operational efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a schematic diagram of the structure of the multi-task resource intelligent allocation system based on deep learning of the present invention; Figure 2 This is a flowchart of the operation of the knowledge graph construction module of the multi-task resource intelligent allocation system based on deep learning of the present invention; Figure 3 This is a flowchart of the operation of the adaptive task identification module of the multi-task resource intelligent allocation system based on deep learning of the present invention; Figure 4 This is a flowchart of the operation of the intelligent dynamic resource allocation module of the multi-task resource intelligent allocation system based on deep learning of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0022] See also Figure 1-Figure 4 As shown, the present invention provides a technical solution: including a data acquisition preprocessing module, a knowledge graph construction module, a context awareness module, an adaptive task identification module, a resource status awareness module, an intelligent dynamic resource allocation module and a feedback optimization adjustment module; The data collection and preprocessing module is used to collect information about task requirements and resource status from multiple sources and preprocess the collected data; The knowledge graph construction module is used to construct a knowledge graph of task and resource relationships based on the preprocessed data; The context perception module is used to perceive the real-time context status according to the real-time environmental conditions and the knowledge graph information; The adaptive task identification module is used to analyze and classify tasks based on the knowledge graph combined with situational state information; The resource status perception module is used to monitor the status of all available resources in real time according to the data acquisition preprocessing module; The intelligent dynamic resource allocation module is used to formulate optimal resource allocation according to the identified tasks and the status of available resources; The feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and make dynamic adjustments.

[0023] The data acquisition preprocessing module is wirelessly connected to the knowledge graph construction module and the resource status perception module, the knowledge graph construction module is wirelessly connected to the context perception module and the adaptive task identification module, the context perception module is wirelessly connected to the adaptive task identification module, the adaptive task identification module and the resource status perception module are wirelessly connected to the intelligent dynamic resource allocation module, and the intelligent dynamic resource allocation module is wirelessly connected to the feedback optimization and adjustment module.

[0024] Among them, the data collection and preprocessing module collects information about task requirements and resource status from multiple sources and preprocesses the collected data; according to the task type and application scenario, a fixed time interval is set to regularly collect data, and the collected data is denoised and filled with missing values ​​for data cleaning, and different data after data cleaning are converted into a unified format, and numerical data are standardized.

[0025] The knowledge graph construction module constructs a knowledge graph of the relationship between tasks and resources based on the preprocessed data; acquires data from the data collection preprocessing module in real time, and sets the task category , classify the collected data x, and the implementation formula is:

[0026] In the formula, Indicates that the input x belongs to the task category The probability of represents the adjustment factor, Represents the semantic similarity function, which measures the input x and the task category The semantic similarity of represents the i-th task, x represents the input data, and j represents the index of all possible task categories; Relationship extraction is performed based on the classified task categories, and the implementation formula is: , In the formula, Represents a given task and And input x, the probability of the existence of relation R, and represents the weight parameter, Represents the evaluation task and The probability of co-occurrence in the same context, It represents the measure of the information strength of the input x about the relation R. represents probability; Set the system perception coefficient , the synergy between measurement task identification classification and relation extraction is implemented as:

[0027] , In the formula, represents the cooperative perception coefficient, represents the normalization constant, For all possible task pairs To sum, Input x belongs to the task category The probability of Represents a given task and And input x, the probability of the existence of relationship R; create a knowledge graph based on task categories and relationship information, and update the knowledge graph in real time based on real-time data.

[0028] The knowledge graph construction module uses the preprocessed data to construct a knowledge graph of task and resource relationships. By acquiring and classifying the data from the data acquisition preprocessing module in real time, it realizes task category identification and further performs relationship extraction to create a dynamically updated knowledge graph. It not only captures explicit entity relationships, but also reveals implicit association patterns and provides deep semantic information support. By introducing the collaborative perception coefficient, it can measure the synergy between task identification classification and relationship extraction, thereby optimizing the structure and content of the knowledge graph and helping to improve the accuracy of task parsing and the rationality of resource allocation.

[0029] Among them, the situational awareness module perceives the real-time situational status based on the real-time environmental conditions and knowledge graph information; obtains the latest data in real time through integrated sensors, API interfaces and other sources, and extracts situational information and rules from the knowledge graph, integrates the real-time data with the extracted situational information to form a comprehensive situational data set, analyzes the integrated situational data information, identifies the key status of the current situation based on the situational analysis results, evaluates the identified status, and automatically adjusts the situational category and judgment rules based on the latest data changes.

[0030] The adaptive task identification module analyzes and classifies tasks according to the knowledge graph combined with the situational status information; receives real-time situational status information from the situational awareness module, and obtains the latest knowledge graph data from the knowledge graph construction module. Let the task description be , the knowledge graph is G, and the task description Combined with the knowledge graph G information, the implicit representation of the task is generated, and the implementation formula is:

[0031] In the formula, The implicit representation vector representing the task, represents the task description, G represents the knowledge graph, Represents the task representation function based on the knowledge graph; Let the situation state be c, and calculate the task category The similarity with situation c is calculated, and the weight of each task category is dynamically adjusted. The implementation formula is: , In the formula, represents the task weight of situational awareness, represents the adjustment factor, Represents the context similarity function, evaluating the task category The matching degree with the current situation c, represents the i-th type of task, and c represents the current situation state.

[0032] Among them, the adaptive task identification module, assuming the parameter matrix is , according to the implicit representation vector of the task amount Projection to each task category The parameter matrix Get , combined with the task weight of context perception, and added to the task representation projection result to form a comprehensive score , converting the comprehensive scores of all categories into probability distributions , the implementation formula is: , In the formula, Represents a given task description and c, belonging to the task category The probability of represents the adjustment factor, Represents a task representation projection, Representation and Task Categories The relevant parameter matrix, represents the task weight of situation awareness; The impact of each task classification is reflected through context information, and the implementation formula is: , In the formula, represents the impact of contextual information on task classification, represents the normalization constant, Represents all possible task categories Perform the summation.

[0033] The adaptive task identification module parses and classifies tasks by combining knowledge graphs and situational status information, significantly improving the accuracy and flexibility of task identification. It first generates an implicit representation vector of the task, then calculates the similarity between the task category and the current situation, and dynamically adjusts the weight of each task category. It not only considers the characteristics of the task itself, but also combines the influence of the current situation. It can more accurately identify and classify tasks in complex environments, and project the task representation onto each task category through the parameter matrix to form a comprehensive score, which is converted into a probability distribution, realizing the probabilistic output of task classification, which helps to improve the efficiency and quality of overall task management.

[0034] Among them, the resource status perception module monitors the status of all available resources in real time according to the data acquisition and preprocessing module; receives real-time resource status data from the data acquisition and preprocessing module, and performs instant evaluation of the key indicators of each resource according to pre-set rules to determine whether its current status is normal, and monitors the resource status in real time.

[0035] Among them, the intelligent dynamic resource allocation module is used to formulate the optimal resource allocation according to the identified tasks and the status of available resources; obtain the identified task list and its characteristics from the adaptive task identification module, and obtain the latest resource status information from the resource status perception module, train the prediction model based on historical data, predict the resource demand change trend in the future period of time, and find the approximate optimal solution through the ant colony algorithm based on the identified tasks and the status of available resources to formulate the optimal resource allocation.

[0036] Among them, the feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and dynamically adjust; status information of each task; collect real-time status information of each task in real time, and continuously obtain the latest resource usage provided by the resource status perception module, and conduct instant evaluation of the current resource allocation effect according to pre-set rules to determine whether the expected goal has been achieved. If the expected goal has not been achieved, the corresponding adjustment measures are automatically triggered. According to the evaluation results and adjustment strategies, the optimal adjustment suggestions are generated, and the resource configuration parameters are dynamically adjusted according to the optimal adjustment suggestions.

[0037] Working principle: By collecting information about task requirements and resource status from multiple sources, the collected data is preprocessed, and a knowledge graph of the relationship between tasks and resources is constructed based on the preprocessed data. Tasks are classified through classification algorithms, and the relationship between tasks is extracted. The knowledge graph is updated using real-time data. According to sources such as integrated sensors and API interfaces, the latest data is obtained in real time, and context-related information and rules are extracted from the knowledge graph. The real-time data is integrated with the context information to form a comprehensive context data set, and the context data is analyzed to identify key states. The context categories and judgment rules are automatically adjusted according to the latest data changes. The adaptive task identification module receives real-time context status information and knowledge graph data, combines the task description and knowledge graph information, generates an implicit representation of the task, calculates the similarity between the task category and the current context, and dynamically adjusts the task category weight, and uses the comprehensive score and probability distribution to determine the task category. The resource status perception module receives real-time resource status data from the data acquisition preprocessing module, and identifies the key indicators of each resource according to pre-set rules. The intelligent dynamic resource allocation module obtains the identified task list and its characteristics and the latest resource status information, uses historical data to train the prediction model, and predicts the trend of resource demand changes in the future. According to the identified tasks and available resource status, it uses optimization algorithms such as ant colony algorithm to find the approximate optimal solution and formulate the optimal resource allocation plan. The feedback optimization and adjustment module collects the real-time status information and resource usage of each task in real time, and conducts an instant evaluation of the current resource allocation effect according to the pre-set rules. If the expected goal is not achieved, the corresponding adjustment measures are automatically triggered. According to the evaluation results and adjustment strategies, the optimal adjustment suggestions are generated, and the resource configuration parameters are dynamically adjusted.

[0038] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0039] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A multi-task resource intelligent allocation system based on deep learning, characterized by: It includes data collection and preprocessing module, knowledge graph construction module, situational awareness module, adaptive task identification module, resource status awareness module, intelligent dynamic resource allocation module and feedback optimization and adjustment module; The data collection and preprocessing module is used to collect information about task requirements and resource status from multiple sources and preprocess the collected data; The knowledge graph construction module is used to construct a knowledge graph of task and resource relationships based on the preprocessed data; The context perception module is used to perceive the real-time context status according to the real-time environmental conditions and the knowledge graph information; The adaptive task identification module is used to analyze and classify tasks based on the knowledge graph combined with situational state information; The resource status perception module is used to monitor the status of all available resources in real time according to the data acquisition preprocessing module; The intelligent dynamic resource allocation module is used to formulate optimal resource allocation according to the identified tasks and the status of available resources; The feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and make dynamic adjustments.

2. The multi-task resource intelligent allocation system based on deep learning according to claim 1 is characterized in that: The data collection and preprocessing module collects information about task requirements and resource status from multiple sources and preprocesses the collected data; according to the task type and application scenario, it sets a fixed time interval to regularly collect data, performs data cleaning by denoising and filling missing values ​​on the collected data, converts different data after data cleaning into a unified format, and standardizes numerical data.

3. The multi-task resource intelligent allocation system based on deep learning according to claim 1 is characterized in that: The knowledge graph construction module constructs a knowledge graph of the relationship between tasks and resources based on the preprocessed data; acquires data from the data collection preprocessing module in real time, and sets the task category , classify the collected data x, and the implementation formula is: ， In the formula, Indicates that the input x belongs to the task category The probability of represents the adjustment factor, Represents the semantic similarity function, which measures the input x and the task category The semantic similarity of represents the i-th task, x represents the input data, and j represents the index of all possible task categories; Relationship extraction is performed based on the classified task categories, and the implementation formula is: ， In the formula, Represents a given task and And input x, the probability of the existence of relation R, and represents the weight parameter, Represents the evaluation task and The probability of co-occurrence in the same context, It represents the measure of the information strength of the input x about the relation R. represents probability; Set the system perception coefficient , the synergy between measurement task identification classification and relation extraction is implemented as: ， In the formula, represents the cooperative perception coefficient, represents the normalization constant, For all possible task pairs To sum, Input x belongs to the task category The probability of Represents a given task and And input x, the probability of the existence of relationship R; create a knowledge graph based on task categories and relationship information, and update the knowledge graph in real time through real-time data.

4. The multi-task resource intelligent allocation system based on deep learning according to claim 1, characterized in that: The context perception module perceives the real-time context status based on the real-time environmental conditions and the knowledge graph information; acquires the latest data in real time through integrated sensors, API interfaces and other sources, extracts context-related information and rules from the knowledge graph, integrates the real-time data with the extracted context information to form a comprehensive context data set, analyzes the integrated context data information, identifies the key status of the current context based on the context analysis results, evaluates the identified status, and automatically adjusts the context category and judgment rules based on the latest data changes.

5. The multi-task resource intelligent allocation system based on deep learning according to claim 1 is characterized in that: The adaptive task identification module analyzes and classifies tasks according to the knowledge graph combined with the situational status information; receives real-time situational status information from the situational awareness module, and obtains the latest knowledge graph data from the knowledge graph construction module. Let the task description be , the knowledge graph is G, and the task description Combined with the knowledge graph G information, the implicit representation of the task is generated, and the implementation formula is: ， In the formula, The implicit representation vector representing the task, represents the task description, G represents the knowledge graph, Represents the task representation function based on the knowledge graph; Let the situation state be c, and calculate the task category The similarity with situation c is calculated, and the weight of each task category is dynamically adjusted. The implementation formula is: ， In the formula, represents the task weight of situational awareness, represents the adjustment factor, Represents the context similarity function, evaluating the task category The matching degree with the current situation c, represents the i-th type of task, and c represents the current situation state.

6. The multi-task resource intelligent allocation system based on deep learning according to claim 5 is characterized in that: The adaptive task identification module assumes that the parameter matrix is , according to the implicit representation vector of the task amount Projection to each task category The parameter matrix Get , combined with the task weight of context awareness, and added to the task representation projection result to form a comprehensive score , converting the combined scores of all categories into a probability distribution , the implementation formula is: ， In the formula, Represents a given task description and c, belonging to the task category The probability of represents the adjustment factor, Represents a task representation projection, Representation and Task Categories The relevant parameter matrix, represents the task weight of situation awareness; The impact of each task classification is reflected through context information, and the implementation formula is: ， In the formula, represents the impact of contextual information on task classification, represents the normalization constant, Represents all possible task categories Perform the summation.

7. The multi-task resource intelligent allocation system based on deep learning according to claim 1, characterized in that: The resource status perception module monitors the status of all available resources in real time according to the data acquisition and preprocessing module; receives real-time resource status data from the data acquisition and preprocessing module, and performs instant evaluation of key indicators of each resource according to pre-set rules to determine whether its current status is normal, and monitors the resource status in real time.

8. The multi-task resource intelligent allocation system based on deep learning according to claim 1, characterized in that: The intelligent dynamic resource allocation module is used to formulate the optimal resource allocation according to the identified tasks and the status of available resources; obtain the identified task list and its characteristics from the adaptive task identification module, and obtain the latest resource status information from the resource status perception module, train the prediction model based on historical data, predict the resource demand change trend in the future, and find the approximate optimal solution through the ant colony algorithm based on the identified tasks and the status of available resources to formulate the optimal resource allocation.

9. The multi-task resource intelligent allocation system based on deep learning according to claim 1, characterized in that: The feedback optimization and adjustment module is used to monitor task execution and resource usage in real time, evaluate resource allocation effects and dynamically adjust the status information of each task; collect the real-time status information of each task in real time, and continuously obtain the latest resource usage provided by the resource status perception module, and conduct instant evaluation of the current resource allocation effect according to pre-set rules to determine whether the expected goal has been achieved. If the expected goal has not been achieved, the corresponding adjustment measures are automatically triggered. According to the evaluation results and adjustment strategies, the optimal adjustment suggestions are generated, and the resource configuration parameters are dynamically adjusted according to the optimal adjustment suggestions.

Images