Control method and device for AI intelligent calculation integration, electronic equipment and storage medium

Abstract

The invention provides an AI intelligent calculation integration control method and device, electronic equipment and a storage medium, and the method comprises the following steps: 1, obtaining AI task request data, carrying out the feature analysis of the AI task request data through a deep learning model, and constructing a multi-dimensional task model; according to the method, the multi-dimensional task model and the resource state model are constructed, and an advanced calculation path generation and optimization algorithm is combined, so that accurate resource allocation is realized, and the resource utilization efficiency is improved; in the task execution process, distributed optical fiber sensing and quantum encryption communication technologies are used for real-time monitoring, a calculation path is adjusted in time by means of a quantum annealing algorithm, waiting and adjusting time caused by link mismatching is shortened, and calculation efficiency is improved; a calculation path is verified and simulated through a quantum simulator, and mechanisms such as changes, knowledge extraction and feedback after tasks are completed and the like are coped with in real time during execution, so that stable execution of the tasks is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and computing technology, and in particular to a control method, device, electronic device and storage medium that integrates AI intelligence and computing. Background Art

[0002] With the rapid development of information technology, artificial intelligence (AI) has been widely used in various fields, such as medical diagnosis, autonomous driving, and financial risk prediction. These application scenarios place extremely high demands on the computing power and efficiency of AI systems. On the one hand, AI algorithms continue to evolve, from traditional machine learning algorithms to today's complex deep learning architectures, and the model complexity and parameter scale are exploding. Taking deep neural networks as an example, the number of network layers can reach hundreds of layers, and the number of parameters is in the hundreds of millions, requiring a large amount of computing resources for training and reasoning. On the other hand, the amount of data is also expanding rapidly. Massive data provides rich information for AI systems, but it also poses huge challenges to data processing and storage. Existing computing control methods have achieved certain results in supporting the operation of AI systems. For example, some rule-based resource scheduling algorithms can guarantee the execution of tasks to a certain extent. However, there are still certain problems: 1. The current resource allocation mechanism is mostly based on experience or simple task priority rules, without in-depth analysis of AI task characteristics. The computing resource requirements of AI tasks vary greatly at different stages. In the early stage of training, a large amount of memory is required to load data, and in the later stage, it is more dependent on the computing power of the processor. The existing methods cannot accurately allocate resources based on this dynamic change, resulting in resource waste or task waiting, affecting the computing efficiency of the system; 2. The AI ​​calculation process includes data reading, preprocessing, model calculation and result output. Currently, the coordination of each link is poor and most of them run independently. If the data reading speed does not match the model calculation speed, the model will be idle waiting for data, or the data preprocessing results will not meet the model calculation requirements and additional adjustment time will be required, reducing the calculation efficiency; 3. Real AI application scenarios are complex and change frequently. Different tasks have different requirements for computing resource types (such as CPU, GPU, TPU, etc.), accuracy and data format. The computing environment is easily affected by hardware failures and network fluctuations. Existing computing control methods lack flexibility and adaptability, and it is difficult to quickly adjust strategies to ensure stable execution of tasks. To this end, a control method, device, electronic device and storage medium integrating AI intelligent computing are proposed. Summary of the invention

[0003] In view of this, the embodiments of the present invention hope to provide a control method, device, electronic device and storage medium integrating AI intelligent computing to solve or alleviate the technical problems existing in the prior art and at least provide a beneficial option.

[0004] In order to solve the above technical problems, a technical solution adopted in this application is: an AI intelligent computing integrated control method, comprising the following steps: Step 1: Obtain AI task request data, perform feature analysis on the AI ​​task request data through a deep learning model, and build a multi-dimensional task model; Step 2: Obtain the hardware and software resource data in the computing system, use quantum-inspired sensing technology to perceive the changes in the microscopic state of resources, combine macro-resource indicators to quantitatively evaluate the resource data, and build a resource state model; Step 3: Based on the multi-dimensional task model and resource status model, generate and optimize the computing path by combining reinforcement learning with the quantum evolutionary algorithm; Step 4: Use a quantum simulator to verify and simulate the computational path and evaluate the computational path by running test cases; Step 5: During the execution of AI tasks, distributed fiber optic sensing and quantum encryption communication technology are used for real-time monitoring, and the calculation path is adjusted according to the monitoring results; Step 6: Use quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms to manage and coordinate task execution data; Step 7: After the task is completed, use machine learning and quantum information processing technology to extract knowledge from the task execution data and generate feedback information.

[0005] As a further preferred embodiment of the present technical solution, in step one, the deep learning model is trained based on historical task data to identify implicit features of the task; the deep learning model is a fusion architecture of a convolutional neural network and a recursive neural network to extract deep features of different types of task data.

[0006] As a further preferred embodiment of the present technical solution, in step 2, the quantum-inspired sensor is based on a superconducting Josephson junction device, analyzes microscopic performance changes of hardware resources by monitoring quantum fluctuations of physical quantities, and uses quantum bit simulation technology to evaluate the quantum optimization space of software resources.

[0007] As a further preferred embodiment of the present technical solution, in step three, the reinforcement learning learns the calculation path by interacting with the system environment feedback model, and the system environment feedback model is based on a quantum reward mechanism; the quantum evolution algorithm evolves the quantum bit population through quantum gate operations and evaluates the quality of the solution according to the fitness function.

[0008] As a further preferred embodiment of the present technical solution, in step 4, the quantum simulator runs test cases of different types, scales and resource status combinations according to a preset hybrid environment parameter configuration, and determines whether the calculation path needs to be adjusted based on the monitoring indicators.

[0009] As a further preferred embodiment of the present technical solution, in step five, the distributed optical fiber sensor obtains task execution information by monitoring changes in optical signals; the quantum encryption communication technology encrypts monitoring data using a quantum key distribution protocol; and the quantum annealing algorithm is used to adjust the calculation path in response to quantum-related abnormal situations.

[0010] As a further preferred embodiment of the present technical solution, in step six, the quantum data storage is based on ion traps or superconducting quantum bits; the quantum state data transmission technology utilizes quantum teleportation technology; the quantum error correction code technology utilizes surface codes; and the quantum machine learning algorithm is used to optimize data preloading and cache calculation paths.

[0011] To solve the above technical problems, another technical solution adopted in this application is: an AI intelligent computing integrated control device, including: a task analysis module, a resource monitoring module, a path generation and optimization module, a path verification and simulation module, a real-time monitoring and adjustment module, a data management and collaboration module, and a feedback and improvement module; The task analysis module is used to receive AI task requests, perform deep feature analysis on AI tasks based on a deep learning model, and build a multi-dimensional task model in combination with traditional task parameters. The deep learning model is trained based on historical task data and is used to identify implicit features of tasks. The resource monitoring module is used to monitor the hardware and software resource status in the computing system in real time, using quantum-inspired sensing technology to perceive changes in the microscopic state of resources, and establishing a resource status model in combination with macroscopic resource indicators; The path generation and optimization module is used to take the task model and the resource state model as input, and generate and optimize the calculation path based on a method combining reinforcement learning and quantum evolution algorithm; The path verification and simulation module is used to verify and simulate the calculation path using a quantum simulator and evaluate the calculation path by running test cases; The real-time monitoring and adjustment module is used to use distributed optical fiber sensing and quantum encryption communication technology to perform real-time monitoring during the execution of AI tasks, and adjust the calculation path according to the monitoring results; The data management and coordination module is used to manage and coordinate task execution data by using quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms; The feedback and improvement module is used to extract knowledge from task execution data and generate feedback information using machine learning and quantum information processing technology after the task is completed.

[0012] To solve the above technical problems, another technical solution adopted by the present application is: an electronic device, the electronic device comprising a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute an AI intelligent computing integrated control method as described in any one of the above items according to the instructions in the program code.

[0013] To solve the above technical problems, another technical solution adopted in the present application is: a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute an AI intelligent computing integrated control method as described in any one of the above items.

[0014] The embodiment of the present invention has the following advantages due to the adoption of the above technical solution: The present invention realizes accurate resource allocation by constructing a multi-dimensional task model and a comprehensive resource status model, and combines advanced computing path generation and optimization algorithms, which can dynamically adjust resources according to the needs of different stages of AI tasks, solve the problem of unreasonable resource allocation, and improve resource utilization efficiency; the present invention uses distributed optical fiber sensing and quantum encryption communication technology to monitor in real time during task execution, and adjusts the computing path in time with the help of quantum annealing algorithm, effectively coordinates the synergy of various links in the computing process, reduces the waiting and adjustment time caused by link mismatch, and improves computing efficiency; the present invention verifies and simulates the computing path through a quantum simulator, responds to changes in real time during execution, and extracts and feedbacks knowledge after the task is completed, etc., which enhances the adaptability to complex task types and changing environments (including hardware failures, network fluctuations, etc.), and ensures the stable execution of tasks. The above overview is for the purpose of the specification only and is not intended to be limited in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be easy to understand by referring to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 A flowchart of an AI intelligent computing integrated control method of the present invention; Figure 2This is a functional module schematic diagram of an AI intelligent computing integrated control device of the present invention; Figure 3 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0018] It should be clear that the following embodiments of the present disclosure are described by specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0019] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0020] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0021] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0022] Figure 1It is a flow chart of an AI intelligent computing integrated control method according to an embodiment of the present invention. It should be noted that if there are substantially the same results, the method of the present application is not limited to Figure 1 The process sequence shown is limited. Figure 1 As shown: A control method integrating AI intelligent computing includes the following steps: Step 1: Obtain AI task request data, perform feature analysis on the AI ​​task request data through a deep learning model, and build a multi-dimensional task model; Specifically, first, the AI ​​system needs to be able to receive task requests from users or clients. These requests may contain information such as task description, input data, expected output, etc. After receiving the task request, the system should perform basic verification to ensure the validity and completeness of the request, for example, check whether the input data meets the format requirements, whether the task description is clear, etc. Then, through the multiple levels of the deep learning model, the deep features of the task are extracted layer by layer. These features may include low-level pixel features, mid-level texture features, and high-level semantic features, etc. In addition to the features extracted by deep learning, it is also necessary to combine traditional task parameters to build a more comprehensive task model. These parameters may include the type, scale, complexity, etc. of the task; finally, the deep features extracted by deep learning are combined with traditional task parameters to construct a task model containing multiple dimensions. This model can more comprehensively reflect the characteristics of the task and provide more accurate information for subsequent task processing.

[0023] Step 2: Obtain the hardware and software resource data in the computing system, use quantum-inspired sensing technology to perceive the changes in the microscopic state of resources, combine macro-resource indicators to quantitatively evaluate the resource data, and build a resource state model; Hardware resource monitoring includes real-time monitoring of parameters such as utilization, temperature, and power consumption of hardware devices such as CPU, GPU, memory, storage, and network. These parameters are crucial for evaluating the current status of the hardware and predicting potential performance bottlenecks. Software resource monitoring includes the operating status, process information, memory allocation, disk I / O, etc. of software such as the operating system, middleware, and application programs. By monitoring these software resources, you can understand the overall performance and resource usage of the system; Macro resource indicators include traditional system performance monitoring indicators such as CPU utilization, memory usage, disk I / O rate, and network throughput. These indicators provide a macro understanding of the overall system performance; By combining the results of micro-state perception with macro-resource indicators, a comprehensive resource state model is established. This model not only contains traditional performance monitoring data, but also integrates the micro-state information provided by quantum-inspired sensing technology, thereby more accurately reflecting the true state of the system.

[0024] Step 3: Based on the multi-dimensional task model and resource status model, generate and optimize the computing path by combining reinforcement learning with the quantum evolutionary algorithm; Among them, reinforcement learning is a machine learning method that learns how to take actions by interacting with the environment. In AI task management, reinforcement learning can be used to learn how to dynamically adjust the computing path according to the task and resource status. Based on the task model and resource status model, the reinforcement learning algorithm can generate a series of computing paths that aim to maximize a long-term goal (such as task completion time, resource utilization, etc.). Through continuous interaction with the environment (i.e., the computing system), the reinforcement learning algorithm can gradually optimize these computing paths to make them more adaptable to the changing task and resource status. The quantum evolutionary algorithm (QEA) is an optimization method that combines quantum computing and evolutionary algorithms. It uses the parallelism of quantum computing and the global search capability of evolutionary algorithms to accelerate the optimization process. Based on the initial calculation path generated by reinforcement learning, the quantum evolutionary algorithm can search for a better path in a larger solution space. By simulating the superposition and entanglement state of quantum bits, QEA can evaluate multiple paths at the same time, thereby finding the global optimal solution faster. QEA can also fuse multiple excellent paths to generate a calculation path with better overall performance. Reinforcement learning and quantum evolutionary algorithm work together; reinforcement learning is responsible for generating the initial computational paths and continuously optimizing these paths by interacting with the environment; while the quantum evolutionary algorithm is responsible for further searching for better paths based on reinforcement learning and fusing these paths to generate the final computational path; as the task and resource status change, the method combining reinforcement learning and quantum evolutionary algorithm can adaptively adjust the computational path to ensure the performance and stability of the system.

[0025] Step 4: Use a quantum simulator to verify and simulate the computational path and evaluate the computational path by running test cases; Specifically, the computational path generated in step three is input into the quantum simulator, and a series of test cases are run on the quantum simulator to simulate the execution of the computational path in the actual environment. These test cases should cover various possible scenarios and boundary conditions. During the simulation, performance data about the computational path is collected, such as task completion time, resource utilization, system stability indicators, etc. It is checked whether the computational path can be successfully executed under the given task and resource constraints. If the path cannot complete the task or causes the system to crash, it needs to be adjusted. Depending on the nature and severity of the problem, it is necessary to return to step three to adjust the computational path, which may include modifying algorithm parameters, optimizing resource allocation paths, or redesigning task scheduling algorithms. Through continuous verification, simulation and adjustment, the computational path is gradually optimized until it meets the needs of actual applications.

[0026] Step 5: During the execution of AI tasks, distributed fiber optic sensing and quantum encryption communication technology are used for real-time monitoring, and the calculation path is adjusted according to the monitoring results; Specifically, distributed fiber optic sensing and quantum encryption communication technologies are integrated into the AI ​​task management system to form a comprehensive real-time monitoring network. Various data during the task execution process are collected through the real-time monitoring network, including system status, task progress, resource utilization, etc. Then, data analysis tools are used to process and analyze these data to identify potential problems and anomalies; the quantum annealing algorithm is used to optimize the calculation path to adapt to new task requirements and system status, which may include adjusting task scheduling, resource allocation, algorithm parameters, etc.; the adjusted calculation path is applied to the AI ​​task management system, and continues to be monitored and evaluated through the real-time monitoring network; during the execution of AI tasks, distributed fiber optic sensing and quantum encryption communication technologies are continuously used for real-time monitoring, and the calculation path is continuously iterated and optimized according to the monitoring results and actual needs to ensure efficient execution of tasks and stable operation of the system.

[0027] Step 6: Use quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms to manage and coordinate task execution data; Specifically, quantum data storage, quantum state data transmission, quantum error correction code technology and quantum machine learning algorithms are combined to form a complete quantum data management framework. This framework can efficiently manage data during the execution of AI tasks and ensure the security, integrity and accuracy of the data. In the execution of distributed AI tasks, the quantum data management framework is used to achieve data collaboration between computing nodes. Through quantum state data transmission and quantum error correction code technology, data can be safely and efficiently transmitted and shared between multiple computing nodes. At the same time, quantum machine learning algorithms are used to accelerate the training and prediction process of the model, thereby improving the computing efficiency and accuracy of the entire system.

[0028] Step 7: After the task is completed, use machine learning and quantum information processing technology to extract knowledge from the task execution data, generate feedback information, and continuously improve the system based on the feedback information; Specifically, after the task is completed, clustering algorithms in machine learning (such as the DBSCAN algorithm, which automatically determines the number of clusters based on data density) and principal component analysis (PCA) are combined with quantum information processing technology to extract knowledge from the task execution data; for example, cluster analysis is used to find out the performance differences of similar task types under different resource allocations, and tasks are clustered according to characteristics such as task type, data size, and resource allocation mode, and performance indicators such as average execution time and accuracy of tasks in different clusters are analyzed; quantum principal component analysis is used to explore the potential relationship between quantum state changes and task results, and by reducing the dimensionality of quantum state measurement data, the factors that have a greater impact on task results are found. Large quantum state components; generate feedback information from the extracted knowledge, which can be in the form of rule updates (such as adjusting resource allocation rules for specific task types), model parameter adjustments (such as optimizing hyperparameters in deep learning models), etc., and continuously improve the system based on the feedback information; for example, if it is found that a certain type of image recognition task has higher accuracy under a specific quantum resource allocation, this resource allocation mode is used as a new rule for the processing of subsequent similar tasks, or the learning rate, convolution kernel size and other parameters of the deep learning model are adjusted according to the feedback to improve the system's processing capabilities for different tasks and the rationality of resource allocation, so as to achieve continuous optimization of the system and adapt to new task requirements.

[0029] In one embodiment, specifically, in step 1, the deep learning model is trained based on historical task data to identify implicit features of the task; the deep learning model is a fusion architecture of a convolutional neural network (CNN) and a recurrent neural network (RNN) to extract deep features of different types of task data; Specifically, for image data, the CNN part can extract local features such as texture, shape, and color distribution. Its convolution kernel size, step size and other parameters are set according to the characteristics of the image data. For example, when processing high-resolution medical images, the convolution kernel size can be set to 3x3 or 5x5, and the step size can be set to 1 or 2 to effectively capture image details. The RNN part captures the semantic associations between different areas of the image through long short-term memory units (LSTM) or gated recurrent units (GRU). The number of neurons in the hidden layer can be set according to the complexity of the image and the task requirements. For example, it can be set to 256 or 512 in complex satellite image analysis tasks. In natural language processing tasks, RNN is responsible for processing the sequence information of sentences. Its time step is determined by the length of the sentence. The time step can be appropriately increased when processing long texts. CNN mines the local features of words, and the width of the convolution kernel can be set according to the dimension of the vocabulary vector. For example, when the dimension of the vocabulary vector is 300, the width of the convolution kernel is set to 2-5. Combining traditional task parameters, such as task type (image recognition, natural language processing, speech recognition, etc.), data scale (data volume, data dimension, etc.), demand characteristics for computing resources (preference for CPU, GPU, memory usage, etc.) and time urgency (whether the task has real-time requirements), a multi-dimensional and high-precision task model is constructed. Among them, task types are classified by predefined labels, data scale is measured by the number of samples and feature dimensions in the data set, computing resource demand characteristics are obtained based on historical task execution data statistics, and time urgency is set according to the application scenario to which the task belongs. The corresponding time threshold is set. For example, the time threshold for real-time video analysis tasks can be set to process a certain number of frames per second.

[0030] In one embodiment, specifically, in step 2, the quantum-inspired sensor is based on a superconducting Josephson junction device, analyzes microscopic performance changes of hardware resources by monitoring quantum fluctuations of physical quantities, and uses quantum bit simulation technology to evaluate the quantum optimization space of software resources; Specifically, in terms of hardware resource assessment, for the CPU in the hardware resources, the quantum fluctuation of its current is monitored. When the fluctuation amplitude exceeds a certain threshold (such as ±0.1nA), it is considered that it may affect the computing frequency; for the memory, by detecting the voltage quantum fluctuation, when the standard deviation of the voltage fluctuation is greater than a certain set value (such as 0.05V), it is judged that a read and write error may occur; for the storage device, the performance change of the disk read and write head is analyzed based on the magnetic flux quantum fluctuation. When the magnetic flux fluctuation change rate exceeds a certain ratio (such as 5%), it indicates the risk of storage performance degradation; In terms of software resource assessment, quantum bit simulation technology is used to analyze the algorithms in the algorithm library. Taking the sorting algorithm as an example, it is converted into quantum bit representation, and the quantum optimization space of its time complexity is calculated by simulating quantum gate operations. If the time complexity reduction ratio under quantum simulation exceeds a certain value (such as 20%), the algorithm is considered to have great quantum optimization potential. These microscopic information are combined with macroscopic resource indicators (such as CPU utilization, memory occupancy, network bandwidth utilization, etc., with thresholds set to 80%, 70%, and 60% as warning values ​​for resource shortage), and a comprehensive resource status model is established through clustering algorithms in machine learning (such as K-means clustering, the number of clusters is set according to resource type and system scale) and neural network models (such as multi-layer perceptron, the number of hidden layer nodes is determined according to the number of input features).

[0031] In one embodiment, specifically, in step three, reinforcement learning learns the calculation path by interacting with the system environment feedback model, and the system environment feedback model is based on a quantum reward mechanism; Specifically, in reinforcement learning, the quantum reward mechanism rewards the agent according to the changes in quantum states during task execution; for example, when the fidelity of the quantum computing part increases by 10%, the utilization rate of quantum resources increases by 15%, or the task execution time is shortened by 20%, corresponding positive reward values ​​(such as +10, +8, +12) are given respectively; on the contrary, if the frequency of quantum state collapse increases, quantum resources are wasted, or task execution errors occur, negative reward values ​​(such as -5, -8, -10) are given; The quantum evolutionary algorithm evolves the quantum bit population through quantum gate operations and evaluates the quality of the solution based on the fitness function; Specifically, in the quantum evolutionary algorithm, quantum gate operations (such as Hadamard gates, CNOT gates, etc.) act on the quantum bit population, and the population size is set according to the complexity of the task and resources. For example, it can be set to 100-200 in complex simulation computing tasks; the quality of the solution is evaluated according to the fitness function, which comprehensively considers the quantum coherence maintenance of resources (by calculating the quantum coherence time and comparing it with the preset threshold, the threshold is set according to the quantum characteristics of the hardware), the minimization of the quantum information entropy of the task (the entropy reduction ratio exceeds a certain value, such as 15% is optimal) and the utilization rate of classical resources (the utilization rate is between 70% and 90% is a reasonable range) and other factors to ensure the optimality of the generated computing path in the quantum-classical hybrid computing environment.

[0032] In one embodiment, specifically, in step 4, the quantum simulator runs test cases of different types, scales, and resource status combinations according to a preset hybrid environment parameter configuration, and determines whether the calculation path needs to be adjusted according to the monitoring indicators; Specifically, the hybrid environment parameters of the quantum simulator include the number of qubits (set according to the complexity of the task and the quantum computing capabilities of the hardware, such as 10-100 qubits), the type of quantum gates (including the commonly used single-bit gates and multi-bit gates), and the simulation parameters of classical computing resources (such as the number of simulated CPU cores, memory size, etc.); Among them, the test cases cover various possible AI task scenarios, from small-scale image classification tasks (the number of image data set samples is less than 1000, and the number of categories is less than 10) to large-scale complex natural language processing tasks (the number of words in the text data set exceeds one million, and the processing tasks include grammatical analysis, semantic understanding, etc.), as well as task execution under different resource load conditions (resource utilization rate ranges from low to high, such as 20%-90%); monitoring indicators include resource utilization rate (quantum resource utilization rate is required to reach more than 60%, and classical resource utilization rate is between 70%-90%), task completion time (compared with the historical optimal time for tasks of the same type, the error is within ±10%), quantum state stability (quantum state collapse frequency is less than 5 times per hour), accuracy of task execution results (such as image recognition accuracy rate reaches more than 90%, and F1 value of natural language processing tasks reaches more than 80%), etc.; if any indicator exceeds the set reasonable range, it is considered that there is a problem with the calculation path, and it is necessary to return to step three for adjustment.

[0033] In one embodiment, specifically, in step five, the distributed optical fiber sensor obtains task execution information by monitoring changes in optical signals; specifically, the working state of the computing node is determined by monitoring changes in the intensity of Rayleigh scattering. When the intensity change exceeds a certain ratio (such as ±15%), it is considered that the node may be abnormal. The Mach-Zehnder interferometer is used to detect the movement of interference fringes to obtain information on data transmission and processing. When the movement of the fringes is greater than a certain threshold (such as 0.1 mm), it indicates that data transmission is abnormal. Quantum encryption communication technology uses quantum key distribution protocols to encrypt monitoring data; specifically, quantum encryption communication technology uses quantum key distribution protocols such as BB84 or E91 to encrypt and transmit monitoring data, and the key length is set according to the sensitivity of the data. For example, the key length of highly sensitive data is set to 2048 bits; when abnormal situations occur, such as task errors caused by quantum bit flips (by comparing the quantum state during task execution with the expected quantum state, a quantum bit flip is determined when the inconsistency rate exceeds 5%), resource performance fluctuations due to quantum noise (resource performance indicator changes exceed the set threshold, such as CPU calculation frequency fluctuations exceeding 10%), or changes in task requirements, such as quantum algorithm parameter adjustments (parameter changes exceed 20%), and new quantum-classical hybrid tasks are added, the calculation path is adjusted with the help of the quantum annealing algorithm; The quantum annealing algorithm is used to adjust the computing path in response to quantum-related anomalies; specifically, the quantum annealing algorithm simulates the lowest energy state of the quantum system during the adiabatic evolution process to find new optimal resource allocation and task scheduling solutions. The annealing time is set according to the scale and complexity of the problem, such as 10-20 seconds in medium-sized problems.

[0034] In one embodiment, specifically, in step six, quantum data storage is based on ion traps or superconducting quantum bits; specifically, in quantum data storage, quantum bits based on ion traps can store multiple quantum states, and their storage capacity is calculated based on the number of ions and the dimension of the quantum state. For example, a single ion can store 2-3 quantum states, and a storage unit composed of multiple ions can realize large-scale quantum data storage; superconducting quantum bits are realized through structures such as Josephson junctions, and their coherence time determines the stability of data storage, requiring the coherence time to be greater than a certain value (such as 100 microseconds); Quantum state data transmission technology uses quantum teleportation technology; specifically, quantum state data transmission technology uses quantum teleportation to achieve instantaneous data transmission. The transmission distance is determined by the characteristics of the quantum entangled state and the performance of the quantum repeater. In the absence of a relay, transmission of hundreds of kilometers can be achieved. Quantum error correction code technology uses surface codes, and its coding efficiency is set according to the importance of the data and the transmission distance. High coding efficiency (such as more than 75%) is used for important data or long-distance transmission; Quantum machine learning algorithms are used to optimize data preloading and caching computing paths; specifically, quantum machine learning algorithms dynamically adjust data preloading and caching computing paths based on task demand predictions and resource status; for example, by predicting the probability of a task accessing a certain type of data (preloading is performed when the probability is greater than 60%) and the frequency of use of cached data (data is retained in the cache if it is accessed more than 10 times per hour), it optimizes data sharing and utilization between different tasks.

[0035] In summary, an AI intelligent computing integrated control method provided by an embodiment of the present invention realizes the intelligent management of the entire chain of AI tasks from reception, analysis, resource scheduling, computing path generation and optimization, real-time monitoring and adjustment, data management and collaboration, to knowledge extraction and feedback improvement after task execution through deep integration of quantum technology and artificial intelligence technology. The method first uses a deep learning model to perform deep feature analysis on the task, and builds a multi-dimensional task model in combination with traditional task parameters; then, by real-time monitoring of the hardware and software resource status in the computing system, a comprehensive resource state model is established using quantum heuristic sensing technology. On this basis, a method combining reinforcement learning and quantum evolution algorithm is used to generate and optimize the computing path, and verification and simulation are performed through a quantum simulator to ensure the feasibility, efficiency and stability of the computing path. During the execution of the task, distributed optical fiber sensing and quantum encryption communication technology are used for real-time monitoring, and the computing path is flexibly adjusted with the help of quantum annealing algorithm to cope with abnormal situations and changes in task requirements. At the same time, the method also uses quantum data storage, quantum state data transmission, quantum error correction code technology and quantum machine learning algorithm to build an efficient quantum data management framework, realizing the safe, efficient management and collaboration of task execution data. Finally, after the task is completed, machine learning and quantum information processing technology are used to extract knowledge from the task execution data, generate feedback information, and continuously improve and optimize the system. The entire control method not only improves the execution efficiency and accuracy of AI tasks, but also enhances the stability and adaptability of the system, providing strong support for the further development of AI technology.

[0036] Figure 2 is a functional module diagram of an AI intelligent computing integrated control device according to an embodiment of the present application, such as Figure 2 As shown, an AI intelligent computing integrated control device includes: a task analysis module, a resource monitoring module, a path generation and optimization module, a path verification and simulation module, a real-time monitoring and adjustment module, a data management and collaboration module, and a feedback and improvement module; The task analysis module is used to receive AI task requests, perform deep feature analysis on AI tasks based on deep learning models, and build a multi-dimensional task model based on traditional task parameters. The deep learning model is trained based on historical task data and is used to identify implicit features of tasks. Resource monitoring module, which is used to monitor the hardware and software resource status in the computing system in real time, using quantum-inspired sensing technology to perceive the changes in the micro-state of resources, and combining macro-resource indicators to establish a resource status model; The path generation and optimization module is used to take the task model and resource state model as input, and generate and optimize the calculation path based on the method combining reinforcement learning and quantum evolution algorithm; The path verification and simulation module is used to verify and simulate the computational path using a quantum simulator and evaluate the computational path by running test cases; The real-time monitoring and adjustment module is used to use distributed fiber optic sensing and quantum encryption communication technology to conduct real-time monitoring during the execution of AI tasks, and adjust the calculation path according to the monitoring results; Data management and coordination module, which is used to manage and coordinate task execution data by using quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms; The feedback and improvement module is used to extract knowledge from task execution data and generate feedback information using machine learning and quantum information processing technology after the task is completed.

[0037] In summary, an AI intelligent computing integrated control device provided by an embodiment of the present invention ensures the stability and reliability of task execution by real-time monitoring and dynamic adjustment of computing strategies. By utilizing quantum technology and machine learning algorithms, the performance and scalability of the system are improved, the automation and intelligent processing of AI tasks are realized, and the task processing efficiency and accuracy are improved. It can be widely used in machine learning, deep learning, artificial intelligence and other fields, and provides an efficient, reliable and scalable solution for the processing of AI tasks.

[0038] For other details about the technical solutions for implementing each module in the AI ​​intelligent computing integrated control device in the above embodiment, please refer to the description of an AI intelligent computing integrated control method in the above embodiment, which will not be repeated here.

[0039] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0040] like Figure 3 The present invention provides a schematic diagram of the structure of an electronic device according to an embodiment of the present invention, which is suitable for implementing the electronic device in the embodiment of the present invention. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0041] like Figure 3As shown, the electronic device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device into a random access memory (RAM). In the RAM, various programs and data required for the operation of the electronic device are also stored. The processor, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0042] Typically, the following devices can be connected to the I / O interface: input devices such as sensors or visual information acquisition devices; output devices such as display screens; storage devices such as tapes, hard disks, etc.; and communication devices. The communication device allows the electronic device to communicate with other devices (such as edge computing devices) wirelessly or by wire to exchange data. Figure 3 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0043] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by the processor, all or part of the steps of an AI intelligent computing integrated control method of an embodiment of the present disclosure are executed.

[0044] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0045] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by the processor, all or part of the steps of an AI intelligent computing integrated control method of each embodiment of the present disclosure are executed.

[0046] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).

[0047] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0048] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0049] In the present disclosure, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The block diagrams of the devices, devices, equipment, and systems involved in the present disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0050] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0051] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0052] Various changes, substitutions, and modifications of the techniques described herein may be made without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of events, means, methods, and actions described above. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.

[0053] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0054] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A control method integrating AI intelligent computing, characterized in that: The following steps are involved: Obtain AI task request data, perform feature analysis on the AI ​​task request data through a deep learning model, and build a multi-dimensional task model; Obtain hardware and software resource data in the computing system, use quantum-inspired sensing technology to perceive changes in the microscopic state of resources, combine macro-resource indicators to quantitatively evaluate resource data, and build a resource state model; Based on the multi-dimensional task model and resource status model, the computing path is generated and optimized by combining reinforcement learning with quantum evolutionary algorithm; Use quantum simulators to verify and simulate computational paths, and evaluate computational paths by running test cases; During the execution of AI tasks, distributed fiber optic sensing and quantum encryption communication technologies are used for real-time monitoring, and the calculation path is adjusted according to the monitoring results; Use quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms to manage and coordinate task execution data; After the task is completed, machine learning and quantum information processing techniques are used to extract knowledge from the task execution data and generate feedback information.

2. The AI ​​intelligent computing integrated control method according to claim 1 is characterized by: The deep learning model is trained based on historical task data and is used to identify implicit features of tasks. The deep learning model is a fusion architecture of convolutional neural networks and recurrent neural networks and is used to extract deep features of different types of task data.

3. The AI ​​intelligent computing integrated control method according to claim 1 is characterized in that: The quantum-inspired sensing technology is based on superconducting Josephson junction devices. It analyzes the microscopic performance changes of hardware resources by monitoring the quantum fluctuations of physical quantities and uses quantum bit simulation technology to evaluate the quantum optimization space of software resources.

4. The AI ​​intelligent computing integrated control method according to claim 1 is characterized in that: The reinforcement learning learns the calculation path by interacting with the system environment feedback model, and the system environment feedback model is built based on the quantum reward mechanism; the quantum evolution algorithm evolves the quantum bit population through quantum gate operations and evaluates the quality of the solution according to the fitness function.

5. The AI ​​intelligent computing integrated control method according to claim 1 is characterized by: The quantum simulator runs test cases of different types, scales and resource status combinations according to preset hybrid environment parameter configurations, and determines whether the calculation path needs to be adjusted based on monitoring indicators.

6. The AI ​​intelligent computing integrated control method according to claim 1 is characterized by: The distributed optical fiber sensor obtains task execution information by monitoring changes in optical signals; the quantum encryption communication technology uses a quantum key distribution protocol to encrypt monitoring data; and the quantum annealing algorithm is used to adjust the calculation path in response to quantum-related anomalies.

7. The AI ​​intelligent computing integrated control method according to claim 1 is characterized by: The quantum data storage is based on ion traps or superconducting quantum bits; the quantum state data transmission technology utilizes quantum teleportation technology; the quantum error correction code technology utilizes surface codes; and the quantum machine learning algorithm is used to optimize data preloading and cache calculation paths.

8. An AI intelligent computing integrated control device, applied to an AI intelligent computing integrated control method according to any one of claims 1 to 7, characterized in that: include: Task analysis module, resource monitoring module, path generation and optimization module, path verification and simulation module, real-time monitoring and adjustment module, data management and collaboration module, and feedback and improvement module; The task analysis module is used to receive AI task requests, perform deep feature analysis on AI tasks based on a deep learning model, and build a multi-dimensional task model in combination with traditional task parameters. The deep learning model is trained based on historical task data and is used to identify implicit features of tasks. The resource monitoring module is used to monitor the hardware and software resource status in the computing system in real time, using quantum-inspired sensing technology to perceive changes in the microscopic state of resources, and establishing a resource status model in combination with macroscopic resource indicators; The path generation and optimization module is used to take the task model and the resource state model as input, and generate and optimize the calculation path based on a method combining reinforcement learning and quantum evolution algorithm; The path verification and simulation module is used to verify and simulate the calculation path using a quantum simulator and evaluate the calculation path by running test cases; The real-time monitoring and adjustment module is used to use distributed optical fiber sensing and quantum encryption communication technology to perform real-time monitoring during the execution of AI tasks, and adjust the calculation path according to the monitoring results; The data management and coordination module is used to manage and coordinate task execution data by using quantum data storage and quantum state data transmission technology, combined with quantum error correction code technology and quantum machine learning algorithms; The feedback and improvement module is used to extract knowledge from task execution data and generate feedback information using machine learning and quantum information processing technology after the task is completed.

9. An electronic device, characterized in that: The electronic device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute an AI intelligent computing integrated control method as described in any one of claims 1-7 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program code, and the program code is used to execute the AI ​​intelligent computing integrated control method described in any one of claims 1-7.

Images