A machine learning based space weather model parameter optimization system and method

Abstract

The application relates to the field of information technology, and particularly provides a space weather model parameter optimization system and method based on machine learning. The space weather model parameter optimization system based on machine learning comprises data input, parameter management, machine learning optimization, model test scheduling and result evaluation modules. The data input module obtains basic configuration, parameter constraint and other data, and sends adjustable parameters to the parameter management module; the parameter management module classifies and stores the adjustable parameters, performs coding conversion and generates parameter combinations, and sends the parameter combinations to the machine learning optimization module; the machine learning optimization module takes the parameter combinations as input and loss value as feedback, iteratively updates the model and samples and outputs candidate parameter combinations; the model test scheduling module initiates back-calculation tests through a visual platform, supports parallel back-calculation and abnormality processing; and the result evaluation module calculates a loss function in combination with observation data and feeds back the loss function, so that parameter automatic optimization is realized.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular to a system and method for optimizing space weather model parameters based on machine learning. Background Technology

[0002] Numerical space weather models are core tools for simulating and forecasting solar activity, interplanetary disturbances, and magnetosphere-ionosphere-thermosphere coupling processes. Their forecast accuracy directly affects the ability to issue early warnings of space weather disasters. However, existing numerical space weather models contain a large number of empirical or uncertain parameters, the values ​​of which have a significant impact on the model output results.

[0003] Currently, adjustments to these parameters primarily rely on manual methods. Technicians select parameter combinations based on experience or trial-and-error, verify the parameter effects through model back-calculation experiments, and then repeatedly refine the parameter values. This traditional manual parameter tuning method has significant drawbacks: firstly, it is extremely inefficient, as the scope of manual exploration is limited in the face of a high-dimensional parameter space, often requiring substantial time and computational resources; secondly, manual parameter tuning is easily limited by subjective experience, making it difficult to break through local optima and find globally optimal parameter combinations, thus hindering the full improvement of model forecast accuracy and limiting the potential of space weather numerical models.

[0004] With the development of artificial intelligence technology, machine learning algorithms have demonstrated efficient global search capabilities in the field of parameter optimization. How to combine machine learning technology with space weather numerical models to build an automated parameter optimization system and address the pain points of traditional manual parameter tuning has become a key requirement for improving the accuracy of space weather forecasts. Summary of the Invention

[0005] One objective of this application is to provide a space weather model parameter optimization system based on machine learning, which at least addresses the problem of how to achieve intelligent and automated optimization of key parameters of space weather numerical models, thereby improving parameter tuning efficiency and effectiveness.

[0006] To achieve the above objectives, some embodiments of this application provide the following aspects:

[0007] In a first aspect, some embodiments of this application also provide a space weather model parameter optimization system based on machine learning, the system including a data input module, a parameter management module, a machine learning optimization module, a model experiment scheduling module, and a result evaluation module;

[0008] The data input module is used to acquire the basic configuration information of the space weather numerical model, the constraint range of the adjustable parameters, the historical observation dataset, and the initial boundary conditions required for model backcalculation, and to send the adjustable parameters to the parameter management module.

[0009] The parameter management module manages the adjustable parameters in real time, including parameter classification and storage, encoding conversion, and generation of parameter combinations; and sends the parameter combinations to the machine learning optimization module.

[0010] The machine learning optimization module is used to take the parameter combination output by the parameter management module as input, the loss value output by the result evaluation module as feedback signal, update the algorithm model iteratively, sample within the parameter constraint range, output the next round of candidate parameter combinations, and send the candidate parameter combinations to the mode experiment scheduling module.

[0011] The model experiment scheduling module is used to receive candidate parameter combinations output by the machine learning optimization module and automatically initiate back-calculation experiments of space weather numerical models through the visualization experiment platform. The visualization experiment platform has a task queue management function, which can dynamically allocate test tasks according to the computing resource status, support parallel back-calculation of multiple parameter combinations, monitor the test progress in real time, and automatically retry or alarm for failed tests.

[0012] The result evaluation module is used to back-calculate the output results of the experiment, and calculate the loss function between the model forecast results and the observation data by combining the historical observation dataset provided by the data input module.

[0013] Secondly, some embodiments of this application also provide a method for optimizing space weather model parameters based on machine learning, the method comprising:

[0014] S1. Import space weather numerical model configuration, adjustable parameter constraint range, historical observation data and initial boundary conditions through the data input module, and the data storage module completes the classification and storage of the data.

[0015] S2. Users set optimization goals, iteration termination conditions, and loss function types through the visual interaction module; the parameter management module encodes and converts adjustable parameters to generate an initial parameter space.

[0016] S3. The machine learning optimization module generates the first batch of candidate parameter combinations based on the initial parameter space and uses a preset machine learning algorithm. The parameter management module then verifies and formats the combinations.

[0017] S4. The model experiment scheduling module receives candidate parameter combinations, automatically initiates space weather numerical model back-calculation experiments, schedules computing resources in parallel to complete multiple sets of experimental tasks, and monitors and records the experimental status in real time.

[0018] S5. The results evaluation module collects the backtesting results of each experiment, compares them with historical observation data to calculate the loss function value, and feeds the loss value back to the machine learning optimization module.

[0019] S6. The machine learning optimization module updates the algorithm model based on the loss value, performs a new round of intelligent sampling in the parameter space, and generates the next batch of candidate parameter combinations; repeat steps S4-S6 until the iteration termination condition is met.

[0020] S7. The system outputs the optimal parameter set that minimizes the loss function. The model prediction effect corresponding to the optimal parameters is displayed through the visualization interaction module. At the same time, the optimal parameter set and optimization process data are stored in the data storage module.

[0021] Thirdly, some embodiments of this application also provide an electronic device, the electronic device comprising: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method described above.

[0022] Fourthly, some embodiments of this application also provide a computer-readable medium having computer program instructions stored thereon, which can be executed by a processor to implement the method described above.

[0023] Fifthly, some embodiments of this application also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method described above.

[0024] Compared to related technologies, the solution provided in this application, a space weather model parameter optimization system based on machine learning, uses a data input module as the system's data access point. This module collects basic configuration information of the space weather numerical model (including model type, operating environment requirements, calculation accuracy thresholds, etc.), the constraint range of adjustable parameters (which can be preset by the user based on the model's physical mechanism, or automatically generated by the system based on historical optimization cases), historical observation datasets (sourced from observation platforms such as satellites and ground observation stations, ensuring data authenticity and timeliness), and the initial boundary conditions required for model backcalculation (such as key initial parameters like initial solar wind velocity, magnetic field strength, and plasma density). After data acquisition, this module simultaneously performs preliminary data verification and formatting, removing abnormally missing or invalid data that exceeds reasonable ranges. The processed adjustable parameters are then transmitted to the parameter management module. The parameter management module categorizes and stores adjustable parameters of different types, such as empirical and uncertain parameters, in an orderly manner for easy retrieval and recall. Encoding conversion transforms the physical values ​​of parameters into an input format compatible with machine learning algorithms, resolving the compatibility issue between algorithms and physical parameters and ensuring effective identification of parameter signals. Parameter combination generation, based on the sampling requirements of the machine learning optimization module, rationally combines parameters, verifies whether the combinations conform to constraints, and eliminates invalid combinations. After completing the above processing, the parameter management module pushes compliant parameter combinations to the machine learning optimization module, ensuring efficient parameter flow. The machine learning optimization module, as the system's computational unit, uses the parameter combinations output by the parameter management module as input and combines them with the loss values ​​fed back by the result evaluation module to construct a closed-loop optimization. This module incorporates various optimization algorithms, including Bayesian optimization, genetic algorithms, and reinforcement learning algorithms, and can adaptively select the optimal algorithm or employ algorithm fusion strategies based on the parameter space dimension and optimization objective. Through iterative updates to the algorithm model, intelligent sampling is performed within the preset parameter constraints to locate potential optimal parameter regions, thereby generating the next round of candidate parameter combinations and sending them to the mode experiment scheduling module, ensuring the scientific rigor and relevance of the optimization direction. The model experiment scheduling module is responsible for receiving candidate parameter combinations and automatically initiating back-calculation experiments of space weather numerical models through a visual experiment platform. This visual experiment platform has comprehensive task queue management functions, can perceive the computing resource load status in real time, dynamically allocate the priority of experiment tasks, support parallel back-calculation of multiple parameter combinations, and significantly improve experiment efficiency. At the same time, it monitors the progress of each experiment in real time, records detailed experiment logs, and automatically triggers a retry mechanism for failed experiments caused by resource conflicts, abnormal initial conditions, etc. If the retry fails, an abnormal alarm is issued in a timely manner to ensure the continuity and stability of the experiment process.The results evaluation module collects the output results of the model backtesting experiment, retrieves the historical observation dataset pre-stored in the data input module, and calculates the difference between the model forecast results and the observed data using a preset loss function (including root mean square error, mean absolute error, correlation coefficient, etc.) to obtain a quantified loss function value. This loss function value is then transmitted as a key feedback signal to the machine learning optimization module, providing a basis for the iterative update of the algorithm model. This completely replaces the traditional manual parameter tuning method, enabling intelligent and automated optimization of key parameters in space weather numerical models. This not only significantly improves parameter tuning efficiency and reduces the cost of manual intervention, but also overcomes the limitations of human experience, effectively avoiding getting trapped in local optima and improving the accuracy of space weather forecasts. Attached Figure Description

[0025] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0026] Figure 1 A schematic diagram of a space weather model parameter optimization system based on machine learning is provided in an embodiment of this application;

[0027] Figure 2 This is a flowchart illustrating a method for optimizing space weather model parameters based on machine learning, as implemented in this application.

[0028] Figure 3 This is a schematic diagram of an exemplary structure of a processor and a memory according to this application;

[0029] Figure 4 This is an exemplary structural diagram of an electronic device according to this application.

[0030] Figure label:

[0031] 11-Data Input Module;

[0032] 12-Parameter Management Module;

[0033] 13-Machine Learning Optimization Module;

[0034] 14-Mode test scheduling module;

[0035] 15-Results Evaluation Module. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0037] See appendix Figure 1 , Figure 1 This is a schematic diagram of a space weather model parameter optimization system based on machine learning, provided as an embodiment of this application. Figure 1 As shown, the main system of the space weather model parameter optimization system based on machine learning in this application embodiment includes a data input module 11, a parameter management module 12, a machine learning optimization module 13, a model experiment scheduling module 14, and a result evaluation module 15.

[0038] The data input module 11 is used to acquire the basic configuration information of the space weather numerical model, the constraint range of the adjustable parameters, the historical observation dataset, and the initial boundary conditions required for model back calculation, and to send the adjustable parameters to the parameter management module 12.

[0039] The parameter management module 12 manages the adjustable parameters in real time, including parameter classification and storage, encoding conversion, and generation of parameter combinations; and sends the parameter combinations to the machine learning optimization module 13.

[0040] The machine learning optimization module 13 is used to take the parameter combination output by the parameter management module 12 as input, the loss value output by the result evaluation module 15 as feedback signal, update the algorithm model iteratively, sample within the parameter constraint range, output the next round of candidate parameter combinations, and send the candidate parameter combinations to the pattern experiment scheduling module 14.

[0041] The model experiment scheduling module 14 is used to receive the candidate parameter combinations output by the machine learning optimization module 13 and automatically initiate back-calculation experiments of space weather numerical models through the visualization experiment platform. The visualization experiment platform has a task queue management function, which can dynamically allocate test tasks according to the computing resource status, support parallel back-calculation of multiple parameter combinations, monitor the test progress in real time, and automatically retry or alarm for failed tests.

[0042] The results evaluation module 15 is used to calculate the output results of the backtest, and, in conjunction with the historical observation dataset provided by the data input module 11, calculate the loss function between the model forecast results and the observation data.

[0043] In this embodiment, the data input module 11 serves as the system's data access point, used to collect basic configuration information of the space weather numerical model (including model type, operating environment requirements, calculation accuracy thresholds, etc.), the constraint range of adjustable parameters (which can be preset by the user based on the model's physical mechanism, or automatically generated by the system based on historical optimization cases), historical observation datasets (sourced from observation platforms such as satellites and ground observation stations, ensuring data authenticity and timeliness), and the initial boundary conditions required for model backcalculation (such as key initial parameters like initial solar wind velocity, magnetic field strength, and plasma density). After completing data acquisition, this module simultaneously performs preliminary data verification and formatting, removing abnormally missing or invalid data that exceeds reasonable ranges, and then transmits the processed adjustable parameters to the parameter management module 12. The parameter management module 12 categorizes and stores adjustable parameters of different types, such as empirical and uncertain parameters, in an orderly manner for easy retrieval and recall. Encoding conversion transforms the physical values ​​of parameters into an input format suitable for machine learning algorithms, resolving the compatibility issue between algorithms and physical parameters and ensuring effective identification of parameter signals. Parameter combination generation, based on the sampling requirements of the machine learning optimization module 13, rationally combines parameters and verifies whether the combinations meet constraints, eliminating invalid combinations. After completing the above processing, the parameter management module 12 pushes compliant parameter combinations to the machine learning optimization module 13, ensuring efficient parameter flow. The machine learning optimization module 13, as the system's computational unit, uses the parameter combinations output by the parameter management module 12 as input and combines them with the loss value fed back by the result evaluation module 15 to construct a closed-loop optimization. This module incorporates various optimization algorithms, including Bayesian optimization, genetic algorithms, and reinforcement learning algorithms, and can adaptively select the optimal algorithm or employ an algorithm fusion strategy based on the parameter space dimension and optimization objective. By iteratively updating the algorithm model and intelligently sampling within the preset parameter constraints, the potential optimal parameter region is located, generating the next round of candidate parameter combinations, which are then sent to the model experiment scheduling module 14 to ensure the scientific rigor and relevance of the optimization direction. The model experiment scheduling module 14 receives the candidate parameter combinations and automatically initiates back-calculation experiments of the space weather numerical model through a visual experiment platform. This visual experiment platform has comprehensive task queue management functions, can perceive the computing resource load status in real time, dynamically allocate the priority of experiment tasks, and support parallel back-calculation of multiple parameter combinations, significantly improving experiment efficiency. Simultaneously, it monitors the progress of each experiment in real time, records detailed experiment logs, and automatically triggers a retry mechanism for failed experiments caused by resource conflicts, abnormal initial conditions, etc. If a retry fails, an abnormal alarm is issued promptly to ensure the continuity and stability of the experiment process.The results evaluation module 15 collects the output results of the model back-calculation experiment, retrieves the historical observation dataset pre-stored in the data input module 11, and calculates the difference between the model forecast results and the observed data using a preset loss function (including root mean square error, mean absolute error, correlation coefficient, etc.) to obtain a quantified loss function value. This loss function value is then transmitted as a key feedback signal to the machine learning optimization module 13, providing a basis for the iterative update of the algorithm model. This completely replaces the traditional manual parameter tuning method, achieving intelligent and automated optimization of key parameters of the space weather numerical model. This not only significantly improves parameter tuning efficiency and reduces the cost of manual intervention, but also overcomes the limitations of human experience, effectively avoiding getting trapped in local optima and improving the accuracy of space weather forecasts.

[0044] In one implementation, the system also includes a visual interaction module;

[0045] The visualization and interaction module is used to present information about the entire parameter optimization process, including parameter spatial distribution, candidate parameter combinations, experimental progress, loss function change curves, optimal parameter set and corresponding model prediction effects.

[0046] In this embodiment, the system also integrates a visualization and interaction module. This module serves as a visual representation of the entire parameter optimization process, intuitively displaying information from each stage of the optimization process. Specifically, this includes the distribution characteristics of the parameter space, details of candidate parameter combinations for each iteration, real-time progress of the model backtesting, the loss function's variation curve with iteration, and a comparison of the final optimal parameter set and its corresponding model prediction performance. This clear visualization allows users to easily monitor the optimization process in real time, quickly assess the optimization results, and provides intuitive data support for potential manual intervention.

[0047] In one implementation, the system further includes a data storage module;

[0048] The data storage module is used to store space weather model parameters, which include input basic data, parameter information, experimental task records, back-calculation results, loss function values, optimal parameter sets, and algorithm model parameters.

[0049] The data storage module adopts a distributed storage architecture.

[0050] In this embodiment, the system also integrates a data storage module, which serves as the core data assurance system for the entire parameter optimization process. This module is responsible for comprehensively storing various parameters and data related to the space weather model. Specifically, this includes basic configuration data imported by the data input module 11, adjustable parameters and constraint ranges, test task records generated by the model test scheduling module 14, backtesting output results, loss function values ​​calculated by the result evaluation module 15, the final optimized parameter set, and algorithm model parameters generated during the operation of the machine learning optimization module 13. To ensure the security, stability, and scalability of data storage, the data storage module adopts a distributed storage architecture, enabling distributed backup and rapid retrieval of data. It also supports long-term storage and reuse of massive amounts of optimization data, providing reliable support for the traceability of the optimization process.

[0051] In one implementation, the adjustable parameters include empirical parameters and uncertain parameters in the model, and the constraint range is preset by the user according to the physical mechanism of the model or automatically generated by the system based on historical optimization data.

[0052] In this embodiment, empirical parameters are mostly derived from historical observation experience, while uncertain parameters originate from unknowns or simplified assumptions in the model's physical process description. To ensure the rationality and relevance of parameter optimization, the constraint range of adjustable parameters is set flexibly: allowing users to preset reasonable ranges based on the core physical mechanisms of the space weather numerical model and past parameter tuning experience, and also supporting the system to perform intelligent analysis based on historical optimization data to automatically generate constraint ranges suitable for the current optimization scenario.

[0053] In one implementation, encoding conversion is used to convert the physical values ​​of parameters into an input format suitable for machine learning algorithms, with the parameter combination generated by the parameter sampling results output by the machine learning optimization module 13.

[0054] In this embodiment, the physical values ​​of adjustable parameters (such as magnetic permeability, diffusion coefficient, and other numerical values ​​with clear physical meaning) are converted into an input format suitable for machine learning algorithms. Since the physical values ​​of parameters often differ in magnitude or have diverse units, the encoding conversion forms a standardized algorithm input vector, ensuring that the machine learning optimization module 13 can accurately identify and efficiently process parameter information. Furthermore, the generation of parameter combinations is not random but strictly based on the parameter sampling results output by the machine learning optimization module 13, combined with parameter constraints to complete the combination construction and validity verification, ensuring that the generated parameter combinations are both scientifically sound and feasible.

[0055] In one implementation, the machine learning optimization module 13 incorporates at least one of the following algorithms: Bayesian optimization algorithm, genetic algorithm, and reinforcement learning algorithm.

[0056] In this embodiment, the machine learning optimization module 13, as the core computing unit for system parameter optimization, incorporates a variety of high-performance optimization algorithms, specifically covering at least one of Bayesian optimization, genetic algorithms, and reinforcement learning algorithms. This module can flexibly adapt algorithm selection according to the actual optimization scenario: Bayesian optimization can be used for low-dimensional parameter spaces or scenarios requiring rapid convergence; for high-dimensional complex parameter spaces, the global search capability of genetic algorithms can be employed; and reinforcement learning algorithms are more suitable if dynamic adaptation to environmental changes during parameter optimization is required. The optional configuration of multiple algorithms ensures that the module can efficiently address the parameter optimization needs of different types of space weather numerical models, providing algorithmic support for accurately locating the globally optimal parameter set.

[0057] In one implementation, the loss function includes at least one of root mean square error, mean absolute error, and correlation coefficient;

[0058] Users can customize the weights or combinations of the loss function according to their optimization needs;

[0059] The calculated loss function value is fed back to the machine learning optimization module 13 as the basis for parameter optimization iteration.

[0060] In this embodiment, the loss function used by the result evaluation module 15 has flexible configuration characteristics, specifically covering at least one of root mean square error, mean absolute error, and correlation coefficient. To adapt to different optimization scenarios and needs, users can define the weight allocation or combination of the loss function themselves. For example, for high-precision forecasting requirements, the weight of root mean square error can be increased, and for long-term statistical optimization, mean absolute error and correlation coefficient can be combined. This module calculates the difference between the forecast results and the observed data using a preset or custom loss function, obtains the quantified loss function value, and then feeds it back to the machine learning optimization module 13. This provides a core basis for the iterative update of the algorithm model and parameter adjustment, ensuring the accuracy of the optimization direction.

[0061] See appendix Figure 2 , Figure 2 This is a flowchart illustrating a method for optimizing space weather model parameters based on machine learning, as described in this application. Figure 2 As shown, the space weather model parameter optimization method based on machine learning in this embodiment of the invention mainly includes the following steps S1-S7:

[0062] Step S1: Import the space weather numerical model configuration, adjustable parameter constraint range, historical observation data and initial boundary conditions through the data input module 11, and the data storage module completes the classification and storage of the data;

[0063] Step S2: The user sets the optimization objective, iteration termination condition and loss function type through the visual interaction module; the parameter management module 12 encodes and converts the adjustable parameters to generate the initial parameter space;

[0064] Step S3: The machine learning optimization module 13 generates the first batch of candidate parameter combinations based on the initial parameter space using a preset machine learning algorithm, and the parameter management module 12 completes the verification and formatting of the combinations.

[0065] Step S4: The model experiment scheduling module 14 receives candidate parameter combinations, automatically initiates space weather numerical model back-calculation experiments, schedules computing resources in parallel to complete multiple sets of experimental tasks, and monitors and records the experimental status in real time.

[0066] Step S5: The result evaluation module 15 collects the backtesting results of each experiment, compares them with historical observation data to calculate the loss function value, and feeds the loss value back to the machine learning optimization module 13;

[0067] Step S6: The machine learning optimization module 13 updates the algorithm model based on the loss value, performs a new round of intelligent sampling in the parameter space, and generates the next batch of candidate parameter combinations; repeat steps S4-S6 until the iteration termination condition is met.

[0068] Step S7: The system outputs the optimal parameter set that minimizes the loss function, displays the model prediction effect corresponding to the optimal parameters through the visualization interaction module, and stores the optimal parameter set and optimization process data in the data storage module.

[0069] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but without changing the core design of the algorithm and process, are also within the scope of protection of this patent.

[0070] It is not difficult to see that this embodiment is a method embodiment corresponding to the system embodiment, and this embodiment can be implemented in conjunction with the system embodiment. The relevant technical details mentioned in the system embodiment are still valid in this embodiment, and will not be repeated here to reduce repetition. Correspondingly, the relevant technical details mentioned in this embodiment can also be applied to the system embodiment.

[0071] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problem proposed in this application; however, this does not mean that other units are absent from this embodiment.

[0072] Furthermore, some embodiments of this application also provide an electronic device. The electronic device can be various forms of digital computer, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device can also be various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices.

[0073] The electronic device includes: one or more processors; and a memory storing computer program instructions that, when executed, cause the processor to perform the steps of the methods provided in any one or more of the above embodiments. Figure 4 An exemplary structural diagram of the electronic device is disclosed. For example... Figure 3 As shown, the electronic device includes one or more processors 1101, a memory 1102, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components are interconnected via different buses and can be mounted on a common motherboard or otherwise as required. The processors can process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple electronic devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). The components, their connections and relationships, and their functions shown herein are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0074] The electronic device may further include an input device 1103 and an output device 1104. The processor 1101, memory 1102, input device 1103, and output device 1104 may be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0075] Input device 1103 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the electronic device, such as a touch screen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 1104 may include a display device, auxiliary lighting device (e.g., LED), and haptic feedback device (e.g., vibration motor). The display device may include, but is not limited to, a liquid crystal display (LCD), a light-emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0076] To provide interaction with the user, the electronic device can be a computer. The computer has: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0077] In this embodiment, a computer-readable medium stores a computer program / instructions that, when executed by a processor, implement the steps of the methods provided in any one or more of the above embodiments. This computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into that device. The aforementioned computer-readable medium carries one or more computer-readable instructions.

[0078] The memory 1102 can serve as a non-transitory computer-readable storage medium, used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions, and modules stored in the memory 1102, thereby implementing the program instructions / modules corresponding to the methods provided in any one or more of the embodiments described above in this application.

[0079] The memory 1102 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 1102 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 1102 may optionally include memory remotely located relative to the processor 1101, and these remote memories can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0080] It should be noted that the computer-readable medium described in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0081] Computer-readable media include permanent and non-permanent, removable and non-removable media, which can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, read-only optical disc (CD-ROM), digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0082] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0083] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. For example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of this application can be executed by a processor to implement the steps or functions described above. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium, such as RAM memory, magnetic or optical drives, floppy disks, or similar devices. Furthermore, some steps or functions of this application can be implemented in hardware, for example, as circuitry that works with a processor to perform the various steps or functions.

[0084] The computer program product provided in this application includes one or more computer programs / instructions. When executed by a processor, these computer programs / instructions generate, in whole or in part, the processes or functions described in this application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0085] The flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-specific system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0086] The scope of this application is defined by the appended claims rather than the foregoing description, and is therefore intended to encompass all variations falling within the meaning and scope of equivalents of the claims. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a device claim may also be implemented by a single unit or device in software or hardware. Terms such as "first," "second," etc., are used only for distinguishing descriptions and do not indicate any particular order, nor should they be construed as indicating or implying relative importance.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily made by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims, and the above embodiments should be regarded as exemplary and non-limiting.

Claims

1. A space weather model parameter optimization system based on machine learning, characterized in that, The system includes a data input module, a parameter management module, a machine learning optimization module, a pattern experiment scheduling module, and a result evaluation module. The data input module is used to acquire the basic configuration information of the space weather numerical model, the constraint range of the adjustable parameters, the historical observation dataset, and the initial boundary conditions required for model backcalculation, and to send the adjustable parameters to the parameter management module. The parameter management module manages the adjustable parameters in real time, including parameter classification and storage, encoding conversion, and generation of parameter combinations; and sends the parameter combinations to the machine learning optimization module. The machine learning optimization module is used to take the parameter combination output by the parameter management module as input, the loss value output by the result evaluation module as feedback signal, update the algorithm model iteratively, sample within the parameter constraint range, output the next round of candidate parameter combinations, and send the candidate parameter combinations to the mode experiment scheduling module. The model experiment scheduling module is used to receive candidate parameter combinations output by the machine learning optimization module and automatically initiate back-calculation experiments of space weather numerical models through the visualization experiment platform. The visualization experiment platform has a task queue management function, which can dynamically allocate test tasks according to the computing resource status, support parallel back-calculation of multiple parameter combinations, monitor the test progress in real time, and automatically retry or alarm for failed tests. The result evaluation module is used to back-calculate the output results of the experiment, and calculate the loss function between the model forecast results and the observation data by combining the historical observation dataset provided by the data input module.

2. The system according to claim 1, characterized in that, The system also includes a visual interaction module; The visualization and interaction module is used to present information about the entire process of parameter optimization, including parameter spatial distribution, candidate parameter combinations, experimental progress, loss function change curves, optimal parameter set and corresponding model prediction effects.

3. The system according to claim 2, characterized in that, The system also includes a data storage module; The data storage module is used to store space weather model parameters, which include input basic data, parameter information, experimental task records, back calculation results, loss function values, optimal parameter sets, and algorithm model parameters. The data storage module adopts a distributed storage architecture.

4. The system according to claim 1, characterized in that, The adjustable parameters include empirical parameters and uncertain parameters in the mode, and the constraint range is preset by the user according to the mode's physical mechanism or automatically generated by the system based on historical optimization data.

5. The system according to claim 1, characterized in that, The encoding conversion is used to convert the physical values ​​of the parameters into an input format suitable for machine learning algorithms. The parameter combination is generated by the parameter sampling results output by the machine learning optimization module.

6. The system according to claim 1, characterized in that, The machine learning optimization module incorporates at least one of the following algorithms: Bayesian optimization algorithm, genetic algorithm, and reinforcement learning algorithm.

7. The system according to claim 1, characterized in that, The loss function includes at least one of root mean square error, mean absolute error, and correlation coefficient. Users can customize the weights or combinations of the loss function according to their optimization needs; The calculated loss function value is fed back to the machine learning optimization module as the basis for iterative parameter optimization.

8. A method for optimizing parameters of a space weather model based on machine learning, characterized in that, Applied to the system as described in any one of claims 1-7, the method comprises: S1. Import space weather numerical model configuration, adjustable parameter constraint range, historical observation data and initial boundary conditions through the data input module, and the data storage module completes the classification and storage of the data. S2. Users set optimization goals, iteration termination conditions, and loss function types through the visual interaction module; the parameter management module encodes and converts adjustable parameters to generate an initial parameter space. S3. The machine learning optimization module generates the first batch of candidate parameter combinations based on the initial parameter space and uses a preset machine learning algorithm. The parameter management module then verifies and formats the combinations. S4. The model experiment scheduling module receives candidate parameter combinations, automatically initiates space weather numerical model back-calculation experiments, schedules computing resources in parallel to complete multiple sets of experimental tasks, and monitors and records the experimental status in real time. S5. The results evaluation module collects the backtesting results of each experiment, compares them with historical observation data to calculate the loss function value, and feeds the loss value back to the machine learning optimization module. S6. The machine learning optimization module updates the algorithm model based on the loss value, performs a new round of intelligent sampling in the parameter space, and generates the next batch of candidate parameter combinations; repeat steps S4-S6 until the iteration termination condition is met. S7. The system outputs the optimal parameter set that minimizes the loss function. The model prediction effect corresponding to the optimal parameters is displayed through the visualization interaction module. At the same time, the optimal parameter set and optimization process data are stored in the data storage module.

9. An electronic device, characterized in that, The device includes: One or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the method as described in claim 8.

10. A computer-readable medium, characterized in that, It stores computer program instructions that can be executed by a processor to implement the method as described in claim 8.

Images