Software architecture supporting super-large scale simulation
Through the combination of dynamic task allocation and principal component analysis, the resource and performance problems of a single process simulation product when dealing with a large number of vehicles are solved, and efficient ultra-large-scale simulation and powerful visualization functions are achieved, supporting horizontal expansion and stability .
Patent Information
- Application Number
- CN202411572644.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-06
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, when a single-process simulation product deals with a large number of vehicles and equipment, the resource and performance problems are serious and cannot support horizontal horizontal scaling, resulting in simulation cycle delays and resource waste.
A software architecture is designed, including a control unit, a scheduling unit and an engine unit. Simulation tasks are generated through user configuration, tasks are dynamically allocated according to task priority and resource availability, simulation calculations are performed using the engine unit, and the results are visualized through the rendering unit. Use principal component analysis to process complex data.
It realizes efficient utilization of resources, supports horizontal expansion, improves simulation efficiency and system stability, can effectively handle large-scale simulation needs, and promotes data analysis and decision-making through powerful visualization functions.
Smart Images

Figure CN119088573A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of large-scale simulation, and in particular to a software architecture supporting ultra-large-scale simulation. Background Art
[0002] At present, typical pure software simulation products are usually single processes, which use multi-threading to control multiple vehicles and equipment in the simulation. The simulation control unit needs to maintain the thread corresponding to each vehicle and poll the status of all vehicles and equipment in a single simulation cycle. This method can meet the needs well when the number of vehicles and equipment is small. However, as the number of vehicles and equipment increases, the performance disadvantages will gradually become prominent; first, when the number of threads maintained by the simulation control unit is large, it will bring a series of resource and performance problems, including: the overhead caused by frequent context switching, resource competition caused by poor synchronization mechanism, memory consumption caused by maintenance threads, and possible unpredictable behavior caused by inter-thread communication. Secondly, when there are many vehicles and equipment in the simulation, the corresponding cost of polling status will also increase, which will not only affect the actual running time of the simulation cycle, but also cause delays in the status of vehicle equipment. Thirdly, using a single simulation control unit does not support horizontal expansion and cannot fully utilize machine resources. Therefore, a software architecture that supports ultra-large-scale simulation is designed. Summary of the invention
[0003] The purpose of the present invention is to provide a software architecture that supports ultra-large-scale simulation to solve the problems of single process, lack of support for horizontal expansion, high load and unbalanced resources proposed in the above background technology.
[0004] To achieve the above object, the present invention aims to provide a software architecture supporting ultra-large-scale simulation, including: A control unit, wherein the control unit generates a simulation task according to user configuration information; A scheduling unit, wherein the scheduling unit allocates the simulation tasks to corresponding engine units according to the priorities of the simulation tasks and resource availability; An engine unit, wherein the engine unit receives a simulation task and performs simulation calculations; A rendering unit is used to obtain simulation calculation results and visualize them.
[0005] As a further improvement of the technical solution, the control unit generates a simulation task according to the user configuration information, including the following steps: S1.1, receiving user configuration information through the interface module, generating a configuration file from the user configuration information and parsing the user configuration information; S1.2. Create a simulation task description document based on the parsed user configuration information, and set task parameters according to the user configuration information. The task parameters include simulation type, simulation model parameters, and output format. S1.3, build a task object containing all necessary information, including task ID, task description, task parameters, and set the initial state of the task; S1.4. Submit the constructed task object to the task queue and transmit it to the scheduling unit.
[0006] As a further improvement of the technical solution, the scheduling unit allocates the simulation task to the corresponding engine unit according to the priority of the task and the availability of resources, including the following steps: S2.1. Determine the priority of each task, understand the resource requirements of each task, and collect other relevant information; S2.2, real-time monitoring of the status of all available resources, including idle and occupied status, and collection and analysis of performance indicators of engine units; S2.3, sorting the tasks according to the priority scheduling algorithm to form new priority information, and dynamically adjusting the task queue according to the new priority information during operation; S2.4. Select resources for allocation based on task priority and resource requirements; S2.5, assigning the task to the selected engine unit and sending the task description information; S2.6. Continuously monitor the execution status of the task to ensure that the task runs as expected. When the task is completed, confirm the task result and release the occupied resources.
[0007] As a further improvement of the technical solution, in S2.3, sorting the tasks according to the priority scheduling algorithm to form new priority information includes the following steps: S2.31. Create a list of all pending tasks. Each task has a unique identifier. Each task should contain at least the following attributes: task name, priority, required resources, and arrival time. S2.32. Define the priority criteria based on the importance and urgency of the task, and assign a specific priority value to each task; S2.33, select the quick sort algorithm to sort the task list, and sort the task list in descending order according to the priority; S2.34. During the task execution process, dynamically adjust the task priority according to the actual execution status of the task and the system status. If the priority changes, re-sort the task list; S2.35. After the sorting is completed, a new task queue is formed, in which the tasks are arranged from high to low priority.
[0008] As a further improvement of the technical solution, the engine unit receives the simulation task and performs the simulation calculation, including the following steps: S3.1, receiving a simulation task from a scheduling unit, parsing task information, and extracting a simulation task description, the simulation task description including a simulation model, parameter configuration, and input data; S3.2, configure the simulation environment according to the simulation task description; S3.3, obtain various model data through the model module, and set model parameters according to the parameters in the simulation task description; S3.4, setting the simulation start state, which includes initializing variables and setting the simulation time starting point; entering the simulation loop, gradually advancing the simulation process through the backward Euler method, performing calculations according to the definition of the model module and the input data, and adjusting the simulation results according to the boundary conditions during the simulation process; handling any abnormal situations that occur during the simulation process; S3.5. Save the simulation results, which include state variables, time series data, and statistics.
[0009] As a further improvement of the technical solution, in S3.4, the backward Euler method is specifically: ; in, Represents the state of the next time step; represents the state of the current time step; h represents the time step; f represents the right-hand side function of the differential equation.
[0010] As a further improvement of the technical solution, in S3.4, adjusting the simulation results according to the boundary conditions includes the following steps: S3.41, determine the start time of the simulation; initialize all relevant state variables, including position, speed, and temperature; set various parameters required for the simulation; S3.42, read the specific information of the boundary conditions from the simulation task description, including the boundary value and the derivative value; S3.43, set periodic boundary conditions; S3.44, set the initial time, and perform calculations in each time step according to the model module definition and input data; S3.45. Check numerical stability after each update.
[0011] As a further improvement of the technical solution, the rendering unit obtains the simulation calculation results and visualizes them, including the following steps: S4.1. Receive simulation calculation result data from the engine unit, where the simulation calculation result data includes state variables, time series data, and statistics; S4.2, parsing the received simulation calculation result data, converting the original simulation calculation result data into a visual form, and converting the data into a visual form includes arranging the time series data into a table form; S4.3, preprocessing the simulation calculation result data; S4.4, mapping the simulation calculation result data to the visualization coordinate system through the principal component analysis method, mapping different colors according to the simulation calculation result data values, and mapping different sizes according to the simulation calculation result data values; S4.5. Create simulation graphics through the visualization tool Matplotlib, and display the simulation graphics in real time on the interface module.
[0012] As a further improvement of the technical solution, in S4.4, the simulation calculation result data is mapped to a visualization coordinate system by a principal component analysis method, comprising the following steps: S4.41, subtracting the mean of the corresponding feature from each simulation calculation result data for preprocessing; S4.42, calculating the covariance matrix of the simulation calculation result data; S4.43, calculate eigenvalues and eigenvectors; S4.44, sort the eigenvectors by the size of the eigenvalue. The larger the eigenvalue, the larger the variance in that direction, that is, the more important that direction is. Select the eigenvectors corresponding to the first k largest eigenvalues as the principal components; S4.45. Use the selected principal components to project the simulation calculation result data into a low-dimensional space.
[0013] As a further improvement of the technical solution, in S4.42, the covariance matrix of the simulation calculation result data is calculated as follows: ; Wherein, C represents the covariance matrix of the simulation calculation result data; N represents the number of samples; X represents the original simulation calculation result data matrix; and T represents the transposition operation.
[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. In the software architecture that supports ultra-large-scale simulation, resources can be used more efficiently by dynamically allocating tasks based on the priority of the task and the resource availability of the current system. At the same time, it allows users to flexibly configure simulation tasks according to their own needs, and the system can intelligently manage and schedule tasks based on these configurations, thereby improving simulation efficiency and reducing waiting time and resource waste. By splitting the overall modules of the simulation system and isolating the resources of each sub-module, the operating environment of each sub-module is independent and has horizontal scalability. While improving the stability of the entire system, it can also meet the actual needs of large-scale simulation.
[0015] 2. The software architecture that supports ultra-large-scale simulation can convert complex simulation calculation results into intuitive visual charts, helping users to understand and analyze simulation results more easily. In addition, by adopting advanced data processing technologies such as principal component analysis, the system can process and display large amounts of data, which is particularly important for understanding multi-dimensional and highly complex simulation results. This powerful visualization function not only enhances the user's understanding of simulation data, but also facilitates further data analysis and decision making. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is the overall flow chart of the present invention; The meaning of each number in the figure is: 1. Control unit; 2. Scheduling unit; 3. Engine unit; 4. Rendering unit. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] Example: See Figure 1 As shown, a software architecture supporting ultra-large-scale simulation is provided, including: The control unit 1 generates a simulation task according to the user configuration information; In this embodiment, the control unit 1 generates a simulation task according to the user configuration information, including the following steps: S1.1, receiving user configuration information through the interface module, generating a configuration file from the user configuration information and parsing the user configuration information; Among them, the interface module is used to display the simulation process and provide visual configuration; S1.2. Create a simulation task description document based on the parsed user configuration information, and set task parameters according to the user configuration information. The task parameters include simulation type, simulation model parameters, and output format. S1.3, build a task object containing all necessary information, including task ID, task description, task parameters, and set the initial state of the task, such as "pending" or "queued"; S1.4, submit the constructed task object to the task queue and transmit it to the scheduling unit 2.
[0019] The scheduling unit 2 allocates the simulation tasks to the corresponding engine units 3 according to the priorities of the simulation tasks and the availability of resources; In this embodiment, the scheduling unit 2 allocates the simulation task to the corresponding engine unit according to the priority of the task and the availability of resources, including the following steps: S2.1. Determine the priority of each task, assign it by user or automatically according to the task type, understand the resource requirements of each task, including CPU, memory, storage space, etc.; collect other relevant information, such as the expected execution time of the task, dependencies, etc.; S2.2, real-time monitoring of the status of all available resources, including idle and occupied status, and collection and analysis of performance indicators of the engine unit 3, such as CPU utilization, memory usage, etc.; S2.3, sorting the tasks according to the priority scheduling algorithm to form new priority information, and dynamically adjusting the task queue according to the new priority information during operation; Among them, by assigning different priorities to different tasks and sorting them according to these priorities, the system can give priority to more critical or urgent tasks, which can ensure that important tasks are processed in a timely manner and avoid delays in critical tasks due to low-priority tasks occupying resources. This helps to improve the responsiveness and efficiency of the entire system, especially when resources are limited, and can better meet the needs of different tasks; as the simulation tasks are executed, their actual needs may change. For example, some tasks may require more time or resources than initially estimated. The function of dynamically adjusting task priorities allows the system to re-evaluate the importance of tasks according to actual conditions and re-sort the task queue accordingly, which enables the system to remain flexible in a changing environment and adjust strategies in a timely manner to cope with emergencies; sorting tasks according to the priority scheduling algorithm to form new priority information includes the following steps: S2.31. Create a list of all pending tasks. Each task has a unique identifier (such as a task ID). Each task should contain at least the following attributes: task name, priority, required resources (such as CPU time, memory requirements, etc.), and arrival time. S2.32. Define the priority criteria based on the importance and urgency of the task, and assign a specific priority value to each task. The larger the value, the higher the priority. S2.33. Select the quick sort algorithm to sort the task list, and sort the task list in descending order according to priority (tasks with higher priority are placed first); S2.34. During task execution, dynamically adjust the priority of tasks according to the actual execution status of tasks and system status. For example, if a task execution time exceeds expectations, its priority can be lowered. If the priority changes, re-sort the task list; S2.35. After the sorting is completed, a new task queue is formed, in which the tasks are arranged from high to low priority; S2.4. Select resources for allocation based on task priority and resource requirements; S2.5, assigning the task to the selected engine unit 3 and sending the task description information; S2.6. Continuously monitor the execution status of the task to ensure that the task runs as expected. When the task is completed, confirm the task result and release the occupied resources.
[0020] The engine unit 3 receives the simulation task and performs the simulation calculation; In this embodiment, the engine unit 3 receives the simulation task and performs simulation calculation, including the following steps: S3.1, receiving a simulation task from the scheduling unit 2, parsing the task information, and extracting a simulation task description, the simulation task description including a simulation model, parameter configuration, and input data; S3.2, configure the simulation environment according to the simulation task description, including loading necessary libraries, setting environment variables, etc.; S3.3, obtain various model data through the model module, and set the model parameters according to the parameters in the simulation task description to ensure that the model correctly reflects the user's needs; Among them, the model module is a module that provides vehicle information in the simulation (including but not limited to posture, position, sensors, etc.); S3.4, set the simulation start state, which includes initializing variables and setting the simulation time starting point; enter the simulation loop, gradually advance the simulation process through the backward Euler method, calculate according to the definition of the model module and the input data, and adjust the simulation results according to the boundary conditions during the simulation process; handle any abnormal conditions that occur during the simulation process, such as numerical instability, overflow, etc.; Among them, the backward Euler method is an implicit method, which means that it can handle rigid problems without causing divergence of the solution. When dealing with systems with large differential time scales, this method can provide better numerical stability and maintain the stability and accuracy of the solution even at large time steps; Due to the stability and convergence characteristics of the backward Euler method, it is very suitable for long-term simulation. In simulations with long time spans, stability is often more important than the accuracy of high-order methods, because even small cumulative errors may lead to inaccurate results in long-term simulations; For large-scale systems, the backward Euler method can be combined with parallel computing technology to speed up the simulation process through distributed computing resources, because large systems can be decomposed into smaller parts and then solved in parallel on different processors; The specific backward Euler method is: ; in, Represents the state of the next time step; Indicates the state of the current time step; h indicates the time step; f indicates the right-hand side function of the differential equation; Furthermore, boundary conditions directly reflect the external constraints or environmental influences of the system. In the simulation process, accurate application of boundary conditions can ensure that the behavior of the model is as consistent as possible with the behavior in the real world, which helps to improve the accuracy and reliability of the simulation and make the results more credible. Different application scenarios may correspond to different boundary conditions. By adjusting the boundary conditions, the same model can be adapted to a variety of different scenarios, increasing the flexibility and wide applicability of the model. The choice of boundary conditions directly affects the physical meaning of the simulation. For example, in fluid mechanics simulation, boundary conditions determine how the fluid interacts with the container wall. If these conditions are not set correctly, results that do not conform to physical laws may result, thereby misleading the analysis conclusions. Adjusting the simulation results according to the boundary conditions includes the following steps: S3.41, determine the start time of the simulation; initialize all relevant state variables, including position, speed, and temperature; set various parameters required for the simulation, such as time step, simulation time, etc.; S3.42, read the specific information of the boundary conditions from the simulation task description, including the boundary value and the derivative value; S3.43, set periodic boundary conditions; Among them, without periodic boundary conditions, the edges of the system may produce unnatural behaviors because the particles or phenomena at these edges will be affected by the external environment. Periodic boundary conditions eliminate these edge effects, making the behavior inside the system more uniform and reducing the artificial interference introduced by boundary conditions. For some phenomena that need to simulate a large or infinite range (such as crystal structure, fluid dynamics, etc.), it may be impractical to directly simulate the entire system. Periodic boundary conditions allow a small representative unit (called a super cell or unit cell) to replace the entire system, thereby simplifying the computational model and reducing the demand for computing resources. The periodic boundary condition is expressed as: ; In the formula, represents the new coordinates of the position point after the periodic boundary conditions are applied in the simulation space; x represents the initial coordinates of the position point in the simulation space; LA represents the period length of the periodic boundary; S3.44, set the initial time, and perform calculations in each time step according to the model module definition and input data; S3.45. Check numerical stability after each update, such as whether the speed or position is reasonable; S3.5. Save the simulation results, which include state variables, time series data, and statistics.
[0021] The rendering unit 4 obtains the simulation calculation results and visualizes them; In this embodiment, the rendering unit 4 obtains the simulation calculation results and visualizes them, including the following steps: S4.1, receiving simulation calculation result data from the engine unit 3, the simulation calculation result data including state variables, time series data, and statistics; S4.2, parsing the received simulation calculation result data, converting the original simulation calculation result data into a visual form, and converting the data into a visual form includes arranging the time series data into a table form; S4.3, preprocessing the simulation calculation result data, including data aggregation, data smoothing and data cleaning; S4.4, mapping the simulation calculation result data to the visualization coordinate system through the principal component analysis method, mapping different colors according to the simulation calculation result data values, and mapping different sizes according to the simulation calculation result data values; Among them, principal component analysis is an effective data dimensionality reduction technique that can convert high-dimensional data into low-dimensional data while retaining the main features of the data. This is particularly useful for visualization, as most visualization tools are limited to two-dimensional or three-dimensional space; principal component analysis can identify correlations in the data and remove this redundant information in a new coordinate system. This can reduce unnecessary variables, make the analysis more concise, and also reduce the computational burden; the visualization results obtained by principal component analysis can provide decision makers with clear information to help them make more informed decisions based on the simulation results. Especially in cases involving multi-objective or multi-factor analysis, principal component analysis can help highlight the most important factors; mapping the simulation calculation result data to the visualization coordinate system through principal component analysis includes the following steps: S4.41, subtracting the mean of the corresponding feature from each simulation calculation result data for preprocessing; S4.42, calculating the covariance matrix of the simulation calculation result data; Furthermore, the covariance matrix of the simulation result data is calculated as follows: ; Wherein, C represents the covariance matrix of the simulation calculation result data; N represents the number of samples; X represents the original simulation calculation result data matrix; and T represents the transposition operation.
[0022] S4.43. Calculate eigenvalues and eigenvectors: , v represents the eigenvector, represents the eigenvalue; S4.44, sort the eigenvectors by the size of the eigenvalue. The larger the eigenvalue, the larger the variance in that direction, that is, the more important that direction is. Select the eigenvectors corresponding to the first k largest eigenvalues as the principal components; S4.45, projecting the simulation calculation result data into a low-dimensional space using the selected principal components; Furthermore, different colors are mapped according to the data values of the simulation calculation results, and different sizes are mapped according to the data values of the simulation calculation results as follows: import numpy as np import matplotlib.pyplot as plt from matplotlib import cm # Simulate simulation data data_values = np.random.rand(100) # Assume we have 100 simulated data points # Data normalization data_normalized = (data_values - np.min(data_values)) / (np.max(data_values) - np.min(data_values)) # Define color mapping cmap = cm.get_cmap('viridis') # Use the 'viridis' color map # Mapping data to colors colors = cmap(data_normalized) # Mapping data to size sizes = data_normalized 100 + 10 # Assume the minimum size is 10 and the maximum is 110 # Create graphics and draw plt.scatter(range(len(data_values)), data_values, c=colors, s=sizes) # Display the graph plt.show() S4.5. Create simulation graphics through the visualization tool Matplotlib, and display the simulation graphics in real time on the interface module.
[0023] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.
Claims
1. A software architecture supporting very large-scale simulation, characterized in that: include: A control unit (1), wherein the control unit (1) generates a simulation task according to user configuration information; A scheduling unit (2), wherein the scheduling unit (2) allocates the simulation task to the corresponding engine unit (3) according to the priority of the simulation task and the availability of resources; An engine unit (3), wherein the engine unit (3) receives a simulation task and performs simulation calculations; A rendering unit (4), wherein the rendering unit (4) obtains simulation calculation results and visualizes them.
2. The software architecture supporting ultra-large-scale simulation according to claim 1, characterized in that: The control unit (1) generates a simulation task according to user configuration information, comprising the following steps: S1.1, receiving user configuration information through the interface module, generating a configuration file from the user configuration information and parsing the user configuration information; S1.
2. Create a simulation task description document based on the parsed user configuration information, and set task parameters according to the user configuration information. The task parameters include simulation type, simulation model parameters, and output format. S1.3, build a task object containing all necessary information, including task ID, task description, task parameters, and set the initial state of the task; S1.
4. Submit the constructed task object to the task queue and transmit it to the scheduling unit (2).
3. The software architecture supporting ultra-large-scale simulation according to claim 1, characterized in that: The scheduling unit (2) allocates the simulation task to the corresponding engine unit (3) according to the priority of the task and the availability of resources, comprising the following steps: S2.
1. Determine the priority of each task, understand the resource requirements of each task, and collect other relevant information; S2.2, real-time monitoring of the status of all available resources, including the idle and occupied status of available resources, and collection and analysis of performance indicators of the engine unit (3); S2.3, sorting the tasks according to the priority scheduling algorithm to form new priority information, and dynamically adjusting the task queue according to the new priority information during operation; S2.
4. Select resources for allocation based on task priority and resource requirements; S2.5, assigning the task to the selected engine unit (3) and sending task description information; S2.
6. Continuously monitor the execution status of the task to ensure that the task runs as expected. When the task is completed, confirm the task result and release the occupied resources.
4. The software architecture supporting ultra-large-scale simulation according to claim 3, characterized in that: In S2.3, the tasks are sorted according to the priority scheduling algorithm to form new priority information, including the following steps: S2.
31. Create a list of all pending tasks. Each task has a unique identifier. Each task should contain at least the following attributes: task name, priority, required resources, and arrival time. S2.
32. Define the priority criteria based on the importance and urgency of the task, and assign a specific priority value to each task; S2.33, select the quick sort algorithm to sort the task list, and sort the task list in descending order according to the priority; S2.
34. During the task execution process, dynamically adjust the task priority according to the actual execution status of the task and the system status. If the priority changes, re-sort the task list; S2.
35. After the sorting is completed, a new task queue is formed, in which the tasks are arranged from high to low priority.
5. The software architecture supporting ultra-large-scale simulation according to claim 1, characterized in that: The engine unit (3) receives the simulation task and performs simulation calculation, including the following steps: S3.1, receiving a simulation task from the scheduling unit (2), parsing the task information, and extracting a simulation task description, wherein the simulation task description includes a simulation model, parameter configuration, and input data; S3.2, configure the simulation environment according to the simulation task description; S3.3, obtain various model data through the model module, and set model parameters according to the parameters in the simulation task description; S3.4, setting the simulation start state, which includes initializing variables and setting the simulation time starting point; entering the simulation loop, gradually advancing the simulation process through the backward Euler method, performing calculations according to the definition of the model module and the input data, and adjusting the simulation results according to the boundary conditions during the simulation process; handling any abnormal situations that occur during the simulation process; S3.
5. Save the simulation results, which include state variables, time series data, and statistics.
6. The software architecture supporting ultra-large-scale simulation according to claim 5, characterized in that: In S3.4, the backward Euler method is specifically: ; in, Represents the state of the next time step; represents the state of the current time step; h represents the time step; f represents the right-hand side function of the differential equation.
7. The software architecture supporting ultra-large-scale simulation according to claim 5, characterized in that: In S3.4, adjusting the simulation results according to the boundary conditions includes the following steps: S3.41, determine the start time of the simulation; initialize all relevant state variables, including position, speed, and temperature; set various parameters required for the simulation; S3.42, read the specific information of the boundary conditions from the simulation task description, including the boundary value and the derivative value; S3.43, set periodic boundary conditions; S3.44, set the initial time, and perform calculations in each time step according to the definition of the model module and the input data; S3.
45. Check numerical stability after each update.
8. The software architecture supporting ultra-large-scale simulation according to claim 1, characterized in that: The rendering unit (4) obtains the simulation calculation results and visualizes them, including the following steps: S4.1, receiving simulation calculation result data from the engine unit (3), the simulation calculation result data including state variables, time series data, and statistics; S4.2, parsing the received simulation calculation result data, converting the original simulation calculation result data into a visual form, and converting the data into a visual form includes arranging the time series data into a table form; S4.3, preprocessing the simulation calculation result data; S4.4, mapping the simulation calculation result data to the visualization coordinate system through the principal component analysis method, mapping different colors according to the simulation calculation result data values, and mapping different sizes according to the simulation calculation result data values; S4.
5. Create simulation graphics through the visualization tool Matplotlib, and display the simulation graphics in real time on the interface module.
9. The software architecture supporting ultra-large-scale simulation according to claim 1, characterized in that: In S4.4, the simulation calculation result data is mapped to a visualization coordinate system by principal component analysis, including the following steps: S4.41, subtracting the mean of the corresponding feature from each simulation calculation result data for preprocessing; S4.42, calculating the covariance matrix of the simulation calculation result data; S4.43, calculate eigenvalues and eigenvectors; S4.44, sort the eigenvectors by the size of the eigenvalue. The larger the eigenvalue, the larger the variance in that direction, that is, the more important that direction is. Select the eigenvectors corresponding to the first k largest eigenvalues as the principal components; S4.
45. Use the selected principal components to project the simulation calculation result data into a low-dimensional space.
10. The software architecture supporting very large-scale simulation according to claim 9, characterized in that: In S4.42, the covariance matrix of the simulation result data is calculated as follows: ; Wherein, C represents the covariance matrix of the simulation calculation result data; N represents the number of samples; X represents the original simulation calculation result data matrix; and T represents the transposition operation.
Citation Information
Patent Citations
Configurable distributed three-dimensional scene simulating system
CN103593181A
Simulation IOT resource service system and method
CN106027319A
Logistics park and storage decision support software platform system
CN116976069A
Airflow organization analogue simulation method and system
CN118520675A
Simulation resource scheduling method based on knowledge graph
CN118550654A
