Engine scheduling method for mixed granularity-oriented digital model

Abstract

The invention relates to an engine scheduling method for a mixed granularity-oriented digital model. The engine scheduling method comprises the following steps of: S1, sorting models, constructing a model interaction relation matrix DSM, grading and sorting simulation models, and adjusting a matrix sequence to determine a model execution priority; s2, a task scheduling strategy is designed, an improved centralized scheduling strategy is adopted, task distribution is globally controlled by a scheduling host, and nodes actively report loads and execute partial task scheduling; s3, based on a frame period of system simulation and a multi-rate distributed simulation node synchronization mechanism, a hybrid task scheduling algorithm based on priority is executed, and the priority sequence is periodic tasks gt; an accidental task gt; and background tasks. According to the invention, engine scheduling of the digital model under the mixed granularity can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of simulation and modeling, and in particular to an engine scheduling method for digital models oriented to mixed granularity. Background Art

[0002] With the continuous growth of the demand for large-scale system simulation in the fields of aviation, aerospace, shipbuilding, and ordnance, the complexity and scale of digital models have shown an exponential increase. The simulation system needs to integrate entity models from different sources, with different granularities (such as high-precision physical models and low-precision behavior models), and of different scales to support multi-disciplinary collaborative simulation. In this context, the task scheduling mechanism of the simulation engine has become the core challenge for the efficient operation of the system.

[0003] In the prior art, simulation modeling and scheduling engines are important components in large-scale system simulation in the fields of aviation, aerospace, shipbuilding, and ordnance. Facing the need for synchronous simulation of entity models from different sources, with different granularities, and of different scales, how to determine the execution order between models, formulate an engine scheduling mechanism, and achieve synchronous scheduling of models with different priorities and different granularities is an urgent problem to be solved. Summary of the Invention

[0004] In view of the above technical problems, the present invention proposes an engine scheduling method for digital models oriented to mixed granularity to achieve engine scheduling of digital models under mixed granularity.

[0005] The technical solution for solving the technical problems of the present invention is: an engine scheduling method for digital models oriented to mixed granularity, including the following steps:

[0006] Step S1, model sorting: construct a model interaction relationship matrix DSM, perform hierarchical sorting on the simulation models, and adjust the matrix order to determine the model execution priority;

[0007] Step S2, design a task scheduling strategy: adopt an improved centralized scheduling strategy, where the scheduling host globally controls task allocation, and the nodes actively report the load and execute part of the task scheduling;

[0008] Step S3, based on the frame period of system simulation and the multi-rate distributed simulation node synchronization mechanism, execute a priority-based hybrid task scheduling algorithm, and the priority order is: periodic task > sporadic task > background task.

[0009] According to a technical solution of the present invention, in step S1, when performing hierarchical sorting on the simulation models, it specifically includes:

[0010] Step S11, construct a model interaction relationship matrix, and use a hierarchical method to take the model without input as the first level and gradually deduce the subsequent models;

[0011] Step S12: Adjust the matrix order and complete the model execution sequence planning through source point and sink point positioning and closed-loop processing.

[0012] According to a technical solution of the present invention, in step S11, the classification method is:

[0013] The model without other model input is regarded as the first level, the data receiving model of the first level model is regarded as the second level model, and so on; and the model that has been graded will no longer participate in the subsequent grading.

[0014] According to a technical solution of the present invention, in step S12, the closed-loop processing includes:

[0015] Perform a coupling operation on the model with directed loops, use the normalization operation to treat the coupled model as a submodule, and select the source point in the submodule to continue sorting.

[0016] According to a technical solution of the present invention, in step S2, the improved centralized scheduling strategy includes:

[0017] Step S21: The node actively reports load information to the scheduling host periodically or in a threshold-triggered manner;

[0018] Step S22: The scheduling host delegates part of the scheduling decision-making power to the node, and the node performs task allocation according to the local state.

[0019] According to a technical solution of the present invention, the threshold-triggered active reporting of load information includes:

[0020] When the load variable of a node exceeds a preset threshold, the load change exceeds a difference threshold, or a request is received from the scheduling host, the node actively reports the load information.

[0021] According to a technical solution of the present invention, in step S3, the multi-rate distributed simulation node synchronization mechanism includes:

[0022] Step S31: The frame period of the system simulation is used as the logical time unit, and the node step length is an integer multiple of the frame period;

[0023] Step S32: The node starts or pauses operation according to the timestamp of the received data and the logical time relationship.

[0024] According to a technical solution of the present invention, the frame period of the system simulation is the greatest common divisor of the step lengths of all nodes, and the scheduling host only interacts with the nodes that meet the step length requirements.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] According to the concept of the present invention, an engine scheduling method for a digital model facing hybrid granularity is proposed, in which different granularity models in the simulation space are sorted according to logical and timing relationships and then allocated to each computing node; an improved centralized scheduling strategy is adopted to determine the task allocation and scheduling method during the simulation process; an appropriate simulation step size is used to formulate the system simulation frame period; a multi-rate distributed simulation node synchronization mechanism is designed, and on this basis, a hybrid task scheduling algorithm based on priority is executed to achieve the engine scheduling of the digital model under hybrid granularity.

[0027] In the present invention, a task scheduling strategy is introduced, and the sorted models are scheduled according to the improved centralized scheduling strategy. Each node actively calculates its own load and operating status regularly and sends them to the scheduling host. Part of the scheduling tasks are handed over to the nodes for processing, so that the overall situation of task allocation and scheduling can be grasped.

[0028] The present invention designs a hybrid task scheduling algorithm based on priority to determine the system simulation frame period. Facing the problem of inconsistent advancement rates of multi-granularity models, a multi-rate distributed simulation node synchronization mechanism is designed and, on this basis, a hybrid task scheduling algorithm based on priority is executed to ensure the integrity, authenticity, and real-time performance of the entire simulation system.

[0029] The present invention has good application prospects in terms of simulation engines, deduction, and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Schematically shows a flowchart of an engine scheduling method for a digital model facing hybrid granularity according to an embodiment of the present invention;

[0031] Figure 2 Schematically shows a model grading process diagram according to an embodiment of the present invention;

[0032] Figure 3 Schematically shows a model DSM matrix according to an embodiment of the present invention;

[0033] Figure 4 Schematically shows a circular model dependency relationship according to an embodiment of the present invention;

[0034] Figure 5 Schematically shows a centralized scheduling model according to an embodiment of the present invention;

[0035] Figure 6 Schematically shows a distributed scheduling model according to an embodiment of the present invention

[0036] Figure 7 Schematically shows an improved centralized scheduling strategy according to an embodiment of the present invention;

[0037] Figure 8 Schematic diagram showing the execution sequence of a hybrid task scheduling algorithm according to an embodiment of the present invention

[0038] Figure 9 Schematic flowchart showing a non-real-time scheduling algorithm according to an embodiment of the present invention. Detailed implementation manners

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings required for use in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0040] The present invention will be described in detail below in conjunction with the accompanying drawings and specific implementation manners. The implementation manners cannot be enumerated one by one here, but the implementation manners of the present invention are not limited to the following implementation manners.

[0041] As Figure 1 shown, an engine scheduling method for a digital model oriented to hybrid granularity according to the present invention includes the following steps:

[0042] Step S1, model sorting

[0043] Construct a model interaction relationship matrix DSM, hierarchically sort the simulation models, and adjust the matrix order to determine the model execution priority.

[0044] In a distributed simulation system, a task refers to a model component with a certain function and independent operation, which is the smallest entity for scheduling. Simulation models involve multiple disciplines, with complex calculations and a huge amount of computational work, so there are mutual influences. Facing the dependency relationships between models, the input-output relationships between models are considered during task decomposition, that is, the problem of the order of task execution is solved. Therefore, before task decomposition, it is necessary to plan the solution order of the models in the simulation application first.

[0045] In step S1, it includes constructing a model interaction relationship matrix and adjusting the matrix order.

[0046] Step S11, construct a model interaction relationship matrix, and take the model without input as the first level through a hierarchical method, and gradually deduce the subsequent models;

[0047] Construct a two-dimensional matrix DSM (Design Structure Matrix) for model interaction. Its dynamic characteristics and the recognition and sorting algorithms for the unique loops and bands can clearly show the information flow between models, and can ensure that the information required for each task is obtained before the execution of this task to the greatest extent. DSM is an n-order matrix used to display the interaction relationships of various elements in the matrix, which is conducive to the visual analysis of complex project relationships. In the matrix, "1" and "0" are used to indicate whether there is a data relationship between models. The models corresponding to the rows marked as "1" in each column represent the external output models of the models corresponding to this column. That is, as shown in the table, models M4 and M2 receive the output data of model M1; "0" indicates that there is no data interaction relationship between the corresponding two models. The DSM matrix contains each model that makes up the simulation application and the way of information exchange between models. From it, it can be seen which models' output information is required at the beginning of a certain model and which models will receive the data generated by this model. In the matrix, all the input sources of the model in a certain row can be found; from a certain column, it can be seen which models (i.e., the models corresponding to 1) receive the output information of this model. The information in the table below the diagonal represents feedforward, while the information above the diagonal represents feedback information.

[0048] / Ml M2 M3 M4 Ml 0 0 1 0 M2 1 0 0 1 M3 0 1 0 0 M4 1 0 0 0

[0049] Table 1

[0050] Before constructing the matrix, it is necessary to perform a primary classification on the models in the experiment to simplify the subsequent operations on the matrix. The specific classification method: The models without input from other models are used as the first level, and the data receiving models of the first-level models are used as the second-level models, and so on. Since there will inevitably be information feedback between models in the experiment, it is stipulated that the models that have been classified will no longer participate in subsequent classifications, and the process is as Figure 2 shown.

[0051] After the classification is completed, the DSM matrix is constructed in the order of increasing model levels. If the number of models in the same level is more than one, then the order of the models in this level can be arranged arbitrarily or written in ascending order according to some existing attributes of the models when constructing the DSM. As Figure 3 shown in the dependency relationship between the 7 models, after classification, the preliminary sorting of the models is: A, B, C, D, E, F, G, and the design structure matrix is constructed accordingly.

[0052] Step S12: Adjust the matrix order and complete the model execution order planning through source point, sink point positioning and closed-loop processing.

[0053] Adjust the source points in the matrix to the first row and the first column of the DSM, and record the candidate source points; adjust the sink points in the matrix to the last row and the last column of the DSM, and record the candidate sink points.

[0054] In step S1, a preliminary sorting of the execution order of the models has been carried out. However, for cases where there is a closed-loop state or multiple models at the same level, the execution order of the models needs to be further adjusted. For structures such as Figure 4 shown, the model in the first row corresponding to the DSM matrix is taken as the source point.

[0055] In the matrix excluding the source point and the sink point, the secondary source point and sink point can be found according to the above method, and the priority order of model execution can be found in turn.

[0056] If the number of models is n and there is no directed cycle, after (n + 1) / 2 times of the above process, the sorting of the models will be obtained.

[0057] If there is a directed loop, then there will be no source point in the matrix after removing the source point. At this time, a decoupling operation is performed on the matrix, and a normalization operation is used to regard the coupled models as a sub-module. Then, one of them is selected as the source point according to the requirements (rules) in the sub-module, and the previous process is continued.

[0058] After the above sorting process, the models in the entire simulation task can be decomposed into independent running modules with a sequential order or sub-modules with a coupling relationship, which is convenient for subsequent task allocation.

[0059] Step S2: Design the task scheduling strategy

[0060] An improved centralized scheduling strategy is adopted, where the scheduling host globally controls the task allocation, and the nodes actively report the load and execute part of the task scheduling.

[0061] Determine the task allocation and scheduling method in the simulation process. After the model sorting, design the model scheduling strategy, map multiple simulation tasks to multiple simulation nodes according to certain rules, and use the improved centralized scheduling strategy to arrange the execution order of the tasks to meet the time requirements of the simulation tasks and achieve the effective utilization of resources.

[0062] Centralized and distributed scheduling strategies

[0063] Scheduling strategies are mainly divided into centralized (such as Figure 5 shown) and distributed (such as Figure 6As shown in the figure. Centralized means that there is a scheduling service host responsible for uniformly scheduling the model and collecting information from each node; distributed means that each node collects information in parallel and sends scheduling requests to other nodes according to its current state, and other nodes respond to the requests according to their current states. Centralized and distributed scheduling each have their own advantages and disadvantages. The main advantage of centralized scheduling is that the scheduling host has global information and the program algorithm implementation is separated from the logic, but the scheduling host is prone to becoming the bottleneck of the scheduling service; the main advantage of distributed scheduling is that the scalability of the node system is good, but it is difficult to track the running of the program.

[0064] Improved centralized scheduling strategy

[0065] The distributed simulation system should select the corresponding scheduling strategy according to the simulation fidelity, simulation tasks, load, and iterative mode of the simulation model. If a distributed scheduling strategy is adopted, since the scheduling nodes and computing nodes of each system are self-contained, the information interaction or communication between them is completely realized by their own iterative rate and interface communication mechanism, so there may be confusion in the interaction information between them, which not only increases the difficulty of the system, but also affects the performance of the entire system. Therefore, the general distributed simulation system adopts a centralized scheduling strategy, which can not only have an overall grasp of the task allocation and scheduling, but also be relatively easy to implement. This patent makes some improvements to alleviate the system bottleneck brought by the centralized strategy, such as Figure 7 As shown in the figure, it is mainly reflected in the following two aspects:

[0066] (1) Change the scheduling host to track and collect the load information of the nodes to each node actively calculating its own load and running state regularly and sending it to the scheduling host;

[0067] (2) Change the scheduling host to be fully responsible for the scheduling tasks to handing over part of the scheduling tasks to the nodes for processing, that is, after the nodes complete the calculation, they make partial task scheduling according to the scheduling mechanism.

[0068] The computing node active reporting strategy can collect load information by combining periodic collection and state change strategy. For example, the computing node needs to send load information to the scheduling node when the load variable exceeds its given threshold, the load change amount exceeds its given difference, or the scheduling node sends a request message.

[0069] After the task allocation and scheduling method are determined, in order to ensure the integrity, authenticity and real-time performance of the entire simulation system function, we also need to study the task startup method and the simulation time advancement algorithm.

[0070] Step S3, design a task scheduling algorithm

[0071] Frame period and multi-rate distributed simulation node synchronization mechanism based on system simulation, execute a hybrid task scheduling algorithm based on priorities, and the priority order is: periodic tasks > sporadic tasks > background tasks.

[0072] After the task allocation and scheduling methods are determined, in order to ensure the integrity, authenticity, and real-time performance of the entire simulation system's functions, it is also necessary to study the task startup method and the simulation time advancement algorithm. First, determine the system simulation frame period. Facing the problem of inconsistent advancement rates of multi-granularity models, design and adopt a multi-rate distributed simulation node synchronization mechanism. On this basis, execute a hybrid task scheduling algorithm based on priorities, and then realize the engine scheduling of the hybrid granularity digital model.

[0073] In step S3, it includes formulating the frame period of system simulation, a hybrid task scheduling algorithm based on priorities, and a node synchronization algorithm in multi-rate distributed simulation.

[0074] Formulate the frame period of system simulation

[0075] During the simulation process, if the system uses a smaller time step as the frame period for scheduling, it not only does not significantly improve the accuracy of the simulation results, but will instead increase system resource consumption; if a larger simulation time step is used for scheduling, although it is beneficial for the simulation clock to advance, it will inevitably sacrifice the accuracy of the simulation results. Therefore, when designing the scheduling module of the distributed simulation system, it is necessary to comprehensively consider the simulation accuracy of the system and the system's load conditions to determine the frame period.

[0076] The solution idea for this problem is: select an appropriate simulation time step so that most (or even all) computing nodes can participate in the calculation at the start of a certain scheduling, and meet the time step requirements of each node. The management host schedules according to this time step. At each scheduling, it only exchanges data with the nodes that meet the time step requirements, rather than all nodes. For example, the greatest common divisor of all time steps can be selected as the system time step. When the task with the largest time step starts a new cycle of operation, all simulation tasks with smaller time steps start a new cycle of operation simultaneously. This algorithm not only ensures the simulation accuracy but also does not increase the workload of the computing nodes, only slightly increasing the workload of the scheduling host for time step matching.

[0077] Hybrid task scheduling algorithm based on priorities

[0078] The distributed simulation system mainly consists of periodic tasks, sporadic tasks, and background tasks. There is both a certain degree of correlation and a certain degree of independence among the three types of tasks. When these three tasks appear simultaneously, it is necessary to reasonably schedule the simulation tasks. The operation sequence diagram of the hybrid tasks is as Figure 8 shown. First, execute the periodic tasks, then execute the sporadic tasks, and finally execute the background tasks with higher priorities.

[0079] To accurately promote the sequential execution of the above simulation tasks, during non-real-time simulation, the scheduling host introduces a promotion flag. Only when all three tasks are completed is the simulation promotion flag available; during real-time simulation, a high-precision clock and a timer are introduced. The timer is created during initialization and the period of the timer is set. The timer is triggered in each period to execute one frame of simulation tasks. The priorities of the three tasks are set, with the periodic task having the highest priority, the sporadic task having the second highest priority, and the background task having the lowest priority. Thus, a cyclic scheduling algorithm based on priority step driving is designed to complete the scheduling of periodic tasks, sporadic tasks, and background tasks in the distributed simulation system. The pseudo-code of the non-real-time scheduler is as follows:

[0080] Input: The saved scheduling module L(k), where k = 0, 1,..., F - 1;

[0081] The aperiodic task queue, the sporadic task waiting queue, and the accepted sporadic task EDF queue;

[0082] Task initialization:

[0083] The current time t = 0;

[0084] The current frame k = 0;

[0085] Do forever

[0086] At time t f Accept the host task instruction;

[0087] CurrentBlock = L(k);

[0088] t = t + 1;

[0089] k = t mod F;

[0090] If no task is released in CurrentBlock, take appropriate actions;

[0091] While (the sporadic task queue is not empty)

[0092] Remove the task at the head of the sporadic task waiting queue;

[0093] Perform an acceptance test on this task;

[0094] If this task is acceptable

[0095] Insert this task into the accepted sporadic task queue in EDF order;

[0096] Else delete this task and notify the application;

[0097] Endwhile

[0098] Wake up the periodic task server to execute the slices in CurrentBlock;

[0099] Sleep until the periodic task server finishes;

[0100] While (the aperiodic task queue is not empty)

[0101] Wake up the task at the head of the aperiodic task;

[0102] Sleep until the aperiodic task finishes;

[0103] Remove this aperiodic task from the queue;

[0104] Endwhile;

[0105] Sleep until the next time step;

[0106] End do

[0107] The flow of the above scheduling algorithm is as Figure 9 shown. First, the program is loaded and initialized. If the non-real-time simulation mode is selected, the system time step is set, and control flags are created to drive the execution of periodic tasks, sporadic tasks, and background tasks.

[0108] Node Synchronization Algorithm in Multi-Rate Distributed Simulation

[0109] In a distributed simulation system, the scheduling algorithm must be able to handle various problems brought about by the spatio-temporal inconsistencies of multi-resolution models.

[0110] Node startup problem: The operation of the simulation node is controlled by input data, that is, the simulation node starts running when it receives new interaction data and waits for the next data update after the operation ends. The interaction data includes the data of the scheduling host and the data of other nodes, and the logical time of the data is controlled by the global time data structure of the system.

[0111] Initialization problem: To minimize the startup delay of the entire simulation system as much as possible, the scheduling host establishes all node interaction relationships completely before initialization and sends initial commands, initialization data, and interaction relationships to all computing nodes during initialization. This requires that all computing nodes start running immediately after receiving the first interaction data and there is no need for interaction data between nodes. The logical clock of the first interaction data is represented by 0.

[0112] Running startup issues: As shown in Figure 7, if the simulation step size of computing node 2 is m times that of computing node 1, then computing node 2 starts the simulation operation of the current timestamp only after receiving the data from computing node 1 m times in a row. At the same time, it updates its logical time to the current system clock and sends the result data after the operation ends. If the simulation step size of computing node 1 is m times that of computing node 2, after receiving m scheduling commands from the scheduling host or other nodes, simulation node 2 transmits the interactive data stream to computing node 1. Simulation node 1 starts the operation immediately after receiving the data and updates its logical time and calculation result. In this way, within the simulation period of simulation node 1, simulation node 2 can complete m simulation operations.

[0113] Non-startup issues: When the logical time of a computing node is greater than the system timestamp, or the system timestamp and the node timestamp are not in an integer ratio, the computing node does not start the operation.

[0114] Taking into account various issues in the multi-rate simulation system and the system scheduling strategy, the basic implementation steps of the node synchronization algorithm for the distributed simulation system are as follows:

[0115] (1) Control all computing nodes to start running synchronously with the clock of the scheduling host.

[0116] (2) Use the frame period of the simulation system as a logical time unit.

[0117] (3) The simulation node sets its own time controller. The node simulation step size T1 has an integer multiple relationship with the system frame period length Tm (T1 = mT m ), and takes 0 as the initial value of the logical time. It periodically sends node information with timestamps to the scheduling host using the UDP transmission method. After each synchronization operation is completed, the logical time of the computing node is incremented by m; the logical time of the scheduling host is incremented by 1 after the current clock operation is completed.

[0118] (4) After receiving the interactive data, the computing node performs state transitions such as initialization, running, and pausing according to different situations.

[0119] Using the above synchronization algorithm can not only ensure the sequential dependency relationship between tasks, but also ensure that the simulation system has a certain degree of parallelism, and at the same time does not increase the resource overhead of the system.

[0120] According to one aspect of the present invention, there is provided an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device runs, the processor executes the one or more computer programs stored in the memory, so that the electronic device executes the engine scheduling method for the digital model with hybrid granularity as described in any one of the above technical solutions.

[0121] According to one aspect of the present invention, there is provided a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the engine scheduling method for a digital model with hybrid granularity as described in any one of the above technical solutions is implemented.

[0122] The present invention discloses a method for quantifying and adjudicating the observation probability of a remote sensing satellite under the influence of multiple factors, including the following steps: Step S1, model sorting, constructing a model interaction relationship matrix DSM, grading and sorting the simulation models, and adjusting the matrix order to determine the model execution priority; Step S2, designing a task scheduling strategy, adopting an improved centralized scheduling strategy, globally controlling task allocation by a scheduling host, and nodes actively reporting loads and executing partial task scheduling; Step S3, based on the frame period of system simulation and the multi-rate distributed simulation node synchronization mechanism, executing a hybrid task scheduling algorithm based on priority, and the priority order is: periodic task > sporadic task > background task. In the present invention, different granularity models in the simulation space are sorted according to logical and timing relationships and then allocated to each computing node; an improved centralized scheduling strategy is adopted to determine the task allocation and scheduling method during the simulation process; an appropriate simulation step size is adopted to formulate the system simulation frame period; a multi-rate distributed simulation node synchronization mechanism is designed, and on this basis, a hybrid task scheduling algorithm based on priority is executed to realize the engine scheduling of the digital model under hybrid granularity.

[0123] In the present invention, a task scheduling strategy is introduced, and the sorted models are scheduled according to the improved centralized scheduling strategy. Each node actively calculates its own load and operating status regularly and sends them to the scheduling host, and hands over part of the scheduling tasks to the nodes for processing, so as to globally grasp the task allocation and scheduling.

[0124] The present invention designs a hybrid task scheduling algorithm based on priority to determine the system simulation frame period. Facing the problem of inconsistent advancement rates of multi-granularity models, a multi-rate distributed simulation node synchronization mechanism is designed, and on this basis, a hybrid task scheduling algorithm based on priority is executed to ensure the integrity, authenticity, and real-time performance of the entire simulation system function.

[0125] The present invention has good application prospects in terms of simulation engines, deduction, and control.

[0126] In addition, it should be noted that the present invention can be provided as a method, an apparatus, or a computer program product. Therefore, the embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media that contain computer-usable program code.

[0127] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal devices generate means for realizing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0128] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks. These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable terminal device provide steps for realizing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0129] It should also be noted that in this article, the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or terminal device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such a process, method, article, or terminal device. Without more limitations, the element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article, or terminal device including the said element.

[0130] Finally, it should be noted that the above description is the preferred embodiment of the present invention. It should be pointed out that although the preferred embodiments of the present invention have been described, for those skilled in the art of this technology, once the basic creative concept of the present invention is known, several improvements and refinements can be made without departing from the principle described in the present invention, and these improvements and refinements should also be regarded as the protection scope of the present invention. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

Claims

1. An engine scheduling method for a digital model oriented to mixed granularity, characterized in that, It includes the following steps: Step S1, model sorting, constructing a model interaction relationship matrix DSM, hierarchically sorting the simulation models, and adjusting the matrix order to determine the model execution priority; Step S2, designing a task scheduling strategy, adopting an improved centralized scheduling strategy, globally controlling task allocation by the scheduling host, and the nodes actively reporting the load and executing partial task scheduling; Step S3, based on the frame period of the system simulation and the multi-rate distributed simulation node synchronization mechanism, executing a priority-based hybrid task scheduling algorithm, and the priority order is: periodic task > sporadic task > background task.

2. The method according to claim 1, characterized in that, In step S1, hierarchically sorting the simulation models specifically includes: Step S11, constructing a model interaction relationship matrix, taking the model without input as the first level through hierarchical classification, and gradually deriving subsequent models; Step S12, adjusting the matrix order, and completing the model execution order planning through source point, sink point positioning and closed-loop processing.

3. The method according to claim 2, wherein In step S11, the hierarchical method is: Taking the model without input from other models as the first level, and taking the data receiving model of the first-level model as the second-level model, and so on; and the models that have been hierarchically classified will no longer participate in subsequent hierarchical classification.

4. The method according to claim 2, wherein In step S12, the closed-loop processing includes: Performing decoupling operation on the model with a directed loop, regarding the coupled model as a sub-module by using the normalization operation, and selecting a source point in the sub-module to continue sorting.

5. The method according to claim 1, wherein, In step S2, the improved centralized scheduling strategy includes: Step S21, the nodes actively report the load information to the scheduling host periodically or triggered by a threshold; Step S22, the scheduling host decentralizes part of the scheduling decision-making power to the nodes, and the nodes perform task allocation according to the local state.

6. The method according to claim 5, characterized in that, The threshold-triggered active reporting of load information includes: When the load variable of the node exceeds the preset threshold, the load change amount exceeds the difference threshold or the node receives a request from the scheduling host, the node actively reports the load information.

7. The method according to claim 1, characterized in that, In step S3, the multi-rate distributed simulation node synchronization mechanism includes: Step S31, taking the frame period of the system simulation as the logical time unit, and the node step size is an integer multiple of the frame period; Step S32, the node starts or pauses the operation according to the time stamp of the received data and the logical time relationship.

8. The method according to claim 7, characterized in that The frame period of the system simulation is the greatest common divisor of all node step sizes, and the scheduling host only performs data interaction with the nodes that meet the step size requirements.

Images