Rail transit service system for adaptive resource adjustment and distribution algorithm thereof

Abstract

The invention discloses an adaptive resource adjustment rail transit service system and a distribution algorithm thereof, the rail transit service system comprises a distributed redundant function framework, a task space and a twin address space, the distributed redundant function framework is composed of a main service and a plurality of function sub-frameworks, the main service is used for converting a configuration file of the task space into tasks and allocating the tasks, the tasks comprise an acquisition task, a calculation task and a driving task, and the twin address space is used for transmitting the allocated tasks to the function subframes. The plurality of function subframes are used for executing the task. The method has the advantages that self-adaptive resource adjustment of the rail transit service can be achieved, and the resource utilization rate and the system operation efficiency are improved to the maximum extent.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit industrial control, and in particular to a rail transit service system with adaptive resource regulation and its allocation algorithm. Background Art

[0002] In the signal control system of rail transit, the system is required to be simple, compact, efficient, with low latency and high stability. Traditional methods usually implement data acquisition, calculation and data service functions through high-performance servers. In the case of a limited number of supervised devices, a high-performance server can achieve decoupling of data service functions through multi-network card, multi-CPU configuration, and combined with an in-memory key-value database. However, when it comes to the transfer of multiple lines, the comprehensive scheduling of transportation hubs, and the data services at the line level and network level, the computing power and network communication efficiency of high-performance servers will become the bottleneck of the system.

[0003] At the same time, considering primary and backup redundancy, adjacent station resource support, and the cloud deployment requirements at the line and network levels, it is obviously impractical to equip dedicated high-performance servers for industrial control applications at each level. In the context of the increasingly complex networked operation of the rail transit system, the issues of cost performance and construction cost are becoming increasingly prominent, becoming the main bottleneck restricting the development of the system.

[0004] Therefore, it is necessary to perform cloud decoupling to achieve the decoupling of software and hardware, data / applications, and system / functions. Only in this way can the application requirements be efficiently met using the cloud-based standard resource pool, and a redundant and stable deployment architecture can be established. Summary of the Invention

[0005] The object of the present invention is to provide a rail transit service system with adaptive resource regulation and its allocation algorithm according to the deficiencies of the above-mentioned prior art. The rail transit service system includes a distributed redundant function framework, a task space, and a twin address space. The distributed redundant function framework consists of a main service and several functional sub-frameworks. The main service is used to convert the configuration file of the task space into tasks and allocate the tasks. The tasks include acquisition tasks, calculation tasks, and driving tasks. The twin address space is used to transmit the allocated tasks to several functional sub-frameworks. The several functional sub-frameworks are used to execute the tasks. Combining with the allocation algorithm, it can achieve adaptive resource regulation of rail transit services, so as to maximize resource utilization and system operation efficiency.

[0006] The object of the present invention is achieved by the following technical solutions:

[0007] An urban rail transit service system with adaptive resource regulation, the urban rail transit service system includes a distributed redundant function framework, a task space, and a twin address space. The distributed redundant function framework consists of a main service and several functional sub-frameworks. The main service is used to convert the configuration file of the task space into tasks and allocate the tasks. The tasks include acquisition tasks, calculation tasks, and driving tasks. The twin address space is used to transmit the allocated tasks to several functional sub-frameworks, and several functional sub-frameworks are used to execute the tasks.

[0008] The task space consists of a task channel and a task nature.

[0009] The twin address space consists of a channel space and a device space.

[0010] Each functional sub-framework can execute one or two or three of the acquisition tasks, the calculation tasks, and the driving tasks.

[0011] The functional sub-frameworks are divided into active frameworks and standby frameworks.

[0012] An allocation algorithm for an urban rail transit service system with adaptive resource regulation, the allocation algorithm adopts an adaptive scheduling algorithm. The adaptive scheduling algorithm means that when the active framework does not fail, the active framework executes all the tasks; when the active framework fails, all the tasks are fed back to the main service, and the main service reallocates some or all of the tasks, and the active framework alone or the active framework and the standby framework together execute all the tasks; wherein, the adaptive scheduling algorithm utilizes a real-time handover mechanism. The real-time handover mechanism means that each functional sub-framework can theoretically perform all the tasks, and taking over other tasks only needs to create a functional module instance through task allocation information.

[0013] The adaptive scheduling algorithm includes the following steps:

[0014] S1: Obtain the status information of the urban rail transit service system, including CPU utilization rate U CPU , memory usage U Mem , network bandwidth U BW and I / O utilization rate U IO ;

[0015] S2: Calculate the priority of each task according to the status information of the urban rail transit service system and the task attributes. The priority P i of the task can be calculated according to the following formula:

[0016] P i =α·W i +β·Ui + γ·R i + δ·(U CPU + U Mem + U BW + U IO ) + λ·D i ;

[0017] Wherein, W i is the importance weight of task i; U i is the proportion of the historical execution time of task i; R i is the computing resource required for task i; α, β, γ, and λ are weight coefficients used to balance the influence of different factors; δ is the system state weight; D i is the task dependency, which is calculated through the task correlation matrix A = [a ij :

[0018]

[0019] S3: Based on the priorities of the tasks and the system state, select an adaptive task scheduling strategy. Assume there are N tasks and M resources, where each task i requires a certain amount of resource x ij (i = 1, 2,..., N, j = 1, 2,..., M). The resource allocation goal is to maximize the task completion volume and minimize the system latency L total :

[0020]

[0021] Wherein, the system latency L total is represented by the following formula:

[0022]

[0023] The resource allocation needs to satisfy the following constraints:

[0024]

[0025] And the task allocation constraints:

[0026]

[0027] S4: According to the adaptive task scheduling strategy, allocate the resources in the system to each task;

[0028] S5: Introduce a feedback mechanism to dynamically adjust the priorities of tasks by comparing the difference between the actual execution latency of the tasks and the expected latency :

[0029]

[0030] In the formula, ε is the feedback adjustment coefficient;

[0031] S6: Continuously repeat the optimization process, and with the help of the genetic algorithm and the feedback mechanism, dynamically adapt to the system state and task requirements, and continuously optimize the resource allocation.

[0032] The advantages of the present invention are: It can achieve adaptive resource adjustment for rail transit services, so as to maximize the resource utilization rate and system operation efficiency. Brief Description of the Drawings

[0033] Figure 1 It is a schematic structural diagram of the rail transit service system of the present invention;

[0034] Figure 2 It is a service structure diagram of the rail transit service system of the present invention;

[0035] Figure 3 It is a flowchart of the allocation algorithm of the present invention. Specific Embodiments

[0036] The following further details the features of the present invention and other related features through embodiments in conjunction with the drawings, so as to facilitate the understanding of those skilled in the same industry:

[0037] Embodiment: As Figure 1 shown, this embodiment relates to a rail transit service system with adaptive resource adjustment. The rail transit service system mainly includes a distributed redundant function framework, a task space, and a twin address space. The task space consists of a task channel and a task nature, and the twin address space consists of a channel space and a device space. The distributed redundant function framework consists of a main service (real-time and historical data services) and several functional sub-frameworks. The main service is used to convert the configuration file of the task space into tasks and allocate the tasks. The tasks include acquisition tasks, calculation tasks, and drive tasks. The main service provides functions such as reading and writing the twin address space, navigating the twin address space, data subscription, and external data services. The twin address space is used to transmit the allocated tasks to several functional sub-frameworks. The twin address space is based on the principle of decoupling, and separates the three functions of acquisition, calculation, and drive of the main service in the traditional architecture from the main service. Several functional sub-frameworks are used to execute tasks. Each functional sub-framework can execute one or two or three of the acquisition tasks, calculation tasks, and drive tasks. The functional sub-frameworks are divided into active frameworks and standby frameworks, and the functional sub-frameworks provide functions such as data acquisition, engineering data calculation, and instruction execution.

[0038] As Figure 1As shown in the figure, this embodiment also has an allocation algorithm for an adaptive resource adjustment rail transit service system. This allocation algorithm adopts an adaptive scheduling algorithm, which means that when the current framework does not fail, the current framework executes all tasks; when the current framework fails, all tasks are fed back to the main service, and the main service reallocates some or all of the tasks. The current framework executes all tasks alone or jointly with the standby framework. In other words, the sum of all tasks is always 100%, and there will be no situation where multiple functional sub-frameworks execute the same task at the same time or some tasks have no functional sub-framework to undertake (each functional sub-framework knows the tasks being executed by other functional sub-frameworks). Among them, tasks are allocated through collection task allocation, computing power task allocation, driving task allocation, and internal instructions, and tasks are fed back through collection task feedback, computing power task feedback, driving task feedback, and instruction feedback. The adaptive scheduling algorithm utilizes a real-time handover mechanism, which means that each functional sub-framework can theoretically perform all tasks, and taking over other tasks only needs to create a functional module instance through task allocation information. Therefore, task drift is theoretically a real-time handover without delay (where module instance creation and communication handshake delays are required, theoretically at the millisecond level).

[0039] As Figure 3 shown, the adaptive scheduling algorithm includes the following steps:

[0040] S1: Obtain the status information of the rail transit service system, including CPU utilization U CPU , memory usage U Mem , network bandwidth U BW and I / O utilization U IO .

[0041] S2: Calculate the priority of each task according to the status information of the rail transit service system and the task attributes (when having a higher priority, resource allocation needs to be considered first). The priority P i of the task can be calculated according to the following formula:

[0042] P i =α·W i +β·U i +γ·R i +δ·(U CPU +U Mem +U BW +U IO )+λ·D i ;

[0043] In the formula, W i is the importance weight of task i; U i is the proportion of the historical execution time of task i; R iis the computing resources required for task i; α, β, γ, and λ are weight coefficients used to balance the influence of different factors; δ is the system state weight; D i is the task dependency, which is calculated through the task correlation matrix A = [a ij :

[0044]

[0045] S3: Based on the task priority and system state, select an adaptive task scheduling strategy. Suppose there are N tasks and M resources, where each task i requires a certain amount of resources x ij (i = 1, 2,..., N, j = 1, 2,..., M). The resource allocation goal is to maximize the task completion quantity and minimize the system delay L total :

[0046]

[0047] In the formula, the system delay L total is represented by the following formula:

[0048]

[0049] The resource allocation needs to meet the following constraints:

[0050]

[0051] And the task allocation constraint:

[0052]

[0053] S4: According to the adaptive task scheduling strategy, allocate the resources in the system to each task.

[0054] S5: Introduce a feedback mechanism. By comparing the actual execution delay of the task with the expected delay to dynamically adjust the task priority:

[0055]

[0056] In the formula, ε is the feedback adjustment coefficient.

[0057] S6: Continuously repeat the optimization process. With the help of the genetic algorithm and the feedback mechanism, dynamically adapt to the system state and task requirements, and continuously optimize the resource allocation.

[0058] Such as Figure 2As shown in the figure, from the perspective of the data platform, the ontology, client, and data source are classified into three levels: digital twin, service mapping, and drive mapping, to present the upstream and downstream data relationships. Specifically, all clients (external data services, HMI clients) are classified into service mapping; the OPC UA server (OPC UA SERVER), historical structured database, and rule platform in the ontology are classified into digital twin; and all lower-level data acquisition and instruction issuance channels are classified into drive mapping.

[0059] Service mapping is applied to HMI clients and external data services. The underlying layer is based on the monitoring data subscription mechanism of the OPC UA client (OPC UA CLIENT), which converts the original message into engineering data based on the digital twin model. On this basis, the external data interface is formed using MQTT (Message Queuing Telemetry Transport, a lightweight publish / subscribe message transport protocol) and WEB SOCKET (a full-duplex communication protocol introduced by HTML5) technologies.

[0060] The digital twin includes three parts: the real-time data service core, historical data service, and operation instruction specification core, which are used to provide real-time data engineering transformation calculation, historical data query, and forward and reverse business logic rule verification. The real-time data service core is based on OPC UA technology, the historical data service combines OPC UA and structured database technologies, and the operation instruction specification core is based on JAVA MARVEN (a cross-platform project management tool for project creation, dependency management, and project information management in the Java platform) and SIDDHI-CEP engine streaming computing technology.

[0061] Drive mapping mainly targets protocols such as MODBUS / IEC104, TCP / UDP, and IT information interfaces such as RESTFUL and SAP, and compiles two major functions: batch read and write polling data acquisition, and instruction reverse compilation, parsing, and issuance and execution.

[0062] The beneficial technical effect of this embodiment is that it can achieve adaptive resource adjustment for rail transit services, so as to maximize resource utilization and system operation efficiency.

[0063] Although the above embodiments have described in detail the concept and implementation of the object of the present invention with reference to the drawings, those of ordinary skill in the art can recognize that various improvements and transformations can still be made to the present invention without departing from the scope defined by the claims, so they will not be elaborated here one by one.

Claims

1. An adaptive resource regulation rail transit business system, characterized in that: The rail transit business system includes a distributed redundant functional framework, a task space and a twin address space. The distributed redundant functional framework consists of a main service and several functional sub-frameworks. The main service is used to convert the configuration file of the task space into tasks and assign the tasks. The tasks include acquisition tasks, computing tasks and driving tasks. The twin address space is used to transfer the assigned tasks to several functional sub-frameworks, and several functional sub-frameworks are used to execute the tasks.

2. The rail transit service system for adaptive resource regulation according to claim 1, wherein: The task space consists of task channels and task properties.

3. An adaptive resource adjustment rail transit service system according to claim 1, characterized in that: The twin address space consists of a channel space and a device space.

4. The rail transit service system for adaptive resource regulation according to claim 1, characterized in that: Each of the functional subframes can execute one, two, or three of the acquisition task, the calculation task, and the driving task.

5. An adaptive resource adjustment rail transit service system as claimed in claim 1, wherein: The functional subframe is divided into an active frame and a standby frame.

6. The allocation algorithm of an adaptive resource regulation rail transit service system according to any one of claims 1 to 5, characterized in that: The allocation algorithm adopts an adaptive scheduling algorithm, which means that when the active framework is not faulty, the active framework executes all the tasks; when the active framework fails, all the tasks are fed back to the main service, and the main service reallocates part or all of the tasks, and the active framework executes all the tasks alone or the active framework and the backup framework together; wherein the adaptive scheduling algorithm utilizes a real-time handover mechanism, which means that each of the functional sub-frameworks can theoretically perform all the tasks, and taking over other tasks only requires creating a functional module instance through task allocation information.

7. The allocation algorithm of an adaptive resource regulation rail transit service system according to claim 6, characterized in that: The adaptive scheduling algorithm comprises the following steps: S1: Obtain the status information of the rail transit business system, including CPU utilization rate U CPU , memory usage U Mem , network bandwidth U BW and I / O utilization rate U IO ; S2: Calculate the priority of each task according to the status information and task attributes of the rail transit service system, where the priority P of the task i can be calculated according to the following formula: P i = α·W i + β·U i + γ·R i + δ·(U CPU + U Mem + U BW + U IO ) + λ·D i ; where, W i is the importance weight of task i; U i is the proportion of the historical execution time of task i; R i is the computing resource required for task i; α, β, γ, and λ are weight coefficients used to balance the influence of different factors; δ is the system state weight; D i is the task dependency, which is calculated through the task correlation matrix A = [a ij ; S3: Based on the priority of the task and the system state, select an adaptive task scheduling strategy. Suppose there are N tasks and M resources, where each task i requires a certain amount of resource x ij (i = 1, 2,..., N, j = 1, 2,..., M), and the resource allocation goal is to maximize the task completion volume and minimize the system latency L total : where the system delay L total is represented by the following formula: Resource allocation must satisfy the following constraints: And the task allocation constraints: S4: Allocate system resources to each task according to the adaptive task scheduling strategy; S5: Introduce a feedback mechanism to dynamically adjust the task priority by comparing the difference between the actual execution latency of the task and the expected latency : Where, ε is the feedback adjustment coefficient; S6: Continuously repeat the optimization process, use genetic algorithms and feedback mechanisms to dynamically adapt to system status and task requirements, and continuously optimize resource allocation.

Images