Rail transit power system edge device computing power optimization method

Abstract

The present disclosure discloses a rail transit power system edge device computing power optimization method, comprising the following steps: obtaining a plurality of rail transit power system data calculation tasks; obtaining a feasible calculation node set corresponding to each data calculation task based on a constraint condition; sorting the plurality of data calculation tasks according to the size of all feasible calculation node sets, and assigning priority to the feasible calculation node set with a small number of feasibility strategies to complete the data calculation task unloading; optimizing each feasible calculation node set to obtain an optimal calculation node; and obtaining the minimum calculation resource required by the optimal calculation node based on a monotonic optimization method to complete the computing power optimization of the rail transit power system edge device.

Description

TECHNICAL FIELD

[0001] The present disclosure belongs to the field of power equipment state diagnosis, and particularly relates to a track traffic power system edge device computing power optimization method. BACKGROUND

[0002] With the rapid development of track traffic in China, the complexity of its power consumption scenarios also makes the complexity of the power system higher and higher. In order to realize real-time monitoring of the operation state of the power grid system, a large number of intelligent terminals and sensing devices are deployed at the network edge. Terminal devices collect data in various aspects of the power grid, including environmental data, electrical information, operating information, etc. The massive data generated by the terminal devices have low information density, and processing all the data by the cloud computing center will cause high load of the cloud computing center and increase communication time consumption. At the same time, some data processing has strict delay, and cloud computing cannot meet the real-time requirements of data analysis, processing and response; therefore, the edge computing mode emerges as the times require.

[0003] Edge computing is a new network architecture and platform that integrates network, computing, storage and application at the network edge close to the data source. Edge computing reduces high latency caused by network congestion and routing by decentralizing storage and computing capabilities to the network edge close to the data source, so as to achieve fast response to terminal requests and real-time processing of massive data. At the same time, through timely filtering, computing and processing of data with low value density, communication overhead is reduced and the pressure on the cloud computing center is reduced. SUMMARY

[0004] In view of the deficiencies in the prior art, the purpose of the present disclosure is to provide a track traffic power system edge device computing power optimization method, which can improve the efficiency of raw data processing while reducing the waste of computing resources.

[0005] To achieve the above purpose, the present disclosure provides the following technical solutions:

[0006] A track traffic power system edge device computing power optimization method, comprising the following steps:

[0007] S100: obtaining a plurality of track traffic power system data computing tasks;

[0008] S200: obtaining a set of feasible computing nodes corresponding to each data computing task based on constraint conditions;

[0009] S300: sorting the plurality of data computing tasks according to the size of all sets of feasible computing nodes, and assigning priority to the set of feasible computing nodes with a small number of feasibility strategies, so as to complete the offloading of data computing tasks;

[0010] S400: optimizing each set of feasible computing nodes to obtain an optimal computing node;

[0011] S500: Obtain the minimum computing resource required by the optimal computing node based on the single optimization method to complete the computing power optimization of the rail transit power system edge device.

[0012] Preferably, in step S100, the data computing task includes: maximum time delay requirement T i，max , task computing amount r ij , and task execution time T ij .

[0013] Preferably, in step S200, the constraint condition is:

[0014]

[0015]

[0016]

[0017] (i∈N and j∈M)

[0018] (i∈N and j∈M)

[0019]

[0020] wherein, rij is the computing resource allocated by device j for task i, r j，max is the maximum allocatable resource of the jth device; T ij is the execution time of the ith task allocated to the jth device, T i，max is the maximum time delay required by task i, o ij is a feasible mode matrix, n is the number of computing tasks, N is the set of terminal devices in the system, N={1, 2, …n}, corresponding to n terminal devices in the system, numbered 1, 2, 3…n, M is the execution mode, M={0, 1, …n}, corresponding to n+1 execution modes in the system, numbered 0 representing the host computer executing the computing task, numbered 1 representing the 1st terminal executing the computing task, numbered 2 representing the 2nd terminal executing the computing task, and so on.

[0021] Preferably, in step S400, the optimization of the feasible computing node set includes the following steps:

[0022] S401: Initialize the feasible computing node set;

[0023] S402: Set the number of iterations, and iteratively optimize the initialized feasible computing node set until the number of iterations is satisfied.

[0024] Preferably, in step S402, the iteration on the initialized feasible computing node set comprises the following steps:

[0025] S4011: calculating the selected probability of each feasible strategy in the feasible computing node set;

[0026] S4012: selecting a strategy according to the probability through a roulette algorithm;

[0027] S4013: updating the pheromone τ;

[0028] S4014: repeating steps S4011 to S4014 until a set iteration number is reached.

[0029] Preferably, in step S4011, the selected probability of each feasible strategy is calculated by the following formula:

[0030]

[0031] wherein τ is the pheromone, the initial value is 500, and μ is the heuristic information, wherein is the total energy consumption of the system for arranging task i on device j, and β is the relative importance of the heuristic information and the pheromone, and is taken as 0.95.

[0032] Preferably, in step S4013, the pheromone τ is updated according to the following formula:

[0033]

[0034] wherein is the pheromone evaporation factor, taken as 0.3, and τ0 is the initial value of the pheromone.

[0035] Compared with the prior art, the present disclosure has the beneficial effects that the present disclosure is mainly used for raw data analysis and extraction of secondary equipment monitoring of rail transit power systems, and reduces the waste of computing resources while improving the efficiency of raw data processing. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a flowchart of an edge device computing power optimization method of a rail transit power system according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] Specific embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. Although specific embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0038] It should be noted that some terms are used in the specification and claims to refer to certain components. Those skilled in the art will understand that the same component can be referred to by different terms. The description and claims should not be construed as being limited to the terminology used. The description and claims should be construed as being limited only by the functional differences between the components. As used throughout the specification and claims, "comprising" or "including" is to be construed as meaning "including but not limited to." The description that follows is intended to describe preferred embodiments of the present disclosure and is not intended to limit the scope of the present disclosure. The scope of the present disclosure is defined by the appended claims.

[0039] In order to facilitate the understanding of the embodiments of the present disclosure, the following will be further explained and described with specific examples in conjunction with the accompanying drawings, and each drawing does not constitute a limitation on the embodiments of the present disclosure.

[0040] In one embodiment, as shown in Figure 1 The present disclosure provides a rail transit power system edge device computing power optimization method, comprising the following steps:

[0041] S100: Obtain a plurality of rail transit power system data computing tasks;

[0042] In this step, it is assumed that the set of terminal devices in the system is N={1, 2,..., n}, and the computing tasks can be assigned to local devices, other terminal devices, and edge servers for execution, a total of n+1 execution modes M={0, 1, 2,..., n}, wherein the edge server can receive tasks from multiple terminals, and the terminal device can only receive one task. A feasible mode matrix o is established, and o ij =1 indicates that the ith task is assigned to the jth device for execution.

[0043] S200: Obtain the feasible computing node set corresponding to each data computing task based on the constraint condition;

[0044] S300: Sort the data computing tasks according to the size of all feasible computing node sets, and assign priority to the feasible computing node set with a small number of feasibility strategies to complete the offloading of data computing tasks;

[0045] S400: Optimize each feasible computing node set to obtain the optimal computing node;

[0046] S500: Obtain the minimum computing resource required by the optimal computing node to complete the computing power optimization of the rail transit power system edge device.

[0047] In another embodiment, in step S100, the data computing task includes: maximum time delay requirement T i，max , task computing amount r ij , and task execution time T ij .

[0048] In another embodiment, in step S200, the constraint condition is:

[0049]

[0050]

[0051]

[0052] (i∈N and j∈M)

[0053] (i∈N and j∈M)

[0054]

[0055] wherein r ij is the computing resource allocated by the device j for the task i, r j，max is the maximum allocatable resource of the jth device; T ij is the execution time of the ith task allocated to the jth device, T i，max is the maximum time delay required for the task i to compute, o ij is the feasible mode matrix, n is the number of computing tasks, N is the set of terminal devices in the system, N={1, 2, …n}, corresponding to n terminal devices in the system, numbered 1, 2, 3, …n, and M is the execution mode, M={0, 1, …n}, corresponding to n+1 execution modes in the system, numbered 0 representing the host computer executing the computing task, numbered 1 representing the 1st terminal device executing the computing task, numbered 2 representing the 2nd terminal device executing the computing task, and so on.

[0056] It should be noted that the constraint condition ensures that each task will be executed and each terminal will only receive one task, and each computing task must be completed within the specified time. According to the constraint condition, the feasible mode matrix o ij is updated, and the updated feasible mode matrix o ij is the set of feasible computing nodes.

[0057] In another embodiment, in step S400, the optimization of the set of feasible computing nodes includes the following steps:

[0058] S401: initialize the set of feasible computing nodes;

[0059] S402: Set the number of iterations, and iterate the initialized feasible computing node set until the number of iterations is met.

[0060] In this embodiment, the feasible computing node set is first initialized, that is, the feasible mode matrix o ij is set to [100, 500], and the priority is set to 200. After setting the number of iterations, the iteration optimization is started, and the iteration optimization process is as follows:

[0061] (1) Calculate the selected probability of each feasible strategy in the feasible computing node set, and the specific calculation formula is as follows:

[0062]

[0063] Wherein, τ is pheromone, the initial value is 500, μ is heuristic information, Wherein is the total system energy consumption of task i arranged on device j, β is the relative importance of heuristic information and pheromone, and is taken as 0.95.

[0064] (2) Select the strategy through the roulette algorithm according to the probability;

[0065] (3) Update the pheromone τ according to the following formula:

[0066]

[0067] Wherein, is the pheromone evaporation factor, taken as 0.3, and τ0 is the initial value of pheromone.

[0068] (4) Repeat steps (1), (2) and (3) until the set number of iterations is met.

[0069] In another embodiment, in step S500, the minimum computing resource required by the computing node is obtained based on the monotonic optimization method.

[0070] In this embodiment, the computing resource and the energy consumption have a strict monotonic increasing relationship, in order to ensure the minimum energy consumption, the resource allocated to each task should be as little as possible. The minimum computing resource required by the task is affected by the mode selection. In the local computing mode, the minimum computing resource required by the task is affected by the task computing amount and the maximum delay constraint, and the minimum required resource is r ij / T i，max ; for the mode sent to other computing nodes, the minimum computing resource required is r ij / T ij When a task is allocated the minimum computing resource, it also meets the requirements of the unloading decision, and finally completes the optimal allocation of the entire edge computing power.

[0071] The technical solutions provided by the present disclosure are described in detail above in combination with specific embodiments, and meanwhile, the description of the above embodiments is only used to help understand the core idea of the present disclosure. For those skilled in the art, according to the idea of the present disclosure, the specific implementation manners and application scopes will be changed. Therefore, the content of the present specification should not be understood as a limitation of the present disclosure.

Claims

1. A method for optimizing the computing power of edge devices in a rail transit power system, comprising the following steps: S100: Acquire several data calculation tasks related to the rail transit power system; S200: Obtain the set of feasible computing nodes corresponding to each data computing task based on constraints; S3 00: Sort the data computation tasks according to the size of all feasible computation node sets, and assign priority values ​​to feasible computation node sets with fewer feasible strategies in order to unload the data computation tasks. S400: Optimize each feasible set of computing nodes to obtain the optimal computing nodes; S500: Based on the monotonic optimization method, the minimum computing resources required for the optimal computing node are obtained to complete the computing power optimization of the edge equipment of the rail transit power system; The method described above is used for the extraction of raw data from the monitoring of secondary equipment in rail transit power systems, which improves the efficiency of raw data processing while reducing the waste of computing resources.

2. The method according to claim 1, wherein, In step S100, the data computation task includes: maximum latency requirement T i,max Task computation amount r ij Task execution time T ij .

3. The method according to claim 1, wherein, In step S200, the constraint condition is: ， ， ， ， ， ， Where, r ij The computing resources allocated by device j to task i, r j,max The maximum allocatable resources for the j-th device; T ij T is the execution time of the i-th task assigned to the j-th device. i,max Calculate the maximum required latency for task i. Let M be the feasible mode matrix, n be the number of computation tasks, N be the set of terminal devices in the system, N={1, 2, ..., n}, corresponding to n terminals in the system, numbered 1, 2, 3, ..., n, and M be the execution mode, M={0, 1, ..., n}, corresponding to n+1 execution modes in the system. Number 0 represents the host computer executing the computation task, number 1 represents terminal 1 executing the computation task, number 2 represents terminal 2 executing the computation task, and so on.

4. The method according to claim 1, wherein, In step S400, optimizing the feasible computation node set includes the following steps: S401: Initialize the feasible computation node set; S402: Set the number of iterations, and iteratively optimize the initialized set of feasible computation nodes until the number of iterations is met.

5. The method according to claim 4, wherein, In step S402, iterating over the initialized set of feasible computational nodes includes the following steps: S4011: Calculate the probability of each feasible strategy being selected in the feasible computing node set; S4012: Select a strategy using a roulette wheel algorithm based on the stated probability; S4013: Update pheromones ; S4014: Repeat steps S4011 to S4014 until the set number of iterations is reached.

6. The method according to claim 5, wherein, In step S4011, the probability of each feasible strategy being selected is calculated using the following formula: ， in, It's a pheromone, with an initial value of 500. It is heuristic information. ,in The total system energy consumption for assigning task i to device j. This represents the relative importance of heuristic information and pheromones, and is set to 0.

95.

7. The method according to claim 5, wherein, In step S4013, the pheromone is updated according to the following formula. : ， in, The pheromone evaporation factor is set to 0.

3. This is the initial value for pheromones.

Images