A computing resource allocation method for smart construction sites

Abstract

The present invention discloses a computing resource allocation method for a smart construction site, including: constructing a smart construction site status detection system; calculating the total time when all smart terminals perform status detection task processing based on historical detection data collected by a smart cloud platform, and the total energy cost when all smart terminals perform status detection task processing, and calculating the overall accuracy of the smart construction site status detection system when performing task unloading; constructing an evaluation fitness index, and converting the task unloading process into a Markov decision behavior based on the evaluation fitness index; solving the Markov decision behavior based on a computing resource allocation method of mean field game theory to obtain a computing resource allocation result. The present invention can reduce the processing time and energy cost of smart terminals when performing detection processing, and improve the processing accuracy and efficiency of detection tasks.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a computing resource allocation method for a smart construction site. Background Art

[0002] In the modern construction industry, smart construction sites can monitor and analyze various conditions on the construction site through technologies such as video surveillance and the Internet of Things. However, due to the complexity and diversity of construction site conditions, traditional resource management methods often cannot respond to dynamic changes on site in real time.

[0003] With the rise of edge computing, more and more computing tasks are being delegated to the edge. Using smart terminals to detect the status of construction sites can not only ensure the real-time detection, but also reduce the computing workload of the cloud platform. However, when using smart terminals to detect the status of construction sites, there is an unreasonable allocation of computing resources, which will increase the processing time and energy cost of the smart terminals during detection and processing, and affect the processing accuracy and efficiency of the detection tasks. Summary of the invention

[0004] The purpose of the present invention is to provide a computing resource allocation method for a smart construction site, so as to reduce the processing time and energy cost of intelligent terminals during detection processing and improve the processing accuracy and efficiency of detection tasks.

[0005] A computing resource allocation method for a smart construction site, comprising:

[0006] Step S1, constructing a smart construction site status detection system, wherein the smart construction site status detection system includes a smart cloud platform and a terminal device layer, wherein the terminal device layer is composed of a plurality of smart terminals carrying edge servers;

[0007] Step S2, based on the historical detection data collected by the smart cloud platform, calculate the total time when all smart terminals perform status detection task processing, as well as the total energy cost when all smart terminals perform status detection task processing, and calculate the overall accuracy of the smart construction site status detection system when performing task offloading;

[0008] Step S3, constructing an evaluation fitness index based on the total time when all intelligent terminals perform status detection task processing, the total energy cost when all intelligent terminals perform status detection task processing, and the overall accuracy of the smart construction site status detection system when performing task unloading, and converting the task unloading process into a Markov decision behavior based on the evaluation fitness index;

[0009] Step S4, solving the Markov decision behavior based on the computing resource allocation method of mean field game theory to obtain a computing resource allocation result.

[0010] According to the computing resource allocation method for smart construction sites provided by the present invention, the total time for all smart terminals to process status detection tasks, the total energy cost for all smart terminals to process status detection tasks, and the overall accuracy of the smart construction site status detection system when unloading tasks are first calculated, and then an evaluation fitness index is constructed. The system resource allocation can be preliminarily sorted according to the size of the evaluation fitness index, and then the task unloading process is converted into a Markov decision behavior, and the decision process is regarded as a mean field game. The computing resource allocation method based on the mean field game theory solves the Markov decision behavior, which can significantly reduce the time and energy cost when unloading detection tasks, improve the overall accuracy of unloading, and then realize the optimal strategy for computing resource allocation. The proposed computing resource allocation method regards each smart terminal as a mean field, and each mean field as a game participant, with the goal of maximizing rewards, which effectively reduces the processing time and energy cost of the smart terminal when performing detection processing, and improves the processing accuracy and efficiency of the detection task. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flow chart of the computing resource allocation method for a smart construction site according to the present invention. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.

[0013] See also Figure 1 The computing resource allocation method for a smart construction site provided in an embodiment of the present invention includes steps S1 to S4.

[0014] Step S1, constructing a smart construction site status detection system, wherein the smart construction site status detection system includes a smart cloud platform and a terminal device layer, and the terminal device layer is composed of multiple smart terminals carrying edge servers.

[0015] Among them, the smart cloud platform is a cloud platform responsible for collecting and presenting the construction site environment status information generated by the smart terminal. n It consists of a smart terminal with an edge server.

[0016] The status detection tasks generated by the intelligent terminal composition are divided into two execution modes: one is to execute locally, and the other is to offload to other intelligent terminals with sufficient computing resources for execution.

[0017] The system performs status detection in a time slot manner. For each time slot, the intelligent terminal receives a set of status detection tasks.

[0018] Step S2, based on the historical detection data collected by the smart cloud platform, calculate the total time taken by all smart terminals to process status detection tasks, as well as the total energy cost of all smart terminals to process status detection tasks, and calculate the overall accuracy of the smart construction site status detection system when offloading tasks.

[0019] The following formula is used to calculate the total time for all intelligent terminals to process the status detection task:

[0020] ;

[0021] ;

[0022] ;

[0023] in, It represents the total time when all intelligent terminals perform status detection tasks. Max represents the maximum value function. Indicates i The total time it takes for an intelligent terminal to process the status detection task. Indicates i The average processing time of a smart terminal for status detection tasks, Indicates i The processing time of task offloading for each intelligent terminal, Indicates i The average response time of smart terminals, P local Indicates the set of smart terminals where the status detection task is executed locally. P load Indicates that the status detection task is offloaded to the smart terminal collection outside the local smart terminal. represents the communication channel bandwidth when the task is offloaded, B represents the communication channel bandwidth of the intelligent terminal, Indicates i The average communication delay of each smart terminal.

[0024] The following formula is used to calculate the total energy cost of all intelligent terminals when performing status detection tasks:

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] represents the total energy cost of all intelligent terminals when processing state detection tasks, n Indicates the total number of smart terminals; Indicates i Whether the state detection task of the first intelligent terminal is offloaded to the j The binary value of a smart terminal, if i The state detection task of the smart terminal is offloaded to the j If there are 1 smart terminal, Take 1, if i The status detection task of the first intelligent terminal is not offloaded to the j If there are 1 smart terminal, Take 0; Indicates i The total energy cost of a smart terminal when processing the state detection task, Indicates i The average energy cost of a smart terminal performing state detection tasks locally, Indicates that the status detection task is offloaded to the i The energy cost of a smart terminal, Indicates i The energy cost of the intelligent terminal receiving state detection task is Indicates i The transmission power level of a smart terminal can be estimated based on the battery usage of the smart terminal.

[0030] The following formula is used to calculate the overall accuracy of the smart construction site status detection system during task offloading:

[0031] ;

[0032] in, It indicates the overall accuracy of the smart construction site status detection system when performing task offloading. Indicates i The average accuracy of a smart terminal when performing state detection tasks.

[0033] Step S3, construct an evaluation fitness index based on the total time when all smart terminals perform status detection task processing, the total energy cost when all smart terminals perform status detection task processing, and the overall accuracy of the smart construction site status detection system when performing task unloading, and convert the task unloading process into a Markov decision behavior based on the evaluation fitness index.

[0034] Among them, the constructed evaluation fitness index is:

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] ;

[0040] in, represents the fitness evaluation index, is the penalty coefficient based on processing time, is the penalty coefficient based on energy cost, min represents the minimum value function, T Indicates the upper limit of task processing time. E Indicates the upper limit of energy cost.

[0041] When offloading detection tasks, it should be ensured that the offloaded tasks processed in each time slot meet the set constraints, that is, and , if the constraints are exceeded, they will be subject to penalty coefficients.

[0042] After obtaining the evaluation fitness index, the system resource allocation can be preliminarily sorted according to the size of the evaluation fitness index.

[0043] In order to obtain the optimal strategy for resource allocation, the present invention transforms the task offloading process into a Markov decision behavior based on the fitness evaluation index, wherein the state space is designed s , Action Space b And the reward function R k , defined as follows:

[0044] ;

[0045] ;

[0046] ;

[0047] in, Indicates the size of the state detection task; , , Respectively represent the first, second, and third tasks when offloading tasks. n The amount of tasks that need to be offloaded to each smart terminal, Indicates the reward coefficient when the smart terminal to be uninstalled is correctly selected.

[0048] Step S4, solving the Markov decision behavior based on the computing resource allocation method of mean field game theory to obtain a computing resource allocation result.

[0049] Among them, for the mean field game, all intelligent terminals in the smart construction site status detection system are regarded as an intelligent agent, the resource allocation process of the system is regarded as a mean field game, and each intelligent agent is regarded as a mean field. When making decisions, the intelligent agent that needs to unload resources interacts with the mean field transformed by other intelligent agents, and the interaction behavior of multiple intelligent agents is converted into game behavior between intelligent agents.

[0050] Defining the overall value action function for:

[0051] ;

[0052] in, Indicates i When a smart terminal performs task offloading Q function, Indicates that when unloading a task i A binary value indicating whether a smart terminal is selected;

[0053] Calculate using the following formula :

[0054] ;

[0055] in, represents the weighted average action value, It represents the deviation range of the mean field game, which conforms to the uniform distribution;

[0056] Will Substitution , and Perform a secondary Tai expansion to obtain the weighted average action value Q function ;

[0057] Again Perform iterative updates, the expression is as follows:

[0058] ;

[0059] in, Indicates x Iteration , Indicates x -1 iteration , represents the learning rate;

[0060] based on , combined with the Boltzmann distribution, we get i The optimal strategy function for task offloading of intelligent terminals for:

[0061] ;

[0062] in, is an exponential factor, indicating the rationality of the smart terminals involved in the offloading. A larger value indicates that the uninstallation is more likely to be performed Larger intelligent terminal uninstall, when When it is smaller, it indicates that the uninstallation mode of the smart terminal is random.

[0063] According to the optimal strategy function It can be seen that when task offloading is performed, every time a time slot game ends, the mean field participating in the game can observe the decision of its related mean field and use Calculate the next action and Choose the optimal decision to make in the next time slot.

[0064] In this embodiment, in order to Asymptotically converges to Nash equilibrium, and the learning rate is adjusted using the following formula To update:

[0065] ;

[0066] in, Indicates x Iteration , Indicates x -1 iteration , represents the minimum value of the learning rate, Indicates the maximum value of the learning rate.

[0067] By continuously updating the learning rate , update the current state to the next state, and get , and then Substitute the latest middle.

[0068] In summary, according to the computing resource allocation method for smart construction sites provided by the present invention, the total time for all smart terminals to process status detection tasks, the total energy cost for all smart terminals to process status detection tasks, and the overall accuracy of the smart construction site status detection system when unloading tasks are first calculated, and then an evaluation fitness index is constructed. The system resource allocation can be preliminarily sorted according to the size of the evaluation fitness index, and then the task unloading process is converted into a Markov decision behavior, and the decision process is regarded as a mean field game. The computing resource allocation method based on the mean field game theory solves the Markov decision behavior, which can significantly reduce the time and energy cost when unloading detection tasks, improve the overall accuracy of unloading, and then realize the optimal strategy for computing resource allocation. The proposed computing resource allocation method regards each smart terminal as a mean field, and each mean field as a game participant, with the goal of maximizing the reward, which effectively reduces the processing time and energy cost of the smart terminal when performing detection processing, and improves the processing accuracy and efficiency of the detection task.

[0069] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0070] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A computing resource allocation method for a smart construction site, characterized in that: include: Step S1, constructing a smart construction site status detection system, wherein the smart construction site status detection system includes a smart cloud platform and a terminal device layer, wherein the terminal device layer is composed of a plurality of smart terminals carrying edge servers; Step S2, based on the historical detection data collected by the smart cloud platform, calculate the total time when all smart terminals perform status detection task processing, as well as the total energy cost when all smart terminals perform status detection task processing, and calculate the overall accuracy of the smart construction site status detection system when performing task offloading; Step S3, constructing an evaluation fitness index based on the total time when all intelligent terminals perform status detection task processing, the total energy cost when all intelligent terminals perform status detection task processing, and the overall accuracy of the smart construction site status detection system when performing task unloading, and converting the task unloading process into a Markov decision behavior based on the evaluation fitness index; Step S4, solving the Markov decision behavior based on the computing resource allocation method of mean field game theory to obtain a computing resource allocation result; In step S2, the total time for all intelligent terminals to process the status detection task is calculated using the following formula: ； ； ； in, It represents the total time when all intelligent terminals perform status detection tasks. Max represents the maximum value function. Indicates i The total time it takes for an intelligent terminal to process the status detection task. Indicates i The average processing time of a smart terminal for status detection tasks, Indicates i The processing time of task offloading for each intelligent terminal, Indicates i The average response time of smart terminals, P local Indicates the set of smart terminals where the status detection task is executed locally. P load Indicates that the status detection task is offloaded to the smart terminal collection outside the local smart terminal. represents the communication channel bandwidth when the task is offloaded, B represents the communication channel bandwidth of the intelligent terminal, Indicates i Average communication delay of smart terminals; In step S2, the total energy cost of all intelligent terminals when performing state detection task processing is calculated using the following formula: ； ； ； ； represents the total energy cost of all intelligent terminals when processing state detection tasks, n Indicates the total number of smart terminals; Indicates i Whether the state detection task of the first intelligent terminal is offloaded to the j The binary value of a smart terminal, if i The state detection task of the smart terminal is offloaded to the j If there are 1 smart terminal, Take 1, if i The status detection task of the first intelligent terminal is not offloaded to the j If there are 1 smart terminal, Take 0; Indicates i The total energy cost of a smart terminal when processing the state detection task, Indicates i The average energy cost of a smart terminal performing state detection tasks locally, Indicates that the status detection task is offloaded to the i The energy cost of a smart terminal, Indicates i The energy cost of the intelligent terminal receiving state detection task is Indicates i The transmission power level of each intelligent terminal; In step S2, the overall accuracy of the smart construction site status detection system during task unloading is calculated using the following formula: ； in, It indicates the overall accuracy of the smart construction site status detection system when performing task offloading. Indicates i The average accuracy of a smart terminal when performing state detection tasks.

2. The computing resource allocation method for a smart construction site according to claim 1, characterized in that: In step S3, the constructed evaluation fitness index is: ； ； ； ； ； in, represents the fitness evaluation index, is the penalty coefficient based on processing time, is the penalty coefficient based on energy cost, min represents the minimum value function, T Indicates the upper limit of task processing time. E Indicates the upper limit of energy cost.

3. The computing resource allocation method for a smart construction site according to claim 2, characterized in that: In step S3, in the process of converting the task offloading process into a Markov decision behavior based on the evaluation fitness index, the state space is designed s , Action Space b And the reward function R k , defined as follows: ； ； ； in, Indicates the size of the state detection task. , , Respectively represent the first, second, and third tasks when offloading tasks. n The amount of tasks that need to be offloaded to each smart terminal, Indicates the reward coefficient when the smart terminal to be uninstalled is correctly selected.

4. The computing resource allocation method for a smart construction site according to claim 3, characterized in that: In step S4, the computational resource allocation method based on the mean field game theory is used to solve the Markov decision behavior, and each intelligent terminal is taken as a mean field to define the overall value action function for: ； in, Indicates i When a smart terminal performs task offloading Q function, Indicates that when unloading a task i A binary value indicating whether a smart terminal is selected; Calculate using the following formula : ； in, represents the weighted average action value, Indicates the deviation range of the mean field game; Will Substitution , and Perform a secondary Tai expansion to obtain the weighted average action value Q function ; Again Perform iterative updates, the expression is as follows: ； in, Indicates x Iteration , Indicates x -1 iteration , represents the learning rate; based on , combined with the Boltzmann distribution, we get i The optimal strategy function for task offloading of intelligent terminals for: ； in, is the exponential factor.

5. The computing resource allocation method for a smart construction site according to claim 4, characterized in that: In step S4, the learning rate is calculated using the following formula: To update: ； in, Indicates x Iteration , Indicates x -1 iteration , represents the minimum value of the learning rate, Indicates the maximum value of the learning rate.

Images