Industrial production edge intelligent resource allocation and task scheduling method and system

Abstract

The invention relates to an industrial production edge intelligent resource allocation and task scheduling method and system, and the method comprises the following steps: calculating the task criticality of any task in an industrial production link based on task time and task resources; meanwhile, a scheduling objective function and constraint conditions of the task are constructed, and elastic resource influence factors for completing the task are calculated; based on a deep learning algorithm, constructing a state space, setting a resource utilization rate as a state in the state space, and calculating the resource utilization rate according to task criticality and elastic resource change; designing a reward function, wherein the reward function corresponds to the resource allocation strategy; initializing a state space, updating an Actor network and a Critic network according to a reward function, and generating a trained strategy agent; and deploying the trained strategy agent in edge equipment in the industrial production environment, and performing resource allocation and task scheduling through the strategy agent.

Description

Technical Field

[0001] The present invention relates to an industrial production edge intelligent resource allocation and task scheduling method and system, belonging to the technical field of industrial production intelligence. Background Art

[0002] With the rapid development of emerging technologies such as artificial intelligence, big data, and the Internet of Things, the demand for computing power has increased dramatically. Especially in industrial-level application scenarios, the demand for highly adaptable and high-performance computing modules is particularly prominent. In order to meet these needs, domestic chip manufacturers have made significant progress in the field of high-performance computing in recent years. They have not only achieved a significant increase in computing power, but also provided optimized support in graphics rendering, video processing, cloud computing, etc.

[0003] However, having powerful hardware is only the first step. How to effectively manage and schedule these resources to achieve optimal performance is the key to building a high-performance computing system. Task scheduling is an important bridge between the application layer and the hardware layer. While ensuring the efficiency of multi-task parallel execution, it also needs to achieve the optimal allocation of resources.

[0004] Although domestic chips have reached the international advanced level in peak performance, the resource utilization rate in actual applications has not been fully utilized. Due to the lack of intelligent task allocation mechanism, some cores or acceleration units may be idle and all available resources cannot be fully utilized, resulting in resource waste. This problem limits the improvement of the overall system performance and affects cost-effectiveness.

[0005] Furthermore, the strict requirements for real-time and reliability in industrial environments have brought additional pressure to task scheduling. Many tasks have extremely high standards for response time and data processing accuracy, and existing scheduling algorithms may not be able to fully meet these requirements, especially when encountering sudden loads or abnormal situations. The stability and recovery capabilities of the system are particularly important. In addition, in edge computing scenarios, devices usually work in relatively harsh environments, and temperature changes under long-term operating conditions have a direct impact on performance. Therefore, it is necessary to consider how to reduce overall energy consumption through intelligent scheduling strategies while ensuring the long-term stable operation of the system. Summary of the invention

[0006] In order to solve the above-mentioned problems existing in the prior art, the present invention proposes an industrial production edge intelligent resource allocation and task scheduling method and system.

[0007] The technical solution of the present invention is as follows:

[0008] On the one hand, the present invention proposes an industrial production edge intelligent resource allocation and task scheduling method, comprising the following steps:

[0009] For any task in the industrial production process, the task criticality is calculated based on the task time and task resources; at the same time, the task scheduling objective function and constraint conditions are constructed, and the elastic resource influencing factors for completing the task are calculated;

[0010] Based on the deep learning algorithm, a state space is constructed, and resource utilization is set as a state in the state space, wherein the resource utilization is calculated according to task criticality and elastic resource changes; and a reward function is designed, wherein the reward function corresponds to a resource allocation strategy;

[0011] Initialize the state space, and update the Actor network and Critic network according to the reward function to generate a trained policy agent;

[0012] Deploy the trained policy agent in the edge device in the industrial production environment, and use the policy agent to perform resource allocation and task scheduling.

[0013] As a preferred implementation, the calculation formula for calculating task criticality based on task time and task resources is:

[0014]

[0015] Among them, E(v i ) represents the criticality of task i, is the earliest start time of task i, represents the latest start time of task i, t i is the execution time of task i, m is the total number of resources consumed by the task, r i j Indicates the jth resource consumed by task i, i Out is the number of elements in the predecessor task set corresponding to task i. i is the out-degree of the number of elements in the set of post-tasks corresponding to task i, τ, ρ, σ j , ι are respectively the preset time weight coefficient, execution weight coefficient, resource weight coefficient and in-degree weight coefficient;

[0016] in:

[0017]

[0018] in, is the set of subsequent tasks of task i, and the number of its elements is represented by the in-degree Out i ; is the latest start time of the subsequent task k, t k is the execution time of task k; Refers to the set of predecessor tasks of task i, and the number of its elements is represented by the in-degree In i ; represents the earliest start time of predecessor task j, t j is the execution time of task j.

[0019] As a preferred implementation, in the step of constructing the scheduling objective function and constraints of the task, the scheduling objective function is:

[0020] minimize(∑t i +∑a i p i ),v i ∈V;

[0021] Among them, minimize represents the minimization operation, t i represents the execution time of task i, a i is the weight coefficient of task i, p i represents the cost required for task i, and V represents the task set;

[0022] Constraints include critical path constraints and composite constraints, where the critical path constraints are:

[0023]

[0024] Among them, v u is the critical task u, CP represents the critical task set on the critical path, t u is the execution time of critical task u, T d is the total time set;

[0025] The composite constraints are:

[0026]

[0027] Among them, g j (t) represents the remaining resource j at time t; b k represents the start time of task k, c l represents the completion time of task l, b i ≥0,T d >0 is a constraint condition, which means that the start time of any task i is not less than time 0 and the total time is greater than 0.

[0028] As a preferred implementation, the formula for calculating the elastic resource influencing factors for completing the task is specifically:

[0029]

[0030] in, represents the elastic resource influencing factors for completing task i, The independent change of each resource quantity affects the task execution time. The changes of multiple resources together affect the task execution time.

[0031] As a preferred implementation, the calculation formula for the resource utilization is:

[0032]

[0033] in, represents the resource utilization of task i at time t, E(v i ) is the task criticality of task i, p i represents the cost required for task i, t i is the execution time of task i, represents the elastic resource influencing factors for completing task i, is the impact of the change in the amount of resources occupied by the predecessor task set of task i on the execution time of task i, T i The maximum execution time of the predecessor and successor tasks of task i.

[0034] As a preferred implementation, the reward function is specifically:

[0035]

[0036] Among them, r t (ζ) is the reward at time t, a is the performance change ratio after adopting the allocation strategy, and are the amount of resources allocated at time 0 and time t respectively; ζ is the cumulative discount parameter for intermediate rewards.

[0037] As a preferred implementation, in the step of updating the Actor network and the Critic network according to the reward function:

[0038] Update the parameters of the Critic network by minimizing the cumulative reward and the valuation reward;

[0039] Update the parameters of the Actor network by sampling policy gradients and estimated rewards.

[0040] On the other hand, the present invention also proposes an industrial production edge intelligent resource allocation and task scheduling system, comprising:

[0041] The resource allocation and task scheduling target model building module is used to calculate the task criticality based on task time and task resources for any task in the industrial production process; at the same time, it builds the task scheduling objective function and constraint conditions, and calculates the elastic resource influencing factors for completing the task;

[0042] A deep learning model building module is used to build a state space based on a deep learning algorithm, set resource utilization as a state in the state space, and calculate the resource utilization according to task criticality and elastic resource changes; and design a reward function, the reward function corresponding to the resource allocation strategy;

[0043] The agent training module is used to initialize the state space and update the Actor network and Critic network according to the reward function to generate a trained policy agent;

[0044] The resource allocation and task scheduling module is used to deploy the trained policy agent in the edge devices in the industrial production environment, and perform resource allocation and task scheduling through the policy agent.

[0045] On the other hand, the present invention also proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for industrial production edge intelligent resource allocation and task scheduling as described in any embodiment of the present invention is implemented.

[0046] On the other hand, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the industrial production edge intelligent resource allocation and task scheduling method as described in any embodiment of the present invention.

[0047] The beneficial effects of the present invention include:

[0048] Based on deep learning algorithms, the present invention optimizes or approximates the optimal resource allocation and task scheduling of edge intelligent computing units in industrial production environments by training strategic agents. By analyzing the criticality of each task, combining the task pre- and post-task set, completion time, and required resource quantity as inputs for agent training, a strategic agent is finally obtained. By deploying the agent online, corresponding strategies can be generated according to the current production environment.

[0049] Additional aspects and advantages of the present invention will be set forth in the following description, and some of them will be apparent from the description, or may be understood by practicing the present invention. In addition, the various aspects and advantages of the present invention may be realized and obtained by the method steps and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a method according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0053] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0054] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0055] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0056] Embodiment 1:

[0057] See also Figure 1 , an industrial production edge intelligent resource allocation and task scheduling method, comprising the following steps:

[0058] S100. For any task in the industrial production process, calculate the task criticality based on the task time and task resources; at the same time, construct the task scheduling objective function and constraint conditions, and calculate the elastic resource influencing factors for completing the task;

[0059] S200, constructing a state space based on a deep learning algorithm, setting resource utilization as a state in the state space, wherein the resource utilization is calculated according to task criticality and elastic resource changes; and designing a reward function, wherein the reward function corresponds to a resource allocation strategy;

[0060] S300, initialize the state space, and update the Actor network and Critic network according to the reward function to generate a trained strategic agent;

[0061] S400. Deploy the trained policy agent in the edge device in the industrial production environment, and perform resource allocation and task scheduling through the policy agent.

[0062] As a preferred implementation of this embodiment, the calculation formula for calculating the task criticality based on task time and task resources is:

[0063]

[0064] Among them, E(v i ) represents the criticality of task i, is the earliest start time of task i, represents the latest start time of task i, t i is the execution time of task i, m is the total number of resources consumed by the task, r i j Indicates the jth resource consumed by task i, i Out is the number of elements in the predecessor task set corresponding to task i. i is the out-degree of the number of elements in the set of post-tasks corresponding to task i, τ, ρ, σ j , ι are respectively the preset time weight coefficient, execution weight coefficient, resource weight coefficient and in-degree weight coefficient;

[0065] in:

[0066]

[0067] in, is the set of subsequent tasks of task i, and the number of its elements is represented by the in-degree Out i ; is the latest start time of the subsequent task k, t k is the execution time of task k; Refers to the set of predecessor tasks of task i, and the number of its elements is represented by the in-degree In i ; represents the earliest start time of predecessor task j, t j is the execution time of task j. Among them, the earliest start time of the first task can be determined as the 0 time when the entire task starts, and the latest start time of the last task can be determined as the maximum deadline of the entire task, that is, the total time.

[0068] As a preferred implementation of this embodiment, since there are associations and logical relationships between tasks, multiple tasks carried out simultaneously at the same time may require the resolution of resource conflicts. In the case of limited resources, this embodiment achieves optimization through a scheduling objective function and uses a linear programming equation to meet the constraints, where the scheduling objective function is:

[0069] minimize(∑t i +∑a i p i),v i ∈V;

[0070] Among them, minimize represents the minimization operation, t i represents the execution time of task i, a i is the weight coefficient of task i, p i represents the cost required for task i, and V represents the task set;

[0071] Constraints include critical path constraints and composite constraints, where the critical path constraints are:

[0072]

[0073] Among them, v u is the critical task u, CP represents the critical task set on the critical path, t u is the execution time of critical task u, T d is the total time set;

[0074] The composite constraints are:

[0075]

[0076] Among them, g j (t) represents the remaining resource j at time t; b k represents the start time of task k, c l represents the completion time of task l, Indicates that the completion time of a task is greater than the maximum completion time of all its predecessor tasks; b i ≥0,T d >0 is a constraint condition, which means that the start time of any task i is not less than time 0 and the total time is greater than 0.

[0077] As a preferred implementation of this embodiment, it is assumed that the task v is completed. i , needs to consume m kinds of resources The execution time is t i , then under the elastic resource condition t i The size varies with the amount of resources, and is expressed by the elastic resource influencing factors for completing the task:

[0078]

[0079] in, represents the elastic resource influencing factors for completing task i, The independent change of each resource quantity affects the task execution time. The changes of multiple resources together affect the task execution time.

[0080] In this embodiment, the DDPG algorithm is used as the deep learning algorithm, and the constructed state space is s(t)={s 1 (t),s 2 (t),…,s n (t)} represents the set of all possible states of the environment. In order to describe the effective resource allocation in the environment, the resource utilization Set to state i (t), where It represents resource utilization, that is, the temporal and spatial proportion of the total resources occupied by task i at time t.

[0081] As a preferred implementation of this embodiment, the calculation formula of the resource utilization is:

[0082]

[0083] in, represents the resource utilization of task i at time t, E(v i ) is the task criticality of task i, p i represents the cost required for task i, t i is the execution time of task i, represents the elastic resource influencing factors for completing task i, is the impact of the change in the amount of resources occupied by the predecessor task set of task i on the execution time of task i, T i The maximum execution time of the predecessor and successor tasks of task i.

[0084] In this embodiment, the action represents the agent's serialized decision on the amount of resources allocated to each task in the environment and maximizes the cumulative reward. The vector form of the action space is:

[0085] A=[a 1 ,a 2 …,a 3 ];

[0086] Among them, the resource allocation strategy is set to actiona t ,a t It is a vector composed of different resource allocation ratios, namely: in Represents the set of predecessor tasks p and the set of successor tasks s of task i. In order to satisfy the constraint that the sum of resource distribution satisfies 1, the output of the Actor network is normalized as follows:

[0087]

[0088] Among them, L u (v) gives the critical path from task u to task v, β 1 and β2 These are two parameters that can be cross-validated. In practice, Actor outputs The probability is extremely low, infinitely close to 0. When the value is less than the threshold λ=10 -5 When is 0.

[0089] As a preferred implementation of this embodiment, in order to optimize the resource allocation strategy, the design of the reward is related to the strategy, and the design of the reward function needs to consider effectively guiding the agent to allocate necessary resources for key tasks. The reward function is defined as follows:

[0090]

[0091] Among them, r t (ζ) is the reward at time t, a is the performance change ratio after the allocation strategy is adopted. If the ratio is positive, a penalty is given, and if the ratio is negative, a reward is given. and are the amount of resources allocated at time 0 and time t respectively, is the limit of minimizing resources, calculated according to the greedy algorithm; ζ is the cumulative discount parameter of the intermediate reward. In order to guide the agent to approach the optimal allocation strategy, the method proposed in this embodiment allocates an intermediate reward in each step of training to correct the next step strategy of the agent, that is, to add an additional penalty

[0092] As a preferred implementation of this embodiment, in the step of updating the Actor network and the Critic network according to the reward function:

[0093] The Critic network minimizes the cumulative reward y i (t) and valuation reward Q(s i (t),a i (t)|θ Q ) to update the parameters:

[0094]

[0095] At the same time, the parameters θ of the Actor network μ Updated by the sampling policy gradient as follows:

[0096]

[0097] For the target Actor network and the target Critic network, θ μ′ and θ Q′ The parameters are updated as follows:

[0098] θ μ′ ←τθ μ +(1-τ)θμ′ ;

[0099] θ Q′ ←τθ Q +(1-τ)θ Q′ ;

[0100] Among them, τ is the weight of the online network parameters

[0101] After training, the output μ(s(i)|θ of the online Actor network μ ) is the resource allocation strategy agent.

[0102] The generation process of the entire strategic agent is as follows:

[0103] Using random theta μ and θ Q Initialize the online Actor network μ(s(t)|θ μ and the online critic network Q(s(t),a(t)|θ Q );

[0104] Using Theta μ′ ←θ μ and θ Q′ ←θ Q Initialize the target Actor network μ ′ (s(t)|θ μ′ and the target critic network Q ′ (s(t),a(t)|θ Q′ )

[0105] Initialize experience replay buffer B

[0106] Initialize the regression process N∈=1.0

[0107] For each iteration, do the following:

[0108] Get the initial state:

[0109] Sampling Action:

[0110] Acquisition strategy:

[0111] Update status:

[0112] Get rewards:

[0113] Updated online Critic network and online Actor network.

[0114] Returns the generated policy agent μ(s(i)|θ μ ).

[0115] Embodiment 2:

[0116] This embodiment proposes an industrial production edge intelligent resource allocation and task scheduling system, including:

[0117] The resource allocation and task scheduling target model building module is used to calculate the task criticality based on the task time and task resources for any task in the industrial production process; at the same time, it builds the task scheduling objective function and constraint conditions, and calculates the elastic resource influencing factors for completing the task; this module is used to implement the function of step S100 in the first embodiment, and will not be repeated here;

[0118] A deep learning model construction module is used to construct a state space based on a deep learning algorithm, set resource utilization as a state in the state space, and calculate the resource utilization according to the task criticality and the elastic resource change; and design a reward function, which corresponds to the resource allocation strategy; this module is used to implement the function of step S200 in the first embodiment, and will not be repeated here;

[0119] The agent training module is used to initialize the state space and update the Actor network and the Critic network according to the reward function to generate a trained strategic agent; this module is used to implement the function of step S300 in the first embodiment, and will not be repeated here;

[0120] The resource allocation and task scheduling module is used to deploy the trained policy agent in the edge device in the industrial production environment, and perform resource allocation and task scheduling through the policy agent; this module is used to implement the function of step S400 in Example 1, which will not be repeated here.

[0121] Embodiment three:

[0122] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for industrial production edge intelligent resource allocation and task scheduling as described in any embodiment of the present invention is implemented.

[0123] Embodiment 4:

[0124] This embodiment proposes a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method for industrial production edge intelligent resource allocation and task scheduling as described in any embodiment of the present invention is implemented.

[0125] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0126] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0127] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0128] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0129] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An industrial production edge intelligent resource allocation and task scheduling method, characterized in that: The following steps are involved: For any task in the industrial production process, the task criticality is calculated based on the task time and task resources; at the same time, the task scheduling objective function and constraint conditions are constructed, and the elastic resource influencing factors for completing the task are calculated; Based on the deep learning algorithm, a state space is constructed, and resource utilization is set as a state in the state space, wherein the resource utilization is calculated according to task criticality and elastic resource changes; and a reward function is designed, wherein the reward function corresponds to a resource allocation strategy; Initialize the state space, and update the Actor network and Critic network according to the reward function to generate a trained policy agent; Deploy the trained policy agent in the edge device in the industrial production environment, and use the policy agent to perform resource allocation and task scheduling.

2. According to claim 1, an industrial production edge intelligent resource allocation and task scheduling method is characterized in that: The calculation formula for calculating task criticality based on task time and task resources is: Among them, E(v i ) represents the criticality of task i, is the earliest start time of task i, represents the latest start time of task i, t i is the execution time of task i, m is the total number of resources consumed by the task, r i j Indicates the jth resource consumed by task i, i Out is the number of elements in the predecessor task set corresponding to task i. i is the out-degree of the number of elements in the set of post-tasks corresponding to task i, τ, ρ, σ j , ι are respectively the preset time weight coefficient, execution weight coefficient, resource weight coefficient and in-degree weight coefficient; in: Among them, V i S is the set of subsequent tasks of task i, and the number of its elements is represented by the in-degree Out i ; is the latest start time of the subsequent task k, t j is the execution time of task k; V i P Refers to the set of predecessor tasks of task i, and the number of its elements is represented by the in-degree In i ; represents the earliest start time of predecessor task j, t j is the execution time of task j.

3. The method for industrial production edge intelligent resource allocation and task scheduling according to claim 1, characterized in that: In the step of constructing the scheduling objective function and constraint conditions of the task, the scheduling objective function is: minimize(∑t i +∑a i p i ),v i ∈V; Among them, minimize represents the minimization operation, t i represents the execution time of task i, a i is the weight coefficient of task i, p i represents the cost required for task i, and V represents the task set; Constraints include critical path constraints and composite constraints, where the critical path constraints are: Among them, v u is the critical task u, CP represents the critical task set on the critical path, t u is the execution time of critical task u, T d is the total time set; The composite constraints are: Among them, g j (t) represents the remaining resource j at time t; b k represents the start time of task k, c l represents the completion time of task l, b i ≥0,T d >0 is a constraint condition, which means that the start time of any task i is not less than time 0 and the total time is greater than 0.

4. The method for industrial production edge intelligent resource allocation and task scheduling according to claim 1, characterized in that: The formula for calculating the elastic resource influencing factors for completing the task is specifically: in, represents the elastic resource influencing factors for completing task i, The independent change of each resource quantity affects the task execution time. The changes of multiple resources together affect the task execution time.

5. The method for industrial production edge intelligent resource allocation and task scheduling according to claim 1, characterized in that: The calculation formula of the resource utilization is: in, represents the resource utilization of task i at time t, E(v i ) is the task criticality of task i, p i represents the cost required for task i, t i is the execution time of task i, represents the elastic resource influencing factors for completing task i, is the impact of the change in the amount of resources occupied by the predecessor task set of task i on the execution time of task i, T i The maximum execution time of the predecessor and successor tasks of task i.

6. The method for industrial production edge intelligent resource allocation and task scheduling according to claim 1, characterized in that: The reward function is specifically: Among them, r t (ζ) is the reward at time t, a is the performance change ratio after adopting the allocation strategy, and are the amount of resources allocated at time 0 and time t respectively; ζ is the cumulative discount parameter for intermediate rewards.

7. The method for industrial production edge intelligent resource allocation and task scheduling according to claim 1, characterized in that: In the step of updating the Actor network and Critic network according to the reward function: Update the parameters of the Critic network by minimizing the cumulative reward and the valuation reward; Update the parameters of the Actor network by sampling policy gradients and estimated rewards.

8. An industrial production edge intelligent resource allocation and task scheduling system, characterized in that: include: The resource allocation and task scheduling target model building module is used to calculate the task criticality based on task time and task resources for any task in the industrial production process; at the same time, it builds the task scheduling objective function and constraint conditions, and calculates the elastic resource influencing factors for completing the task; A deep learning model building module is used to build a state space based on a deep learning algorithm, set resource utilization as a state in the state space, and calculate the resource utilization according to task criticality and elastic resource changes; and design a reward function, the reward function corresponding to the resource allocation strategy; The agent training module is used to initialize the state space and update the Actor network and Critic network according to the reward function to generate a trained policy agent; The resource allocation and task scheduling module is used to deploy the trained policy agent in the edge devices in the industrial production environment, and perform resource allocation and task scheduling through the policy agent.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the industrial production edge intelligent resource allocation and task scheduling method as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the industrial production edge intelligent resource allocation and task scheduling method as described in any one of claims 1 to 7.

Images